<%BANNER%>

Clause-Final Negation in Brazilian Portuguese

Permanent Link: http://ufdc.ufl.edu/UFE0041949/00001

Material Information

Title: Clause-Final Negation in Brazilian Portuguese
Physical Description: 1 online resource (238 p.)
Language: english
Creator: Hansen, Quinn
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: brazilian, clause, discourse, negation, pragmatics, semantics, syntax
Linguistics -- Dissertations, Academic -- UF
Genre: Linguistics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: My study explores Clause-Final Negation in Brazilian Portuguese, within a Minimalist Program framework. Brazilian Portuguese makes use of three types of sentential negation. In Type 1, the negative marker appears preverbally (i). Type 2 has both a preverbal marker and a clause-final marker (ii). Type 3 has only a clause final negative marker (iii). i. Eu na tildeo quero o bolo TYPE 1 1S NEG want.PRS.1S DET cake I do not want the cake. ii. Eu na tildeo quero o bolo na tildeo TYPE 2 1S NEG want.PRS.1S DET cake NEG I do not want the cake. iii. Eu quero o bolo na tildeo TYPE 3 1S want.PRS.1SG DET cake NEG I do not want the cake. In this work, I analyze preverbal negation as the head of a functional projection NegP within a split IP scenario (Pollock 1989, Costa and Galves 2002). I analyze Clause-Final Negation as resulting from leftward syntactic movement of the whole IP to the specifier of a Topic Phrase which is headed by na tildeo. Because the IP moves to the left, na tildeo is clause-final. The movement of IP is motivated by the discourse status of the IP. Following Schwenter (2005), I propose that Clause-Final Negation is only licit when the proposition that it negates is discourse-old when it is a kind of topic. The movement analysis accounts for both syntactic and semantic/pragmatic restrictions on clause-final negation.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Quinn Hansen.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Potsdam, Eric H.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041949:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041949/00001

Material Information

Title: Clause-Final Negation in Brazilian Portuguese
Physical Description: 1 online resource (238 p.)
Language: english
Creator: Hansen, Quinn
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: brazilian, clause, discourse, negation, pragmatics, semantics, syntax
Linguistics -- Dissertations, Academic -- UF
Genre: Linguistics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: My study explores Clause-Final Negation in Brazilian Portuguese, within a Minimalist Program framework. Brazilian Portuguese makes use of three types of sentential negation. In Type 1, the negative marker appears preverbally (i). Type 2 has both a preverbal marker and a clause-final marker (ii). Type 3 has only a clause final negative marker (iii). i. Eu na tildeo quero o bolo TYPE 1 1S NEG want.PRS.1S DET cake I do not want the cake. ii. Eu na tildeo quero o bolo na tildeo TYPE 2 1S NEG want.PRS.1S DET cake NEG I do not want the cake. iii. Eu quero o bolo na tildeo TYPE 3 1S want.PRS.1SG DET cake NEG I do not want the cake. In this work, I analyze preverbal negation as the head of a functional projection NegP within a split IP scenario (Pollock 1989, Costa and Galves 2002). I analyze Clause-Final Negation as resulting from leftward syntactic movement of the whole IP to the specifier of a Topic Phrase which is headed by na tildeo. Because the IP moves to the left, na tildeo is clause-final. The movement of IP is motivated by the discourse status of the IP. Following Schwenter (2005), I propose that Clause-Final Negation is only licit when the proposition that it negates is discourse-old when it is a kind of topic. The movement analysis accounts for both syntactic and semantic/pragmatic restrictions on clause-final negation.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Quinn Hansen.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Potsdam, Eric H.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041949:00001


This item has the following downloads:


Full Text





CLAUSE-FINAL NEGATION IN BRAZILIAN PORTUGUESE


By

QUINN MCCOY HANSEN















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2010

































2010 Quinn McCoy Hansen

































To my light and my flower
Para LQcia e Sienna Yasmin
Amo-vos









ACKNOWLEDGMENTS

I want to thank Eric Potsdam, my supervisory committee chair, for teaching me the

art and science of conducting and producing linguistic research.

Additionally, I thank Dr. Brent Henderson, Dr. Gary Miller, Dr. Elizabeth Ginway,

and Dr. Benjamin Hebblethwaite for being part of my committee. I thank Dr. Henderson

for his patience with my questions, most of which were elementary. Dr. Miller deserves

thanks for always providing stimulating debate and conversation in and out of class. I

thank Dr. Hebblethwaite for introducing me to the world of Creole Studies and showing

me with concrete evidence how linguistics can have a direct impact on societies.

Finally, I thank Dr. Ginway for allowing me to teach Portuguese, and allowing me the

freedom to run the courses as best I saw fit.

I am extremely grateful to all my Brazilian Portuguese consultants, both here in the

United States and in Brazil. Particularly, I thank Sharon Barkley and Andrea Ferreira for

their judgments and comments about use. Also, here in the States, I thank Cilene

Rodrigues and Ana Bastos for sending me articles in Portuguese and adding their

grammaticality judgments. I thank Megan Armstrong and Scott Schwenter for their

comments on my work. In Brazil, I want to thank Julia Ramos, Florbela Silva, Divino,

Ednilson, a Nordestina Valentona, Jose Fernandes and his employees at NorteSul, and

Abilio Joaquim Fernandes. Also, I thank the producers of the show Sem Meias-

Palavras for producing a terribly low-class show that provided several hundred

examples of Clause-Final Negation.

I owe many thanks to all the people in the Linguistics Program that taught me in

diverse ways, from formal classroom settings to informal chats.









I thank my parents for showing me that I can do whatever I want in life, and for

teaching me that reading and learning are the keys to happiness.

I thank my daughter Sienna Yasmin for having patience with me, even when I had

to stare at a book or a screen for several hours straight. I hope that someday she will

understand the sacrifices that she has made for me.

I thank my wife LOcia, without whom I would have never begun this journey. Also,

I thank her for putting up with hours of questions about Brazilian Portuguese and

negative sentences, all the while, smiling brightly.









TABLE OF CONTENTS

page

A C K N O W LED G M ENTS ........... ............................ ..... .......................... ............... 4

LIST OF TABLES ............. ............ ............................. 9

LIST OF ABBREVIATIONS..................... .......... .............................. 10

A B S T R A C T ...................................................................................................... 1 2

CHAPTER

1 INTRODUCTION ...................................................................... ......... ................... 14

1.1 Research Questions ... ............ .. ................................... 14
1.2 Domain of Investigation: Brazilian Portuguese ......... ... ............................. 16
1.3 Three Types of Sentential Negation ........................... ......... .................. 17
1.4 Past Corpus Studies of Clause-Final Negation ............ ........ .................. 20
1.4.1 Roncarti 1996 .......................................... ....... 20
1.4.2 Da Cunha 1996, 2001, 2007 ............................ ... ............. .......... 21
1.4.3 Alkmim 1999.................................... ............... 21
1.4.4 Camargos 2000 ............................................................. 22
1.4.5 S ousa 2004 ................................. ......... ............................ 23
1.4.6 Cavalcante 2007............................. .................... 23
1.4.7 Summary of Corpus Studies.................. ............................... ..... 24
1.5 Syntactic and Semantic/Pragmatic Restrictions in BP for T2 and T3............. 25
1.5.1 Discourse Requirements for T2 and T3 ............ ......... ................... 25
1.5.2 Em bedded Clauses .......... ... ...... ........................... ......... ...... 26
1.5.3 Relative Clauses............................... .................. 28
1.5.4 Topics 31
1.5.5 W h-Questions......................................... ........ 32
1.5.6 N-W ords and Negative Concord ................... ......... .................... 33
1.5.7 Imperatives ................ ........................... 36
1.5.8 N P I Idiom s.............................. ............... 37
1.5 .9 S um m ary of the D ata .................................................. ... .. ............... 38
1.6 Structure of the Study ......................................... ................ .............. 39

2 THEORETICAL FOUNDATIONS...................... ...... ........................... 42

2.1 Introduction ...................................................................... ......... ................... 42
2.2 Minimalism ......... .......... 2.......... ...... ........ 42
2.2.1 Fundam ental Principles of M inim alism .............. ............ .................. 42
2.2.2 Probe-Goal System and Feature Agreement .................................... 45
2.3 CP: The Left Periphery............................... .................... 47
2.3.1 Rizzi's Split CP .......... ......... ................................... ......... 47
2.3.2 The Split CP and BP................................... ................. 49


6









2 .4 C o n c lu s io n ......... ......... ............................................................................. 5 3

3 THE CLAUSE STRUCTURE OF BRAZILIAN PORTUGUESE ........................... 54

3.1 Introduction .................. ......... .......... ......... 54
3 .2 S subjects in B P ................................ ................ ........... ..... ............... 55
3.2.1 The Debate over the Romance Preverbal Subject Position.............. 55
3.2.2 BP's Preverbal Subject.............................. ............... 63
3.2.2.1 Tem poral Adverbs ............................. .............. ........... 64
3.2.2.2 W eak Pronouns ............... .... ........................... ........... 65
3.2.2.3 Quantifiers ........................ ..................... 66
3.3 V erbs and V erb M ovem ent.................................................... ... .. ............... 68
3.3.1 IP 69
3.3.2 N on-S plit IP ......... ............................ ............. .............. .............. 72
3.3.3 A grP in B P ..... ................ ................................. .......... 74
3.4 Fronting in BP ....................... ......................... ........ ........ 84
3.5 W h-Q questions: Fronted and In-Situ..................................................... ....... 89
3.5.1 Types and Conditions for Wh-Words: Movement and In-Situ.............. 89
3.5.2 Deriving W h-Questions in BP .................................... ...................... 92
3 .6 C o nclusions ............................................. ...... ........... 97

4 PREVERBAL NEGATIO N................................................ ..................... 100

4.1 Introduction ............................................................. .............. 100
4.2 The Syntax of Negation................................................. ............... 100
4.2.1 NegP 101
4.2.2 The Location of NegP .............................. ................... 105
4.2.2.1 Laka ................ ........ ............... ........... 105
4.2.2.2 Zanuttini ............ ......... ................... 108
4.3 Preverbal Negation in BP ......................................................................... 111
4.3.1 The Phonology of the Preverbal Negative Marker........................... 112
4.3.2 Mioto 1992 & 1998 ...................................... .. ............... 114
4.3.3 M artins 1994 and Nam iuti 2008.................................... ............... 117
4.3.4 The structural Position of NegP in BP ............................................ 120
4.4 C conclusion ............. .............. ................. ............... ................ 122

5 THE PRAGMATICS AND SEMANTICS OF CLAUSE-FINAL NEGATION .......... 124

5 .1 Introd auction ............................................................ ... .. .......................... 124
5.2 Previous Approaches to Clause-Final Negation: Semantics/Pragmatics ..... 125
5.2 .1 E m phasis.................. ........................ ...... .......... ........... 125
5.2.2 Presuppositional or Contrary-to-Fact ...................... ..... ............. 130
5.2.3 Discourse-Old ......... ......... .. ....... .......... ................... 133
5.3 A applying the Theory ...................... ....... ......... .. ............................ 145
5.4 Conclusion .......................... ......... .... .................. 149









6 THE SYNTAX OF CLAUSE-FINAL NEGATION ......................................... 150

6.1 Introduction ....................... ................... ................................. 150
6.2 Previous A ccounts........................................... .................. 150
6.2.1 M artins 1997 ................................................................................ 150
6.2.2 Fonseca 2004 ........................................ ................ .............. 158
6.2.3 Cavalcante 2007 .................. ......................... ...... .. ... ....... 161
6.3 A Proposal for Clause-Final Negation in Brazilian Portuguese ................. 166
6.3.1 Derivation of Clause-Final Negation ............ .................................. 169
6.3.2 W h-W words ...................................... .......... 175
6.3.3 Relative Clauses............................... .................. 178
6.3.4 Em bedded Clauses ................. ....................... ............... 181
6.3.5 Topicalization................................... 182
6.3.6 Imperatives ............ ....... .... ........................... 184
6.3.7 Semantic Interpretation of Clause-Final Negation........................... 186
6.4 C conclusion .............................................. .............. ................ 192

7 NEGATIVE CONCORD IN BRAZILIAN PORTUGUESE ................................... 194

7.1 Introduction ......................................... ................................ 194
7 .2 N e g active C o nco rd ......................................................................... 194
7.2.1 Definition and Types of Negative Concord ............... .... ............ 195
7.2.2 N-W ords are Negative Q uantifiers................................................... 200
7.2.3 N-Words are Non-Negative ........... ...................... ...... 204
7.2.4 Zeijlstra's Feature-Based Approach to Negative Concord................ 207
7.2.5 Summary .................. ..... ........ ................... ...... 211
7.3 N-Words and Negative Concord in Brazilian Portuguese........................... 211
7.3.1 BP N-W words ........... ............... ...... ............ .......... 211
7.3.2 Feature Agreement and BP N-Words............. ................ .......... 215
7.4 Negative Concord and Clause-Final Negation in Brazilian Portuguese ....... 217
7.5 C conclusion ...................................... .... ................................. 221

8 CONCLUSION ................. ..... ....... .... ..... .... .................. 222

LIST O F REFER ENC ES ......... ................. ........................................ .............. 224

BIOGRAPHICAL SKETCH .......... ........ .. ................. .... ...................... 238









LIST OF TABLES

Table page

1-1 Syntactic Restrictions for BP CFN ......... ......................... ... ............... 38

3-1 Loss of Verbal Morphology in BP ................................ ............... 73

6-1 Syntactic Restrictions for BP CFN ................ ....... ..... ................... .............. 168

7-1 N -W words and C FN .......................................................... .............. 218









LIST OF ABBREVIATIONS

* Ungrammatical/Unacceptable

[*] Strong Feature

? Degraded Grammaticality

?? Highly Degraded Grammaticality

/ Grammatical

Negative/Negation

1 First Person

2 Second Person

3 Third Person

AGR Agreement

asp Aspect Marker

BP Brazilian Portuguese

CFN Clause-final Negation

cl Clitic

CLLD Clitic Left Dislocation

comp Complementizer

DenP Denial Phrase

det Determiner

DN Double Negative

emph Emphasis

f Feminine

FQ Floating Quantifier

Foc Focus Phrase

fut Future Tense









imp Imperative

imper Past Imperfect

[iNEG] Interpretable negative feature

inf Infinite

IP Inflectional Phrase

JC Jespersen Cycle

m Masculine

NC Negative Concord

neg Negative Marker

NPI Negative Polarity Item

NSL Null Subject Language

Op Syntactic operator

p Plural

pos Positive

prep Preposition

pst Past Tense

QP Quantifier Phrase

s Singular

sbj Subjunctive

spec Specifier

T Tense

t Trace

[uNEG] Uninterpretable Negative Feature

VP Verb Phrase

[WH] Interrogative Feature









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

CLAUSE-FINAL NEGATION IN BRAZILIAN PORTUGUESE

By

Quinn McCoy Hansen

August 2010

Chair: Eric Potsdam
Major: Linguistics


My study explores Clause-Final Negation in Brazilian Portuguese, within a

Minimalist Program framework. Brazilian Portuguese makes use of three types of

sentential negation. In Type 1, the negative marker appears preverbally (i). Type 2 has

both a preverbal marker and a clause-final marker (ii). Type 3 has only a clause final

negative marker (iii).

i. Eu nao quero o bolo TYPE 1
1S NEG want.PRS.1S DET cake
'I do not want the cake.'

ii. Eu nao quero o bolo nao TYPE 2
1S NEG want.PRS.1S DET cake NEG
'I do not want the cake.'

iii. Eu quero o bolo nao TYPE 3
1s want.PRS. SGthe cake NEG
'I do not want the cake.'


In this work, I analyze preverbal negation as the head of a functional projection

NegP within a split IP scenario (Pollock 1989, Costa and Galves 2002). I analyze

Clause-Final Negation as resulting from leftward syntactic movement of the whole IP to

the specifier of a Topic Phrase which is headed by nao. Because the IP moves to the









left, nao is clause-final. The movement of IP is motivated by the discourse status of the

IP. Following Schwenter (2005), I propose that Clause-Final Negation is only licit when

the proposition that it negates is discourse-old-when it is a kind of topic. The

movement analysis accounts for both syntactic and semantic/pragmatic restrictions on

clause-final negation.









CHAPTER 1
INTRODUCTION

1.1 Research Questions

The goal of this dissertation is to examine syntactic, semantic, and pragmatic

issues surrounding the analysis of sentential negation in Brazilian Portuguese (BP

henceforth). Specifically, this work explores Clause-Final Negation (CFN) in BP and

gives a syntactic account within a minimalist framework. Negation and other topics

related to polarity have been a part of linguistic study since well before the generative

approach began (Jespersen 1917) because they aid in understanding syntactic

structures, semantics, and scope (Lasnik 1972, Emonds 1978, Pollock 1989 Laka 1990,

Zanuttini 1997, Zeijlstra 2004).

BP has three options for sentential negation. In what I call Type 1 (T1), the

negative marker appears preverbally as seen in (1)a. Type 2 (T2) has both a preverbal

marker and a postverbal marker that is clause-final, example (1)b. Type 3 (T3) only has

a clause-final negative marker, example (1)c.


(1) a. Eu nao quero o bolo TYPE 1 (T1)
1S NEG want.PRS.1S DET cake
'I do not want the cake.'

b. Eu nao quero o bolo nao TYPE 2 (T2)
1S NEG want.PRS.1S DET cake NEG
'I do not want the cake.'

c. Eu quero o bolo nao TYPE 3 (T3)
1s want.PRS.1s DET cake NEG
'I do not want the cake.'

The existence of CFN seen in BP, as in T2 and T3, is not common for Romance

languages. These patterns have not been widely discussed in the literature, and the fact









that all three types exist in a single language is also unusual (see Schwegler 1991 b for

other Romance languages with clause-final negation).

The two main goals of this dissertation are 1) to document the three types of

negation in BP and 2) to understand the three types of negation with a minimalist

syntactic framework.


This dissertation answers the following questions:


Question 1: What are the patterns of negation in BP and how do they differ from
European Portuguese and other Romance languages?


Question 2: What is the syntactic analysis of the different sentential negation types?
a. What is the syntactic status of nao (i.e. head, phrase, or specifier)?
b. How can a syntactic analysis of BP negation be understood within a
minimalist framework?
c. What is the distribution of negative words and how does negative
concord work in BP?

The rest of this chapter answers Question 1 by presenting the empirical domain of

investigation. The subsequent chapters of this work answer Question 2 and sub-

questions (a), (b), and (c).

The remainder of this chapter is divided into four sections. Section 1.2 addresses

the domain of investigation, how data was collected and analyzed, and this section

introduces Brazilian Portuguese generally. Section 1.3 discusses the different types of

sentential negation crosslinguistically. Here I present examples of the three types of

negation seen in BP that are used in other languages. Section 1.4 discusses previous

corpus studies that have looked at CFN in BP. Section 1.5 lays out the principal data









for this dissertation. This section shows the syntactic structures that allow CFN in BP

as well as the structures that prohibit it. Additionally, this section introduces the

discourse situations where CFN is grammatical but infelicitous. The final section of this

chapter summarizes the data and gives an overview of the structure of this dissertation.

1.2 Domain of Investigation: Brazilian Portuguese

BP is a Romance language spoken by the majority of Brazilians, and government

numbers estimate that there are more than 200 million speakers of BP

(www.brasil.gov.br). The number of varieties is undefined, but BP can roughly be broken

up into 5 main dialects, northern, northeastern, central, southeastern, and southern

(Azevedo 2005).

With respect to negation, I assume that all three types of negation are found

among all dialects and socio-economic classes (Schwenter 2005). However, this fact is

not overtly admitted by the upper-class because they see T3 as 'caipira' or 'country

bumpkin' speech (Barme 2000). To counter this stereotype and to collect data for this

study, I use naturalistic data from reality TV, popular talk shows, and internet chat

rooms in addition to grammaticality judgments on elicited data and data from past

corpus studies presented in this chapter. The grammaticality judgments were made

principally by Portuguese speakers residing in Brazil, although Brazilians living in the

United States also provided some data.

The history of T2 and T3, clause-final negation, is not well known. Because of the

lack of written material that references this type of negation, I will not be concerned with

the historical aspects. Furthermore, CFN in BP is relatively new as a topic of study

evidenced by the fact that the first scientific attempt to understand this phenomenon in

BP was Schwegler (1991 a).









1.3 Three Types of Sentential Negation

The three types of sentential negation in BP are repeated below. By types of

sentential negation I merely mean the linear position of the negative marker in

relationship to the verb. Sentences in (1), repeated here, semantically represent the

same propositional meaning (Schwenter 2005:1429).


a. Eu nao quero
1S NEG want.PRS.1S
'I do not want the cake.'

b. Eu nao quero
1S NEG want.PRS.1S
'I do not want the cake.'

c. Eu quero o
1s want.PRS.1S DET
'I do not want the cake.'


0
DET


0
DET


bolo
cake


bolo
cake


bolo nao
cake NEG


nao
NEG


T1, preverbal negation, is found in Spanish and other Romance languages and many

researchers argue that these negative markers are syntactically similar (Mioto 1992,

Zanuttini 1997, 2001, Zeijlstra 2004).


T1


(3) Yo no quiero
1S NEG want.PRS.1S
'I do not want the book.'


(4) Maria non studia.
Maria NEG study.PRES.3SG
'Maria doesn't study'


(5) Tu nao fizeste
2s NEG do.PST.2S
'You didn't do the test'


el libro
DET book


o exame
DET test


European Portuguese


Spanish


Italian










T2 is not common in other Romance languages; however, it exists in some

Portuguese Creoles, such as Sao Tome Creole, as well as in some Spanish dialects of

the Americas, for example Choc6 Spanish of Columbia (Schwegler 1991a:95, 1991b).


T2


(6) No me gust6 eso alia no Choc6 Spanish
NEG 1s please.PST.3s that there NEG
'I don't like that over there'


(7) A na kuvida n6 fa Sao Tome Creole
IMPER NEG invite 1P NEG
'They didn't invite us' (Ferraz 1979:68)

Another Indo-European language, Afrikaans, has a T2 construction similar to BP

(Biberauer 2009).


(8) Ek het nie boeke gelees nie Afrikaans
1s have.PRs.1s NEG books read NEG
'I have not read books'

Negative constructions similar to T2 are found in other Romance languages, such

as French, Northern Italian dialects, and Catalan. The difference is that the position of

the second negative marker is postverbal and not clause-final, as in the French

example with 'the meat' being clause final and the negative marker postverbal (Zanuttini

1997:17):


(9) Jean n'aime pas la viande French
Jean NEG-love.3S NEG DET meat
'Jean doesn't like the meat'










Berini & Remat (1996) shows that the position of the CFN marker in BP is indeed

clause-final and not postverbal. The difference between the BP and most other

Romance languages is that there is almost no limit to the length of material between the

two instances of negation, as seen in (10).


(10) A Sienna ndo quer saber como 6 que ele chegou a minha casa nao.
DET Sienna NEG want.3s know.INF how is that he arrive.PST.3S PREP 1s house NEG
'Sienna doesn't want to know how he arrived at my house.'


T3 negation is rare; nonetheless, there are certain cases in Romance where it

does appear, as in Milanese, a dialect of Italian, (11) (Zanuttini 1997:5). Also, it is found

in some languages of West Africa such as Ewe and Fon (Dryer 2007). Example (12) is

from the Nigerian language Birom (Bouquiaux (1970:386) cited in Dryer (2007)).

Example (13) shows T3 in Lung'iye, a Portuguese Creole spoken on the island of Sao

Tome and Principe.


T3

(11) el I'ha scrivuu no Milanese
3s 3s.CL-has.3s written NEG
'He hasn't written'


(12) yen a-tos ney wet Birom
3P AORIST-bear children NEG
'They have not given birth to children'


(13) na sa podi da ci fa Lung'iye
1SASP able give 2s NEG
'I can't give it to you' (GCnther 1973:78)









These examples help position BP within the spectrum of the different negation

strategies types. Although no single negative type is extraordinary, the fact that all

three currently exist in one language is rare. That is, some languages allow more than

one negative type, but they do not synchronically allow the three types that exist in BP.

1.4 Past Corpus Studies of Clause-Final Negation

This section discusses some corpus studies of BP directly related to negation.

These studies document the frequency of use for each negation type. It should be

noted that in nearly all cases, T3 is the least frequently used type, followed by T2. T1 is

par excellence the norm for negation in BP. Most researchers in these studies collected

their data from the northeastern states of Brazil. Although the population of these states

is generally perceived as using CFN more than other parts of Brazil, T2 and T3 negation

takes place in all regions of Brazil (Scherre 2007, Schwegler 1991a, Schwenter 2005,

and Cavalcante 2007). I conclude this section by commenting on the collective results

of these studies.

1.4.1 Roncarti 1996

Roncarti (1996) is a spoken corpus study of the BP dialect in the northeastern city

of Fortaleza. One of the main purposes of this work is to understand how the

expression sei nao 'I don't know (lit. I know not)' came about. Working within a

functionalist perspective, Roncarti proposes that this expression came about either

through de-grammaticalization of a syntactic element to a discourse element or through

the lexicalization of a syntactic element to a lexical item. Her research studied the

syntactic situations in which each type of negation is used, such as type of clause, null

subject, verbal repetition etc.









Her corpus totaled 822 sentence, with nine of these unanalyzed. Her study points

to an overall preference for T1 at 77%. Table (14) shows the division of frequency.


(14) Roncarti 1996:


T1 T2 T3 Total

Tokens 625 149 39 813

Frequency 77% 18% 5% 100%

1.4.2 Da Cunha 1996, 2001, 2007

Like Roncarti, Da Cunha uses functionalist paradigms to understand the rise of the

clause-final negative marker and what she calls the 'weakening' of the preverbal marker

in terms of phonology and semantics. Her corpus is from another northeastern city,

Natal. She uses not only spoken data but also a written corpus from 12 consultants

from three different levels of education in terms of highest level achieved: elementary,

high school, and post-high school education. Da Cunha's theoretical conclusions are

presented in chapter five. As with the other studies, T1 is the dominant method for

negation; Da Cunha tabulated a low T3 rate, merely 9 (.6%) out of 1465 total tokens.


(15) Da Cunha 1996:

T1 T2 T3 Total

Tokens 1298 158 9 1465

Frequency 88.6% 10.8% .6% 100%



1.4.3 Alkmim 1999

Alkmim analyzes social factors that might influence the rise and use of clause-final

negation, such as race, age, city of origin, etc. Her study took place in two rural towns,









Mariana and Pombal, both of which are located in the southwestern state Minas Gerais.

Both are largely Afro-Brazilian towns populated by those that are 'illiterate and in a

lower social class.' The idea that clause-final negation is related to Afro-Brazilians is

central to her work. Alkmim postulates that linguistic contact between slaves from West

Africa and the Portuguese during the slave period of Brazil produced a change in how to

negate a sentence. The numbers in table (16) are similar to other studies.


(16) Alkmim 1999:

T1 T2 T3 Total

Tokens 1791 491 40 2322

Frequency 77.1% 21.2% 1.7% 100%


1.4.4 Camargos 2000

Camargos also works in Minas Gerais, although not as geographically specialized

as Alkmim. He differs from some of the previous works because he also considers

sentences with other preverbal negative elements as being instances of T2 negation

(e.g. nunca 'never', ningu6m 'nobody', etc). The other authors did not count these as

part of their corpus because their main goal was to investigate the weakening of the

preverbal marker.

Much like Roncarti, Camargos looks at the syntactic situations where each type of

negation occurs. He also takes into account social factors to understand the use and

function of clause-final negation.










(17) Camargos 2000:


T1 T2 T3 Total

Tokens 687 265 28 980

Frequency 70% 27% 3% 100%

1.4.5 Sousa 2004

Sousa (2004) studies the small Afro-Brazilian town of Helvicia in the state of

Salvador in northeastern Brazil. In her study, following Schwegler's analysis for Rio de

Janeiro and Salvador, she does not include T3 as a type in its own right. Instead, she

claims that T3 is merely T2 with deletion of the preverbal marker. It is therefore difficult

to make too many claims about her findings for the purposes of this dissertation.

Nonetheless, it stills points to two-thirds majority of T1.


(18) Sousa 2004:

T1 T2 and T3 Total

Tokens 943 465 1408

Frequency 67% 33% 100%


1.4.6 Cavalcante 2007

Cavalcante's work is part of a larger effort to understand the connection of the

creolization of Portuguese to the current state of BP. The Projeto Vertentes do

Portugues Rural da Bahia out of the Federal University of Bahia analyzes rural dialects

in the state of Salvador where there is a large Afro-Brazilian population. Cavalcante

concentrated on three communities: Cinzento, Rio de Contas and Sape. He conducted

six interviews in each town with consultants who were not only born in the town, but









whose parents were born there as well. The consultants were also divided into three

age groups: 20 to 40, 40 to 60, and 60 and above. Cavalcante's study makes some

important contributions to this dissertation, and chapter six examines them extensively.

Of all the studies presented, his is the study where T1 has the lowest frequency of 66%,

much like Sousa's in Helvicia.


(19) Cavalcante 2007

T1 T2 T3 Total

Tokens 1343 568 115 2026

Frequency 66% 28% 6% 100%

1.4.7 Summary of Corpus Studies

These sections briefly explained the purpose and results of six corpus studies that

looked at CFN. In most cases, these authors did not attempt to make any specific

syntactic hypotheses; however, they each have their conclusions, some of which I treat

in chapters five and six. The data show considerable differences among the studies.

These differences could be attributed to regional and socio-ethnic factors. For example,

in the cases where clause-final negation is most prevalent, the main focus of the study

was rural Afro-Brazilians (Sousa 2004, Cavalcante 2007). In other cases, the studies

were more of a cross-section of the Brazilian population, which points to the depth of

use of clause-final negation. Additionally, in some cases the studies limited themselves

to typical nao... nao sentences, which could greatly reduce the numbers of occurrences

(Cavalcante 2007).

From the data presented above, it is apparent that T1 is clearly the predominant

form of sentential negation in BP. A conclusion from the next section is that T1 is









allowed in all contexts in which negation is used (Cavalcante 2007), and that T2 and T3

are marked forms (Schwenter 2005). As stated in the first section of this chapter, one

goal is to understand the motivations for T2 and T3 both syntactically and

semantic/pragmatically. This section has given quantitative evidence that a difference

exists among the three types of negation, thus validating the reason for this study.

1.5 Syntactic and Semantic/Pragmatic Restrictions in BP for T2 and T3

An important part of this work determines the syntactic and semantic differences

between the types. As mentioned, T1 is allowed in all contexts in BP (Schwenter 2005),

and in every case below, if the sentence is ungrammatical with CFN, its remedy is T1

negation (Biberauer & Cyrino 2009). Thus, the preverbal negative marker differs from

the clause-final nao (T2 and T3), which is restricted in its use. Below are instances

where T2 and T3 are restricted. This section begins by discussing the

semantic/pragmatic restrictions. Following that, I present several syntactic situations

where the syntactic structure restricts the use of T2 and T3. This includes a brief

discussion as to how these two types differ.

1.5.1 Discourse Requirements for T2 and T3

The main difference between T1 and the other two types is that T2 and T3 are

linked to the discourse (Schwenter 2005). This means that for T2 and T3 to be

pragmatically felicitous, the proposition must have already been introduced into the

discourse. These topics are explained in greater detail in chapter five where I explain

the discourse requirements for CFN. Simply put, if the information being negated is not

already introduced into the discourse, then the clause-final negative marker is not licit.

In questions like (20), often termed 'out-of-the-blue questions', what is being negated in









the answer has not been introduced into the discourse and therefore T2 and T3 are not


felicitous


(20) A. 0 que aconteceu?
DET what happened
'What happened?

B. Eu nao terminei o trabalho
1S NEG finish.PST.1S DET work
'I didn't finish the work'


(#nao)
NEG


If the negated information has been introduced into the discourse, then the use of

clause-final negation becomes felicitous. In the following case the information 'finish the

work' has been introduced in to the discourse.


(21) A. Voc6 terminou o trabalho?
2s finish.PST.2S DET work
'Did you finish the work?'


B. Eu nao terminei o trabalho
1S NEG finish.PST.1S DET work
'I didn't finish the work'


(nao)
NEG


1.5.2 Embedded Clauses

Some researchers have stated that T2 and T3 behave differently in embedded

clauses (Schwegler 1991a, Cavalcante 2007). These researchers claim that T2 is

permitted in embedded clauses and T3 is not. (22) gives an example of T2 in an

embedded clause.


(22) A LOcia pediu para os homes no fazerem
DET Lucia ask.PST.3S PREP DET men NEG do.INF.3P
'LOcia asked for the men not to do the work'
*'Lucia didn't ask for the men not to do the work'


o trabalho
DET work


nao.
NEG










Example (23) is an attempt at placing T3 in an embedded clause. The

interpretation of T3 in this case is that the negative marker has scope over the whole

sentence and not just the embedded clause.


(23) A Lucia pediu para os homes fazerem o trabalho nao.
DET Lucia ask.PST.3S PREP DET men do.INF.3P DET work NEG
'Lucia didn't ask for the men to do the work'
*'Lucia asked for the men to not work'


While the interpretation of sentence (23) is of negation of the whole sentence,

chapter six shows that this is deceptive because T3 is syntactically allowed in

embedded clauses. Because previous researchers didn't address the discourse

requirements, the next examples show that if all the discourse requirements necessary

to license T3 are in place, then T3 is allowed. In the case below, the question

introduces a direct activation of the information 'he finished the test', and in this case it

is possible to use T3 in the embedded clause.


(24) Cesar told Ana that he finished the test.

(25) Eu acho que ele terminou o exame nao
1s think that 3s finish.PST.3S DET test NEG
'I think that he didn't finish the test'


(26) shows that if there is CFN and the main clause has T1, then the sentence

final nao is interpreted as part of the main clause. This is another reason why other

researchers have claimed that T3 cannot be in complement clauses or in other

embedded clauses (Schwegler 1991a, Cavalcante 2007). The interpretation of (26) is

(b). The preference for this interpretation might be due to processing, where this









example would is interpreted as having one instance of T2 in the main clause instead of

two instances of negation, on in each of the matrix and embedded clauses. To negate

the embedded clause and the main clause then the final nao must not be present.


(26) A Sienna nao ouviu que a Maria disse uma mentira nao.
DET Sienna NEG heard.3s that DET Maria told.3s DET lie NEG
a. *'Sienna didn't hear that Maria didn't tell a lie'
b. 'Sienna didn't hear that Maria told a lie'


(27) has T1 in the main clause and T2 in the embedded clause.1 This sentence is

marginal. The remedy to this is given in (28), where the sentence final position is not

used, i.e. T1 in both clauses.


(27) ??A Lucia nao pediu para os homes nao fazerem o trabalho nao.
DET Lucia NEG ask.PST.3S PREP DET men NEG do.INF.3P DET work NEG
'LQcia didn't ask for the men not to do the work'

(28) A Lucia nao pediu para os homes nao fazerem o trabalho.
DET Lucia NEG ask.PST.3S PREP DET men NEG do.INF.3P DET work
'LQcia didn't ask for the men not to do the work'

1.5.3 Relative Clauses

Relative clauses place restrictions on the appearance of T2 and T3. In relative

clauses in object position only T1 and T2 are grammatical, and T2 is only grammatical if

the relative clause is sentence final. T1 and T2 negation of a relative clause in object

position is given in (29) for T1, and (30) for T2 (Cavalcante 2007); additionally, (31)

shows a sentence where the negative marker is not sentence final but is followed by a

prepositional phrase 'with the work' and is therefore ungrammatical.



1 Schwegler (1991a)'s data are similar to what I have here, in that when there are three naos, then the
sentence is difficult to understand. In some cases, my consultants reject it all together.









(29) O Joao ajudou o menino
DET Joao helped.3s DET boy
'John helped the boy that doesn't

(30) O Joao ajudou o menino
DET Joao helped.3s DET boy
'Joao helped the boy that doesn't

(31) *O Joao ajudou o menino
DET Joao helped.3s DET boy
'Joao helped the boy that doesn't


que nao usa
that NEG use.3s
wear glasses'

que nao usa
that NEG use.3s
wear glasses'


6culos
glasses


6culos nao
glasses NEG


que nao usa 6culos nao com a tarefa.
that NEG use.3s glasses NEG PREP DET work
wear glasses with the work'


T3 is never allowed in relative clauses. When there is only a sentence final

negative marker, then the scope of negation is over the whole sentence (i.e. it is

interpreted at T1 in the main clause) and not the relative clause as exemplified in (33).


(32) O Joao ajudou o menino que usa 6culos nao
DET Joao helped.3s DET boy that use.3s glasses NEG
*'Joao helped the boy that doesn't wear glasses'
'Joao didn't help the boy that wears glasses'


(33) has the relative clause in an indirect object position negated using T3, and it

is ungrammatical. (34) has the same sentence negated with T2, and again, it is

ungrammatical. Finally, in (35), T1 is grammatical, as is expected from its unmarked

nature. In each case, the relative clause is before the direct object and the sentence is

ungrammatical. The brackets delineate the noun phrase containing the relative clause.


(33) O Joao deu pro [menino que usa 6culos nao] o livro
DET Joao gave.3s PREP.DET boy that use.3s glasses NEG DET book
*'Joao gave the boy that doesn't wear glasses the book'

(34) O Joao deu pro [menino que nao usa 6culos nao] o livro
DET Joao gave.3s PREP.DET boy that NEG use.3s glasses NEG DET book
*'Joao gave the boy that doesn't wear glasses the book'









(35) O Joao deu pro [menino que no usa
DET Joao gave.3s PREP.DET boy that NEG use.3s
'Joao gave the boy that doesn't wear glasses the book'


6culos ] o livro
glasses DET book


When a relative clause in subject position is negated with T2, then the sentence is

ungrammatical, seen in (36). I bracket off the subject to show that the second nao is

within the relative clause and it is not negating the matrix verb ajudar'to help'. As

expected, T3 is not allowed in subject relative clauses either, as evidenced by (37). T1

would be grammatical.


(36) [O menino
DET boy
*'The boy that

(37) [O menino
DET boy
*'The boy that


que nao usa 6culos nao] ajudou
that NEG use.3s glasses NEG helped.3s
doesn't wear glasses helped Joao'


que usa 6culos nao]
that use.3s glasses NEG
doesn't wear glasses helped Joao'


ajudou
helped.3s


BP allows subject doubling where the subject is repeated through the use of a

pronoun. The subject is 'he' and 'the boy' is the doubled subject.


(38) O menino, ele ajudou
DET boy 3s helped.3s
'The boy, he helped John'


o Jo0o
DET Jo0o


In these cases, T2 in the doubled subject is also ungrammatical, seen in (39).


(39) *[O menino que nao usa 6culos nao] ele ajudou
DET boy that NEG use.3s glasses NEG 3s helped.3s
'The boy that doesn't wear glasses, he helped Joao'


o Jo0o
DET Jo0o


o Jo0o
DET Jo0o


o Jo0o
DET Jo0o









1.5.4 Topics

BP has topic-like expressions of various constituents (Martins & Nunes 2006;

Galves 1998). A fronted topic object DP is given in (40).


(40) Este filme, voc6 pode ver
DET movie, 2s can.2s see.INF
'This movie, you should see'


Fronting is possible with T1, T2, and T3 in (41), (42), and (43) respectively.

Cavalcante (2007) claims that topicalization as a result of fronting in T3 sentences is not

allowed, but I have found no evidence for this.


(41) Este filme, voc6 nao pode ver
DET movie, 2s NEG can.2s see.INF
'This movie, you can't see it'

(42) Este filme, voc6 nao pode ver nao
DET movie, 2s NEG can.2s see.INF NEG
'This movie, you can't see it'

(43) Este Filme, voc6 pode ver nao.
DET movie, 2s can.2s see.INF NEG
'This movie, you can't see it'


BP also has verbal fronting. In these cases, the same types of restrictions hold, i.e. all

types are grammatical. (44) gives an affirmative declarative sentence with verbal

fronting, and the next three example show the three types of negation with fronted verb.


(44) Falar, ela fala Ingles
Speak.INF, 3SF speak.3s English
'As for speaking, she speaks English'

(45) Falar, ela nao fala Ingles T1
Speak.INF, 3SF NEG speak. 3s English
'As for speaking, she doesn't speak English'










(46) Falar, ela nao fala Ingles nao
Speak.INF, 3SF NEG speak. 3s English NEG
'As for speaking, she doesn't speak English'


(47) Falar, ela fala Ingles nao T3
Speak.INF, 3SF speak. 3s English NEG
'As for speaking, she doesn't speak English'


Sentence (48) illustrates that the topicalized VP cannot be negated (Bastos-Gee 2009).


(48) *Nao falar (nao), ela nao fala
NEG Speak.INF NEG, 3SF NEG speak. 3s
'As for speaking, she doesn't speak English'


Ingles
English


(nao)
NEG


1.5.5 Wh-Questions

CFN (i.e. T2 and T3) is ungrammatical with wh-questions in which the wh-phrase

is fronted, (49), (50) for T2, and for T3 (51), (52). All of the following examples can be

made grammatical by eliminating the clause-final nao. BP has several strategies for

formulating wh-questions, and the examples below show that neither clefted questions

nor non-clefted questions allow T2/T3 CFN.

T2


(49) *Por que (e que) voce nao quis
Why is that 2s NEG want.PST.2S
'Why don't you want to eat the pizza?'


comer a pizza
eat.INF DET pizza


(50) *O que
What
'What is


(e que) voc6 nao
is that 2s NEG
it that you don't want to eat?'


quer
want.2s


comer nao?
eat.INF NEG


(51) *Por que (e que) voce quis comer
Why is that 2s want.PST.2S eat.INF
'Why don't you want to eat the pizza?'


nao?
NEG


a pizza
DET pizza


nao?
NEG









(52) *O que (e que) voce quer comer nao?
What is that 2s want.2s eat.INF NEG
'What is it that you don't want to eat?'


BP also has wh-in-situ questions (Zocca 2008, Pires & Taylor 2007). (54) gives T1.


(53) Voce comprou o que?
2s bought.PST.2s what
'What did you buy?

(54) Voce nao comprou o que?
2s NEG bought.PST.2s what
'What didn't you buy?'


Examples (55) shows that T2 in wh-in-situ questions is highly degraded. (56)

shows that T3 is ungrammatical.


(55) ??Voc6 nao comprou o que nao?
2s NEG bought. PST.2S what NEG
'Which books didn't you buy?'

(56) *Voc6 comprou o que nao?
2s bought. PST.2S what NEG
'Which books didn't you buy?'

1.5.6 N-Words and Negative Concord

In BP, so-called n-words (as in Laka 1990, Giannakidou 2006) follow similar

patterns found in other Romance languages; however, some differences can be seen

with T2 and T3. Chapter seven treats in detail the idea of Negative Concord (NC). For

now, the term n-words are taken to be a theory-neutral term, meaning words with

apparent negative morphology like nobody in English, nessuno in Italian, niemand in

Dutch, etc.

NC is a situation where negation appears to have multiple surface manifestations

(seen on the negative marker and an n-word), but is only interpreted once (Giannakidou









2006:328). BP, Spanish, Italian, and European Portuguese are considered non-strict

NC languages. These languages do not allow negative markers (non, no, nao) to be c-

commanded by n-words. These languages require n-words to be c-commanded by

negative markers in most cases, for example postverbal subject n-words or object n-

words (Zanuttini 1997; Giannakidou 2000, 2006; Zeijlstra 2004).

First, I present the data with n-words as subjects. (57) and (59) illustrate the

behavior of non-strict negative concord in Spanish and BP with T1. In these examples

the subject n-word c-commands the negative marker and the sentence has a double

negative reading, not a negative concord reading. To eliminate the double negative

reading, one omits the negative markers no and nao as in examples (58) and (60).


(57) Nadie no ayud6 a Juan Diego Spanish
Nobody NEG helped PREP Juan Diego
'Nobody didn't help Juan Diego'
*'Nobody helped Juan Diego'

(58) Nadie ayud6 a Juan Diego
Nobody helped PREP Juan Diego
'Nobody help Juan Diego'

(59) Ninguem nao ajudou o Abilio2 BP
Nobody NEG help.PST.3S DET Abilio
'Nobody didn't help Abilio'
*'Nobody helped Abilio'


2 Martinez (2006) states that the following is allowed in Mineiro Portuguese (State of Minas Gerais)

i. Ningu6m nao veio
Nobody NEG came
'nobody came'
While she points to other studies, Alkmim (2001), this is the only example of the n-word and the preverbal
negative marker not invoking a negative reading. All the preliminary work suggests that this is dialectal
and is not seen in most dialects of BP. If it turns out that this claim is false, then more analysis would be
required.









(60) Ninguem ajudou o
Nobody help.PST.3S DET
'Nobody help Abilio'


Abilio
Abilio


T2 negation and subject n-words behave as expected, with a double negative

reading due to the n-word c-commanding the negative marker.


(61) Ninguem nao ajudou
Nobody NEG help.PST.3S
'Nobody didn't help Abilio'
*'Nobody helped Abilio'


0
DET


Abilio nao
Abilio NEG


The facts are different with T3. While the n-word appears to c-command the

negative marker, (62), a double negative reading does not result, in contrast to T1 and

T2, as seen above.


(62) Ninguem ajudou o Abilio
Nobody help.PST.3S DET Abilio
'Nobody helped Abilio'


nao
NEG


Object n-words give grammatical NC readings with T1 (63) and T2 (64), but not

with T3 (65). Independent of the clause-final negation in (65), the sentence is

ungrammatical.


(63) 0 Abilio nao ajudou
DET Abilio NEG help.PST.3s
'Ablilio didn't help anyone'

(64) 0 Abilio nao ajudou
DET Abilio NEG help.PST.3s
'Ablilio didn't help anyone'


(65) *O Abilio
DET Abilio
'Ablilio didn't help


ajudou
help.PST.3s
anyone'


ninguem
nobody


ninguem
nobody


ninguem
nobody


nao
NEG


nao
NEG










1.5.7 Imperatives

BP allows all three types of negation in negative imperatives. BP has two types of

imperatives: "true negative imperatives" and "suppletive imperatives". True negative

imperatives are formed from positive imperatives that are negated using the typical

negative marker. Suppletive imperatives, found in some Romance languages like

Spanish, are commands formed using non-imperative verbal morphology, such as the

subjunctive. (66) illustrates a Spanish imperative. This clause cannot be negated

directly, (67), rather a subjunctive verb form is used to negate an imperative, (68).


(66) Lee eso! Spanish
Read.2s.iMP that
'Read that!'

(67) *No lee eso!
NEG read.2s.iMP that
'Read that!'

(68) No leas eso!
NEG read.2s.SBJ that
'Don't read that'


Like Spanish and European Portuguese (EP), suppletive negative imperatives are

grammatical in BP. (71) shows that BP also allows true negative imperatives, although

this is not the case in EP.3 EP does not allow negative imperatives without use of the

subjunctive.

(69) L6 isso! EP and BP
Read.2s.iMP that
'Read that!'

3 This type of example would not be considered an imperative by Zanuttini (1994) and Postma & van der
Wurff (2007). They require that imperatives have different morphology from the other verb types, i.e
declarative or subjunctive. Since BP's imperatives are identical to declaratives, these researchers would
not consider them to be true imperatives.










(70) Nao leias isso EP and BP
NEG Read.2s.SBJ that
'Don't read that!'

(71) Nao l~ isso *EP and BP
NEG read.2s.IMP that
'Don't read that!'


(72) and (73) show that T2 and T3 are also allowed in imperatives (Postma & van

der Wurff 2007, Schwegler 1991 a).


(72) Nao l~ isso nao BP
NEG Read.2s.IMP that NEG
'Don't read that!'

(73) L6 isso nao BP
Read.2s.IMP that NEG
'Don't read that!'

1.5.8 NPI Idioms

NPI idioms are idioms that are felicitous only in the presence of an NPI licensor.

In the case of BP the licensor is generally the preverbal negative marker. The three

types of negation show a difference in their ability to license NPI idioms (Biberauer &

Cyrino 2009). The following examples show that T2 allows the idiomatic meaning and

T3 does not. The expression nao tem um chinelo pra botar um prego literally means 'to

not have a flip-flop to nail a nail with', seen in (74). As an idiomatic expression it means

'to not have anything'. This meaning is retained in the T2 sentence (75), but lost in (76)

with T3.


(74) Ele nao tem um chinelo pra botar um prego T1
3s NEG have DET flip-flop PREP.DET throw.INF DET nail
'He doesn't have anything'









(75) Ele no tem um chinelo pra botar
3s NEG have DET flip-flop PREP.DET throw.INF
'He doesn't have anything'

(76) Ele tem um chinelo pra botar
3s have DET flip-flop PREP.DET throw.INF
'He doesn't have a flip-flop to nail a nail with'
*'He doesn't have anything'


um prego nao
DET nail NEG


um prego nao
DET nail NEG


1.5.9 Summary of the Data

The following table shows the situations where each type is used. Note that /

means grammatical and X means ungrammatical. For relative clauses, T2 is licit in

object relative clauses, but only if they are sentence final, thus the *

Table 1-1. Syntactic Restrictions for BP CFN
Environment T1 T2


Discourse Requirements

Allowed in Relative Clauses in Subject Position

Allowed in Relative Clauses in Object Position

Allowed in Fronted Wh-questions

Allowed in in-situ Wh-questions

Allowed with Subject N-words

Allowed with Object N-words

Licenses NPI Idiomatic Expressions

Allowed in Embedded Clauses

Allowed in Imperatives

Compatible with Topicalization


x


/ X X
x x


yX/ X
x x

x x

x x

4 x

Sx









The data raise a number of questions that are the focus of the chapters that follow.

I argue that the answers to these questions contribute to an explanation of the

distributional differences and similarities presented above.


i. What are the semantic and pragmatic differences between the three types
of negation?

ii. What is the analysis of the interaction of n-words and sentential negation
in BP?

iii. What is structural analysis of the three types of negation?

1.6 Structure of the Study

The chapters of this work are organized as follows. Chapter two is a brief

overview of the theoretical framework to which this study adheres. This chapter

highlights the fundamental principles of minimalism and minimalist syntax. The main

goal of this chapter is to set up the theoretical base from which I can properly explain

sentential and clause-final negation in BP.

Chapter three provides an overview of the clause structure of BP. In this chapter, I

discuss the syntactic positions of the subject, verb, topics, wh-words, and some

adverbs. The conclusions of this chapter have several repercussions for the overall

analysis of clause-final negation. First, I argue for a split IP in BP consisting of AgrP

dominating TP. The subject is in spec,AgrP and the verb is in T (Costa & Galves 2002).

I also argue for a split CP in BP as well. Fronted DP topics are shown to be base-

generated in the specifier of a Topic Phrase and fronted wh-phrases are analyzed as

occupying the specifier of a Focus Phrase, which occurs below TopicP. Wh-in-situ in

BP is analyzed as not involving movement at all. Additionally, wh-in-situ has a









discourse-based restriction on it and it only occurs if elements of the question are part of

the Common Ground (Pires & Taylor 2007).

Chapter four concerns itself with the preverbal negative marker. I present the

main theoretical approaches to understanding the syntax of the preverbal marker in

Romance languages. The chapter extends the clause structure from chapter three and

proposes a position for the preverbal negative marker as the head of NegP between

AgrP and TP.

Chapter five discusses one of the main topics of this dissertation, the

semantic/pragmatic status of clause-final negation. Here, I discuss previous attempts

at understanding CFN in terms of semantics and pragmatics. I reject several previous

analyses of CFN as insufficient to explain the data (Schwegler 1991a, Da Cunha 2007).

Instead, I follow Schwenter's interpretation of Prince's (1992) Discourse-Information

Status theory. Roughly, what is negated with CFN must have been introduced into the

discourse for it to be felicitous.

Chapter six discusses the second main topic of this dissertation, the syntactic

status of clause-final negation. This chapter discusses past attempts to understand the

syntax of CFN. I argue that previous accounts do not fully capture the requirements

and restrictions on CFN. The final part of chapter six introduces my proposal for CFN in

BP. I propose that CFN is phrasal movement of the whole AgrP to a specifier position

of a lower TopicP phrase in the left periphery. The head of this lower topic phrase is a

nao for negative sentence and a sim for some positive sentences.

Chapter seven tests the syntactic proposals made in chapters five and six using

Negative Concord and n-words. In this chapter, I discuss the status of n-words in BP.









From there, I combine the analysis from chapter six with the analysis of Negative

Concord to argue for the existence of three negative formatives in BP. The three

markers are: a preverbal nao with an interpretable negative feature, a clause-final ndo

with an uninterpretable negative feature, and a clause-final nao with an interpretable

negative feature feature. The first marker appears in T1 and T2, the second appears in

T2 only, and the third appears exclusively in T3.

Chapter eight is a summary and a conclusion. It summarizes the findings of the

dissertation and highlights the major theoretical implications.









CHAPTER 2
THEORETICAL FOUNDATIONS

2.1 Introduction

This chapter's main goal is to present and explain the theoretical assumptions

used in this dissertation. A main purpose of this chapter is to describe the linguistic

tools that will be used to understand negation in Brazilian Portuguese, both preverbal

and clause-final. Two areas of linguistic theory are covered: certain aspects of

minimalist syntax and the left periphery. Section 2.2 covers the Minimalist Program and

some aspects of minimalism that will be relevant in the analyses in later chapters.

Section 2.3 discusses the left periphery, making specific use of the proposals from Rizzi

(1997).

2.2 Minimalism

This work assumes a rather conservative version of minimalist syntax from the

1990's and early 2000's (Adger 2003, Radford 2004), which includes X-bar theory

(Carnie 2002) and certain mechanisms that are used to drive movement in the syntax

(Chomsky 1995, 2000, 2001, Hornstein 2001, Hornstein, Nunes & Grohmann 2005,

BoSkovi6 & Lasnik 2007). Section 2.2.1 presents the main theoretical ideas and

motivations behind a minimalist syntax. Section 2.2.2 presents my assumptions about

the Probe-Goal system and Feature Checking.

2.2.1 Fundamental Principles of Minimalism

The minimalist program has as a research goal "simplicity, elegance, parsimony,

and naturalness" (Hornstein 2001:4). These somewhat abstract terms are concretely

defined by Hornstein, Nunes & Grohmann (2005:ch 1) (Haddad 2007:24):









i. Naturalness implies that only notions that correspond to self-evident facts about
language should be preserved.
ii. Simplicity follows from naturalness. If only natural notions are preserved and all
other theory internal notions are removed, the grammar becomes simpler.
Further, given two theories A and B that are equal in every way except that A has
fewer rules than B, A is considered superior.
iii. Economy is pertinent to derivations and derivational rules. Everything else being
equal, a derivational step that requires the least effort (e.g., fewer steps) and that
happens only when necessary is optimal.


Coupled with these goals and definitions, the minimalist program assumes some

important ideas about language. Hornstein (2001) presents these ideas as follows:

i. sentences are the basic linguistic units
ii. sentences are pairings of sounds and meaning
iii. there is no upper bound to the number of sentences in any given [Natural
Language]
iv. sentences show displacement properties in the sense that expressions
pronounced in one position are interpreted in another
v. sentences are composed of words organized into larger units with
hierarchical structure, i.e. phrases.


Chomsky and others propose that the Language Faculty (FL) of the mind

generates sentences or linguistic expressions. FL has interfaces with the articulatory-

perceptual (AP) system and the Conceptual-Intentional (CI) system. Phonological Form

(PF) is the representation at the interface between the FL and the Articulatory-

perceptual system and Logical Form (LF) is the representation at the interface between

the Language Faculty and the Conceptual-Intentional system. These two

representations yield the pronounced form and meaning of a sentence, and these are









the only two levels of representation in the theory. Many visual models attempt to show

the relationships just mentioned, and the model that is assumed generally by those

working in a minimalist framework is as follows (Chomsky 1995, Hornstein, Nunes, &

Grohmann 2005):

i. Lexicon

Numeration

Spell-Out--Phonological Form (PF)

Logical Form (LF)


Chomsky explains that lexical items (LIs) bundle three types of features:

phonological, semantic, and formal and the lexicon is the storage of these lexical items.

Phonological features include the information needed to be interpreted at PF, i.e.

roughly what would be considered the sounds of the International Phonetic Alphabet.

Semantic features are features, such as [+animate], which are interpreted at LF.

Finally, formal features are things to which the computation system is sensitive, and

they can be interpretable or uninterpretable (see section 2.2.2 for more details). It is

important to note that the lexicon of a language may have lexical entries that are

phonologically null. These null elements are interpreted at LF even though they have

no pronounced form.

The numeration is a selection of elements from the lexicon selected to take part in

the derivation. Structure building operations apply to elements in the numeration, after

which the sentence is Spelled-out, i.e. sent to the interfaces. Spell-out is where the

phonological features are mapped onto PF and the other features continue towards LF.

Syntactic operations that occur before Spell-out are considered overt operations









because they are visible through the phonology, and operations after Spell-out are

covert because they are not visible to the phonology. At Logical Form the system

determines whether a sentence is well-formed or not, including whether all the

uninterpretable formal features have been checked.

I assume four main computational or syntactic operations: Select, Merge, Agree,

and Copy. Select essentially picks something from the lexicon. Merge takes two

syntactic objects and combines them into one. Merge is the operation that builds

syntactic trees. The operation Agree "establishes a relation (agreement, Case-checking)

between an LI and a feature F in some restricted search space (its domain)" (Chomsky

2000:100). Finally, Copy duplicates a linguistic object and makes it available for Merge.

Movement in the syntax is a result of Copy plus Merge. I will leave aside any technical

discussion of Select, Merge, and Copy and return to Agree below.

Another main idea relevant to this current work is that a clause is divided into

three domains: the left periphery, or CP, the inflectional domain, or IP, and the verb

phrase domain, or vP (Chomsky 2000, Grohmann 2003). These categories have been

shown to be made up of subparts. The representative works for each are: Rizzi (1997)

for the CP, Pollock (1989) for IP, and Larson (1988) for VP. I discuss the left periphery

in BP in section 2.3 of this chapter.

2.2.2 Probe-Goal System and Feature Agreement

An important aspect of syntactic representations is formal features and agreement

among these features. Features on lexical items can be of two kinds: interpretable,

represented by [iF], or uninterpretable, represented as [uF], where F is some

morphosyntactic feature such as person, number, negation, topic, focus, question, etc.

An interpretable feature is one that has an interpretation at LF and an uninterpretable









feature is one that does not. For example, phi-features (person, number) are

interpretable on DPs but uninterpretable on agreeing heads such as adjectives and

verbs. An uninterpretable feature is illegitimate at LF and violates what is called the

Principle of Full Interpretation, which states that all syntactic objects should be fully

interpretable at the interfaces (PF and LF) (Chomsky 1995). To avoid a violation of Full

Interpretation, uninterpretable features must be eliminated in the syntax before they

reach LF. The mechanism that removes uninterpretable features is called Feature

Agreement. Informally, Feature Agreement, or Agree, allows an uninterpretable feature

to check itself against an interpretable counterpart and thus become invisible. Agree is

accomplished through a system of probes and goals. A probe is a feature that is

uninterpretable [uF]. The probe searches its syntactic domain to find a corresponding

interpretable feature [iF], which is called the goal. If the probe finds a goal and the two

features match, the uninterpretable [uF] is checked and deleted.

The syntactic domain in which a probe searches for a goal is its c-command

domain; a probe looks down to find a goal. In addition, I adopt a proposal put forth in

Rezac (2003) which argues that although probes normally look within their c-command

domain to find a goal, if no goal is found, then the probe can wait for additional material

to be merged into the derivation. This means that the field of search ultimately available

to a probe includes elements higher in the structure, outside it's c-command domain.

This upward search possibility for probes will be employed to understand Negative

Concord (see section 7.2.4),following work by Zeijlstra (2004).

Since Agree can happen at a distance and doesn't require movement to result in

feature checking, something additional must cause elements to move. For the









purposes here, I claim that what forces movement is that a feature on a probe can be

strong, which requires that it be checked in a more local configuration than with Agree

alone. A strong uninterpretable feature triggers movement of something lower in the

structure to the specifier position of the probe carrying the strong feature. A strong

feature is represented as [F*] and must be checked in a specifier-head configuration.

In this section I have briefly discussed the major framework and linguistic tools that

this dissertation employs. Additional syntactic machinery will be presented as

necessary.

2.3 CP: The Left Periphery

This section presents the theoretical assumptions about the left periphery that

have risen due, in part, to Rizzi (1997). Section 2.3.1 introduces Rizzi's theory. Then, I

take the idea of an expanded CP and apply it to BP in section 2.3.2. By doing this, I

show that Rizzi's proposal is compatible with BP. In fact, as later chapters show, an

expanded CP is fundamental to a proper understanding of BP clause structure and

CFN.

2.3.1 Rizzi's Split CP

CP is taken to be the region of the clause where certain grammatical notions such

as topic, focus, clause type, etc. are expressed (Chomsky 1995). Chomsky, referring to

Cheng (1991, 1997), states that C serves as a "force indicator", where force is

understood to mean clause type, such as declarative, interrogative, imperative, etc.

(Chomsky 1995:69). Rizzi's (1997) work develops this idea by showing that CP actually

consists of a number of functional projections, each of which independently hosts

elements related to topic, focus, clause type, etc.









Before Rizzi's work, for example in Lectures on Government and Binding

(Chomsky 1981), the left periphery was a single CP projection that housed a host of

different elements, including wh-phrases, complementizers of varying types, topics, and

focused elements. In the early 1990s Lasnik & Saito (1992) and McCloskey (1991)

proposed that topics and sentential adverbs adjoin to the main clause. However, in

light of cross-linguistic evidence, this adjunction that some proposed for topics and

sentential adverbs was not very constrained, and as such, behavior of this type seemed

fit to be eliminated from the grammar. Rizzi (1997) and others after him (Beninca 2001,

Beninca & Polleto 2004, Cinque 1999, 2002, Rizzi 2004, Polleto 2000) proposed that

CP be split into several X-bar compatible projections, such as TopicP, FocusP, and

ForceP, among others. This split created dedicated head and specifier positions for

elements commonly found in the CP region.

The Force projection is what determines the type of clause. The Topic projection

hosts topics of the sentence and what follows is assumed to be the comment. The

Focus projection hosts focused elements and what follows is assumed to be the

presupposition. Finally, the Finite projection determines the (non-)finiteness of the TP.

Rizzi's structure looks like (1) (see Rizzi 2001:289 for the inclusion of an Int(errogative)

Phrase). The indicates that TopicPs can iterate (but see Beninca & Polleto 2004 for

evidence against this).


(1) ForceP TopicP* IntP TopicP* FocusP TopicP* FinP


Rizzi assumes that not all projections in the split CP are necessarily present in all

clauses. "It is reasonable to assume that the topic-focus system is present in a structure









only 'if needed' (Rizzi 1997: 288)". Rizzi relies on Agree to fill these positions. If the

topic position is filled, for example, then a strong uninterpretable topic feature [uTop*]

must exist on the head of TopicP to drive movement to this position, as discussed

above.

2.3.2 The Split CP and BP

This section provides evidence for Rizzi's various projections in BP. In Rizzi's

approach, finite complementizers are inserted into Force. The Italian sentence (2) and

the BP sentence (4) have the complementizers che and que, respectively, in Force,

which causes the embedded clause to be finite. Rizzi contrasts these complementizers

with prepositional complementizers, such as Italian di, which introduce non-finite

clauses. While BP does not have a construction exactly like the Italian example (3), it

does use a preposition-like complementizer to introduce non-finite clauses as in (5).

Rizzi claims that di in Italian is Fin; however, I suggest below that this is not the case for

BP para.


(2) Credo che loro apprezzerebbero molto il tuo libro.
Believe.1s COMP 3P appreciate.FUT.3P much DET 2s book
'I believe that they will appreciate your book very much'

(3) Credo di loro apprezzare molto il tuo libro.
Believe.1s PREP 3P appreciate.INF much DET 2s book
'I believe that they appreciate your book very much'

(4) A Sienna pediu que eles trabalhassem.
DET Sienna ask.PST.3S COMP 3p work.SBJ.PST.3P
'Sienna asked that they work'

(5) A Sienna pediu para eles trabalharem.
DET Sienna ask.PST.3S PREP 3P work.INF.3P
'Sienna asked that they work'









Rizzi's claim is that the Italian complementizers che and di cannot be the head of

the same projection because of how they interact with topics and focused elements. In

the case of finite complementizer che, topics must necessarily follow che. With the non-

finite complementizer di, topics must necessarily precede di. The topics are underlined

and the complementizer is bolded in the following examples. Notice that in (6) che

precedes the topic; whereas, in (7) and (8) the di must follow the topic.


(6) Credo
Believe. 1s
'I believe that

(7) *Credo
Believe. 1s
'I believe that

(8) Credo
Believe. 1s
'I believe that


che il tuo libro
COMP DET 2s book
your book, they will

di il tuo libro
PREP DET 2s book
your book, they will

il tuo libro di
DET 2s book PREP
your book, they will


loro lo apprezzerebbero molto.
3P CL.3S appreciate.FUT.3P much
appreciate it a lot.'

apprezzarlo molto.
appreciate. INF-CL much
appreciate it a lot.

apprezzarlo molto
appreciate. INF-CL much
appreciate it a lot.'


In the case of the BP finite complementizer que, BP behaves like Italian in that

topics cannot precede que. Assuming that que is the head of ForceP, this shows that

the order of the functional heads is Force Topic, as Rizzi proposes.


(9) Acho que os meninos o Joao viu
Believe.1s COMP DET boys DET John saw.PST.3S
'I believe that the boys John saw them in the market'

(10) *Acho os meninos que o Joao viu
Believe.1s DET boys COMP DET John saw.PST.3S


no mercado.
PREP.DET market


no mercado.
PREP.DET market'


Based on examples such as (7) and (8), Rizzi argues for a head called FinP,

which is where he places the prepositional complementizer di in Italian. BP does not

seem to have any overt evidence of this. The interaction of topics with BP para is the









same as with que in that topics cannot precede para. The examples below are

ungrammatical on the interpretation where the topic preceding the complementizer is

associated with the embedded clause.


(11) *A Sienna pediu no mercado que eles trabalhassem.
DET Sienna ask.PST.3S PREP.DET market that 3p work.sUBJ.PST.3P
'Sienna asked that in the market they work' (They work in the market)

(12) *A Sienna pediu no mercado para eles trabalharem.
DET Sienna ask.PST.3S PREP.DET market PREP 3P work.INF.3P
'Sienna asked that in the market they work'(They work in the market)


From this, I assume that both para and que in BP realize Force and Fin is not overtly

realized in BP.

Rizzi (1997, 2001) also shows that topics can precede fronted wh-words. Again,

the same can be said of BP. In the Italian example in (13) a Gianni 'to Gianni' is fronted

to the left of the wh-word. The same happens in the BP example in (14), with pra Lu 'to

Lu' to the left of the wh-phrase.


(13) A Gianni, che cosa gli hai ditto?
PREP Gianni, what thing DAT.3S have.2s said
'To Gianni, what did you say'

(14) Pra LQ, o que que voc6 deu?
PREP.DET Lu, what that 2s gave.PST.2S
'To LQ, what did you give?'


Rizzi assumes that the landing site of wh-phrases is the specifier of a focus

phrase, FocusP, below TopicP and I will follow his conclusion.4 Evidence for this claim

comes from the incompatibility of wh-phrases with a fronted focused phrase:5


4 Section 3.5 will explain that the que in example (14) is not the same as que the complementizer. Rather
this is a Q operator.










(15) *PRO ZE o que que voce deu (nao pro Lionel)
PREP.DET Ze what that 2s give.PST.2S (NEG PREP.DET Lionel)
'TO ZE what did you give (not to Lionel)'

(16) o que que PRO ZE voce deu (nao pro Lionel)
what that PREP.DET Ze 2s give.PST.2S (NEG PREP.DET Lionel)
'TO ZE what did you give (not to Lionel)'


(15) has the focused PP before the wh-word, and (16) has it following the wh-

word. In both cases, the result ungrammatical. Rizzi takes this fact to mean that the

target for wh-fronting is also FocusP, thus the mutually exclusive nature of these two

items. In addition, only one focused phrase is allowed, confirming that FocusP cannot

iterate:


(17) *PRO CESAR O LIVRO eu vou dar (nao pra Sienna o caderno)
PREP.DET Cesar DET book 1s go.1s give.INF (NEG PREP.DET Sienna DET folder)
'The book, to Cesar, tomorrow, I will give (not to Sienna the folder)'


In contrast, a clause may contain multiple topics, indicating that TopicP does

iterate. (18) has two topics, 'the book' and 'to Cesar'.


(18) 0 livro pro Cesar amanh5 eu vou dar
DET book PREP.DET Cesar tomorrow 1s go.1s give.INF
'The book, to Cesar, tomorrow, I will give'


These data support the claim that fronted topics and focus elements are located in

distinct positions in BP, much like in Italian. Ordering these XPs, we observe ForceP -

TopP* WH/FocusP. Chapter six will discuss a second topic position below FocusP.

The fact that there are two topic positions is predicted in Rizzi's work, and it is

5 Ambar (1999) explains how focus works in European Portuguese, and the results are similar for BP.
The meaning of focus for our purpose here is 'new information', as opposed to 'old information'.









something that will aid in explaining CFN. The topics presented in this chapter are

related to topics discussed further in chapter three. There, I give evidence to show that

the higher topics are base-generated. Chapter six explains that the lower topics are

cases of movement of AgrP to the specifier of the lower TopP. To distinguish the two

topic positions, I will label them Topic1 and Topic2. This gives following the structure for

BP:


(19) ForceP Topicl P* WH/FocusP Topic2P


2.4 Conclusion

This chapter's goal was to present the theoretical base for the discussion of

negation in Brazilian Portuguese. I have presented some assumptions regarding Agree

that will be useful in what follows. The second half of this chapter dedicated itself to

Rizzi's expanded CP (1997). Section 2.3.2 showed that Brazilian Portuguese, much

like its cousins in the Romance family, is compatible with Rizzi's theory. This is

important because the idea of an expanded CP is central to the proposals that this

dissertation makes.









CHAPTER 3
THE CLAUSE STRUCTURE OF BRAZILIAN PORTUGUESE

3.1 Introduction

The main goal of this chapter is to examine some elements of the clause structure

of Brazilian Portuguese (BP). A proper understanding of Portuguese clause structure is

imperative for a complete analysis of the negative marker and to understand the

syntactic proposals in later chapters. I discuss the syntactic positions of the subject,

verb, adverbs, topics, and wh-words, and the structures I present are assumed

throughout this dissertation. The structures presented build upon the existing work

related to Portuguese clause structure, and are aligned to the theories discussed in

chapter two (Galves 1989, Mioto 1992, Kato 1999, Costa & Galves 2002, Silva 2004,

Rodrigues 2004, Kato 2009, Nunes 2009).

This chapter is organized as follows. Section 3.2 deals with the subject and its

positions. In section 3.2.1, I consider the option debated in the literature regarding the

preverbal subject position as either an A- or A'- position and argue that it is an A-

position. Section 3.2.2 uses adverbs and floating quantifiers to further investigate the A-

position. Section 3.3 investigates Portuguese verb movement and the position of the

verb. Sections 3.4 and 3.5 discuss the syntax of topics and wh-questions, respectively.

The concluding section of this chapter summarizes the results. As a way of anticipating

the conclusions that come out during this chapter, the clause-structure that I propose for

BP is represented by the basic schematic in (i).

i. [ForceP Complementizer [Topl P topic-phrase [ FocusP wh-phrase [Top2P6
topic-phrase [AgrP subject [NegP nao [TP Vi [vP subject [VP Vi ... ]]]]]]]]

6 This lower topic position will be discussed in chapter six. The higher topic position is the main focus of
the topics discussed in this chapter.











Wh-phrases and topics occupy A' positions in the left periphery of the clause. The

inflectional layer below these projections consists of at least two projections, AgrP and

TP. The preverbal subject occupies the specifier of the higher projection, spec,AgrP,

which is an A-position. The verb raises from V to the head of the lower projection, T, but

no higher (i.e. not to Agr). The inflectional layer dominates vP and VP, where the

external argument, the verb, and verbal complements originate.

3.2 Subjects in BP

This section addresses the theoretical debate surrounding the preverbal subject

position in Romance generally and BP specifically. For other Romance languages,

recent works suggest that the preverbal subject is an A- position or an A' topic position.

From this perspective, I analyze the preverbal subject position in BP. I reject the A'

position analysis for BP, however, and conclude that the BP preverbal subject position

is a canonical A-position.

3.2.1 The Debate over the Romance Preverbal Subject Position

Many suggest that BP is transforming from a null subject language to a non-null

subject language (see all the articles in Kato & Negrao 2001 and Rodrigues 2004). As

such, BP finds itself in the middle of an interesting debate among linguists as to the

position of preverbal subjects in null subject languages (NSL) (Belletti 1990 for Italian,

Alexiadou & Anagnostopoulou 1998 for Greek, Barbosa 1995, 2004 for European

Portuguese, Ordo6ez 1997, Ordo6ez & Treviio 1999 for Spanish, Cardinaletti 2004 for

an extended discussion on the various subject positions that possibly exist in natural









language7). The canonical word order for Romance Languages in general is SVO, and

the debate centers around whether the preverbal subject position is an A- or and A'-

position. This debate is useful in the case of BP because it helps inform the possible

subject positions. I discuss the two main analyses below, first A'-position and then A-

position.

Researchers have argued that the preverbal subject occupies an A' topic position

in the left periphery of the clause, outside of the inflectional domain (Zubizarreta 1998,

Ord6iez 1997, Alexiadou & Anagnostopoulou 1998, and others). Alexiadou and

Anagnostopoulou (1998) give a theoretical argument in favor of the preverbal subject

being in an A' position. They argue that in NSLs like Greek and Spanish, the EPP is

satisfied by movement of the verb to T. That is, verbal agreement morphology in such

languages "includes a nominal element" which satisfies the EPP (Alexiadou &

Anagnostopoulou 1998:516). DP movement to spec, IP is thus unnecessary for EPP

satisfaction. Further, they propose that subjects can check their Case features in their

base position inside vP. Thus, subjects also do not need to move to spec,TP for Case

reasons. Given that there is no motivation for A-movement of the subject to spec,TP, it

must be the case that the preverbal subject is not in an A-position. Rather, they

conclude, it is in an A' topic position (see also Ord6iez 1997, Ord6iez & Trevino 1999).

Movement to this A' position is motivated by discourse considerations, perhaps a topic

feature on the subject (Costa 2004:13).


7 Cardinaletti (2004) discusses several subject positions. She bases this on data from comparing the
Romance languages to things like double subject constructions in Icelandic. While it might be true that
several subject positions are reserved for different types of subjects, as far as I can tell these are all in the
same spot in BP, which is above TP. Essentially a subject in BP could move to spec of TP and move
again to some other subject position higher up, which is essentially what AgrP does. I will not pursue this
line of thought in this work, rather see the facts presented in Kato (2000) and Rodrigues (2004).









Ord6oez & Treviio (1999) provide an empirical argument for this conclusion based

on ellipsis data in Spanish. In each of the following examples, the second conjunct

contains a stranded constituent and the polarity item tambi6n, 'also'. According to the

authors, in (1) the stranded phrase is the subject, in (2) it is an object, and in (3) it is an

indirect object.


(1) El le dio unos libros a Pia y Pepe tambien [le dio unos libros a Pia]
3s CL-gave.3s some books PREP Pia and Pepe also [cL-gave.3s some books PREP Pia]
'He gave some books to Pia and Pepe also [gave some books to Pia]'

(2) Unos libros le dio a Pia y unos cuadros tambien [le dio a Pia]
Some books CL-gave.3s PREP Pia and some paintings too [cL-gave.3s PREP Pia]
'He gave some books to Pia and some paintings also [he gave to Pia]'

(3) A Pia le dio unos libros y a Sara tambien [le dio unos libros]
PREP Pia CL-gave.3s some books and PREP Sara too [cL-gave.3s some books]
'To Pia, he gave some books and to Sara also [he gave some books]'


Ord6oez & Treviio 1999 argue that in (2) and (3) a phrase has been fronted

followed by ellipsis of the constituent following that phrase. More specifically, they

assume that the (in)direct object has been topicalized to spec,TopP and the elided

constituent is TP (following Lobeck 1995), the complement to Top:

(4) ... [TopP unos cuadrosi [Top' Top [T-P le dio ti a Pia]]]

In order to give the same analysis to (1), they claim that the subject must also be

in an A' position, as it appears to occur in the same position as the other verbal

elements in (2) and (3). If the subject were lower, in IP, it would be included in the

ellipsis site. Their claim is also crucially dependent on an analysis of null-subject

languages, where the subject need not necessarily move to spec, IP. As such, if there

is anything in a preverbal position, they claim that element has been fronted, even the

subject.









Alexiadou and Anagnostopoulou (1998:503) provide another argument from

Spanish, based on the complementary distribution of preverbal subjects with some

preverbal adverbs. (5) shows that when the adverb temprano and the subject Julia are

both to the left of the verb the sentence is ungrammatical. When just one precedes the

verb however, the result is grammatical, (6).


(5) *Temprano Julia salia de casa
early Julia leave.IMP.3s PREP house

(6) Temprano salia Julia de casa
Early leave.IMP.3s Julia PREP house
'Julia used to leave/was leaving early from the house'


Ordo6ez (1997) shows that the same type of asymmetry also exists with negative

quantifiers and subjects. (7) demonstrates that Spanish allows negative quantifiers in a

preverbal subject position. They can also be fronted from object position, as in (8),

where the object nada has been fronted. If both a negative quantifier and the subject

are to the left of the verb, however, the sentence is ungrammatical, (9)


(7) Nadie le debe la renta a Maria.
nobody CL-owes.3S DET rent PREP Maria
'Nobody owes Maria the rent'

(8) Nada le debe Juan a sus amigos
nothing CL-owes.3s Juan PREP his friends
'Juan owes nothing to his friends'

(9) *Nada Juan le debe a sus amigos
nothing Juan CL-owes.3S PREP 3s friends
'Juan owes nothing to his friends'


Such data can be accounted for if there is one A'-position before the verb and it

can be occupied by a fronted element (temporal adverb or negative quantifier) or by the









subject but not both. If the subject had its own A-position, such complementary

distribution would be unexpected.

Following Uribe-Echevarria (1992)'s original observations for Spanish, Ord6oez

(1997) also argues that the subject is an A'- position based on scope. According to him,

if the preverbal subject were in an A' position, then its scope would freeze (c.f. Barss

1986). By this he claims that when moved to an A'-position, a quantificational element

is no longer in a position where it scopally interacts with another quantificational

element in the sentence. Thus, in example (10) where the subject is postverbal, scopal

ambiguity arises due to the possibility of the subject raising at LF above aqui6n. In this

case, the universal quantifier has both wide and narrow scope. The wide scope

meaning is that each senator loves a different person. In the narrow scope meaning, all

senators love the same person.


(10) Aquien dices que amaba cada senador?
PREP.whom say.2s that loved.3s each senator
'Who did you say each senator loved?'

Whom > each
Each > whom

(11) no longer has a scopal ambiguity because the subject has moved to the left of the

verb. In this case the only reading is one where cada senador loves the same person,

i.e. narrow scope reading of the universal quantifier, whom > each. Ord6oez claims that

this is more evidence in favor of the preverbal subject being in an A'-position.

(11) Aquien dices que cada senador amaba?
PREP.whom say.2s that each senator loved.3s
'Who did you say each senator loved?'

Whom > Each
*Each > Whom











Alexiadou & Anagnostopoulou (1998) continue Ordo6ez's line of thinking and

claim that multiple quantifiers also are evidence for an A'-position. Under their analysis,

when a quantifier moves to an A-position its position allows it to interact with other

quantificational elements in the sentence and as such creates scopal ambiguities (e.g.

van Riemsdijk & Williams 1986, May 1985). The relationship that allows ambiguity does

not exist when the quantifier is realized in an A'-position. If the preverbal subject in

Greek is an A'-position, then this explains why 'some' can only have wide scope in (12),

but not in (13) where scopal ambiguity remains (Alexiadou & Anagnostopoulou 1998).


(12) Kapios fititis stihiothetise kathe arthro
Some student filed every article
'Some student filed every article' some>every
'*For every article, there is some student that filed it' *every>some


(13) Stihiothetise kapios fititis kathe arthro
filed some student every article
'Some student filed every article' some>every
'For every article, there is some student that filed it' every>some


Based on these pieces of evidence and others, Ordo6ez & Trevino, Alexiadou &

Anagnostopoulou, and others conclude that the subject in Spanish is in the left

periphery using Rizzi's (1997) terminology, and thus the subject position is an A'-

position. This type of argument is the reason that many believe that a preverbal

subject is in an A'-position in the CP domain.

Although there are many arguments in favor of an A' subject position, they are not

universally accepted (Costa 2004, Holmberg 2005, Roberts 2004). Rizzi (1982) first

proposed a traditional analysis in which he argues that subjects in Italian are in an A-









position, and often spec, IP is considered an A-position. More recently, Goodall (2001)

for Spanish and Costa (2004) for European Portuguese have defended this claim,

arguing against the evidence presented above. One piece of evidence for the A-

position analysis comes from "neutral" word order. Adragao & Costa (2004:1) claim that

in EP the neutral word order is SV(O) (see Cardinaletti 1997 for similar facts in Italian).

Out-of-the-blue and all-focus questions demonstrate this, where the entire answer is

new information. SV(O) word order is obligatory in such situations. Any word order but

SV(O) is infelicitous:


(14) O que e que aconteceu?
'What happened?"

(15) 0 Pedro partiu o brago
DET Pedro broke.3s DET arm
'Pedro broke his arm'

(16) #Partiu o Pedro o brago
Broke.3s DET Pedro DET arm
'Broke Pedro his arm'

(17) #0 brago, o Pedro partiu-o
DET arm DET Pedro broke.3s-cL
'His arm, Pedro broke it'


The infelicity of (16) and (17) suggests that these are not neutral word orders and

that some constituent is inappropriately marked as new or old information here. SVO

order in (15), in contrast, is acceptable because no constituent has been singled out

with special discourse status. If this is correct, then the preverbal subject position

cannot be a discourse-oriented A'-position. Instead, it is an ordinary A-position which

generally carries no discourse restrictions.









Following Rizzi (1997), Goodall (2001) argues that bare quantifiers are not able to

be topics, as in (18). In this example, the fronted PP 'to nobody' is clearly in an A'-

position and is ungrammatical.


(18) *A nadie, Juan lo ha visto
PREP nobody, Juan CL has.3s seen
'No one, Juan has seen'


However, bare quantifiers can be subjects. If the subject were left-dislocated, then

sentence (19) would be ungrammatical. Since, the subject is in spec, IP, the sentence is

grammatical.


(19) Nadie ha visto a Juan
nobody has.3s seen PREP Juan
'No one has seen Juan'


Additionally, Casielles (1997) shows that bare nouns, such as ninfos 'children' in

Spanish can appear postverbally as subject, but they cannot be preverbal subjects.


(20) Jugaban nifos en el parque
Play.IMP.3P children PREP DET park
'Children were playing in the park'


(21) *Nihos jugaban en el parque
children play.IMP.3P PREP DET park


Bare nouns are also allowed in topic positions. In this example the bare noun

libros 'books' is topicalized.

(22) Yo a el libros no le dejo
1S PREP 3s book NEG CL-lend.ls
'as for me, books I don't lend him'











Casielles concludes that it is unlikely that preverbal subjects and topics share the

same syntactic position in Spanish because bare nouns can be topics, but cannot be

preverbal subjects.

In summary, the issue of whether the preverbal position in NSLs like Spanish and

EP is an A- or A'-position has not been decided. In the next section, I examine BP in this

light. I argue that the facts are clearer for BP and the preverbal subject position is an A-

position.

3.2.2 BP's Preverbal Subject8

The position of the subject in BP is fixed and only a preverbal position is allowed

(Costa 1998, Kato 2000, Silva 2001, Silva 2004):

(23) a. SVO

O Cesar consertou o carro
DET Cesar fixed DET car
'Cesar fixed the car'

b. *VSO

*Consertou o Cesar o carro
fixed DET Cesar DET car
'Cesar fixed the car'




8 I will not discuss postverbal subjects in the work. Briefly, BP subjects can be to the right of the verb; however, VS
order is restricted, for the most part, to unaccusative verbs (Silva 2001, Silva 2004). Since these are objects of the
verb, it is assumed that they are allowed to stay in-situ.

i) A gua congelou ii) Congelou a agua.
det water froze.3sg det water froze.3sg
'The water froze.' 'The water froze'
BP does not allow postverbal subjects with transitive or unergative verbs (Silva 2001:84).
iii) *Leu a revista a Bia iv) *Leu a Bia a revista
Read the magazine the Bia Read the Bia the magazine

v) *Brinca a crianca em casa vi) *Brinca em casa a crianga
Plays a child in house Plays in house a child









c. *VOS

*Consertou o carro o Cesar
fixed DET car DET Cesar
'Cesar fixed the car'


In what follows I claim that for BP the subject position is unambiguously an A-

position, spec, IP, in agreement with Silva (2004), Barbosa (2004), Martins & Nunes

(2006), Pires (2007), and others. The arguments come from non-complementarity

between preverbal subjects and temporal adverbs, doubled subjects, and preverbal

quantifiers.

3.2.2.1 Temporal Adverbs

Recall that Alexiadou & Anagnostopoulou (1998) claim that complementarity

between preverbal subjects and temporal adverbs in Spanish points toward the

preverbal subject and adverb both occupying an A'-position in the left periphery. This

same complementarity does not exist in BP as example (24)b shows. This sentence is

similar to the Spanish example (5), repeated as (24)a from above, that dealt with

temporal adverbs like temprano, 'early'. In Spanish, the temporal adverb and the

subject can not both be to the left of the verb. BP does allow both a temporal adverb

and the subject to appear left of the verb, as in (24)b (Pires 2007).


(24) a. *Temprano Julia salia de casa Spanish
early Julia leave PREP house
'Early, Julia left from home'

b. Hoje cedo a Julia saiu de carro BP
Today early DET Julia left.3SG PREP car
'Early today, Julia left by car'









This suggests that the temporal adverb and the subject do not compete for the

same structural position, which is compatible with the subject being in an A position.

3.2.2.2 Weak Pronouns

Weak Pronouns also point to an A position for the subject. Phonologically reduced

pronouns are referred to as weak pronouns (Cardinaletti & Starke 1999). Weak

pronouns are subject to syntactic restrictions whereas full pronouns are not, and this is

the case for BP. An example of a phonologically reduced pronoun is ce, which cannot

occur in isolation, (25).


(25) A. Quem fez o trabalho?
Who did.3s DET work
'Who did the work'

B. *CG / Voce
2s
'You'


This pronoun cannot occur in the left periphery as (26) and (28) show. The full

form voce can occur in any position, seen in (27). This phonologically reduced pronoun

can occur in the subject position as a doubled subject, example (27) (Kato 1999:28 for

reference to phonologically-reduced pronouns in spec TP, Pires (2007:132) for the

examples below).


(26) *C6 / Voc6, eu vi
2s is saw. s
'you, I saw'

(27) Voce, o seu pai c6 pode convidar
2s, DET your father 2s AUX invite.INF
'As for you, your father, you can invite'

(28) *C6, o seu pai c6 pode convidar
2s, DET your father 2s AUX invite.INF











The phonologically reduced pronouns do not have to occur in conjunction with

some full pronoun or NP in the left periphery, as example (29) shows. Here the

pronoun is the subject and appears alone.


(29) c6 pode convidar ele
2s CAN.2S invite.INF 3s
'You can invite him'


This evidence suggests that a phonologically reduced subject pronoun is in an A

position. While this does not guarantee that full pronouns and other NPs are also in an

A position, it confirms that an A-subject position exists and this position is filled at least

regularly by phonologically-reduced pronouns.

3.2.2.3 Quantifiers

Much like the phonologically-reduced pronouns, bare quantifiers cannot be left-

dislocated in BP. In fact, it is well known that non-referential expressions like quantifiers

and n-words cannot be left-dislocated in many languages. Take the examples from

French. French allows doubled subjects much like Portuguese. The examples here

show that the doubled subjects quelqu'un and personnel cannot be in the left periphery,

(30) and (32), but they can be in the subject position, (31) and (33).


(30) *Quelqu'un il vient.
someone he comes

(31) Quelqu'un vient.
someone comes
'Someone is coming.'

(32) *Personne il n'a rien dit.
no-one he not-has nothing said









(33) Personne n'a rien dit.
no-one not-has nothing said
No one said anything.'

The same can be said of quantifiers in BP (Pires 2007:126). They are grammatical

in subject position, (34) and (36), but not fronted (35) and (37):9


(34) Alguem pintou esse quadro
Someone painted.3s that painting
'Someone painted that painting'

(35) ??Alguem, o Joao nao ajudou
Someone, DET John NEG helped.3s
'Someone, John didn't help.'

(36) Ninguem viu o Jo0o.
Nobody saw. 3s DET John
'Nobody saw John'


(37) *Ninguem, o Joao
Nobody, DET John
'Nobody, John saw'


(nao) viu.
(NEG) saw.3s


These examples verify that while an A'-position in the left periphery is not open for

quantifiers, the subject position is. This data argues that BP subjects are not in an A'-

position, but in an A-position.

Related to quantified expressions is Anagnostopoulou & Alexiadou (1998)'s claim

that, in Greek, preverbal subject QPs have unambiguous scope, and postverbal QPs

have ambiguous scope. This was presented in examples (12) for a preverbal subject,

repeated in (38), and (13) for a postverbal subject. Different from Greek, BP maintains

a scopal ambiguity when the subject is preverbal, which supports the claim that the


9 Belletti claims that Italian indefinite quantifiers need to receive emphatic stress to be grammatical in
subject position; however, this has been shown to not be the case for BP (Costa and Galves 2002).









subject is in an A-position because it is believed that an A-position allows for scopal

ambiguity.


(38) Kapios fititis stihiothetise kathe arthro
Some student filed every article
'Some student filed every article' some>every
'*For every article, there is some student that filed it' *every>some

(39) Alguns estudantes leram todos os artigos
Some students read.3P all DET articles
'Some students read all the articles' some>all
'For all the articles, there is some student that read them' all>some


This section has demonstrated that the preverbal subject in BP is an A-position.

The evidence for this included the possibility of weak pronouns and quantifiers, neither

of which can appear in A'-positions.

3.3 Verbs and Verb Movement

This section addresses more precisely the position of the verb in BP. A discussion

about verb movement in BP must include a discussion about the internal structure of IP

which is subject to debate. This section discusses that debate and locates BP within

that context. For Romance languages, the debate is over Pollock (1989)'s split IP.

Specifically, Pollock claims that IP is broken up into different functional heads such as

AgrP and TP, where AgrP is higher in the derivation than TP. With regards to the ideas

for BP, researchers argue two proposals:

i) BP's IP is not split into T(ense)P and Agr(eement)P, there is a single

TP projection, the verb moves to T, and the subject moves to spec,TP

(Kato 1999, Rodrigues 2004);









ii) IP is split into TP and AgrP, with AgrP above TP, and the verb moves

to T and the subject moves to spec,AgrP (Costa & Galves 2002, Silva

2004).

I follow (ii) and conclude that the subject in BP is in spec,AgrP. To do this, I

explain the debate and its consequences for understanding BP. Section two of this

discussion looks at the evidence for a Split IP. First, I discuss floating quantifiers as a

diagnostic for a split IP. Then I discuss some adverbs and their relationship to the

verb's movement.

3.3.1 IP

Pollock (1989)'s argument for split IP was based on word order differences

between English and French. Pollock argues that the English negative marker not and

the French marker pas indicate that the verbs for each language are in a different

positions in finite sentences. In French, the verb moves above the negative marker,

whereas in English the verb does not move. Instead, English uses an auxiliary verb do

to occupy the position occupied by aime in French.


(40) Jean n'aime pas Marie
John NEG-loves NEG Mary
'John doesn't love Mary'

(41) *John likes not Mary
(42) John does not like Mary


(43) *Jean ne pas aime Marie
John NEG NEG loves Mary
'John doesn't love Mary'


Pollock explains using adverbs that the verbs in each language are in different

structural positions (examples from Belletti 2004). Examples (44), (45), and (46) show









that some adverbs and the negative marker are to the right of finite verbs and to the left

of non-finite verbs. Example (47) confirms that the adverb can be to right of the non-

finite verb, but the negative marker cannot, as in (48). Pollock claims French verbs can

occupy at least two positions in IP: a higher position for finite verbs; and a lower position

for non-finite verbs.


(44) Jean recontre
John meets

(45) Jean essaye de
John tries

(46) Jean essaye de
John tries

(47) Jean essaye de
John tries

(48) *Jean essaye de ne
John tries


souvent
often

souvent
often

ne pas
not

recontrer
to meet

recontrer
to meet


The contrasts with English are clear. The English examples show that English

verbs do not move to either of the positions suggested for French. Notice that in cases

(50), (52), and (54) the adverb or the negative marker must precede the verb.


(49) John often meets Mary
(50) *John meets often Mary
(51) John tries to often meet Mary
(52) *John tries to meet often Mary
(53) John tries not to meet Mary
(54) *John tries to meet not Mary.



Based on these and other pieces of data, Pollock concludes:


Marie
Mary

recontrer
to meet

recontrer
to meet

souvent
often

pas Marie
not Mary.


Marie
Mary

Marie
Mary

Marie
Mary









(55) English: lexical verbs do not move to either inflectional position, these positions
are filled by modal and auxiliary verbs

(56) French: lexical verbs must move to the higher inflectional position in finite clauses,
and nonfinite verbs optionally move to the lower inflectional position only.


Thus, in French finite clauses it is claimed that the verb undergoes long movement

to the highest projection in IP, and in non-finite clauses, the verb moves to a lower

projection or doesn't move at all. For example, contrasting example (40) with (57), in

the first case the verb moves past the negative marker, and in the second the verb does

not move past the negative marker. However, again contrasting (47) and (57), the first

allows the non-finite verb appear to the left of the adverb, but not the negative marker.

This movement above the adverb but below the negative marker could be considered

short-movement.


(57) Ne pas sembler heureux est une condition pour ecrire des romans.
NEG NEG seem.INF happy is DET condition PREP write.INF DET novels'
Not to seem happy is a prerequisite for writing novels.'


Generally for Romance, the two main projections posited as making up IP are

TenseP and Agr(eement)P. This is based on the make-up of Romance verbs because

they, in most cases, overtly exhibit these two functional morphemes. The Italian

example and the Spanish show the morphemic breakdown of the verb, separating tense

and person/number agreement.

(58) Canta-va-no Italian
Root-Tense-Person
Sing-IMP-3P
'They were singing'










(59) Llega-ba-n Spanish
Root-Tense-Person
Arrive-IMP-3P
'They were arriving'


While there is evidence for split IP in French, some authors do not believe that the

split necessarily exists in all languages (latridou 1990, Thrainsson 1996, Bobaljik and

Thrainsson 1998). latridou suggests that while some languages may have TP and

AgrP, there needs to be evidence in each language for such a split (latridou 1990:553).

Thrainsson agrees by positing The Real Minimalist Principle: Assume only those

functional categories that you have evidence for (1996:261). Because of objections

raised by these linguists and others that have come about since Pollock's paper,

linguists debate about the make up of IP for BP.

3.3.2 Non-Split IP

One of the goals of Chomsky (1995) is to eliminate AgrP as a functional clausal

head as it has no interpretation, and only things that receive interpretation can be

clausal heads. Following Chomsky (1995), Kato (2000) argues that AgrP is not part of

BP and the subject is realized in spec,TP. Her argument is based on the changes in

verbal morphology that have been observed in the last one hundred years. Duarte

(1995), among many others, shows that BP verbal morphology has weakened to the

point where it is changing BP from a null subject language to a non-null subject

language. Since this change, a more rigid order of SVO is observed, as seen in section

two. The next chart, table 3-1 shows the weakening of agreement morphology in BP

over time where paradigm one indicates the colonial period (1500-1820s), paradigm two

is the post-colonial period (1820s-1900s), and finally paradigm three represents the









present day (Duarte 2000:19). BP goes from having six agreement conjugations to

three. Notice that ------ means the form is no longer in use.

Table 3-1 Loss of Verbal Morphology in BP
Person/Number Pronouns Paradigm 1 Paradigm 2 Paradigm 3
1st singular Eu am o am o am o

2nd singular Tu ama s --
Voce ama ama ama

3rd singular Ela/ele ama ama ama

1st plural N6s ama mos ama mos
A gente ama ama

2nd plural V6s ama is --
Voces ama m ama m ama m

3rd plural Eles/Elas ama m ama m ama m


Since the weakening began, weak pronouns have begun to develop (Galves 1998, Kato

1999). According to Kato, both of these factors lead to the reinterpretation of Agr as a

lexical item having lost syntactic independence, i.e. it is no longer an independent

projection like TP (Kato 2000). Thus, IP is not split along the AgrP TP distinction.

Additionally, as some have noted Agr is relational, not functional, and therefore

does not head its own projection. Chomsky (1995: 240, 349-55) and others argue that

Agr is fundamentally different from such functional categories as Tense and Negation.

He argues that, unlike T, C, and D, which have [+interpretable] features, Agr has no

semantic properties but consists of [-interpretable] formal features only. Moreover, "Agr

exists only when it has strong features. Agr is nothing more than an indication of a

position that must be occupied at once by overt operations" (Chomsky 1995: 351).









3.3.3 AgrP in BP

While some argue in favor of an unsplit IP, based on empirical evidence, I assume

that despite significant change in the verbal morphology, the change did not cause AgrP

to be eliminated from the grammar. Here I follow Galves (1998) and Costa & Galves

(2002) who propose that there is still a projection between CP and TP necessary to

account for cross-linguistic differences. In fact, BP needs a projection like AgrP to

understand certain facts about the position of the verb, the subject, floating quantifiers,

and adverbs. The BP facts parallel the facts in French non-finite clauses, which showed

short verb movement (Costa and Galves 2002). This contrasts with languages such as

French finite clauses, which have long verb movement, or English, which has no overt

verb movement.

A diagnostic tool used to determine the structure of IP is the presence of floating

quantifiers (FQ) such as BP todo 'all' (see Vicente 2006 for a complete discussion of

FQs in BP). I follow Vicente (2006) and Muller et al. (2007) who convincingly argue for

FQ as a diagnostic tool to understand BP structure. Sportiche (1988) proposes that

FQs initially form a constituent with the DP they modify. When the DP moves, the FQ

can be "stranded" in positions that the DP moves through. The stranding results when a

subconstituent of the DP moves further. FQs thus mark positions through which a DP

has moved.

Sportiche's analysis is based on a series of conclusions from the 80's (Bobaljik

2003:6):









(60) Floating Quantifiers:

(i) FQs appeared to modify DPs in the same way as DP-initial Qs;
(ii) FQs in some languages display determiner-like agreement with the DP they
modify;
(iii) FQs surface in the left periphery of (certain) maximal projections, especially
VP;
(iv) the relationship between an FQ and the DP it modifies obeys an anaphor-like
locality condition.


Assuming the Predicate-Internal Subject Hypothesis (Koopman and Sportiche

1991), then a quantifier should be allowed to be stranded in positions below the surface

position of the subject. This is what we see in the BP examples below from Brito

(2001:78). The lowest position in which the FQ can appear is to the right of the main

verb, in (63). The FQ is stranded in the subject's base position, spec,vP. Both the

subject and verb start inside vP but move out above spec,vP, as I explain below. (64) is

ungrammatical with todas to the right of the object.


(61) Todas as mulhers comeram a lagosta
All DET women eat.PST.3P DET lobster
'All of the women ate the lobster with their hands'

(62) As mulhers todas comeram a lagosta
(63) As mulhers comeram todas a lagosta
(64) *As mulhers comeram a lagosta todas


The crucial example in support of a split IP is (62), where the verb and the subject are

separated by the FQ. As Costa & Galves (2002) suggest, if IP is split into at least AgrP

and TP, then the grammaticality of (62) is predicted. I argue that the verb moves to T

and the subject moves to the spec,TP. After that, the subject moves up to spec,AgrP,

and the FQ is stranded in spec,TP, with the verb remaining in T, hence the separation

of the verb from the subject.










(65) AgrP

DP
As Mulhersi


Agr'


Agr

[todas ti]j


T
comeramii


This example contrasts with (63), where the FQ is to the right of the verb. This example

and the tree in (66) show that if the subject is generated in vP, then for the FQ to be to

the right of the verb, the verb must move outside of vP. In this case, I have argued that

it does move, and it moves to T.


(66) AgrP


DP
As Mulhersi


Agr'


Agr


T
comeramii
DP
todas ti









These facts about FQs support a split IP in BP. If the floating quantifier is between

the subject and the verb and both are outside vP, then there must be a position

between the subject and the verb where a DP can land. I claim that this position is

spec,TP in a split IP scenario.

One could argue that the FQ is merely in spec, vP, the original position of the

subject, and the subject is in the specifier of a non-split IP; however, this analysis runs

into problems with (63). (63) provides evidence that the verb moves to a position higher

than the FQ, which is no lower than spec,vP, meaning that the verb moves out of vP.

For there to be a specifier position for the FQ and a head position for the verb above vP

but below the surface position of the subject requires another projection, as provided in

a split IP scenario.

Adverbs further support the claim that the verb moves to T, necessitating a split IP.

Galves (1994) and Costa & Galves (2002) show that there are at least two adverb

positions in a simple BP clause, preverbal and postverbal:


(67) a. 0 Jo0o beija frequentemente a Maria.
DET Jo0o kiss.3s frequently DET Maria
'Joao kisses often Maria'

b. 0 Jo0o frequentemente beija a Maria.
DET Jo0o frequently kiss.3s DET Maria
'Joao often kisses Maria'

This pattern is different from what was observed for French and English in section

3.3.1. Recall that the English adverb must be to the left of the verb and the French

adverb to the right.

(68) Jean rencontre souvent Marie
John meets often Mary
'John often meets Mary'










(69) *Jean souvent rencontre Marie
John often meets Mary
'John often meets Mary'

(70) John often meets Mary

(71) *John meets often Mary

Costa (2004) argued that this difference between Portuguese and French and English is

due to short verb movement in Portuguese. Verb movement according to Costa is short

in this case because IP is split into more than one functional projection. Portuguese

verbs move to the lower head position in a split IP, hence short verb movement. His

analysis claims that the subject is in spec,AgrP, and the verb moves only to T, which is

below AgrP. The difference between sentences (a) and (b) in (67) is that the adverb

adjoins in two different places. In the case of sentence (a) the adverb is in a low

position, and for (b) the adverb is in a high position. The structure below represents the

syntactic positions referred to as 'high' and 'low'.


(72) AgrP

DP Agr'
O Pedroi
Agr TP

High AdvP TP

T vP

Low AdvP vP


Costa & Galves (2002) use high and low adverbs to show that the verb is in T. In

Portuguese a low adverb is an adverb that surfaces only postverbally. Some examples









of these are bem 'well' and atentamente 'carefully'. As they can only appear

postverbally, Costa & Galves claim that the verb must move above these low adverbs,

which is the case in (73). Notice that (74)10 is ungrammatical with the verb to the right

of the low adverb. Thus, the verb must move past these low adverbs in sentences like

(73). Sentence (73) is represented in (75) where the low adverb adjoins to vP.


(73) 0 Pedro leu bem/atentamente o livro
DET Peter read.3s well/carefully DET book
'Peter read well/carefully the book'

(74) *O Pedro bem/atentamente leu o livro

(75) AgrP

DP Agr'
O Pedroi
Agr TP

ti T'

T vP
leuii
AdvP vP
bem
ti tii o livro


Next, Costa & Galves show that the verb can appear between two adverbs. The

case of (76) has the low adverb bem 'well' and what they call a high adverb, ontem

'yesterday'. Different from the low adverb, the high adverb adjoins to TP seen in (77),

causing the verb to be to its right.




10 In the case of example (74), the sentence would be grammatical if there was a pause. Costa & Galves
are assuming no pause.









(76) Os meninos ontem leram bem o livro
DET children yesterday read.PST.3P well DET book
'Yesterday the children read well the book'

(77) AgrP

DP Agr'
O Pedroi
Agr TP

AdvP TP
ontem
DP T'
ti
T vP
leuii
AdvP vP
bem
ti tii o livro


Finally, Costa & Galves show that certain adverbs can adjoin either high or low

and, depending on their position of adjunction, they can yield a subject-oriented reading

or a manner reading (Jackendoff 1972). A subject-oriented meaning refers to the

subject having the quality that the adverbs suggest. A manner-oriented meaning is one

where the adverb is more closely linked to the meaning of the action that the verb

conveys. A subject oriented reading corresponds to a high adverb and TP adjunction.

The adverb inteligentemente in (78) adjoins to TP.


(78) 0 Pedro inteligentemente leu o livro (Subject-oriented/*manner)
DET Peter intelligently read.3sG DET book
'It is clever of Pedro to read the book'










(79) AgrP


DP
O Pedroi


Agr'


Agr

AdvP
inteligentemente


A manner-oriented reading results from a low adverb that is adjoined to vP.

Sentence (80) has the same adverb inteligentemente but in this case it adjoins to vP.


(80) 0 Pedro leu inteligentemente o livro (*Subject-oriented/Manner)
DET Peter read.3sG intelligently DET book
'Pedro reads the book in a clever manner'


(81) AgrP


DP
O Pedroi


Agr'


Agr


AdvP
inteligentemente


Given these data, Costa & Galves (2002) claim that the verb must move outside the vP

in BP. Not only must it move outside the vP, but it must move to a position that is not

high in the IP domain. To show that the movement must be to a landing site low in the









IP domain, they contrast the Portuguese data with French. In French, the verb always

precedes the adverb regardless of the adverb's interpretation, as in (82). This is

possible because in French the verb must move all the way to Agr. The ambiguity

between the readings can arise because the verb is obligatorily in a higher projection

above both the TP-adjoined and vP-adjoined positions.


(82) Pierre lit intelligemment le livre (Subject-oriented/Manner)
Peter read.3s intelligently DET book
'Peter intelligently reads the book.'


(83) AgrP

DP Agr'
Pierrei
Agr TP (-Subject-Oriented
litii
Adv TP
intelligemment
ti T'

T vP (-Manner

AdvP vP
Intelligemment


If one assumes that the adjunction positions of the adverb is the same for French as it is

in Portuguese, i.e. TP for subject-oriented and vP for manner, then the cross-linguistic

differences follow from differences in verb movement in French versus BP.

The picture is further supported by the interaction of adverbs like 'intelligently'

with negation, which is below TP and above vP. Here the negative marker pas

distinguishes the two adjunction sites. (84) has the adverb adjoined to TP and negation









follows the adverb. In (86) the adverb is adjoined to the vP and negation precedes the


adverb.

Subject-oriented/*manner


(84) Pierre ne lit intelligemment
Pierre NEG reads intelligently
'Pierre doesn't read intelligently the book'


pas le livre
NEG DET book


(85) AgrP


Agr'


Agr T
ne litii
AdvP
intelligemment
ti


NegP


Neg
pas


(86) Pierre ne lit pas intelligemment
Pierre NEG reads NEG intelligently
'Pierre doesn't read intelligently the book'


le livre
DET book


(87) AgrP


Agr'


Agr
ne litii


NegP


Neg
pas


AdvP
intelligemment


DP
Pierrei


*Subject-oriented/Manner


Pierrei










Given these facts and those from the previous two sections, I have argued that IP

is split in BP, following Pollock (1989), into at least AgrP and TP. The subject moves to

the specifier of the higher projection, spec,AgrP, and the verb exhibits short movement

to T, the head of the lower projection. Adverb placement data supported this picture.

3.4 Fronting in BP

Fronting in Portuguese differs from other languages in the Romance family and

languages such as English. I show that BP uses base-generated topics. This can be

contrasted with English, which uses movement topicalization, and Spanish, which uses

Clitic-Left Dislocation (CLLD). Base-generated topics are merged into the derivation in

the higher topic position discussed in chapter two.

Below is an English sentence where a DP is fronted and leaves behind a trace.. I

call such a derivation topicalization.


(88) This booki, I like ti


CLLD occurs when a DP is fronted and subsequently a pronominal clitic marker

appears. This is seen in Spanish (89), where este libro is fronted and a clitic lo

appears. Notice that if the sentence does not have a clitic, it is ungrammatical, as in

(90). This shows that Spanish does not allow topicalization as English does, where only

a null trace is left behind.


(89) Este libro, yo lo lei Spanish
this book, 1S CL read.1s
'This book I read it'

(90) *Este libro, yo lei
this book, 1s read. 1s
'This book, I read'










It has been claimed that fronting in BP involves left dislocation. Different from Spanish,

a clitic does not appear but a resumptive pronoun does, (91).


(91) Este livro, eu li ele BP
this book, 1s read. 1s 3s
'This book I read it'


However, in BP the resumptive pronoun is not necessary, seen in example (92),

and it is here that Portuguese differs from other Romance languages. Cases like (92)

suggest that BP has topicalization much like English.


(92) Este livro, eu li
this book 1s read. s
'This book, I read'


BP topicalization not only occurs with DPs but also with VPs, which is another area that

differentiates Portuguese from the many other Romances languages. When a VNP is

topicalized, it is in the infinitive form as in (93). The topicalized VP can also contain

complements of the verb.


(93) Dormir ele dorme
sleep.INF, 3s sleep.3s
'As for sleeping, he sleeps'

(94) Cantar fado ele canta
sing.INF fado 3s sings.3s
'As for singing fado, he sings'

I only treat fronting of DPs and dispense with discussing VP fronting11 because

both cases are allowed with CFN sentences.


11 For a complete analysis of VP fronting in BP, see Bastos (2001, 2009).









Two analyses for BP fronting are: movement or base-generation. Both English

Topicalization and CLLD make use of a movement solution. I briefly discuss the

movement analyses, and from there I present the base-generated analysis. The facts

about base-generated topics become useful in understanding my analysis of clause-

final negation presented in chapter six.

Kato (2003) and Kato & Raposo (2007) (K&R) argue for a traditional movement

analysis for fronting constructions in BP. K&R propose that DP fronting in BP is really

CLLD, much like Spanish, but in this case the clitic is null. They claim this in order to fill

out the apparently defective clitic paradigm. Third person clitics exist but are null. Their

analysis originates with Uriagereka's (1995) work on Spanish Clitic Doubling. He

proposes that the clitic and the associated NP start in a spec-head relation in DP.

From there, the NP moves to a position higher in the derivation, and the clitic is

stranded somewhere along the way. The derivation for a Spanish sentence like (95), is

given in (96).

(95) A Juan, le doy un libro
PREP John CL.3s give.ls DET book
'I give the book to John'

(96) a. doy un libro [Juan le]
b. [Juan le] i doy un libro ti
c. Juanj [tj le] i doy un libro ti {merge of a}
d. a Juanj [tj le]i doy un libro ti


K & R claim that BP is different from Spanish in that the DP moves to the left but the

clitic stays in-situ. It is stranded in the DP's base position, not in a preverbal position.


(97) 0 Joao, eu vi ontem
DET John 1s saw CL yesterday
'John, I saw (him) yesterday'









(98) a. eu vi [Joao 0] ontem
b. [Joao]i eu vi [DP ti 0]


Kato & Raposo (2007) proposes that such dislocation that leaves behind a

resumptive pronominal clitic is not subject to island constraints (Ross 1967), thus the

BP topic construction is insensitive to islands. (99) shows a topic that appears to have

moved from inside a complex subject (Kato & Raposo 2007:208).


(99) Esse boloi o rapaz [que trouxe agora mesmo da pastelaria] era teu afilhado
this cake, DET boy [COMP brought CL now EMPH PREP.DET bakery] was your godson
'this cake, the boy who brought (it) just now from the bakery was your godson.'


Although K & R claim that the resumptive clitic rescues the structure from an

island violation, Cinque (1990) points out that pronominal clitics do not save sentences

from island effects. Furthermore, it is not clear what would be meant by a null clitic.

Traditionally, clitics are phonologically dependent, and it is difficult to understand how a

null clitic could exist because it has no phonology. Also, it is never quite clear why in the

case of Spanish the clitic moves with the NP, but in Portuguese the clitic stays in-situ.

For these reasons, I move away from a movement analysis of BP fronting and

propose that BP fronting is an instance of a base-generated topic in spec,TopP, with a

resumptive pronoun, sometimes null, generated in the argument position. One reason

to think that a movement analysis is incorrect for BP comes from the same example that

K & R used in (99). The DP moves out of an island but does not violate Ross's island

constraints. This fact follows if fronted DPs are base-generated. Additional evidence for

this analysis comes from the absence of third person clitic pronouns for both the

accusative or dative case noted by K&R. Recall from section 3.2, that BP has









developed weak pronouns in the sense of Cardinaletti & Starke (1999) (Galves 1998).

The examples for second person voce, c6, were given above. In the case of third

person, both plural and singular, weak pronouns exist but have no phonological

realization (Galves 1989, Schwenter & Silva 2002, Kato 2003). Evidence for the null

third person object pronouns comes not only from topic structures, but many other types

of sentences. For example, the following sentences are an exchange between two

people. Notice that in (101) there is no overt object nor is there a topicalized DP. In this

case, a null pronoun is in the object position. Although, John is the topic of the

exchange, there is no DP in sentence in the topic position, and the sentence is still

grammatical because of the null pronoun (Galves 1989).


(100) Joao vem?
John comes
'Is John coming'


(101) Ninguem convidou pro.
n-body invited pro
"Nobody invited him"


The behavior of wh-movement in BP also supports a base-generation analysis of

fronting. In BP, wh-movement is not allowed out of island, different from what has been

just seen for BP topics. Compare (102) and (103). In the first case, a null pronoun is

generated in the argument position and the topic is generated in the left periphery.

Because there is no movement, there is no island violation. The second case has wh-

movement, leaving a trace, which results in an island violation.

(102)Estes CDs, voces encontraram uma loja que vende pro
These CDs, 2P find.pst.2P DET store that sell.3s pro
'These CDs, y'all found a store that sells them'










(103)*Que CD1 voces encontraram uma loja que vende ti?
Which CD 2P find.pst.2P DET store that sell.3s
"Which CD have you found a shop that sells it?


The evidence given suggests that topics in BP are based-generated and a resumptive

pronominal element, either null or not, is in the argument position.

3.5 Wh-Questions: Fronted and In-Situ

BP makes use of both fronted and in-situ wh-questions. In this section, I discuss

the derivations for both types of questions. There have been several different opinions

through the more recent years of generative grammar as to the nature of wh-questions

in BP. Kato & Raposo (1996) claim that all fronted wh-questions are clefted

expressions. Kato & Mioto (2005), Pires & Taylor (2007), Zocca (2007) argue for

understanding them as feature agreement (although the mechanics of each of these

differ), and in this work, I follow the claims made by Pires & Taylor and Kato & Mioto. I

first give a brief outline of the different types of wh-questions in BP. After that, I explain

the derivation for both movement and in-situ questions.

3.5.1 Types and Conditions for Wh-Words: Movement and In-Situ

Not only does BP have fronted and in-situ wh-questions, the fronted questions

come in four versions. The examples below illustrate both fronted and in-situ questions.

The first set has fronted wh-words, (104) (107), and the question is 'what did you eat?'.

The second set is a BP sentence with a wh-in-situ, (108), which is 'you ate what?'. The

first version is a fronted wh-question with no complementizer or copula. The second

version has the complementizer que following the wh-phrase. The third version has the

wh-word fronted with a clefted construction involving 6 que 'is that'. The fourth version

combines the second and third. Examples (106) and (107) both appear to be clefted









constructions, and I do not treat them here (see Kato & Mioto 2005a,b for an analysis of


these constructions).


(104)0 que
DET what

(105)0 que
DET what

(106)0 que
DET what

(107)0 que
DET What


voce comeu?
2s ate.2s

que voce comeu?
COMP 2s ate. 2s


e que
is COMP


voce
2s


que e que
COMP is COMP


comeu?
ate. 2s


voce comeu?
2s ate. 2s


'What did you eat?'


(108)Voce comeu
2s ate. 2s
'You ate what?'


o que?
DET what


Some additional facts are important to note. First, the interrogative

complementizer que is optional with fronted wh-words, (109) and (110), but impossible

with wh-in-situ, (111) and (112) (Hornstein et al. 2005).


(109)Como voce
how 2s
'How did you


fez isso?
do.PSTthat
do that?'


(110)Voce fez isso
2s do.PST that
'How did you do that?'


(111)Como que
how that
'How did you


voce fez
2s do.PST
do that?'


como?
how


isso?
that










(112)*Que voce fez isso como?
that 2s do.PST that how
'How did you do that?'

Second, it has been noted that whether the complementizer is null or overt, wh-

movement within embedded interrogative clauses is obligatory (Zocca 2007). The

examples below illustrate this with the interrogative verb perguntar 'to ask'. This verb

forces wh-movement in its complement, which is why (114) is ungrammatical, as the

wh-phrase stays in-situ.


(113)Eu perguntei como (que) voce fez isso.
1s ask.PST.1s how (that) 2s did that
'I asked how you did that.'

(114)*Eu perguntei voce fez isso como.
1s ask.PST.1s 2s did that how
*'I asked you did that how.'


Third, BP wh-movement is sensitive to islands while wh-in-situ is not. Thus,

sentence (117) where the wh-phrase moved out of an island is ungrammatical.

However, if the wh-word does not move, then it is licit even in island contexts. (118) has

que livro 'which book' in a complex NP island and is still grammatical.


(115)Que livro voce disse que ela comprou?
which book 2s said that 3sF bought
'Which book did you say that she bought?'

(116)Voc6 disse que ela comprou que livro?
2s said that 3sF bought which book
'Which book did you say that she bought?'

(117)*Que livro voce conversou com o autor [que escreveu]?
which book 2s talked PREP DET author that wrote
'Which book did you converse with the author that wrote?'









(118)Voc6 conversou com o autor [que escreveu que livro]?
2s talked PREP DET author that wrote which book
'Which book did you converse with the author that wrote?'


Finally, unlike other Romance languages, there is no subject-verb inversion in

main or embedded clauses with wh-movement. This type of inversion is seen in

European Portuguese; however, in BP the meaning changes completely, as seen in

sentence (120).

(119)0 que comeu a LQcia? European Portuguese
What comeu. 3s DET LQcia
'What did LQcia eat?

(120) O que comeu a LQcia BP
what ate.3s DET LQcia
'What ate LQcia?'
*'What did LQcia eat?'

3.5.2 Deriving Wh-Questions in BP

In this section I outline an approach to wh-words, both fronted and in-situ. The

main conclusions for both types of wh-questions are as follows.

For fronted wh-questions, I claim that:

i) an element in Foc with a strong, uninterpretable wh-feature [uWH*] triggers
movement;
ii) the wh-word with a [iWH] feature must move to a position where it satisfies the
strong features of the element in Foc.
For Wh-in-situ in BP:

i) Wh-in-situ in BP has a null Q element inserted into TopP;

ii) the wh-word has an [iWH] feature;

iii) Q unselectively binds with the wh-word;

iv) the null Q in TopP has some discourse features indicating that the question is
part of the Common Ground.











Point iv) for the in-situ question requires special attention. I discuss the notion of

Common Ground (Stalnaker 1978), which is used by Pires & Taylor (2007) to

understand the seeming optionality of wh-in-situ. I follow their proposal that the null Q

carries a discourse feature that allows wh-in-situ, and if the null Q is present, then

movement is not allowed.

First, to explain fronted wh-expressions in BP, I use feature checking and Agree

outlined in chapter two. I begin here with an overt focus head que. Example (105),

repeated here with a derivation in (122), contains this head, which has an [uWH*] which

triggers movement of the wh-word with an [iWH] feature to spec,Foc(us)P.


(121)0 que [iWH] que [uWH*] voc6 comeu?
DET what COMP 2s ate.2s
'What did you eat?


(122) FocP

DP Foc'
O que[iWH]k
Foc AgrP
que [uWH*]
DP Agr'
vocei
Agr TP

ti T'

T vP
comeu
tk









This overt head must check its strong feature, and therefore, if there is no movement

the derivation crashes:


(123)*que voc6 comeu o que?
COMP 2s ate.2s DET what
You ate what?


Like example 0, there are cases where there is no overt complementizer and yet the

wh-phrase is fronted. In these cases, the overt Focus head is deleted after movement

of the wh-phrase (Kato & Mioto 2005); however, the same mechanisms motivate

fronting, i.e. the [uWH*] feature of the deleted head.


(124)0 que que voc6 comeu?
DET what comp 2s ate.2s
'What did you eat?


In addition to the Focus head that forces movement, I claim that BP has a null

head with [uWH] which is not strong (see Pires & Taylor 2007, Zocca 2007 for a similar

account for BP). This is similar to what was proposed for French, Chinese and other

wh-in-situ languages (BoSkovi6 1999, Aoun & Li 2000). The null head does not require

movement, evidenced by the fact that wh-in-situ interrogatives can evade island

violations. What differentiates the null head that does not cause movement and the

overt que which does is their discourse contributions. The null head associated with

wh-in-situ, as mentioned, is related to Common Ground and discourse-old status.

Pires & Taylor (P & T) (2007) proposes that single wh-questions can have the wh-

phrase in-situ, provided that semantic and pragmatic requirements are met. If these

semantic and pragmatic requirements are not met, then, the wh-expression must be









fronted. P & T's analysis depends upon Stalnaker's (1978, 2002) Common Ground,

which they define as "information that was previously given in the discourse or in the

extralinguistic context ... and which is shared (or assumed by the speaker to be shared)

by speaker and hearer" (P & T 2007:5). For BP, "wh-in-situ is possible if the

information that is being requested is expected by the speaker to be part of the

Common Ground" (2007:6). When talking about the Common Ground, Stalnaker states

that "the presuppositions of a speaker are the propositions whose truth he takes for

granted as part of the background of the conversation" (1978:149). This means that for

in-situ questions in BP there is an answer to the question. Thus, in the example below,

'you know who in Sao Paulo', is similar to 'you know someone in Sao Paulo' as there

exists a person that you know in Sao Paulo. In both cases the answer is non-null, and

the speaker presupposes the existence of a person that the addressee knows in Sao

Paulo.

P & L give the following interaction as an example. The infelicity of this example

stems from the fact the information being requested (i.e. knowing someone in Sao

Paulo) is not part of the Common Ground, having not been overtly introduced into the

discourse.

[You approach a colleague at work and ask, out of the blue:]

(125) #Voc6 conhece quem em Sao Paulo?
2s know.2s who PREP Sao Paulo
'You know who in Sao Paulo'

This same interaction can be made felicitous in two ways, by fronting the wh-word

or by giving a discourse context shared by both the speaker and the hearer.

Importantly, the same sentence with a fronted wh-word is grammatical and felicitous,









although my consultants all commented that this would mostly likely result in a question

such as: Are you going to Sao Paulo?

[You approach a colleague at work and ask, out of the blue:]

(126)Quem voce conhece em Sao Paulo?
who 2s know.2s PREP Sao Paulo
'You know who in Sao Paulo'


The following exchange shows how the discourse requirement affects the felicity

of the sentence. In this instance, the Common Ground is contains information that

allows the use of in-situ wh-words; namely, it is part of the Common Ground that 'you

visited Sao Paulo', and the information requested by the speaker is presupposed to

already exist and be implied in the discourse.

[You approach a colleague at work and comment:]

(127) A. Eu visitei Sao Paulo este fim de semana
1s visited. 1s Sao Paulo this end PREP week
'I visited Sao Paulo this weekend'

(128) B. Voce conhece quem em Sao Paulo?
2s know.2s who PREP Sao Paulo
'You know who in Sao Paulo?'


The analysis that I assign to such examples is given in (129). The connection to the

Common Ground is represented by a null question operator Q in the Top head which

encapsulates this restriction.12 Although I do not develop this connection, the givenness

requirement on parts of the wh-question are somehow related to topicality. This Q

operator binds the in-situ wh-phrase. Since no strong feature is present, the wh-phrase

does not move.


12 In chapter six, I will revise this analysis somewhat and claim that the Q operator is actually located in
the specifier of a lower topic phrase, Top2P.









(129) TopP

Top AgrP
Q
DP Agr'
vocei
Agr TP

ti T'

T vP
conhece /
quem[iwH] em Sao Paulo


Que cannot occur in this structure with wh-in-situ, (112), because that head has a

strong feature which would force movement. I account for the fact that wh-in-situ is

impossible in embedded clauses by simply stipulating that the Q Top head is

unavailable in selected contexts.

To summarize, BP has two interrogative heads in the left periphery: one overt one

in Foc that forces movement for feature checking and a second that is covert, related to

the discourse structure, which does not allow movement.

3.6 Conclusions

This chapter has addressed the clause structure for BP. This chapter focused

principally on the CP and IP domains. The topics addressed here were presented to

understand five sub-questions related to BP clause structure.

i. What is the subject position?
ii. What is the position of the finite verb? Is there any verb movement?
iii. What is/are the structural positions) of topics?
iv. What is the structure of wh-movement?
v. What is the structure of wh-in-situ









I have argued that BP employs a split IP (Pollock 1989). I claimed that the subject

is the specifier of AgrP. I showed that these elements cannot be in an A'-position in the

left periphery, which has been suggested by some to be the location where preverbal

subject resides in null-subject languages. Section three shows that the verb starts

inside VP and moves up to T. I argued for short movement of the verb to T (Costa &

Galves 2002). Floating quantifiers as well as adverbs showed that this is the case.

Section four discussed BP's system of topicalization. I proposed that topic structures in

BP are base-generated with the topic phrase in spec,TopP binding a resumptive

pronoun, which may be null in the third person. Finally, I gave an overview of wh-words

and questions in BP. The distinction was made between fronted and in-situ questions.

In the case of fronted wh-words, the fronted word moves to the specifier of FocP, which

is headed by que with a strong [uWH*] feature. When que is not present in wh-

questions with a fronted wh-phrase, it has been deleted. In wh-in-situ questions, this

head does not appear. Instead, there is a null operator in Top which binds the in-situ

wh-phrase.

This chapter has been developed to show the basic clause structure of Brazilian

Portuguese. Below is a tree structure that gives a basic map of the structural position of

the phrases discussed in this chapter. Notice here again that there are two topic

positions: one above FocP and one below. The Topl P above FocP is the position

discussed in this chapter. The other Top2P, mentioned in chapter two, is discussed in

detail in chapters six and seven, and it relates the CFN.









(130) ForceP

Force Topl P

Top2 FocP

Foc Top2P

Top2 Agr

Subject Agr'

Agr TP

T'

T vP
verb









CHAPTER 4
PREVERBAL NEGATION

4.1 Introduction

As this work compares the differences between preverbal and postverbal negative

markers, I begin with an analysis of preverbal negation. Several linguists have

examined negation through the lens of generative grammar, especially since Pollock's

work on IP (Pollock 1989, latridou 1990, Laka 1990, Belletti 1990, Zanuttini 1991, Mioto

1992, among many others). In this chapter, I discuss negation in general and BP

specifically. Section 4.2 discusses negation and the different approaches to

understanding syntactic issues concerning negation. Part of this section reviews some

ideas regarding negation in Romance languages. Because the topic of negation is

larger than the scope of this work, I omit discussion on the semantics of negation.

Therefore, I focus solely on some of the more common syntactic analyses of negation.

Section 4.3 moves the discussion to Brazilian Portuguese and the preverbal negative

marker. I leave the clause-final negative marker for chapters five and six and negative

concord and n-words for chapter seven. Other negative elements such as NPIs and

semi-negatives, words like sem 'without', are not discussed here.

4.2 The Syntax of Negation

The main goal of this section is to introduce sentential negation and briefly discuss

how sentential negation is understood in generative syntax. Section 4.2.1 discusses the

functional projection NegP. Here I discuss the status of certain preverbal negative

markers such as nao for BP, no in Spanish, and non for Italian. To do this, I continue

the discussion from chapter three about Pollock's (1989) work. Section 4.2.2 presents

overviews of the theories related to the location of NegP in Romance languages.


100









4.2.1 NegP

While many works predate Pollock (1989), such as Jackendoff (1972), Lasnik

(1972), Emonds (1978), he provides evidence that negation is a separate functional

category, NegP. Pollock expanded IP and posited a syntactic location for NegP. In

chapter three, I mentioned his (1989) work as an argument for separating IP into AgrP

and TP projections. Here I extend that discussion to include the data that he provided

with regards to negation.

Building on Chomsky (1976) and Emonds (1978), Pollock observes the differences

between the position of negative markers in English and French, much like he does with

adverbs. In English, the negative marker must precede a lexical verb, whereas in

French it follows the verb.


(1) John doesn't likes Mary
(2) *John likes not Mary

(3) *Jean (ne) pas aime Marie
'John doesn't love Mary'

(4) Jean (ne) aime pas Marie
'John doesn't love Mary'


In English, the negative marker must come after auxiliary verbs and before lexical

verbs. This differs from French where the negative marker must come to the right of

both auxiliary and finite lexical verbs. To understand this difference and to keep in line

with linguistic universals, Pollock proposes that the negative marker for English and

French are in a similar position, and the verb's movement differs. His conclusion is that

in English auxiliary verbs are in a position similar to that of lexical verbs in French, seen

in a comparison of (5) and (6) with (4).


101









(5) He is not coming
(6) He must not come


He also proposes that Neg houses preverbal negative markers, such as ne in French,

and spec,NegP is the locus for negative adverbs, pas in French. This line of reasoning

and his analysis were adopted by several linguists (Laka 1990, Chomsky 1995, Lasnik

1999).

(7) TP

Spec T'

T NegP

Spec Neg'
pas
Neg AgrP
ne
Spec Agr'

Agr VP


Spanish, Italian, and Portuguese all have obligatory preverbal negative markers,

and based on Pollock's arguments, the negative marker for each of these should be in

Neg. In addition to that, Haegeman (1995) claims that negative markers such as no

and nao are not only in Neg, but they are also generated there. Zanuttini13 (2001)

claims that the preverbal negative marker is a head and that it is the head of NegP.

Following Kayne (1989), she assumes that syntactic heads interfere with movement of

other heads. This is based on what is known as the Head Movement Constraint which

is formalized by Travis as follows:


13 For arguments against having the preverbal negative marker be a head see Rowlett (1998) who gives
evidence from French against some of Zanuttini's claims.


102









(8) Head Movement Constraint
An Xo may only move into the Yo which properly governs it. (Travis 1984:131)

Thus, head movement is strictly local and always moves one head to the closest head.

If the preverbal negative marker blocks head movement, then it is the head of NegP.

Kayne showed that French pronominal clitics that correspond to arguments of an

embedded clause cliticize to the matrix verb in causative constructions. This is seen in

example (9), where the clitic la cliticizes to the matrix verb fait of the main clause

(Zanuttini 2001:524).


(9) Jean la fait manger a Paul
John 3s makes eat.INF PREP Paul
'John makes Paul eat it'


The clitic is not allowed to cliticize to the matrix verb if there is a negative marker, in this

case ne, blocking the movement.


(10) *Jean I'a fait ne pas manger a I'enfant
John CL-has made NEG NEG eat.INF PREP DET-child
'John made the child not eat it'


Zanuttini also bases her evidence for the head status of the preverbal markers in Italian

and Spanish on clitic climbing. She shows with Italian data that a clitic pronoun can

climb to the matrix clause if it is not blocked by the negative marker. I give similar

evidence from Spanish. Example (11) shows that the clitic may cliticize to the verb of

the infinitival clause. (12) shows the clitic pronoun climbing to the matrix verb. (13)

introduces the negative marker no which causes the clitic to remain in the infinitival

clause. (14) is the key to Zanuttini's argument because here the clitic is not allowed to


103









climb to the matrix clause as in (12). She claims that the reason for the

ungrammaticality of (14) is the head status of no which blocks the clitic.


(11) Juan quiere
John wants
'John wants to see it'

(12) Juan lo quiere
John CL wants


verlo
see.INF-CL


ver
see.INF


(13) Juan quiere no
John wants NEG
'John wants to not see it'

(14) Juan lo quiere
John CL wants
'John wants to not see it'


verlo
see.INF-CL


no ver
NEG see.INF


Clitic climbing can also be used to argue that BP nao is a base-generated head of

NegP:


(15) Joao quer te ver
John wants CL.2S see.INFL
'John wants to see you'


(16) Joao te quer
John CL wants
'John wants to see you'


ver
see.INF


(17) Joao quer nao te ver
John wants NEG CL.2S see.INFL
'John wants to not see you'


(18) *Joao te quer ndo
John CL.2s wants NEG
'John wants to not see you'


ver
see.INF


104









This section has shown that the preverbal negative markers no, nao, and non

block clitic climbing, and this blockage is related to head movement. This implies that

these preverbal negative markers occupy a head position.

4.2.2 The Location of NegP

While Pollock's work was a success in that it highlighted the importance of a

functional projection specifically for negation, problems began to arise as to the actual

syntactic location of NegP. While his claims seemed strong when comparing two

languages like French and English, cross-linguistic evidence suggested that NegP could

possibly be in different positions in different languages. Some researchers continued

this line of reasoning and claimed that NegP was not the only functional projection

responsible for negation. This section discusses two works related to these ideas.

Section 4.2.2.1 presents Laka (1990) and discusses her arguments for the addition of

ZP, which is a projection responsible for polarity. 4.2.2.2 discusses Zanuttini's work on

negation in Romance. Both of these researchers present theories that linguists

debating negation in Brazilian Portuguese have turned to.

4.2.2.1 Laka

Laka (1990) continues along the same lines as Pollock, and she looks to account

for the differences between English and Basque. She claims that while there is a

category responsible for negation, it is not fixed and it is not solely for negation.

Consequently, NegP, which she renames ZP, does not necessarily have to be within IP,

but can be above it or below it (Ouhalla 1990). This can first be seen in the variation

that exists among languages with regards to the syntactic position of negation markers,

but as Laka argues, also with regards to positive/emphatic markers because there

exists a certain parallel between negative constructions and emphatic affirmative ones.


105









The first set of English sentences shows that an emphatic positive DID appears to be in

the same structural position as negation.


(19) He DID come (Emphatic)
(20) He didn't come.


Laka then gives more complex evidence that negation and emphatic verbal markers are

syntactically similar. She shows that in affirmative sentences in Basque the auxiliary

verb always follows the main verb, (21), and never the reverse (22), and no element can

intervene between the verbal elements (23).


(21) Etxea erori da
House fallen has
'The house has fallen'

(22) Etxea da erori
House has fallen


(23) Erori etxea da
Fallen house has (Laka 1990: 14, 18-19)


Nevertheless, the presence of the negative sentential marker ez alters this order. The

auxiliary and the negative marker both precede the verb (24) and the necessity of the

auxiliary and the verb to be adjacent is lost (25). In negative sentences, the negative

marker and the auxiliary must be adjacent (26) and (27).


(24) Etxea EZ da erori
house NEG has fallen
'The house hasn't fallen'

(25) EZ da etxea erori
NEG has house fallen


106









(26) *EZ etxea erori da
NEG house fallen has

(27) *EZ etxea da erori
NEG house has fallen (Laka 1990: 25-26)


Examples (28), (29), and (30) show that the same mechanism that causes the negative

marker and the auxiliary verb to be adjacent applies to emphatic elements. If the

preverbal subject is adjacent to the auxiliary verb, it must be emphatic, as in (30) or the

sentence is ungrammatical, as in (29).


(28) Mari ez da joan.
Maria NEG has left
'Maria hasn't left'

(29) *Mari da joan
Mari has left
'Maria has left'

(30) MARI da joan
Maria has left
'MARIA has left'


Given this similarity, Laka proposes that both are instances of a specific functional

category ZP (SigmaP) which represents some sort of "Speech Act". She claims that

IP's position is language specific, contrary to Pollock's claims of a universal negative

locus. Laka presents evidence from English and Basque that argues for two different

positions for ZP. She notes that in Basque where ellipsis occurs, only the negative

marker remains and the auxiliary is deleted (Laka 1990:33).


(31) Marik liburua erosi du eta Peruk ez
Maria book bought has and Peter NEG
'Maria bought the book but Peter didn't'


107










English does not allow the negative marker to remain alone, and must be

accompanied by the auxiliary verb do (Laka 1990:32). In example (32), everything to

the right of the auxiliary is elided. In (33), everything to the right of the negative marker

is elided as well, but in (34), English does not allow ellipsis with just the negative

marker.


(32) Mary didn't buy a book, but Peter did
(33) Mary bought a book and Peter didn't
(34) Mary bought a book and Peter not


Laka argues that this shows a difference in the structural positions of Basque's IP and

English's ZP. Basque's IP is above TP, (35), and in English, IP is below TP, (36).


(35) Basque


(36) English


IP

n't


4.2.2.2 Zanuttini

Zanuttini (1997, 2001) continues along similar lines as Laka by arguing that the

location of NegP is not as rigid as Pollock originally proposed. Her proposal for NegP


108









takes into account the different types of negative markers that exist. Through a

typological approach, she determines that there are four classes of negative markers: (i)

negative adverbs, (ii) strong preverbal negative markers, (iii) weak preverbal negative

markers, and (iv) negative markers that are part of the verbal morphology. She is

principally concerned with the distinction between strong and weak negative markers,

noting that several Romance languages have two negative markers in a sentence. In

the cases of two negative markers, one of the markers is strong and the other is weak.

For her, a strong negative marker is one that can negate the sentence independent of

any other negative element. The weak negative markers must be in conjunction with

some other negative marker. More than one position needs to be posited because

there are two elements. For her, a strong marker is the head of a NegP which she

refers to as Pol(arity)P because the "label PolP suggests that the projection contains not

only negative elements but also markers of emphatic affirmation, which in some cases

can be shown to be in complementary distributions with sentential negative markers"

(Zanuttini (1997:22). The weak negative marker is similar to the position that Pollock

(1989) proposes for pas in NegP. Her analysis involves several factors including

cliticization of the weak preverbal negative marker to the verb; however, If the preverbal

element is weak then it is adjoined to the VP and cliticizes to the verb. If the negative

marker is strong, then it is generated in Neg :


109









(37) NegP


Neg'

Neg AgrP
Strong Marker
Agr'

Agr FP-1

F-1'

F1 VP

Weak Marker F1 FP-2 VP

Furthermore, Zanuttini extends the number of possible positions for negative markers.

Not only can NegP dominate IP or be dominated by IP as Laka proposes, but as Pollock

suggests, it can be within a split IP, and IP can be split into more than just two

categories. Zanuttini claims, based on adverbs and Cinque's hierarchy (1999), there

are at least four positions for negative markers in relation to IP, i.e. one above IP, two

within IP and one below IP.


(38) [NegP1 [TP1 [NegP2 [TP2 [NegP3 [AspPperf [Aspgen/prog [NegP4 ]]]]]]]]


Two of the positions are seen when looking at standard Italian and Romagnolo, a

dialect of Italian. Standard Italian has a strong negative marker, non, that is generated

in NegP1 (just like Spanish no). Romagnolo has a weak preverbal clitic negative

marker that generated below the verb and moves up to the verb and cliticizes to it. The

postverbal negative marker is in one of the other NegP and the verb with the negative

clitic move past it to render the Neg-V Neg order (example (40) from Zanuttini 1997:17).


110









(39) Maria non studia
Mary NEG studies.3s
'Mary doesn't study'

(40) An's dis brisa aksi
SUBJ.CL-NEG.CL say NEG like-that
'one doesn't say it that way'


4.3 Preverbal Negation in BP

The negative marker nao is similar to other preverbal negative markers in

Romance languages, such as Italian and Spanish (Belletti 1990, Laka 1990, Mioto

1992, Zanuttini 1997, Zeijlstra 2004). Following Belletti (1990), Mioto (1992) points out

three options for the structural position of negation in BP: NegP dominates IP (i); NegP

is somewhere within the IP, between flexional nodes (ii); IP dominates NegP (iii).


(41) i. CP [NEGP] [AgrP] [TP] [VP]
ii. CP [AgrP] [NEGP] [TP] [VP]
iii. CP [AgrP] [TP] [NEGP] [VP]


This section looks at these positions and determines where negation falls in BP.

Section 4.3.1 discusses the different phonological forms of the negative marker. This

becomes relevant in chapter five when discussing the rise and reason for clause-final

negation. Sections 4.3.2 and 4.3.3 discuss two previous studies of BP negation. The

proposals of this chapter are merely for preverbal negation. Chapter six continues the

discussion of sentential negation and CFN, combining the information from each

chapter. Following the review of past accounts, I lay out evidence which shows that the

preverbal negative marker in BP is the head of NegP, which is between AgrP and TP (ii

from (41)). To do that I follow some of the previous authors' claims. Namely, as Pollock


111









shows, the NegP is within the IP domain; however, that does not mean that there are

not other syntactic positions for NegP, much like Zanuttini (1997) and Zeijlstra (2004)

show. Rather, BP's negative marker is generated within IP.

4.3.1 The Phonology of the Preverbal Negative Marker

The phonological weight of the negative marker has come into question in BP.

Specifically, some claim that the marker has weakened semantically as well as

phonologically, like the French preverbal negative marker, to the point of needing

postverbal negative emphasis (Schwegler 1991a, da Cunha 1996, Vitral 1999, Martins

1997, Fonseca 2002). In chapter five, I show that although there is phonological

reduction, this has not resulted in any syntactic change in sentential negation. This

section addresses the issue of phonological reduction and presents the relevant data.

This is important because as chapter six shows, only the preverbal negative marker

allows phonological reduction, and the clause-final marker is always the full form nao.

As mentioned, other Romance languages have weak negative markers, usually

clitics, in addition to strong ones. Generally, a language has either a strong negative

marker or a strong and weak negative marker. However, nao in BP has a certain

duplicity about it. For example, Sousa (2007) and Namiuti (2008) both suggest

independently that nao is a clitic, suggesting that it is a weak negative marker like ne.

In spite of its clitic nature, it can negate the sentence by itself much like Spanish and

Italian. Namiuti (2008), in looking at the history of the negative marker in BP, reflects

and discusses the non-clitic/clitic nature of the negative marker in BP. Sousa (2007)

uses a corpus study to examine the different phonological forms of nao. She concludes


112









that there are essentially five forms with the reduced form nu [nu] and num [nu], both

clitics being the most common:


(42) Sousa (2007:57)

Nao Num Nu u N' Total

130 301 454 15 18 918



Ndo is often considered a strong negative marker, meaning that it is not a clitic (Mioto

1992, Martins 1994, Sousa 2007). The other variants of the negative marker are

assumed to be clitics. Mioto claims that the clitic variants are not allowed to stand

alone, as in the response to a direction question, in example (43).


(43) Q: 0 Jo0o agrediu o Pedro?
DET John attacked DET Peter
'Did John attack Peter'

A: Nao/*num, nu, u, n'


The idea that the preverbal marker is a 'weak' clitic supports many linguists' claims

about the status of the preverbal marker and CFN. However, recall from chapter one

that T1 negation is the overwhelming preferred choice for negation. Combine that

information with the data from Sousa which shows that the clitic negative markers are

the most common forms for the preverbal negative marker, and an argument against

the 'weak' preverbal negative marker can be made. Namely, although there has been

phonological reduction, the preverbal negative marker still carries the syntactic and

semantic properties to negate a sentence. This topic is continued in chapters five and

six where I discuss previous theories of CFN.


113









4.3.2 Mioto 1992 & 1998

Mioto claims that NegP is above the IP domain. Before giving linguistic evidence

for his claim, he begins by asserting that NegP is a [+l]nflectional because of its

relationship to IP which is not split according to him. This relationship is based on the

fact that nao has the ability to change the truth value of a sentence, i.e. cause a

proposition p to become -p. He also claims that the unbreakable nao+clitic+verb order

is based on movement of lower inflectional items to Neg. Thus, the structure that he

proposes is:


(44) NegP

Subject Neg'

Neg IP
n~o
ti I'

I VP


Mioto argues that his structure comes from three principal pieces of evidence. The first

piece of evidence is related to what Laka (1990) claimed for Basque and Spanish, i.e.

these languages license a null IP in coordinated sentences such as (45). In this

example, nao is the head of a phrase, and everything below is deleted. Given this

example, Mioto claims that NegP is a functional category, as it allows ellipsis (Lobeck

1995). Recall that he claims that IP is not split as Pollock claims.


(45) A Sienna viu a JQlia mas a LQcia nao [viu a JQlia]
DET Sienna saw.3s DET Julia but DET Lucia NEG [saw. 3S DET Julia]
'Sienna saw Julia but Lucia didn't [see Julia]


114









The second reason for placing NegP above IP is to properly understand the clitic

nature of the negative marker. Mioto claims that only the structure NegP dominating IP

is descriptively accurate in capturing the duplicity of the clitic nature of nao. He

continues and claims that if NegP were within IP, "the behavior of nao in BP would not

permit placement of it into the class of clitics that can take care of an empty IP" (Namiuti

2008:167). Therefore, BP allows sentences where nao is not deleted in the TP.14

Where the clitic nao is 'attached' to the verb in a lower position, then a sentence like

(46) would not be allowed because in this sentence the negative marker would have to

be deleted along with the verb.


(46) A Sienna viu a JQlia mas nao o Abilio
DET Sienna saw.3s DET Julia but NEG DET Ablilio
'Sienna saw Julia but didn't (see) Ablilio'


The third reason that Mioto claims NegP is above IP is because of the difficulty of

inserting a negative adverb into the structure, since the verb moves at least to TP.

According to him, if NegP is above IP, then it becomes easier to understand why a

sentence like (47) is grammatical. In this sentence, he claims that the negative nunca

'never' is above IP and is allowed. This is because, if the negative adverbial nunca is

to the right of the verb the sentence becomes degraded, as seen in example (48).

(47) A Sienna nunca chora
DET Sienna never cries.3s
'Sienna never cries'

(48) ??A Sienna come nunca cenouras
DET Sienna eat.3s never carrots
'Sienna never eats carrots'

14 This is allowed presumably because movement is adjunction, and the adjoined verb would be deleted
leaving the object, below, intact.


115










A problem with Mioto's proposal is the subject position. According to him, the

subject in affirmative and negative sentences has two different positions. In the

affirmative sentence it is in spec, IP, and in negative sentences, the subject is in

spec,NegP. However, this problem could be resolved by using Laka's IP which houses

both negative and in most cases, abstract positive markers.

Another problem with his analysis comes with sentences that make use of modal

auxiliary verb poder 'to can/be able to/might'. The auxiliary is often overlooked in its

dual role. This is seen in the examples below. The positive sentences appear the

same, but do have different meanings depending on the context.


(49) Ele pode fazer o trabalho
3s can do. INF DET work.
'He can/is able do the work'

(50) Ele pode fazer o trabalho
3s can do.INF DET work.
'He might do the work'


The difference of syntax and meaning is clearer when the negative marker is added to

the sentences. In the first case, the sentence is stating that the person is not capable of

doing the work. The second sentence by having the modal verb to the left of the

negative marker means that the person might not do the work, although it might be the

case that he is capable of doing the work. In fact, as example (53) shows, the auxiliary

poder can be combined with poder 'to be able to'.


(51) Ele nao pode fazer o trabalho
3s NEG can do. INF DET work.
'He can't/ is unable do the work'


116









(52) Ele pode nao fazer o trabalho
3s can NEG do.INF DET work.
'He might not do the work'

(53) Ele pode nao poder fazer o trabalho
3s can NEG BE.ABLE.INF do.INF DET work.
'He might not be able to do the work'

The examples show that it is not always the case that the negative marker

precedes the verb and its auxiliaries. Thus, although it was not mentioned in chapter

three, IP supposedly can be split into several projections. If this is the case and if NegP

is within the IP, then sentences like (52) exist where the negative marker is not directly

after the subject, as Mioto claims, but is split by the insertion of a auxiliary.

4.3.3 Martins 1994 and Namiuti 2008

Martins (1994) and Namiuti (2008) follow Mioto in stating that negation is above

IP. However, Martins claims that NegP is not a correct denomination for what happens

in BP. Rather, she chooses to follow Laka (1990), and concludes that IP is the correct

nomenclature for BP. Her main objection, much like an objection in the previous

section is what to do with the subject. Mioto's proposal argues for two different subject

positions, and for Laka and Martins, this is undesirable. Martins' basic structure is seen

in example (54). Notice that in her structure, as well as Namiuti's, the negative marker

is generated in NegP, which is in the IP domain. The negative marker then moves to

EP to check its 'strong negative feature' (more on this in chapter six).


117









(54) jP

Subject

Y AgrP
naoii
ti Agr'

Agr NegP

Neg TP
tii
ti T'

T VP


By having the negative marker originate in NegP, within the IP domain, Martins

(1994) and Namiuti (2008) claim the structure and movement are able to properly

account for the clitic nature of the negative marker and its interaction with pronominal

clitics. By way of example, Martins (55) shows the order of neg-clitic-verb, and (56)

shows that that pronominal clitics must be adjacent to the verb. If, as Martins claims,

the negative marker must move to Y, then this would always prevent the pronominal

clitic from being separated from the verb.


(55) A Sienna nao te viu.
DET Sienna NEG CL.2S saw.3s
'Sienna didn't see you'

(56) *A Sienna te nao viu
DET Sienna CL.2S NEG saw.3s
'Sienna didn't see you'


Additionally, the movement of the negative marker explains the ungrammaticality of

sentences like (58). (57) shows that a negative marker must be adjacent to the verb


118









and that nothing else besides clitic pronouns can intervene. The adverbial expression

na noite passada 'last night' is not allowed to intervene, (58).


(57) A Sienna na noite passada nao quis dormir
DET Sienna PREP.DET night last NEG wanted.3s sleep.INF
'Sienna last night didn't want to sleep.'

(58) *A Sienna nao na noite passada quis dormir
NEG Sienna NEG PREP.DET night last wanted.3s sleep. INF
'Sienna last night didn't want to sleep.'


Much like Laka's claims for English and Basque, Martins claims that there is

evidence for a positive-type marker in Portuguese. In Portuguese the affirmation or

positive indicator of the existence of I comes from verbal responses to direct questions.

Martins argues that in English did in an emphatic sentence is the head of ZP. She then

continues by positing verb movement from I to Y. This supposedly always occurs, but is

more apparent, according to the author, with direct questions. To answer a direct

question in Portuguese affirmatively, sim 'yes' is not used; instead, the verb is used.

Martins says that this behavior supports the idea of the verb rising to the head of IP to

check its positive features. As an example, (59) is asked. The felicitous responses to

this are (60) and (61). (62) has the representative tree for (60).


(59) Did you go?

(60) Fui
Went. 1s
'I went'/'yes'

(61) N o fui.
NEG went. 1s
'I didn't go'


119










(62) jP



Y AgrP
fuii
Agr'

Agr TP

T'

T
ti

However, it has been shown elsewhere that this type of answer is not necessarily

related to iP (Santos 2004), but rather to some topic position in the CP or deletion of

VP (Schwenter 2005). Martins (1997) bases her arguments on evidence from post-

clausal negation in BP that the verb moves from V to Top.

While Martins might be right in her objections to Mioto's multiple subject positions,

there does not seem to be any reason to not allow more than one subject position in BP

(Costa 2001 posits several positions for the subject in European Portuguese). In

addition to this, as shown in the previous section, there are cases where the negative

marker appears to not move to I because the subject and the negative marker can be

separated by a modal auxiliary like poder.

4.3.4 The structural Position of NegP in BP

I now take the information given in the previous sections and apply it to the facts

concerning BP negation. It is crucial that I lay out the exact position of the negative

marker in a BP sentence because the subsequent chapters use this information as a


120









basis for understanding clause-final negation and negative concord. Below is a review

of the facts from above. The negative marker in BP is:


i. preverbal
ii. adjacent to the verb and only separated by pronominal clitics
iii. a functional category that allows ellipsis


Other relevant facts discussed in chapter three are:


iv. the subject is in spec,AgrP
v. the verb only moves to T (i.e. short movement)


First, as mentioned, Mioto's structure is problematic due to two subject positions.

This is remedied by Martins' proposal to use ZP, which simplifies the grammar by

postulating a single subject position; however, Martins then proposes two negation

positions. So, although the subject has been simplified, negation requires movement.

Given these considerations and iv and v above, I conclude that nao is the head of

NegP, which appears between AgrP and TP:

(63) AgrP

subject Agr'

Agr NegP

Neg TP
no
T vP
verb


This structure correctly places negation in a preverbal position without any movement of

the negative element. It will license ellipsis by having TP elide.


121









This structure can also account for the data with modal auxiliaries and the

negative marker seen above. Example (64) is a repeat from above and this example

has the modal to the left of the negative marker. This sentence has the meaning of

'might not', and it is not an example of constituent negation. Constituent negation in this

case would require stressing the negative marker and would render the meaning 'he

can NOT do the work'. The syntactic tree structure is given in (65).

(64) Ele pode nao fazer o trabalho
3s can NEG do.INF DET work.
'He might not do the work'

(65) AgrP

Elei Agr'

Agr XP

X'

X NegP
pode
Neg TP
n~o
ti fazer o trabalho


Assuming a functional head position for the modal above TP and NegP but below AgrP

within a split IP (Costa & Galves 2002) correctly allows this sentence.

4.4 Conclusion

The principal goal of this chapter was to review some ideas about syntactic

negation and to examine the preverbal negative marker in BP. I discussed general

accounts to understanding the syntax of negation, and the syntax of negation in BP. I

have shown that nao is a syntactic head that is generated as the head of NegP and that

NegP is between AgrP and TP in BP. Much like Zeijlstra (2004), Laka (1990), and


122









others, I am not claiming the NegP must necessarily be the same in all languages, but

rather, there are language specific positions.


123









CHAPTER 5
THE PRAGMATICS AND SEMANTICS OF CLAUSE-FINAL NEGATION

5.1 Introduction

This chapter analyzes semantic and pragmatic restrictions on the use of clause-

final negation in Brazilian Portuguese (BP). The goal is to properly account for the

felicity of sentences that contain clause-final negation (CFN) within a

semantic/pragmatic framework. Ideally, this chapter sets up a theoretical base from

which to understand the syntactic structural facts of CFN in BP.

As mentioned in chapter one, the following examples are instances of CFN.15 (1)

shows that the clause-final nao can cooccur with the preverbal negative marker nao.

This type of construction is referred to in the literature as T2. (2), also a version of T2,

has some other negative item, a so-called n-word, in a preverbal position. Finally,

different from the first two examples is (3) T3, where there is no preverbal negative

marker.


(1) A LQ nao come almogo em casa nao. T2
DET Lu NEG eat.3s lunch PREP house NEG
'Lu doesn't eat lunch at home'

(2) 0 Cesar nunca faz o trabalho nao. T2
DET Cesar never do.3s DET work NEG
'Cesar never does the work'

(3) A Sienna quer comer nao T3
DET Sienna want.3s eat.INF NEG
'Sienna doesn't want to eat'

Section 5.2 discusses three previous approaches to understanding the semantic

and pragmatic restrictions on CFN: section 5.2.1 looks at emphasis (Roncarti 1996,

15 Because preverbal negation was treated in chapter four, I do not specifically address it here; however it
does play a role in understanding CFN. This idea is pursued in chapter six.


124









Barme 2000); section 5.2.2 investigates presupposition or contrary to fact (Schwegler

1991a, 1996; da Cunha 1996, 2001, 2007), and Section 5.2.3 looks at discourse status

(Prince 1992, Schwenter 2005, 2006, Armstrong 2008). In each case, I present a

theory and show how it has been applied to BP. I ultimately choose Schwenter's (2005)

discourse status or discourse-old proposal for CFN in BP, and I show in section 5.3 that

this accounts for several un-elicited sentences, as well as ruling out infelicitous

sentences.

5.2 Previous Approaches to Clause-Final Negation: Semantics/Pragmatics

5.2.1 Emphasis

The idea that the CFN is emphatic comes from relating BP to the process known

as the Jespersen Cycle (JC) (Roncarti 1996). Simply put, the JC is a cycle that involves

a change that takes place with the sentential negative marker changing its syntactic

position, usually from a preverbal to a postverbal position, or vice versa. In many

cases, JC is observed when a preverbal negative marker loses its inherent negative

semantic force, and often this loss of negative force is related to phonological reduction.

This loss of force motivates the rise of a postverbal negative element to reemphasize

negation in the sentence. French is a classic example of this, in which the postverbal

emphatic marker becomes the sole negative element in the sentence, and the original

preverbal maker disappears. Below, each French16 example is followed by a BP


16 Zeijlstra (2004) gives a more complete breakdown of the stages that make up the Jespersen Cycle:

Phase I: Negation is only expressed by a single negative marker that is attached to the finite verb.
Phase II: The negative marker that is attached to the finite verb becomes phonologically too weak to
express negation by itself and a second negative adverb becomes optionally available.
Phase III: Sentential negation is obligatory expressed by the negative marker that is attached to the finite
verb and the adverbial negative marker.


125









sentence that is assumed to showed the JC in BP (Vitral 1999). The sentences all

mean: 'I don't know'.


(4) a. je ne sais stage 1: verbal negative marker (NM)
a'. eu nao sei

b. je ne sais pas stage 2: preverbal NM & postverbal emphatic adverb
b'. eu num sei nao

c. je ne sais pas stage 3: preverbal NM & obligatory postverbal NM
c'. eu num sei nao

d. je sais pas stage 4: lost of preverbal NM & only postverbal NM
d'. eu sei nao


Recall from chapter four, the preverbal NM has weakened phonologically from

[naw] to [nu]/[no]/[n] (Roncarti 1996, Vitral 1999, Sousa 2007). Roncarti claims that this

weakening, as in French, has caused the addition of a postverbal negative marker to

appear (see Lobato 1986 for an early discussion of this). Roncarti claims that CFN is

emphatic because it does not cause a double negation reading, just as the addition of

pas in French does not give a double negation reading. That is, CFN does not appear

to interact semantically with the preverbal negative marker. As such, the sentence ele

nao comeu o pao nao does NOT mean that 'he didn't not eat the bread'. Taking the

example sentences from (4), according to those that take BP negation to be changing



Phase IV: The negative adverb is the obligatory marker for negation and the use of the negative marker
that is attached to the finite verb becomes optional.
Phase V: The negative adverb is the only available negative marker. The negative marker that is attached
to the finite verb is no longer available.
Phase VI: The negative marker is available in two forms: it can appear either as negative adverb or as a
negative marker that is attached on the finite verb, though sometimes simultaneously.
Phase VII: Negation is only expressed by a single negative marker that is attached to the finite verb.


126









within the context of the JC, BP is moving from stage 2, where the clause-final nao is

emphatic, to stage 3, where the nao is obligatory (Schwegler 1988, 1991a). Not only

does this appear to be the JC, there is evidence from the positive marker that this is a

case of an emphatic element. Example (5) shows that the positive marker sim may

appear in the same clause-final position as the negative marker. The argument here is

that this clause-final position in Portuguese is used for emphasis (Barme 2000).


(5) Ele comeu o pao sim
3s eat. PST.3S DET bread yes
'He did eat the bread'


In addition to this, CFN is seen as emphatic or intensifying because of its ability to occur

with preverbal negative words like nunca 'never' (Barme 2000). Nunca is sufficient to

negate a sentence without additional support, as (6) shows.


(6) Ela nunca come pao
3s.F never eat.3s bread
'She never eats bread'


The argument is that nunca is not phonetically weak and, by extension, not

semantically or syntactically weak, and therefore needs no reinforcement. Yet, CFN is

allowed in situations with nunca. This follows from Schwegler's claim that there is a

"constant and universal psycholinguistic need for negative emphasizers" (1988: 36),

which is why example (7) has CFN.


(7) Ela nunca come pao nao.
3s.F never eat.3s bread NEG
'She never eats bread'


127









There are problems with this analysis, however. First, this line of thinking hinges on the

idea that the preverbal negative marker is weakening to the point of needing some sort

of support, as in French. The data reviewed in chapter one regarding usages of the

preverbal marker suggest that the preverbal marker is not weak, i.e. roughly 75% of

negative sentence rely solely on preverbal negation. If the preverbal negative marker

were weak, then the use of type one negation would most likely be lower because it

would need to be supported by some emphatic negative element.

Second, in previous attempts to understand CFN in terms of emphasis,

researchers have failed to define the term adequately. As Schwenter notes, emphasis is

not properly defined by those (i.e. Barme, Roncarti) that use it to describe BP negation

(Schwenter 2005, 2006). In many cases, there are two types of emphasis that are

being discussed: i) reinforcement a weak negative marker; ii) intensification of the

negative force of the sentence.

On the one hand, emphasis is used as a reinforcing tool for negative markers that

have phonologically or semantically weakened. This view stems from a particular

conception of the Jespersen Cycle and the desire to assimilate the BP data with

previously studied cases of negation within the Romance family. However, since it has

been shown in chapter one that T1 is still overwhelmingly the strategy used for

negation, it is difficult to argue that the postverbal negative marker is needed to negate

a sentence. The purported necessity of a postverbal marker entails that preverbal

negative markers are no longer strong enough to negate a sentence on their own, which

is not the case (Cavalcante 2007).


128









The other use of the term is a more traditional one adopted by Schwenter,

following Israel (1998): emphasis in language is "informative because it exceeds what

one would normally expect to be asserted" (1998: 47). This means that a sentence like

he didn't do a lick of work could entail that he didn't do work but not vice versa. This is

because a lick of exceeds the normal expectations of not working in this case. A typical

strategy used for emphasis in Portuguese is through the use of mesmo 'same'. In the

sentence below, mesmo indicates an emphatic meaning towards the positive or

negative semantic meaning. Notice that in both cases, the emphasis adds to the

meaning and suggests that the person worked (or didn't work) beyond the normal

expectations i.e. in the first case she did all the work, even though she didn't have to,

and in the second case, she didn't do anything, not even pick up one piece of trash from

the ground.


(8) Ela trabalhou mesmo.
3s.F work.PST.3s EMPH
'She REALLY worked'

(9) Ela nao trabalhou mesmo.
3s.F NEG work.PST.3S EMPH
'She didn't work AT ALL'


By taking the definition that Israel gives and applying it to CFN, one can see that

CFN is not a case of emphasis. See example (10) below. In this example, B is

negating the accusation, and B goes one step farther to emphasize the fact that he

didn't even break the cup. Use of the CFN is not felicitous in this instance. Also, the

negative markers in both parts of the sentence could have an emphatic stress, but the


129









clause-final negative marker can neither be stressed nor can it even appear in this

situation.


(10) A. Voc6 derrubou a agua.
2s spill.PST.2S DET water
'You spilled the water'

B. Eu nao derrubei a agua, e nao quebei o copo (#nao)
1 S NEG spill.PST.1S DET water and NEG break.PST.1s DET cup NEG
'I didn't spill the water, and I didn't break the cup'


Finally, as Schwenter notes, even accepting that T2 negation is emphatic, there still

exists the problem of T3 negation. In this case, there is no preverbal negative marker,

as such it is difficult to understand how this type of negation is emphasizing the

sentential negation, because this is not present.

5.2.2 Presuppositional or Contrary-to-Fact

Another approach to BP negation is one that takes CFN to be linked to

presuppositions. This claim originally comes from Schwegler (1991a) and is similar to

what da Cunha currently proposes (1996, 2007). Schwegler argues that CFN is used in

sentences that "presuppose a previous affirmative assertion or assumption which [it]

seek[s] to contradict."17 (1991a: 194) That is to say, BP speakers use CFN when they

want to contradict some presupposition. For the purposes here, to presuppose

something means to assume something (Saeed 1997), or a presupposition is a set of

assumptions made by participants in conversation (Stalnaker 1978). Accordingly, a

sentence 'your son is funny' presupposes that you have a son. Similarly, if (11) below is



17 This is also the analysis that Zanuttini (1997) gives for several marked negative forms in some Italian
dialects.


130









uttered, Schwegler's claim argues that there is a presupposition 'He lies to his parents'

that this sentence looks to deny or contradict.


(11) Ele nao mentiu para os pais dele nao.
3s NEG lie.PST.3S PREP DET parents prep.3S NEG
'He didn't lie to his parents'

Schwegler gives two examples to show how CFN is used to contradict a

presupposition. He claims that the A sentences below do not presuppose and therefore

"do not challenge any previous statement assumption or implication on the matter

involved, they simply assert a proposition", and the B sentences "presuppose a previous

affirmative assertion or assumption which they seek to contradict"(1991a: 194).


(12)


A. Quando estive no Rio, nao fui na praia.
When was.1sg PREP.DET Rio, NEG went. 1s PREP.DET beach
'When I was in Rio, I didn't go to the beach'

B. Quando estive no Rio, nao fui na praia nao.
When was.lsg PREP.DET Rio, NEG went.1s PREP.DET beach NEG
'When I was in Rio, I didn't go to the beach
(contrary to others or my own beliefs)'


(13) A. O Brasil nao e um pais rico
DET Brasil NEG is DET country NEG
'Brazil isn't a rich country'

B. O Brasil nao e um pais rico nao.
DET Brasil NEG is DET country rich NEG
'Brazil isn't a rich country (contrary to others or my own beliefs)'


This analysis is weak for two reasons. First, if the presupposition is already negative,

then the use of CFN yield a double negation reading according to Schwegler's own

explanation. In the example below, A gives her beliefs that B didn't do the work.

According to Schwegler, if CFN were to be used, it would try to contradict the


131









presupposition, i.e. that B didn't do it. This is not the case, however. Instead, CFN is

used in sentences that agree with the presupposition, and with the sentence in general.


(14) A. Eu sei que voc6 noo fez o trabalho
1s know.ls COMP 2s NEG do.PST.2S DET work
'I know that you didn't do the work'

B. Eu nao fiz o trabalho nao
1S NEG do.PST.1S DET work NEG
'I didn't do the work'


The second problem with this analysis, as mentioned in Schwenter (2005), is that

there are cases of where the use of CFN does not contradict an assumption or

presupposition. One case where presupposition does not work is in presuppositional

denials. Presuppositional denials are denials where negation is interpreted as applying

to some presupposition of a prior utterance, not to its asserted content. In these cases,

CFN is not possible. The examples below show this to be the case (Schwenter 2005).

The presupposition of the A example is that 'John smoked in the past', and this is

contrasted with the assertion, that 'John has stopped smoking' (Stalnaker 1978, 2002).

According to Schwegler's analysis, CFN should be licit in the B example because it

negates the presupposition 'John smoked'. This is not the case, as CFN is not allowed.


(15) A: 0 Joao ja deixou de fumar.
DET Jo0o already stop.PST.3S PREP smoke.INF
'John has stopped smoking'

B: Ele nunca fumou (#nao).
3s never smoke.PST.3S NEG
'He never smoked'


132









This pair of sentences contrasts with the pair in (16). CFN is not negating the

presupposition but rather the assertion that John has stopped smoking. Evidence that

the sentence is not negating the presupposition is seen in the second half of the

sentence, where B agrees with A by claiming that 'John still smokes' which also has as

a presupposition 'John smoked in the past'.


(16) A: O Joao ja deixou de fumar.
DET Jo0o already stop.PST.3S PREP smoke.INF
'John has stopped smoking'

B: Ele nao deixou de fumar (nao), ele ainda fuma.
3s NEG stopped PREP smoke.INF (NEG), 3s still smokes.
'He hasn't stopped smoking, he still smokes'


This section has shown that although there are claims that CFN is used to

contradict a presupposition, this is not what allows CFN to be licit in BP. Rather, the

data given has shown that CFN is related to what is asserted in the discourse and not

necessarily to what is presupposed.

5.2.3 Discourse-Old

I now lay out the discourse theory that is assumed throughout the remainder of

this work. Schwenter's 2005 proposal claims that the discourse information structure

developed by Prince (1981, 1992) is sufficient to explain the data for BP CFN. After

presenting Prince's ideas, I present Schwenter's analysis of CFN in BP. I show that his

application of Prince's theory more broadly captures the semantic/pragmatic restrictions

on CFN in BP.

Prince's goal is to properly describe the information structure of discourse. She

has a far reaching definition of discourse that not only refers to the things said but also

to the extra-linguistic elements, such as the context (Prince 1981). This means that


133









discourse can be interpreted very broadly and include many facts and ideas that aid in

the interpretation of the linguistic utterances. Prince also separates discourse into two

categories: new information and old information (Prince 1981, Horn 1989). She takes

this separation one step further and distinguishes between hearer information and

discourse information. That is to say, something can be old/new with regards to the

hearer, and something can be old/new in relationship to the discourse. As for the status

of the information with relationship to the hearer, Prince states that "information, by

which is here generally meant entities/referents, may be old/new with respect to (the

speaker's beliefs about) the hearer's beliefs" (1992:5 emphasis mine).18 Essentially,

Prince tries to show that hearer-status (i.e. the hearer's beliefs, etc.) can be "partially"

independent from the discourse-status; however, the inverse is not true. If an entity is in

the discourse, then it is necessarily within the realm of the hearer's knowledge, belief,

etc. To explain the point, Prince gives the following examples with an accompanying

chart in (19) (Prince 1992:12).


(17) a. I'm waiting for it to be noon so I can call someone in California.
b. I figure she'll be up by 9, her time.

(18) a. I'm waiting for it to be noon so I can call Sandy Thompson.
a. I figure Sandylshe'll be up by 9, her time.


In (17), someone is a new entity to the discourse and it is new to the hearer;

however, in the case of (18), Sandy Thompson is new to the discourse, but it is not new

to the hearer, as it is assumed from this example that the hearer knows who Sandy

Thompson is. (17) and (18) both have she, which in these sentences is now old to the

18 The hearer's beliefs might differ from what the speaker believes about the hearer's beliefs, but this is
not relevant for BP CFN, as such this will not be pursued.


134









discourse and old to the hearer. The chart in (19) shows Prince's organization of

discourse information. Notice that if something is discourse old it would be impossible

for it to be new to the hearer; thus, the Hearer New/Discourse Old cell of the table is

blank.

(19) Hearer- and Discourse-status of a discourse entity:

Discourse New Discourse Old
Hearer New Brand new:
(17)a. someone

Hearer Old Unused Evoked:
(18)a. Sandy Thompson (17)b. she
(18)b. Sandy/she


In addition to things mentioned in the discourse, Prince claims that NPs can be

discourse old if they are inferable from the context. She gives the following example

which mentions the Bastille. Because the Bastille is an edifice, then it is inferable that

the Bastille has a door. Therefore, B is felicitous because the context lets the hearer

infer the door (Prince 1992:8).


(20) A. He passed by the door of the Bastille and the door was painted purple.
B. He passed by the Bastille and the door was painted purple.

By the same token, example (21) is not felicitous because it is not inferable from

mentioning the Bastille that there is the trunk. This example shows that the discourse-

old requirement is not activated because the trunk is neither old to the hearer nor old to

the discourse.


(21) #He passed by the Bastille and the trunk was painted purple.


135









It is important that the ideas of presupposed information and discourse-information

are distinct, even though they might at times intersect. The question can be asked:

How is discourse-old different from presupposed information? To answer this, recall

from the previous section what a presupposition is. Presupposition is related to the

beliefs or assumptions related to assertions. The crucial point is that a presupposition

may or may not enter the discourse. If it does not enter the discourse, it cannot be

discourse-old according to Prince's definition. Example (22), repeated from above,

shows the difference between presupposition and discourse-old. In case of sentence A,

the presupposition is 'John smoked in the past' and the discourse-old assertion is 'John

has stopped smoking'. Sentence B denies the presupposition 'John smoked in the

past', but it does not deny the discourse-old assertion that 'John stopped smoking'.

Sentence B' denies the discourse-old assertion 'John stopped smoking' and makes no

direct claim about the presupposition 'John smoked in the past'.


(22) A: 0 Joao ja deixou de fumar.
DET Jo0o already stop.PST.3S PREP smoke.INF
'John has stopped smoking'

B: Ele nunca fumou
3s never smoke.PST.3s
'He never smoked.'

B'. Ele nao deixou de fumar
3S NEG stop.PST.3S PREP smoke.INF
'He didn't stop smoking'

There is still another problem that lingers. Prince mentions the fact that discourse-

information can be old if it is inferable. If something is inferable, then most likely that

inference is due to assumptions that the speakers hold. In the case above, people

presuppose that Bastilles have doors, and this presupposition is based on the fact that


136









people assume that buildings have doors. This is where the discussion encounters

problems with inferable information, or better yet some presuppositions. CFN allows

some inferable information to be negated using CFN, but there are instances where

CFN is not allowed. The reason for this is related to a quasi-scale of grading for how

readily inferable the information is. If the information is readily inferable, doors on a

building, then it is allowed. However, if the information is not readily inferable, like

chairs in a house, then CFN is not allowed. More on this is seen in examples (39) and

(40). For now, it is sufficient to say that CFN is used to deny or negate presuppositions,

but it is not allowed with all presupposition, thus giving the advantage to the discourse-

information theory outlined for BP in the following paragraphs.

Schwenter attempts to extend Prince's analysis of NPs to whole propositions that

are negated. He claims that CFN is licit when what is being negated is discourse-old,

as in Prince's examples (17) and (18). Additionally, the discourse-old 'feature' can be

on any part of the sentence. Part of Schwenter's goal is to show that emphasis and

presupposition are not proper descriptions of CFN in BP. He separates his proposal

from Schwegler and da Cunha's proposals by separating out hearer-old information and

discourse-old information. Recall that discourse-old and hearer-old do not mean the

same thing. That is, a proposition can be hearer-old, this is like a presupposition, but

not discourse-old, i.e. present in the current discourse. It is the discourse-old status

of/within the proposition that licenses the felicity of a sentence with CFN. For a

proposition to be discourse-old, it would necessarily need to be uttered or inferred from

the context, such as a Bastille having a door given in Prince's examples above.


137









Schwenter extends Prince's examples, which generally include only NPs, to larger

pieces of the sentence. However, he doesn't fully define what a discourse-old

proposition is. As such, for the purposes of this work, I define a proposition somewhat

broadly because it has to include unspoken discourse as well as larger phrases such as

VP/AgrP/etc... To give a complete definition of proposition for this work, I use the

sentence below as an example:


(23) Sienna ate bean soup.


For a sentence like the one above, the proposition is Sienna ate bean soup. It is a

proposition because it has a truth-value that can be calculated. Following Prince's

proposals, if Sienna ate bean soup is uttered then it, the whole proposition, is discourse-

old. Consequently, I call these propositions discourse-old propositions. Also, if the

proposition (23) is uttered, then it has been directly activated in the discourse and it is

discourse-old.

BP CFN extends the meaning of discourse-old propositions to include phrases

and sentences inferred from the initial proposition. Inferred is the word that Prince uses

but here I also say that BP CFN is connected to propositions that are entailed from the

initial proposition. Propositions that are inferred or entailed from the initial proposition

are considered to be indirectly activated in the discourse. However, although not

directly activated in the discourse, they are discourse-old, much like the door in Prince's

Bastille example.

Indirectly activated discourse-old propositions, as mentioned above, are related

to what can be entailed from the initial proposition. Below are examples of inferred


138









discourse-old propositions. First, sentence (23) above entails sentences (24) and (25),

and in an even more abstract way (26):

(24) Sienna ate something

(25) Someone ate bean soup.

(26) Someone ate something.

Using a quantifier like something potentially raises the number of possibilities almost

infinitely. Thus, for a discourse-old proposition, the something must be 'related' to the

original lexical idea. For example, it would not be possible to infer that Sienna ate a

kitchen table, but this is not related to bean soup or even food. Again, this is much like

Prince's example of the Bastille door, which is inferable from the discourse-old NP the

Bastille. That is, something is 'related' if it is interpreted as part of some item in the

proposition. The door on the Bastille is related because it is interpreted that a Bastille

has door, i.e. it is an inherent quality of Bastilles to have doors. The idea of a chest in a

Bastille cannot be inferred from the Bastille because a chest is not interpreted as part of

the Bastille.

Also, much like Prince's theory, other variables, especially the postverbal ones, if

not explicitly stated can be inferable. One can infer from the proposition that not only

Sienna ate soup but also:


(27) Sienna ate soup with a spoon.


Other prepositional phrases or adverbial phrases 'related' to the original proposition are

also inferable propositions that originated from the discourse-old proposition.


139









Schwenter illustrates the discourse-old requirement for CFN in BP with examples

(28) and (29) (2005:1434). (28) shows that if the proposition, Someone turn off the

stove', has not been introduced into the discourse, then the final nao is infelicitous.


[A speaker is walking down the street talking to her friend about a recent soccer
game. Suddenly she remembers that she forgot to turn off the stove and says:]


(28) Nossa! Eu nao desliguei o fogao (#nao)!
Ours! 1s NEG turn.off.PST.1S DET stove (#NEG)
'Damn! I didn't turn off the stove!'


Notice the difference in the next example where the proposition 'you turn off the stove'

is introduced into the discourse through a simple dialogue.


[same situation as (28), the friend (a) asks (b):]


(29) a. Voc6 desligou o fog5o, ne?
2SG turned.off.2s DET stove, right
'You turned off the stove, right?'

b. Nossa! (Nao) desliguei nao!
Ours! NEG turn.off.PST.1s NEG
'Damn! I didn't turn it off!'

Recall that according to Prince's definition of discourse that the information does

not necessarily have to be introduced verbally. If there is something in the conversation

that has been introduced either verbally or not, then CFN is felicitous. In this case, the

idea of 'dropping the papers' or 'that you dropped the papers', which is reduced to the

pro-from isso, is what activates the CFN. It can be inferred that dropping the papers or

doing something wrong or bad is generally not good, and when people do something

bad they feel bad. Again, this is like Prince's Bastille-door example. The same can be


140









said with doing an act that is bad, where it can be inferred that people feel bad because

of a certain act.


[A Woman drops a stack of papers while working with her friend. The women,
referring to the event (dropping the papers) says:]

(30) Esquenta por causa disso nao19
Heat.up.IMP.2s PREP cause PREP.DEM NEG
'Don't worry about it'


Now note, example (31) shows that even if the discourse is in progress, if the

information is not present (discourse-old), then CFN is not felicitous. This is because in

this case the proposition of 'going to the beach' is not introduced into the discourse, and

as such is not discourse-old.


[A conversation about a recent trip to Rio de Janeiro]

(31) A. O que voce nao fez no Rio que queria fazer?
DET what 2s NEG did.2s PREP.DET Rio that want.IMP.2s do.INF
'What didn't you do in Rio that you wanted to do?

B. Eu nao fui a praia (#nao).
1S NEG went.1S PREP.DET beach (NEG)
'I didn't go to the beach.'


This conversation could be imagined in a different way, where the two friends spoke

before the trip about the plans that B has made, one of which was going to a beach like

Copacabana. Now that the trip is over, A asks the question. Even though the

information is hearer-old, it is not licit to use CFN because it is not discourse-old, i.e. not

present in the discourse.

19 i.e. i. Esquenta por causa do fato de voc6 ter deixado cair a folhas nao.

'Don't worry about the fact that you let the papers fall'


141









Schwenter also makes a distinction between T2 and T3, however in both cases,

the motivation for CFN is related the discourse-information status of the proposition.

His initial questions about this arise due to the actual percentage of usage found in

corpus studies. For example, in da Cunha's study and spoken corpus from

northeastern Brazil, there are only nine examples of T3 out of a total of 1465 negatives,

a mere .6% (1996, 2001, 2007). Keep in mind that generally it is assumed that the

dialects of northeastern Brazil are where CFN occurs most (Schwenter 2005).

Roncarti's (1996) corpus had only 39 tokens out of 813, 4.7%. In each case, T2 has

much higher rates of use, 10.8% in da Cunha and 17.9% in Roncarti; although, these

pale in comparison to T1, which is always about or above 75%. Additionally, these low

numbers for T3 almost always included repetition of the same verb previously

mentioned in the discourse with no object present. Roncarti's study shows that 72%,

28/39 of all the T3 sentences, were similar to the next example, where T3 occurs in

response to a direct question and the same verb is used. Roncarti also states that the

other tokens, 11/39, were the expression sei nao 'I don't know' which is considered by

the author to be a lexicalized form due to its similarity with another lexicalized form sei

/ 'I don't know (literally I know there)'.20

(32) A. Voc6 gostou da palestra da Maria?
2s liked PREP.DET presentation PREP.DET Maria
'Did you like Maria's presentation?

B. gostei nao
liked.1 s NEG
'I didn't like it"


20 This form should not be confused with CFN in BP. The expression sei I" only means 'I don't know'.
The adverb Ia 'there' is only valid as a negative item in this context. Interestingly, quero I/ saber 'I don't
want to know' is also grammatical. This is related to chapter three where the position of the verb was
discussed and adverbs helped show verb movement to T.


142











Schwenter shows that if the verb is different in examples like (32) then T3 is not licit;

however, T2 is. The author shows this in (33) where it is inferable that A believes (i.e.

pragmatically presupposes) that B went to Maria's talk (2005:1449). The proposition

that is being negated is X went to Maria's presentation, and because of this Schwenter

claims that T2 is allowed.


(33) A. Voc6 gostou da palestra da Maria?
2s liked PREP.DET presentation PREP.DET Maria
'Did you like Maria's presentation?

B. # fui nao T3
Went. 1 s NEG
'I didn't go'

C. Nao fui nao T2
NEG went. 1S NEG
'I didn't go'


Because T3 is not allowed in these situations, Schwenter claims that T3 is a subset of

T2 in that it is discourse linked but it can only be used when the discourse old

proposition is directly activated, meaning it must be used with a repetition of the verb in

the proposition.

Biberauer & Cyrino (2009) also notice difference between T2 and T3. Although

they do make not mention of the semantic/pragmatic mechanics that allow CFN, they

show that T3 can only be used in the context of a presupposition denial. T2 does not

have this restriction, as Schwenter states, and it can be inferred from the context. In the

example sentences from Biberauer & Cyrino, a question is asked that allows inferred

information to become discourse-old (Schwenter's term, not theirs), and therefore, T2 is


143









licit. For the examples below, the response 'I cannot find my cat' infers that people

normally fell emotional when they lose a pet. This type of context inference is not

available in T3 CFN. This suggests that T2 and T3 are different and it is another

example that T3 is limited in its discourse function to directly activated discourse

information. In the case of this situation, it must be mentioned that for B to be felicitous,

it must be the case that A knows that B has a cat, and that the cat is loved by B. If A is

not aware of these facts, then in neither case is CFN licit. However, if A knows this then

B can use T2 CFN because it can be inferred that sadness and a certain desperation

might accompany the loss of a pet. Again, even knowing the facts of this exchange, it is

not licit to use T3 because the discourse information has not been directly activated.


(34) A. Por que voc6 esta desesperada? O que aconteceu?
why 2s ARE.2 desperate.F? what happen.PST.3s
'Why are you desperate? What's happened?'

B. Eu nao t6 achando minha gatinha nao T2
is NEG am finding 1s pussycat NEG
'I cannot find my pussycat'

B'. *T6 achando minha gatinha nao T3
am finding ls pussycat NEG
'I cannot find my pussycat'

One question for Schwenter's proposal remains. He cites the example, seen

above in (15), as a reason to favor discourse-old over presupposition. What is

presupposed in A's sentence is that 'John smoked in the past'. Because the negation in

the second clause applies to the presupposed content and not the proposition 'John has

stopped smoking', the use of CFN is not felicitous. Thus, in this case the

presuppositional aspect of the sentence does not license CFN.


144









(35) A: 0 Joao ja deixou de fumar.
DET Jo0o already stop.PST.3S PREP smoke.INF
'John has stopped smoking'

B: Ele nunca fumou (#nao).
3SG never smoke.PST.3S NEG
'He never smoked.'


Schwenter's proposal accounts for when CFN is felicitous and explains when it is

not. However, it does not explain why T3 is allowed to negate a sentence on its own or

why T2 is not a double negative. For the purposes of his proposal it is not necessary to

explain these ideas (see section 6.3.7).

5.3 Applying the Theory

This section presents additional evidence to show that Schwenter's discourse-old

model for BP negation is correct. Although I have explained Schwenter's proposal,

here I add to his examples to show that the discourse-old proposal captures all sorts of

different BP data. I will be using data from previous researchers as well as data that I

collected through the use of reality TV shows and data I gathered and tested through

grammaticality judgments from my consultants both in Brazil and in the US.

Additionally, while I reject the claims that emphasis and presupposition/contrary-to-fact

are the primary motivations for CFN, I do not claim that the CFN cannot be used with

these intended meanings. Rather, I claim that Schwenter's use of Prince's theory is a

more precise use of linguistic tools to account for more data and it can also be

implemented to give a greater understanding of the syntactic forces behind these

structures.

The first example comes from a television program where the reporter is asking

the woman if her ex-husband's wife is going to get custody of the child. Notice that here


145









the direct question is answered, and then the women claims that the child will not be

taken away. In this case, the CFN is licensed because 'she take-away your son' is

discourse-old. While this might be an emphatic use of CFN, it is not contrary to fact or

presupposition on the part of the reporter. His question is related to the ex-husband's

wife wanting to take away the child, and the women responds by stating what she thinks

will happen, although, she acknowledges that the ex-husband's wife wants to gain

custody.


(36) A. Ela ta quereno tira o seu filho, num e isso?21
3S.F is wanting take.INF DET 2s son, NEG is that?
'She wants to take your son, isn't that right?'

B. e, e isso, mai ela num tira no porque...
is, is that, but 3S.F NEG take NEG because...
'It is, that's it, but she won't take him because...'


Another example comes from a monologue, where a wife is yelling at her husband.

Here she is explaining that he is always following her around and is getting into her

business. She introduces proposition 'you look for me' which entails the proposition

'someone look for someone', and later is able to use CFN because this has already

been activated in the discourse.












21 Taken from youtube.com on April 23, 2009
http://www.youtube.com/watch?v=PMKRWOxzSdg=1&playnext_from=PL&index=5


146










(37) Tu me tiraste... porque eu tava numa boa e eu nao queria que ...
2s 1s took.2s because 1s was PREP.DET good and 1s NEG wanted that

andar de madrugada atras de mim porque
...walk.INF PREP early.morning PREP PREP 1s because

eu nunca andei atras de tu nao.22
1s never walked.1s PREP PREP 2S NEG

'You took me...because I was doing great, and I didn't want you to look for me in
the early morning because I never was looking for you.'


Finally, I present one more instance of CFN captured in writing. On an informal

blog, a woman discusses a breakup that she recently suffered. She is able to use CFN

because the discourse-old proposition 'he to speak with me' is mentioned, and thus

activates CFN through the inferred proposition 'someone to speak with someone'.


(38) ...e teve um dia q fui em uma pagode perto
and have.PST DET day that was. 1s PREP DET pagode close

da minha casa ne, q ele esperou acabar p/ tentar
PREP. DET 1s house NEG.is, that 3s waited finish.INF PREP try.INF

falar comigo, e ai eu no falei com ele nao.23
talk.INF PREP.1s and there 1s NEG speak.ls PREP 3S NEG

'And there was a day that I went to a Pagode close to my house, right, and he
waited until the party finished to talk to me, and so I didn't talk to him'


I now give some examples where the discourse-old status is not activated and as

such, the CFN is not felicitous. (39) introduces 'painting the house' into the discourse;

however, nothing about painting chairs was mentioned, therefore the CFN is not

22 Taken from youtube.com on April 29, 2009
http://www.youtube.com/watch?v=2JeDRh2YF5E&feature=related
23 Taken from yahoo brazil on April 29, 2009
http://br.answers.yahoo.com/question/index?qid=20090324060211AAmhic3


147









felicitous in B. That is, although the discourse-old proposition 'they paint the house' was

introduced, 'paint the chairs' cannot be inferred because chairs is not an inherent

property of a house, as walls, doors, and windows are.


(39) A. Eu passei pela casa e pintaram
1s passed. s PREP. DET house and painted.3P
'I went by the house, and did they paint the walls?'


a parede?
DET wall


B. Pintaram, mas nao pintaram as cadeiras (#nao.)
Painted.3P, but NEG painted.3P DET chairs NEG
'They did, but they didn't paint the chairs'


Prince predicts that inferred discourse items can be negated which means that

sentence (40) is allowed because a door is inferred to be part of a house. This is

different from example (39) above, where painting chairs cannot be inferred from

painting a house. That is, chairs are not directly relevant to painting a house, unless

they were mentioned in the discourse, which is not the case here. Painting a door or

walls is directly related to painting a house as each of the make-up part of the house,

i.e. the physical structure.


(40) A. Eu passei pela casa e pintaram
1s passed. 1s PREP. DET house and painted.3P
'I went by the house, and did they paint the walls?'

C. Pintaram, mas nao pintaram a porta nao.
Painted.3P, but NEG painted.3P DET door NEG
'They did, but they didn't paint the door'


a parede?
DET wall


This section explained the different analysis used to explain the semantic/pragmatic

reasons for CFN. It was shown that neither the emphasis nor the presupposition claims

are sufficient to capture all of the data. Schwenter's discourse-old application of


148









Prince's information structure theory is able to account for BP and CFN. The section

also argues that T2 and T3 differ. That is, T3 appears to be a subset of T2 that is

limited to directly activated proposition.

5.4 Conclusion

This chapter's goal was to given an explanation to the discourse restrictions that

affect the felicity of CFN sentence in BP. I have shown that previous attempts to

understand CFN in terms semantics/pragmatics have not be able to fully capture the

meaning and use of CFN. I presented Schwenter's (2005) use of Prince's Discourse-

Information theory. His results were that CFN in BP is licensed if what is being negated

is discourse-old information. Discourse-old information was defined as a proposition

that had been previously introduced into the discourse, either verbally or non-verbally.

After presented his theories, I applied the discourse-old theory to several new

sentences of BP and showed that this theory is capable of explaining the felicitous as

well as the infelicitous sentences.


149









CHAPTER 6
THE SYNTAX OF CLAUSE-FINAL NEGATION

6.1 Introduction

An analysis of BP negation should account for the semantic/pragmatic restrictions

presented in chapter five as well as the syntactic restrictions presented in chapter one.

A syntactic theory should account for the fact that CFN is not allowed in wh-questions,

relative clauses. It should also account for the licensing of n-words. Finally, an analysis

should account for the acceptability of CFN in topic structures, imperatives, direct

questions, and embedded clauses. Section 6.2 discusses previous analyses of BP

CFN and the extent to which they capture the syntactic patterns from BP data. In

section 6.3, I introduce my own analysis of CFN which builds on these earlier proposals

and the semantic/pragmatic theory argued for in chapter five.

6.2 Previous Accounts

6.2.1 Martins 1997

Martins (1997) discusses sentential negation in BP within a minimalist framework.

She, like those who propose an emphasis explanation, assumes that the change that

has taken place in BP negation is directly related to the Jespersen Cycle. She agrees

with those in the emphatic camp, discussed in section 5.2.1, that the postverbal

negative maker is an emphatic negative marker that came about due to the preverbal

marker weakening to the point of not negating a sentence on its own because of the

weakened negative feature (Zanuttini 1997). In fact, Martins claims that preverbal

negation is not sufficient to negate a sentence, which equates BP with modern French

(1997:24).


150









To structurally explain CFN in BP, Martins discusses European Portuguese and

what she labels Standard BP. She assumes that European Portuguese and Standard

BP have the following clause structure:

(1) CP

C ZP

Z AgrP

Agr NegP

Neg TP

T VP

She claims that two XPs are responsible for negation: NegP (Pollock 1989), where

the negative markers are inserted, and ZP, similar to Laka (1990), where the negative

markers check their negative features. The negative marker is inserted in NegP and

subsequently moves to EP to check its strong negative feature. For Martins, feature

checking describes certain strong morpho-syntactic features on heads and phrases (like

[+neg] on nao) that need to be checked against some matching feature elsewhere in the

structure in order to license the derivation. This is done through movement of a lexical

item that carries a feature to some position in the syntactic derivation where it can check

that feature. Feature checking forces the item to move to a position that is structurally

local, and the tree in (2) shows the movement of the negative marker in a European

Portuguese or Standard BP sentence.


151










(2) EP


Y AgrP
naoi
Agr NegP

Neg TP
ti


For a positive sentence, Martins proposes movement of the verb, with a [+pos]

feature, to I to check the [+pos] feature on t. The subject moves to spec,ZP. All of this

happens, according to Martins, because the verb needs to check its positive feature.

(3) gives a positive declarative sentence where the verb moves to Z and then the

subject moves to spec, P.



(3) Joao viu a LQcia
John saw DET LQcia
'John saw LQcia'

(4) YP

DP
0 Jooi
AgrP
viu2
Spec Agr'
ti
Agr TP

t2 a LQcia


A negative version of the sentence has nao moving from the inserted position, NegP, to

an adjoined position in AgrP. From there, the negative marker moves to the head of

YP, and the subject also moves to spec,YP.


152










(5) Joao nao viu a LQcia
John NEG saw.3s DET LQcia
John didn't see Lucia'


(6) YP

DP
O Jooi
AgrP
nao2
ti Agr'

Agr NegP

Neg Agr Neg TP
t2 ViU3 t2
T VP
t3
tl,t3 LQcia

According to Martins, BP is different from European Portuguese and Standard BP.

She claims that the phonological weakening of the negative marker from nao [naw] to

num [nu], mentioned in chapter four, indicates that there has been a loss of strong

features, therefore num (the phonologically weak form) does not move to Z before the

sentence is pronounced. The derivation of a BP sentence (5) is (7) where the negative

marker does not move to X. Although she does not mention this, she assumes that the

verb in positive sentences still moves to X, as this is present in all derivations, and the

subject likewise moves to the specifier position. At LF the negative marker moves to X

to check its negative features.


153










(7) -P


DP '
O Jooi
Z AgrP

Spec Agr'
ti
Agr NegP

Neg Agr Neg TP
nao viu3 t2
a LQcia


Martins proposes a second instance of nao that corresponds to the emphatic

postverbal nao which is what gives rise to T2. This second nao is not in NegP, where

the weakened nao resides, but rather in some adverbial position that is part of the VP

shell. She claims the postverbal nao is clause-final because of the Linear

Correspondence Axiom (LCA) developed by Kayne (1994). The nao needs to be in the

lowest position that is c-commanded by everything else because the sentence precedes

the clause-final nao. She claims that BP has a complex VP with several XPs or multiple

VPs as Larson (1988) suggests seen in (8) (1997:34).


154









(8) vP

Subject

v


Object

V


Indirect Object

AdvP
Nao


That being the case, (9) has the structure in (10), where the CFN is an adverb adjoined

to VP.


155









(9) A Sienna nao comeu o bolo nao
DET Sienna NEG eat.PAST.3S DET cake NEG
'Sienna didn't eat the cake.'


(10) YP


DP '
A Sienna_ -
Y AgrP

Spec Agr'
ti
Agr NegP

Neg TP
nao
Spec T'

T vP
comeu2 ,
v VP

NP V
o bolo
AdvP V
nao t2


While assuming that T1 and T2 are essentially the same in function, the author

notes that T3 is generally an answer to direct yes/no questions; therefore, it needs to be

treated differently. (11) includes direct yes/no question and a T3 answer.


(11) a. O Jo0o vai?
DET John go.3s
'Is John going?'

b. Vai nao
go.3s NEG
'No'


156









Martins sees this type of structure as a topic/comment structure where the verb is

the topic and the negation is the commentary on the topic (Barme 2000). To

syntactically explain this, as well as to differentiate it from T2 in the syntax, she

proposes movement of T to a topic phrase above CP.24 According to Martins, the verb

acts as the topic of the sentence in answers to BP direct questions. In this case, nao is

generated in NegP, but now it has a strong feature, and so moves to XP to check its

strong feature.



(12) TopP


Top
vail


AgrP


nao2


Agr


NegP


Neg
t2


Spec


24 Martins wrote around the same time as Rizzi (1997) so the fact that the TopP is above CP might be
merely the fact that Rizzi's proposals hadn't made it to her paper. The fact that TopP is above CP means
little for her analysis here.


157









Martins makes three essential claims about CFN in BP:

i) the negative marker has weakened (with regards to her claims about the
JC) to the point that it needs reinforcement
ii) a postverbal AdvP is generated to reinforce the weakened negative
marker. This adjunct is a strong nao, similar to other adverbials seen in
languages that undergo the JC
iii) T3 is a case of topicalization (making it different from T2), where the verb
moves to a topic position (TopP) and the negative marker moves to Z.


Martins' distinction between T2 and T3 is important because it highlights the

difference in the contexts that these generally occur; nonetheless, she does run into

some problems with her proposal. Recall that the two kinds of restrictions on BP CFN:

semantic/pragmatic and syntactic. First, Martins does not fully explain why in normal

cases nao has weak features and does not move to EP, but in cases of T3, it does.

Second, she assumes that T2 can occur in all syntactic contexts just like T1 and,

consequently, her analysis does not explain why CFN is not allowed in wh-questions.

With other emphatic markers in BP, the empathic marker is allowed in wh-questions.

The question remains: why do wh-questions not permit her emphatic negative marker?

Third, Martins' analysis does not take into account the semantic/pragmatic restrictions

on CFN; they are not captured. T2 negation might be an instance of adverbial

adjunction, but she does not explain why in some cases the postverbal nao is not

allowed depending on the discourse conditions.

6.2.2 Fonseca 2004

Fonseca (2004) analyzes negation within a minimalist framework. Much like

Martins, she assumes two projections that deal with negative markers, PolarityP (PolP)

and NegP. PolP is similar to Martins' EP in that it is a functional projection with polarity


158









values, i.e. negative or positive, and NegP is similar to that in Pollock (1989). Fonseca

claims that features need to be checked before Spell-Out, and the need to check

features is what causes movement visible at PF. For her, weak features do not move at

PF, but rather at LF.

Fonseca claims that the BP preverbal nao is weak in two regards. First, based on

phonological weakening and the Jespersen Cycle, nao is weak, it is a clitic that attaches

to the verb (comparable to what happens in French), and it is generated in Neg.

Second, and related to the first according to Fonseca, the clitic negative marker has

weak negative features, akin to [uNEG] features, that must be checked before Spell-

Out. To check its negative feature, the negative marker moves from Neg to the

specifier of PolP. According to her, the only way to do this for BP is TP movement to

spec, PolP where it forms a spec-head relationship with another negative marker. The

head of PolP is the second negative marker nao that needs to form a checking

relationship, but this marker has strong features ([iNEG]), which allows it and the clitic

negative marker to match their negative features (Fonseca 2004:16). This is seen in

the simple derivation below of sentence (14). While the preverbal negative marker is

the element that carries the [uNEG] feature, the whole verbal complex must move to

match this feature because the negative marker clitizes to the verb. However, she

extends this further and claims that TP has the negative feature, and this is the reason

the entire clause to move to spec,PolP.


(13) Eu num vi nao.
1S NEG saw. 1s NEG
'I didn't see (it)'


159










(14)


PolP

TP[uNEG] Pol'


Eu num-vii


Pol TP
nao [iNEG] ti


To account for T3 negation, Fonseca posits deletion of the preverbal nao, based

on Jespersen Cycle weakening, which means that T2 and T3 are essentially the same

according to her. Fonseca's main claims include the idea that the preverbal marker is

weakened and therefore is unable to negate a sentence by itself, again like in French,

and the postverbal negative marker takes over as the semantically negative element in

the sentence. Her main claims are summarized as follows:

i) The preverbal marker is too weak to negate the sentence on its own, i.e. it
has a [uNEG] feature.
ii) The preverbal nao cliticizes to the verb
iii) A strong marker with [iNEG] feature is generated in Pol.
iv) The entire TP moves to spec,PolP in order to check the negative feature
in TP
I believe that Fonseca is right in trying to align BP with what is known about other

Romance languages in that she claims that the preverbal marker is weak and unable to

negate a sentence on its own. However, her proposal runs into technical and

theoretical problems and inconsistencies. Syntactically, she claims that the weak

marker moves at LF and that the same space reserved for LF movement is occupied by

the postverbal nao. More important however, is the claim that the preverbal marker is

weak. This on its own is difficult to defend because all the corpus studies related to

negation in BP show that T1, preverbal only, is the overwhelming choice for negation.


160









Even though there are problems with the syntactic development, a larger problem is her

lack of pragmatic/semantic discussion. She does not give any explanation of the

pragmatic restrictions on T2 and T3. However, she does identify the need to have

movement of a large part of the structure to a higher position in the tree. Although she

doesn't mention the syntactic environments that block CFN, namely wh-questions and

relative clauses, Fonseca's structure does block the A'-movements involved in these

constructions. The TP movement that she argues for creates an island out of which

additional movement is prohibited. This last observation, although not made directly by

Fonseca, is part of the analysis I ultimately propose for BP CFN.

6.2.3 Cavalcante 2007

More recently, Cavalcante (2007) continues with Martins' claim that T2 and T3 are

syntactically different. Cavalcante uses several different corpus studies to determine

where all three types occur. From there, within a minimalist framework, he proposes

structural positions for the postverbal nao. The author determines, based on the

different corpus studies, that T2 occurs in main clauses and in embedded clauses, and

T3 cannot occur in embedded clauses or with topic constructions. To account for the

different types of CFN in BP, he proposes the use of a new type of XP, a Denial Phrase

(DenP). For Cavalcante, DenP accounts for T2 and T3 in main clauses, but it doesn't

resolve the issue of T2 in embedded clauses. He proposes that for T2 in embedded

clauses, there is nao that adjoins to TP following which TP moves to the specifier of

TopP. I explain his mechanisms below.

Following Rizzi's (1997) idea of an expanded CP, Cavalcante positions DenP

above ForceP and assumes NegP to be below TP and above VP. His structure for BP

is (15).


161









(15) DenP ForceP FinP IP/TP NegP -VP

Crucial for Cavalcante's proposal is the existence of a Denial Phrase above

ForceP. The head of DenP is filled by an abstract phonologically null element that has

some denial feature checked by the insertion of nao in to the specifier. He explains that

DenP is a category responsible for codifying discourse information that "refers to a

confirmation or a denial of a previous presupposition" (Cavalcante 2007:129).25 He

claims that nao in the following sentences is in DenP. Nao in this sentence denies the

previous presupposition.


(16) A. E mora aqui ou a senhora mora em Btro luga?
And lives here or DET lady lives PREP other place
And does she live here or does she live in another place'

B. Nao, ela mora... ela morava... depois desse home dai...
NEG, 3S.F lives 3S.F live.IMP after PREP.DEM man PREP.ADV
'No, she lives... she used to live... after that man from over there'


Because DenP is above CP, he states that DenP cannot be selected which

means, according to him, that DenP does not have scope over the sentence, and

therefore, it does not change the polarity of the sentence (2007:129). As DenP is

above ForceP, T3 does not occur in embedded clauses because then the negative

marker would be outside the clause. He continues by stating that DenP does not

change the polarity of a sentence, but rather denies the previous presupposition, which

could be positive or negative. This means that in the case of T2, there is no semantic

25 "DenP 6 a categoria que aloja as profrases afirmativas e negatives (como SIM, E e NAO) utilizadas em
contextos de resposta a uma pergunta direta ou de assentimento ou denegagAo de uma declaragAo
realizada anteriormente" (Cavalcante 2007:129)
"DenP is a category that houses affirmative and negative prophrases (such as YES, IS, and NO) used in
the context where they are the response to a direct question or assertion, or denial of a previous
declaration."


162









confusion because the postverbal negative marker is there only to confirm the

presupposition. The wording of his definition prevents a double negative interpretation

in T2 contexts.

Cavalcante gives sentence (17) as an example of T2 in the main clause. To

generate a typical T2 sentence, BP uses a DenP to insert the denial marker nao as the

specifier. Recall that the Den is filled by what Cavalcante calls an abstract element.

Above the DenP is a TopicP, also outside of CP. His TopP above CP is also headed by

an abstract head that is phonologically null. TP moves to the specifier of TopP where

TP checks some topic feature with the abstract element in the head (Cavalcante

2007:149). Note that Cavalcante does not mention what a sentence like (17) would

mean. His only claim about T2 is that the final nao does not change the polarity of the

sentence. Based on his comments about T3, given with the next set of examples, I

assume that a sentence like this means: I confirm the presupposition that I didn't buy

the house.


(17) Eu num comprei a casa nao
1S NEG bought. 1s DET house NEG
'I didn't buy the house'

(18) TopP

TP [+top] Top'

Eu num comprei a casai Top DenP

nao Den'

Den CP

C TP
ti


163









Different from T2, DenP does not confirm a phrase in T3, but rather denies it.

Specifically he claims that "final negation in [V nao] as a predication that does not have

the function of altering the truth value of a proposition, but to express a denial of the

presupposition (Cavalcante 2007:140).26 Thus, in T3 sentences, NegP, which hosts

preverbal negation, is not present. Under this assumption, the clause-final negative

marker denies a presupposition. Example (19) means something like 'I bought the

house. That is not a true statement', which is different from what he claims for T2.


(19) Eu comprei a casa nao
1 s bought. 1s DET house NEG
'I didn't buy the house'

Cavalcante claims that DenP and his derivation are able to account for all

instances of negation in BP except for T2 in embedded clauses, because DenP is

outside the main clause, i.e. above it. Recall, Cavalcante states that the DenP cannot

be selected, and therefore according to him, is unavailable for embedded clauses, and

this rules out T3 in embedded clauses (Cavalcante 2007:141).



(20) ?*Ele disse que conseguiu nao
3s said.3s that able.to.PST.3S NEG
'He said that he wasn't able to.'

To get around the problem of T2 negation in embedded clauses, he suggests that

a second nao adjoins above TP and below FinP and then the TP raises to Topic



26 "ISso implica que, at6 a ativagAo do nivel DenP, acima de ForceP, o sistema computacional realize
uma derivagAo em tudo semelhante a de uma sentenga afirmativa, o que 6 compativel corn a intuigAo de
que a sentenga funciona como um t6pico e a negagAo final em [V nao] como uma predicagAo que nao
tem a fungAo de alterar o valor de verdade da proposigAo, mas expressar uma denegagAo sobre uma
pressuposigao." (Cavalcante 2007:140)


164









Phrases above FinP (Cavalcante 2007:149). He never explains why this mechanism is

not available in main clauses for T2.



(21) ForceP


Force TopP
que

TP[+neg]i TopP


pro num viajou Top FinP


Fin TP


no TP


ti

Cavalcante's study is helpful because he begins by systematically looking at

where each type occurs. As such, he is able to separate out the occurrences of T2 and

T3, and identify, through corpus studies, the syntactic situations that allow these types

of negation. By doing this, he concludes that BP syntax generates T2 and T3 in the

same manner in main clauses. Cavalcante also recognizes the fact that T2 and T3

seem to be linked to discourse, and so he developed DenP, which is a projection high in

the derivation. His claims have their problems, however. Namely, his introduction of

DenialP to the derivation is difficult to follow. He explains what it means to be a denial

and not a negation, but is not consistent in using his own definition. In the case of T2,

he states that the denial acts as affirmation of a negative sentence. In contrast, DenP is

used as a denial of a positive sentence for T3. So while these positions, lexical items,


165









and discourse functions are the same, the semantic meaning changes to the point of

being opposites. Additionally, the evidence that DenP cannot be selected is based

solely on T3 data, i.e. T3 cannot be in embedded clauses. Stating that DenP is not

selected does not seem to be compatible with what he states about it not having scope.

Generally scope is seen to be related to syntactic hierarchy. As DenP is higher up in the

derivation, it would then have scope over the whole sentence, contrary to what

Cavalcante claims.

The fact that T2 has two different manifestations is also problematic. Cavalcante

never answers the questions: what does the CFN contribute in embedded clauses?

Does it have the same affirmation of negative presupposition? If the answer is yes, why

doesn't this nao appear in the denial phrase as is the case in main clauses. If the

answer is no, what semantic information does CFN carry, if any, in embedded clauses.

By separating T2 into two different derivations he causes confusion.

Finally, Cavalcante argues that topics are not licit in T3, but are in T2. His

argument suffers because he also claims that T2 and T3 are syntactically identical in

main clauses. This is a problem for his analysis, which he admits he cannot solve

(2007:144).

6.3 A Proposal for Clause-Final Negation in Brazilian Portuguese

My proposal builds on these previous analyses coupled with the information

presented in chapter five. I assume like Cavalcante that T2 and T3 are related to the

discourse (Schwenter 2005). Also, I follow Martins and Cavalcante in that T3 sentences

should involve topicalization, although I extend Martins' analysis to include topicalization

of propositions and not just the verb, as in Cavalcante's analysis. The principal

advantages of my analysis over the previously discussed ones are:


166









i) All of the syntactic restrictions are accounted for

ii) It provides a unified account of T2, which differs from Cavalcante who
separates T2 into main and embedded clauses, and Martins who also
suggests topicalization only for T3

iii) The semantics of T2 and T3 are dealt with in a way that permits both
types to occur motivated by the same semantic/pragmatic theory

iv) It provides a more complete analysis of the semantic values of both
negative markers in relationship to n-words


To support these claims, I give a structural analysis that is in line with what I

accept from Schwenter (2005) as presented in section 5.2.3, and I account for the facts

presented in chapter one. As a reminder, I present below a list of the syntactic

environments relevant to CFN and whether they are grammatical or not. This table

assumes that the discourse requirements are met. A 1 represents grammatical, and X

represents ungrammatical. For T2 under Allowed in Object Relative Clauses, I put /*,

which refers to the obligatory sentence final position of the postverbal negative marker;

however, half of my consultants indicate that these sentences are ungrammatical.

Nonetheless, similar to previous authors, I treat them as grammatical.


167









Table 6-1. Syntactic Restrictions for BP CFN
Environment T1 T2 T3
Discourse Requirements X /

Allowed in Relative Clauses in Subject Position X X

Allowed in Relative Clauses in Object Position / X

Allowed in Fronted Wh-questions X X

Allowed in in-situ Wh-questions X X

Allowed with Subject N-words X X /

Allowed with Object N-words X

Licenses NPI Idiomatic Expressions X

Allowed in Embedded Clauses

Allowed in Imperatives

Compatible with Topicalization


The structural analysis specifically addresses the ban on T2 and T3 with wh-questions

and relative clauses. It also addresses the differences between T2 and T3 in their

interpretation and their licensing of NPI idiomatic expressions. Finally, these structures

need to be checked against the facts related to n-words, which I discuss in chapter

seven.

The syntactic proposal is that there is an XP below ForceP and above AgrP which

houses a clause-final nao. Ndo is the head of this phrase, and I call it a Top(ic) Phrase.

This nao ends up in a clause-final position when the entire clause, AgrP, moves to the

specifier of TopP. In the case of T3, the head has an [iNeg] feature that negates the

sentence it is in. With T2, this head has an [uNeg] feature that must be licensed by

another negative element in the sentence.


168









The rest of this section is divided as follows. First I give the derivations of basic

clauses containing CFN. From there, I show how my proposal accounts for the data

represented by the table above. Principally, I discuss wh-words, relative clauses,

topicalization, and imperatives. I hold off on n-words and discuss these in chapter

seven. Section 6.3.7 ends by explaining the semantic differences between T2 and T3,

and this section includes data regarding the licensing of idiomatic expressions by CFN.

6.3.1 Derivation of Clause-Final Negation

I argue that T3 is formed by moving AgrP to the specifier of a lower TopP. This

relates to the discussion in section 5.2.3 where I claimed that CFN is used when the

proposition is discourse-old. It is this proposition that is being moved. Recall also, as

laid out in chapter two, the use of strong features to drive movement. If an

uninterpretable feature [uF*] is strong, it must be checked by an interpretable instance

of the feature in a spec-head configuration. If [uF] is weak, the feature can check

against [iF] if it c-commands [iF].

In BP's CFN sentence, the head of TopP is nao, which is a topic marker that has a

[iNEG] feature and a strong [uTOP*] feature. This TopP is the lower topic phrase briefly

introduced in chapter two. The reason that this must be a lower topic phrase will be

discussed in section 6.3.5. I assume that in derivations involving CFN, AgrP has a

[iTOP] feature that checks the [uTOP*] feature on Top. AgrP has the [iTOP] feature

because this is the discourse-old proposition. In the derivation, AgrP[iTOP] moves to

the specifier of TopP to satisfy the strong feature [uTOP*] on nao in Top. (22) gives a

generic blue-print of a T3 structure, and (23) gives a T3 sentence and its derivation.


169









(22) ForceP

Force TopP

AgrPi Top'

TOp[uTOP*,iNEG]
nao


(23) a. Gosto dele
Like.ls PREP.3S
'I don't like him'

b. ForceP

Force TopP


AgrP[iTOP]i
A


nao
NEG


Top'


Top [uTOP*,iNEG]
nao


c. ForceP

Force TopP

AgrP[iTOP]i

gosto dele


AgrP [iTOP]

gosto dele


Top'

Top[uTOP*,iNEG]
nao


The basic idea, in line with the discourse-old hypothesis, is that if an AgrP contains

discourse-old information then it can be topicalized in BP. By stating this, I am claiming

that CFN is a case of proposition topicalization, where the topic marker has two

features, a polarity feature that can be either negative [NEG] or positive [POS], and a

strong uninterpretable topic feature [uTOP*] that triggers movement. The polarity


170


AgrP
ti









feature on the discourse-old topic marker is not only seen with negative sentences, but

also with positive sentences that make use of sim 'yes'. The same type of discourse

requirements apply for the positive marker as they do for the negative marker. If the

proposition is not discourse-old then it is infelicitous to use sim, as in (24). Note that

there is no pause preceding sim. (25) would then take the entire sentence as the topic

and move it to the specifier of this topic phrase with sim as the head. A full translation

would be something like 'as for me visiting the beaches of Rio, that is a true statement'.


(24) Visitei as praias (#sim)
Visit.PST.1S DET beaches (yes)
'I visited the beaches'

(25) a. Voc6 vistou as praias do Rio?
2svisit.PST.2s DET beaches PREP.DET Rio
'Did you visit the beaches of Rio'


b. Visitei as praias do
Visit.PST.1S DET beaches PREP.DET
'I visited the beaches'

(26) ForceP

Force TopP


Rio sim.
Rio yes


Top'


Top[uTOP*,iPOS]
sim

(27) ForceP

Force TopP

AgrP[iTOP]i
visited as praias do Rio
visitei as praias do Rio


AgrP[iTOP]

visitei as praias do Rio




Top'

TOp[uTOP*, iPOS] AgrP
sim ti


171











This sim is only allowed in contexts where the proposition is the topic, as in 'visit the

beaches of Rio'. If this proposition is not the topic, but a focused or new element, then

the sentence becomes infelicitous with sim or nao. Syntactically this means that the

feature [iTOP] is not present on AgrP and therefore there is nothing to check the

uninterpretable [uTOP*] on Top, which can then not be present.

T2 is a combination of T1, discussed in chapter four, and the movement operation

seen in T3. The preverbal marker is generated in its usual position, Neg, and the CFN

marker is in Top. Unlike with T3, however, the CFN marker in this case does not have a

[iNEG] feature but rather a [uNEG] feature. This means that the CFN marker in T2 is

not sufficient to negate a sentence on its own (for a similar claim about Afrikaans see

Biberauer 2009). I discuss the semantic issues related to the two negative markers

further in section 6.3.7. Because the feature is uninterpretable, it must be checked by a

semantically contentful [iNEG] elsewhere in the structure. The movement of AgrP is still

motivated by a strong feature [uTOP*] on the CFN marker. Again, I point out here that

AgrP has a [iTOP] feature, but additionally, it also contains the [iNEG] feature from the

preverbal negative marker. I propose that [iNEG] percolates from Neg to AgrP because

the proposition as a whole is negative. It is this instance of [iNEG] that will check the

uninterpretable [uNEG] feature on Top. (28) is a basic blue-print of a clause with T2:


172










(28) ForceP


Force TopP

AgrP[iTOP, iNEG]i Top'

subject Agr' Top[uTOP*,uNEG] AgrP
Sn5o ti
Agr NegP

Neg[iNEG] TP
nso
T'

T VP
verb /


The blue-print shows that after movement, AgrP c-commands the CFN marker. This

position is what allows the [uNEG] feature to be checked. As chapter seven will show,

for negative elements in BP to check their features, the [iNEG] feature must c-command

the [uNEG] feature (Zeijlstra 2004, 2008). Thus, feature checking of the negative

features must take place after Spell-Out and cannot take place prior to movement or

under reconstruction.

Example (29) is a sentence that contains T2: Top is filled by nao, the discourse

requirement is met, and AgrP moves to spec,TopP.


(29) a. Ela nao viu os meninos nao
3s NEG see.PST.3S DET kids NEG
'She didn't see the kids'


173









b. ForceP

Force TopP


Top'


TOp[uTOP*,uNEG]
nao


AgrP[iTOP,iNEG]

Ela Agr'

Agr NegP

Neg TP
nao /


T VP
viu A
...os meninos


c. ForceP

Force TopP

AgrP[iTOP,iNEG]i

Ela Agr'

Agr N

N
n


Top'

TOp[uTOP*,uNEG] ti
nao
egP

eg[iNEG] TP
Io


T VP
viu /
...os meninos


Now that the derivations have been given, I explain how my proposal accounts for

the syntactic restrictions observed in BP CFN.


174









6.3.2 Wh-Words

The analysis I propose prevents T2 and T3 with wh-questions. I first discuss

fronted wh-questions and then in-situ wh-questions. Fronted wh-words with CFN are

ungrammatical because my proposal has AgrP move to the specifier of the lower TopP,

which creates an island (Ross 1967, Szabolcsi 2006). Specifically, the

ungrammaticality of CFN in sentences with wh-words follows from Huang's (1982)

Condition on Extraction Domain (CED) defined in 0. Properly governed means

governed by a lexical head.


(30) CED
A phrase A may be extracted out of a domain B only if B is properly governed


The CED means that extraction out of complements is allowed, but not out of

adjuncts and phrases in specifier positions because these positions are not properly

governed. The movement of AgrP to the specifier position of TopicP thus creates a

specifier island and the wh-phrase inside of AgrP cannot be extracted to a higher

position. Example (31) shows a sentence that is ungrammatical and (32) shows the

derivation. Notice that the moved AgrP creates a specifier island that prevents

extraction to the landing site for wh-words, which is spec,FocusP, above TopP. The

bolded, fronted wh-word is co-indexed with the trace that it leaves behind in the fronted

AgrP.


(31) *O que que voc6 (nao) visitou nao?
What that 2s (NEG) visited NEG
'What didn't you visit?'


175









(32) *FocP

0 quei

Foc
que


Foc'


TopP

AgrP[iTOP]ii

voc6 Agr'

Agr


TopP

Top[uTOP*]
nao
NegP

Neg[iNEG] TP
no


T VP
visitou A
...ti


This analysis thus correctly predicts that T2 and T3 are incompatible with wh-

movement.

Chapter 3 showed that wh-in-situ does not obey islands and does not involve

movement:


(33) Voc6 conversou com o autor [que escreveu que livro]?
2s talk.PST.2S PREP DET author that write.PST.3S which book
'Which book did you talk to the author who wrote?'


It was also pointed out that wh-in-situ in BP is constrained by semantic/pragmatic

restrictions. BP wh-in-situ is only allowed if certain discourse requirements are met.

Much like Schwenter's (2005) proposal for BP CFN, Pires & Taylor (2007) proposed the

use of discourse to understand syntax. They claimed that the in-situ situations are

licensed by information in the Common Ground. They define Common Ground as


176









"information that was previously given in the discourse or in the extralinguistic

context... and which is shared (or assumed by the speaker to be shared) by speaker

and hearer" (2007:5). This means that in-situ wh-questions are only felicitous when the

proposition being questioned is discourse-old. For example, in a sentence such as, 'I

made dessert', the discourse-old proposition is 'I made X'. From this one could

question:


(34) Voc6 fez que tipo de sobremesa?
2s do.PST.3s what type PREP dessert
'You made what kind of dessert?'


The felicity changes when what is being questioned is not discourse-old, as in

(35). The example is grammatical, but infelicitous. It can be made felicitous by fronting

the wh-word.


[You approach a colleague at work and ask, out of the blue:]


(35) #Voc6 conhece quem em Sao Paulo?
2s know.2s who PREP Sao Paulo
'You know who in Sao Paulo?'

The following exchanges shows how the discourse requirements affect the felicity

of the sentence. With the appropriate context, the above example becomes felicitous:


[You approach a colleague at work and comment A, to which the colleague
responds with B:]


(36) A. Eu visitei Sao Paulo este fim de semana
1s visit.PST.1s Sao Paulo DET end PREP week
'I visited Sao Paulo this weekend'


177










B. Voce conhece quem em Sao Paulo?
2s know.2s who PREP Sao Paulo
'You know who in Sao Paulo?'

The example below shows that the discourse restriction on the use of CFN is similar.

(37) A. Voce conheceu alguem em Sao Paulo este fim de semana
2s know.PST.2s somebody prep Sao Paulo DET end PREP week
'Did you meet anybody in Sao Paulo this weekend'

B. Eu nao conheci ninguem em Sao Paulo este fim de semana nao.
1 s NEG know. PST. 1s nobody PREP Sao Paulo DET end PREP week NEG
'I didn't meet anybody in Sao Paulo this week end'


The similarities between the discourse restrictions on wh-in-situ and CFN suggest

that these two constructions make use of the same mechanisms and by extension the

same structural position. I propose that CFN and wh-in-situ make use of a single

syntactic position in the left periphery, the specifier of the lower TopP. I previously

argued that there is no movement for wh-in-situ; rather, there is a Q operator linked to

the discourse in a similar way that the CFN is linked. I propose that there is

complementary distribution between AgrP topicalization and the Q operator because

both are located in the specifier of the lower TopP, which I called Top2P in chapter

two.27 There can be one or the other, but not both. Thus, CFN and wh-in-situ are

incompatible.

6.3.3 Relative Clauses

BP prohibits relative clauses containing T2 and T3 with the exception that T2 is

possible in relative clauses if the relative clause is in a sentence final position. The

derivation argued for here does ban the use of CFN in relatives, and this makes relative

27 This is an updating of the analysis in chapter 3 where I tentatively proposed that the Q operator was in
the head position of the higher Topl P. This change does not affect the results.


178









clauses in object position with T2 a mystery. Before addressing the problem with object

relatives, I give a current account of relative clauses in BP and show that the analysis in

this chapter account for the ban on relative clauses with CFN.

Lessa de Oliveira (2008) and Nunes & Kato (2009) give an account of BP relative

clauses within a minimalist framework where they propose a raising analysis for relative

clauses28 (Kayne 1994). A structure based on their analysis is:


(38) a pessoa que o professor viu
DET person that DET professor see.PST.3S
'The person that the professor saw'

(39) DP

a ForceP

DP ForceP

pessoai DPk Force AgrP
A
que ti o professor viu tk

Again, much like wh-phrases, relative clauses involve a type of movement that is

blocked once AgrP moves to spec,TopP with CFN. This is seen in example (41). AgrP

moves to a specifier position which gives rise to CFN. The DP pessoa cannot be

extracted out of this AgrP due to the same CED violation seen in wh-movement.


(40) a pessoa que o professor nao viu (*nso)
DET person that DET professor NEG see.PST.3S NEG
'the person that the professor didn't see'





28 Both works also deal with non-standard and free relative clauses, but the idea that there is raising out
CP to a higher projection is present. As such, I only present what they call standard relative clauses.


179










(41) *DP


a ForceP

DP ForceP

pessoai DPk Force TopP

que ti AgrPm Top'

o professor nao viu tk nao tm


The prediction that CFN in impossible in relative clauses is largely correct. The

table introduced earlier showed that T3 is completely impossible in relative clauses and

T2 is impossible, except if the relative clause happens to be at the end of the clause:


(42) Todo mundo gosta de morangos
All world like.3s PREP strawberries
'Everybody like strawberries'

(43) Eu ensinei portugues para uma pessoa que nao gosta de morangos nao
1s taught. 1s Portuguese PREP DET person that NEG likes PREP strawberries NEG
'I taught Portuguese to a person that doesn't like strawberries'


I do not have an explanation for the why this example is grammatical. The theory

introduced above predicts that it should be not be. Note that if the object is not the last

element in the sentence, then the sentence becomes ungrammatical. I leave the

explanation of this contrast for future work.


(44) *Eu ensinei portugues para uma pessoa que nao gosta
1staught.1s Portuguese PREP DET person that NEG likes

de morangos nao ontem
PREP strawberries NEG yesterday

'I taught Portuguese yesterday to a person that doesn't like strawberries'


180











6.3.4 Embedded Clauses

Others before this work have claimed that T3 does not exist in embedded clauses

(Cavalcante 2007). Recall that in chapter one I have shown that T3 does exist in

embedded clauses if the discourse requirement is met. The problem here is that

although T3 can exist in embedded clauses, there are no cases of them occurring in

natural speech. The sentences that I presented were created in conjunction with a

scenario that would permit the discourse-old information to be in the embedded clause.

After creating such a scenario, my consultants generally agreed that T3 is allowed in

embedded clauses. This situation points to a possible difference between T2 and T3.

As stated in section 5.2.3, T3 is restricted to directly activated discourse-information.

This seems difficult to occur naturally, and generally if there is CFN in an embedded

clause, it is T2.

Although not naturally occurring, T3 can exist, and my proposal predicts that this

structure is syntactically possible, because nothing prevents topics in embedded

clauses. Below is an example where the discourse requirement is met and the

sentence is grammatical. As expected from the discussion, AgrP moves to a Topic

position, and negative marker is in the clause-final position.



(45) Eu acho que ele terminou o exame no
1s think that 3s finished DET test NEG
'I think that he didn't finish the test'


181









(46) IP

Eu I'

I VP
acho
V ForceP

Force TopP
que
AgrP[iTOP]ii Top'

ele Agr' Top tii
Zf naO[uTOP*]
Agr VP
terminou /
..o exame

6.3.5 Topicalization

Another area in which this dissertation differs from previous analyses is in allowing

topic constructions with CFN. In chapter 3 I argued that topics in BP are base-generated

in the specifier of the higher spec,Topl P and co-indexed with a resumptive pronoun,

null or overt. Different from wh-movement, topic fronting with CFN is grammatical

because no movement is involved. This requires that at least two topic positions exist in

BP: a lower one used for CFN and a higher one for 'typical' topicalization. The following

examples show the derivation of a sentence that contains both a topicalized DP and

CFN. In this case, the CED is not violated because there is no movement of the topic

out of AgrP.


(47) A Eliza, eu nao vi nao
DET Eliza, 1S NEG saw. s NEG
'As for Eliza, I didn't she her'


182









(48) Top1P

A Elizai Topl'

Top1 Top2P

AgrP[iTOP]ii Top2'

eu Agr' Top2 tii
Zf nao[uTOP*]
Agr TP

nao vi ... pro;

Separation of the two kinds of topic phrases, Topl P versus Top2P, is necessary to

avoid generating ungrammatical word orders. In particular, the position for base-

generated topics must be above the position to which AgrP moves in order avoid getting

the two topics in the wrong order:


(49) *Eu nao vi noo a Eliza
1S NEG saw. 1s NEG DET Eliza
'As for Eliza, I didn't see her'


Rizzi proposed that there is potentially more than one topic position; however,

the above structure does not instantiate TopP recursion in the sense he describes.

Rather, different kinds of topics must be ordered in the left periphery (Beninca & Poletto

2004). In the case of BP, the data shows that fronted AgrPs are lower topics and base-

generated DPs are higher topics. I have encoded this observation by simply labeling

the two kinds of topic phrases as Topl P and Top2P. Even before this, Aissen (1992)

argued for an ordering of topics in Mayan Tzotzil, and this has been shown to work in

other Mayan languages (Hansen 2005). As seen earlier, these topic phrases are also

separated by FocusP.


183









6.3.6 Imperatives

BP allows all three types of negation in imperatives. Two verb forms are used in

BP (negative) imperatives: subjunctive or bare verb (Scherre 2007). I repeat some

examples from chapter one. Dialectal differences exist between the use of the two

types of imperatives; however, this does not influence the outcome, and as such I use

the bare verb form.

A. Bare Verb B. Subjunctive

(50) Nao Faz isso Nao faga isso T1
NEG do that NEG do.SUBJ that

(51) Nao faz isso nao Nao faga isso nao T2
NEG do that NEG NEG do.SUBJ that NEG


(52) Faz isso nao Faga isso nao T3
Do that NEG do.suBJ that NEG
'Don't do that' 'Don't do that'


I only illustrate imperatives with T2, but the same results apply to T3. The same

discourse restrictions apply, and the next examples show this. This first example shows

that inferred discourse information is allowed, as seen in declarative sentences. B is

the derivation, again similar to declaratives.


[A mother to a young daughter who is writing on the walls with new crayons]

(53) A. Nao faz isso nao
NEG do that NEG
'Don't do that'


184









B. ForceP


Force TopP

AgrPi Top'

Nao faz isso Top AgrP
n5o ti

If there is nothing in the conversation that triggers the discourse-old information then

CFN is not licit (Schwenter 2005).


(54) Ano que vem a gente vai viajar a Salvador
Year that comes DET people go travel.INF PREP Salvador
'Next year we are going travel to Salvador'

(55) Quando voces chegarem la, nao vao ao Restaurante Irina (#nao)
When 2P arrive.FUT.SUBJ there NEG GO.2P PREP.DET Restaurant Irina NEG
'When y'all get there, don't go to Restaurant Irina'


The syntax of BP imperatives is similar to what has been proposed for the Slavic

languages (Han 2001). Unlike other Romance languages which are assumed to have I-

to-C movement in imperative clauses (Rivero & Terzi 1995, Han 2001, Zeijlstra 2006),

BP verbs do not move in imperative clauses, and this is based on the evidence from

object clitics. In Spanish, the object clitic is to the right of the verb with positive

imperatives (57), unlike a declarative sentence, when it is to the left (56).

(56) Maria lo ley6. Declarative
Maria CL.M READ.3SG
'Mary read it'

(57) Leelo! Imperative
Read.2s.IMP-CL.M
'Read it!'

In order for the verb to be on the left of the clitic, it must move to some position higher in

the tree that deals with (imperative) mood, generally considered to be C. BP clitics are


185









always to the left of the verb suggesting that there is no movement of the verb to a

higher spot in the derivation.


(58) A Maria me da o livro. Declarative
DET Maria CL.1s gives DET book
'Mary gives me the book'

(59) A. Me da! Imperative
CL.1s gives
'Give me!'

B. Me da ele! Imperative
CL.1s gives
'Give me it!'


Because the verb stays in its position in IP, as in a normal declarative, CFN is licit in

imperative sentences.

6.3.7 Semantic Interpretation of Clause-Final Negation

The previous sections showed how the current proposal accounts for the syntactic

restrictions related to CFN, and in most cases, T2 and T3 are identical. They differ

syntactically in two areas: negative concord and idiomatic expressions. Negative

concord is the topic of chapter seven and is treated there. Here I discuss idiomatic

expressions and how they further our understanding of the semantic interpretation of T2

and T3. Before discussing idiomatic expressions, I explain the theoretical problems

related to the interpretation of sentences containing CFN. From there, I show how

idioms help delineate the differences between T2 and T3, but how this separation

creates a problem related to double negation.

T1 is the unmarked strategy and T3 is the marked strategy, and in both cases,

there is only one negative marker for the sentence. A problem arises with T2, as this

structure contains two negative markers, leading researchers to question how these


186









sentences do not evoke a double negative reading (Schwegler 1991a, da Cunha 2007).

One claim is that BP is traveling through the Jespersen Cycle, and the preverbal

negative marker no longer has a negative feature strong enough to negate a sentence

on its own. It is well documented fact that negative markers lose their negative force.

However, in the case of BP, this seems unlikely because in all corpus studies T1 is still

the main method of negation (Cavalcante 2007), suggesting that the preverbal has not

lost its semantically strong negative feature.

Another attempt to understand the lack of double negation in T2 comes from

Cavalcante (2007). First, he attempts to explain T3 by claiming the clause-final is a

denial of the asserted positive sentence. This line of thinking follows what Horn has

called metalinguistic negation (Horn 1989), and Cavalcante takes the topic nao as some

sort of metalinguistic negation. Metalinguistic negation usually suggests the denial of

an assertion or the refusal to accept as assertion. The classic example reused by Horn

(1989) is 'The king of France is not bald (because) there is no king of France'.

Essentially, negation in this sense is not about a part of the sentence but rather the

whole utterance, i.e. the king is not bald not because he doesn't have hair, but he is not

bald because he doesn't exist, and therefore cannot be either bald nor have a head full

of hair. While not exactly like metalinguistic negation as defined by Horn, this type of

interpretation comes close to what T3 means. Example (60) explains what Cavalcante

means when he uses denial in BP CFN. This sentence has T3 CFN, and the positive

sentence is 'John ran two miles'. If we take the CFN to be a denial, the meaning of this

sentence is: 'John ran two miles, that is not true', or 'John ran two miles, I deny that

fact'.


187









(60) O Joao correu duas milhas nao
DET Jo0o ran.3s two miles NEG
'John didn't run two miles.'


The major problem with this hypothesis, as noted by Cavalcante himself, comes

with T2 and its meaning. If T3 is metalinguistic negation, where the utterance as a

whole is negated or denied then his analysis leads to an incorrect interpretation.

Sentence (61)a involves T2, and (b) translates the sentence to a possible interpretation.

Notice that the possible interpretation gives a double negative, meaning that 'John did

run two miles', when (a) clearly states that he didn't.


(61) a. O Jo0o nao correu duas milhas nao
DET Jo0o NEG ran.3s two miles NEG
'John didn't run two miles.'

b. John didn't run two miles. That is not true (that John didn't run 2 miles)


Though Cavalcante mentions this fact, he doesn't pursue a solution. To explain

this, I propose taking T2 to be a case of negative spread (Den Besten 1986,

Giannakidou 2006). Negative spread is where the negative feature of some items, like

n-words and negative markers, spreads over the whole sentence. For example, in non-

strict negative concord languages, the negative marker must c-command the n-word,

unless the n-word is in the subject position, as in (62) (see chapter seven for more

discussion).


(62) O Abilio nao ajudou ninguem
DET Abilio NEG helped nobody
'Ablilio didn't help anyone'


188









However, the n-word can be c-commanded by another n-word.


(63) Ninguem ajudou ninguem
nobody helped nobody
'Nobody helped anyone'


These examples illustrate the idea that the features spread to the rest of the sentence

without causing a double negation reading. T2 seems to be a situation similar to this,

where the first negative marker spreads its negative meaning to the rest of the negative

items in the sentence, because these items are not by themselves inherently negative.

Thus in the previous sections I claimed that the CFN maker in T2 carries an [uNEG]

feature and must be c-commanded by some element with a [iNEG] feature. In the case

of CFN AgrP carries the [iNEG] feature as the feature is able to percolate up from the

preverbal negative marker. I continue with this line of reasoning in chapter seven, and

now I focus on NPI idiomatic expressions and T3.

Biberauer & Cyrino (2009) notes that T3 does not license NPI idiomatic

expressions whereas T2 does, and I believe that this observation is related to the

semantic meaning of the negative markers. This is important because it is one area

that demonstrates that T1 and T3 are not only different in terms of discourse, but also in

terms of inherent semantic meaning and semantic contribution to a sentence. This also

shows that the semantics of T3 and T2 are not equal. They give the following example.

A utters the sentence 'John is rich'. B disagrees with A and uses an NPI idiom to

express the fact that 'John is not rich'. The expression in Portuguese literally translates

to 'He doesn't have a cent/penny with a hole in it' which means 'he doesn't have any

money'. The translation that Biberauer & Cyrino give is 'he doesn't have a red-cent'.


189









Notice that in the case of T1, B, the sentence maintains its idiomatic meaning. The

same can be said about T2, B', but in B", the situation is different because of the use of

T3. In the case of B", the figurative meaning is lost, and only a literal meaning remains.

Notice C, a sentence containing T3, and how it would most likely have to finish for it to

be grammatical in this situation, albeit perhaps not pragmatically felicitous.


(64) A. O Joao e rico!
DET John is rich
"John is rich!"

B. 0 que? ele nao tem um tostao furado! T1
what 3s NEG has DET cent with-a-hole
'What? He doesn't have red-cent' (i.e. he doesn't have anything)

B'. 0 que? ele nao tem um tostao furado nao! T2
what 3s NEG has DET cent with-a-hole NEG
'What? He doesn't have a red-cent'(i.e. he doesn't have anything)

B". *O que? ele tem um tostao furado nao! T3
what 3s has DET cent with-a-hole NEG
'What? He doesn't have a red-cent. (i.e. he doesn't have anything)


C. Ele tem um tostao furado nao, ele tem um inteiro! T3
3s has DET cent with-a-hole NEG, 3s has DET whole
'He doesn't have a red cent; he has a BLUE one' (i.e. literal meaning)


The fact that T3 does not license NPI idioms in the way that T1/2 do suggests that

it is different. For it to negate a sentence on its own, as T1 does, it needs to have an

interpretable negative feature, which I propose it does. Like T1, T3 has a [iNEG]

feature, but in the case of NPI idioms the [iNEG] feature is apparently not in an

appropriate structure position to license the idiom chunk. Thus, the idiom loses its

idiomatic meaning and has only a literal meaning. As discussed above, negative

elements with an [uNEG] feature must be licensed by a negative element with an [iNEG]


190









feature. I follow Zeiljstra 2004, 2008 and assume that [iNEG] must c-command [uNEG]

in order for [uNEG] to be checked. Since in the case of T3, the NPI idiom is not c-

commanded by the CFN marker, the NPI interpretation is not available. This state of

affairs is seen below:

(65) *TopP

AgrPi Top'

DP Agr' Top AgrP
A nio[iNEG] ti
Ele Agr TP

T VP
tem
um tostao furado[uNEG]

T2 two is different because the licensing of the NPI is done by the preverbal

negative marker, which has an [iNEG] feature. The CFN marker in Top is irrelevant for

the licensing of the NPI but is itself licensed by the same interpretable negative feature

on AgrP. In the following structure, [iNEG] c-commands and checks both instances of

[uNEG].



(66) TopP

AgrP[iNEG]i Top'

DP Agr' Top AgrP
A/ no[uNEG] ti
Ele Agr NegP

Neg[iNEG] TP
nao
T VP
tem
um tostao furado[uNEG]


191









Because of the facts presented in this section, I argue that T3 and T2 participate

in the same derivation but differ in their morphosyntactic features. The nao in T2 is a

negative item that is similar to an n-word and cannot negate a sentence on its on; it is

[uNEG]. The nao in T3 is a negative item that can negate a sentence on its own-it is

[iNEG]-but is highly restricted to a sub-set of discourse linked information and is not in

a position to license NPI idioms. These claims and the claims about the [NEG] feature

that the negative markers have are further explored with respect to n-words in chapter

seven. The behavior n-word licensing supports the current proposal.

There is one problem with this arrangement, however. The theoretical possibility

exists for a T1 negative marker, [iNEG], and a T3 negative marker, [iNEG], to be in the

same structure. Such a sentence should then have a double negation reading. A

double negation reading for the purposes here and in chapter seven refers to the idea

that two [iNEG] features in the same sentence would cancel each other out causing the

sentence to receive a positive interpretation. Below is a hypothetical illustration:

(67) O Joao nao[iNEG] correu duas milhas nao[iNEG]
DET Jo0o NEG ran.3s two miles NEG
'John didn't not run two miles' i.e. he did run two miles.

Native speakers indicate that such sentences never have a double negation

interpretation however and this is problematic for my proposal. I will need to leave this

issue unresolved in hopes that it can be solved later.

6.4 Conclusion

This chapter provided a syntactic analysis of Clause-Final Negation in Brazilian

Portuguese. Section 6.2 presented some previous attempts to understand CFN

syntactically. I then claimed that CFN in BP is derived through fronting AgrP to the

specifier of a lower Topic Phrase in the left periphery. The head of the topic phrase is


192









either the negative marker nao or the positive marker sim. I concluded this chapter by

proposing that this head can be realized by ndo[uNEG] or ndo[iNEG]. The former is

found in T2 and the latter in T3. It remains a problem why the latter cannot also be used

in T2.

In order to account for the BP data, I relied on both syntactic mechanisms and the

semantic/pragmatic restriction from chapter five. We saw that the analysis accounts for

the syntactic restrictions from chapter one. In contrast to Cavalcante's approach, there

was no need to provide different analyses for CFN in main and embedded clauses. In

agreement with others, however, I am forced to claim that there are two CFN markers

which differ in the interpretability of their negative feature.


193









CHAPTER 7
NEGATIVE CONCORD IN BRAZILIAN PORTUGUESE

7.1 Introduction

The principal goal of this chapter is to discuss the concepts of Negative Concord

(NC), N-words, NC in Brazilian Portuguese (BP), and how NC and Clause-final

Negation (CFN) interact. I divide this chapter into three main sections. Section 7.2

discusses the general ideas surrounding NC, n-words, and the differences between

strict and non-strict NC Languages. I also discuss the debate regarding the semantic

and syntactic nature of n-words and negative concord. Section 7.3 shows that BP is a

non-strict NC language and makes explicit which words are n-words in BP. Section 7.4

addresses the nature of the CFN marker and its relationship to n-words. This is a

continuation of the discussion in chapters five and six, and here I finalize my discussion

of CFN, semantics, and syntax.

7.2 Negative Concord

Negative Concord (NC) is a highly studied phenomenon, and this fact is especially

true within the field of Romance Linguistics. The intense study of NC has lead to

several theoretical disputes. This section addresses these debates. First, I define and

give examples of the different two types of NC, i.e. strict and non-strict. From there, I

lay out the principal analyses for a general understanding of NC both semantically and

syntactically.

The main issue concerning n-words is whether they are negative or not. In some

languages, such as the Germanic languages, n-words are in fact negative. This is the

reason that, in all cases, an n-word in Standard English and a sentential marker give

rise to a Double Negation (DN) reading. However, it has been more difficult to decide


194









whether n-words in NC languages are inherently negative. There are essentially two

main options: either n-words are negative (Zanuttini 1991, Haegeman 1995, Haegeman

& Zanuttini 1996, De Swart & Sag 2002); or n-words are non-negative (Laka 1990,

Ladusaw 1992, Giannakidou 2000, Zeijlstra 2004, 2008). A third option claims that n-

words are ambiguous, however this is not treated here29 (Van der Wouden & Zwarts

1993, Van der Wouden 1997, Herburger 2001). For this dissertation, I follow Zeijlstra

(2004), who proposes a version of the non-negative option based more on syntax and

feature agreement than semantics. I present Zeijlstra's claims that NC involves an

Agree relationship. According to him n-words in NC languages have an [uNEG] feature

while in DN languages they have an [iNEG] feature.

7.2.1 Definition and Types of Negative Concord

Early in the study of Portuguese and negation, Dias (1917) (cf. Pereira 2000)

explained what NC is stating that, 'From the co-occurrence of two negatives a positive

does not result' (Dias 1917:308). This occurs in many languages, such as the Slavic

languages where n-words (defined below) require the presence of some negative

marker in order for them to be negative. In the case of (1), the n-word nikomu must be

accompanied by the negative marker on the verb, ne-. Romance languages are also

known for having NC, seen in example (2), where the n-word nadie requires the

presence of the preverbal negative marker no.


(1) Milan nikomu nevola Czech
Milan nobody NEG-call
'Milan doesn't call anybody' (Zeijlstra 2004:121)


29 See Zeijlstra 2004 for a presentation of the ambiguity analysis and the reasons that it cannot be cross-
linguistically applicable.


195










(2) Juan no Ilam6
Juan NEG called.3SG
'Juan didn't call anybody'


a nadie
PREP nobody


N-words are words that enter into a relationship, to be discussed below, with a

negative marker in NC situations (Laka 1990, Giannakidou 2000, 2006, Zeijlstra 2008).

Giannakidou gives a more formal definition that I follow here (2006:328):


(3) N-word
An expression a is an n-word iff:
a. a can be used in structures containing sentential negation or another a-
expression yielding a reading equivalent to one logical negation; and

b. a can provide a negative fragment answer


Giannakidou then defines a fragment answer as the following:



(4) Fragment Answer
An answer a to a wh-question Q is a fragment answer iff:
c. a corresponds in form to the wh-XP constituent in Q; and
d. a is interpreted as a proposition

An example of a fragment answer from Spanish is (5). The question's answer is an

n-word fragment.


(5) a. Quien hizo la tarea?
Who did.3s DET homework
'who did the homework?'

b. Nadie
'nobody'


196


Spanish









In addition, n-words can often be identified by what appears to be negative

morphology, such as the nie- in niemand 'nobody' in Dutch, or the ni- in ningun 'none'

for Spanish; however, this is not strictly necessary. French has several n-words with no

apparent negative morphology, such as personnel 'nobody'.

The debate surrounding NC is: if an n-word is negative, then two negative items

appear in a sentence, the n-word and the sentential negative marker. That being the

case, sentences with an n-word and a negative marker should yield a double negation

(DN) reading. The interpretation in NC is that of only one negative operator in the

sentence, however. This is not what happens in many Germanic languages evidenced

by Standard English. If the negative marker and an n-word are both present, then DN

arises. (6) gives an English sentence with an n-word and a negative marker, and (7)

gives a logical representation. Notice that in this case, there are two negative elements,

hence DN. DN means that two negatives are independent of each other and as such,

they cancel each other out, which is why a double negative gives a positive reading in

some dialects of English (Horn 2002:296 for a definition of double negatives).


(6) Maria doesn't see nobody (i.e. Maria sees somebody)

(7) E 3x SEE(m,x)30 Double Negation


NC does not have the same logical structure as a DN sentence does. The logical

representation for the NC sentence (2) is given below, and even though there are

(apparently) two negative elements in the sentence, only one negative operator is

interpreted semantically.


30 There are other ways to represent this; however they are not relevant here.


197










(8) x CALL(j,x)


This is NC, i.e. two negative elements coalescing to yield only one semantic negation

(Giannakidou 2000, 2006, Zeijlstra 2008).

NC languages are divided into two types based on empirical patterns: strict NC

language and non-strict NC languages. The same two languages from above, Czech

and Spanish, serve as examples of each type of NC. Strict NC languages are

languages that always require a sentential negative marker to be present when the

sentence contains an n-word (Giannakidou 2006:352). In (9), if the negative marker is

not present, then the sentence is ungrammatical.


(9) Dnes nikdo *(ne)vola Czech
Today nobody NEG.calls
'Today nobody is calling' (Zeijlstra 2008:3)


In many Romance language however, this generalization does not hold, and these

languages are known as non-strict NC languages. This is because the n-word does not

always require the presence of a negative marker. In Italian for example, if the n-word

is the subject, then the negative marker is not needed, as in example (10). In fact in

these cases, if the negative marker and a preverbal n-word appear, then the DN reading

arises, (11).

(10) Nessuno ha arrivato Italian
nobody has.3s arrived
'Nobody arrived'

(11) Nessuno no ha arrivato
nobody NEG has.3s arrived
*'Nobody arrived'
'Nobody didn't arrive' (i.e. everybody did arrive)


198


Negative Concord











A preverbal n-word is sufficient to license a postverbal n-word in non-strict NC

languages, and example (12) shows this with nadie licensing nunca and nada.


(12) Nadie nunca hace nada Spanish
Nobody never does.3s nothing
'Nobody ever does anything'


Also in non-strict languages, the n-word needs to be an independent DP constituent

(Giannakidou 2006). If the n-word is a sub-part of a DP constituent then a double

negation reading may arise, or at the least the NC reading isn't readily available. The

example below from Giannakidou (2006:354) shows that when nessuno is inside a DP,

the sentence is very degraded. This is different from a strict NC language because in

strict NC languages, in all cases, if there is an n-word there must be a sentential

negative marker.


(13) ??Nessuno student ha letto nessun libro Italian
No student had.3s read no book
'No student read any book'


Strict and Non-strict NC languages also differ in their strategies for forming

negative imperatives. Strict NC languages may allow what are called true negative

imperatives (Zeijlstra 2008). True negative imperatives are formed from positive

imperatives that are negated using the typical negative marker (Zanuttini 1997, Han

2001, Zeijlstra 2008). This is seen in the example below from Polish (Zeijlstra 2008:24).


(14) Pracuj! Polish
Work.2s.IMP
'Work!'


199










(15) Nie pracuj!
NEG work.2S.IMP
'Don't work!'


While strict NC languages allow true negative imperatives, non-strict languages do not

have true negative imperatives in (17) (Zeijlstra 2008). Instead of true negative

imperatives, suppletive imperatives are formed using non-imperative morphology,

generally the subjunctive form of the verb, as seen in the Spanish example in (18).


(16) Lee eso Spanish
Read.IMP.2s that
'Read that!'

(17) *No lee eso!
NEG read.IMP.2SG that

(18) No leas eso
NEG read.suB.2s that
'Don't read that!'


This section has shown two types of NC languages: strict and non-strict.

Following traditional definitions I have shown that strict NC languages must always have

a negative marker present when an n-word is used. Additionally, strict NC languages

allow true negative imperatives. Non-strict NC languages allow subject n-words to exist

without the presence of a negative marker. Negative imperatives in non-strict

languages are formed using suppletive morphology instead of imperative morphology.

7.2.2 N-Words are Negative Quantifiers

As mentioned above, there are two main claims about the inherent negativity of n-

words. The first hypothesis claims that n-words are negative. A principal proponent of

this idea is Zanuttini (1991), supported by Haegeman & Zanuttini (1996), who claims


200









that n-words works much like wh-words. First, much like NC, multiple wh-words are

interpreted as one question where each wh-word binds a different variable. These two

wh-binders become one in the syntax, much like the interpretation of one negative

operator in NC. Second, like wh-movement, movement of a negative operator to the left

periphery causes subject-auxiliary inversion, such as never had I been so scared.

Third, both wh-words and n-words license negative polarity items (NPI), such as

anybody in examples (19) and (20).


(19) Who loves anybody?

(20) Nobody saw anybody.


Based on this evidence, Haegeman & Zanuttini (1996) determine that the interpretation

of n-words is like that of wh-words. They propose a negative absorption rule based on

Higginbotham & May's (1981) wh-absorption rule (Giannakidou 2006:334). This rule is

(21) and states that every instance of negation coalesces to one instance of negation at

LF. Thus, for Zanuttini (1991), n-words are negative quantifiers that go through a

process of neg-absorption31 in a NC situation. Example (22) has the neg-absorption

rule where there is one n-word and a negative operator, i.e. a sentential negative

marker.


(21) [Vx-][Vy-][Vz-]=[Vx,y,z]
(22) [Vx-]- = [Vx]-



31 Another part of this is factorization. Factorization is what "happens when two quantifying elements
raise to the same projection under quantifierr raising] in order to turn from monadic quantifiers into one n-
ary polydic quantifier. In the case of WH, polyadic quantification take place after factorization, whereby
the interrogative operator of the second wh-element is transmitted into the first wh-operator (Zeijlstra
2004:193). This process for Haegemann & Zanuttini (1996) would be the same for negative operators.


201









For the syntax, they posit a Neg-Criterion for n-words, similar to Rizzi's (1991) Wh-

Criterion for wh-phrases.


(23) Neg-Criterion
a. A NEG-operator must be in a Spec-head configuration with X [NEG]
b. An X [NEG] must be in Spec-head configuration with a NEG-operator
Whereby the following definitions hold:
NEG-operator: a negative phrase in scope position
Scope position: left-peripheral A'-position [Spec,XP] or [YP,XP].



What the Neg-Criterion does is force all n-words to the specifier of some negative head.

Notice that the sentential negative marker, in many cases, is the head of NegP (see

chapter four, Namiuti 2008, and Mioto 1992 for the position of nao in BP). Under this

analysis, an n-word is a negative quantificational operator and it moves to spec,NegP to

form a spec-head relationship with Neg.32 This is done so that the negative quantifiers

have scope over the sentence and negation. By moving all the n-words to spec,NegP,

the result is essentially (21), at which point neg-absorption can take place over the

syntactic representation.

N-words contrast with negative polarity items like any- in English. N-words are

allowed to be fragment answers, as stated above (Zanuttini 1991, Giannakidou 2006),

but NPIs are not. Example (24) shows that the NPI anybody is not licit as a fragment

answer, whereas, the negative nobody is. Rather, polarity items like any must be

licensed by some operator (Moscati 2006).


32 If there are no n-words in a negative clause, a sentential negative marker would be the head of NegP
and this means that for the Neg-Criterion to be satisfied, some null negative operator must be in spec,
NegP.


202










(24) a. Who came to the party?
b. *anybody/nobody


Some problems arise with Zanuttini's analysis. First, Giannakidou (1997) shows

that the relationship between wh-words and n-words is not as uniform as Zanuttini

claims. She explains that there are also empirical problems equating subject-auxiliary

inversion in these two instances. For example, not every language has subject-auxiliary

inversion in negative fronting, such as Spanish and Portuguese, even though it is still

used with wh-questions.

Zeijlstra (2004) points outs other problems related to the the law of DN. Recall

that one of the reasons that Zanuttini & Haegeman argue for the negative quantifier

account of n-words is because of their similarities to wh-phenomena. Zeijlstra explains

that while multiple wh-words cross-linguistically are interpreted at LF as a single wh-

operator, there is a great deal of variation with regards to multiple negative operators.

Because n-words are negative quantifiers, the law of DN should apply to all n-words as

it does in English. Zeijlstra states that "negative quantifiers" obey the law of DN only in

a restricted set of languages; this is an indicator that n-words in NC languages are not

semantically negative in the same way that they are in DN languages (2004:199).

Finally, as Pereira (2000) points out, there are instances where n-words are not

negative, and vice versa, where non-n-words are negative. Herburger (2001) gives

(25) as an example where an n-word is interpreted as positive. If the n-word was a

negative quantifier, then it should have a negative interpretation, contrary to fact.. At the

same time, there are examples as in (26) (Bosque 1980 and Herburger 2001) where

non-negative phrases are interpreted as negative.


203










(25) Antes de hacer nada, debes lavarle las manos
before PREP do.INF n-thing must.2S wash.INF.CL DET hands
'Before doing anything, you should wash his hands.'
(Herburger 2001:297)

(26) En modo alguno se puede tolerar tal actitud!
in manner some CL AUX.3s tolerate.INF such attitude
'Under no circumstances can one tolerate such an attitude!'
(Herburger 2001:293)


Herburger also points out that in many cases n-words serve as negative polarity items

(NPI). She gives example (27) using sin 'without', a downward entailing preposition, to

show n-words behave like NPIs. This behavior of NC n-words is similar to the behavior

of the English word any, assumed to be a negative polarity item, and not similar to

negative quantifiers. Notice that sin has a different interaction with the sentential

negative marker no because these two together cause a double negative reading, (28)

(Herburger 2001:297).


(27) Pedro compr6 el terreno sin contarselo a nadie
Pedro buy.PST.3S DET land without tell.INF.CL.CL PREP n-body
'Peter bought the land without telling anybody'


(28) Pedro compr6 el terreno sin no contarselo a nadie
Pedro buy.PST.3S DET land without NEG tell.INF.CL.CL PREP n-body
'Peter bought the land without not telling anybody'
i.e. Peter bought the land, and told everyone.


7.2.3 N-Words are Non-Negative

Ladusaw (1992) is a strong proponent of a second hypothesis concerning the

negativity of n-words. In response to Haegeman & Zanuttini (1991), he claims that n-

words are non-negative indefinites, and he argues that Haegeman & Zanuttini do not

have enough semantic support for a negative quantifier account of n-words.


204









Furthermore, he claims that n-words do not have any quantificational force in their own

right. Ladusaw concludes that n-words are unbound variables, equating them to NPIs.

While Ladusaw admits that n-words and NPIs are different, the principles that govern

each are similar.

Ladusaw claims that n-words are only licensed if the specifier position or the head

position of NegP is filled. In the case of strict NC languages, this requirement is always

satisfied by the sentential negative marker. For non-strict NC languages, this

requirement is satisfied by a sentential negative marker, most likely in Neg, or by some

n-word in the specifier position of NegP. An abstract element appears in non-strict NC

languages when an n-word is preverbal.


(29) Non ha telefonato a nessuno
[NegP [ Neg non ][TP ha telefonato a nessuno]
'Nobody has called'

(30) Nessuno ha telefonato a nessuno
[NegP Nessuno [ Neg OP ][TP ha telefonato a nessuno]
'Nobody called anyone'


Another approach to n-words as non-negatives comes from Giannakidou's (1998,

2000) work. She makes similar claims to Ladusaw's in that she takes n-words in NC

languages to be non-negative. Her approach differs in that she claims that n-words are

NPIs. This helps distinguish between DN languages and NC languages. Giannakidou's

(2006) approach is different from her previous works in that she does not claim that n-

words are necessarily indefinites. Rather, she believes that in some NC languages n-

words are existential quantifiers and in others they can be universal quantifiers. She


205









also concludes that within a language n-words might be ambiguous between universal

readings and existential readings (Giannakidou 2006).

An apparent problem with this approach is that preverbal n-words and n-word

fragment answers in non-strict languages have a negative meaning. In both of these

cases, there is no negative operator to license the negative meaning. Take the Spanish

example repeated from above:


(31) a. Quien hizo la tarea?
who did.3s DET homework
'Who did the homework?'

b. Nadie
'nobody'


Giannakidou (2000) claims that looking at fragment answers in this way is deceptive.

She proposes that these are really instances of ellipsis, similar to conjunctions or

disjunctions. Take the Spanish sentence from Zeijlstra (2004:212) where the disjunct is

negative but doesn't have sentential negative marker to license it. In this case, what

has happened, according to Zeijlstra, is that part of the elided material is the negative

marker. Thus, the sentence in (32) comes from (33). The possibility of fragment

answers would follow from the same logic.


(32) Me caso contigo o con nadie
1s marry.1s PREP.2S or PREP nobody
'I marry you or nobody'

(33) Me caso contigo o no me caso con nadie
1s marry. 1SG PREP.2S or NEG 1s marry. 1s PREP nobody
'I marry you or I marry nobody'


206









Despite this explanation, Zeijlstra (2006) concludes that, based on Wantabe

(2005), this kind of ellipsis is not allowed because the elided material does not have the

same semantic identity. That is, the elided XP contains a negative marker that is not

present in the first part of the sentence. Giannakidou (2006) explains that this isn't a

problem for her analysis, citing Merchant (2001).

7.2.4 Zeijlstra's Feature-Based Approach to Negative Concord

Continuing in the spirit of Ladusaw and Giannakidou, Zeijlstra (2004, 2006, 2008)

claims that n-words are non-negative in a traditional sense but that this claim should be

viewed within a minimalist perspective and understood syntactically. I follow Zeijlstra's

analysis of n-words and NC. First, I lay out some general assumptions that Zeijlstra

makes. As a way of showing how his theories work, I proceed to apply his analysis to a

non-strict NC language.

Zeijlstra's goal is to show that NC can be more easily understood if it is seen

through the lens of minimalist feature agreement. He claims that both n-words and

sentential negative markers have some [NEG] feature (Chomsky 2000) that needs to be

checked/matched by something in the syntax with the same feature. Recall Chomsky's

claim (2000:3):


(34) 'We therefore have a relation Agree holding between a and /, where a has
interpretable inflectional features and / has uninterpretable ones, which delete under
Agree.'


Zeijlstra sees the benefit of feature agreement for the semantic properties of person and

number that are both on the subject and the verb. Therefore, he applies the same

principle of Agree to negation, and he subsequently formulates a theory for NC based

on the agreement of interpretable and uninterpretable features (2008:20):


207










(35) 'NC is an Agree relation between a single feature [iNEG] and one or more
features [uNEG].'


(35) is based on three critical assumptions about feature agreement and NC. First,

certain elements of NC languages are only 'formally' negative, and although they share

some of the morpho-syntactic properties of negation, they themselves are not negative

(Ladusaw 1992). These n-words carry an [uNEG] feature that must be checked.

Syntactically, they introduce a free variable, making them indefinites (Heim 1982).

Second, Zeijlstra follows the proposal of Multiple Agree (Hiraiwa 2001). Multiple

Agree is where a single interpretable feature may establish an Agree relationship with

multiple elements with the same uninterpretable feature. The same conditions on Agree

must still be met; however, Zeijlstra departs from the standard version of Agree

introduced in chapter two in one notable way. In contrast to the claim from chapter two

that the uninterpretable probe must c-command an interpretable goal, feature checking

for NC works in a top-down manner, and the element with the [iNEG] feature must c-

command the [uNEG] elements in the derivation (Adger 2003).

Third, the element with the [iNEG] feature may be covert. If some overt element

with an [uNEG] feature is present, there must be a corresponding element with [iNEG]

even if it is not phonologically realized. In Zeijlstra's analysis, this element is a negative

operator, whose position can vary across languages.

Based on these assumptions, Zeijlstra proceeds to show how this application of

Agree can account for not only strict and non-strict NC languages, but also for DN

languages, such as Standard English. Because BP is a non-strict NC language, I do

not treat strict NC languages nor DN languages here. However, where DN arises, both


208









the negative marker and the n-words have [iNEG] features. Hence, the DN reading in

English is to due the fact that both the n-word and the negative marker are interpretable

and no feature checking is involved. Likewise, strict NC language are languages whose

sentential negative marker and n-words all carry an [uNEG] feature and all sentences

with negation require a abstract negative operator with an [iNEG] feature.

Zeijlstra explains that non-strict NC languages have an [uNEG] feature on n-words

and that the sentential negative marker has an [iNEG] feature. To obtain a DN reading

with subject n-words, he posits an abstract negative operator with an [iNEG] feature in

sentences containing a preverbal n-word. The position of the abstract negative operator

is not in spec,NegP but rather some position above AgrP.

Italian serves as a proto-typical non-strict language. Non is sufficient to provide

semantic negation because it has an [iNEG] feature. In (36) the [iNEG] feature c-

commands and checks the [uNEG] feature of the n-word, resulting in a grammatical

sentence.


(36) a. Gianni non telefona a nessuno Italian
Gianni NEG call.3SG PREP nobody
'Gianni doesn't call anybody'


b. [IP Gianni [NegP non[iNEG] telefona [vP a nessuno[uNEG]]]]



With preverbal n-words, a negative operator with an [iNEG] feature is present and

checks the [uNEG] features of the subject n-word. The merger of the null operator only

occurs in sentences with a subject n-word-an important stipulation that I adopt without

comment.


209









(37) A. Nessuno ha arrivato
n-body has.3s arrived
'Nobody arrived'

B. [IP Op- [iNEG] [IP nessuno [uNEG] [I' ha arrivato [VP...]]]]


The position of the negative operator is not fully defined in Zeijlstra's work. Crucially, it

must be above IP so that it can check an [uNEG] feature on subject n-words. In what

follows, I show it adjoined to IP/AgrP.

The null operator appears in sentences with a subject n-word and sentential

negation, (38). In this case, both the null operator and the negative marker have an

[iNEG] feature, and the two interpretable negative features in the sentence lead to a

double negation reading.


(38) Op- Nessuno no ha arrivato
n-body NEG has.3s arrived
*'Nobody arrived'
'Nobody didn't arrive' (i.e. everybody did arrive)

Finally, as noted above, there are cases where an n-word in non-strict NC

language is apparently not licensed by anything. In Zeijlstra's model, this means that

the [uNEG] feature is not getting checked. Recall the Spanish example from above:


(39) Antes de hacer nada, debes lavarle las manos
before PREP do.INF n-thing must.2S wash.INF.CL DET hands
'Before doing anything, you should wash his hands.'


According to Zeijlstra's proposal, the only way that this could be allowed is if something

was checking the [uNEG] feature of nada. He claims that in these situations, the null

negative operator is again inserted.


210









(40) [[NegP Op-[iNEG][PP Antes de hacer nada,]][TP debes lavarle las manos]]]
NEG before PREP do.INF n-thing must.2S wash.INF.CL DET hands
'Before doing anything, you should wash his hands.'


7.2.5 Summary

This section has discussed two main approaches to understanding what n-words

are. The first hypothesis is that n-words are negative universal quantifiers; this is

Zanuttini's (1991) approach. The second hypothesis is the one in Ladusaw (1992),

Giannakidou (1998, 2000), and Zeijlstra (2004, 2008), which claims that n-words are

non-negative existential or universal quantifiers. I presented Zeijlstra's implementation

of this hypothesis in which n-words are non-negative but have an uninterpretable

[uNEG] feature that must be checked under Agree against an interpretable [iNEG]

feature.

7.3 N-Words and Negative Concord in Brazilian Portuguese

I now discuss n-words and NC in BP. Section 7.3.1 argues that BP is a non-strict

NC language. Section 7.3.2 applies Zeijlstra's theory to BP.

7.3.1 BP N-Words

Following Giannakidou's formal definition of n-words, given in (3), the following BP

words are n-words: ningu6m 'nobody', nada 'nothing', nunca 'never', nenhum 'no/any',

nenhuma pessoa 'nobody' (Peres 2000). Based on examples (41) through (43), this

chapter assumes that BP is a non-strict NC language as defined in section 7.2.1.

Example (41) shows the non-strict nature of BP n-words where a single n-word can

occur preverbally as the subject and carry a negative meaning without the presence of a

sentential negative marker.


211









(41) a. Ninguem
Nobody


'nobody came'


veio
came.3s


conseguia


Nothing was.capable.IMPERF.3s
'Nothing was able to solve the problem'


c. Ela nunca
3sF never
'She never calls me'


resolver


o problema


resolve.INF DET problem


me liga
1s calls.3s


d. Nenhum remedio curou
No medicine healed.3s
'No medicine healed the patient.'


As with Spanish and Italian, if there is a preverbal negative marker, a DN reading

arises.33


(42) Ninguem nao ajudou a Sienna
Nobody NEG helped.3s DET Sienna
'Nobody didn't helped Sienna (i.e. everybody helped Sienna)'


The examples in (43) show that there must be a preverbal sentential negative marker

when BP n-words are in a postverbal position.34




33 Martinez (2006) states that the following is allowed in Mineiro Portuguese (State of Minas Gerais)

i. Ningu6m nao veio
Nobody neg came
'Nobody came'

While she points to other studies, Alkmim (2001), this is the only example of the n-word and the preverbal
negative marker not invoking a double negative reading. All the preliminary work suggests that this is
dialectal and is not seen in most dialects of BP. If it turns out that this claim is false, then more analysis
would be required.
34 Cavalcante (2007) suggests that there are instances where the postverbal n-word does not co-occur
with a preverbal negative marker. This appears to be dialectal and restricted to the northeast of Brazil.
This could be due to BP n-words being inherently negative, contrary to what I assume above (Pereira
2000, Peres 2000, Matos 2003, Giannakidou 2006).


212


b. Nada


o doente
DET sick. person










A LOcia
DET LOcia
'LOcia didn't


b. A LOcia
DET LOcia
'LOcia didn't

c. A LOcia
DET LOcia
'LOcia didn't

d. A LOcia
DET LOcia


*(nao)
NEG
see anybody'


viu
saw.3s


*(nao) leu
NEG read.3s
read anything'

*(nao) leu
NEG read.3s
read any books'


*(nao)
NEG


me liga
1s read.3s


ninguem
nobody


nada
nothing


nenhum
no


livro
book


nunca
never


'LOcia doesn't ever call me'


Like other non-strict NC languages, a preverbal n-word is sufficient to license postverbal

n-words, as in (44). Like NC languages in general, BP allows several n-words to be

present in the sentence without any double negation reading, (45).


(44) Ninguem leu
Nobody read.3sG
'Nobody read any books'


nenhum livro
none book


(45) Ninguem nunca deu nada para ninguem em lado nenhum.
Nobody never gave.3sG nothing PREP nobody PREP side none
'Nobody ever gave anything to anybody anywhere.'


The second part of Giannakidou's definition deals with fragment answers.

Although I give only two examples, all of the words just given are valid fragment

answers in BP.


Quem fez
who did.3s
'who did the work'


o trabalho?
DET work


b. Ninguem
'Nobody'


213


(43) a.


(46) a.










(47) a. 0 que que voc6 fez?
What that 2s did.2s
'What did you do?'

b. Nada
'Nothing'

BP departs from the behavior of traditional non-strict languages, such as Spanish and

European Portuguese, in imperatives. Recall that according to Zeijlstra, strict NC allows

true negative imperatives, while non-strict languages do not. Although BP clearly

patterns with non-strict languages in other ways, it does allow true negative imperatives.

Example (48) has an imperative, and the next example, (49), negates the imperative by

simply adding the negative marker. Note however, that BP has both true negative

imperatives and suppletive imperatives, example (50) (Scherre 2007).


(48) L6 isso
Read.2sG.IMP that
'Read that!'

(49) Nao Il isso
NEG Read.2sG.IMP that
'Don't read that!'

(50) Nao leia isso
NEG Read.2sG.suBJ that
'Don't read that!'


Another place where BP differs form other non-strict NC languages is if the n-word

is part of a DP constituent. For other languages in this group, these types of sentences

are not grammatical, as shown in example (13) for Italian. BP has no problem with

these types of sentences, and in each case a NC reading is available.


214










(51) Nenhum estudante leu nenhum livro
no student read.3sG no book
'No student read any book'


Even though the evidence appears to leave the status of BP as a non-strict NC

language open to debate, I will assume that it is a non-strict NC language. My reason is

that preverbal n-words in BP cannot co-occur with the sentential negative marker

without yielding a DN reading. This seems to be the most important criterion and it

suggests that n-words in BP behave much like those in other Romance languages

(Pereira 2000, Zeijlstra 2008).

7.3.2 Feature Agreement and BP N-Words

In this section I apply Zeijlstra's analysis to BP.35 Following his analysis of non-

strict NC languages, I assume that BP n-words have an [uNEG] feature. BP works the

same as Italian, described above in (36) and (37). In (52), the negative marker has an

[iNEG] feature which checks the [uNEG] feature on the n-word ningu6m. (53) gives a

syntactic representation.


(52) Joao nao ligou para ninguem
John NEG calls PREP nobody
'John doesn't call anybody'

(53) AgrP

DP NegP

Jo0o n5o[iNEG] TP

ligou para ninguem[uNEG]

35 Zeijlstra, Giannakidou, Zanuttini all analyze Romance languages but in most cases their discussions of
Portuguese are based on European Portuguese. As such, some of the conclusions that they come up
with do not directly apply to Brazilian Portuguese.


215











The following example show that preverbal n-words in BP behave like n-words in

Italian, triggering the introduction of a null negative operator that carries an [iNEG]

feature.

(54) a. Ninguem veio
Nobody came.3s
'nobody came'

(55) AgrP

Op-[iNEG] AgrP

DP Agr'

Ninguem[uNEG] veio

As in other non-strict languages, the phenomenon of negative spread is also captured in

BP through Zeijlstra's adoption of Multiple Agree. The next sentence has several n-

words, all of which check their uninterpretable feature with the null operator.


(56) Ninguem nunca deu nada para ninguem.
Nobody never gave.3sG nothing PREP nobody
'Nobody ever gave anything to anybody.'

(57) AgrP

Op-[iNEG] AgrP

DP Agr'

Ninguem[uNEG] nunca[uNEG] deu nada[uNEG] para ninguem[uNEG]


216









The co-occurrence of the null operator and the sentential negative operator in BP

causes a double negation reading. This is due to the presence of two [iNEG] features in

the sentence.

(58) Ninguem nao ajudou a Sienna
Nobody NEG helped.3s DET Sienna
'Nobody didn't helped Sienna (i.e. everybody helped Sienna)'

(59) XP

Op-[iNEG] AgrP

DP Agr'

Ninguem[uNEG] Agr NegP

Neg TP
n5o[iNEG]
ajudou a Sienna

7.4 Negative Concord and Clause-Final Negation in Brazilian Portuguese

This section returns to the discussion that began at the end of chapter six where

questions arose regarding the semantic contributions of CFN and the different clause-

final naos in T2 and T3. Specifically, I argued that the CFN marker in T2 has a [uNEG]

feature, which equates it with n-words. Given Zeijlstra's system, this [uNEG] feature is

checked when AgrP bearing an interpretable [iNEG] moves to spec,TopP and c-

commands the Top head. The T3 nao is used as a negative element in denials which

can negate a sentence without any support. Here the negative marker has an [iNEG]

feature, much like the preverbal negative marker, and does not need to be checked.

My analysis of BP NC supports the claims of chapter six. Recall the syntactic

environments in which CFN licenses an n-word:


217









Table 7-1. N-Words and CFN
Environment
Allowed with Subject N-words

Allowed with Object N-words

Compatible with NPI Idiomatic Expressions


T1 T2
X X

-\ -\


S X


The table above indicates that CFN has no effect on the licensing of n-words. The

licensing picture is the same whether CFN is present or not. If the n-word is postverbal,

then preverbal negation must be present and CFN can be present or not, i.e. an object

n-word is licensed by T1 or T2 but not T3. When the n-word is the subject, preverbal

negation cannot be present and CFN can be present or not, i.e. a subject n-word cannot

appear with T1 and T2, only with T3. These patterns are illustrated by the following

examples:


(60) A Sienna nao ajudou ninguem
DET Sienna NEG helped.3s nobody
'Sienna didn't help anyone'

(61) Ninguem ajudou o Abilio
Nobody helped.3s DET Ablilio
'Nobody helped Ablilio'


(nao)
NEG


(nao)
NEG


In the structure in (62), corresponding to (60), the [iNEG] feature of preverbal nao in

Neg checks the [uNEG] feature on the postverbal n-word. The [iNEG] feature that

percolated to AgrP checks the [uNEG] on the CFN in Top.


218










(62) TopP


AgrP[iNEG,iTOP]i Top'

DP Agr' Top AgrP
A no[uNEG,uTOP*] ti
A Sienna Agr NegP

Neg[iNEG] TP
n~o
T vP
ajudou
...ninguem[uNEG]

In (63), corresponding to (61), the subject n-word's [uNEG] feature is checked by the

null operator. The CFN's [uNEG] feature is again checked by the [iNEG] feature on

AgrP.


(63) TopP

AgrP[iNEG,iTOP]i Top'

Op-[iNEG] AgrP Top AgrP
Sno[uNEG,uTOP*] ti
DP Agr'

Ninguem[uNEG] Agr TP

T vP
ajudou
...o Ablilio


This analysis equates n-words and T2 CFN by suggesting that both have an

[uNEG] feature that must be checked by an [iNEG] feature elsewhere in the structure.

As such T2 is a case of negative concord.


219









The theory also correctly predicts that T3, with [iNEG] in Top, will not license an

object n-word. After AgrP moves, the Top head, even though it has an [iNEG] feature,

does not c-command any n-words inside AgrP. Consequently, their [uNEG] features will

remain unchecked. I demonstrate this claim by showing how it blocks the

ungrammatical example in (64). As can be seen in the structure below, Top does not c-

command and therefore does not check the n-word's [uNEG] feature.

(64) *O Ze ajudou ninguem n5o
DET Ze helped.3s nobody NEG
'Ze helped nobody'


(65) *TopP

AgrP[uTOP]i Top'

DP Agr' Top ti
A nao[iNEG,uTOP*]
O Ze Agr TP

ajudou ninguem[uNEG]


The ungrammaticality of (64) thus has the same explanation as T3's inability to

license an NPI idiom that we saw in chapter six.

Unfortunately, the same DN problem from chapter six arises in these structures.

Because there are two CFN markers, either one could be inserted into the derivation.

However, if the T3 negative marker, which is [iNEG], were inserted into the derivations

in (62) and (63), then a double negative reading would incorrectly result as the clause

would have two [iNEG] features. This is the same problem from chapter six which I

have not solved. Neverless, it seems necessary to keep T2 and T3 distinct analytically

because we saw that T3 is subject to a stricter discourse restriction than T2.


220










7.5 Conclusion

This chapter has discussed NC in BP. The data show that BP is a non-strict NC

language similar to Spanish and Italian (Zeijlstra 2004). I followed Zeijlstra in analyzing

the licensing of n-words as an instance of syntactic Agree in which an [iNEG] must c-

command and check any n-word's [uNEG] feature. Three interpretable elements are in

principle available to do this checking: preverbal negation, a null negative operator

which is present just in case an n-word is preverbal, and CFN ndo[iNEG] in Top. In

practice however, CFN cannot license an n-word on its own because it does not c-

command into AgrP once AgrP has moved to spec,TopP. The analysis of CFN from

chapter six thus correctly accounts for the interaction of CFN and n-word licensing.

CFN has no affect on n-word licensing.36






36 Of particular interest to future work is question of diachronic change of n-words. Some BP dialects vary
on the level of Negative Concord. Consider the following examples from Cavalcante (2007) where the n-
word is not licensed by any c-commanding negative element:

(i) Fui cobrar nada dele
Went.ls charge. INF nothing PREP.3S.M
'I went to get nothing'

(ii) Veio ninguem
Came.3s nobody
'Nobody came'

(iii)Tou sabendo de nada
Am knowing PREP nothing
'I know nothing'

Since some languages change from strict negative concord languages to double negation
languages, it is interesting to understand this process in more detail and make some cross-dialectal
comparisons within BP. Also of interest is the frequency of clause-final negation in dialects with
different types of n-words. This work claims that clause-final negation is found in all sectors of
Brazil; however, there are definite regional differences concerning n-words that should be examined.
Specifically, the relationship between n-words and clause-final negation in northern and northeastern
dialects appear to suggest a relationship between frequency of clause-final negation and [iNEG]
carrying n-words.


221









CHAPTER 8
CONCLUSION

In this work, I sought to understand Clause-Final Negation in Brazilian Portuguese

within the Minimalist Program. Clause-Final Negation (CFN) is a somewhat rare form of

negation, and this is especially so within the Indo-European family of languages.

Brazilian Portuguese has three types of negation and two of these include clause-final

negative markers. CFN is thus a part of Brazilian Portuguese that is in need of study.

The work began in chapter one by laying out the data and setting the stage for a

discussion that involved both syntactic as well as semantic/pragmatic elements.

Chapters two, three, and four furthered the discussion by implementing a theoretical

framework and clause structure for BP. In these chapters I gave evidence for the

clause structure of BP that is needed to understand negation generally and CFN in

particular.

Chapter five discussed the semantic and pragmatic restrictions on the use of CFN.

I showed, following Schwenter, that CFN can only be fully understood through a proper

delineation of the discourse factors that influence the felicity of CFN sentences.

Chapter six offered a syntactic analysis of CFN. I argued that the negative marker

ndo comes to be clause-final when the entire clause, AgrP, fronts to the left nao, which

is a topic head in the left periphery. The movement of AgrP yields a topic interpretation

for the clause. The analysis opens the door to further examination of topicalization in

BP. In particular, this work provides evidence that large phrases can be topicalized.

This discussion of clause topicalization promises to be an area which will show an even

tighter link between discourse and syntax.


222









Chapter six posited two CFN markers. One is interpretable, bearing an [iNEG]

feature, and the other is uninterpretable, being specified as [uNEG]. The need for two

heads motivated three observations: 1) partially distinct discourse licensing restrictions

on T2 versus T3, 2) the need to have a semantically negative interpretation in T3

sentences, and 3) the need to avoid a Double Negation reading in T2 sentences. An

interpretable head with [iNEG] is used in T3 environments in order to give such

sentences a negative interpretation. On the other hand, the uninterpretable head with

[uNEG] was required in order to prevent double negation readings with T2.

Unfortunately, this picture could still not prevent the T3 head from also being used in

combination with preverbal negation, resulting in a double negation reading. I left this

problem for future work.

Chapter seven closed the discussion by testing the theory from chapter six in the

domain of Negative Concord (NC). I applied Zeijlstra's theory of NC to BP and the

theory in conjunction with the analysis of CFN from chapter six was able to explain why

CFN does not have any influence on the licensing of n-words in BP.


223









LIST OF REFERENCES


Adger, David. 2003. Core syntax: A Minimalist approach. New York: Oxford
University Press.

Adragao, Maria do Mar and Joao Costa. 2004. On the Status of Preverbal
Subjects in Null Subject Languages: Evidence from Acquisition. In Proceedings
of GALA 2003, ed. Jacqueline van Kampen and Sergio Baauw, 69-80. The
Hague: Utrecht University, Lot Publications.

Aissen, Judith. 1992. Topics and Mayan. Language 68, 1:43-80.

Alexiadou, Artemis and Elena Anagnostopoulou. 1998. Parametrizing Agr: Word
order, Verb-Movement and EPP-checking. Natural Language and Linguistic
Theory 16:491-539.

Alkmim, Monica. 2001. As negatives sentenciales no dialeto mineiro: uma
abordagem varacionista. Doctoral dissertation, Universidade Federal de Minas
Gerais.

Ambar, Manuela. 1999. Aspects of the Syntax of Focus in Portuguese. In The
Grammar of Focus, eds. Georges Rebuschi and Laurice Tuller, 23-53.
Amsterdam/Philadelphia: John Benjamins.

Aoun, Joseph and Audrey Li .2000. Scope, structure, and expert systems: A
reply to Kuno et al. Language 76:133-55.

Armstrong, Meghan E. 2008. Pragmatic Restrictions on Affirmative Response
Choice in Brazilian Portuguese. In Selected Proceedings of the 10th Hispanic
Linguistics Symposium, ed. Joyce Bruhn de Garavito and Elena Valenzuela,
288-299. Somerville, MA: Cascadilla Proceedings Project.

Azevedo, Milton. 2005. Portuguese. A linguistic introduction. Cambridge, United
Kingdom: Cambridge University Press.

Barbosa, Pilar. 1995. Null Subjects. Doctoral dissertation, Massachusetts
Institute of Technology.

Barbosa, Pilar. 2004. Two Kinds of Subject pro. Paper presented at the 14th
Colloquium of Generative Linguistics, University of Oporto.

Barme, Stefan. 2000. Negagao (e afirmacgo) no portugues brasileiro falado
informal. ms.,University of Mainz.

Barss, Andrew. 1986. Chains and anaphoric dependence. Doctoral dissertation,
Massachusetts Institute of Technology.


224









Bastos, Ana. 2001. Fazer, eu fago! Topicalizacgo de Constituintes Verbais em
Portugues Brasileiro. Master Thesis, Universidade de Campinas.

Bastos-Gee, Ana. 2009. Topicalization of Verbal Projections in Brazilian
Portuguese. In Minimalist Essays on Brazilian Syntax, ed. Jairo Nunes, 161-190.
Amsterdam: John Benjamins.

Belletti, Adriana. 1990. Generalized Verb Movement. Turin: Rosenberg e Sellier.

Belletti, Adriana. 2004. Aspects of the Low IP Area. In The Structure of CP and
IP. The Cartography of Syntactic Structures, Vol. 2, ed. Luigi Rizzi, 16-51.
Oxford: Oxford University Press.

Beninca, Paola, 2001. The position of topic and focus in the left periphery. In
Current Studies in Italian Syntax. Essays Offered to Lorenzo Renzi, ed. Cinque,
Guglielmo and Giampaolo Salvi, 39-64. Amsterdam: Elsevier.

Beninca, Paola and Cecilia Poletto. 2004. Topic, Focus and V2. Defining the CP
sublayers, In The Structure of CP and IP. The Cartography of Syntactic
Structures, Vol. 2, ed. Luigi Rizzi, 52-75. Oxford: Oxford University Press.

Bernini, Giuliano and Paolo Ramat. 1996. Negative Sentences in the Languages
of Europe: A Typological Approach (Empirical Approaches to Language
Typology 16). Berlin: Mouton de Gruyter.

Besten, Hans den. 1986, Double Negation and the Genesis of Afrikaans. In
Substrata versus Universals in Creole Languages, ed. Pieter Muysken and
Norval Smith, 185-230. Amsterdam: John Benjamins.

Biberauer, Theresa. 2009. Jespersen off course? The case of contemporary
Afrikaans negation. In Cyclical Change, ed. Elly Van Gelderen, 91-132.
Amsterdam: John Benjamins.

Biberauer, Theresa and Sonia Cyrino. 2009. Negative developments in
Afrikaans and Brazilian Portuguese, paper presented at the 19th Colloquium on
Generative Grammar, University of the Basque Country, Vitoria-Gasteiz.


Bobaljik, Jonathan. 2003. Floating Quantifiers: Handle With Care. In The Second
Glot International State-of-The-Article Book, eds. Lisa Cheng and Rint Sybesma,
107-148. Berlin: Mouton de Gruyter.

Bobaljik, Jonathan and Hoskuldur Thrainsson. 1998. 'Two Heads Aren't Always
Better than One', Syntax 1.1:37-71.


225









Boskovi6, Zeljko. 1999. On multiple feature checking: multiple Wh-fronting and
multiple head-movement. In Working Minimalism, ed. Samuel Epstein and
Norbert Hornstein, 159-87. Cambridge: MIT Press.

Boskovi6, Zeljko and Howard. Lasnik. 2007. Minimalist Syntax: The Essential
Readings. Oxford: Blackwell.

Bosque, Ignacio. 1980. Sobre la Negaci6n. Madrid: Catedra.

Bouquiaux, Luc. 1970. La Langue Birom. Paris: Bibliotheque de la
Faculte de Philosophie et Lettres de I'Universite de Liege.

Brito, Ana Maria. 2001. Clause Structure, Subject Positions and Verb Movement.
In Romance Languages and Linguistic Theory. ed. D'Hulst et al., 63-86.
Amsterdam: John Benjamins.

Camargos, Marcelo Luiz de. 2002. Teoria de C6pia e Negagao. Masters thesis,
Universidade Federal de Minas Gerais.
http://www. ufop. br/ichs/conifes/anais/LCA/clca03. htm

Cardinaletti, Anna. 1997. Subjects and Clause Structure. In The New
Comparative Syntax ed, Liliane Haegeman. London: Longman.

Cardinaletti, Anna. 2004. Towards a cartography of subject positions. In The
Structure of CP and IP. The Cartography of Syntactic Structures, vol. 2, ed. Luigi
Rizzi. Oxford and New York: Oxford University Press

Cardinaletti, Anna & Michal Starke. 1999. The Typology of Structural Deficiency:
A Case Study of the Three Classes of Pronouns. In Clitics in the Languages of
Europe, ed. Henk van Riemsdijk, 145-234.Berlin: Walter de Gruyter.

Carnie, Andrew. 2006. Syntax: A Generative Introduction. Oxford: Wiley
Blackwell Publishers.

Casielles, Eugenia. 1997. Topic, Focus, and Bare Nominals in Spanish. Doctoral
dissertation. University of Massachusetts, Amherst.

Cavalcante, Rerisson. 2007. A negagao p6s-verbal no portugues brasileiro:
analise descritiva e te6rica de dialetos rurais de afro-descendentes. Master
thesis, Universidade Federal da Bahia.

Cinque, Guglielmo, 1999. Adverbs and Functional Heads. A Cross-Linguistic
Perspective. New York: Oxford University Press.

Cinque, Guglielmo (Ed.), 2002. Functional Structure in DP and IP. The
Cartography of Syntactic Structures, Vol. 1. New York: Oxford University Press.


226









Cheng, Lisa L.-S. 1991. On the Typology of Wh-questions, Doctoral dissertation,
Massachusetts Institute of Technology.

Cheng, Lisa L-S.1997. On the Typology of Wh-Questions. New York: Garland
Publishing.

Chomsky, Noam. 1976. Essays on Form and Interpretation. Amsterdam:North-
Holland.

Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht:
Mouton de Gruyter.

Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: The MIT
Press.

Chomsky, Noam. 2000. Minimalist Inquiries: the Framework. In Step by Step.
Essays in Honor of Howard Lasnik, ed. Roger Martin, David Michaels and Juan
Uriagereka, 89-155. Cambridge, Mass: MIT Press.

Chomsky, Noam. 2001. Derivation by phase. In Ken Hale: A life in language, ed.
Michael Kenstowicz. Cambridge, Mass.: MIT Press.

Cinque, Guglielmo. 1990. Types of A'-Dependencies. Cambridge: MIT Press.

Costa, Joao. 1998. Word order variation : a constraint-based approach. The
Hague: Holland Academic Graphics.

Costa, Joao. 2000. (ed.). Portuguese syntax New comparative studies. Oxford:
Oxford University Press.

Costa, Joao. 2001. Spec, IP ou Deslocado? Pr6s e contras das duas analises
dos sujeitos pre-verbais. DELTA, 17:283-304.

Costa, Joao. 2004. Subject Positions and Interfaces: The Case of European
Portuguese. The Hague: Mouton de Gruyter.

Costa, Joao and Charlotte Galves. 2002. External subjects in two varieties of
Portuguese: evidence for a non-unified analysis. In Romance Languages and
Linguistic Theory 2000, ed. Claire Beyssade et al., 109-125. Amsterdam: Johns
Benjamins.

Da Cunha, Maria. 1996. Gramaticalizagco dos Mecanismos de Negacgo
em Natal. In Gramaticalizagdo no Portugues do Brasil, ed. Maria Martelotta et al.,
167-189. Rio de Janeiro:Tempo Brasileiro.

Da Cunha, Maria. 2001. O Modelo das Motivag6es Competidoras no Dominio
Funcional da Negacgo. DELTA 17:1-30.


227










Da Cunha, Maria. 2007. Grammaticalization of the strategies of negation in
Brazilian Portuguese. Journal of Pragmatics 39:1638-1653.

Den Besten, H. 1986. Double negation and the genesis of Afrikaans. In Substrata
versus universals in Creole languages. Papers from the Amsterdam Creole
workshop, April 1985, ed. Pieter Muysken and Norval Smith, 185-230.
Amsterdam: John Benjamins.

De Swart, Henriette and Ivan Sag. 2002. Negative Concord in France. Linguistics
& Philosophy 25: 373-417.

Dryer, Matthew S. 2007. Verb-Object-Negative Order in Central Africa. In
Negation Patterns in West Africa, ed. Norbert Cyffer, Erwin Ebermann, and
Georg Ziegelmeyer. Ms.

Duarte, Maria. 1995. A Perda do Principio "Evite pronome" no Portugues
Brasileiro. Doctoral dissertation, Universidade de Campinas.

Duarte, Maria. 2000. The loss of the 'avoid pronoun' principle in Brazilian
Portuguese. In Brazilian Portuguese and the Null Subject Parameter, ed. Mary
Kato et. al.,17-36. Frankfurt-Madrid: Vervuert-lberoamericana.

Emonds, Joseph. 1978. The verbal complex V'-V in French. Linguistic Inquiry
9:151-75.

Ferraz, Luiz Ivens. 1979. The Creole of Sdo Tom6. Johannesburg:
Witwatersrand University Press.

Fonseca, Helena. 2004. Marcador negative final no Portugues Brasileiro.
Cadernos de estudos lingaisticos. 46: 5-19.

Galves, Charlotte. 1989. O objeto nulo no portugues brasileiro: percurso de uma
pesquisa. Cadernos de Estudos Lingoisticos 17:65-90.

Galves, Charlotte. 1998. T6picos e sujeitos, Pronomes e Concordancia no
Portugues Brasileiro. Cadernos de Estudos Lingoisticos, 34:19-31.

Giannakidou, Anastasia. 1998. Polarity sensitivity as (non)veridical dependency.
Amsterdam: John Benjamins.

Giannakidou, Anastasia. 2000. Negative ... Concord? Natural Language and
Linguistic Theory 18:457-523.


228









Giannakidou, Anastasia. 2006. N-words and negative concord. In The
Linguistics Companion vol. III, ed. Martin Everaert and Henk van Riemsdijk. 327-
391. Oxford: Blackwell.

Goodall, Grant. 2001. The EPP in Spanish. Objects and Other Subjects, eds.
William Davies and Stanley Dubinsky, 193-223. Amsterdam:Kluwer.

Grohmann, Kleanthes. 2003. Prolific Domains: On the Anti-Locality of Movement
Dependencies. Amsterdam: John Benjamins.

Gunther, Wilfried. 1973. Das Portugiesische Kreolisch der l/ha do Principe.
Marburg an der Lahn: Marburger Studien zur Afrika und Asienkunde.

Haddad, Youssef. 2007. Adjunct Control in Telegu and Assamese. Doctoral
dissertation. University of Florida.

Haegeman, Liliane. 1995. The syntax of negation. Cambridge Studies in
Linguistics 75. Cambridge: Cambridge University Press.

Haegeman, Liliane and Raffaella Zanuttini. 1991. Negative Heads and the Neg-
Criterion. The Linguistic Review 8:233-251.

Haegeman, Liliane and Raffaella Zanuttini. 1996. Negative Concord in West
Flemish. In Parameters and Functional Heads, eds. Belleti Adriana and Luigi
Rizzi,117-180. Oxford: Oxford University Press.

Han, C-H. 2001. Force, negation and imperatives. The Linguistic Review 18,
289-325.

Hansen, Quinn. 2005. Basic Word Order: Kaqchikel Mayan. Masters thesis,
Universidade Nova de Lisboa.

Hiraiwa, Ken. 2001. Multiple Agreement and the Defective Intervention Effect,
Ms. MIT.

Heim, Irene. 1982. The semantics of definite and indefinite noun phrases.
Doctoral dissertation, University of Massachusetts, Amherst.

Herburger, Elena. 2001. The negative concord puzzle revisited. Natural
Language Semantics 9:289-333.

Higginbotham, James and Robert May. 1981. Questions, Quantifiers, and
Crossing. The Linguistic Review 1, 41-80.

Holmberg, Anders. 2005. Is there a little pro? Evidence from Finnish. Linguistic
Inquiry 33, 533-64.


229










Horn, Laurence. 1989. A Natural History of Negation. Chicago: The University of
Chicago Press.

Horn, Laurence. 2002. Assertoric inertia and NPI licensing. CLS 38: 55-82.

Hornstein, Norbert. 2001. Move! A Minimalist theory of construal: Generative
syntax. United Kingdom: Blackwell.

Hornstein, Norbert; Jairo Nunes and Kleanthes Grohmann. 2005.
Understanding Minimalism. Cambrigde: Cambrigde University Press.

Huang, C.-T. J. 1982. Logical Relations in Chinese and the Theory of Grammar.
Doctoral dissertation, Massachusetts Institute of Technology.

latridou, Sabine. 1990. About Agr(P). Linguistic Inquiry 21: 551-577.

Israel, Michael. 1998. The rhetoric of grammar: scalar reasoning and polarity
sensitivity. Doctoral dissertation, University of California, San Diego.

Jespersen, Otto. 1917. Negation in English and other Languages. Copenhagen:
A.F.H0st.

Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar.
Cambridge: MIT Press.

Kato, Mary. 1999. Strong and Weak pronominals in the Null Subject Parameter.
Probus 11: 1-37.

Kato, Mary. 2000. The partial pro-drop nature and the restricted VS order in
Brazilian Portuguese. In Brazilian Portuguese and the Null Subject Parameter,
ed. Mary Kato et. al., 207-240. Frankfurt-Madrid: Vervuert-lberoamericana.

Kato, Mary. 2003. Null objects and VP ellipsis in European and Brazilian
Portguese. In Romance Languages and Linguistic Theory 2001, ed. Josep Quer
et al.,131-154. Amsterdam: John Benjamins.

Kato, Mary. 2009. Mudanga de ordem e gramaticalizacgo na evolucgo das
estruturas de foco no Portugues Brasileiro. Estudos LingOisticos 38: 375-385.

Kato, Mary and Carlos Mioto. 2005. A multi-evidence analysis of European and
Brazilian Portuguese Wh-questions. In Linguistic evidence: empirical, theoretical
and computational perspectives, ed. S Kepser and Marcelo Reis, 307-328.
Berlin: Mouton de Gruyter.


230









Kato, Mary and Eduardo Raposo. 1996. European and Brazilian Portuguese
word order: questions, focus and topic constructions. In Aspects of Romance
Linguistics, eds. C. Parodi et. al., 267-77. Washington: Georgetown University
Press.

Kato, Mary and Eduardo Raposo. 2007. Topicalization in European and Brazilian
Portuguese. In Romance Linguistics 2006. ed. Jose Camacho et. al., 205-218.
Amsterdam: John Benjamins.

Kayne, Richard. 1989. Notes on English Agreement. CIEFL 1: 40-67.

Kayne, Richard. 1994: The Antisymmetry of Syntax. Cambridge, MA: MIT Press.

Koopman, Hilda. And Dominique Sportiche. 1991. The position of subjects.
Lingua 85.1:211-258.

Ladusaw, William A. 1992. Expressing negation. In Proceedings of the Second
Conference on Semantics and Linguistic Theory, ed. Chris Barker and David
Dowty, 237-259. Columbus: Department of Linguistics, Ohio State University.

Laka, Itziar. 1990. Negation in Syntax: on the Nature of Functional Categories
and Projections. Doctoral dissertation, Massachusetts Institute of Technology

Larson, Richard .1988. On the double object construction. Linguistic Inquiry
19:335-391

Lasnik, Howard. 1972. Analyses of negation in English. Doctoral dissertation,
Massachusetts Institute of Technology.

Lasnik, Howard. 1999. Minimalist Analysis. Oxford: Blackwell.

Lasnik, Howard and Mamoru Saito. 1992. Move a : Conditions on its application
and output. Cambridge: The MIT Press.

Lessa de Oliveira, Ana. 2008. Oraq6es Relativas do Portugues: quest6es
te6ricas e fatos de aquisigco. Doctoral Dissertation, Universidade de
Campinas.

Lobato, Lucia Maria Pinheiro. 1986. Sintaxe gerativa do portugues. Da teoria
padrdo a teoria da reg6ncia e ligagdo. Belo Horizonte:Vigilia.

Lobeck, Anne, 1995. Ellipsis: Functional Heads, Licensing, and Identification.
New York: Oxford University Press.

Martins, Eneida. 1997. Sentential negation in spoken Brazilian Portuguese.
Washington: Georgetown University.


231










Martins, Ana. 1994. Cliticos na hist6ria do portugues. Doctoral dissertation,
Universidade de Lisboa.

Martins, Ana and Jairo Nunes. 2006. Raising Issues in Brazilian and European
Portuguese. Journal of Portuguese Linguistics 4: 53-77.

Martins, Ana and Jairo Nunes. 2008. Apparent Hyper-raising in Brazilian
Portuguese: Base-Generation of Topics and Long Distance Agreement. In
Selected papers from Edges in Syntax, Cyprus 2006, ed. Phoevos Panagiotidis.
Oxford: Oxford University Press.

Martinez, Cristina. 2005. Concordancia de negaci6n en el portugues minero.
Pterod6ctilo: Revista de Arte, Literatura y Lingoistica 3:26-35.

Matos, Gabriela. 2003. Aspectos sintacticos da negacgo. In Gram6tica da lingua
portuguesa. ed. Maria Helena Mira Mateus, 767-793. Lisboa: Caminho.

May, Robert. 1985. Logical Form: Its Structure and Derivation. Cambridge: MIT
Press.

McCloskey, James. 1991. Clause structure, ellipsis and proper government in
Irish. Lingua 85:259-302.

Merchant, Jason. 2001. The Syntax of Silence: Sluicing, Islands, and the Theory
of Ellipsis. Oxford: Oxford University Press.

Mioto, Carlos. 1992. Negacao Sentential no Portugues Brasileiro. Doctoral
disseration, Universidade de Campinas.

Mioto, Carlos. 1998. Preposig6es: NOcleos lexicais ou funcionais? In Lingoistica
e Ensino, eds. Luis Cabral et al., 165-179. Florian6polis, Editora Insular.

Moscati, Vincenzo. 2006. The Scope of Negation. Doctoral dissertation,
University of Siena.

Muller, Ana, et. al. 2007. The Brazilian Portuguese Quantifier "Todo" in
Distributive and in Collective Contexts. DELTA 23:71-95.

Namiuti, Cristiane. 2008. Aspectos da hist6ria gramatical do portugues:
Interpolagco, negacgo e mudanga. Doctoral disseration, Universidade de
Campinas.

Nunes, Jairo (ed). 2009. Minimalist Essays on Brazilian Portuguese Syntax.
Amsterdam: John Benjamins.


232









Nunes, Jairo and Mary Kato. 2009. A Uniform Raising Analysis for Standard and
Non-Standard Relative Clauses. In Minimalist Essays on Brazilian Syntax, ed.
Jairo Nunes, 93-120. Amsterdam: John Benjamins.

Ordo6ez, Francisco. 1997. Word Order and Clause Structure in Spanish and
other Romance Languages. Doctoral dissertation CUNY.

Ordo6ez, Francisco and Eduardo Treviio. 1999. Left-dislocated subjects and the
pro-drop parameter: A case study of Spanish. Lingua, 107: 39-68.

Ouhalla, Jamal. 1990. Sentential negation, Relativised minimality, and the
aspectual status of auxiliaries. The Linguistic Review 7: 183-231.

Pereira, Iris. 2000. Para uma definicgo do process de concordancia negative
em portugues europeu. In XV Encontro Nacional da Associagco Portuguesa de
Linguistica : actas, ed. Rui Castro and Pilar Barbosa, 125-149. Lisboa:
Associacgo Portuguesa de Linguistica.

Peres, Joao. 2000. Towards a Comprehensive View of Negative Concord. In
Meaning Through Language Contrast, ed. K.M Jaszczolt and Ken Turner, 29-44.
Philadelphia: John Benjamins.

Pires, Acrisio. 2007. The subject, it is here! The Varying Structural Positions of
Preverbal Subjects. DELTA 23:113-147.

Pires, Acrisio and Heather Taylor. 2007. The Syntax of Wh-in-situ and Common
Ground. Paper presentation Penn Linguistics Colloquium 31, University of
Pennsylvania, Feb 23-25.

Poletto, Cecilia, 2000. The Higher Functional Field. Evidence from Northern
Italian Dialects. New York: Oxford University Press.

Pollock, Jean-Yves. 1989. Verb Movement, Universal Grammar, and the
Structure of IP. Linguistic Inquiry 20:365-424.

Postma, Gertjan, and Wim van der Wurff. 2007. How to say No and Don't:
Negative Imperatives in Romance and Germanic. In Imperative Clauses in
Generative Grammar: Studies Offered to Frits Beukema, ed. Wim van der Wurff
73-110. Amsterdam: John Benjamins

Prince, Ellen. 1981. Toward a taxonomy of given-new information. In Radical
Pragmatics, ed. Peter Cole, 223-255. New York: Academic Press.

Prince, Ellen. 1992. The ZPG letter: subjects, definiteness, and information-
Status. In Discourse Description: Diverse Analyses of a Fundraising Text, eds.
Sally Thompson and W. Mann, 295-325. Amsterdam: Benjamins.


233










Radford, Andrew. 2004. Minimalist Syntax: Exploring the structure of English.
Cambridge: Cambrige University Press.

Rezac, Milan. 2003. The Fine Structure of Cyclic Agree. Syntax 6: 156-82.

Rivero, Maria-Luisa and Arhonto Terzi. 1995. Imperatives, V-movement, and
Logical Mood. Journal of Linguistics, 31:301-332.

Rizzi, Luigi. 1982. Issues in Italian Syntax. Dordrecht: Kluwer.

Rizzi, Luigi. 1997. The Fine Structure of the Left Periphery. In Elements of
Grammar, ed. Liliane Haegeman, 281-338. Dordrecht: Kluwer.

Rizzi, Luigi, 2001. On the position "Int(errogative)" in the left periphery of the
clause. In Current Studies in Italian Syntax. Essays Offered to Lorenzo Renzi.
ed. Guglielmo Cinque and Giampaolo Salvi, 287-296. Amsterdam: Elsevier.

Rizzi, Luigi (Ed). 2004. The Structure of CP and IP. The Cartography of Syntactic
Structures, Vol. 2. New York: Oxford University Press.

Roberts, lan. 2004. The C-System in Brythonic Celtic Languages, V2, and the
EPP. In The Structure of IP and CP, ed. Luigi Rizzi, 297-328. Oxford/New York:
Oxford University Press.

Rodrigues, Cilene. 2004. Loss of Morphology and A-Movement out of Case
Domains. Doctoral dissertation. University of Maryland.

Roncarti, Claudia. 1996. A Negagao no Portugues Falado. In Variagdo e
Discurso ed. Alzira Tavares de Macedo et al., 97-112. Rio de Janeiro: Tempo
Brasileiro.

Ross, John. 1967. Constraints on Variables in Syntax. Doctoral dissertation,
Massachusetts Institute of Technology.

Rowlett, Paul. 1998. Sentential Negation in French. New York/Oxford: Oxford
University Press.

Saeed, John. 1997. Semantics. Oxford: Blackwell Publishers.

Santos, Ana LQcia. 2004. How do children say 'Yes' in European Portuguese?. In
Proceedings of GALA 2003Jacqueline, eds, van Kampen and Sergio Baauw.
Utrecht: Utrecht University.

Scherre, Maria Marta Pereira. 2007. Aspectos sincr6nicos e diacr6nicos do
imperative gramatical no portugues brasileiro. Alfa, 51:189-222.


234










Schwegler, Armin. 1988.Word-order Changes in Predicate Negation Strategies in
Romance Languages. Diachronica 2:21-58.

Schwegler, Armin. 1991a. Predicate Negation in Contemporary Brazilian
Portuguese-a Change in Progress. Orbis 34:187-214.

Schwegler, Armin. 1991b. Negation in Palenquero. Synchrony. Journal of Pidgin
and Creole Languages 6:165-214.

Schwenter, Scott. 2005. The pragmatics of negation in Brazilian Portuguese.
Lingua 115:1427-1456.

Schwenter, Scott. 2006. Fine-Tuning Jespersen's Cycle. In Drawing the
Boundaries of Meaning: Neo- Gricean Studies in Pragmatics and Semantics in
Honor of Laurence R. Horn, ed. Betty J. Birner and Gregory Ward. Amsterdam:
John Benjamins.

Schwenter, Scott and Glaucia Silva. 2002. Overt vs. null direct objects in
spoken Brazilian Portuguese: a semantic/pragmatic account. Hispania, 85:577-
586.

Silva, Glaucia. 2001. Word order in Brazilian Portuguese : a minimalist analysis.
The Hague: Mouton de Gruyter.

Silva, Claudia. 2004. A natureza de AGR e suas implicag6es na ordem VS: um
estudo comparative entire o portugues brasileiro e o portugues europeu. Doctoral
dissertation, Universidade Federal de Alagoas.

Sportiche, Dominique. 1988. A theory of floating quantifiers and its corollaries for
constituent structure. Linguistic Inquiry 19: 425-449.

Sousa, Arivaldo Sacramento. 2004. As estruturas de negagao em uma
comunidade rural afro-brasileira: Helvecia BA.
http://www.vertentes.ufba.br/souza.doc

Sousa, Liliana. 2007. Formas Reduzidas de Itens Negativos no Portugues
Brasiliero. Masters thesis. Universidade Federal de Minas Gerais.

Stalnaker, Robert. 1978. Assertion. In Pragmatics: Syntax and Semantics
Volume 9. ed. P.Cole. New York: Academic Press.

Stalnaker, Robert. 2002. Common Ground. Linguistics and Philosophy 25: 701-
21.


235









Szabolcsi, Anna. 2006. Strong vs. weak islands. In The Blackwell Companion to
Syntax, ed. Everaert and van Riemsdijk. 4:479-532.

Thrainsson, Hoskuldur. 1996. On the (non)-universality of functional projections.
In Minimal ideas, ed. W. Abraham et. al., 253-81. Amsterdam: John Benjamins.

Travis, Lisa. 1984. Parameters and Effects of Word Order Variation. Doctoral
dissertation, MIT.

Uriagereka, Juan. 1995. Aspects of the Syntax of Clitic Placement in Western
Romance. Linguistic Inquiry, 26: 79-123.

Uribe-Echevarria, Maria. 1992. On the Structural Positions of Subjects in
Spanish, and their Consequences for Quantification. In Syntactic Theory and
Basque Syntax eds. Jose Lakarra and Julio de Urbina, 447-493, San Sebastian:
ASJU.

van Riemsdijk, Henk and Edwin Williams. 1986. Introduction to the Theory of
Grammar. Current Studies in Linguistics. Cambridge: MIT Press.

Vicente, Helena. 2006. O Quantificador Flutuante Todos no Portugues Brasileiro
e no Ingles: Uma Abordagem Gerativa. Universidade de Brasilia.

Vitral, Lorenzo. 1999. A negacao: teoria da checagem e mudanga linguistica.
DELTA, 15:1.

Wouden, Ton, van der and Frans Zwarts. 1993. A Semantic Analysis of Negative
Concord. In Semantics and Linguistic Theory III, 202-219. Ithaca:Cornell
University Press.

Wouden, Ton, van der. 1997. Negative Contexts. London: Routledge.

Zanuttini, Raffaella. 1991. Syntactic properties of sentential negation. A
comparative study of Romance languages. Doctoral dissertation, University of
Pennsylvania.

Zanuttini, Raffaela. 1997. Negation and clausal structure. A Comparative Study
of Romance languages. New York, Oxford: Oxford University Press.

Zanuttini, Raffaela. 2001. Sentential Negation. In The Handbook of
Contemporary Syntactic Theory, ed. Mark Baltin and Chris Collins, 511-535.
Maiden, MA: Blackwell Publishers.

Zeijlstra, Hedde. 2004. Sentential Negation and Negative Concord. Doctoral
dissertation, University of Amsterdam.


236









Zeijlstra, Hedde. 2006. The Ban on True Negative Imperatives Empirical Issues
in Formal Syntax and Semantics 6: 405-424

Zeijlstra, Hedde. 2008. Negative Concord in Syantactic Agreement. Ms.

Zocca, Cynthia. 2007. And the wh goes where. Ms. University of
Connecticutt.

Zubizarreta, Maria Luisa. 1998. Prosody, Focus, and Word Order. Cambridge,
MA: MIT Press.


237









BIOGRAPHICAL SKETCH

Quinn McCoy Hansen was born in Salt Lake City, Utah. Before graduating from

the University of Texas at Austin, he studied at universities in Hawaii, Mexico, Portugal

and Spain. In May 2003, he graduated with honors with a degree in Spanish and

Portuguese languages and literatures from the University of Texas at Austin, College of

Liberal Arts. He subsequently studied at the Universidade Nova de Lisboa, Portugal

where in 2005 he received the Grau de Mestre in linguistics for his work Basic Word

Order in Kaqchikel which received the highest mark of muito bom. After studying in

Lisbon, Quinn moved to Aleppo, Syria where he was employed as a first grade teacher

at the National School of Aleppo. He began his journey towards earning a Doctorate

degree in linguistics at the University of Florida in August 2006. Since arriving at the

University of Florida, he has taught Portuguese in the Department of Spanish and

Portuguese, Quinn has also taught linguistics for the Program in Linguistics. In 2009,

he received the award for outstanding Portuguese Teaching Assistant. In 2010, Quinn

received a prestigious Graduate Student Teaching Award from the University of Florida

and the College of Liberal Arts and Sciences. In the Fall of 2010, he will teach

Portuguese at the University of Florida.


238





PAGE 1

CLAUSE-FINAL NEGATION IN BRAZILIAN PORTUGUESE By QUINN MCCOY HANSEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010 1

PAGE 2

2010 Quinn McCoy Hansen 2

PAGE 3

To my light and my flower Para Lcia e Sienna Yasmin Amo-vos 3

PAGE 4

ACKNOWLEDGMENTS I want to thank Eric Potsdam my supervisory committee c hair, for teaching me the art and science of conducting and producing linguistic research. Additionally, I thank Dr. Brent Henderson, Dr. Gary Miller, Dr. Elizabeth Ginway, and Dr. Benjamin Hebblethwaite for being part of my committee. I thank Dr. Henderson for his patience with my questions, most of wh ich were elementary. Dr. Miller deserves thanks for always providing stimulating debate and conversation in and out of class. I thank Dr. Hebblethwaite for introducing me to the world of Creole Studies and showing me with concrete evidence how linguistics can have a direct impact on societies. Finally, I thank Dr. Ginway for allowing me to teach Portuguese, and allowing me the freedom to run the courses as best I saw fit. I am extremely grateful to all my Brazilian Portuguese co nsultants, both here in the United States and in Brazil. Particularly, I thank Sharon Barkley and Andra Ferreira for their judgments and comments about use. Also, here in the States, I thank Cilene Rodrigues and Ana Bastos for sending me articles in Portuguese and adding their grammaticality judgments. I thank Megan Armstrong and Scott Schwenter for their comments on my work. In Brazil, I want to thank Jlia Ramos, Florbela Silva, Divino, Ednilson, a Nordestina Valentona, Jos Fernandes and his employees at NorteSul, and Ablio Joaquim Fernandes. Also, I thank the producers of the show Sem MeiasPalavras for producing a terribly low-class show that provided several hundred examples of Clause-Final Negation. I owe many thanks to all the people in the Linguistics Program that taught me in diverse ways, from formal classr oom settings to informal chats. 4

PAGE 5

I thank my parents for showi ng me that I can do whatever I want in life, and for teaching me that reading and learning are the keys to happiness. I thank my daughter Sienna Yasmin for hav ing patience with me, even when I had to stare at a book or a screen for several hours straight I hope that someday she will understand the sacrifices t hat she has made for me. I thank my wife Lcia, without whom I w ould have never begun this journey. Also, I thank her for putting up with hours of questions about Brazilian Portuguese and negative sentences, all the while, smiling brightly. 5

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS ..................................................................................................4 LIST OF TABLES ............................................................................................................9 LIST OF ABBREVIATIONS ...........................................................................................10 ABSTRACT ...................................................................................................................12 CHAPTER 1 INTRODUC TION....................................................................................................14 1.1 Research Questions .......................................................................................14 1.2 Domain of Investigation: Brazilian Portuguese ...............................................16 1.3 Three Types of Sentential Negation ...............................................................17 1.4 Past Corpus Studies of Clause-Final Negation ..............................................20 1.4.1 Roncarti 1996 ......................................................................................20 1.4.2 Da Cunha 1996, 2001, 2007 ...............................................................21 1.4.3 Alkmim 1999 ........................................................................................21 1.4.4 Camargos 2000 ...................................................................................22 1.4.5 Sousa 2004 .........................................................................................23 1.4.6 Cavalcante 2007 ..................................................................................23 1.4.7 Summary of Corpus Studies ................................................................24 1.5 Syntactic and Semantic/Pragmatic Restrictions in BP for T2 and T3 .............25 1.5.1 Discourse Requirements for T2 and T3 ...............................................25 1.5.2 Embedded Clauses .............................................................................26 1.5.3 Relative Clauses ..................................................................................28 1.5.4 Topics 31 1.5.5 Wh-Questions ......................................................................................32 1.5.6 N-Words and Negative Concord ..........................................................33 1.5.7 Imperatives ..........................................................................................36 1.5.8 NPI Idioms ...........................................................................................37 1.5.9 Summary of the Data ...........................................................................38 1.6 Structure of the Study .....................................................................................39 2 THEORETICAL FO UNDATION S............................................................................42 2.1 Introduction ....................................................................................................42 2.2 Minimalism .....................................................................................................42 2.2.1 Fundamental Principles of Minimalism ................................................42 2.2.2 Probe-Goal System an d Feature Agreement ......................................45 2.3 CP: The Left Periphery ...................................................................................47 2.3.1 Rizzis Split CP ....................................................................................47 2.3.2 The Split CP and BP ............................................................................49 6

PAGE 7

2.4 Conclusion .....................................................................................................53 3 THE CLAUSE STRUCTURE OF BRAZILIAN PO RTUGUESE...............................54 3.1 Introduction ....................................................................................................54 3.2 Subjects in BP ................................................................................................55 3.2.1 The Debate over the Romance Preverbal Subject Position .................55 3.2.2 BPs Preverbal Subject ........................................................................63 3.2.2.1 Temporal Adverbs ..................................................................64 3.2.2.2 Weak Pronouns ......................................................................65 3.2.2.3 Quantifiers ..............................................................................66 3.3 Verbs and Verb Movement .............................................................................68 3.3.1 IP 69 3.3.2 Non-Split IP .........................................................................................72 3.3.3 AgrP in BP ...........................................................................................74 3.4 Fronting in BP ................................................................................................84 3.5 Wh-Questions: Fronted and In-Situ ................................................................89 3.5.1 Types and Conditions for Wh-Words: Movement and In-Situ ..............89 3.5.2 Deriving Wh-Questions in BP ..............................................................92 3.6 Conclusions ....................................................................................................97 4 PREVERBAL NE GATION.....................................................................................100 4.1 Introduction ..................................................................................................100 4.2 The Syntax of Negation ................................................................................100 4.2.1 NegP 101 4.2.2 The Location of NegP ........................................................................105 4.2.2.1 Laka ......................................................................................105 4.2.2.2 Zanuttini ................................................................................108 4.3 Preverbal Negation in BP .............................................................................111 4.3.1 The Phonology of the Pr everbal Negative Marker .............................112 4.3.2 Mioto 1992 & 1998 ............................................................................114 4.3.3 Martins 1994 and Namiuti 2008 .........................................................117 4.3.4 The structural Position of NegP in BP ...............................................120 4.4 Conclusion ...................................................................................................122 5 THE PRAGMATICS AND SEMANTICS OF CLAUSE-FINAL NEGATION...........124 5.1 Introduction ..................................................................................................124 5.2 Previous Approaches to Clause-Final Negation: Semantics/Pragmatics .....125 5.2.1 Emphasis ...........................................................................................125 5.2.2 Presuppositional or Contrary-to-Fact .................................................130 5.2.3 Discourse-Old ....................................................................................133 5.3 Applying the Theory .....................................................................................145 5.4 Conclusion ...................................................................................................149 7

PAGE 8

6 THE SYNTAX OF CLAUSE -FINAL NEGA TION...................................................150 6.1 Introduction ..................................................................................................150 6.2 Previous Accounts ........................................................................................150 6.2.1 Martins 1997 ......................................................................................150 6.2.2 Fonseca 2004 ....................................................................................158 6.2.3 Cavalcante 2007 ................................................................................161 6.3 A Proposal for Clause-Final Negation in Brazilian Portuguese ....................166 6.3.1 Derivation of Clause-Final Negation ..................................................169 6.3.2 Wh-Words ..........................................................................................175 6.3.3 Relative Clauses ................................................................................178 6.3.4 Embedded Clauses ...........................................................................181 6.3.5 Topicalization .....................................................................................182 6.3.6 Imperatives ........................................................................................184 6.3.7 Semantic Interpretation of Clause-Final Negation .............................186 6.4 Conclusion ...................................................................................................192 7 NEGATIVE CONCORD IN BR AZILIAN PORT UGUESE......................................194 7.1 Introduction ..................................................................................................194 7.2 Negative Concord .........................................................................................194 7.2.1 Definition and Types of Negative Concord ........................................195 7.2.2 N-Words are Negative Quantifiers .....................................................200 7.2.3 N-Words are Non-Negative ...............................................................204 7.2.4 Zeijlstras Feature-Based A pproach to Negative Concord .................207 7.2.5 Summary ...........................................................................................211 7.3 N-Words and Negative Concord in Brazilian Portuguese .............................211 7.3.1 BP N-Words .......................................................................................211 7.3.2 Feature Agreement and BP N-Words ................................................215 7.4 Negative Concord and Clause-Final Negation in Brazilian Portuguese .......217 7.5 Conclusion ...................................................................................................221 8 CONCLUS ION......................................................................................................222 LIST OF REFERENCES .............................................................................................224 BIOGRAPHICAL SKETCH ..........................................................................................238 8

PAGE 9

LIST OF TABLES Table page 1-1 Syntactic Restrictions for BP CFN ......................................................................38 3-1 Loss of Verbal Morphology in BP .......................................................................73 6-1 Syntactic Restrictions for BP CFN ....................................................................168 7-1 N-Words and CFN ............................................................................................218 9

PAGE 10

LIST OF ABBREVIATIONS Ungrammatical/Unacceptable [*] Strong Feature ? Degraded Grammaticality ?? Highly Degraded Grammaticality Grammatical Negative/Negation 1 First Person 2 Second Person 3 Third Person AGR Agreement asp Aspect Marker BP Brazilian Portuguese CFN Clause-final Negation cl Clitic CLLD Clitic Left Dislocation comp Complementizer DenP Denial Phrase det Determiner DN Double Negative emph Emphasis f Feminine FQ Floating Quantifier Foc Focus Phrase fut Future Tense 10

PAGE 11

imp Imperative imper Past Imperfect [iNEG] Interpretable negative feature inf Infinite IP Inflectional Phrase JC Jespersen Cycle m Masculine NC Negative Concord neg Negative Marker NPI Negative Polarity Item NSL Null Subject Language Op Syntactic operator p Plural pos Positive prep Preposition pst Past Tense QP Quantifier Phrase s Singular sbj Subjunctive spec Specifier T Tense t Trace [uNEG] Uninterpretable Negative Feature VP Verb Phrase [WH] Interrogative Feature 11

PAGE 12

Abstract of Dissertation Pr esented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy CLAUSE-FINAL NEGATION IN BRAZILIAN PORTUGUESE By Quinn McCoy Hansen August 2010 Chair: Eric Potsdam Major: Linguistics My study explores Clause-Final Negat ion in Brazilian Portuguese, within a Minimalist Program framework. Brazilian Portuguese make s use of three types of sentential negation. In Type 1, the negative marker appears preverbally (i). Type 2 has both a preverbal marker and a clause-final ma rker (ii). Type 3 has only a clause final negative marker (iii). i. Eu no quero o bolo TYPE 1 1S NEG want.PRS.1S DET cake I do not want the cake. ii. Eu no quero o bolo no TYPE 2 1S NEG want.PRS.1S DET cake NEG I do not want the cake. iii. Eu quero o bolo no TYPE 3 1S want.PRS.1SG the cake NEG I do not want the cake. In this work, I analyze preverbal negation as the head of a functional projection NegP within a split IP scenario (Pollo ck 1989, Costa and Galves 2002). I analyze Clause-Final Negation as resulting from leftwa rd syntactic movement of the whole IP to the specifier of a Topic Phrase which is headed by no Because the IP moves to the 12

PAGE 13

left, no is clause-final. The movement of IP is motivated by t he discourse status of the IP. Following Schwenter (2005), I propose that Clause-Final Negation is only licit when the proposition that it negates is discourse-oldwhen it is a kind of topic. The movement analysis accounts for both syntactic and semantic/pragmatic restrictions on clause-final negation. 13

PAGE 14

CHAPTER 1 INTRODUCTION 1.1 Research Questions The goal of this dissertation is to ex amine syntactic, semantic, and pragmatic issues surrounding the analysis of sentent ial negation in Brazilian Portuguese (BP henceforth). Specifically, this work expl ores Clause-Final Negation (CFN) in BP and gives a syntactic account within a minima list framework. Negation and other topics related to polarity have been a part of lingui stic study since well before the generative approach began (Jespersen 1917) because t hey aid in understanding syntactic structures, semantics, and scope (Lasnik 1972, Emonds 1978, Pollock 1989 Laka 1990, Zanuttini 1997, Zeijlstra 2004). BP has three options for sentential negati on. In what I call Type 1 (T1), the negative marker appears preverbally as seen in (1) a. Type 2 (T2) has both a preverbal marker and a postverbal marker that is clause-final, example (1) b. Type 3 (T3) only has a clause-final negative marker, example (1) c. (1) a. Eu no quero o bolo TYPE 1 (T1) 1S NEG want.PRS.1S DET cake I do not want the cake. b. Eu no quero o bolo no TYPE 2 (T2) 1S NEG want.PRS.1S DET cake NEG I do not want the cake. c. Eu quero o bolo no TYPE 3 (T3) 1S want.PRS.1S DET cake NEG I do not want the cake. The existence of CFN seen in BP, as in T2 and T3, is not common for Romance languages. These patterns have not been widely discussed in the literature, and the fact 14

PAGE 15

that all three types exist in a single langua ge is also unusual (see Schwegler 1991b for other Romance languages with clause-final negation). The two main goals of this dissertation are 1) to document the three types of negation in BP and 2) to understand the thre e types of negation with a minimalist syntactic framework. This dissertation answers the following questions: Question 1 : What are the patterns of negation in BP and how do they differ from European Portuguese and other Romance languages? Question 2 : What is the syntactic analysis of the different sentential negation types? a. What is the syntactic status of no (i.e. head, phrase, or specifier)? b. How can a syntactic analysis of BP negation be understood within a minimalist framework? c. What is the distribution of negative words and how does negative concord work in BP? The rest of this chapter answers Question 1 by presenting the empirical domain of investigation. The subsequent chapters of this work answer Question 2 and subquestions (a), (b), and (c). The remainder of this chapter is di vided into four sections. Section 1.2 addresses the domain of investigation, how data wa s collected and analyzed, and this section introduces Brazilian Port uguese generally. Section 1.3 discusses the different types of sentential negation crosslinguistically. Here I present examples of the three types of negation seen in BP that are used in other languages. Section 1.4 discusses previous corpus studies that have look ed at CFN in BP. Section 1.5 lays out the principal data 15

PAGE 16

for this dissertation. This section shows t he syntactic structures that allow CFN in BP as well as the structures that prohibit i t. Additionally, this section introduces the discourse situations where CFN is grammatical but infelicitous. The final section of this chapter summarizes the data and gives an overview of the structure of this dissertation. 1.2 Domain of Investigation: Brazilian Portuguese BP is a Romance language spoken by the majority of Brazilians, and government numbers estimate that there are mo re than 200 million speakers of BP ( www.brasil.gov.br ). The number of varieties is undefined, but BP can roughly be broken up into 5 main dialects, northern, northeas tern, central, southeastern, and southern (Azevedo 2005). With respect to negation, I assume that all three types of negation are found among all dialects and socio-eco nomic classes (Schwenter 2005). However, this fact is not overtly admitted by the upper-class because they see T3 as caipira or country bumpkin speech (Barme 2000). To counter th is stereotype and to collect data for this study, I use naturalistic data from reality TV popular talk shows, and internet chat rooms in addition to grammaticality judgm ents on elicited data and data from past corpus studies presented in this chapter The grammaticality judgments were made principally by Portuguese speakers residing in Brazil, although Braz ilians living in the United States also provided some data. The history of T2 and T3, clause-final negati on, is not well known. Because of the lack of written material that references this type of negat ion, I will not be concerned with the historical aspects. Furthermore, CFN in BP is relatively new as a topic of study evidenced by the fact that the first scientific attempt to understand this phenomenon in BP was Schwegler (1991a). 16

PAGE 17

1.3 Three Types of Sentential Negation The three types of sentential negation in BP are repeated below. By types of sentential negation I merely mean the linear position of the negative marker in relationship to the verb. Sentences in (1) repeated here, semantically represent the same propositional meaning (Schwenter 2005:1429). (2) a. Eu no quero o bolo T1 1S NEG want.PRS.1S DET cake I do not want the cake. b. Eu no quero o bolo no T2 1S NEG want.PRS.1S DET cake NEG I do not want the cake. c. Eu quero o bolo no T3 1S want.PRS.1S DET cake NEG I do not want the cake. T1, preverbal negation, is found in Spanish and other Romance languages and many researchers argue that these negative marker s are syntactically similar (Mioto 1992, Zanuttini 1997, 2001, Zeijlstra 2004). T1 (3) Yo no quiero el libro Spanish 1S NEG want.PRS.1S DET book I do not want the book. (4) Maria non studia. Italian Maria NEG study.PRES.3SG Maria doesnt study (5) Tu no fizeste o exame European Portuguese 2S NEG do.PST.2S DET test You didnt do the test 17

PAGE 18

T2 is not common in other Romance lan guages; however, it exists in some Portuguese Creoles, such as So Tom Creole, as well as in some Spanish dialects of the Americas, for example Choc Spanish of Columbia (Schwegler 1991a:95, 1991b). T2 (6) No me gust eso all no Choc Spanish NEG 1S please.PST.3S that there NEG I dont like that over there (7) A na kuvida n fa So Tom Creole IMPER NEG invite 1P NEG They didnt invite us (Ferraz 1979:68) Another Indo-European language, Afrikaans, has a T2 construction similar to BP (Biberauer 2009). (8) Ek het nie boeke gelees nie Afrikaans 1S have.PRS.1S NEG books read NEG I have not read books Negative constructions similar to T2 ar e found in other Romance languages, such as French, Northern Italian dial ects, and Catalan. The differ ence is that the position of the second negative marker is postverbal and not clause-final, as in the French example with the meat being clause final and the negative marker postverbal (Zanuttini 1997:17): (9) Jean n aime pas la viande French Jean NEG-love.3S NEG DET meat Jean doesnt like the meat 18

PAGE 19

Berini & Remat (1996) shows that the position of the CFN marker in BP is indeed clause-final and not postverbal. The diffe rence between the BP and most other Romance languages is that there is almost no limit to th e length of material between the two instances of negation, as seen in (10) (10) A Sienna no quer saber como que ele chegou a minha casa no DET Sienna NEG want.3S know.INF how is that he arrive.PST.3S PREP 1S house NEG Sienna doesnt want to know how he arrived at my house. T3 negation is rare; nonetheless, there ar e certain cases in Romance where it does appear, as in Milanese, a dialect of Italian, (11) (Zanuttini 1997:5). Also, it is found in some languages of West Africa such as Ewe and Fon (Dryer 2007). Example (12) is from the Nigerian language Bi rom (Bouquiaux (1970:386) ci ted in Dryer (2007)). Example (13) shows T3 in Lungiye, a Portuguese Creole spoken on the island of So Tom and Principe. T3 (11) el lha scrivuu no Milanese 3S 3S.CL-has.3S written NEG He hasnt written (12) Birom 3P AORIST-bear children NEG They have not given birth to children (13) na sa podi d ci fa Lungiye 1S ASP able give 2S NEG I cant give it to you (Gnther 1973:78) 19

PAGE 20

These examples help position BP within the spectrum of the different negation strategies types. Although no single negative type is extraordinary, the fact that all three currently exist in one language is rare. That is, some languages allow more than one negative type, but they do not synchronically allow the three types that exist in BP. 1.4 Past Corpus Studies of Clause-Final Negation This section discusses some corpus studies of BP directly related to negation. These studies document the frequency of us e for each negation type. It should be noted that in nearly all cases, T3 is the leas t frequently used type, followed by T2. T1 is par excellence the norm for negation in BP. Most re searchers in these studies collected their data from the northeastern states of Br azil. Although the population of these states is generally perceived as using CFN more than other parts of Brazil, T2 and T3 negation takes place in all regions of Brazil (Scherre 2007, Schwegler 1991a, Schwenter 2005, and Cavalcante 2007). I conclude this secti on by commenting on the collective results of these studies. 1.4.1 Roncarti 1996 Roncarti (1996) is a spoken corpus study of the BP dialect in the northeastern city of Fortaleza. One of the main purposes of this work is to understand how the expression sei no I dont know (lit. I know not ) came about. Working within a functionalist perspective, Roncarti proposes that this expression came about either through de-grammaticalization of a syntactic element to a di scourse element or through the lexicalization of a syntactic element to a lexical item. Her research studied the syntactic situations in which each type of negation is used, such as type of clause, null subject, verbal repetition etc. 20

PAGE 21

Her corpus totaled 822 sentence, with nine of these unanalyzed. Her study points to an overall preference for T1 at 77%. Table (14) shows the division of frequency. (14) Roncarti 1996: T1 T2 T3 Total Tokens 625 149 39 813 Frequency 77% 18% 5% 100% 1.4.2 Da Cunha 1996, 2001, 2007 Like Roncarti, Da Cunha uses functionalis t paradigms to understand the rise of the clause-final negative marker and what she calls the weakening of the preverbal marker in terms of phonology and semant ics. Her corpus is from another northeastern city, Natal. She uses not only spoken data but al so a written corpus from 12 consultants from three different levels of education in terms of highest level achieved: elementary, high school, and post-high school education. Da Cunhas theoretical conclusions are presented in chapter five. As with the ot her studies, T1 is the dominant method for negation; Da Cunha tabulated a low T3 rate, merely 9 (.6%) out of 1465 total tokens. (15) Da Cunha 1996: T1 T2 T3 Total Tokens 1298 158 9 1465 Frequency 88.6% 10.8% .6% 100% 1.4.3 Alkmim 1999 Alkmim analyzes social factors that might influence the rise and use of clause-final negation, such as race, age, city of origin, etc. Her study took place in two rural towns, 21

PAGE 22

Mariana and Pombal, both of which are located in the southwestern state Minas Gerais. Both are largely Afro-Brazilian towns populated by those that are illiterate and in a lower social class. The idea that clause-fi nal negation is related to Afro-Brazilians is central to her work. Alkmim postulates that linguistic contact between slaves from West Africa and the Portuguese during the slave per iod of Brazil produced a change in how to negate a sentence. The numbers in table (16) are similar to other studies. (16) Alkmim 1999: T1 T2 T3 Total Tokens 1791 491 40 2322 Frequency 77.1% 21.2% 1.7% 100% 1.4.4 Camargos 2000 Camargos also works in Minas Gerais, although not as geographically specialized as Alkmim. He differs from some of t he previous works because he also considers sentences with other preverbal negative elem ents as being instances of T2 negation (e.g. nunca never, ningum nobody, etc). T he other authors did not count these as part of their corpus because their main goal was to investigate the weakening of the preverbal marker. Much like Roncarti, Camargos looks at t he syntactic situations where each type of negation occurs. He also takes into account social factors to understand the use and function of clause-final negation. 22

PAGE 23

(17) Camargos 2000: T1 T2 T3 Total Tokens 687 265 28 980 Frequency 70% 27% 3% 100% 1.4.5 Sousa 2004 Sousa (2004) studies the small Afro-Brazilian town of Helvicia in the state of Salvador in northeastern Brazil. In her study, following Schweglers analysis for Rio de Janeiro and Salvador, she does not include T3 as a type in its own right. Instead, she claims that T3 is merely T2 with deletion of the preverbal marker. It is therefore difficult to make too many claims about her findings for the purposes of this dissertation. Nonetheless, it stills points to tw o-thirds majority of T1. (18) Sousa 2004: T1 T2 and T3 Total Tokens 943 465 1408 Frequency 67% 33% 100% 1.4.6 Cavalcante 2007 Cavalcantes work is part of a larger effort to understand the connection of the creolization of Portuguese to the current state of BP. The Projeto Vertentes do Portugus Rural da Bahia out of the Federal Un iversity of Bahia analyzes rural dialects in the state of Salvador wher e there is a large Afro-Braz ilian population. Cavalcante concentrated on three communities: Cinzento, Rio de Contas and Sap. He conducted six interviews in each town with consultants who were not only born in the town, but 23

PAGE 24

whose parents were born there as well. The consultants were also divided into three age groups: 20 to 40, 40 to 60, and 60 and abov e. Cavalcantes study makes some important contributions to this dissertation, and chapter six examines them extensively. Of all the studies presented, his is the study where T1 has the lowest frequency of 66%, much like Sousas in Helvicia. (19) Cavalcante 2007 T1 T2 T3 Total Tokens 1343 568 115 2026 Frequency 66% 28% 6% 100% 1.4.7 Summary of Corpus Studies These sections briefly explained the purpose and results of six corpus studies that looked at CFN. In most ca ses, these authors did not atte mpt to make any specific syntactic hypotheses; however, they each have their conclusions, some of which I treat in chapters five and six. The data show considerable differences among the studies. These differences could be attributed to regi onal and socio-ethnic fact ors. For example, in the cases where clause-final negation is mo st prevalent, the main focus of the study was rural Afro-Brazilians (Sousa 2004, Cavalcant e 2007). In other cases, the studies were more of a cross-section of the Brazil ian population, which po ints to the depth of use of clause-final negation. Additionally, in some cases the studies limited themselves to typical no no sentences, which could greatly reduce the numbers of occurrences (Cavalcante 2007). From the data presented above, it is apparent that T1 is clearly the predominant form of sentential negation in BP. A conclu sion from the next sect ion is that T1 is 24

PAGE 25

allowed in all contexts in which negation is used (Cavalc ante 2007), and that T2 and T3 are marked forms (Schw enter 2005). As stated in the fi rst section of this chapter, one goal is to understand the motivations for T2 and T3 both syntactically and semantic/pragmatically. This section has given quantitative evidence that a difference exists among the three types of negation, t hus validating the reason for this study. 1.5 Syntactic and Semantic/Pragmatic Restrictions in BP for T2 and T3 An important part of this work determines the syntactic and semantic differences between the types. As mentioned, T1 is allowed in all contex ts in BP (Schwenter 2005), and in every case below, if the sentence is ungrammatical with CFN, its remedy is T1 negation (Biberauer & Cyrino 2009). Thus, t he preverbal negative ma rker differs from the clause-final no (T2 and T3), which is restricted in its use. Below are instances where T2 and T3 are restricted. Th is section begins by discussing the semantic/pragmatic restricti ons. Following that, I present several syntactic situations where the syntactic structure restricts the use of T2 and T3. This includes a brief discussion as to how these two types differ. 1.5.1 Discourse Requirements for T2 and T3 The main difference between T1 and the other two types is that T2 and T3 are linked to the discourse (Sch wenter 2005). This means that for T2 and T3 to be pragmatically felicitous, the proposition must have already been introduced into the discourse. These topics are explained in great er detail in chapter five where I explain the discourse requirements for CFN. Simply put, if the information being negated is not already introduced into the discourse, then the clause-final negative marker is not licit. In questions like (20) often termed out-of-the-blue que stions, what is being negated in 25

PAGE 26

the answer has not been introduced into the di scourse and therefore T2 and T3 are not felicitous (20) A. O que aconteceu? DET what happened What happened? B. Eu no terminei o trabalho (#no) 1S NEG finish.PST.1S DET work NEG I didnt finish the work If the negated information has been introduced in to the discourse, then the use of clause-final negation becomes felicitous. In t he following case the information finish the work has been introduced in to the discourse. (21) A. Voc terminou o trabalho? 2S finish.PST.2S DET work Did you finish the work? B. Eu no terminei o trabalho (no) 1S NEG finish.PST.1S DET work NEG I didnt finish the work 1.5.2 Embedded Clauses Some researchers have stated that T2 and T3 behave differently in embedded clauses (Schwegler 1991a, Cavalcante 2007). These researchers claim that T2 is permitted in embedded clauses and T3 is not. (22) gives an example of T2 in an embedded clause. (22) A Lcia pediu para os homens no fazerem o trabalho no. DET L cia ask.PST.3S PREP DET men NEG do.INF.3P DET work NEG Lcia asked for the men not to do the work *Lucia didnt ask for t he men not to do the work 26

PAGE 27

Example (23) is an attempt at placing T3 in an embedded clause. The interpretation of T3 in this case is that the negative marker has scope over the whole sentence and not just the embedded clause. (23) A Lcia pediu para os homens fazerem o trabalho no. DET Lcia ask.PST.3S PREP DET men do.INF.3P DET work NEG Lcia didnt ask for the men to do the work *Lcia asked for the men to not work While the interpretation of sentence (23) is of negation of the whole sentence, chapter six shows that this is deceptive because T3 is syntactically allowed in embedded clauses. Because previous rese archers didnt address the discourse requirements, the next examples show that if all the discourse requirements necessary to license T3 are in place, then T3 is allowed. In the case below, the question introduces a direct activation of the information he finished the test, and in this case it is possible to use T3 in the embedded clause. (24) Csar told Ana that he finished the test. (25) Eu acho que ele terminou o exame no 1S think that 3S finish.PST.3S DET test NEG I think that he didnt finish the test (26) shows that if there is CFN and t he main clause has T1, then the sentence final no is interpreted as part of the main cl ause. This is another reason why other researchers have claimed that T3 cannot be in complement clauses or in other embedded clauses (Schwegler 1991a, Cavalc ante 2007). The interpretation of (26) is (b). The preference for th is interpretation might be due to processing, where this 27

PAGE 28

example would is interpreted as having one inst ance of T2 in the main clause instead of two instances of negation, on in each of the matrix and embedded clauses. To negate the embedded clause and the main clause then the final no must not be present. (26) A Sienna no ouviu que a Maria disse uma mentira no. DET Sienna NEG heard.3S that DET Maria told.3S DET lie NEG a. *Sienna didnt hear that Maria didnt tell a lie b. Sienna didnt hear t hat Maria told a lie (27) has T1 in the main clause and T2 in the embedded clause. 1 This sentence is marginal. The remedy to this is given in (28) where the sentence final position is not used, i.e. T1 in both clauses. (27) ??A Lcia no pediu para os homens no fazerem o trabalho no. DET Lcia NEG ask.PST.3S PREP DET men NEG do.INF.3P DET work NEG Lcia didnt ask for t he men not to do the work (28) A Lcia no pediu para os homens no fazerem o trabalho. DET Lcia NEG ask.PST.3S PREP DET men NEG do.INF.3P DET work Lcia didnt ask for t he men not to do the work 1.5.3 Relative Clauses Relative clauses place restrictions on t he appearance of T2 and T3. In relative clauses in object position only T1 and T2 are gr ammatical, and T2 is only grammatical if the relative clause is sentence final. T1 and T2 negation of a relative clause in object position is given in (29) for T1, and (30) for T2 (Cavalcante 2007); additionally, (31) shows a sentence where the negative marker is not sentence final but is followed by a prepositional phrase with the work and is therefore ungrammatical. 1 Schwegler (1991a)s data are similar to what I have here, in that when there are three nos then the sentence is difficult to understand. In some cases, my consultants reject it all together. 28

PAGE 29

(29) O Joo ajudou o menino que no usa culos DET Joo helped.3S DET boy that NEG use.3S glasses John helped the boy that doesnt wear glasses (30) O Joo ajudou o menino que no usa culos no DET Joo helped.3S DET boy that NEG use.3S glasses NEG Joo helped the boy that doesnt wear glasses (31) *O Joo ajudou o menino que no usa culos no com a tarefa DET Joo helped.3S DET boy that NEG use.3S glasses NEG PREP DET work Joo helped the boy that doesnt wear glasses with the work T3 is never allowed in relative clauses. When there is only a sentence final negative marker, then the scope of negation is over the whole sentence (i.e. it is interpreted at T1 in the main clause) and not the relative clause as exemplified in (33) (32) O Joo ajudou o menino que usa culos no DET Joo helped.3S DET boy that use.3S glasses NEG *Joo helped the boy that doesnt wear glasses Joo didnt help the boy that wears glasses (33) has the relative clause in an indirect object position negated using T3, and it is ungrammatical. (34) has the same sentence negated with T2, and again, it is ungrammatical. Finally, in (35) T1 is grammatical, as is expected from its unmarked nature. In each case, the relative clause is before the direct obj ect and the sentence is ungrammatical. The brackets delineate the noun phrase containing the relative clause. (33) O Joo deu pro [menino que usa culos no ] o livro DET Joo gave.3S PREP.DET boy that use.3S glasses NEG DET book *Joo gave the boy that doesnt wear glasses the book (34) O Joo deu pro [menino que no usa culos no ] o livro DET Joo gave.3S PREP.DET boy that NEG use.3S glasses NEG DET book *Joo gave the boy that doesnt wear glasses the book 29

PAGE 30

(35) O Joo deu pro [menino que no usa culos ] o livro DET Joo gave.3S PREP.DET boy that NEG use.3S glasses DET book Joo gave the boy that doesn t wear glasses the book When a relative clause in subject position is negated with T2, then the sentence is ungrammatical, seen in (36) I bracket off the subjec t to show that the second no is within the relative clause and it is not negating the matrix verb ajudar to help. As expected, T3 is not allowed in subject re lative clauses either, as evidenced by (37) T1 would be grammatical. (36) [O menino que no usa culos no ] ajudou o Joo DET boy that NEG use.3S glasses NEG helped.3S DET Joo *The boy that doesnt wear glasses helped Joo (37) [O menino que usa culos no] ajudou o Joo DET boy that use.3S glasses NEG helped.3S DET Joo *The boy that doesnt wear glasses helped Joo BP allows subject doubling where the subj ect is repeated through the use of a pronoun. The subject is he and the boy is the doubled subject. (38) O menino, ele ajudou o Joo DET boy 3S helped.3S DET Joo The boy, he helped John In these cases, T2 in the doubled s ubject is also ungrammatical, seen in (39) (39) *[O menino que no usa culos no ] ele ajudou o Joo DET boy that NEG use.3S glasses NEG 3S helped.3S DET Joo The boy that doesnt wear glasses, he helped Joo 30

PAGE 31

1.5.4 Topics BP has topic-like expressions of various constituents (Martins & Nunes 2006; Galves 1998). A fronted topic object DP is given in (40) (40) Este filme, voc pode ver DET movie, 2S can.2S see.INF This movie, you should see Fronting is possible with T1, T2, and T3 in (41) (42) and (43) respectively. Cavalcante (2007) claims that to picalization as a result of fronting in T3 sentences is not allowed, but I have found no evidence for this. (41) Este filme, voc no pode ver DET movie, 2S NEG can.2S see.INF This movie, you cant see it (42) Este filme, voc no pode ver no DET movie, 2S NEG can.2S see.INF NEG This movie, you cant see it (43) Este Filme, voc pode ver no. DET movie, 2S can.2S see.INF NEG This movie, you cant see it BP also has verbal fronting. In these cases, the same types of restrictions hold, i.e. all types are grammatical. (44) gives an affirmative declarative sentence with verbal fronting, and the next three example show the three types of negation with fronted verb. (44) Falar, ela fala Ingls Speak.INF, 3SF speak.3S English As for speaking, she speaks English (45) Falar, ela no fala Ingls T1 Speak.INF, 3SF NEG speak. 3S English As for speaking, she doesnt speak English 31

PAGE 32

(46) Falar, ela no fala Ingls no T2 Speak.INF, 3SF NEG speak. 3S English NEG As for speaking, she doesnt speak English (47) Falar, ela fala Ingls no T3 Speak.INF, 3SF speak. 3S English NEG As for speaking, she doesnt speak English Sentence (48) illustrates that the topicalized VP cannot be negated (Bastos-Gee 2009). (48) No falar (no) ela no fala Ingls ( no) NEG Speak.INF NEG, 3SF NEG speak. 3S English NEG As for speaking, she doesnt speak English 1.5.5 Wh-Questions CFN (i.e. T2 and T3) is ungrammatical wit h wh-questions in which the wh-phrase is fronted, (49) (50) for T2, and for T3 (51) (52) All of the following examples can be made grammatical by elim inating the clause-final no. BP has several strategies for formulating wh-questions, and the examples bel ow show that neither clefted questions nor non-clefted questions allow T2/T3 CFN. T2 (49) *Por que ( que) voc no quis comer a pizza no? Why is that 2S NEG want.PST.2S eat.INF DET pizza NEG Why dont you want to eat the pizza? (50) *O que ( que) voc no quer comer no? What is that 2S NEG want.2S eat.INF NEG What is it that you dont want to eat? T3 (51) *Por que ( que) voc quis comer a pizza no? Why is that 2S want.PST.2S eat.INF DET pizza NEG Why dont you want to eat the pizza? 32

PAGE 33

(52) *O que ( que) voc quer comer no? What is that 2S want.2S eat.INF NEG What is it that you dont want to eat? BP also has wh-in-situ questions (Zocca 2008, Pires & Taylor 2007). (54) gives T1. (53) Voc comprou o que? 2S bought.PST.2S what What did you buy? (54) Voc no comprou o que? 2S NEG bought.PST.2S what What didnt you buy? Examples (55) shows that T2 in wh-in-situ questions is highly degraded. (56) shows that T3 is ungrammatical. (55) ??Voc no comprou o que no? 2S NEG bought. PST.2S what NEG Which books didnt you buy? (56) *Voc comprou o que no? 2S bought. PST.2S what NEG Which books didnt you buy? 1.5.6 N-Words and Negative Concord In BP, so-called n-words (as in Laka 1990, Giannakidou 2006) follow similar patterns found in other Romance languages; however, some differences can be seen with T2 and T3. Chapter seven treats in deta il the idea of Negative Concord (NC). For now, the term n-words are taken to be a theory-neutral term, meaning words with apparent negative morphology like nobody in English, nessuno in Italian, niemand in Dutch, etc. NC is a situation where negation appears to have multiple surface manifestations (seen on the negative marker and an n-word), but is only interpreted once (Giannakidou 33

PAGE 34

2006:328). BP, Spanish, Italian, and Europ ean Portuguese are cons idered non-strict NC languages. These languages do not allow negative markers (non, no, no ) to be ccommanded by n-words. These languages r equire n-words to be c-commanded by negative markers in most cases, for exampl e postverbal subject n-words or object nwords (Zanuttini 1997; Giannakidou 2000, 2006; Zeijlstra 2004). First, I present the data with n-words as subjects. (57) and (59) illustrate the behavior of non-strict negative co ncord in Spanish and BP with T1. In these examples the subject n-word c-commands the negativ e marker and the sentence has a double negative reading, not a negative concord reading. To elim inate the double negative reading, one omits the negative markers no and no as in examples (58) and (60) (57) Nadie no ayud a Juan Diego Spanish Nobody NEG helped PREP Juan Diego Nobody didnt help Juan Diego *Nobody helped Juan Diego (58) Nadie ayud a Juan Diego Nobody helped PREP Juan Diego Nobody help Juan Diego (59) Ningum no ajudou o Ablio 2 BP Nobody NEG help.PST.3S DET Ablio Nobody didnt help Ablio *Nobody helped Ablio 2 Martinez (2006) states that the following is allo wed in Mineiro Portuguese (State of Minas Gerais) i. Ningum no veio Nobody NEG came nobody came While she points to other studies, Alkmim (2001), this is the only example of the n-word and the preverbal negative marker not invoking a negative reading. All the preliminary work suggests that this is dialectal and is not seen in most dialects of BP. If it turns out that this claim is false, then more analysis would be required. 34

PAGE 35

(60) Ningum ajudou o Ablio Nobody help.PST.3S DET Ablio Nobody help Ablio T2 negation and subject n-words behave as expected, with a double negative reading due to the n-word ccommanding the negative marker. (61) Ningum no ajudou o Ablio no BP Nobody NEG help.PST.3S DET Ablio NEG Nobody didnt help Ablio *Nobody helped Ablio The facts are different with T3. While the n-word appears to c-command the negative marker, (62) a double negative reading does not result, in contrast to T1 and T2, as seen above. (62) Ningum ajudou o Ablio no Nobody help.PST.3S DET Ablio NEG Nobody helped Ablio Object n-words give grammatical NC readings with T1 (63) and T2 (64) but not with T3 (65) Independent of the cl ause-final negation in (65) the sentence is ungrammatical. (63) O Ablio no ajudou ningum DET Ablio NEG help.PST.3S nobody Abllio didnt help anyone (64) O Ablio no ajudou ningum no DET Ablio NEG help.PST.3S nobody NEG Abllio didnt help anyone (65) *O Ablio ajudou ningum no DET Ablio help.PST.3S nobody NEG Abllio didnt help anyone 35

PAGE 36

1.5.7 Imperatives BP allows all three types of negation in ne gative imperatives. BP has two types of imperatives: true negative imperatives and suppletive imper atives. True negative imperatives are formed from positive imperatives that are negated using the typical negative marker. Suppletive imperatives found in some Romance languages like Spanish, are commands formed using non-imperative verbal morphology, such as the subjunctive. (66) illustrates a Spanish imperative. This clause cannot be negated directly, (67) rather a subjunctive verb fo rm is used to negate an imperative, (68) (66) Lee eso! Spanish Read.2S.IMP that Read that! (67) *No lee eso! NEG read.2S.IMP that Read that! (68) No leas eso! NEG read.2S.SBJ that Dont read that Like Spanish and European Portuguese (EP) suppletive negative imperatives are grammatical in BP. (71) shows that BP also allows tr ue negative imperatives, although this is not the case in EP. 3 EP does not allow negative im peratives without use of the subjunctive. (69) L isso! EP and BP Read.2S.IMP that Read that! 3 This type of example would not be considered an im perative by Zanuttini (1994) and Postma & van der Wurff (2007). They require that imperatives have di fferent morphology from the other verb types, i.e declarative or subjunctive. Since BPs imperatives are identical to declaratives, these researchers would not consider them to be true imperatives. 36

PAGE 37

(70) No leias isso EP and BP NEG Read.2S.SBJ that Dont read that! (71) No l isso *EP and BP NEG read.2S.IMP that Dont read that! (72) and (73) show that T2 and T3 are also allowed in imperatives (Postma & van der Wurff 2007, Schwegler 1991a). (72) No l isso no BP NEG Read.2S.IMP that NEG Dont read that! (73) L isso no BP Read.2S.IMP that NEG Dont read that! 1.5.8 NPI Idioms NPI idioms are idioms that are felicitous only in the presence of an NPI licensor. In the case of BP the licensor is generally the preverbal negative marker. The three types of negation show a difference in thei r ability to license NPI idioms (Biberauer & Cyrino 2009). The following examples show that T2 allows the idiomatic meaning and T3 does not. The expression no tem um chinelo pra botar um prego literally means to not have a flip-flop to nail a nail with, seen in (74) As an idiomatic expression it means to not have anything. This meaning is retained in the T2 sentence (75) but lost in (76) with T3. (74) Ele no tem um chinelo pra botar um prego T1 3S NEG have DET flip-flop PREP.DET throw.INF DET nail He doesnt have anything 37

PAGE 38

(75) Ele no tem um chinelo pr a botar um prego no T2 3S NEG have DET flip-flop PREP.DET throw.INF DET nail NEG He doesnt have anything (76) Ele tem um chinel o pra botar um prego no T3 3S have DET flip-flop PREP.DET throw.INF DET nail NEG He doesnt have a flip-flop to nail a nail with *He doesnt have anything 1.5.9 Summary of the Data The following table shows the situations where each type is used. Note that means grammatical and X means ungrammatical. For relative clauses, T2 is licit in object relative clauses, but only if t hey are sentence final, thus the *. Table 1-1. Syntactic Restrictions for BP CFN Environment T1 T2 T3 Discourse Requirements X Allowed in Relative Clauses in Subject Position X X Allowed in Relative Clauses in Object Position X Allowed in Fronted Wh-questions X X Allowed in in-situ Wh-questions X X Allowed with Subject N-words X X Allowed with Object N-words X Licenses NPI Idiomatic Expressions X Allowed in Embedded Clauses Allowed in Imperatives Compatible with Topicalization 38

PAGE 39

The data raise a number of questions that ar e the focus of the c hapters that follow. I argue that the answers to these questions contribute to an explanation of the distributional differences and similarities presented above. i. What are the semantic and pragmatic differences between the three types of negation? ii. What is the analysis of the interaction of nwords and sentential negation in BP? iii. What is structural analysis of the three types of negation? 1.6 Structure of the Study The chapters of this work are organiz ed as follows. Chapter two is a brief overview of the theoretical framework to which this study adheres. This chapter highlights the fundamental prin ciples of minimalism and minimalist syntax. The main goal of this chapter is to set up the theoret ical base from which I can properly explain sentential and clausefinal negation in BP. Chapter three provides an overview of the cl ause structure of BP. In this chapter, I discuss the syntactic positions of the s ubject, verb, topics, wh-words, and some adverbs. The conclusions of this chapter have several repercussions for the overall analysis of clause-final negation. First, I argue for a split IP in BP consisting of AgrP dominating TP. The subject is in spec,AgrP and the verb is in T (Costa & Galves 2002). I also argue for a split CP in BP as well. Fronted DP topics are shown to be basegenerated in the specifier of a Topic Phra se and fronted wh-phrases are analyzed as occupying the specifier of a Focus Phrase, wh ich occurs below TopicP. Wh-in-situ in BP is analyzed as not involving movement at all. Additionally, wh-in-situ has a 39

PAGE 40

discourse-based restriction on it and it only occu rs if elements of the question are part of the Common Ground (Pires & Taylor 2007). Chapter four concerns itse lf with the preverbal negativ e marker. I present the main theoretical approaches to understanding the syntax of the preverbal marker in Romance languages. The chapter extends t he clause structure from chapter three and proposes a position for the preverbal negat ive marker as the head of NegP between AgrP and TP. Chapter five discusses one of the main topics of this dissertation, the semantic/pragmatic status of clause-final ne gation. Here, I discuss previous attempts at understanding CFN in terms of semantics and pragmatics. I reject several previous analyses of CFN as insufficient to explain the data (Schwegler 1991a, Da Cunha 2007). Instead, I follow Schwenters interpretation of Princes (1992) Discourse-Information Status theory. Roughly, w hat is negated with CFN must have been introduced into the discourse for it to be felicitous. Chapter six discusses the second main topi c of this dissertation, the syntactic status of clause-final negation. This chapt er discusses past attempts to understand the syntax of CFN. I argue that previous acc ounts do not fully capt ure the requirements and restrictions on CFN. The final part of chapter six introduces my proposal for CFN in BP. I propose that CFN is phrasal movement of the whole AgrP to a specifier position of a lower TopicP phrase in the left periphery. The head of this lower topic phrase is a no for negative sentence and a sim for some positive sentences. Chapter seven tests the syn tactic proposals made in c hapters five and six using Negative Concord and n-words. In this chapter, I discuss the status of n-words in BP. 40

PAGE 41

From there, I combine the analysis from chapter six with the analysis of Negative Concord to argue for the existence of thr ee negative formatives in BP. The three markers are: a preverbal no with an interpretable negative feature, a clause-final no with an uninterpretable negative f eature, and a clause-final no with an interpretable negative feature feature. The first mark er appears in T1 and T2, the second appears in T2 only, and the third appears exclusively in T3. Chapter eight is a summary and a conclu sion. It summarizes the findings of the dissertation and highlights the ma jor theoretical implications. 41

PAGE 42

CHAPTER 2 THEORETICAL FOUNDATIONS 2.1 Introduction This chapters main goal is to present and explain the theoretical assumptions used in this dissertation. A main purpose of this chapter is to describe the linguistic tools that will be used to understand negation in Brazilian Portuguese, both preverbal and clause-final. Two areas of linguistic theory are covered: certain aspects of minimalist syntax and the left periphery. Section 2.2 covers the Minimalist Program and some aspects of minimalism that will be rele vant in the analyses in later chapters. Section 2.3 discusses the left periphery, making spec ific use of the pr oposals from Rizzi (1997). 2.2 Minimalism This work assumes a rather conservative version of minimalist syntax from the 1990s and early 2000s (Adger 2003, Radfor d 2004), which includes X-bar theory (Carnie 2002) and certain mec hanisms that are used to driv e movement in the syntax (Chomsky 1995, 2000, 2001, Hornstein 2001, Hornstein, Nunes & Grohmann 2005, Bokovi & Lasnik 2007). Section 2.2.1 presents the main theoretical ideas and motivations behind a minimalist syntax. Section 2.2.2 presents my assumptions about the Probe-Goal system and Feature Checking. 2.2.1 Fundamental Principles of Minimalism The minimalist program has as a research goal simplicity, elegance, parsimony, and naturalness (Hornstein 200 1:4). These somewhat abstract terms are concretely defined by Hornstein, Nunes & Gr ohmann (2005:ch 1) (Haddad 2007:24): 42

PAGE 43

i. Naturalness implies that only notions that co rrespond to self-evident facts about language should be preserved. ii. Simplicity follows from naturalness. If only nat ural notions are preserved and all other theory internal notions are removed, the grammar becomes simpler. Further, given two theories A and B that ar e equal in every way except that A has fewer rules than B, A is considered superior. iii. Economy is pertinent to derivations and derivat ional rules. Everything else being equal, a derivational step that requires the least effort (e.g., fewer steps) and that happens only when necessary is optimal. Coupled with these goals and definitions, the minimalist program assumes some important ideas about language. Hornstein (2001) presents these ideas as follows: i. sentences are the basic linguistic units ii. sentences are pairi ngs of sounds and meaning iii. there is no upper bound to the number of sentences in any given [Natural Language] iv. sentences show displacement proper ties in the sense that expressions pronounced in one position are interpreted in another v. sentences are composed of word s organized into larger units with hierarchical structure, i.e. phrases. Chomsky and others propose that the Language Faculty (FL) of the mind generates sentences or linguistic expressions. FL has interfaces with the articulatoryperceptual (AP) system and t he Conceptual-Intentional (C I) system. Phonological Form (PF) is the representation at the interface between the FL and the Articulatoryperceptual system and Logical Form (LF) is the representation at the interface between the Language Faculty and the Conceptual-I ntentional system. These two representations yield the pronounced form a nd meaning of a sentence, and these are 43

PAGE 44

the only two levels of representation in the t heory. Many visual models attempt to show the relationships just ment ioned, and the model that is assumed generally by those working in a minimalist framework is as follows (Chomsky 1995, Hornstein, Nunes, & Grohmann 2005): i. Lexicon Numeration Spell-Out Phonological Form (PF) Logical Form (LF) Chomsky explains that lexical items (LIs) bundle three types of features: phonological, semantic, and formal and the lexic on is the storage of these lexical items. Phonological features include the informa tion needed to be interpreted at PF, i.e. roughly what would be consider ed the sounds of the Intern ational Phonetic Alphabet. Semantic features are features, such as [+animate], which are interpreted at LF. Finally, formal features are things to whic h the computation system is sensitive, and they can be interpretable or uninterpretable (see section 2.2.2 for more details). It is important to note that the lexicon of a language may have lexical entries that are phonologically null. These null elements are interpreted at LF even though they have no pronounced form. The numeration is a selection of elements from the lexicon selected to take part in the derivation. Structure build ing operations apply to element s in the numeration, after which the sentence is Spelled-out, i.e. sent to the interfaces. Spell-out is where the phonological features are mapped onto PF and the other features continue towards LF. Syntactic operations that occur before Spell-out are considered overt operations 44

PAGE 45

because they are visible through the phonolo gy, and operations after Spell-out are covert because they are not visible to the phonology. At Logical Form the system determines whether a sentence is well-fo rmed or not, including whether all the uninterpretable formal feat ures have been checked. I assume four main computational or syntactic operations: Select, Merge, Agree and Copy Select essentially picks something from the lexicon. Merge takes two syntactic objects and combines them into one. Merge is the operation that builds syntactic trees. The operation Agree estab lishes a relation (agreement, Case-checking) between an LI and a feature F in some rest ricted search space (its domain) (Chomsky 2000:100). Finally, Copy duplicates a linguistic object and makes it available for Merge. Movement in the syntax is a result of Copy plus Merge. I will leave aside any technical discussion of Select, Merge, and C opy and return to Agree below. Another main idea relevant to this curr ent work is that a clause is divided into three domains: the left periphery, or CP, t he inflectional domain, or IP, and the verb phrase domain, or vP (Chomsky 2000, Grohm ann 2003). These categories have been shown to be made up of subparts. The represent ative works for each are: Rizzi (1997) for the CP, Pollock (1989) fo r IP, and Larson (1988) for VP. I discuss the left periphery in BP in section 2.3 of this chapter. 2.2.2 Probe-Goal System and Feature Agreement An important aspect of syntact ic representations is form al features and agreement among these features. Features on lexical items can be of two kinds: interpretable, represented by [iF], or uni nterpretable, represented as [uF], where F is some morphosyntactic feature such as person, number, negation, topi c, focus, question, etc. An interpretable feature is one that has an interpretation at LF and an uninterpretable 45

PAGE 46

feature is one that does not. For example phi-features (per son, number) are interpretable on DPs but uninterpretable on agreeing heads such as adjectives and verbs. An uninterpretable feature is illegitimate at LF and vi olates what is called the Principle of Full Interpretation, which states that all synta ctic objects should be fully interpretable at the interfaces (PF and LF) (Chomsky 1995). To avoid a violation of Full Interpretation, uninterpretable features must be eliminated in the syntax before they reach LF. The mechanism that remove s uninterpretable f eatures is called Feature Agreement Informally, Feature Agreement, or Ag ree, allows an uninterpretable feature to check itself against an interpretable counterpart and thus become invisible. Agree is accomplished through a system of probes and goals. A probe is a feature that is uninterpretable [uF]. The probe searches its syntactic domain to find a corresponding interpretable feature [iF], which is called the goal. If the probe finds a goal and the two features match, the uninterpretable [uF] is checked and deleted. The syntactic domain in which a probe searches for a goal is its c-command domain; a probe looks down to find a goal. In addition, I adopt a proposal put forth in Rezac (2003) which argues that although probes normally look within their c-command domain to find a goal, if no goal is found, then the probe can wait for additional material to be merged into the derivation. This means t hat the field of search ultimately available to a probe includes elements higher in the structure, outside its c-command domain. This upward search possibility for probes will be employed to understand Negative Concord (see section 7.2.4 ),following work by Zeijlstra (2004). Since Agree can happen at a distance and does nt require movement to result in feature checking, something additional mu st cause elements to move. For the 46

PAGE 47

purposes here, I claim that what forces mo vement is that a f eature on a probe can be strong, which requires that it be checked in a more local configuration than with Agree alone. A strong uninterpretable feature triggers movement of something lower in the structure to the specifier pos ition of the probe carrying the strong feature. A strong feature is represented as [F*] and must be c hecked in a specifier-head configuration. In this section I have briefly discussed t he major framework and linguistic tools that this dissertation employs. Additional syntactic machinery will be presented as necessary. 2.3 CP: The Left Periphery This section presents the theoretical assumptions about the left periphery that have risen due, in part, to Rizzi (1997). Section 2.3.1 introduces Rizzis theory. Then, I take the idea of an expanded CP and apply it to BP in section 2.3.2 By doing this, I show that Rizzis proposal is compatible wi th BP. In fact, as la ter chapters show, an expanded CP is fundamental to a proper understanding of BP clause structure and CFN. 2.3.1 Rizzis Split CP CP is taken to be the region of the claus e where certain grammatical notions such as topic, focus, clause type, etc. are ex pressed (Chomsky 1995). Chomsky, referring to Cheng (1991, 1997), states that C serves as a force indicator, where force is understood to mean clause type, such as declarative, interrogative, imperative, etc. (Chomsky 1995:69). Rizzis (1997) work develops this idea by showing that CP actually consists of a number of f unctional projections, each of which independently hosts elements related to topic, focus, clause type, etc. 47

PAGE 48

Before Rizzis work, for example in Lectures on Government and Binding (Chomsky 1981), the left periphery was a single CP projection that housed a host of different elements, including wh-phrases, co mplementizers of varying types, topics, and focused elements. In the early 1990s Lasnik & Saito (1992) and McCloskey (1991) proposed that topics and sententia l adverbs adjoin to the main clause. However, in light of cross-linguistic evidence, this adjunction that some proposed for topics and sentential adverbs was not very constrained, and as such, behavior of this type seemed fit to be eliminated from the grammar Rizzi (1997) and others after him (Beninc 2001, Beninc & Polleto 2004, Ci nque 1999, 2002, Rizzi 2004, Po lleto 2000) proposed that CP be split into several X-bar compatible projections, such as TopicP, FocusP, and ForceP, among others. This split created dedicated head and specifier positions for elements commonly found in the CP region. The Force projection is what determines t he type of clause. The Topic projection hosts topics of the sentence and w hat follows is assumed to be the comment The Focus projection hosts focused elements and what follows is assumed to be the presupposition. Finally, the Fini te projection determines the (non-)finiteness of the TP. Rizzis structure looks like (1) (see Rizzi 2001:289 for the incl usion of an Int(errogative) Phrase). The indicates that TopicPs c an iterate (but see Beninc & Polleto 2004 for evidence against this). (1) ForceP TopicP* IntP Topi cP* FocusP TopicP* FinP Rizzi assumes that not all projections in t he split CP are necessarily present in all clauses. It is reasonable to a ssume that the topic-focus system is present in a structure 48

PAGE 49

only if needed (Rizzi 1997: 288). Rizzi relie s on Agree to fill these positions. If the topic position is filled, for example, then a strong uninterpr etable topic feature [uTop*] must exist on the head of TopicP to drive movement to this position, as discussed above. 2.3.2 The Split CP and BP This section provides evidence for Rizzis various projections in BP. In Rizzis approach, finite complementizers are insert ed into Force. The Italian sentence (2) and the BP sentence (4) have the complementizers che and que respectively, in Force, which causes the embedded clause to be finite. Rizzi contrasts these complementizers with prepositional complement izers, such as Italian di which introduce non-finite clauses. While BP does not have a construction exactly like the Italian example (3) it does use a preposition-like co mplementizer to introduce non-finite clauses as in (5) Rizzi claims that di in Italian is Fin; however, I suggest below that this is not the case for BP para (2) Credo che loro apprezzerebbero molto il tuo libro. Believe.1S COMP 3P appreciate.FUT.3P much DET 2S book I believe that they will appreciate your book very much (3) Credo di loro apprezzare molto il tuo libro. Believe.1S PREP 3P appreciate.INF much DET 2S book I believe that they appreciate your book very much (4) A Sienna pediu que eles trabalhassem. DET Sienna ask.PST.3S COMP 3p work.SBJ.PST.3P Sienna asked that they work (5) A Sienna pediu para eles trabalharem. DET Sienna ask.PST.3S PREP 3P work.INF.3P Sienna asked that they work 49

PAGE 50

Rizzis claim is that the Italian complementizers che and di cannot be the head of the same projection because of how they inte ract with topics and focused elements. In the case of finite complementizer che, topics must necessarily follow che. With the nonfinite complementizer di topics must necessarily precede di The topics are underlined and the complementizer is bolded in the following examples. Notice that in (6) che precedes the topic; whereas, in (7) and (8) the di must follow the topic. (6) Credo che il tuo libro loro lo apprezzerebbero molto. Believe.1S COMP DET 2S book 3P CL.3S appreciate.FUT.3P much I believe that your book, they will appreciate it a lot. (7) *Credo di il tuo libro apprezzarlo molto. Believe.1S PREP DET 2S book appreciate.INF-CL much I believe that your book, they will appreciate it a lot. (8) Credo il tuo libro di apprezzarlo molto Believe.1S DET 2S book PREP appreciate.INF-CL much I believe that your book, they will appreciate it a lot. In the case of the BP finite complementizer que BP behaves like Italian in that topics cannot precede que Assuming that que is the head of ForceP, this shows that the order of the functional heads is Force Topic, as Rizzi proposes. (9) Acho que os meninos o Joo viu no mercado. Believe.1S COMP DET boys DET John saw.PST.3S PREP.DET market I believe that the boys J ohn saw them in the market (10) *Acho os meninos que o Joo viu no mercado. Believe.1S DET boys COMP DET John saw.PST.3S PREP.DET market Based on examples such as (7) and (8) Rizzi argues for a head called FinP, which is where he places the prepositional co mplementizer di in Italian. BP does not seem to have any overt evidence of th is. The interaction of topics with BP para is the 50

PAGE 51

same as with que in that topics cannot precede para The examples below are ungrammatical on the interpretation where the topic preceding the complementizer is associated with the embedded clause. (11) *A Sienna pediu no mercado que eles trabalhassem. DET Sienna ask.PST.3S PREP.DET market that 3p work.SUBJ.PST.3P Sienna asked that in the market they work (They work in the market) (12) *A Sienna pediu no mercado para eles trabalharem. DET Sienna ask.PST.3S PREP.DET market PREP 3P work.INF.3P Sienna asked that in the market they work(They work in the market) From this, I assume that both para and que in BP realize Force and Fin is not overtly realized in BP. Rizzi (1997, 2001) also shows that topi cs can precede fronted wh-words. Again, the same can be said of BP. In the Italian example in (13) a Gianni to Gianni is fronted to the left of the wh-word. T he same happens in the BP example in (14) with pra L to L to the left of the wh-phrase. (13) A Gianni, che cosa gli hai ditto? PREP Gianni, what thing DAT.3S have.2S said To Gianni, what did you say (14) Pra L, o que que voc deu? PREP.DET L, what that 2S gave.PST.2S To L, what did you give? Rizzi assumes that the landi ng site of wh-phrases is the specifier of a focus phrase, FocusP, below TopicP and I will follow his conclusion. 4 Evidence for this claim comes from the incompatibility of wh-phrases with a fronted focused phrase: 5 4 Section 3.5 will explain that the que in example (14) is not the same as que the complementizer. Rather this is a Q operator. 51

PAGE 52

(15) *PRO ZE o que que voc deu (no pro Lionel) PREP.DET Ze what that 2S give.PST.2S (NEG PREP.DET Lionel) TO ZE what did you give (not to Lionel) (16) o que que PRO ZE voc deu (no pro Lionel) what that PREP.DET Ze 2S give.PST.2S (NEG PREP.DET Lionel) TO ZE what did you give (not to Lionel) (15) has the focused PP before the wh-word, and (16) has it following the whword. In both cases, the result ungrammatical. Rizzi takes this fact to mean that the target for wh-fronting is also FocusP, thus the mutually exclusiv e nature of these two items. In addition, only one focused phrase is allowed, confirming that FocusP cannot iterate: (17) *PRO CESAR O LIVRO eu vou dar (no pra Sienna o caderno) PREP.DET Cesar DET book 1S go.1S give.INF (NEG PREP.DET Sienna DET folder) The book, to Cesar, tomorrow, I will give (not to Sienna the folder) In contrast, a clause may contain multip le topics, indicating that TopicP does iterate. (18) has two topics, the book and to Cesar. (18) O livro pro Csar amanh eu vou dar DET book PREP.DET Cesar tomorrow 1S go.1S give.INF The book, to Cesar, tomorrow, I will give These data support the claim that fronted topics and focus elements are located in distinct positions in BP, much like in Itali an. Ordering these XPs, we observe ForceP TopP* WH/FocusP. Chapter six will discuss a second topic position below FocusP. The fact that there are two topic positions is predicted in Rizzis work, and it is 5 Ambar (1999) explains how focus works in Europea n Portuguese, and the results are similar for BP. The meaning of focus for our purpose here is new information, as opposed to old information. 52

PAGE 53

something that will aid in explaining CFN. The topics presented in this chapter are related to topics discussed furthe r in chapter three. There, I give evidence to show that the higher topics are base-generat ed. Chapter six explains that the lower topics are cases of movement of AgrP to the specifier of the lower TopP. To distinguish the two topic positions, I will label them Topic1 and T opic2. This gives following the structure for BP: (19) ForceP Topic1P* WH/FocusP Topic2P 2.4 Conclusion This chapters goal was to present the theoretical base for the discussion of negation in Brazilian Portuguese. I have presented some assumptions regarding Agree that will be useful in what follows. The sec ond half of this chapter dedicated itself to Rizzis expanded CP (1997). Section 2.3.2 showed that Brazilian Portuguese, much like its cousins in the Romance family, is compatible with Rizzis theory. This is important because the idea of an expanded CP is central to the proposals that this dissertation makes. 53

PAGE 54

CHAPTER 3 THE CLAUSE STRUCTURE OF BRAZILIAN PORTUGUESE 3.1 Introduction The main goal of this chapter is to exam ine some elements of the clause structure of Brazilian Portuguese (BP). A proper understanding of Port uguese clause structure is imperative for a complete analysis of the negative marker and to understand the syntactic proposals in later chapters. I di scuss the syntactic positions of the subject, verb, adverbs, topics, and wh-words, and th e structures I present are assumed throughout this dissertation. The struct ures presented build upon the existing work related to Portuguese clause structure, and are aligned to the theories discussed in chapter two (Galves 1989, Mioto 1992, Kato 1999, Costa & Galves 2002, Silva 2004, Rodrigues 2004, Kato 2009, Nunes 2009). This chapter is organized as follows. Section 3.2 deals with the subject and its positions. In section 3.2.1 I consider the option debated in the literature regarding the preverbal subject position as either an Aor A'position and argue that it is an Aposition. Section 3.2.2 uses adverbs and floating quantifie rs to further investigate the Aposition. Section 3.3 investigates Portuguese verb mo vement and the position of the verb. Sections 3.4 and 3.5 discuss the syntax of topics and wh-questions, respectively. The concluding section of this chapter summarizes the results. As a way of anticipating the conclusions that come out during this c hapter, the clause-struct ure that I propose for BP is represented by the basic schematic in (i). i. [ForceP Complementizer [Top1P t opic-phrase [ FocusP wh-phrase [Top2P 6 topic-phrase [AgrP subjectk [NegP no [TP Vi [vP subjectk [VP Vi ]]]]]]]] 6 This lower topic position will be disc ussed in chapter six. The higher topic position is the main focus of the topics discussed in this chapter. 54

PAGE 55

Wh-phrases and topics occupy A' positions in the left periphery of the clause. The inflectional layer below these projections consists of at l east two projections, AgrP and TP. The preverbal subject occupies the specifier of the higher projection, spec,AgrP, which is an A-position. The verb raises from V to the head of the lower projection, T, but no higher (i.e. not to Agr). The inflectional layer dominates vP and VP, where the external argument, the verb, and ve rbal complements originate. 3.2 Subjects in BP This section addresses t he theoretical debate surroundi ng the preverbal subject position in Romance generally and BP spec ifically. For other Romance languages, recent works suggest that the preverbal subjec t is an Aposition or an A' topic position. From this perspective, I analyze the preverbal subject position in BP I reject the A' position analysis for BP, however, and concl ude that the BP preverbal subject position is a canonical A-position. 3.2.1 The Debate over the Ro mance Preverbal Subject Position Many suggest that BP is transforming fr om a null subject language to a non-null subject language (see all the articles in Ka to & Negro 2001 and Rodrigues 2004). As such, BP finds itself in the middle of an in teresting debate among linguists as to the position of preverbal subjects in null subject languages (NSL) (Belle tti 1990 for Italian, Alexiadou & Anagnostopoulou 1998 for Gr eek, Barbosa 1995, 2004 for European Portuguese, Ordez 1997, Ordez & Trevio 1999 for Spanish, Cardinaletti 2004 for an extended discussion on the various subject positions that possibly exist in natural 55

PAGE 56

language 7 ). The canonical word order for Ro mance Languages in general is SVO, and the debate centers around whethe r the preverbal subject po sition is an Aor and A'position. This debate is useful in the ca se of BP because it he lps inform the possible subject positions. I discuss the two main analyses below, first A'-position and then Aposition. Researchers have argued that the preverbal subject occupies an A' topic position in the left periphery of the cl ause, outside of the inflecti onal domain (Zubizarreta 1998, Ordez 1997, Alexiadou & Anagnostopoulou 1998, and other s). Alexiadou and Anagnostopoulou (1998) give a theoretical argument in favo r of the preverbal subject being in an A' position. They argue that in NSLs like Greek and Spanish, the EPP is satisfied by movement of the verb to T. That is, verbal agreement morphology in such languages includes a nominal element which satisfies the EPP (Alexiadou & Anagnostopoulou 1998:516). DP movement to spec,IP is thus unnecessary for EPP satisfaction. Further, they propose that subjec ts can check their Case features in their base position inside vP. Thus, subjects al so do not need to move to spec,TP for Case reasons. Given that there is no motivation for A-movement of the subj ect to spec,TP, it must be the case that the preverbal subjec t is not in an A-position. Rather, they conclude, it is in an A' topic position (see also Ordez 1997, Ordez & Trevino 1999). Movement to this A' position is motivated by discourse considerations, perhaps a topic feature on the subject (Costa 2004:13). 7 Cardinaletti (2004) discusses several subject positions. She bases this on data from comparing the Romance languages to things like double subject constr uctions in Icelandic. While it might be true that several subject positions are reserved for different types of subjects, as far as I can tell these are all in the same spot in BP, which is above TP. Essentially a subject in BP could move to spec of TP and move again to some other subject position higher up, which is essentially what AgrP d oes. I will not pursue this line of thought in this work, rather see the facts presented in Kato (2000) and Rodrigues (2004). 56

PAGE 57

Ordez & Trevio (1999) provide an empiri cal argument for this conclusion based on ellipsis data in Spanish. In each of the following examples, the second conjunct contains a stranded constit uent and the polarity item tambin also. According to the authors, in (1) the stranded phrase is the subject, in (2) it is an object, and in (3) it is an indirect object. (1) El le dio unos libros a Pia y Pepe tambin [le dio unos libros a Pia] 3S CL-gave.3S some books PREP Pia and Pepe also [CL-gave.3S some books PREP Pia] He gave some books to Pia and Pepe also [gave some books to Pia] (2) Unos libros le dio a Pia y unos cuadros tambin [le dio a Pia] Some books CL-gave.3S PREP Pia and some paintings too [CL-gave.3S PREP Pia] He gave some books to Pia and some paintings also [he gave to Pia] (3) A Pia le dio unos libros y a Sara tambin [le dio unos libros] PREP Pia CL-gave.3S some books and PREP Sara too [CL-gave.3S some books] To Pia, he gave some books and to Sara also [he gave some books] Ordez & Trevio 1999 argue that in (2) and (3) a phrase has been fronted followed by ellipsis of the constituent followi ng that phrase. More specifically, they assume that the (in)direct object has been topicalized to spec,TopP and the elided constituent is TP (following Lobeck 1995), the complement to Top: (4) [TopP unos cuadrosi [Top' Top [ TP le dio ti a Pia]]] In order to give the same analysis to (1) they claim that the subject must also be in an A' position, as it appears to occur in the same position as the other verbal elements in (2) and (3) If the subject were lower, in IP, it would be included in the ellipsis site. Their claim is also cr ucially dependent on an analysis of null-subject languages, where the subject need not necessarily move to spec IP. As such, if there is anything in a preverbal position, they cl aim that element has been fronted, even the subject. 57

PAGE 58

Alexiadou and Anagnostopoulou (1998:503) provide another argument from Spanish, based on the complementary distribution of prev erbal subjects with some preverbal adverbs. (5) shows that when the adverb temprano and the subject Julia are both to the left of the verb the sentence is ungrammatical. When just one precedes the verb however, the result is grammatical, (6) (5) *Temprano Julia sala de casa early Julia leave.IMP.3S PREP house (6) Temprano sala Julia de casa Early leave.IMP.3S Julia PREP house Julia used to leave/was l eaving early from the house Ordez (1997) shows that the same type of asymmetry also exists with negative quantifiers and subjects. (7) demonstrates that Spanish allo ws negative quantifiers in a preverbal subject position. They can also be fronted from objec t position, as in (8) where the object nada has been fronted. If both a negative quantifier and the subject are to the left of the verb, however, the sentence is ungrammatical, (9) (7) Nadie le debe la renta a Mara. nobody CL-owes.3S DET rent PREP Maria Nobody owes Maria the rent (8) Nada le debe Juan a sus amigos nothing CL-owes.3S Juan PREP his friends Juan owes nothing to his friends (9) *Nada Juan le debe a sus amigos nothing Juan CL-owes.3S PREP 3S friends Juan owes nothing to his friends Such data can be accounted for if there is one A'-position before the verb and it can be occupied by a fronted element (tempora l adverb or negative quantifier) or by the 58

PAGE 59

subject but not both. If t he subject had its own A-position, such complementary distribution would be unexpected. Following Uribe-Echevarria (1992)s original observations for Spanish, Ordez (1997) also argues that the s ubject is an A'position based on scope. According to him, if the preverbal subject were in an A' pos ition, then its scope wo uld freeze (c.f. Barss 1986). By this he claims that when moved to an A'-position, a quantificational element is no longer in a position where it scopa lly interacts with anot her quantificational element in the sentence. Thus, in example (10) where the subject is postverbal, scopal ambiguity arises due to the possibility of the subject raising at LF above aquin In this case, the universal quantifier has bot h wide and narrow scope. The wide scope meaning is that each senator lo ves a different person. In the narrow scope meaning, all senators love the same person. (10) Aquin dices que amaba cada senador? PREP.whom say.2S that loved.3S each senator Who did you say each senator loved? Whom > each Each > whom (11) no longer has a scopal ambiguity because the subject has moved to the left of the verb. In this case the only reading is one where cada senador loves the same person, i.e. narrow scope reading of t he universal quantifier, whom > each. Ordez claims that this is more evidence in favor of the pr everbal subject being in an A'-position. (11) Aquin dices que cada senador amaba? PREP.whom say.2S that each senator loved.3S Who did you say each senator loved? Whom > Each *Each > Whom 59

PAGE 60

Alexiadou & Anagnostopoulou (1 998) continue Ordezs line of thinking and claim that multiple quantifiers also are evidence for an A'position. Under their analysis, when a quantifier moves to an A-position its pos ition allows it to interact with other quantificational elements in t he sentence and as such creates scopal ambiguities (e.g. van Riemsdijk & Williams 1986, May 1985). The relationship that allows ambiguity does not exist when the quantifier is realized in an A'-position. If the preverbal subject in Greek is an A'-position, then this explai ns why some can only have wide scope in (12) but not in (13) where scopal ambiguity remains (Alexiadou & Anagnostopoulou 1998). (12) Kapios fititis st ihiothetise kathe arthro Some student filed every article Some student filed every article some>every *For every article, there is some student that filed it *every>some (13) Stihiothetise kapios fititis kathe arthro filed some student every article Some student filed every article some>every For every article, there is some student that filed it every>some Based on these pieces of evidence and ot hers, Ordez & Trevino, Alexiadou & Anagnostopoulou, and others conclude that the subject in Spanish is in the left periphery using Rizzis (1997) terminology, and thus the subject position is an A'position. This type of ar gument is the reason that many believe that a preverbal subject is in an A'-position in the CP domain. Although there are many argum ents in favor of an A' subject position, they are not universally accepted (Costa 2004, Holmberg 2005, Roberts 2004). Rizzi (1982) first proposed a traditional analysis in which he argues that subjects in Italian are in an A60

PAGE 61

position, and often spec,IP is considered an A-position. More recently, Goodall (2001) for Spanish and Costa (2004) for Eur opean Portuguese have defended this claim, arguing against the evidence presented above One piece of evidence for the Aposition analysis comes from "neutral" word order. Adrago & Costa (2004:1) claim that in EP the neutral word order is SV(O) (see Card inaletti 1997 for similar facts in Italian). Out-of-the-blue and all-focus questions demonstrate this, where the entire answer is new information. SV(O) word order is obligator y in such situations. Any word order but SV(O) is infelicitous: (14) O que que aconteceu? What happened? (15) O Pedro partiu o brao DET Pedro broke.3S DET arm Pedro broke his arm (16) #Partiu o Pedro o brao Broke.3S DET Pedro DET arm Broke Pedro his arm (17) #O brao, o Pedro partiu-o DET arm DET Pedro broke.3S-CL His arm, Pedro broke it The infelicity of (16) and (17) suggests that these are not neutral word orders and that some constituent is i nappropriately marked as new or old information here. SVO order in (15) in contrast, is acceptable because no constituent has been singled out with special discourse status. If this is correct, then t he preverbal subject position cannot be a discourse-oriented A'-position. Instead, it is an ordi nary A-position which generally carries no discourse restrictions. 61

PAGE 62

Following Rizzi (1997), Goodall (2001) argues that bare quantifiers are not able to be topics, as in (18) In this example, the front ed PP to nobody is clearly in an A'position and is ungrammatical. (18) *A nadie, Juan lo ha visto PREP nobody, Juan CL has.3S seen No one, Juan has seen However, bare quantifiers can be subjects. If the subject were left-dislocated, then sentence (19) would be ungrammatical. Since, the su bject is in spec,IP, the sentence is grammatical. (19) Nadie ha visto a Juan nobody has.3S seen PREP Juan No one has seen Juan Additionally, Casielles (1997) shows that bare nouns, such as nios children in Spanish can appear postverbally as subject, but they cannot be preverbal subjects. (20) Jugaban nios en el parque Play.IMP.3P children PREP DET park Children were playing in the park (21) *Nios jugaban en el parque children play.IMP.3P PREP DET park Bare nouns are also allowed in topic posit ions. In this example the bare noun libros books is topicalized. (22) Yo a l libros no le dejo 1S PREP 3S book NEG CL-lend.1S as for me, books I dont lend him 62

PAGE 63

Casielles concludes that it is unlikely t hat preverbal subjects and topics share the same syntactic position in Spanish becaus e bare nouns can be topics, but cannot be preverbal subjects. In summary, the issue of whether the prev erbal position in NSLs like Spanish and EP is an Aor A'-position has not been decided. In the next section, I examine BP in this light. I argue that the facts ar e clearer for BP and the preverba l subject position is an Aposition. 3.2.2 BPs Preverbal Subject 8 The position of the subject in BP is fix ed and only a preverbal position is allowed (Costa 1998, Kato 2000, Silva 2001, Silva 2004): (23) a. SVO O Csar consertou o carro DET Csar fixed DET car Csar fixed the car b. *VSO *Consertou o Csar o carro fixed DET Csar DET car Csar fixed the car 8 I will not discuss postverbal subjects in the work. Briefly, BP subjects can be to the right of the verb; however, VS order is restricted, for the most part, to unaccusative verbs (Silva 2001, Silva 2004). Since these are objects of the verb, it is assumed that they are allowed to stay in-situ. i) A gua congelou ii) Congelou a gua. det water froze.3sg det water froze.3sg The water froze. The water froze BP does not allow postverbal subjects with tr ansitive or unergative verbs (Silva 2001:84). iii) *Leu a revista a Bia iv) *Leu a Bia a revista Read the magazine the Bia Read the Bia the magazine v) *Brinca a criana em casa vi) *Brinca em casa a criana Plays a child in house Plays in house a child 63

PAGE 64

c. *VOS *Consertou o carro o Csar fixed DET car DET Csar Csar fixed the car In what follows I claim that for BP the subject position is unambiguously an Aposition, spec,IP, in agreement with Silva (2004), Barbosa (2004), Martins & Nunes (2006), Pires (2007), and other s. The arguments come fr om non-complementarity between preverbal subjects and temporal adverbs, doubled subjects, and preverbal quantifiers. 3.2.2.1 Temporal Adverbs Recall that Alexiadou & Anagnostopoulou (1998) claim that complementarity between preverbal subjects and temporal adverbs in Spanish points toward the preverbal subject and adverb bot h occupying an A-position in the left periphery. This same complementarity does not exist in BP as example (24) b shows. This sentence is similar to the Spanish example (5) repeated as (24) a from above, that dealt with temporal adverbs like temprano, early. In Spanish, the temporal adverb and the subject can not both be to the left of t he verb. BP does allow both a temporal adverb and the subject to appear left of the verb, as in (24) b (Pires 2007). (24) a. *Temprano Julia sa lia de casa Spanish early Julia leave PREP house Early, Julia left from home b. Hoje cedo a Jlia saiu de carro BP Today early DET Julia left.3SG PREP car Early today, Julia left by car 64

PAGE 65

This suggests that the temporal adver b and the subject do not compete for the same structural position, whic h is compatible with the subject being in an A position. 3.2.2.2 Weak Pronouns Weak Pronouns also point to an A position for the subject. Phonologically reduced pronouns are referred to as weak pronouns (C ardinaletti & Starke 1999). Weak pronouns are subject to syntactic restrictions whereas full pronouns are not, and this is the case for BP. An example of a phonologically reduced pronoun is c which cannot occur in isolation, (25) (25) A. Quem fez o trabalho? Who did.3S DET work Who did the work B. *C / Voc 2S You This pronoun cannot occur in the left periphery as (26) and (28) show. The full form voc can occur in any position, seen in (27) This phonologically reduced pronoun can occur in the subject posit ion as a doubled subject, example (27) (Kato 1999:28 for reference to phonologically-reduced pronouns in spec TP, Pires (2007:132) for the examples below). (26) *C / Voc, eu vi 2S 1S saw.1S you, I saw (27) Voc, o seu pai c pode convidar 2S, DET your father 2S AUX invite.INF As for you, your father, you can invite (28) *C, o seu pai c pode convidar 2S, DET your father 2S AUX invite.INF 65

PAGE 66

The phonologically reduced pronouns do not have to occur in conjunction with some full pronoun or NP in the left periphery, as example (29) shows. Here the pronoun is the subject and appears alone. (29) c pode convidar ele 2S CAN.2S invite.INF 3S You can invite him This evidence suggests that a phonologically reduced subject pronoun is in an A position. While this does not guarantee that full pronouns and other NPs are also in an A position, it confirms that an A-subject position exists and this position is filled at least regularly by phonologically-reduced pronouns. 3.2.2.3 Quantifiers Much like the phonologically-reduced pr onouns, bare quantifiers cannot be leftdislocated in BP. In fact, it is well known that non-referential expressions like quantifiers and n-words cannot be left-dislocated in many languages. Take the examples from French. French allows doubled subjects mu ch like Portuguese. The examples here show that the doubled subjects quelquun and personne cannot be in the left periphery, (30) and (32) but they can be in the subject position, (31) and (33) (30) *Quelquun il vient. someone he comes (31) Quelquun vient. someone comes Someone is coming. (32) *Personne il na rien dit. no-one he not-has nothing said 66

PAGE 67

(33) Personne na rien dit. no-one not-has nothing said No one said anything. The same can be said of quantifiers in BP (Pires 2007:126). They are grammatical in subject position, (34) and (36) but not fronted (35) and (37) : 9 (34) Algum pintou esse quadro Someone painted.3S that painting Someone painted that painting (35) ??Algum, o Joo no ajudou Someone, DET John NEG helped.3S Someone, John didnt help. (36) Ningum viu o Joo. Nobody saw. 3S DET John Nobody saw John (37) *Ningum, o Joo (no) viu. Nobody, DET John (NEG) saw.3S Nobody, John saw These examples verify that while an A'position in the left periphery is not open for quantifiers, the subject position is. This dat a argues that BP subjects are not in an Aposition, but in an A-position. Related to quantified expressions is A nagnostopoulou & Alexiadou (1998)s claim that, in Greek, preverba l subject QPs have unambiguous scope, and postverbal QPs have ambiguous scope. This was presented in examples (12) for a preverbal subject, repeated in (38) and (13) for a postverbal subject. Diffe rent from Greek, BP maintains a scopal ambiguity when the subject is prev erbal, which supports the claim that the 9 Belletti claims that Italian indefinite quantifiers need to receive emphatic stress to be grammatical in subject position; however, this has been shown to not be the case for BP (Costa and Galves 2002). 67

PAGE 68

subject is in an A-position because it is be lieved that an A-positi on allows for scopal ambiguity. (38) Kapios fititis st ihiothetise kathe arthro Some student filed every article Some student filed every article some>every *For every article, there is some student that filed it *every>some (39) Alguns estudantes le ram todos os artigos Some students read.3P all DET articles Some students read all the articles some>all For all the articles, there is some student that read them all>some This section has demonstrated that the pr everbal subject in BP is an A-position. The evidence for this included the possibilit y of weak pronouns and quantifiers, neither of which can appear in A'-positions. 3.3 Verbs and Verb Movement This section addresses more precisely the pos ition of the verb in BP. A discussion about verb movement in BP must include a di scussion about the internal structure of IP which is subject to debate. This secti on discusses that debate and locates BP within that context. For Romance languages, t he debate is over Pollock (1989)s split IP. Specifically, Pollock claims that IP is broken up into different functional heads such as AgrP and TP, where AgrP is higher in the der ivation than TP. With regards to the ideas for BP, researchers argue two proposals: i) BPs IP is not split into T(ense)P and Agr( eement)P, there is a single TP projection, the verb moves to T and the subject moves to spec,TP (Kato 1999, Rodrigues 2004); 68

PAGE 69

ii) IP is split into TP and AgrP, with AgrP above TP, and the verb moves to T and the subject moves to spec,A grP (Costa & Galves 2002, Silva 2004). I follow (ii) and conclude that the subject in BP is in spec,AgrP. To do this, I explain the debate and its consequences for understanding BP. Section two of this discussion looks at the evidence for a Split IP First, I discuss floating quantifiers as a diagnostic for a split IP. Then I discuss so me adverbs and their relationship to the verbs movement. 3.3.1 IP Pollock (1989)s argument for split IP was based on word order differences between English and French. Pollock argues that the Englis h negative marker not and the French marker pas indicate that the verbs for each language are in a different positions in finite sentences. In Frenc h, the verb moves above the negative marker, whereas in English the verb does not move. Instead, English uses an auxiliary verb do to occupy the position occupied by aime in French. (40) Jean naime pas Marie John NEG-loves NEG Mary John doesnt love Mary (41) *John likes not Mary (42) John does not like Mary (43) *Jean ne pas aime Marie John NEG NEG loves Mary John doesnt love Mary Pollock explains using adverbs that t he verbs in each language are in different structural positions (examples from Belletti 2004). Examples (44) (45) and (46) show 69

PAGE 70

that some adverbs and the negative marker are to the right of finite verbs and to the left of non-finite verbs. Example (47) confirms that the adverb can be to right of the nonfinite verb, but the negative marker cannot, as in (48) Pollock claims French verbs can occupy at least two positions in IP: a higher position for finite verbs; and a lower position for non-finite verbs. (44) Jean recontre souvent Marie John meets often Mary (45) Jean essaye de souvent recontrer Marie John tries often to meet Mary (46) Jean essaye de ne pas recontrer Marie John tries not to meet Mary (47) Jean essaye de recontrer souvent Marie John tries to meet often Mary (48) *Jean essaye de ne recontrer pas Marie John tries to meet not Mary. The contrasts with English are clear. T he English examples show that English verbs do not move to either of the positions suggested for Frenc h. Notice that in cases (50) (52) and (54) the adverb or the negative ma rker must precede the verb. (49) John often meets Mary (50) *John meets often Mary (51) John tries to often meet Mary (52) *John tries to meet often Mary (53) John tries not to meet Mary (54) *John tries to meet not Mary. Based on these and other pieces of data, Pollock concludes: 70

PAGE 71

(55) English: lexical verbs do not move to either inflectional position, these positions are filled by modal and auxiliary verbs (56) French: lexical verbs must move to the hi gher inflectional positio n in finite clauses, and nonfinite verbs optionally move to the lower inflectional position only. Thus, in French finite clauses it is cl aimed that the verb undergoes long movement to the highest projection in IP, and in non-fi nite clauses, the verb moves to a lower projection or doesnt move at all. For example, contrasting example (40) with (57) in the first case the verb moves past the negative marker, and in the second the verb does not move past the negative marker. However, again contrasting (47) and (57) the first allows the non-finite verb appear to the left of the adverb, but not the negative marker. This movement above the adverb but below the negative marker c ould be considered short-movement. (57) Ne pas sembler heureux est une condition pour crire des romans. NEG NEG seem.INF happy is DET condition PREP write.INF DET novels Not to seem happy is a prerequisite for writing novels.' Generally for Romance, the two main projections posited as making up IP are TenseP and Agr(eement)P. This is based on the make-up of Romance verbs because they, in most cases, over tly exhibit these two functi onal morphemes. The Italian example and the Spanish show the morphemic breakdown of the verb, separating tense and person/number agreement. (58) Canta-va-no Italian Root-Tense-Person Sing-IMP-3P They were singing 71

PAGE 72

(59) Llega-ba-n Spanish Root-Tense-Person Arrive-IMP-3P They were arriving While there is evidence for split IP in French, some authors do not believe that the split necessarily exists in all languages (Iatridou 1990, Thrinsson 1996, Bobaljik and Thrinsson 1998). Iatridou suggests that while some languages may have TP and AgrP, there needs to be evidence in each la nguage for such a split (Iatridou 1990:553). Thrinsson agrees by positing The Real Minimalist Principle : Assume only those functional categories that you have evidenc e for (1996:261). Because of objections raised by these linguists and others that have come about since Pollocks paper, linguists debate about the ma ke up of IP for BP. 3.3.2 Non-Split IP One of the goals of Chomsky (1995) is to eliminate AgrP as a functional clausal head as it has no interpretation, and only things that receive interpretation can be clausal heads. Following Chomsky (1995), Kato (2000) argues that AgrP is not part of BP and the subject is realized in spec,TP. Her argument is based on the changes in verbal morphology that have been observed in the last one hundred years. Duarte (1995), among many others, shows that BP verbal morphology has weakened to the point where it is changing BP from a null subject language to a non-null subject language. Since this change, a more rigid order of SVO is observed, as seen in section two. The next chart, table 3-1 shows t he weakening of agreement morphology in BP over time where paradigm one indicates the colonial period (1500-1820s), paradigm two is the post-colonial period (1820s-1900s), and finally paradigm three represents the 72

PAGE 73

present day (Duarte 2000:19). BP goes fr om having six agreement conjugations to three. Notice that -----means the form is no longer in use. Table 3-1 Loss of Verbal Morphology in BP Person/Number Pronouns Paradigm 1 Paradigm 2 Paradigm 3 1 st singular 2 nd singular 3 rd singular 1 st plural 2 nd plural 3 rd plural Eu Tu Voc Ela/ele Ns A gente Vs Vocs Eles/Elas am o ama s ama ama ama mos ama is ama m ama m am o -----ama ama ama mos ama -----ama m ama m am o -----ama ama -----ama -----ama m ama m Since the weakening began, weak pronouns have begun to develop (Galves 1998, Kato 1999). According to Kato, both of these factor s lead to the reinterpretation of Agr as a lexical item having lost syn tactic independence, i.e. it is no longer an independent projection like TP (Kato 2000). Thus, IP is not split along the AgrP TP distinction. Additionally, as some have noted Agr is relational, not functi onal, and therefore does not head its own projection. Chomsky (1995: 240, 349) and others argue that Agr is fundamentally different from such functional categories as Tense and Negation. He argues that, unlike T, C, and D, which have [+interpretable] features, Agr has no semantic properties but consists of [interpre table] formal features only. Moreover, Agr exists only when it has strong features. Agr is nothing more than an indication of a position that must be occupied at once by overt operations (Chomsky 1995: 351). 73

PAGE 74

3.3.3 AgrP in BP While some argue in favor of an unsplit IP based on empirical evidence, I assume that despite significant change in the verbal morphology, the chan ge did not cause AgrP to be eliminated from the grammar. Here I follow Galves (1998) and Costa & Galves (2002) who propose that ther e is still a projection betwe en CP and TP necessary to account for cross-linguistic differences. In fact, BP needs a projection like AgrP to understand certain facts about th e position of the verb, the subject, floating quantifiers, and adverbs. The BP facts parallel the facts in French non-finite clauses, which showed short verb movement (Costa and Galves 2002) This contrasts with languages such as French finite clauses, which have long verb movement, or English, which has no overt verb movement. A diagnostic tool used to determine the stru cture of IP is the presence of floating quantifiers (FQ) such as BP todo all (see Vicente 2006 for a complete discussion of FQs in BP). I follow Vicent e (2006) and Muller et al. (2007) who convincingly argue for FQ as a diagnostic tool to understand BP stru cture. Sportiche (1988) proposes that FQs initially form a constituent with the DP they modify. When the DP moves, the FQ can be stranded in positions that the DP mo ves through. The stranding results when a subconstituent of the DP move s further. FQs thus mark positions through which a DP has moved. Sportiches analysis is based on a series of conclusions from the 80s (Bobaljik 2003:6): 74

PAGE 75

(60) Floating Quantifiers: (i) FQs appeared to modify DPs in the same way as DP-initial Qs; (ii) FQs in some languages display de terminer-like agreement with the DP they modify; (iii) FQs surface in the left periphery of (certain) maximal projections, especially VP; (iv) the relationship between an FQ and the DP it modifies obeys an anaphor-like locality condition. Assuming the Predicate-Internal Subjec t Hypothesis (Koopman and Sportiche 1991), then a quantifier should be allowed to be stranded in positions below the surface position of the subject. This is what we see in the BP examples below from Brito (2001:78). The lowest position in which the FQ can appear is to the right of the main verb, in (63) The FQ is stranded in the subjects base position, spec, vP. Both the subject and verb start inside vP but move out above spec, vP, as I explain below. (64) is ungrammatical with todas to the right of the object. (61) Todas as mulhers comeram a lagosta All DET women eat.PST.3P DET lobster All of the women ate the lobster with their hands (62) As mulhers todas comeram a lagosta (63) As mulhers comeram todas a lagosta (64) *As mulhers comeram a lagosta todas The crucial example in s upport of a split IP is (62) where the verb and the subject are separated by the FQ. As Costa & Galves (2002) suggest, if IP is split into at least AgrP and TP, then the grammaticality of (62) is predicted. I argue that the verb moves to T and the subject moves to the spec,TP. After that, the subject moves up to spec,AgrP, and the FQ is stranded in spec,TP, with the verb remaining in T, hence the separation of the verb from the subject. 75

PAGE 76

(65) AgrP DP Agr As Mulhersi Agr TP [ todas ti]j T T V P comeramii DP V tj V VP tii This example contrasts with (63) where the FQ is to the right of the verb. This example and the tree in (66) show that if the subject is generated in vP, then for the FQ to be to the right of the verb, the verb must move outside of vP. In this case, I have argued that it does move, and it moves to T. (66) AgrP DP Agr As Mulhersi Agr TP ti T T V P comeramii DP V todas ti V VP tii 76

PAGE 77

These facts about FQs support a split IP in BP. If the floating q uantifier is between the subject and the verb and both are outside vP, then there must be a position between the subject and the verb where a DP c an land. I claim that this position is spec,TP in a split IP scenario. One could argue that the FQ is merely in spec,vP, the original position of the subject, and the subject is in the specifier of a non-split IP; however, this analysis runs into problems with (63) (63) provides evidence that the ve rb moves to a position higher than the FQ, which is no lower than spec,vP, meaning that the verb moves out of vP. For there to be a specifier position for the FQ and a head position for the verb above vP but below the surface position of the subject requires another projection, as provided in a split IP scenario. Adverbs further support the claim that the ve rb moves to T, necessitating a split IP. Galves (1994) and Costa & Galves (2002) show that there are at least two adverb positions in a simple BP claus e, preverbal and postverbal: (67) a. O Joo beija frequentemente a Maria. DET Joo kiss.3S frequently DET Maria Joo kisses often Maria b. O Joo frequentemente beija a Maria. DET Joo frequently kiss.3S DET Maria Joo often kisses Maria This pattern is different from what wa s observed for French and English in section 3.3.1 Recall that the Englis h adverb must be to the le ft of the verb and the French adverb to the right. (68) Jean rencontre souvent Marie John meets often Mary John often meets Mary 77

PAGE 78

(69) *Jean souvent rencontre Marie John often meets Mary John often meets Mary (70) John often meets Mary (71) *John meets often Mary Costa (2004) argued that this difference bet ween Portuguese and Frenc h and English is due to short verb movement in Portuguese. Ve rb movement according to Costa is short in this case because IP is split into more than one functional projection. Portuguese verbs move to the lower head position in a split IP, hence short ve rb movement. His analysis claims that the subject is in spec,A grP, and the verb moves only to T, which is below AgrP. The difference between sentences (a) and (b) in (67) is that the adverb adjoins in two different places. In the ca se of sentence (a) the adverb is in a low position, and for (b) the adverb is in a high position. The structure below represents the syntactic positions referred to as high and low. (72) AgrP DP Agr O Pedroi Agr TP High AdvP TP T vP Low AdvP vP Costa & Galves (2002) use high and low adverb s to show that the ve rb is in T. In Portuguese a low adverb is an adverb that surf aces only postverbally. Some examples 78

PAGE 79

of these are bem well and atentamente carefully. As they can only appear postverbally, Costa & Galves claim that t he verb must move above these low adverbs, which is the case in (73) Notice that (74) 10 is ungrammatical with the verb to the right of the low adverb. Thus, the verb must move past these low adverbs in sentences like (73) Sentence (73) is represented in (75) where the low adverb adjoins to vP. (73) O Pedro leu bem/atentamente o livro DET Peter read.3S well/carefully DET book Peter read well/carefully the book (74) *O Pedro bem/atentamente leu o livro (75) AgrP DP Agr O Pedroi Agr TP ti T T vP leuii AdvP vP bem ti tii o livro Next, Costa & Galves show that the verb can appear between two adverbs. The case of (76) has the low adverb bem well and what they call a high adverb, ontem yesterday. Different from the low adverb, the high adverb adjoins to TP seen in (77) causing the verb to be to its right. 10 In the case of example (74), the sentence would be grammatical if there was a pause. Costa & Galves are assuming no pause. 79

PAGE 80

(76) Os meninos ontem leram bem o livro DET children yesterday read.PST.3P well DET book Yesterday the children read well the book (77) AgrP DP Agr O Pedroi Agr TP AdvP TP ontem DP T ti T vP leuii AdvP vP bem ti tii o livro Finally, Costa & Galves show that certai n adverbs can adjoin either high or low and, depending on their position of adjunction, t hey can yield a subject-oriented reading or a manner reading (Jackendoff 1972). A s ubject-oriented meaning refers to the subject having the quality that the adverbs suggest. A manne r-oriented meaning is one where the adverb is more closely linked to the meaning of the action that the verb conveys. A subject oriented reading corres ponds to a high adverb and TP adjunction. The adverb inteligentemente in (78) adjoins to TP. (78) O Pedro inteligentemente leu o livro (Subject-oriented/*manner) DET Peter intelligently read.3SG DET book It is clever of P edro to read the book 80

PAGE 81

(79) AgrP DP Agr O Pedroi Agr TP AdvP TP inteligentemente DP T ti T vP leu ... A manner-oriented reading results from a low adverb that is adjoined to vP. Sentence (80) has the same adverb inteligentemente but in this case it adjoins to vP. (80) O Pedro leu inteligentemente o livro (*Subject-oriented/Manner) DET Peter read.3SG intelligently DET book Pedro reads the book in a clever manner (81) AgrP DP Agr O Pedroi Agr TP ti T T vP leu AdvP vP inteligentemente Given these data, Costa & Galves (2002) cl aim that the verb must move outside the vP in BP. Not only must it move outside the vP, but it must move to a position that is not high in the IP domain. To show that the mo vement must be to a l anding site low in the 81

PAGE 82

IP domain, they contrast the Portuguese data with French. In French, the verb always precedes the adverb regardless of the adverbs interpretation, as in (82) This is possible because in French the verb must mo ve all the way to Agr. The ambiguity between the readings can arise because the verb is obligatorily in a higher projection above both the TP-adjoined and vP-adjoined positions. (82) Pierre lit intelligemment le livre (Subject-oriented/Manner) Peter read.3S intelligently DET book Peter intelligently reads the book. (83) AgrP DP Agr Pierrei Agr TP Subject-Oriented litii Adv TP intelligemment ti T T vP Manner AdvP vP Intelligemment If one assumes that the adjunction pos itions of the adverb is the same for French as it is in Portuguese, i.e. TP for subject-oriented and vP for manner, then the cross-linguistic differences follow from differences in verb movement in French versus BP. The picture is further supported by the in teraction of adverbs like intelligently with negation, which is below TP and above vP. Here the negative marker pas distinguishes the two adjunction sites. (84) has the adverb adjoined to TP and negation 82

PAGE 83

follows the adverb. In (86) the adverb is adjoined to the vP and negation precedes the adverb. Subject-oriented /*manner (84) Pierre ne lit intelligemment pas le livre Pierre NEG reads intelligently NEG DET book Pierre doesnt read intelligently the book (85) AgrP DP Agr Pierrei Agr TP ne litii AdvP TP intelligemment ti T T NegP tii Neg vP pas *Subject-oriented/ Manner (86) Pierre ne lit pas intelligemment le livre Pierre NEG reads NEG intelligently DET book Pierre doesnt read intelligently the book (87) AgrP DP Agr Pierrei Agr TP ne litii ti T T NegP tii Neg vP pas AdvP vP intelligemment 83

PAGE 84

Given these facts and those from the prev ious two sections, I have argued that IP is split in BP, following Pollock (1989), into at least AgrP and TP. The subject moves to the specifier of the hi gher projection, spec,AgrP, and the verb exhibits short movement to T, the head of the lower pr ojection. Adverb placement data supported this picture. 3.4 Fronting in BP Fronting in Portuguese differs from other languages in the Romance family and languages such as English. I show that BP uses base-generated topics. This can be contrasted with English, whic h uses movement topicalization, and Spanish, which uses Clitic-Left Dislocation (CLLD). Base-generat ed topics are merged into the derivation in the higher topic position di scussed in chapter two. Below is an English sentence where a DP is fronted and leaves behind a trace.. I call such a derivation topicalization. (88) This booki, I like ti CLLD occurs when a DP is fronted and subsequently a pronominal clitic marker appears. This is seen in Spanish (89) where este libro is fronted and a clitic lo appears. Notice that if the sentence does not have a clitic, it is ungrammatical, as in (90) This shows that Spanish does not allow topicalization as English does, where only a null trace is left behind. (89) Este libro, yo lo le Spanish this book, 1S CL read.1S This book I read it (90) *Este libro, yo le this book, 1S read.1S This book, I read 84

PAGE 85

It has been claimed that fronting in BP involves left dislocation. Diffe rent from Spanish, a clitic does not appear but a resumptive pronoun does, (91) (91) Este livro, eu li ele BP this book, 1S read.1S 3S This book I read it However, in BP the resumptive pronoun is not necessary, seen in example (92) and it is here that Portuguese differs from other Romance languages. Cases like (92) suggest that BP has topicaliz ation much like English. (92) Este livro, eu li this book 1S read.1S This book, I read BP topicalization not only occurs with DPs but also with VPs, which is another area that differentiates Portuguese from the many ot her Romances languages. When a V/VP is topicalized, it is in the infinitive form as in (93) The topicalized VP can also contain complements of the verb. (93) Dormir ele dorme sleep.INF, 3S sleep.3S As for sleeping, he sleeps (94) Cantar fado ele canta sing.INF fado 3S sings.3S As for singing fado, he sings I only treat fronting of DPs and di spense with discussing VP fronting 11 because both cases are allowed with CFN sentences. 11 For a complete analysis of VP fronting in BP, see Bastos (2001, 2009). 85

PAGE 86

Two analyses for BP fronting are: movem ent or base-generation. Both English Topicalization and CLLD make use of a movement solution. I briefly discuss the movement analyses, and from there I present the base-gener ated analysis. The facts about base-generated topics become useful in understanding my analysis of clausefinal negation presented in chapter six. Kato (2003) and Kato & Raposo (2007) ( K&R) argue for a traditional movement analysis for fronting constructions in BP. K&R propose that DP fronting in BP is really CLLD, much like Spanish, but in this case the clit ic is null. They claim this in order to fill out the apparently defective clitic paradigm. Th ird person clitics exist but are null. Their analysis originates with Uriagerekas (1995) work on Spanish Clitic Doubling. He proposes that the clitic and the associated NP start in a spec-head relation in DP. From there, the NP moves to a position hi gher in the derivati on, and the clitic is stranded somewhere along the way. The deriv ation for a Spanish sentence like (95) is given in (96) (95) A Juan, le doy un libro PREP John CL.3S give.1S DET book I give the book to John (96) a. doy un libro [Juan le] b. [Juan le] i doy un libro ti c. Juanj [tj le] i doy un libro ti {merge of a} d. a Juanj [tj le]i doy un libro ti K & R claim that BP is different from Spanish in that the DP moves to the left but the clitic stays in-situ. It is stranded in the DPs base position, not in a preverbal position. (97) O Joo, eu vi ___ ontem DET John 1S saw CL yesterday John, I saw (him) yesterday 86

PAGE 87

(98) a. eu vi [Joo ] ontem b. [Joo]i eu vi [DP ti ] Kato & Raposo (2007) proposes that su ch dislocation that leaves behind a resumptive pronominal clitic is not subject to island constraints (Ross 1967), thus the BP topic construction is insensitive to islands. (99) shows a topic that appears to have moved from inside a complex subject (Kato & Raposo 2007:208). (99) Esse bolo1 o rapaz [que trouxe __ agora mesm o da pastelaria] era teu afilhado this cake, DET boy [COMP brought CL now EMPH PREP.DET bakery] was your godson this cake, the boy who brought (it) just now from the bakery was your godson. Although K & R claim that the resumptive clitic rescues the structure from an island violation, Cinque (1990) points out that pronominal clitics do not save sentences from island effects. Furthermo re, it is not clear what would be meant by a null clitic. Traditionally, clitics are phonol ogically dependent, and it is difficult to understand how a null clitic could exist because it has no phonology. Also, it is never quite clear why in the case of Spanish the clitic moves with the NP, but in Portuguese the clitic stays in-situ. For these reasons, I move away from a movement analysis of BP fronting and propose that BP fronting is an instance of a base-generated topic in spec,TopP, with a resumptive pronoun, sometimes null, generated in the argument position. One reason to think that a movement analysis is incorrect for BP comes from the same example that K & R used in (99) The DP moves out of an island but does not violate Rosss island constraints. This fact follows if fronted DPs are base-generat ed. Additional evidence for this analysis comes from the absence of third person clitic pronouns for both the accusative or dative case noted by K&R. Recall from section 3.2 that BP has 87

PAGE 88

developed weak pronouns in t he sense of Cardinaletti & Starke (1999) (Galves 1998). The examples for second person voc, c, were given above. In the case of third person, both plural and si ngular, weak pronouns exis t but have no phonological realization (Galves 1989, Schwenter & Silv a 2002, Kato 2003). Evidence for the null third person object pronouns comes not only from topic structures, but many other types of sentences. For example, the follo wing sentences are an exchange between two people. Notice that in (101) there is no overt object nor is there a topicalized DP. In this case, a null pronoun is in the object position. Although, John is the topic of the exchange, there is no DP in sentence in t he topic position, and the sentence is still grammatical because of t he null pronoun (Galves 1989). (100) Joo vem? John comes Is John coming (101) Ningum convidou pro n-body invited pro Nobody invited him The behavior of wh-movement in BP also supports a base-generation analysis of fronting. In BP, wh-movement is not allowed out of island, di fferent from what has been just seen for BP topics. Compare (102) and (103) In the first case, a null pronoun is generated in the argument position and the t opic is generated in the left periphery. Because there is no movement, there is no is land violation. The second case has whmovement, leaving a trace, which results in an island violation. (102) Estes CDs, vocs encontraram uma loja que vende pro These CDs, 2P find.pst.2P DET store that sell.3S pro These CDs, yall found a st ore that sells them 88

PAGE 89

(103) *Que CD1 vocs encontraram uma loja que vende t1? Which CD 2P find.pst.2P DET store that sell.3S Which CD have you found a shop that sells it? The evidence given suggests that topics in BP are based-generated and a resumptive pronominal element, either null or no t, is in the argument position. 3.5 Wh-Questions: Fronted and In-Situ BP makes use of both fronted and in-situ wh -questions. In this section, I discuss the derivations for both types of questions. There have been several different opinions through the more recent years of generative grammar as to the nature of wh-questions in BP. Kato & Raposo (1996) claim t hat all fronted wh-questions are clefted expressions. Kato & Mi oto (2005), Pires & Taylor ( 2007), Zocca (2007) argue for understanding them as featur e agreement (although the mec hanics of each of these differ), and in this work, I follow the claims m ade by Pires & Taylor and Kato & Mioto. I first give a brief outline of the different types of wh-questions in BP. After that, I explain the derivation for both move ment and in-situ questions. 3.5.1 Types and Conditions for Wh-Words: Movement and In-Situ Not only does BP have fronted and in-situ wh-questions, the fronted questions come in four versions. The examples below illustrate both fronted and in-situ questions. The first set has fronted wh-words, (104) (107) and the question is what did you eat?. The second set is a BP sentence with a wh-in-situ, (108) which is you ate what?. The first version is a fronted wh-question with no complementizer or copula. The second version has the complementizer que following the wh-phrase The third version has the wh-word fronted with a clefted construction involving que 'is that'. The fourth version combines the second and third. Examples (106) and (107) both appear to be clefted 89

PAGE 90

constructions, and I do not treat them here (s ee Kato & Mioto 2005a,b for an analysis of these constructions). (104) O que voc comeu? DET what 2S ate.2S (105) O que que voc comeu? DET what COMP 2S ate. 2S (106) O que que voc comeu? DET what is COMP 2S ate. 2S (107) O que que que voc comeu? DET What COMP is COMP 2S ate. 2S What did you eat? (108) Voc comeu o que? 2S ate. 2S DET what You ate what? Some additional facts are important to note. First, the interrogative complementizer que is optional with fronted wh-words, (109) and (110) but impossible with wh-in-situ, (111) and (112) (Hornstein et al. 2005). (109) Como voc fez isso? how 2S do.PST that How did you do that? (110) Voc fez isso como? 2S do.PST that how How did you do that? (111) Como que voc fez isso? how that 2S do.PST that How did you do that? 90

PAGE 91

(112) Que voc fez isso como? that 2S do.PST that how How did you do that? Second, it has been noted that whether the complementizer is null or overt, whmovement within embedded inte rrogative clauses is obligatory (Zocca 2007). The examples below illustrate this with the interrogative verb perguntar to ask. This verb forces wh-movement in its complement, which is why (114) is ungrammatical, as the wh-phrase stays in-situ. (113) Eu perguntei como (que) voc fez isso. 1S ask.PST.1S how (that) 2S did that I asked how you did that. (114) *Eu perguntei voc fez isso como. 1S ask.PST.1S 2S did that how *I asked you did that how. Third, BP wh-movement is sensitive to islands while wh-in-situ is not. Thus, sentence (117) where the wh-phrase moved out of an island is ungrammatical. However, if the wh-word does not move, t hen it is licit even in island contexts. (118) has que livro which book in a complex NP island and is still grammatical. (115) Que livro voc disse que ela comprou? which book 2S said that 3SF bought Which book did you say that she bought? (116) Voc disse que ela comprou que livro? 2S said that 3SF bought which book Which book did you say that she bought? (117) *Que livro voc conversou com o autor [que escreveu]? which book 2S talked PREP DET author that wrote Which book did you converse with the author that wrote? 91

PAGE 92

(118) Voc conversou com o autor [que escreveu que livro]? 2S talked PREP DET author that wrot e which book Which book did you converse with the author that wrote? Finally, unlike other Romance languages, t here is no subject-verb inversion in main or embedded clauses with wh-movement. This type of inversion is seen in European Portuguese; however, in BP the meani ng changes completely, as seen in sentence (120) (119) O que comeu a Lcia? European Portuguese What comeu. 3S DET Lcia What did Lcia eat? (120) O que comeu a Lcia BP what ate.3S DET Lcia What ate Lcia? *What did Lcia eat? 3.5.2 Deriving Wh-Questions in BP In this section I outline an approach to wh -words, both fronted and in-situ. The main conclusions for both types of wh-questions are as follows. For fronted wh-questions, I claim that: i) an element in Foc with a strong, uninterpretable wh-featur e [uWH*] triggers movement; ii) the wh-word with a [iWH] f eature must move to a position where it satisfies the strong features of t he element in Foc. For Wh-in-situ in BP: i) Wh-in-situ in BP has a null Q element inserted into TopP; ii) the wh-word has an [iWH] feature; iii) Q unselectively binds with the wh-word; iv) the null Q in TopP has some discourse features indicating that the question is part of the Common Ground. 92

PAGE 93

Point iv) for the in-situ question requires s pecial attention. I discuss the notion of Common Ground (Stalnaker 1978), which is used by Pires & Taylor (2007) to understand the seeming opt ionality of wh-in-situ. I follo w their proposal that the null Q carries a discourse feature that allows wh -in-situ, and if the null Q is present, then movement is not allowed. First, to explain fronted wh-expressions in BP, I use feature checking and Agree outlined in chapter two. I begin here with an overt focus head que Example (105) repeated here with a derivation in (122) contains this head, which has an [uWH*] which triggers movement of the wh-word with an [iWH] feature to spec,Foc(us)P. (121) O que [iWH] que [uWH*] voc comeu? DET what COMP 2S ate.2S What did you eat? (122) FocP DP Foc O que[iWH] k Foc AgrP que [uWH*] DP Agr voci Agr TP ti T T vP comeu ... t k 93

PAGE 94

This overt head must check its strong feature, and therefore, if there is no movement the derivation crashes: (123) *que voc comeu o que? COMP 2S ate.2S DET what You ate what? Like example 0 there are cases where there is no overt complementizer and yet the wh-phrase is fronted. In t hese cases, the overt Focus h ead is deleted after movement of the wh-phrase (Kato & Mioto 2005); how ever, the same mechanisms motivate fronting, i.e. the [uWH*] feature of t he deleted head. (124) O que qu e voc comeu? DET what comp 2S ate.2S What did you eat? In addition to the Focus head that forces movement, I claim that BP has a null head with [uWH] which is not strong (see Pires & Taylor 2007, Zocca 2007 for a similar account for BP). This is similar to what was proposed for French, Chinese and other wh-in-situ languages (Bokovi 1999, Aoun & Li 2000). The null head does not require movement, evidenced by the fact that wh -in-situ interrogatives can evade island violations. What differentiates the null head that does not ca use movement and the overt que which does is their discourse contri butions. The null head associated with wh-in-situ, as mentioned, is related to Common Ground and discourse-old status. Pires & Taylor (P & T) (2007) proposes that single wh-questi ons can have the whphrase in-situ, provided that semantic and pragmatic requirements are met. If these semantic and pragmatic require ments are not met, then, the wh-expression must be 94

PAGE 95

fronted. P & Ts analys is depends upon Stalnakers (1978, 2002) Common Ground, which they define as information that was pr eviously given in the discourse or in the extralinguistic context and which is shared (or assumed by the speaker to be shared) by speaker and hearer (P & T 2007:5). Fo r BP, wh-in-situ is possible if the information that is being requested is expec ted by the speaker to be part of the Common Ground (2007:6). When talking about the Common Ground, Stalnaker states that the presuppositions of a speaker are the propositions whose truth he takes for granted as part of the backgroun d of the conversation (1978: 149). This means that for in-situ questions in BP there is an answer to the question. Thus, in the example below, you know who in So Paulo is similar to you know someone in So Paulo as there exists a person that you know in So Paulo. In both cases the answer is non-null, and the speaker presupposes the existence of a person that the addressee knows in So Paulo. P & L give the following interaction as an example. The infelicity of this example stems from the fact the information being requested (i.e. knowing someone in So Paulo) is not part of the Common Ground, having not been over tly introduced into the discourse. [ You approach a colleague at work and ask, out of the blue :] (125) #Voc conhece quem em So Paulo? 2S know.2S who PREP So Paulo You know who in So Paulo This same interaction can be made felicitou s in two ways, by fronting the wh-word or by giving a discourse context shar ed by both the speaker and the hearer. Importantly, the same sentenc e with a fronted wh-word is grammatical and felicitous, 95

PAGE 96

although my consultants all commented that this would mostly likely result in a question such as: Are you going to So Paulo? [ You approach a colleague at work and ask, out of the blue :] (126) Quem voc conhece em So Paulo? who 2S know.2S PREP So Paulo You know who in So Paulo The following exchange shows how the disc ourse requirement af fects the felicity of the sentence. In this instance, the Common Ground is contains information that allows the use of in-situ wh-words; namely, it is part of the Co mmon Ground that you visited So Paulo, and the information r equested by the speaker is presupposed to already exist and be implied in the discourse. [ You approach a colleague at work and comment :] (127) A. Eu visitei S o Paulo este fim de semana 1S visted.1S So Paulo this end PREP week I visited So Paulo this weekend (128) B. Voc conhece quem em So Paulo? 2S know.2S who PREP So Paulo You know who in So Paulo? The analysis that I assign to such examples is given in (129) The connection to the Common Ground is represented by a null ques tion operator Q in the Top head which encapsulates this restriction. 12 Although I do not develop this connection, the givenness requirement on parts of the wh-question ar e somehow related to topicality. This Q operator binds the in-situ wh-phrase. Since no strong featur e is present, the wh-phrase does not move. 12 In chapter six, I will revise this analysis somewhat and claim that the Q operator is actually located in the specifier of a lower topic phrase, Top2P. 96

PAGE 97

(129) TopP Top AgrP Q DP Agr voci Agr TP ti T T vP conhece quem[iWH] em So Paulo Que cannot occur in this structure with wh-in-situ, (112) because that head has a strong feature which would force movement. I account for the fact that wh-in-situ is impossible in embedded clauses by simply stipulating that the Q Top head is unavailable in selected contexts. To summarize, BP has two interrogative heads in the left periphery: one overt one in Foc that forces movement for feature c hecking and a second that is covert, related to the discourse structure, which does not allow movement. 3.6 Conclusions This chapter has addressed the clause stru cture for BP. This chapter focused principally on the CP and IP domains. T he topics addressed here were presented to understand five sub-questions rela ted to BP clause structure. i. What is the subject position? ii. What is the position of the finite verb? Is there any verb movement? iii. What is/are the structur al position(s) of topics? iv. What is the structure of wh-movement? v. What is the structure of wh-in-situ 97

PAGE 98

I have argued that BP employs a split IP (Pollock 1989). I claimed that the subject is the specifier of AgrP. I showed that thes e elements cannot be in an A'-position in the left periphery, which has been suggested by some to be the location where preverbal subject resides in null-subject languages. Se ction three shows that the verb starts inside VP and moves up to T. I argued for s hort movement of the verb to T (Costa & Galves 2002). Floating quantifiers as well as adverbs showed that this is the case. Section four discussed BPs system of topicaliz ation. I proposed that topic structures in BP are base-generated with the topic phrase in spec,TopP binding a resumptive pronoun, which may be null in the third person. Finally, I gave an overview of wh-words and questions in BP. The distinction was made between fronted and in-situ questions. In the case of fronted wh-wor ds, the fronted word moves to the specifier of FocP, which is headed by que with a strong [uWH*] feature. When que is not present in whquestions with a fronted wh-phras e, it has been deleted. In wh-in-situ questions, this head does not appear. Instead, ther e is a null operator in T op which binds the in-situ wh-phrase. This chapter has been developed to show the basic clause structure of Brazilian Portuguese. Below is a tree structure that give s a basic map of the st ructural position of the phrases discussed in this chapter. No tice here again that there are two topic positions: one above FocP and one below. The Top1P above FocP is the position discussed in this chapter. The other Top2P, mentioned in chapter two, is discussed in detail in chapters six and sev en, and it relates the CFN. 98

PAGE 99

(130) ForceP Force Top1P Top2 FocP Foc Top2P Top2 Agr Subject Agr Agr TP T T vP verb 99

PAGE 100

CHAPTER 4 PREVERBAL NEGATION 4.1 Introduction As this work compares the differences between preverbal and postverbal negative markers, I begin with an analysis of pr everbal negation. Several linguists have examined negation through the lens of generative grammar, especially since Pollocks work on IP (Pollock 1989, Iatridou 1990, Laka 1990, Belletti 1990, Zanuttini 1991, Mioto 1992, among many others). In this chapter I discuss negation in general and BP specifically. Section 4.2 discusses negation and the different approaches to understanding syntactic issues concerning negation. Part of this section reviews some ideas regarding negation in Romance langua ges. Because the topic of negation is larger than the scope of this work, I omit discussion on the semantics of negation. Therefore, I focus solely on some of the more common syntactic analyses of negation. Section 4.3 moves the discussion to Brazilian Portuguese and the preverbal negative marker. I leave the clause-final negative ma rker for chapters fi ve and six and negative concord and n-words for chapt er seven. Other negative elements such as NPIs and semi-negatives, words like sem without, are not discussed here. 4.2 The Syntax of Negation The main goal of this section is to in troduce sentential negation and briefly discuss how sentential negation is understood in generative syntax. Section 4.2.1 discusses the functional projection NegP. Here I discuss the status of certai n preverbal negative markers such as no for BP, no in Spanish, and non for Italian. To do this, I continue the discussion from chapter three about Pollocks (1989) work. Section 4.2.2 presents overviews of the theories related to the location of NegP in Romance languages. 100

PAGE 101

4.2.1 NegP While many works predate Pollock (1989), such as Jackendoff (1972), Lasnik (1972), Emonds (1978), he provides evidence that negation is a separate functional category, NegP. Pollock expanded IP and posited a syntactic location for NegP. In chapter three, I mentioned his (1989) work as an argument for separating IP into AgrP and TP projections. Here I extend that discu ssion to include the data that he provided with regards to negation. Building on Chomsky (1976) and Emonds ( 1978), Pollock observes the differences between the position of negative markers in English and French, much like he does with adverbs. In English, the negative marker must precede a lexical verb, whereas in French it follows the verb. (1) John doesnt likes Mary (2) *John likes not Mary (3) *Jean (ne) pas aime Marie John doesnt love Mary (4) Jean (ne) aime pas Marie John doesnt love Mary In English, the negative marker must come after auxiliary verbs and before lexical verbs. This differs from French where the negative marker must come to the right of both auxiliary and finite lexical verbs. To understand this difference and to keep in line with linguistic universals, Pollock proposes that the negative marker for English and French are in a similar position, and the verbs movement differs. His conclusion is that in English auxiliary verbs are in a position similar to that of lexical verbs in French, seen in a comparison of (5) and (6) with (4) 101

PAGE 102

(5) He is not coming (6) He must not come He also proposes that Neg houses preverbal negative markers, such as ne in French, and spec,NegP is the locus for negative adverbs, pas in French. Th is line of reasoning and his analysis were adopted by several linguists (Laka 1990, Chomsky 1995, Lasnik 1999). (7) TP Spec T T NegP Spec Neg pas Neg AgrP ne Spec Agr Agr VP Spanish, Italian, and Portuguese all have obligatory preverbal negative markers, and based on Pollocks arguments, the negative marker for each of these should be in Neg. In addition to that, Haegeman (1995) claims that negative markers such as no and no are not only in Neg, but they are also generated there. Zanuttini 13 (2001) claims that the preverbal negat ive marker is a head and that it is the head of NegP. Following Kayne (1989), she as sumes that syntactic heads in terfere with movement of other heads. This is based on what is known as the Head Movement Constraint which is formalized by Travis as follows: 13 For arguments against having the preverbal negative marker be a head see Rowlett (1998) who gives evidence from French against some of Zanuttinis claims. 102

PAGE 103

(8) Head Movement Constraint An X0 may only move into the Y0 which properly governs it. (Travis 1984:131) Thus, head movement is strictly local and always moves one head to the closest head. If the preverbal negative marker blocks head movement, then it is the head of NegP. Kayne showed that French pronominal clit ics that correspond to arguments of an embedded clause cliticize to the matrix verb in causative constructions. This is seen in example (9) where the clitic la cliticizes to the matrix verb fait of the main clause (Zanuttini 2001:524). (9) Jean la fait manger Paul John 3S makes eat.INF PREP Paul John makes Paul eat it The clitic is not allowed to clit icize to the matrix verb if t here is a negative marker, in this case ne, blocking the movement. (10) *Jean la fait ne pas manger lenfant John CL-has made NEG NEG eat.INF PREP DET-child John made the child not eat it Zanuttini also bases her evidence for the head status of the preverbal markers in Italian and Spanish on clitic climbing. She shows with Italian data that a clitic pronoun can climb to the matrix clause if it is not bl ocked by the negative marker. I give similar evidence from Spanish. Example (11) shows that the clitic may cliticize to the verb of the infinitival clause. (12) shows the clitic pronoun climbing to the matrix verb. (13) introduces the negative marker no which causes the clitic to remain in the infinitival clause. (14) is the key to Zanuttinis argument bec ause here the clitic is not allowed to 103

PAGE 104

climb to the matrix clause as in (12) She claims that the reason for the ungrammaticality of (14) is the head status of no which blocks the clitic. (11) Juan quiere verlo John wants see.INF-CL John wants to see it (12) Juan lo quiere ver John CL wants see.INF (13) Juan quiere no verlo John wants NEG see.INF-CL John wants to not see it (14) Juan lo quiere no ver John CL wants NEG see.INF John wants to not see it Clitic climbing can also be used to argue that BP no is a base-generated head of NegP: (15) Joo quer te ver John wants CL.2S see.INFL John wants to see you (16) Joo te quer ver John CL wants see.INF John wants to see you (17) Joo quer no te ver John wants NEG CL.2S see.INFL John wants to not see you (18) Joo te quer no ver John CL.2S wants NEG see.INF John wants to not see you 104

PAGE 105

This section has shown that the preverbal negative markers no, no and non block clitic climbing, and this blockage is re lated to head movement. This implies that these preverbal negative markers occupy a head position. 4.2.2 The Location of NegP While Pollocks work was a success in t hat it highlighted the importance of a functional projection specifically for negation, problems began to arise as to the actual syntactic location of NegP. While his claims seemed strong when comparing two languages like French and English, cross-lingui stic evidence suggest ed that NegP could possibly be in different positions in different languages. Some researchers continued this line of reasoning and cl aimed that NegP was not t he only functional projection responsible for negation. This section discu sses two works related to these ideas. Section 4.2.2.1 presents Laka (1990) and discusses her arguments for the addition of P, which is a projection responsible for polarity. 4.2.2.2 discusses Zanuttinis work on negation in Romance. Both of these re searchers present t heories that linguists debating negation in Brazilian Portuguese have turned to. 4.2.2.1 Laka Laka (1990) continues along the same lines as Pollock, and she looks to account for the differences between English and Basque. She claims that while there is a category responsible for negation, it is not fixed and it is not solely for negation. Consequently, NegP, which she renames P, does not necessarily have to be within IP, but can be above it or below it (Ouhalla 1990). This can first be seen in the variation that exists among languages with regards to the syntactic position of negation markers, but as Laka argues, also with regards to positive/emphatic markers because there exists a certain parallel between negative cons tructions and emphatic affirmative ones. 105

PAGE 106

The first set of English sentences shows that an emphatic positive DID appears to be in the same structural position as negation. (19) He DID come (Emphatic) (20) He didnt come. Laka then gives more complex evidence that negation and emphatic verbal markers are syntactically similar. She shows that in affirmative sentences in Basque the auxiliary verb always follows the main verb, (21) and never the reverse (22) and no element can intervene between the verbal elements (23) (21) Etxea erori da House fallen has The house has fallen (22) Etxea da erori House has fallen (23) Erori etxea da Fallen house has (Laka 1990: 14, 18-19) Nevertheless, the presence of the negative sentential marker ez alters this order. The auxiliary and the negative marker both precede the verb (24) and the necessity of the auxiliary and the verb to be adjacent is lost (25) In negative sentences, the negative marker and the auxilia ry must be adjacent (26) and (27) (24) Etxea EZ da erori house NEG has fallen The house hasnt fallen (25) EZ da etxea erori NEG has house fallen 106

PAGE 107

(26) EZ etxea erori da NEG house fallen has (27) EZ etxea da erori NEG house has fallen (Laka 1990: 25-26) Examples (28) (29) and (30) show that the same mechanism that causes the negative marker and the auxiliary verb to be adjac ent applies to emphatic elements. If the preverbal subject is adjacent to the auxiliary verb, it must be emphatic, as in (30) or the sentence is ungrammatical, as in (29) (28) Mari ez da joan. Maria NEG has left Maria hasnt left (29) *Mari da joan Mari has left Maria has left (30) MARI da joan Maria has left MARIA has left Given this similarity, Laka proposes that bot h are instances of a specific functional category P (SigmaP) which represents some sort of Speech Act. She claims that Ps position is language specific, contrary to Pollocks claims of a universal negative locus. Laka presents evidence from English and Basque that argues for two different positions for P. She notes that in Basque wher e ellipsis occurs, only the negative marker remains and the auxilia ry is deleted (Laka 1990:33). (31) Marik liburua erosi du eta Peruk ez Maria book bought has and Peter NEG Maria bought the book but Peter didnt 107

PAGE 108

English does not allow the negative marker to remain alone, and must be accompanied by the auxiliary verb do (Laka 1990:32). In example (32) everything to the right of the auxiliary is elided. In (33) everything to the right of the negative marker is elided as well, but in (34) English does not allow ellipsis with just the negative marker. (32) Mary didnt buy a book, but Peter did (33) Mary bought a book and Peter didnt (34) Mary bought a book and Peter not Laka argues that this shows a difference in the structural positions of Basques P and Englishs P. Basques P is above TP, (35) and in English, P is below TP, (36) (35) Basque P TP ez (36) English TP T T P did nt 4.2.2.2 Zanuttini Zanuttini (1997, 2001) continues along simila r lines as Laka by arguing that the location of NegP is not as rigid as Pollock or iginally proposed. Her proposal for NegP 108

PAGE 109

takes into account the different types of negative markers that exist. Through a typological approach, she determines that there are four classe s of negative markers: (i) negative adverbs, (ii) strong preverbal negative markers, ( iii) weak preverbal negative markers, and (iv) negative markers that are part of the verbal morphology. She is principally concerned with t he distinction between strong and weak negative markers, noting that several Romance languages have two negative markers in a sentence. In the cases of two negative marker s, one of the markers is st rong and the other is weak. For her, a strong negative mark er is one that can negate t he sentence independent of any other negative element. The weak negativ e markers must be in conjunction with some other negative marker. More t han one position needs to be posited because there are two elements. Fo r her, a strong marker is the head of a NegP which she refers to as Pol(arity)P because the label PolP suggests that the projection contains not only negative elements but also markers of em phatic affirmation, which in some cases can be shown to be in complementary distri butions with sentential negative markers (Zanuttini (1997:22). The weak negative marker is similar to the position that Pollock (1989) proposes for pas in NegP. Her analysis involves several factors including cliticization of the weak preverbal negative marker to the verb; however, If the preverbal element is weak then it is adjoined to the VP and cliticizes to the ve rb. If the negative marker is strong, then it is generated in Neg : 109

PAGE 110

(37) NegP Neg Neg AgrP Strong Marker Agr Agr FP-1 F-1 F1 VP Weak Marker F1 FP-2 VP Furthermore, Zanuttini extends the number of possible positions for negative markers. Not only can NegP dominate IP or be dominat ed by IP as Laka proposes, but as Pollock suggests, it can be within a split IP, and IP can be split into more than just two categories. Zanuttini claims, based on adverbs and Cinques hierarchy (1999), there are at least four positions for negative mark ers in relation to IP, i.e. one above IP, two within IP and one below IP. (38) [NegP1 [TP1 [NegP2 [TP2 [NegP3 [AspPperf [Aspgen/prog [NegP4 ]]]]]]]] Two of the positions are seen when l ooking at standard Italian and Romagnolo, a dialect of Italian. Standard Italia n has a strong negative marker, non that is generated in NegP1 (just like Spanish no). Romagnolo has a weak preverbal clitic negative marker that generated below the verb and moves up to the verb and cliticizes to it. The postverbal negative marker is in one of the other NegP and the ve rb with the negative clitic move past it to render the Neg-V Neg order (example (40) from Zanuttini 1997:17). 110

PAGE 111

(39) Maria non studia Mary NEG studies.3S Mary doesnt study (40) Ans dis brisa aks SUBJ.CL-NEG.CL say NEG like-that one doesnt say it that way 4.3 Preverbal Negation in BP The negative marker no is similar to other prev erbal negative markers in Romance languages, such as Italian and Spani sh (Belletti 1990, Laka 1990, Mioto 1992, Zanuttini 1997, Zeijlstra 2004). Followi ng Belletti (1990), Mi oto (1992) points out three options for the structural position of negation in BP: N egP dominates IP (i); NegP is somewhere within the IP, between flexio nal nodes (ii); IP dominates NegP (iii). (41) i. CP [NEGP] [AgrP] [TP] [VP] ii. CP [AgrP] [NEGP] [TP] [VP] iii. CP [AgrP] [TP] [NEGP] [VP] This section looks at these positions and determines where negation falls in BP. Section 4.3.1 discusses the different phonological fo rms of the negative marker. This becomes relevant in chapter five when disc ussing the rise and reason for clause-final negation. Sections 4.3.2 and 4.3.3 discuss two previous studies of BP negation. The proposals of this chapter are merely for pr everbal negation. Chapt er six continues the discussion of sentential negation and CFN, combining the information from each chapter. Following the review of past accounts, I lay out ev idence which shows that the preverbal negative marker in BP is the head of NegP, which is between AgrP and TP (ii from (41) ). To do that I follow some of the previous authors claims. Namely, as Pollock 111

PAGE 112

shows, the NegP is within the IP domain; however, that does not mean that there are not other syntactic positions for NegP, mu ch like Zanuttini (1997) and Zeijlstra (2004) show. Rather, BPs negative mark er is generated within IP. 4.3.1 The Phonology of th e Preverbal Negative Marker The phonological weight of the negative marker has come into question in BP. Specifically, some claim that the mark er has weakened semantically as well as phonologically, like the French preverbal ne gative marker, to the point of needing postverbal negative emphasis (Schwegler 1991a, da Cunha 1996, Vitral 1999, Martins 1997, Fonseca 2002). In chapter five, I s how that although ther e is phonological reduction, this has not resulted in any syn tactic change in sentential negation. This section addresses the issue of phonological reduction and presents the relevant data. This is important because as chapter six shows, only the preverbal negative marker allows phonological reducti on, and the clause-final marker is always the full form no As mentioned, other Romance languages ha ve weak negative markers, usually clitics, in addition to strong ones. Genera lly, a language has ei ther a strong negative marker or a strong and weak negative marker. However, no in BP has a certain duplicity about it. For ex ample, Sousa (2007) and Namiuti (2008) both suggest independently that no is a clitic, suggesting that it is a weak negative marker like ne. In spite of its clitic nature, it can negate the sentence by itself much like Spanish and Italian. Namiuti (2008), in looking at the hist ory of the negative mark er in BP, reflects and discusses the non-clitic/clit ic nature of the negative mark er in BP. Sousa (2007) uses a corpus study to examine the different phonological forms of no She concludes 112

PAGE 113

that there are essentially fi ve forms with the reduced form nu [nu] and num [], both clitics being the most common: (42) Sousa (2007:57) No Num Nu u N Total 130 301 454 15 18 918 No is often considered a strong negative marker meaning that it is not a clitic (Mioto 1992, Martins 1994, Sousa 2007). The other variants of the negative marker are assumed to be clitics. Mioto claims that the clitic variants are not allowed to stand alone, as in the response to a direction question, in example (43) (43) Q: O Joo agrediu o Pedro? DET John attacked DET Peter Did John attack Peter A: No/*num, nu, u, n The idea that the preverbal marker is a weak clitic supports many linguists claims about the status of the preverbal marker and CFN. However, recall from chapter one that T1 negation is the overwhelming preferred choice for negation. Combine that information with the data from Sousa which s hows that the clitic negative markers are the most common forms for the preverbal negative mark er, and an argument against the weak preverbal negative marker can be made. Namely, although there has been phonological reduction, the preverbal negative marker still carries the syntactic and semantic properties to negate a sentence. This topic is continued in chapters five and six where I discuss previ ous theories of CFN. 113

PAGE 114

4.3.2 Mioto 1992 & 1998 Mioto claims that NegP is above the IP dom ain. Before giving linguistic evidence for his claim, he begins by asserting that NegP is a [+I]nflectional because of its relationship to IP which is not split accordi ng to him. This relationship is based on the fact that no has the ability to change the truth value of a sentence, i.e. cause a proposition p to become ~p. He also claims that the unbreakable no+clitic+verb order is based on movement of lower inflectional it ems to Neg. Thus, the structure that he proposes is: (44) NegP Subjecti Neg Neg IP no ti I I VP Mioto argues that his structure comes from three principal piec es of evidence. The first piece of evidence is related to what Laka (1990) claimed for Basque and Spanish, i.e. these languages license a null IP in coordinated sentences such as (45) In this example, no is the head of a phrase, and everythi ng below is deleted. Given this example, Mioto claims that NegP is a functi onal category, as it allows ellipsis (Lobeck 1995). Recall that he claims that IP is not split as Pollock claims. (45) A Sienna viu a Jlia mas a Lcia no [viu a Jlia] DET Sienna saw.3S DET Julia but DET Lucia NEG [saw. 3S DET Julia] Sienna saw Julia but Lucia didnt [see Julia] 114

PAGE 115

The second reason for placing NegP above IP is to properly understand the clitic nature of the negative marker. Mioto claims t hat only the structure NegP dominating IP is descriptively accurate in capturi ng the duplicity of the clitic nature of no He continues and claims that if NegP were within IP, the behavior of no in BP would not permit placement of it into the class of cliti cs that can take care of an empty IP (Namiuti 2008:167). Therefore, BP a llows sentences where no is not deleted in the TP. 14 Where the clitic no is attached to the verb in a lower position, then a sentence like (46) would not be allowed because in this sentence the negative marker would have to be deleted along with the verb. (46) A Sienna viu a Jlia mas no o Ablio DET Sienna saw.3S DET Julia but NEG DET Abllio Sienna saw Julia but di dnt (see) Abllio The third reason that Mioto claims NegP is above IP is because of the difficulty of inserting a negative adverb into the structure, since the verb move s at least to TP. According to him, if NegP is above IP then it becomes easier to understand why a sentence like (47) is grammatical. In this se ntence, he claims that the negative nunca never is above IP and is a llowed. This is because, if the negative adverbial nunca is to the right of the verb the sentence becomes degr aded, as seen in example (48) (47) A Sienna nunca chora DET Sienna never cries.3S Sienna never cries (48) ??A Sienna come nunca cenouras DET Sienna eat.3S never carrots Sienna never eats carrots 14 This is allowed presumably because movement is adjunction, and the adjoined verb would be deleted leaving the object, below, intact. 115

PAGE 116

A problem with Miotos proposal is the subj ect position. According to him, the subject in affirmative and negative sentences has two different positions. In the affirmative sentence it is in spec,IP, and in negative sentences, the subject is in spec,NegP. However, this problem could be resolved by using Lakas P which houses both negative and in most cases, abstract positive markers. Another problem with his analysis comes with sentences that make use of modal auxiliary verb poder to can/be able to/might. The auxiliary is often overlooked in its dual role. This is seen in the examples below. The positive sentences appear the same, but do have different mean ings depending on the context. (49) Ele pode fazer o trabalho 3S can do. INF DET work. He can/is able do the work (50) Ele pode fazer o trabalho 3S can do.INF DET work. He might do the work The difference of syntax and meaning is clearer when the negative marker is added to the sentences. In the first case, the sentence is stating that the person is not capable of doing the work. The second sentence by having the modal verb to the left of the negative marker means that t he person might not do the work, although it might be the case that he is capable of doing the work. In fact, as example (53) shows, the auxiliary poder can be combined with poder to be able to. (51) Ele no pode fazer o trabalho 3S NEG can do. INF DET work. He cant/ is unable do the work 116

PAGE 117

(52) Ele pode no fazer o trabalho 3S can NEG do.INF DET work. He might not do the work (53) Ele pode no poder fazer o trabalho 3S can NEG BE.ABLE.INF do.INF DET work. He might not be able to do the work The examples show that it is not a lways the case that the negative marker precedes the verb and its auxiliaries. Thus although it was not mentioned in chapter three, IP supposedly can be split into several proj ections. If this is the case and if NegP is within the IP, then sentences like (52) exist where the negative marker is not directly after the subject, as Mioto cl aims, but is split by the insertion of a auxiliary. 4.3.3 Martins 1994 and Namiuti 2008 Martins (1994) and Namiuti (2008) follow Mioto in stating that negation is above IP. However, Martin s claims that NegP is not a co rrect denomination for what happens in BP. Rather, she chooses to follow Laka (1990), and concludes that P is the correct nomenclature for BP. Her main objection, much like an objection in the previous section is what to do with t he subject. Miotos proposal argues for two different subject positions, and for Laka and Martins, this is undes irable. Martins basic structure is seen in example (54) Notice that in her structure, as well as Namiutis, the negative marker is generated in NegP, which is in the IP domain. The negative marker then moves to P to check its strong negative feature (more on this in chapter six). 117

PAGE 118

(54) P Subjecti AgrP noii ti Agr Agr NegP Neg TP tii ti T T VP By having the negative marker originate in NegP, within the IP domain, Martins (1994) and Namiuti (2008) claim the stru cture and movement are able to properly account for the clitic nature of the negative marker and it s interaction with pronominal clitics. By way of example, Martins (55) shows the order of neg-clitic-verb, and (56) shows that that pronominal clitics must be adja cent to the verb. If, as Martins claims, the negative marker must move to then this would always prevent the pronominal clitic from being separated from the verb. (55) A Sienna no te viu. DET Sienna NEG CL.2S saw.3S Sienna didnt see you (56) *A Sienna te no viu DET Sienna CL.2S NEG saw.3S Sienna didnt see you Additionally, the movement of the negative marker explai ns the ungrammaticality of sentences like (58) (57) shows that a negative marker mu st be adjacent to the verb 118

PAGE 119

and that nothing else besides clitic pronouns can intervene. The adverbial expression na noite passada last night is not allowed to intervene, (58) (57) A Sienna na noite passada no quis dormir DET Sienna PREP.DET night last NEG wanted.3S sleep.INF Sienna last night didnt want to sleep. (58) *A Sienna no na noite passada quis dormir NEG Sienna NEG PREP.DET night last wanted.3S sleep. INF Sienna last night didnt want to sleep. Much like Lakas claims for English and Ba sque, Martins claims that there is evidence for a positive-type marker in Po rtuguese. In Portugues e the affirmation or positive indicator of the existence of comes from verbal respons es to direct questions. Martins argues that in English did in an emphatic sentence is the head of P. She then continues by positing verb movement from I to This supposedly always occurs, but is more apparent, according to the author, with di rect questions. To answer a direct question in Portuguese affirmatively, sim yes is not used; instead, the verb is used. Martins says that this behavior supports the idea of the verb rising to the head of P to check its positive featur es. As an example, (59) is asked. The felicitous responses to this are (60) and (61) (62) has the representative tree for (60) (59) Did you go? (60) Fui Went.1S I went/yes (61) No fui. NEG went.1S I didnt go 119

PAGE 120

(62) P AgrP fuii Agr Agr TP T T ti However, it has been shown elsewhere that this type of answer is not necessarily related to P (Santos 2004), but rather to some topi c position in the CP or deletion of VP (Schwenter 2005). Martin s (1997) bases her arguments on evidence from postclausal negation in BP that the ve rb moves from V to Top. While Martins might be right in her objections to Miotos multiple subject positions, there does not seem to be any reason to not allow more than one subject position in BP (Costa 2001 posits several positions for t he subject in European Portuguese). In addition to this, as shown in the previous section, there are cases where the negative marker appears to not move to because the subject and the negative marker can be separated by a modal auxiliary like poder 4.3.4 The structural Position of NegP in BP I now take the information given in the pr evious sections and apply it to the facts concerning BP negation. It is crucial that I lay out the exact position of the negative marker in a BP sentence because the subsequent chapters use this information as a 120

PAGE 121

basis for understanding clause-final negation and nega tive concord. Below is a review of the facts from above. The negative marker in BP is: i. preverbal ii. adjacent to the verb and only separated by pronominal clitics iii. a functional category that allows ellipsis Other relevant facts discuss ed in chapter three are: iv. the subject is in spec,AgrP v. the verb only moves to T (i.e. short movement) First, as mentioned, Miotos structure is problematic due to two subject positions. This is remedied by Martins proposal to use P, which simplifies the grammar by postulating a single subject position; how ever, Martins then proposes two negation positions. So, although the subject has been simplified, negation requires movement. Given these considerations and iv and v above, I conclude that no is the head of NegP, which appears between AgrP and TP: (63) AgrP subject Agr Agr NegP Neg TP no T vP verb This structure correctly places negation in a preverbal position without any movement of the negative element. It will license elli psis by having TP elide. 121

PAGE 122

This structure can also account for the data with modal auxiliaries and the negative marker seen above. Example (64) is a repeat from above and this example has the modal to the left of the negative ma rker. This sentence has the meaning of might not, and it is not an ex ample of constituent negation. Constituent negation in this case would require stressing the negative marker and would render the meaning he can NOT do the work. The syntactic tree structure is given in (65) (64) Ele pode no fazer o trabalho 3S can NEG do.INF DET work. He might not do the work (65) AgrP Elei Agr Agr XP X X NegP pode Neg TP no ti fazer o trabalho Assuming a functional head position for the m odal above TP and NegP but below AgrP within a split IP (Costa & Galves 2002) correctly allows this sentence. 4.4 Conclusion The principal goal of this chapter was to review some ideas about syntactic negation and to examine the preverbal negative marker in BP. I discussed general accounts to understanding the syntax of negati on, and the syntax of negation in BP. I have shown that no is a syntactic head that is generated as the head of NegP and that NegP is between AgrP and TP in BP. Mu ch like Zeijlstra (2004), Laka (1990), and 122

PAGE 123

others, I am not claiming the NegP must nec essarily be the same in all languages, but rather, there are langu age specific positions. 123

PAGE 124

CHAPTER 5 THE PRAGMATICS AND SEMANTICS OF CLAUSE-FINAL NEGATION 5.1 Introduction This chapter analyzes semantic and pragmat ic restrictions on the use of clausefinal negation in Brazilian Portuguese (BP). The goal is to properly account for the felicity of sentences that contai n clause-final negation (CFN) within a semantic/pragmatic framework. Ideally, this chapter sets up a theoretical base from which to understand the syntactic st ructural facts of CFN in BP. As mentioned in chapter one, the follo wing examples are instances of CFN. 15 (1) shows that the clause-final no can cooccur with the pr everbal negative marker no This type of construction is referred to in the literature as T2. (2) also a version of T2, has some other negative item, a so-called n-word, in a preverbal position. Finally, different from the fi rst two examples is (3) T3, where there is no preverbal negative marker. (1) A L no come almoo em casa no. T2 DET Lu NEG eat.3S lunch PREP house NEG Lu doesnt eat lunch at home (2) O Csar nunca faz o trabalho no. T2 DET Cesar never do.3S DET work NEG Cesar never does the work (3) A Sienna quer comer no T3 DET Sienna want.3S eat.INF NEG Sienna doesnt want to eat Section 5.2 discusses three previous approaches to understanding the semantic and pragmatic restrictions on CFN: section 5.2.1 looks at emphasis (Roncarti 1996, 15 Because preverbal negation was treated in chapter four I do not specifically address it here; however it does play a role in understanding CFN. This idea is pursued in chapter six. 124

PAGE 125

Barme 2000); section 5.2.2 investigates presupposition or contrary to fa ct (Schwegler 1991a, 1996; da Cunha 1996, 2001, 2007), and Section 5.2.3 looks at discourse status (Prince 1992, Schwenter 2005, 2006, Arms trong 2008). In each case, I present a theory and show how it has been applied to BP. I ultimately choose Schwenters (2005) discourse status or discourse-old proposal for CFN in BP, and I show in section 5.3 that this accounts for several un-elicited sentences, as well as ruling out infelicitous sentences. 5.2 Previous Approaches to Clause-Fina l Negation: Semantics/Pragmatics 5.2.1 Emphasis The idea that the CFN is emphatic comes from relating BP to the process known as the Jespersen Cycle (JC) (R oncarti 1996). Simply put, the JC is a cycle that involves a change that takes place with the sententia l negative marker changing its syntactic position, usually from a prever bal to a postverbal position, or vice versa. In many cases, JC is observed when a preverbal negative marker loses its inherent negative semantic force, and often this loss of negative force is related to phonological reduction. This loss of force motivates the rise of a postverbal negative element to reemphasize negation in the sentence. French is a classi c example of this, in which the postverbal emphatic marker becomes the sole negative el ement in the sentence, and the original preverbal maker disappears. Below, each French 16 example is followed by a BP 16 Zeijlstra (2004) gives a more comp lete breakdown of the stages th at make up the Jespersen Cycle: Phase I: Negation is only expressed by a single negativ e marker that is attached to the finite verb. Phase II: The negative marker that is attached to the finite verb becomes phonologically too weak to express negation by itself and a second negat ive adverb becomes optionally available. Phase III: Sentential negation is obligator y expressed by the negative marker that is attached to the finite verb and the adverbial negative marker. 125

PAGE 126

sentence that is assumed to showed the JC in BP (Vitral 1999). The sentences all mean: I dont know. (4) a. je ne sais stage 1: verbal negative marker (NM) a. eu no sei b. je ne sais pas stage 2: preverbal NM & postverbal emphatic adverb b. eu num sei no c. je ne sais pas stage 3: preverbal NM & obligatory postverbal NM c. eu num sei no d. je sais pas stage 4: lost of preverbal NM & only postverbal NM d. eu sei no Recall from chapter four, the preverbal NM has weakened phonologically from [nw] to [nu]/[nu ]/[n] (Roncarti 1996, Vitr al 1999, Sousa 2007). Ronc arti claims that this weakening, as in French, has caused the addition of a postverbal negative marker to appear (see Lobato 1986 for an early discussion of th is). Roncarti claims that CFN is emphatic because it does not cause a double negation reading, just as the addition of pas in French does not give a double negation reading. That is, CFN does not appear to interact semantically wit h the preverbal negative marker. As such, the sentence ele no comeu o po no does NOT mean that he didnt not eat the bread. Taking the example sentences from (4) according to those that take BP negation to be changing Phase IV: The negative adverb is the obligatory marker for negation and the use of the negative marker that is attached to the finite verb becomes optional. Phase V: The negative adverb is the only available negative marker. The negative marker that is attached to the finite verb is no longer available. Phase VI: The negative marker is avai lable in two forms: it can appear ei ther as negative adverb or as a negative marker that is attached on the fini te verb, though sometimes simultaneously. Phase VII: Negation is only expressed by a single negative marker that is attached to the finite verb. 126

PAGE 127

within the context of the JC, BP is movi ng from stage 2, wher e the clause-final no is emphatic, to stage 3, where the no is obligatory (Schwegler 1988, 1991a). Not only does this appear to be the JC, there is evidence from the positive marker that this is a case of an emphatic element. Example (5) shows that the positive marker sim may appear in the same clause-final position as the negative marker. The argument here is that this clause-final position in Po rtuguese is used for emphasis (Barme 2000). (5) Ele comeu o po sim 3S eat.PST.3S DET bread yes He did eat the bread In addition to this, CFN is seen as emphatic or intensifying because of its ability to occur with preverbal negative words like nunca never (Barme 2000). Nunca is sufficient to negate a sentence without additional support, as (6) shows. (6) Ela nunca come po 3S.F never eat.3S bread She never eats bread The argument is that nunca is not phonetically weak and, by extension, not semantically or syntactically weak, and ther efore needs no reinforcement. Yet, CFN is allowed in situations with nunca. This follows from Schweglers claim that there is a constant and universal psycholinguistic need for negative emphasizers (1988: 36), which is why example (7) has CFN. (7) Ela nunca come po no. 3S.F never eat.3S bread NEG She never eats bread 127

PAGE 128

There are problems with this analysis, however. First, this line of thinking hinges on the idea that the preverbal negative marker is w eakening to the point of needing some sort of support, as in French. The data reviewed in chapter one regarding usages of the preverbal marker suggest that the preverbal marker is not weak, i.e. roughly 75% of negative sentence rely solely on preverbal ne gation. If the preverbal negative marker were weak, then the use of type one negation would most likely be lower because it would need to be supported by so me emphatic negative element. Second, in previous attempts to understand CFN in terms of emphasis, researchers have failed to define the term adequately. As Schwenter notes, emphasis is not properly defined by those (i.e. Barme, Ronc arti) that use it to describe BP negation (Schwenter 2005, 2006). In many cases, there are two types of emphasis that are being discussed: i) reinforcement a weak negative marker; ii) intensification of the negative force of the sentence. On the one hand, emphasis is used as a reinforcing tool for negative markers that have phonologically or semantic ally weakened. This view stems from a particular conception of the Jespersen Cycle and t he desire to assimilate the BP data with previously studied cases of negat ion within the Romance family. However, since it has been shown in chapter one that T1 is st ill overwhelmingly the strategy used for negation, it is difficult to argue that t he postverbal negative marker is needed to negate a sentence. The purported necessity of a po stverbal marker entails that preverbal negative markers are no longer strong enough to negate a sentence on their own, which is not the case (Cavalcante 2007). 128

PAGE 129

The other use of the term is a more traditional one adopted by Schwenter, following Israel (1998): emphasis in language is informative because it exceeds what one would normally expect to be asserted (1998: 47). This means that a sentence like he didnt do a lick of work could entail that he didnt do work but not vice versa. This is because a lick of exceeds the normal expectations of not working in this case. A typical strategy used for emphasis in Portuguese is through the use of mesmo same. In the sentence below, mesmo indicates an emphatic meaning towards the positive or negative semantic meaning. No tice that in both cases, the emphasis adds to the meaning and suggests that the person worked (or didnt work) beyond the normal expectations i.e. in the firs t case she did all the work, even though she didnt have to, and in the second case, she didnt do anything, not even pick up one piece of trash from the ground. (8) Ela trabalhou mesmo. 3S.F work.PST.3S EMPH She REALLY worked (9) Ela no trabalhou mesmo. 3S.F NEG work.PST.3S EMPH She didnt work AT ALL By taking the definition that Israel gives and applying it to CFN, one can see that CFN is not a case of emphasis. See example (10) below. In this example, B is negating the accusation, and B goes one step fa rther to emphasize the fact that he didnt even break the cup. Use of the CFN is not felicitous in this instance. Also, the negative markers in both parts of the sentenc e could have an emphatic stress, but the 129

PAGE 130

clause-final negative marker can neither be stressed nor can it even appear in this situation. (10) A. Voc derrubou a gua. 2S spill.PST.2S DET water You spilled the water B. Eu no derrubei a gua, e no quebei o copo (#no) 1 S NEG spill.PST.1S DET water and NEG break.PST.1S DET cup NEG I didnt spill the water, and I didnt break the cup Finally, as Schwenter notes, even accepting that T2 negation is emphatic, there still exists the problem of T3 negation. In this case, there is no pr everbal negative marker, as such it is difficult to understand how this type of negation is emphasizing the sentential negation, because this is not present. 5.2.2 Presuppositional or Contrary-to-Fact Another approach to BP negation is one that takes CFN to be linked to presuppositions. This claim originally comes from Schwegler (1991a) and is similar to what da Cunha currently proposes (1996, 2007). Schwegler argues that CFN is used in sentences that presuppose a previous affi rmative assertion or assumption which [it] seek[s] to contradict. 17 (1991a: 194) That is to say, BP speakers use CFN when they want to contradict some presupposition. For the purposes here, to presuppose something means to assume something (S aeed 1997), or a presupposition is a set of assumptions made by participants in conver sation (Stalnaker 1978). Accordingly, a sentence your son is funny presupposes t hat you have a son. Similarly, if (11) below is 17 This is also the analysis that Zanuttini (1997) gi ves for several marked negative forms in some Italian dialects. 130

PAGE 131

uttered, Schweglers claim ar gues that there is a presupposition He lies to his parents that this sentence looks to deny or contradict. (11) Ele no mentiu par a os pais dele no. 3S NEG lie.PST.3S PREP DET parents prep.3S NEG He didnt lie to his parents Schwegler gives two examples to show how CFN is used to contradict a presupposition. He claims that the A sentences below do not presuppose and therefore do not challenge any previ ous statement assumption or implication on the matter involved, they simply assert a proposition, and the B sentences presuppose a previous affirmative assertion or assu mption which they seek to contradict(1991a: 194). (12) A. Quando estive no Rio, no fui na praia. When was.1sg PREP.DET Rio, NEG went.1S PREP.DET beach When I was in Rio, I didnt go to the beach B. Quando estive no Rio, no fui na praia no. When was.1sg PREP.DET Rio, NEG went.1S PREP.DET beach NEG When I was in Rio, I didnt go to the beach (contrary to others or my own beliefs) (13) A. O Brasil no um pais rico DET Brasil NEG is DET country NEG Brazil isnt a rich country B. O Brasil no um pais rico no. DET Brasil NEG is DET country rich NEG Brazil isnt a rich country (contra ry to others or my own beliefs) This analysis is weak for two reasons. First, if the presupposition is already negative, then the use of CFN yield a double negati on reading according to Schweglers own explanation. In the exam ple below, A gives her beliefs that B didnt do the work. According to Schwegler, if CFN were to be used, it would try to contradict the 131

PAGE 132

presupposition, i.e. that B didnt do it. This is not the case, however. Instead, CFN is used in sentences that agree with the presuppos ition, and with the sentence in general. (14) A. Eu sei que voc no fez o trabalho 1S know.1s COMP 2s NEG do.PST.2S DET work I know that you didnt do the work B. Eu no fiz o trabalho no 1S NEG do.PST.1S DET work NEG I didnt do the work The second problem with this analysis, as mentioned in Schwenter (2005), is that there are cases of where the use of CF N does not contradict an assumption or presupposition. One case where presupposit ion does not work is in presuppositional denials. Presuppositional denia ls are denials where negation is interpreted as applying to some presupposition of a prio r utterance, not to its assert ed content. In these cases, CFN is not possible. The examples below s how this to be the case (Schwenter 2005). The presupposition of the A example is t hat John smoked in the past, and this is contrasted with the assertion, that John has stopped smoking (Stalnaker 1978, 2002). According to Schweglers analysis, CFN shoul d be licit in the B example because it negates the presupposition John smoked. This is not the case, as CFN is not allowed. (15) A: O Joo j deixou de fumar. DET Joo already stop.PST.3S PREP smoke.INF John has stopped smoking B: Ele nunca fumou (#no). 3S never smoke.PST.3S NEG He never smoked' 132

PAGE 133

This pair of sentences contrasts with the pair in (16) CFN is not negating the presupposition but rather the assertion that John has stopped smoking. Evidence that the sentence is not negating the presuppositi on is seen in the second half of the sentence, where B agrees with A by claiming that John still smokes which also has as a presupposition John smoked in the past. (16) A: O Joo j deixou de fumar. DET Joo already stop.PST.3S PREP smoke.INF John has stopped smoking B: Ele no deixou de fumar (no), ele ainda fuma. 3S NEG stopped PREP smoke.INF (NEG), 3S still smokes. He hasnt stopped smoking, he still smokes This section has shown that although t here are claims that CFN is used to contradict a presupposition, this is not what allows CFN to be licit in BP. Rather, the data given has shown that CFN is related to what is asserted in the discourse and not necessarily to what is presupposed. 5.2.3 Discourse-Old I now lay out the discourse theory that is assumed throughout the remainder of this work. Schwenters 2005 proposal claims that the discourse information structure developed by Prince (1981, 1992) is sufficient to explain the data for BP CFN. After presenting Princes ideas, I present Schwenter s analysis of CFN in BP. I show that his application of Princes theory more broadly captures the semantic/pragmatic restrictions on CFN in BP. Princes goal is to properly describe the information structure of discourse. She has a far reaching definition of discourse that not only refers to the things said but also to the extra-linguistic elements, such as the context (Prince 1981) This means that 133

PAGE 134

discourse can be interpreted very broadly and include many facts and ideas that aid in the interpretation of the linguistic utterances. Prince also separates discourse into two categories: new information and old informa tion (Prince 1981, Horn 1989). She takes this separation one step further and dist inguishes between hearer information and discourse information. That is to say, something can be old/new with regards to the hearer, and something can be old/new in relations hip to the discourse. As for the status of the information with relationship to the he arer, Prince states t hat information, by which is here generally meant entities/ref erents, may be old/new with respect to (the speaker's beliefs about) the hearer's beliefs (1992:5 emphasis mine). 18 Essentially, Prince tries to show that hearer-status (i.e the hearers beliefs, etc.) can be partially independent from the discourse-status; however, the inverse is not true. If an entity is in the discourse, then it is necessarily within the realm of the hearers knowledge, belief, etc. To explain the point, Prince gives the following examples with an accompanying chart in (19) (Prince 1992:12). (17) a. I'm waiting for it to be noon so I can call someone in California b. I figure she 'll be up by 9, her time. (18) a. I'm waiting for it to be noon so I can call Sandy Thompson a. I figure Sandy / she'll be up by 9, her time. In (17) someone is a new entity to the discour se and it is new to the hearer; however, in the case of (18) Sandy Thompson is new to the discourse, but it is not new to the hearer, as it is assumed from this example that the hearer knows who Sandy Thompson is. (17) and (18) both have she, which in these sentences is now old to the 18 The hearers beliefs might differ from what the spea ker believes about the hearers beliefs, but this is not relevant for BP CFN, as such this will not be pursued. 134

PAGE 135

discourse and old to t he hearer. The chart in (19) shows Princes organization of discourse information. Notice that if something is discourse old it would be impossible for it to be new to the hearer; thus, the Hearer New/Discourse Old cell of the table is blank. (19) Hearerand Discourse-status of a discourse entity: Discourse New Discourse Old Hearer New Brand new: (17) a. someone Hearer Old Unused (18) a. Sandy Thompson Evoked: (17) b. she (18) b Sandy/she In addition to things mentioned in the disc ourse, Prince claims that NPs can be discourse old if they are inferable from the context. She gives the following example which mentions the Bastille. Because the Bast ille is an edifice, then it is inferable that the Bastille has a door. Therefore, B is fe licitous because the context lets the hearer infer the door (Prince 1992:8). (20) A. He passed by the door of the Bastille and the door was painted purple. B. He passed by the Bastille and the door was painted purple. By the same token, example (21) is not felicitous because it is not inferable from mentioning the Bast ille that there is the trunk This example shows that the discourseold requirement is not activated because the trunk is neither old to the hearer nor old to the discourse. (21) #He passed by the Bastille and the trunk was painted purple. 135

PAGE 136

It is important that the ideas of pr esupposed information and discourse-information are distinct, even though they might at ti mes intersect. The question can be asked: How is discourse-old different from presupposed information? To answer this, recall from the previous section w hat a presupposition is. Presupposition is related to the beliefs or assumptions related to assertions. The crucial poi nt is that a presupposition may or may not enter the disc ourse. If it does not enter the discourse, it cannot be discourse-old according to Pr inces definition. Example (22) repeated from above, shows the difference between presupposition and di scourse-old. In case of sentence A, the presupposition is John smoked in the past and the discourse-old assertion is John has stopped smoking. Sentence B denies the presupposition John smoked in the past, but it does not deny the discourse-o ld assertion that John stopped smoking. Sentence B denies the discourse-old asse rtion John stopped smoking and makes no direct claim about the presupposit ion John smoked in the past. (22) A: O Joo j deixou de fumar. DET Joo already stop.PST.3S PREP smoke.INF John has stopped smoking B: Ele nunca fumou 3S never smoke.PST.3S He never smoked.' B. Ele no deixou de fumar 3S NEG stop.PST.3S PREP smoke.INF He didnt stop smoking There is still another problem t hat lingers. Prince mentions the fact that discourseinformation can be old if it is inferable. If something is inferable, then most likely that inference is due to assumptions that the speakers hold. In the case above, people presuppose that Bastilles have doors, and this presupposition is based on the fact that 136

PAGE 137

people assume that buildings have doors. This is where the discussion encounters problems with inferable information, or bette r yet some presuppositions. CFN allows some inferable information to be negated usi ng CFN, but there are instances where CFN is not allowed. The reason for this is related to a quasi-scale of grading for how readily inferable the informati on is. If the information is readily inferable, doors on a building, then it is allowed. However, if the information is not readily inferable, like chairs in a house, then CFN is not allowed. More on this is seen in examples (39) and (40) For now, it is sufficient to say that CFN is used to deny or negate presuppositions, but it is not allowed with a ll presupposition, thus giving the advantage to the discourseinformation theory outlined for BP in the following paragraphs. Schwenter attempts to extend Princes analysis of NPs to whole propositions that are negated. He claims that CFN is licit wh en what is being negated is discourse-old, as in Princes examples (17) and (18) Additionally, the discourse-old feature can be on any part of the sentence. Part of Schwenters goal is to show that emphasis and presupposition are not proper descriptions of CFN in BP. He separates his proposal from Schwegler and da Cunhas proposals by separating out hearer-old information and discourse-old information. Recall that discourse-old and hearer-old do not mean the same thing. That is, a proposition can be hearer-old, this is like a presupposition, but not discourse-old, i.e. present in the current discourse. It is the discourse-old status of/within the proposition that licenses the felicity of a sentence with CFN. For a proposition to be discourse-old, it would necessarily need to be uttered or inferred from the context, such as a Bastille having a door given in Princes examples above. 137

PAGE 138

Schwenter extends Princes examples, whic h generally include only NPs, to larger pieces of the sentence. However, he doesnt fully define what a discourse-old proposition is. As such, for the purposes of this work, I define a proposition somewhat broadly because it has to include unspoken disc ourse as well as larger phrases such as VP/AgrP/etc... To give a complete definit ion of proposition for this work, I use the sentence below as an example: (23) Sienna ate bean soup. For a sentence like the one above, the proposition is Sienna ate bean soup It is a proposition because it has a truth-value that can be calculated. Following Princes proposals, if Sienna ate bean soup is uttered then it, the whol e proposition, is discourseold. Consequently, I call these propositions discourse-old propositions Also, if the proposition (23) is uttered, then it has been directly activated in the discourse and it is discourse-old. BP CFN extends the meaning of discourse -old propositions to include phrases and sentences inferred from the initial proposition. Inferred is the word that Prince uses but here I also say that BP CFN is connected to propositions that are entailed from the initial proposition. Pr opositions that are inferred or entai led from the init ial proposition are considered to be indirectly activat ed in the discourse. However, although not directly activated in the discourse, they are discourse-old, much like the door in Princes Bastille example. Indirectly activated discourse-old propositions, as mentioned above, are related to what can be entailed from the initial pr oposition. Below are examples of inferred 138

PAGE 139

discourse-old propositions. First, sentence (23) above entails sentences (24) and (25) and in an even more abstract way (26) : (24) Sienna ate something (25) Someone ate bean soup. (26) Someone ate something. Using a quantifier like something potentially raises the number of possibilities almost infinitely. Thus, for a discourse-old proposit ion, the something must be related to the original lexical idea. For example, it would not be possible to infer that Sienna ate a kitchen table but this is not related to bean soup or even food Again, this is much like Princes example of the Bastille door which is inferable from the discourse-old NP the Bastille That is, something is related if it is interpreted as part of some item in the proposition. The door on the Bast ille is related because it is interpreted that a Bastille has door, i.e. it is an inherent quality of Bastilles to have doors. The idea of a chest in a Bastille cannot be inferred from the Bastille because a chest is not interpreted as part of the Bastille Also, much like Princes theory, other va riables, especially the postverbal ones, if not explicitly stated can be in ferable. One can infer from the proposition that not only Sienna ate soup but also : (27) Sienna ate soup with a spoon Other prepositional phrases or adverbial phrases related to the original proposition are also inferable propositions that originat ed from the discourse-old proposition. 139

PAGE 140

Schwenter illustrates the discourse-old r equirement for CFN in BP with examples (28) and (29) (2005:1434). (28) shows that if the proposit ion, Someone turn off the stove, has not been introduced into the discourse, then the final no is infelicitous. [A speaker is walking down the street ta lking to her friend about a recent soccer game. Suddenly she remembers that she fo rgot to turn off the stove and says:] (28) Nossa! Eu no desliguei o fogo (#no)! Ours! 1S NEG turn.off.PST.1S DET stove (#NEG) Damn! I didnt turn off the stove! Notice the difference in the next example w here the proposition you turn off the stove is introduced into the discourse through a simple dialogue. [same situation as (28) the friend (a) asks (b):] (29) a. Voc des ligou o fogo, n? 2SG turned.off.2S DET stove, right You turned off the stove, right? b. Nossa! (No) desliguei no! Ours! NEG turn.off.PST.1S NEG Damn! I didnt turn it off! Recall that according to Princes definit ion of discourse that the information does not necessarily have to be introduced verbally. If there is something in the conversation that has been introduced either verbally or not, then CFN is felicitous. In this case, the idea of dropping the papers or that you dropped the papers, which is reduced to the pro-from isso, is what activates the CFN. It c an be inferred that dr opping the papers or doing something wrong or bad is generally not good, and when people do something bad they feel bad. Again, this is like Princes Bastille-door example. The same can be 140

PAGE 141

said with doing an act that is bad, where it can be inferred that people feel bad because of a certain act. [A Woman drops a stack of papers while working with her friend. The women, referring to the event (d ropping the papers) says:] (30) Esquenta por causa disso no 19 Heat.up.IMP.2S PREP cause PREP.DEM NEG Dont worry about it Now note, example (31) shows that even if the discour se is in progress, if the information is not present (disc ourse-old), then CFN is not felic itous. This is because in this case the proposition of going to the beach is not introduced into the discourse, and as such is not discourse-old. [A conversation about a recent trip to Rio de Janeiro] (31) A. O que voc no fez no Rio que queria fazer? DET what 2S NEG did.2S PREP.DET Rio that want.IMP.2S do.INF What didnt you do in Rio that you wanted to do? B. Eu no fui praia (#no). 1S NEG went.1S PREP.DET beach (NEG) I didnt go to the beach. This conversation could be imagined in a diffe rent way, where the two friends spoke before the trip about the plans that B has made, one of which was going to a beach like Copacabana. Now that the trip is over, A asks t he question. Even though the information is hearer-old, it is not licit to use CFN because it is not discourse-old, i.e. not present in the discourse. 19 i.e. i. Esquenta por causa do fato de voc ter deixado cair a folhas no. Dont worry about the fact that you let the papers fall 141

PAGE 142

Schwenter also makes a distinction bet ween T2 and T3, however in both cases, the motivation for CFN is related the discour se-information status of the proposition. His initial questions about this arise due to the actual percentage of usage found in corpus studies. For example, in da Cunhas study and spoken corpus from northeastern Brazil, there are only nine examples of T3 out of a total of 1465 negatives, a mere .6% (1996, 2001, 2007). Keep in mind that generally it is assumed that the dialects of northeastern Brazil are where CFN occurs most (Schwenter 2005). Roncartis (1996) corpus had only 39 tokens out of 813, 4.7%. In each case, T2 has much higher rates of use, 10.8% in da Cunha and 17.9% in Roncarti; although, these pale in comparison to T1, which is always a bout or above 75%. Additionally, these low numbers for T3 almost always included repet ition of the same verb previously mentioned in the discourse wit h no object present. Roncartis study shows that 72%, 28/39 of all the T3 sentences, were similar to the next example, where T3 occurs in response to a direct question an d the same verb is used. Ro ncarti also states that the other tokens, 11/39, were the expression sei no I dont know which is considered by the author to be a lexicalized form due to its similarity with another lexicalized form sei l I dont know (literally I know there) 20 (32) A. Voc gostou da palestra da Maria? 2S liked PREP.DET presentation PREP.DET Maria Did you like Marias presentation? B. gostei no liked.1S NEG I didnt like it 20 This form should not be confused with CFN in BP. The expression sei l only means I dont know. The adverb l there is only valid as a negat ive item in this context. Interestingly, quero l saber I dont want to know is also grammatical. This is related to chapter three where the position of the verb was discussed and adverbs helped show verb movement to T. 142

PAGE 143

Schwenter shows that if the verb is different in examples like (32) then T3 is not licit; however, T2 is. The author shows this in (33) where it is inferable that A believes (i.e. pragmatically presupposes) t hat B went to Marias talk (2005:1449). The proposition that is being negated is X went to Marias presentation, and because of this Schwenter claims that T2 is allowed. (33) A. Voc gostou da palestra da Maria? 2S liked PREP.DET presentation PREP.DET Maria Did you like Marias presentation? B. # fui no T3 Went.1S NEG I didnt go C. No fui no T2 NEG went.1S NEG I didnt go Because T3 is not allowed in these situations, Schwenter claims that T3 is a subset of T2 in that it is discourse linked but it can only be used when the discourse old proposition is directly activated, meaning it must be used with a repetition of the verb in the proposition. Biberauer & Cyrino (2009) also notice difference between T2 and T3. Although they do make not mention of the semantic/pragmatic mec hanics that allow CFN, they show that T3 can only be used in the cont ext of a presupposition denial. T2 does not have this restriction, as Schwenter states, a nd it can be inferred from the context. In the example sentences from Biber auer & Cyrino, a question is a sked that allows inferred information to become discourse-old (Schwenters term, not theirs), and therefore, T2 is 143

PAGE 144

licit. For the examples below, the response I cannot find my cat infers that people normally fell emotional when they lose a pe t. This type of context inference is not available in T3 CFN. This suggests that T2 and T3 ar e different and it is another example that T3 is limited in its discourse function to directly activated discourse information. In the case of th is situation, it must be menti oned that for B to be felicitous, it must be the case that A know s that B has a cat, and that the ca t is loved by B. If A is not aware of these facts, then in neither case is CFN licit. However, if A knows this then B can use T2 CFN because it can be infe rred that sadness and a certain desperation might accompany the loss of a pet. Again, ev en knowing the facts of this exchange, it is not licit to use T3 because the discourse in formation has not been directly activated. (34) A. Por que voc es t desesperada? O que aconteceu? why 2S ARE.2 desperate.F? what happen.PST.3S Why are you desperat e? Whats happened? B. Eu no t achando minha gatinha no T2 1S NEG am finding 1S pussycat NEG I cannot find my pussycat B. *T achando minha gatinha no T3 am finding 1S pussycat NEG I cannot find my pussycat One question for Schwenters proposal rema ins. He cites the example, seen above in (15) as a reason to favor discourse-old over presupposition. What is presupposed in As sentence is that John smok ed in the past. Because the negation in the second clause applies to the presupposed content and not the proposition John has stopped smoking, the use of CFN is not felicitous. Thus, in this case the presuppositional aspect of the sentence does not license CFN. 144

PAGE 145

(35) A: O Joo j deixou de fumar. DET Joo already stop.PST.3S PREP smoke.INF John has stopped smoking B: Ele nunca fumou (#no). 3SG never smoke.PST.3S NEG He never smoked.' Schwenters proposal accounts for when CFN is felicitous and explains when it is not. However, it does not explain why T3 is allowed to negate a sentence on its own or why T2 is not a double negative. For the purposes of his proposal it is not necessary to explain these ideas (see section 6.3.7 ). 5.3 Applying the Theory This section presents additional evidence to show that Schwenters discourse-old model for BP negation is correct. Although I have explained Schwenters proposal, here I add to his examples to show that the discourse-old proposal captures all sorts of different BP data. I will be using data from previous researc hers as well as data that I collected through the use of reality TV shows and data I gathered and tested through grammaticality judgments from my consultants both in Brazil and in the US. Additionally, while I reject the claims t hat emphasis and presupposit ion/contrary-to-fact are the primary motivations for CFN, I do not claim that the CFN cannot be used with these intended meanings. Rather, I claim that Schwenters us e of Princes theory is a more precise use of linguistic tools to account for more data and it can also be implemented to give a great er understanding of the syntactic forces behind these structures. The first example comes from a television program where the reporter is asking the woman if her ex-husbands wi fe is going to get custody of the child. Notice that here 145

PAGE 146

the direct question is answered, and then t he women claims that the child will not be taken away. In this case, the CFN is lic ensed because she take-away your son is discourse-old. While this might be an emphatic us e of CFN, it is not contrary to fact or presupposition on the part of t he reporter. His question is related to the ex-husbands wife wanting to take away the child, and the women responds by stating what she thinks will happen, although, she ackn owledges that the ex-husbands wife wants to gain custody. (36) A. Ela t quereno tir o seu filho, num isso? 21 3S.F is wanting take.INF DET 2S son, NEG is that? She wants to take your son, isnt that right? B. isso, mai ela num tir no porque... is, is that, but 3S.F NEG take NEG because... It is, thats it, but s he wont take him because Another example comes from a monologue, w here a wife is yelling at her husband. Here she is explaining that he is always following her around and is getting into her business. She introduces proposition you lo ok for me which entails the proposition someone look for someone, and later is abl e to use CFN because this has already been activated in the discourse. 21 Taken from youtube.com on April 23, 2009 http://www.youtube.com/watch?v=PMKRWOx zSdg=1&playnext_from=PL&index=5 146

PAGE 147

(37) Tu me tiraste porque eu tava numa boa e eu no queria que ... 2S 1S took.2S because 1S was PREP.DET good and 1S NEG wanted that andar de madrugada atrs de mim porque ...walk.INF PREP early.morning PREP PREP 1S because eu nunca andei atrs de tu no. 22 1S never walked.1S PREP PREP 2S NEG You took me...because I was doing great, and I didnt want you to look for me in the early morning because I never was looking for you. Finally, I present one more instance of CFN captured in writing. On an informal blog, a woman discusses a breakup that she rec ently suffered. She is able to use CFN because the discourse-old proposition he to speak with me is mentioned, and thus activates CFN through the inferred proposit ion someone to speak with someone. (38) ...e teve um dia q fui em uma pagode perto and have.PST DET day that was.1S PREP DET pagode close da minha casa ne, q ele esperou acabar p/ tentar PREP. DET 1S house NEG.is, that 3S waited finish.INF PREP try.INF falar comigo, e ai eu no falei com ele no. 23 talk.INF PREP.1S and there 1S NEG speak.1S PREP 3S NEG And there was a day that I went to a Pagode close to my house, right, and he waited until the party finished to talk to me, and so I didn t talk to him I now give some examples where the disc ourse-old status is not activated and as such, the CFN is not felicitous. (39) introduces painting the house into the discourse; however, nothing about painting chairs wa s mentioned, therefor e the CFN is not 22 Taken from youtube.com on April 29, 2009 http://www.youtube.com/watch?v=2JeDRh2YF5E&feature=related 23 Taken from yahoo brazil on April 29, 2009 http://br.answers.yahoo.com/question /index?qid=200 90324060211AAmhic3 147

PAGE 148

felicitous in B. That is, although the discour se-old proposition they paint the house was introduced, paint the chairs cannot be inferred because chairs is not an inherent property of a house, as wa lls, doors, and windows are. (39) A. Eu passei pela casa e pintaram a parede? 1S passed.1S PREP. DET house and painted.3P DET wall I went by the house, and did they paint the walls? B. Pintaram, mas no pintar am as cadeiras (#no.) Painted.3P, but NEG painted.3P DET chairs NEG They did, but they didnt paint the chairs Prince predicts that inferred discourse items can be negated which means that sentence (40) is allowed because a door is inferr ed to be part of a house. This is different from example (39) above, where painting chairs cannot be inferred from painting a house. That is, chairs are not di rectly relevant to painting a house, unless they were mentioned in the di scourse, which is not the case here. Painting a door or walls is directly related to painting a hous e as each of the make-up part of the house, i.e. the physical structure. (40) A. Eu passei pela casa e pintaram a parede? 1S passed.1S PREP. DET house and painted.3P DET wall I went by the house, and did they paint the walls? C. Pintaram, mas no pintaram a porta no. Painted.3P, but NEG painted.3P DET door NEG They did, but they didnt paint the door This section explained the different analysis used to explain the semantic/pragmatic reasons for CFN. It was shown that neit her the emphasis nor t he presupposition claims are sufficient to capture all of the data. Schwenters discourse-old application of 148

PAGE 149

Princes information structure theory is able to account for BP and CFN. The section also argues that T2 and T3 differ. That is, T3 appears to be a subset of T2 that is limited to directly activated proposition. 5.4 Conclusion This chapters goal was to given an explanat ion to the discourse restrictions that affect the felicity of CFN s entence in BP. I have shown th at previous attempts to understand CFN in terms semantics/pragmati cs have not be able to fully capture the meaning and use of CFN. I presented Schwent ers (2005) use of Princes DiscourseInformation theory. His result s were that CFN in BP is licensed if what is being negated is discourse-old information. Discourse-o ld information was defined as a proposition that had been previously introduced into the discourse, either verbally or non-verbally. After presented his theories, I applied the discourse-old theory to several new sentences of BP and showed that this theory is capable of explaining the felicitous as well as the infelicitous sentences. 149

PAGE 150

CHAPTER 6 THE SYNTAX OF CLAUSE-FINAL NEGATION 6.1 Introduction An analysis of BP negation should account fo r the semantic/pragmatic restrictions presented in chapter five as well as the syntact ic restrictions presented in chapter one. A syntactic theory should account for the fact that CFN is not allowed in wh-questions, relative clauses. It should also account fo r the licensing of n-words. Finally, an analysis should account for the acceptability of CFN in topic structures, imperatives, direct questions, and embedded clauses. Section 6.2 discusses previous analyses of BP CFN and the extent to which they capture the syntactic patterns from BP data. In section 6.3 I introduce my own analysis of CFN which builds on these earlier proposals and the semantic/pragmatic theor y argued for in chapter five. 6.2 Previous Accounts 6.2.1 Martins 1997 Martins (1997) discusses sentential negation in BP within a minima list framework. She, like those who propose an emphasis explanation, assumes that the change that has taken place in BP negation is directly re lated to the Jespersen Cycle. She agrees with those in the emphatic camp, discussed in section 5.2.1 that the postverbal negative maker is an emphatic negative marker that came about due to the preverbal marker weakening to the point of not negat ing a sentence on its own because of the weakened negative feature (Zanuttini 1997). In fact, Martins clai ms that preverbal negation is not sufficient to negate a sentence, which equates BP with modern French (1997:24). 150

PAGE 151

To structurally explain CFN in BP Martins discusses European Portuguese and what she labels Standard BP. She assumes that European Portuguese and Standard BP have the following clause structure: (1) CP C P AgrP Agr NegP Neg TP T VP She claims that two XPs are responsible for negation: NegP (Pollock 1989), where the negative markers are inserted, and P, similar to Laka (1990), where the negative markers check their negative features. T he negative marker is inserted in NegP and subsequently moves to P to check its strong negative feat ure. For Ma rtins, feature checking describes certain strong morpho-sy ntactic features on heads and phrases (like [+neg] on no ) that need to be checked against some matching feature elsewhere in the structure in order to license the derivation. This is done th rough movement of a lexical item that carries a feature to some position in the syntactic derivation where it can check that feature. Feature checking forces the item to move to a position that is structurally local, and the tree in (2) shows the movement of the negative marker in a European Portuguese or Standard BP sentence. 151

PAGE 152

(2) P AgrP noi Agr NegP Neg TP t1 For a positive sentence, Martins proposes movement of the verb, with a [+pos] feature, to to check the [+pos] feature on The subject moves to spec, P. All of this happens, according to Martins, because the ve rb needs to check its positive feature. (3) gives a positive declarative s entence where the verb moves to and then the subject moves to spec, P. (3) Joo viu a Lcia John saw DET Lcia John saw Lcia (4) P DP O Joo1 AgrP viu2 Spec Agr t1 Agr TP t2 a Lcia A negative version of the sentence has no moving from the inserted position, NegP, to an adjoined position in AgrP. From there, the negative marker moves to the head of P, and the subject also moves to spec, P. 152

PAGE 153

(5) Joo no viu a Lcia John NEG saw.3S DET Lcia John didnt see Lucia (6) P DP O Joo1 AgrP no2 t1 Agr Agr NegP Neg Agr Neg TP t2 viu3 t2 T VP t3 t1, t3 Lcia According to Martins, BP is different from European Portuguese and Standard BP. She claims that the phonological weak ening of the negative marker from no [nw] to num [nu], mentioned in chapter four, indicate s that there has been a loss of strong features, therefore num (the phonologically weak form) does not move to before the sentence is pronounced. The derivation of a BP sentence (5) is (7) where the negative marker does not move to Although she does not menti on this, she assumes that the verb in positive sentences still moves to as this is present in all derivations, and the subject likewise moves to the specifier posit ion. At LF the negative marker moves to to check its negative features. 153

PAGE 154

(7) P DP O Joo1 AgrP Spec Agr t1 Agr NegP Neg Agr Neg TP no viu3 t2 a Lcia Martins proposes a second instance of no that corresponds to the emphatic postverbal no which is what gives rise to T2. This second no is not in NegP, where the weakened no resides, but rather in some adverbi al position that is part of the VP shell. She claims the postverbal no is clause-final because of the Linear Correspondence Axiom (LCA) developed by Kayne (1994). The no needs to be in the lowest position that is c-commanded by ev erything else because the sentence precedes the clause-final no She claims that BP has a complex VP with several XPs or multiple VPs as Larson (1988) suggests seen in (8) (1997:34). 154

PAGE 155

(8) vP Subject v v VP Object V V VP Indirect Object V AdvP V No V That being the case, (9) has the structure in (10) where the CFN is an adverb adjoined to VP. 155

PAGE 156

(9) A Sienna no comeu o bolo no DET Sienna NEG eat.PAST.3S DET cake NEG Sienna didnt eat the cake. (10) P DP A Sienna AgrP Spec Agr t1 Agr NegP Neg TP no Spec T T vP comeu2 v VP NP V o bolo AdvP V no t2 While assuming that T1 and T2 are essent ially the same in function, the author notes that T3 is generally an answer to direct yes/no questions; therefore, it needs to be treated differently. (11) includes direct yes/no question and a T3 answer. (11) a. O Joo vai? DET John go.3S Is John going? b. Vai no go.3S NEG No 156

PAGE 157

Martins sees this type of structure as a topic/comment structur e where the verb is the topic and the negation is the comment ary on the topic (Barme 2000). To syntactically explain this, as well as to di fferentiate it from T2 in the syntax, she proposes movement of T to a topic phrase above CP. 24 According to Martins, the verb acts as the topic of the sentence in answe rs to BP direct questions. In this case, no is generated in NegP, but now it has a strong feat ure, and so moves to P to check its strong feature. (12) TopP Top CP vai1 C P AgrP no2 Agr NegP Neg TP t2 Spec T T vP t1 24 Martins wrote around the same time as Rizzi (1997) so the fact that the TopP is above CP might be merely the fact that Rizzis proposals hadnt made it to her paper. The fact that TopP is above CP means little for her analysis here. 157

PAGE 158

Martins makes three essentia l claims about CFN in BP: i) the negative marker has weakened (with regards to her claims about the JC) to the point that it needs reinforcement ii) a postverbal AdvP is generated to reinforce the weakened negative marker. This adjunct is a strong no, similar to other adverbials seen in languages that undergo the JC iii) T3 is a case of topicalization (maki ng it different from T2), where the verb moves to a topic position (TopP) and the negative marker moves to Martins distinction between T2 and T3 is important because it highlights the difference in the contexts that these general ly occur; nonetheless, she does run into some problems with her proposal. Recall that the two kinds of restrictions on BP CFN: semantic/pragmatic and syntactic. First, Ma rtins does not fully explain why in normal cases no has weak features and does not move to P, but in cases of T3, it does. Second, she assumes that T2 can occur in all syntactic contexts just like T1 and, consequently, her analysis does not explain why CFN is not allowed in wh-questions. With other emphatic markers in BP, the empathic marker is allowed in wh-questions. The question remains: why do wh-questions not permit her emphatic negative marker? Third, Martins analysis does not take into account the semantic/pr agmatic restrictions on CFN; they are not capt ured. T2 negation might be an instance of adverbial adjunction, but she does not explain why in some cases the postverbal no is not allowed depending on the di scourse conditions. 6.2.2 Fonseca 2004 Fonseca (2004) analyzes negation within a minimalist framework. Much like Martins, she assumes two projections that deal with negative markers, PolarityP (PolP) and NegP. PolP is similar to Martins P in that it is a functi onal projection with polarity 158

PAGE 159

values, i.e. negative or positive, and NegP is similar to that in Po llock (1989). Fonseca claims that features need to be check ed before Spell-Out, and the need to check features is what causes movement visible at PF. For her, weak features do not move at PF, but rather at LF. Fonseca claims that the BP preverbal no is weak in two regards. First, based on phonological weakening and the Jespersen Cycle, no is weak, it is a clitic that attaches to the verb (comparable to what happens in French), and it is generated in Neg. Second, and related to the first according to Fonseca, the clitic negative marker has weak negative features, akin to [uNEG] feat ures, that must be checked before SpellOut. To check its negative feature, t he negative marker moves from Neg to the specifier of PolP. According to her, the only way to do this for BP is TP movement to spec,PolP where it forms a spec-head rela tionship with another negative marker. The head of PolP is the second negative marker no that needs to form a checking relationship, but this marker has strong feat ures ([iNEG]), which allows it and the clitic negative marker to match their negative feat ures (Fonseca 2004:16). This is seen in the simple derivation below of sentence (14) While the preverbal negative marker is the element that carries the [uNEG] feature, the whole ver bal complex must move to match this feature because the negative marker clitizes to the verb. However, she extends this further and claims that TP has the negative feature, and this is the reason the entire clause to move to spec,PolP. (13) Eu num vi no. 1S NEG saw.1S NEG I didnt see (it) 159

PAGE 160

(14) PolP TP[uNEG] Pol Eu num-vii Pol TP no [iNEG] ti To account for T3 negation, Fonsec a posits deletion of the preverbal no based on Jespersen Cycle weakening, which means t hat T2 and T3 are essentially the same according to her. Fonsecas main claims include the idea that the preverbal marker is weakened and therefore is unable to negate a sentence by itself, again like in French, and the postverbal negative marker takes over as the semantically negative element in the sentence. Her main clai ms are summarized as follows: i) The preverbal marker is too weak to negate the sentence on its own, i.e. it has a [uNEG] feature. ii) The preverbal no cliticizes to the verb iii) A strong marker with [iNEG] feature is generated in Pol. iv) The entire TP moves to spec,PolP in order to check the negative feature in TP I believe that Fonseca is right in trying to align BP with what is known about other Romance languages in that she claims that the preverbal marker is weak and unable to negate a sentence on its own. However, her proposal runs into technical and theoretical problems and inconsistencies. Syntactically, she claims that the weak marker moves at LF and that the same spac e reserved for LF movement is occupied by the postverbal no More important however, is the cl aim that the preverbal marker is weak. This on its own is difficult to def end because all the corpus studies related to negation in BP show that T1, preverbal only, is the overwhelming choice for negation. 160

PAGE 161

Even though there are pr oblems with the syntactic development, a larger problem is her lack of pragmatic/semantic discussion. S he does not give any explanation of the pragmatic restrictions on T2 and T3. Ho wever, she does identify the need to have movement of a large part of the structure to a higher position in the tree. Although she doesnt mention the syntactic environments that block CFN, namely wh-questions and relative clauses, Fonsecas structure does block the A-movements involved in these constructions. The TP movement that she argues for creates an island out of which additional movement is prohibite d. This last observation, although not made directly by Fonseca, is part of the analysis I ultimately propose for BP CFN. 6.2.3 Cavalcante 2007 More recently, Cavalcante (2007) continues with Martins claim that T2 and T3 are syntactically different. Cavalcante uses se veral different corpus studies to determine where all three types occur. From there, within a minimalist framework, he proposes structural positions for the postverbal no The author determines, based on the different corpus studies, that T2 occurs in main clauses and in embedded clauses, and T3 cannot occur in embedded clauses or with topic constructions. To account for the different types of CFN in BP, he proposes the use of a new type of XP, a Denial Phrase (DenP). For Cavalcante, DenP accounts for T2 and T3 in main clauses, but it doesnt resolve the issue of T2 in embedded clauses. He proposes that for T2 in embedded clauses, there is no that adjoins to TP following whic h TP moves to the specifier of TopP. I explain his mechanisms below. Following Rizzis (1997) idea of an expanded CP, Cavalcante positions DenP above ForceP and assumes NegP to be below TP and above VP. His structure for BP is (15) 161

PAGE 162

(15) DenP ForceP FinP IP/TP NegP -VP Crucial for Cavalcantes proposal is the existence of a Denial Phrase above ForceP. The head of DenP is filled by an abs tract phonologically null element that has some denial feature check ed by the insertion of no in to the specifier. He explains that DenP is a category responsible for codifyi ng discourse information that refers to a confirmation or a denial of a previous presupposition (Cav alcante 2007:129). 25 He claims that no in the following sentences is in DenP. No in this sentence denies the previous presupposition. (16) A. E mora aqui ou a senhora mora em tro lug? And lives here or DET lady lives PREP other place And does she live here or does she live in another place B. No ela mora... ela morava... depois desse home da... NEG, 3S.F lives 3S.F live.IMP after PREP.DEM man PREP.ADV No she livesshe used to liveafte r that man from over there Because DenP is above CP, he states that DenP cannot be selected which means, according to him, that DenP does not have scope over the sentence, and therefore, it does not change the polarity of the sentence (2007:129). As DenP is above ForceP, T3 does not occur in em bedded clauses because then the negative marker would be outside the clause. He c ontinues by stating that DenP does not change the polarity of a sentenc e, but rather denies the pr evious presupposition, which could be positive or negative. This means that in the case of T2, there is no semantic 25 DenP a categoria que aloja as profrases afirmati vas e negativas (como SIM, e NO) utilizadas em contextos de resposta a uma pergunta direta ou de assentimento ou denegao de uma declarao realizada anteriormente (Cavalcante 2007:129) DenP is a category that houses afirmative and negative prophrases (such as YES, IS, and NO) used in the context where they are the response to a direct question or assertion, or denial of a previous declaration. 162

PAGE 163

confusion because the postverbal negative marker is there only to confirm the presupposition. The wording of his definitio n prevents a double negative interpretation in T2 contexts. Cavalcante gives sentence (17) as an example of T2 in the main clause. To generate a typical T2 sentence, BP uses a DenP to insert the denial marker no as the specifier. Recall that the Den is filled by what Cavalcante calls an abstract element. Above the DenP is a TopicP, also outside of CP. His TopP above CP is also headed by an abstract head that is phonologically null. TP moves to the specifier of TopP where TP checks some topic feature with the abstract element in the head (Cavalcante 2007:149). Note that Cavalcante does not mention what a sentence like (17) would mean. His only claim about T2 is that the final no does not change the polarity of the sentence. Based on his comments about T3, given with the next set of examples, I assume that a sentence like this means: I confirm the presupposition that I didnt buy the house. (17) Eu num comprei a casa no 1S NEG bought.1S DET house NEG I didnt buy the house (18) TopP TP [+top] Top Eu num comprei a casai Top DenP no Den Den CP C TP t1 163

PAGE 164

Different from T2, DenP does not confirm a phrase in T3, but rather denies it. Specifically he claims that final negation in [V no] as a predication that does not have the function of altering the trut h value of a proposition, but to express a denial of the presupposition (Cavalcante 2007:140). 26 Thus, in T3 sentences, NegP, which hosts preverbal negation, is not present. Under this assumption, the clause-final negative marker denies a presupposition. Example (19) means something like I bought the house. That is not a true statement, which is different from what he claims for T2. (19) Eu comprei a casa no 1S bought.1S DET house NEG I didnt buy the house Cavalcante claims that DenP and his der ivation are able to account for all instances of negation in BP except for T2 in embedded clauses, because DenP is outside the main clause, i.e. above it. Reca ll, Cavalcante states that the DenP cannot be selected, and therefore according to him, is unavailable for embedded clauses, and this rules out T3 in embedded clauses (Cavalcante 2007:141). (20) ?*Ele disse que conseguiu no 3S said.3S that able.to.PST.3S NEG He said that he wasnt able to. To get around the problem of T2 negati on in embedded clauses, he suggests that a second no adjoins above TP and below FinP and then the TP raises to Topic 26 Isso implica que, at a ativao do nvel DenP, ac ima de ForceP, o sistema computacional realiza uma derivao em tudo semelhante de uma sentena afirmativa, o que compatvel com a intuio de que a sentena funciona como um tpico e a negao final em [V no] como uma predicao que no tem a funo de alterar o valor de verdade da pro posio, mas expressar uma denegao sobre uma pressuposio. (C avalcante 2007:140) 164

PAGE 165

Phrases above FinP (Cavalcante 2007:149). He never explains why this mechanism is not available in main clauses for T2. (21) ForceP Force TopP que TP[+neg]i TopP pro num viajou Top FinP Fin TP no TP t1 Cavalcantes study is helpful because he begins by systematically looking at where each type occurs. As such, he is able to separate out the occurrences of T2 and T3, and identify, through corpus studies, the syn tactic situations that allow these types of negation. By doing this, he concludes that BP syntax generates T2 and T3 in the same manner in main clauses. Cavalcante al so recognizes the fact that T2 and T3 seem to be linked to discourse, and so he developed DenP, which is a projection high in the derivation. His claims have their problems, however. Namely, his introduction of DenialP to the derivation is difficult to follo w. He explains what it means to be a denial and not a negation, but is not consistent in usi ng his own definition. In the case of T2, he states that the denial acts as affirmation of a negative sentence. In contrast, DenP is used as a denial of a positive sentence for T3. So while these posit ions, lexical items, 165

PAGE 166

and discourse functions are the same, the semantic meaning changes to the point of being opposites. Additionally the evidence that DenP cannot be selected is based solely on T3 data, i.e. T3 cannot be in em bedded clauses. Stating that DenP is not selected does not seem to be compatible with what he states about it not having scope. Generally scope is seen to be related to syntact ic hierarchy. As DenP is higher up in the derivation, it would then have scope over the whole sentence, contrary to what Cavalcante claims. The fact that T2 has two different manife stations is also problematic. Cavalcante never answers the questions: what does the CFN contribute in embedded clauses? Does it have the same affirmation of negative presupposition? If the answer is yes, why doesnt this no appear in the denial phrase as is the case in main clauses. If the answer is no, what semantic information does CFN carry, if any, in embedded clauses. By separating T2 into two different derivations he causes confusion. Finally, Cavalcante argues that topics are not licit in T3, but are in T2. His argument suffers because he also claims that T2 and T3 are syntactically identical in main clauses. This is a problem for his analysis, which he admits he cannot solve (2007:144). 6.3 A Proposal for Clause-Final Negation in Brazilian Portuguese My proposal builds on these previous analyses coupled with the information presented in chapter five. I assume like Ca valcante that T2 and T3 are related to the discourse (Schwenter 2005). Also, I follow Martins and Cavalcante in that T3 sentences should involve topicalization, although I exte nd Martins analysis to include topicalization of propositions and not just the verb, as in Cavalcantes analysis. The principal advantages of my analysis over the previously discussed ones are: 166

PAGE 167

i) All of the syntactic restrictions are accounted for ii) It provides a unified a ccount of T2, which differs from Cavalcante who separates T2 into main and embedded clauses, and Martins who also suggests topicalization only for T3 iii) The semantics of T2 and T3 are dealt with in a way that permits both types to occur motivated by the same semantic/pragmatic theory iv) It provides a more complete analysi s of the semantic values of both negative markers in relationship to n-words To support these claims, I give a structural analysis that is in line with what I accept from Schwenter (2005) as presented in section 5.2.3 and I account for the facts presented in chapter one. As a reminder, I present below a list of the syntactic environments relevant to CFN and whether they are grammatical or not. This table assumes that the discourse requirements are met. A represents grammatical, and X represents ungrammatical. For T2 under Allo wed in Object Relative Clauses, I put *, which refers to the obligatory sentence final position of the postverbal negative marker; however, half of my consultants indicate that these sentences are ungrammatical. Nonetheless, similar to previous author s, I treat them as grammatical. 167

PAGE 168

Table 6-1. Syntactic Restrictions for BP CFN Environment T1 T2 T3 Discourse Requirements X Allowed in Relative Clauses in Subject Position X X Allowed in Relative Clauses in Object Position X Allowed in Fronted Wh-questions X X Allowed in in-situ Wh-questions X X Allowed with Subject N-words X X Allowed with Object N-words X Licenses NPI Idiomatic Expressions X Allowed in Embedded Clauses Allowed in Imperatives Compatible with Topicalization The structural analysis specifically addresses the ban on T2 and T3 with wh-questions and relative clauses. It also addresses the differences between T2 and T3 in their interpretation and their licensing of NPI idioma tic expressions. Finally, these structures need to be checked against the facts related to n-words, which I discuss in chapter seven. The syntactic proposal is t hat there is an XP below ForceP and above AgrP which houses a clause-final no No is the head of this phrase, and I call it a Top(ic) Phrase. This no ends up in a clause-final position when the entire clause, AgrP, moves to the specifier of TopP. In the case of T3, t he head has an [iNeg] feat ure that negates the sentence it is in. With T2, this head has an [uNeg] feature that must be licensed by another negative element in the sentence. 168

PAGE 169

The rest of this section is divided as follo ws. First I give the derivations of basic clauses containing CFN. From there, I show how my proposal accounts for the data represented by the table above. Principally, I discuss wh-w ords, relative clauses, topicalization, and imperatives. I hold off on n-words and discuss these in chapter seven. Section 6.3.7 ends by explaining the semantic differences between T2 and T3, and this section includes data regarding the lic ensing of idiomatic expressions by CFN. 6.3.1 Derivation of Clause-Final Negation I argue that T3 is formed by moving AgrP to the specifier of a lower TopP. This relates to the discussion in section 5.2.3 where I claimed that CFN is used when the proposition is discourse-old. It is this proposition that is being moved. Recall also, as laid out in chapter two, the use of strong features to drive movement. If an uninterpretable feature [uF*] is strong, it must be checked by an interpretable instance of the feature in a spec-head configuration. If [uF] is weak, the feature can check against [iF] if it c-commands [iF]. In BPs CFN sentence, the head of TopP is no, which is a topic marker that has a [iNEG] feature and a strong [uTOP*] feature. Th is TopP is the lower topic phrase briefly introduced in chapter two. The reason that this must be a lower topic phrase will be discussed in section 6.3.5 I assume that in derivati ons involving CFN, AgrP has a [iTOP] feature that checks the [uTOP*] feat ure on Top. AgrP ha s the [iTOP] feature because this is the discourse-old proposition. In the derivation, AgrP[iTOP] moves to the specifier of TopP to satisf y the strong feature [uTOP*] on no in Top. (22) gives a generic blue-print of a T3 structure, and (23) gives a T3 sentence and its derivation. 169

PAGE 170

(22) ForceP Force TopP AgrPi Top Top[uTOP*,iNEG] AgrP[iTOP]i no (23) a. Gosto dele no Like.1S PREP.3S NEG I dont like him b. ForceP Force TopP Top Top [uTOP*,iNEG] AgrP [iTOP] no gosto dele c. ForceP Force TopP AgrP[iTOP]i Top gosto dele Top[uTOP*,iNEG] AgrP no t i The basic idea, in line with t he discourse-old hypothesis, is that if an AgrP contains discourse-old information then it can be topicalized in BP. By stating this, I am claiming that CFN is a case of proposition topica lization, where the topic marker has two features, a polarity f eature that can be either negative [NEG] or positive [POS], and a strong uninterpretable topic f eature [uTOP*] that triggers movement. The polarity 170

PAGE 171

feature on the discourse-old topic marker is not only seen with negative sentences, but also with positive sentences that make use of sim yes. The same type of discourse requirements apply for the positive marker as they do for the negative marker. If the proposition is not discourse-old then it is infelicitous to use sim, as in (24) Note that there is no pause preceding sim (25) would then take the entire sentence as the topic and move it to the specifier of this topic phrase with sim as the head. A full translation would be something like as fo r me visiting the beaches of Rio that is a true statement. (24) Visitei as praias (#sim) Visit.PST.1S DET beaches (yes) I visited the beaches (25) a. Voc vistou as praias do Rio? 2S visit.PST.2S DET beaches PREP.DET Rio Did you visit the beaches of Rio b. Visitei as praias do Rio sim. Visit.PST.1S DET beaches PREP.DET Rio yes I visited the beaches (26) ForceP Force TopP Top Top[uTOP*,iPOS] AgrP[iTOP] sim visitei as praias do Rio (27) ForceP Force TopP AgrP[iTOP]i Top visitei as praias do Rio Top[uTOP*, iPOS] AgrP sim ti 171

PAGE 172

This sim is only allowed in contexts where the proposition is the topic, as in visit the beaches of Rio. If this proposition is not th e topic, but a focused or new element, then the sentence becomes infelicitous with sim or no Syntactically this means that the feature [iTOP] is not present on AgrP and therefore there is nothing to check the uninterpretable [uTOP*] on Top, which can then not be present. T2 is a combination of T1, discussed in chapter four, and t he movement operation seen in T3. The preverbal marker is generat ed in its usual position, Neg, and the CFN marker is in Top. Unlike with T3, however, the CFN marker in this case does not have a [iNEG] feature but rather a [uNE G] feature. This means that the CFN marker in T2 is not sufficient to negate a sentence on its own (for a similar claim about Afrikaans see Biberauer 2009). I discuss the semantic iss ues related to the two negative markers further in section 6.3.7 Because the feature is uninterpretable, it must be checked by a semantically contentful [iNEG] elsewhere in the structure. The movem ent of AgrP is still motivated by a strong featur e [uTOP*] on the CFN marker. A gain, I point out here that AgrP has a [iTOP] feature, but additionally, it also contains the [iNEG] feature from the preverbal negative marker. I propose that [iNEG] percolates from Neg to AgrP because the proposition as a whole is negative. It is this instance of [iNEG] that will check the uninterpretable [uNEG] feature on Top. (28) is a basic blue-print of a clause with T2: 172

PAGE 173

(28) ForceP Force TopP AgrP[iTOP, iNEG]i Top subject Agr Top[uTOP*,uNEG] AgrP no ti Agr NegP Neg[iNEG] TP no T T VP verb The blue-print shows that after movement, AgrP c-commands the CFN marker. This position is what allows the [uNEG] feature to be checked. As chapter seven will show, for negative elements in BP to check their fe atures, the [iNEG] feature must c-command the [uNEG] feature (Zeijlstra 2004, 2008). Thus, feature checking of the negative features must take place after Spell-Out and cannot take place prior to movement or under reconstruction. Example (29) is a sentence that contains T2: Top is filled by no the discourse requirement is met, and AgrP moves to spec,TopP. (29) a. Ela no viu os meninos no 3S NEG see.PST.3S DET kids NEG She didnt see the kids 173

PAGE 174

b. ForceP Force TopP Top Top[uTOP*,uNEG] AgrP[iTOP,iNEG] no Ela Agr Agr NegP Neg TP no T T VP viu ...os meninos c. ForceP Force TopP AgrP[iTOP,iNEG]i Top Ela Agr Top[uTOP*,uNEG] ti no Agr NegP Neg[iNEG] TP no T T VP viu os meninos Now that the derivations hav e been given, I explain how my proposal accounts for the syntactic restrictions observed in BP CFN. 174

PAGE 175

6.3.2 Wh-Words The analysis I propose prevents T2 and T3 with wh-questions. I first discuss fronted wh-questions and then in-situ wh-que stions. Fronted wh-words with CFN are ungrammatical because my proposal has AgrP mo ve to the specifier of the lower TopP, which creates an island (Ross 1967, Szabolcsi 2006). Specifically, the ungrammaticality of CFN in sentences with wh-words follows from Huangs (1982) Condition on Extraction Domain (CED) defined in 0 Properly governed means governed by a lexical head. (30) CED A phrase A may be extracted out of a domain B only if B is properly governed The CED means that extraction out of co mplements is allowed, but not out of adjuncts and phrases in specifier positions because these positions are not properly governed. The movement of AgrP to the s pecifier position of TopicP thus creates a specifier island and the wh-phrase inside of AgrP cannot be extracted to a higher position. Example (31) shows a sentence that is ungrammatical and (32) shows the derivation. Notice that the moved AgrP creates a specifier island that prevents extraction to the landing site for wh-words which is spec,FocusP, above TopP. The bolded, fronted wh-word is co-indexed with the trace that it leaves behind in the fronted AgrP. (31) *O que que voc (no) visitou no? What that 2S (NEG) visited NEG What didnt you visit? 175

PAGE 176

(32) *FocP O quei Foc Foc TopP que AgrP[iTOP]ii TopP voc Agr Top[uTOP*] tii no Agr NegP Neg[iNEG] TP no T T VP visitou ti This analysis thus correctly predicts that T2 and T3 are incompatible with whmovement. Chapter 3 showed that wh-in-situ does not obey islands and does not involve movement: (33) Voc conversou com o autor [que escreveu que livro]? 2S talk.PST.2S PREP DET author that write.PST.3S which book Which book did you talk to the author who wrote? It was also pointed out that wh-in-situ in BP is constrained by semantic/pragmatic restrictions. BP wh-in-situ is only allowed if certain discourse requirements are met. Much like Schwenters (2005) proposal for BP CFN, Pires & Taylor (2007) proposed the use of discourse to understand syntax. They claimed that the in-situ situations are licensed by information in the Common Ground. They define Common Ground as 176

PAGE 177

information that was previously given in the discourse or in the extralinguistic contextand which is shared (or assumed by the speaker to be shared) by speaker and hearer (2007:5). This means that in-s itu wh-questions are only felicitous when the proposition being questioned is discourse-old. For example, in a sentence such as, I made dessert, the discourse-old proposition is I made X. From this one could question: (34) Voc fez que tipo de sobremesa? 2S do.PST.3S what type PREP dessert You made what kind of dessert? The felicity changes when what is being questioned is not discourse-old, as in (35) The example is grammatical, but infelici tous. It can be made felicitous by fronting the wh-word. [ You approach a colleague at work and ask, out of the blue :] (35) #Voc conhece quem em So Paulo? 2S know.2S who PREP So Paulo You know who in So Paulo? The following exchanges shows how the disc ourse requirements a ffect the felicity of the sentence. With the appropriate c ontext, the above example becomes felicitous: [ You approach a colleague at work and co mment A, to which the colleague responds with B :] (36) A. Eu visitei So Paulo este fim de semana 1S visit.PST.1S So Paulo DET end PREP week I visited So Paulo this weekend 177

PAGE 178

B. Voc conhece quem em So Paulo? 2S know.2S who PREP So Paulo You know who in So Paulo? The example below shows that the discourse restriction on the use of CFN is similar. (37) A. Voc conheceu algum em So Paulo este fim de semana 2s know.PST.2S somebody prep So Paulo DET end PREP week Did you meet anybody in So Paulo this weekend B. Eu no conheci ningum em S o Paulo este fim de semana no. 1S NEG know.PST.1S nobody PREP So Paulo DET end PREP week NEG I didnt meet anybody in So Paulo this week end The similarities between the discourse re strictions on wh-in-situ and CFN suggest that these two constructions make use of the same mechanisms and by extension the same structural position. I propose that CFN and wh-in-situ make use of a single syntactic position in the left periphery, the specifier of the lower TopP. I previously argued that there is no movement for wh-in-si tu; rather, there is a Q operator linked to the discourse in a similar way that the CFN is linked. I propose that there is complementary distribution between AgrP topicalization and the Q operator because both are located in the specifier of the lo wer TopP, which I called Top2P in chapter two. 27 There can be one or the other, but not both. Thus, CFN and wh-in-situ are incompatible. 6.3.3 Relative Clauses BP prohibits relative clauses containing T2 and T3 with the exception that T2 is possible in relative clauses if the relative clause is in a sentence final position. The derivation argued for here does ban the use of CFN in relatives, and this makes relative 27 This is an updating of the analysis in chapter 3 wh ere I tentatively proposed t hat the Q operator was in the head position of the higher Top1P. This change does not affect the results. 178

PAGE 179

clauses in object position with T2 a mystery. Before add ressing the problem with object relatives, I give a current a ccount of relative clauses in BP and show that the analysis in this chapter account for the ban on relative clauses with CFN. Lessa de Oliveira (2008) and Nunes & Kato (2009) give an account of BP relative clauses within a minimalist framework where they propose a raising analysis for relative clauses 28 (Kayne 1994). A structure based on their analysis is: (38) a pessoa que o professor viu DET person that DET professor see.PST.3S The person that t he professor saw (39) DP a ForceP DP ForceP pessoal DPk Force AgrP que t1 o professor viu tk Again, much like wh-phrases, relative claus es involve a type of movement that is blocked once AgrP moves to spec,TopP with CFN. This is seen in example (41) AgrP moves to a specifier position which gives rise to CFN. The DP pessoa cannot be extracted out of this AgrP due to the same CED violation seen in wh-movement. (40) a pessoa que o professor no viu (*no) DET person that DET professor NEG see.PST.3S NEG the person that the professor didnt see 28 Both works also deal with non-standard and free rela tive clauses, but the idea that there is raising out CP to a higher projection is present. As such, I onl y present what they call standard relative clauses. 179

PAGE 180

(41) *DP a ForceP DP ForceP pessoal DPk Force TopP que t1 AgrPm Top o professo no viu tk no tm The prediction that CFN in impossible in relative clauses is largely correct. The table introduced earlier showed that T3 is comp letely impossible in relative clauses and T2 is impossible, except if the relative clause happens to be at the end of the clause: (42) Todo mundo gosta de morangos All world like.3S PREP strawberries Everybody like strawberries (43) Eu ensinei portugus para uma pessoa que no gosta de morangos no 1S taught.1S Portuguese PREP DET person that NEG likes PREP strawberries NEG I taught Portuguese to a person that doesnt like strawberries I do not have an explanation for the why th is example is grammatical. The theory introduced above predicts that it should be not be. Note that if the object is not the last element in the sentence, then the sent ence becomes ungrammatical. I leave the explanation of this cont rast for future work. (44) *Eu ensinei portugus para uma pessoa que no gosta 1S taught.1S Portuguese PREP DET person that NEG likes de morangos no ontem PREP strawberries NEG yesterday I taught Portuguese yesterday to a per son that doesnt like strawberries 180

PAGE 181

6.3.4 Embedded Clauses Others before this work have claimed that T3 does not exist in embedded clauses (Cavalcante 2007). Recall that in chapter one I have shown that T3 does exist in embedded clauses if the discourse requirement is met. The problem here is that although T3 can exist in embedded clauses, there are no cases of them occurring in natural speech. The sentences that I pres ented were created in conjunction with a scenario that would permit the discourse-old information to be in the embedded clause. After creating such a scenario, my consultants generally agreed that T3 is allowed in embedded clauses. This situation points to a possible difference between T2 and T3. As stated in section 5.2.3 T3 is restricted to directly activated discourse-information. This seems difficult to occur naturally, and generally if there is CFN in an embedded clause, it is T2. Although not naturally occurri ng, T3 can exist, and my propos al predicts that this structure is syntactically possible, because nothing prevents topics in embedded clauses. Below is an example where the discourse requirement is met and the sentence is grammatical. As expected fr om the discussion, AgrP moves to a Topic position, and negative marker is in the clause-final position. (45) Eu acho que ele terminou o exame no 1S think that 3S finished DET test NEG I think that he didn t finish the test 181

PAGE 182

(46) IP Eu I I VP acho V ForceP Force TopP que AgrP[iTOP]ii Top ele Agr Top tii no[uTOP*] Agr VP terminou o exame 6.3.5 Topicalization Another area in which this dissertation diffe rs from previous analyses is in allowing topic constructions with CFN. In chapter 3 I argued that topics in BP are base-generated in the specifier of the higher spec,Top1P and co-indexed with a resumptive pronoun, null or overt. Different from wh-movemen t, topic fronting with CFN is grammatical because no movement is involved. This require s that at least two topic positions exist in BP: a lower one used for CFN and a higher one for typical topicalization. The following examples show the derivation of a sentenc e that contains both a topicalized DP and CFN. In this case, the CED is not violat ed because there is no mo vement of the topic out of AgrP. (47) A Eliza, eu no vi no DET Eliza, 1S NEG saw.1S NEG As for Eliza, I didnt she her 182

PAGE 183

(48) Top1P A Eliza i Top1 Top1 Top2P AgrP[iTOP]ii Top2 eu Agr Top2 tii no[uTOP*] Agr TP no vi pro i Separation of the two kinds of topic phrases, Top1P versus Top2P, is necessary to avoid generating ungrammatical word orders. In particular, the position for basegenerated topics must be above the position to which AgrP moves in order avoid getting the two topics in the wrong order: (49) *Eu no vi no a Eliza 1S NEG saw.1S NEG DET Eliza As for Eliza, I didnt see her Rizzi proposed that there is potentially more than one topic position; however, the above structure does not instantiate TopP recursion in the sense he describes. Rather, different kinds of t opics must be ordered in the left periphery (Beninc & Poletto 2004). In the case of BP, the data shows that fronted Ag rPs are lower topics and basegenerated DPs are higher topics. I have encoded this observation by simply labeling the two kinds of topic phrases as Top1P and Top2P. Even before this, Aissen (1992) argued for an ordering of topics in Mayan Tzotzil, and this has been shown to work in other Mayan languages (Hansen 2005). As seen earlier, thes e topic phrases are also separated by FocusP. 183

PAGE 184

6.3.6 Imperatives BP allows all three types of negation in impe ratives. Two verb forms are used in BP (negative) imperatives: subjunctive or bar e verb (Scherre 2007). I repeat some examples from chapter one. Dialectal di fferences exist between the use of the two types of imperatives; however, this does not influence the outcome, and as such I use the bare verb form. A. Bare Verb B. Subjunctive (50) No Faz isso No faa isso T1 NEG do that NEG do.SUBJ that (51) No faz isso no No faa isso no T2 NEG do that NEG NEG do.SUBJ that NEG (52) Faz isso no Faa isso no T3 Do that NEG do.SUBJ that NEG Dont do that Dont do that I only illustrate imperatives with T2, but t he same results apply to T3. The same discourse restrictions apply, and the next examples show this. This first example shows that inferred discourse information is allowed, as seen in declarative sentences. B is the derivation, again similar to declaratives. [A mother to a young daughter who is wr iting on the walls with new crayons] (53) A. No faz isso no NEG do that NEG Dont do that 184

PAGE 185

B. ForceP Force TopP AgrPi Top No faz isso Top AgrP no ti If there is nothing in the conversation that triggers the discourse-old information then CFN is not licit (Schwenter 2005). (54) Ano que vem a gente vai viajar a Salvador Year that comes DET people go travel.INF PREP Salvador Next year we are goi ng travel to Salvador (55) Quando vocs chegarem l, no vo ao Restaurante Irina (#no) When 2P arrive.FUT.SUBJ there NEG GO.2P PREP.DET Restaurant Irina NEG When yall get ther e, dont go to Restaurant Irina The syntax of BP imperatives is similar to what has been proposed for the Slavic languages (Han 2001). Unlike other Romance languages which are assumed to have Ito-C movement in imperative clauses (Rivero & Terzi 1995, Han 2001, Zeijlstra 2006), BP verbs do not move in imperative clauses, and this is based on the evidence from object clitics. In Spanish, the object clitic is to the right of t he verb with positive imperatives (57) unlike a declarative sentence, when it is to the left (56) (56) Maria lo ley. Declarative Maria CL.M READ.3SG Mary read it (57) Leelo! Imperative Read.2S.IMP-CL.M Read it! In order for the verb to be on t he left of the clitic, it must mo ve to some position higher in the tree that deals with (imperative) mood, generally cons idered to be C. BP clitics are 185

PAGE 186

always to the left of the verb suggesting that there is no movement of the verb to a higher spot in the derivation. (58) A Maria me d o livro. Declarative DET Maria CL.1S gives DET book Mary gives me the book (59) A. Me d! Imperative CL.1S gives Give me! B. Me d ele! Imperative CL.1S gives Give me it! Because the verb stays in its position in IP, as in a normal declarative, CFN is licit in imperative sentences. 6.3.7 Semantic Interpretation of Clause-Final Negation The previous sections showed how the cu rrent proposal accounts for the syntactic restrictions related to CFN, and in most ca ses, T2 and T3 are identical. They differ syntactically in two areas: negative concord and idiomatic expressions. Negative concord is the topic of chapt er seven and is treated there. Here I discuss idiomatic expressions and how they further our understanding of the sem antic interpretation of T2 and T3. Before discussing idiomatic expres sions, I explain the theoretical problems related to the interpretation of sentences c ontaining CFN. From there, I show how idioms help delineate the differences betwe en T2 and T3, but how this separation creates a problem related to double negation. T1 is the unmarked strategy and T3 is the marked strategy, and in both cases, there is only one negative marker for the sent ence. A problem arises with T2, as this structure contains two negative markers, leading researchers to question how these 186

PAGE 187

sentences do not evoke a double negative reading (Schwegler 1991a, da Cunha 2007). One claim is that BP is traveling thr ough the Jespersen Cycle, and the preverbal negative marker no longer has a negative feature strong enough to negate a sentence on its own. It is well documented fact that negative markers lose their negative force. However, in the case of BP, this seems unlik ely because in all corpus studies T1 is still the main method of negation (Cavalcante 2007) suggesting that the preverbal has not lost its semantically strong negative feature. Another attempt to understand the lack of double negation in T2 comes from Cavalcante (2007). First, he a ttempts to explain T3 by cl aiming the clause-final is a denial of the asserted positive sentence. This line of thinking follows what Horn has called metalinguistic negation (Horn 1989), and Cavalcante takes the topic no as some sort of metalinguistic negation. Metalingu istic negation usually suggests the denial of an assertion or the refusal to accept as asse rtion. The classic example reused by Horn (1989) is The king of France is not bald (because) there is no king of France. Essentially, negation in this sense is not about a part of the sentence but rather the whole utterance, i.e. the king is not bald not because he doesnt have hair, but he is not bald because he doesnt exist, and therefore c annot be either bald nor have a head full of hair. While not exactly like metalinguistic negation as defined by Horn, this type of interpretation comes close to what T3 means. Example (60) explains what Cavalcante means when he uses denial in BP CFN. This sentence has T3 CFN, and the positive sentence is John ran two miles. If we take the CFN to be a denial, the meaning of this sentence is: John ran two miles, that is not true, or John ran two miles, I deny that fact. 187

PAGE 188

(60) O Joo correu duas milhas no DET Joo ran.3S two miles NEG John didnt run two miles. The major problem with this hypothesis, as noted by Cavalcante himself, comes with T2 and its meaning. If T3 is metalinguistic negation, where the utterance as a whole is negated or denied then his analysis leads to an incorrect interpretation. Sentence (61) a involves T2, and (b) translates the sentence to a possible interpretation. Notice that the possible interpretation gives a double negative, meaning that John did run two miles, when (a) clearly states that he didnt. (61) a. O Joo no correu duas milhas no DET Joo NEG ran.3S two miles NEG John didnt run two miles. b. John didnt run two miles. That is not true (that John didnt run 2 miles) Though Cavalcante mentions this fact, he doesnt pursue a solution. To explain this, I propose taking T2 to be a case of negative spread (Den Besten 1986, Giannakidou 2006). Negative spread is where the negative feature of some items, like n-words and negative markers, spreads over t he whole sentence. For example, in nonstrict negative concord languages, the negativ e marker must c-command the n-word, unless the n-word is in t he subject position, as in (62) (see chapter seven for more discussion). (62) O Ablio no ajudou ningum DET Ablio NEG helped nobody Abllio didnt help anyone 188

PAGE 189

However, the n-word can be c-commanded by another n-word. (63) Ningum ajudou ningum nobody helped nobody Nobody helped anyone These examples illustrate the i dea that the features spread to the rest of the sentence without causing a double negation reading. T2 seems to be a situation similar to this, where the first negative marker spreads its negative meaning to the rest of the negative items in the sentence, because these items are not by themse lves inherently negative. Thus in the previous sections I claimed that the CFN maker in T2 carries an [uNEG] feature and must be c-commanded by some element with a [iNEG] feature. In the case of CFN AgrP carries the [iNE G] feature as the feature is able to percolate up from the preverbal negative marker. I continue with this line of r easoning in chapter seven, and now I focus on NPI idioma tic expressions and T3. Biberauer & Cyrino (2009) notes that T3 does not license NPI idiomatic expressions whereas T2 does, and I believe t hat this observation is related to the semantic meaning of the negative markers. This is important because it is one area that demonstrates that T1 and T3 are not only different in terms of discourse, but also in terms of inherent semantic m eaning and semantic contribution to a sentence. This also shows that the semantics of T3 and T2 are not equal. They give the following example. A utters the sentence John is rich. B disagrees with A and uses an NPI idiom to express the fact that John is not rich. The expression in Portuguese literally translates to He doesnt have a cent/penny with a hole in it which means he doesnt have any money. The translation that Biberauer & Cyrino give is he doesnt have a red-cent. 189

PAGE 190

Notice that in the case of T1, B, the s entence maintains its idiomatic meaning. The same can be said about T2, B, but in B, t he situation is different because of the use of T3. In the case of B, the figurative meaning is lost, and onl y a literal meaning remains. Notice C, a sentence containing T3, and how it w ould most likely have to finish for it to be grammatical in this situation, albei t perhaps not pragmatically felicitous. (64) A. O Joo rico! DET John is rich John is rich! B. O que? ele no tem um tosto furado! T1 what 3S NEG has DET cent with-a-hole What? He doesnt have red-cent (i.e. he doesnt have anything) B. O que? ele no tem um tosto furado no! T2 what 3S NEG has DET cent with-a-hole NEG What? He doesnt have a red-cent(i.e. he doesnt have anything) B. *O que? ele tem um tosto furado no! T3 what 3S has DET cent with-a-hole NEG What? He doesnt have a red-cen t. (i.e. he doesn t have anything) C. Ele tem um tosto furado no, ele tem um inteiro! T3 3S has DET cent with-a-hole NEG, 3S has DET whole He doesnt have a red cent; he has a BL UE one (i.e. literal meaning) The fact that T3 does not license NPI idioms in the way that T1/2 do suggests that it is different. For it to negate a sentence on its own, as T1 does, it needs to have an interpretable negative feature, which I pr opose it does. Like T1, T3 has a [iNEG] feature, but in the case of NPI idioms the [iNEG] feature is apparently not in an appropriate structure position to license t he idiom chunk. Thus, the idiom loses its idiomatic meaning and has only a literal meaning. As discussed above, negative elements with an [uNEG] featur e must be licensed by a negat ive element with an [iNEG] 190

PAGE 191

feature. I follow Zeiljstra 2004, 2008 and assume that [iNEG] must c-command [uNEG] in order for [uNEG] to be checked. Since in the case of T3, the NPI idiom is not ccommanded by the CFN marker, the NPI interpreta tion is not available. This state of affairs is seen below: (65) *TopP AgrPi Top DP Agr Top AgrP no[iNEG] ti Ele Agr TP T VP tem um tosto furado[uNEG] T2 two is different because the licensing of the NPI is done by the preverbal negative marker, which has an [iNEG] feature. The CFN marker in Top is irrelevant for the licensing of the NPI but is itself licens ed by the same interpretable negative feature on AgrP. In the following structure, [iNEG] c-commands and checks both instances of [uNEG]. (66) TopP AgrP[iNEG]i Top DP Agr Top AgrP no[uNEG] ti Ele Agr NegP Neg[iNEG] TP no T VP tem um tosto furado[uNEG] 191

PAGE 192

Because of the facts presented in this se ction, I argue that T3 and T2 participate in the same derivation but differ in their morphosyntactic features. The no in T2 is a negative item that is similar to an n-word and cannot negate a sentence on its on; it is [uNEG]. The no in T3 is a negative item that c an negate a sentence on its ownit is [iNEG]but is highly restricted to a sub-set of discourse linked information and is not in a position to license NPI idioms. These cl aims and the claims about the [NEG] feature that the negative markers have are further explored with res pect to n-words in chapter seven. The behavior n-word licensi ng supports the current proposal. There is one problem with this arrangement, however. The theoretical possibility exists for a T1 negative marker, [iNEG], and a T3 negative marker, [iNEG], to be in the same structure. Such a sentence s hould then have a double negation reading. A double negation reading for the pu rposes here and in chapter seven refers to the idea that two [iNEG] features in the same sent ence would cancel each other out causing the sentence to receive a positive interpretation. Below is a hypothetical illustration: (67) O Joo no[iNEG] correu duas milhas no[iNEG] DET Joo NEG ran.3S two miles NEG John didnt not run two miles i.e. he did run two miles. Native speakers indicate that such sentences never have a double negation interpretation however and this is problemat ic for my proposal. I will need to leave this issue unresolved in hopes that it can be solved later. 6.4 Conclusion This chapter provided a syntactic analysi s of Clause-Final Negation in Brazilian Portuguese. Section 6.2 presented some previous attempts to understand CFN syntactically. I then claimed that CFN in BP is derived through fronting AgrP to the specifier of a lower Topic Phrase in the le ft periphery. The head of the topic phrase is 192

PAGE 193

either the negative marker no or the positive marker sim I concluded this chapter by proposing that this head can be realized by no [uNEG] or no [iNEG]. The former is found in T2 and the latter in T3. It remains a problem why the latter cannot also be used in T2. In order to account for t he BP data, I relied on both syn tactic mechanisms and the semantic/pragmatic restriction from chapter five. We saw that the analysis accounts for the syntactic restrictions from chapter one. In contrast to Cava lcantes approach, there was no need to provide different analyses fo r CFN in main and embedded clauses. In agreement with others, however, I am forced to claim that there are two CFN markers which differ in the interpretab ility of their negative feature. 193

PAGE 194

CHAPTER 7 NEGATIVE CONCORD IN BRAZILIAN PORTUGUESE 7.1 Introduction The principal goal of this chapter is to discuss the concepts of Negative Concord (NC), N-words, NC in Brazilian Port uguese (BP), and how NC and Clause-final Negation (CFN) interact. I divide this chapt er into three main sections. Section 7.2 discusses the general ideas surrounding NC, n-words, and the differences between strict and non-strict NC Langu ages. I also discuss the debate regarding the semantic and syntactic nature of n-words and negative concord. Section 7.3 shows that BP is a non-strict NC language and makes explicit wh ich words are n-words in BP. Section 7.4 addresses the nature of the CFN marker and its relationship to n-words. This is a continuation of the discussion in chapters five and six, and here I finalize my discussion of CFN, semantics, and syntax. 7.2 Negative Concord Negative Concord (NC) is a highly studied phenomenon, and this fact is especially true within the field of Romance Linguistics. The intense study of NC has lead to several theoretical disputes. This secti on addresses these debates. First, I define and give examples of the different two types of NC, i.e. strict and non-strict. From there, I lay out the principal analyse s for a general understanding of NC both semantically and syntactically. The main issue concerning n-words is whet her they are negative or not. In some languages, such as the Germanic languages, n-wo rds are in fact negative. This is the reason that, in all cases, an n-word in St andard English and a sentential marker give rise to a Double Negation (DN) reading. However, it has been more difficult to decide 194

PAGE 195

whether n-words in NC languages are inherent ly negative. There are essentially two main options: either n-wo rds are negative (Zanuttini 19 91, Haegeman 1995, Haegeman & Zanuttini 1996, De Swart & Sag 2002); or n-words are non-negative (Laka 1990, Ladusaw 1992, Giannakidou 2000, Zeijlstra 2004, 2008). A third option claims that nwords are ambiguous, however this is not treated here 29 (Van der Wouden & Zwarts 1993, Van der Wouden 1997, Herburger 2001). For this dissertation, I follow Zeijlstra (2004), who proposes a version of the nonnegative option based more on syntax and feature agreement than semanti cs. I present Zeijlstras claims that NC involves an Agree relationship. According to him nwords in NC languages have an [uNEG] feature while in DN languages they have an [iNEG] feature. 7.2.1 Definition and Types of Negative Concord Early in the study of Portuguese and negat ion, Dias (1917) (cf. Pereira 2000) explained what NC is stating that, From the co-occurrence of two negatives a positive does not result (Dias 1917:308). This occurs in many languages, such as the Slavic languages where n-words (defined below) r equire the presence of some negative marker in order for them to be negative. In the case of (1) the n-word nikomu must be accompanied by the negative marker on the verb, ne-. Romance languages are also known for having NC, seen in example (2) where the n-word nadie requires the presences of the preverbal negative marker no. (1) Milan nikomu nevol Czech Milan nobody NEG-call Milan doesnt call anybody (Zeijlstra 2004:121) 29 See Zeijlstra 2004 for a presentation of the ambiguity analysis and the reasons that it cannot be crosslinguistically applicable. 195

PAGE 196

(2) Juan no llam a nadie Spanish Juan NEG called.3SG PREP nobody Juan didnt call anybody N-words are words that enter into a re lationship, to be discussed below, with a negative marker in NC situations (Laka 1990, Giannakidou 2000, 2006, Ze ijlstra 2008). Giannakidou gives a more formal defin ition that I follow here (2006:328): (3) N word An expression is an n-word iff : a. can be used in structures containing sentential negation or another expression yielding a reading equivalent to one logical negation; and b. can provide a negative fragment answer Giannakidou then defines a fragm ent answer as the following: (4) Fragment Answer An answer to a wh-question Q is a fragment answer iff: c. corresponds in form to the wh-XP constituent in Q; and d. is interpreted as a proposition An example of a fragment answer from Spanish is (5) The questions answer is an n-word fragment. (5) a. Quien hizo la tarea? Who did.3S DET homework who did the homework? b. Nadie nobody 196

PAGE 197

In addition, n-words can often be i dentified by what appears to be negative morphology, such as the nie in niemand nobody in Dutch, or the ni in ningn none for Spanish; however, this is not strictly nec essary. French has several n-words with no apparent negative morphology, such as personne nobody. The debate surrounding NC is: if an n-word is negative, then two negative items appear in a sentence, the n-wo rd and the sententia l negative marker. That being the case, sentences with an n-word and a negativ e marker should yield a double negation (DN) reading. The interpretation in NC is that of only one negative operator in the sentence, however. This is not what happ ens in many Germanic languages evidenced by Standard English. If t he negative marker and an n-word are both present, then DN arises. (6) gives an English sentence with an n-word and a negative marker, and (7) gives a logical representation. Notice that in this case, there are two negative elements, hence DN. DN means that tw o negatives are independent of each other and as such, they cancel each other out, which is why a double negative gives a positive reading in some dialects of English (Horn 2002: 296 for a definition of double negatives). (6) Maria doesnt see nobody (i.e. Maria sees somebody) (7) x SEE(m,x) 30 Double Negation NC does not have the same logical structur e as a DN sentence does. The logical representation for the NC sentence (2) is given below, and even though there are (apparently) two negative elements in the sentence, only one negative operator is interpreted semantically. 30 There are other ways to represent this; however they are not relevant here. 197

PAGE 198

(8) x CALL(j,x) Negative Concord This is NC, i.e. two negative elements c oalescing to yield only one semantic negation (Giannakidou 2000, 2006, Zeijlstra 2008). NC languages are divided into two types based on empirical patterns: strict NC language and non-strict NC languages. The same two languages from above, Czech and Spanish, serve as examples of eac h type of NC. Strict NC languages are languages that always require a sentential negative marker to be present when the sentence contains an n-word (Giannakidou 2006:352). In (9) if the negative marker is not present, then the sent ence is ungrammatical. (9) Dnes nikdo *( ne)vol Czech Today nobody NEG.calls Today nobody is calling (Zeijlstra 2008:3) In many Romance language however, this generalization does not hold, and these languages are known as non-stri ct NC languages. This is because the n-word does not always require the presence of a negative marker. In Italian for exam ple, if the n-word is the subject, then the negative mark er is not needed, as in example (10) In fact in these cases, if the negative marker and a preverbal n-word appear, then the DN reading arises, (11) (10) Nessuno ha arrivato Italian nobody has.3S arrived Nobody arrived (11) Nessuno no ha arrivato nobody NEG has.3S arrived *Nobody arrived Nobody didnt arrive (i.e everybody did arrive) 198

PAGE 199

A preverbal n-word is sufficient to licens e a postverbal n-word in non-strict NC languages, and example (12) shows this with nadie licensing nunca and nada (12) Nadie nunca hace nada Spanish Nobody never does.3S nothing Nobody ever does anything Also in non-strict languages, the n-word needs to be an independent DP constituent (Giannakidou 2006). If the n-word is a s ub-part of a DP constituent then a double negation reading may arise, or at the least the NC reading isn t readily available. The example below from Giannakidou (2006:354) shows that when nessuno is inside a DP, the sentence is very degraded. This is di fferent from a strict NC language because in strict NC languages, in all cases, if there is an n-word there must be a sentential negative marker. (13) ??Nessuno studente ha letto nessun libro Italian No student had.3S read no book No student read any book Strict and Non-strict NC lan guages also differ in thei r strategies for forming negative imperatives. Strict NC languages may allow what are called true negative imperatives (Zeijlstra 2 008). True negative imperatives are formed from positive imperatives that are negated using the typical negative ma rker (Zanuttini 1997, Han 2001, Zeijlstra 2008). This is seen in the example below from Polish (Zeijlstra 2008:24). (14) Pracuj! Polish Work.2S.IMP Work! 199

PAGE 200

(15) Nie pracuj! NEG work.2S.IMP Dont work! While strict NC languages allow true negativ e imperatives, non-strict languages do not have true negative imperatives in (17) (Zeijlstra 2008). Instead of true negative imperatives, suppletive imperatives are fo rmed using non-imper ative morphology, generally the subjunctive form of the verb as seen in the Spanish example in (18) (16) Lee eso Spanish Read.IMP.2S that Read that! (17) *No lee eso! NEG read.IMP.2SG that (18) No leas eso NEG read.SUB.2S that Dont read that! This section has shown two types of NC languages: strict and non-strict. Following traditional definitions I have shown that strict NC languages must always have a negative marker present when an n-word is used. Additionally, strict NC languages allow true negative imperatives. Non-strict NC languages allow subject n-words to exist without the presence of a negative marker. Negative imperativ es in non-strict languages are formed using suppletive mor phology instead of imperative morphology. 7.2.2 N-Words are Negative Quantifiers As mentioned above, there are two main claims about the inherent negativity of nwords. The first hypothesis claims that nwords are negative. A principal proponent of this idea is Zanuttini (1991), supported by Haegeman & Zanuttini (1996), who claims 200

PAGE 201

that n-words works much like wh-words. Fi rst, much like NC, multiple wh-words are interpreted as one question where each wh-wor d binds a different variable. These two wh-binders become one in the syntax, much like the interpretation of one negative operator in NC. Second, like wh-movement, movement of a negative operator to the left periphery causes subject-auxiliary inversion, such as never had I been so scared Third, both wh-words and n-words license negative polarity item s (NPI), such as anybody in examples (19) and (20) (19) Who loves anybody? (20) Nobody saw anybody. Based on this evidence, Haegeman & Zanuttini (1996) determine that the interpretation of n-words is like that of wh-words. They propose a negative absorption rule based on Higginbotham & Mays (1981) wh-absorption ru le (Giannakidou 2006:334) This rule is (21) and states that every instance of negation coalesces to one instance of negation at LF. Thus, for Zanuttini (1991), n-words are negative quantifiers that go through a process of neg-absorption 31 in a NC situation. Example (22) has the neg-absorption rule where there is one n-word and a negativ e operator, i.e. a sentential negative marker. (21) [ x ][ y ][ z ]=[ x,y,z ] (22) [ x ] [ x] 31 Another part of this is factorization. Factor ization is what happens when two quantifying elements raise to the same projection under [quantifier raising] in order to turn from monadic quantifiers into one nary polydic quantifier. In the case of WH, polyadic quantification take place after factorization, whereby the interrogative operator of the second wh-element is transmitted into the first wh-operator (Zeijlstra 2004:193). This process for Haegemann & Zanuttini (1996) would be the same for negative operators. 201

PAGE 202

For the syntax, they posit a Neg-Criterion fo r n-words, similar to Rizzis (1991) WhCriterion for wh-phrases. (23) Neg-Criterion a. A NEG-operator must be in a S pec-head configuration with X [NEG] b. An X [NEG] must be in Spec-h ead configuration wit h a NEG-operator Whereby the following definitions hold: NEG-operator: a negative phr ase in scope position Scope position: left-peripheral A-position [Spec,XP] or [YP,XP]. What the Neg-Criterion does is force all n-words to the spec ifier of some negative head. Notice that the sentential negative marker, in many cases, is the head of NegP (see chapter four, Namiuti 2008, and Mioto 1992 for the position of no in BP). Under this analysis, an n-word is a negative quantificational operator and it moves to spec,NegP to form a spec-head relationship with Neg. 32 This is done so that the negative quantifiers have scope over the sentence and negation. By moving all the n-words to spec,NegP, the result is essentially (21) at which point neg-absorptio n can take place over the syntactic representation. N-words contrast with negat ive polarity items like anyin English. N-words are allowed to be fragment answers, as stat ed above (Zanuttini 1991, Giannakidou 2006), but NPIs are not. Example (24) shows that the NPI anybody is not licit as a fragment answer, whereas, the negative nobody is. Rather, polarity items like any must be licensed by some operator (Moscati 2006). 32 If there are no n-words in a negative clause, a s entential negative marker would be the head of NegP and this means that for the Neg-Criterion to be satisf ied, some null negative operator must be in spec, NegP. 202

PAGE 203

(24) a. Who came to the party? b. *anybody/nobody Some problems arise with Z anuttinis analysis. Firs t, Giannakidou (1997) shows that the relationship between wh-words and n-words is not as uniform as Zanuttini claims. She explains that there are also empirical problems equat ing subject-auxiliary inversion in these two instances. For exam ple, not every language has subject-auxiliary inversion in negative fronting, such as S panish and Portuguese, even though it is still used with wh-questions. Zeijlstra (2004) points outs other problems related to t he the law of DN. Recall that one of the reasons that Zanuttini & Haegeman argue for the negative quantifier account of n-words is because of their similarities to whphenomena. Zeijls tra explains that while multiple wh-words cross-linguistica lly are interpreted at LF as a single whoperator, there is a great deal of variation wit h regards to multiple negative operators. Because n-words are negative quantifiers, the law of DN should apply to all n-words as it does in English. Zeijlstra states that negativ e quantifiers obey the law of DN only in a restricted set of languages; this is an indi cator that n-words in NC languages are not semantically negative in the same way t hat they are in DN languages (2004:199). Finally, as Pereira (2000) points out, t here are instances where n-words are not negative, and vice versa, where non-n-words are negative. Herburger (2001) gives (25) as an example where an n-word is interpre ted as positive. If the n-word was a negative quantifier, then it shoul d have a negative interpretation, contrary to fact.. At the same time, there are examples as in (26) (Bosque 1980 and Herburger 2001) where non-negative phrases are interpreted as negative. 203

PAGE 204

(25) Antes de hacer nada debes lavarle las manos before PREP do.INF n-thing must.2S wash.INF.CL DET hands Before doing anything, you should wash his hands. (Herburger 2001:297) (26) En modo alguno se puede tolerar tal actitud! in manner some CL AUX.3S tolerate.INF such attitude Under no circumstances can one to lerate such an attitude! (Herburger 2001:293) Herburger also points out that in many cases n-words se rve as negative polarity items (NPI). She gives example (27) using sin without, a downward entailing preposition, to show n-words behave like NPIs. This behavio r of NC n-words is similar to the behavior of the English word any assumed to be a negative polar ity item, and not similar to negative quantifiers. Notice that sin has a different intera ction with the sentential negative marker no because these two together c ause a double negative reading, (28) (Herburger 2001:297). (27) Pedro compr el terreno sin contrselo a nadie Pedro buy.PST.3S DET land without tell.INF.CL.CL PREP n-body Peter bought the land wit hout telling anybody (28) Pedro compr el terreno sin no contrselo a nadie Pedro buy.PST.3S DET land without NEG tell.INF.CL.CL PREP n-body Peter bought the land wit hout not telling anybody i.e. Peter bought the land, and told everyone. 7.2.3 N-Words are Non-Negative Ladusaw (1992) is a strong proponent of a second hypothesis concerning the negativity of n-words. In response to H aegeman & Zanuttini (1991), he claims that nwords are non-negative indefinites, and he argues that Haegeman & Zanuttini do not have enough semantic support for a negativ e quantifier account of n-words. 204

PAGE 205

Furthermore, he claims that n-words do not have any quantificational force in their own right. Ladusaw concludes that n-words are unbound variables, equati ng them to NPIs. While Ladusaw admits that n-words and NPIs ar e different, the principles that govern each are similar. Ladusaw claims that n-words are only licens ed if the specifier position or the head position of NegP is filled. In the case of strict NC lang uages, this requirement is always satisfied by the sentential negative mark er. For non-strict NC languages, this requirement is satisfied by a sentential negativ e marker, most likely in Neg, or by some n-word in the specifier position of NegP. An abstract element appears in non-strict NC languages when an n-word is preverbal. (29) Non ha telefonato a nessuno [NegP [ Neg non ][TP ha telefonato a nessuno] Nobody has called (30) Nessuno ha telefonato a nessuno [NegP Nessuno [ Neg OP ][TP ha telefonato a nessuno] Nobody called anyone Another approach to n-words as non-negatives comes from Giannakidous (1998, 2000) work. She makes similar claims to Ladusaws in that she takes n-words in NC languages to be non-negative. Her approach differs in that she claims that n-words are NPIs. This helps distinguish between DN languages and NC languages. Giannakidous (2006) approach is different from her previous works in that she does not claim that nwords are necessarily indefinites. Rather, she believes that in some NC languages nwords are existential quantifiers and in others they can be universal quantifiers. She 205

PAGE 206

also concludes that within a language nwords might be ambiguous between universal readings and existential readings (Giannakidou 2006). An apparent problem with this approach is that preverbal n-words and n-word fragment answers in non-strict languages have a negative meaning. In both of these cases, there is no negative operator to licens e the negative meaning. Take the Spanish example repeated from above: (31) a. Quien hizo la tarea? who did.3S DET homework Who did the homework? b. Nadie nobody Giannakidou (2000) claims that looking at frag ment answers in this way is deceptive. She proposes that these are r eally instances of ellipsis, similar to conjunctions or disjunctions. Take the Spanish sentence fr om Zeijlstra (2004:212) where the disjunct is negative but doesnt have sentential negative marker to license it. In this case, what has happened, according to Zeijlstra, is that part of the elided material is the negative marker. Thus, the sentence in (32) comes from (33) The possibility of fragment answers would follow from the same logic. (32) Me caso contigo o con nadie 1S marry.1S PREP.2S or PREP nobody I marry you or nobody (33) Me caso contigo o no me caso con nadie 1S marry.1SG PREP.2S or NEG 1S marry.1S PREP nobody I marry you or I marry nobody 206

PAGE 207

Despite this explanation, Zeijlstra (2006) concludes that, based on Wantabe (2005), this kind of ellipsis is not allowed because the elided mate rial does not have the same semantic identity. That is, the eli ded XP contains a negative marker that is not present in the first part of the sentence. Gi annakidou (2006) explains that this isnt a problem for her analysis, citing Merchant (2001). 7.2.4 Zeijlstras Feature-Based Approach to Negative Concord Continuing in the spirit of Ladusaw and Giannakidou, Zeijlstra (2004, 2006, 2008) claims that n-words are non-negati ve in a traditional sense but that this claim should be viewed within a minimalist perspective and under stood syntactically. I follow Zeijlstras analysis of n-words and NC. First, I lay out some general assumptions that Zeijlstra makes. As a way of showing how his theorie s work, I proceed to apply his analysis to a non-strict NC language. Zeijlstras goal is to show that NC can be more easily understoo d if it is seen through the lens of minimalist feature agreement. He claims that both n-words and sentential negative markers have some [NE G] feature (Chomsky 2000) that needs to be checked/matched by something in the syntax with the same feature. Recall Chomskys claim (2000:3): (34) We therefore have a relation Agree holding between and where has interpretable inflec tional features and has uninterpretable ones, which delete under Agree. Zeijlstra sees the benefit of feature agreement for the semant ic properties of person and number that are both on the subject and the verb. Theref ore, he applies the same principle of Agree to negation, and he s ubsequently formulates a theory for NC based on the agreement of interpretable and uninterpretable features (2008:20): 207

PAGE 208

(35) NC is an Agree relation between a single feature [iNEG] and one or more features [uNEG]. (35) is based on three critical assumptions about feature agreement and NC. First, certain elements of NC languages are only f ormally negative, and although they share some of the morpho-syntactic properties of negation, they themse lves are not negative (Ladusaw 1992). These n-words carry an [uNEG] feature that must be checked. Syntactically, they introduce a free variabl e, making them indefinites (Heim 1982). Second, Zeijlstra follows the proposal of Multiple Agree (Hiraiwa 2001). Multiple Agree is where a single interpretable feature may establish an Agree relationship with multiple elements with the same uninterpretable feature. The same conditions on Agree must still be met; however, Zeijlstra departs from the standard version of Agree introduced in chapter two in one notable way. In contrast to the claim from chapter two that the uninterpretable pr obe must c-command an interpretable goal, feature checking for NC works in a top-down manner, and the element with the [iNEG] feature must ccommand the [uNEG] elements in the derivation (Adger 2003). Third, the element with the [ iNEG] feature may be covert. If some overt element with an [uNEG] feature is present, there must be a corre sponding element with [iNEG] even if it is not phonologically realized. In Zeijlstras analysi s, this element is a negative operator, whose position can vary across languages. Based on these assumptions, Zeijlstra proc eeds to show how this application of Agree can account for not only strict and non-strict NC languages, but also for DN languages, such as Standard English. Bec ause BP is a non-strict NC language, I do not treat strict NC languages nor DN languages here. However, where DN arises, both 208

PAGE 209

the negative marker and the n-words have [iNE G] features. Hence, the DN reading in English is to due the fact that both the nword and the negative marker are interpretable and no feature checking is involved. Like wise, strict NC language are languages whose sentential negative marker and n-words all carry an [uNEG] feature and all sentences with negation require a abstr act negative operator wit h an [iNEG] feature. Zeijlstra explains that nonstrict NC languages have an [uNEG] feature on n-words and that the sentential negative marker has an [iNEG] feature. To obtain a DN reading with subject n-words, he posit s an abstract negative operator with an [iNEG] feature in sentences containing a preverbal n-word. The position of the abstract negative operator is not in spec,NegP but rather some position above AgrP. Italian serves as a proto-typical non-strict language. Non is sufficient to provide semantic negation because it has an [iNEG] feature. In (36) the [iNEG] feature ccommands and checks the [uNEG] feature of the n-word, re sulting in a grammatical sentence. (36) a. Gianni non telefona a nessuno Italian Gianni NEG call.3SG PREP nobody Gianni doesnt call anybody b. [IP Gianni [NegP non[iNEG] telefona [ vP a nessuno[uNEG]]]] With preverbal n-words, a negative operator with an [iNEG] featur e is present and checks the [uNEG] features of the subject nword. The merger of the null operator only occurs in sentences with a subject n-word an important stipulat ion that I adopt without comment. 209

PAGE 210

(37) A. Nessuno ha arrivato n-body has.3S arrived Nobody arrived B. [IP Op [iNEG] [IP nessuno [uNEG] [I' ha arrivato [VP...]]]] The position of the negat ive operator is not fully defined in Zeijlstras work. Crucially, it must be above IP so that it can check an [uNEG] feature on subject n-words. In what follows, I show it adjoined to IP/AgrP. The null operator appears in sentences with a subject n-word and sentential negation, (38) In this case, both the null operat or and the negative marker have an [iNEG] feature, and the two interpretable negat ive features in the sentence lead to a double negation reading. (38) Op Nessuno no ha arrivato n-body NEG has.3S arrived *Nobody arrived Nobody didnt arrive (i.e everybody did arrive) Finally, as noted above, there are case s where an n-word in non-strict NC language is apparently not licensed by anything. In Zeijlstras model, this means that the [uNEG] feature is not getting checked. Recall the Spanish example from above: (39) Antes de hacer nada debes lavarle las manos before PREP do.INF n-thing must.2S wash.INF.CL DET hands Before doing anything, you should wash his hands. According to Zeijlstras proposal, the only way that this could be allo wed is if something was checking the [uNEG] feature of nada He claims that in these situations, the null negative operator is again inserted. 210

PAGE 211

(40) [[NegP Op [iNEG][PP Antes de hacer nada, ]][TP debes lavarle las manos]]] NEG before PREP do.INF n-thing must.2S wash.INF.CL DET hands Before doing anything, you should wash his hands. 7.2.5 Summary This section has discussed two main appr oaches to understanding what n-words are. The first hypothesis is that n-word s are negative universal quantifiers; this is Zanuttinis (1991) approach. The second hypothesis is the one in Ladusaw (1992), Giannakidou (1998, 2000), and Zeijlstra (2004, 2008), which claims that n-words are non-negative existential or universal quantifiers I presented Zeijls tras implementation of this hypothesis in which n-words are non-negative but have an uninterpretable [uNEG] feature that must be checked under Agree against an interpretable [iNEG] feature. 7.3 N-Words and Negative Concord in Brazilian Portuguese I now discuss n-words and NC in BP. Section 7.3.1 argues that BP is a non-strict NC language. Section 7.3.2 applies Zeijlstr as theory to BP. 7.3.1 BP N-Words Following Giannakidous formal defin ition of n-words, given in (3) the following BP words are n-words: ningum nobody, nada nothing, nunca never, nenhum no/any, nenhuma pessoa nobody (Peres 2000). Based on examples (41) through (43) this chapter assumes that BP is a non-stri ct NC language as def ined in section 7.2.1 Example (41) shows the non-strict nature of BP n-words wher e a single n-word can occur preverbally as the subject and carry a negative meaning without the presence of a sentential negative marker. 211

PAGE 212

(41) a. Ningum veio Nobody came.3S nobody came b. Nada conseguia resolver o problema Nothing was.capable.IMPERF.3S resolve.INF DET problem Nothing was able to solve the problem c. Ela nunca me liga 3SF never 1S calls.3S She never calls me d. Nenhum remedio curou o doente No medicine healed.3S DET sick.person No medicine healed the patient. As with Spanish and Italian, if there is a preverbal negative marker, a DN reading arises. 33 (42) Ningum no ajudou a Sienna Nobody NEG helped.3S DET Sienna Nobody didnt helped Sienna (i.e. everybody helped Sienna) The examples in (43) show that there must be a prev erbal sentential negative marker when BP n-words are in a postverbal position. 34 33 Martinez (2006) states that the following is allo wed in Mineiro Portuguese (State of Minas Gerais) i. Ningum no veio Nobody neg came Nobody came While she points to other studies, Alkmim (2001), this is the only example of the n-word and the preverbal negative marker not invoking a double negative reading. All the preliminary work suggests that this is dialectal and is not seen in most dialects of BP. If it turns out that this claim is false, then more analysis would be required. 34 Cavalcante (2007) suggests that t here are instances where the postverbal n-word does not co-occur with a preverbal negative marker. This appears to be dialectal and restricted to the northeast of Brazil. This could be due to BP n-words being inherently negativ e, contrary to what I assume above (Pereira 2000, Peres 2000, Matos 2003, Giannakidou 2006). 212

PAGE 213

(43) a. A Lcia *(no) viu ningum DET Lcia NEG saw.3S nobody Lcia didnt see anybody b. A Lcia *(no) leu nada DET Lcia NEG read.3S nothing Lcia didnt read anything c. A Lcia *(no) leu nenhum livro DET Lcia NEG read.3S no book Lcia didnt read any books d. A Lcia *(no) me liga nunca DET Lcia NEG 1S read.3S never Lcia doesnt ever call me Like other non-strict NC langu ages, a preverbal n-word is sufficient to license postverbal n-words, as in (44) Like NC languages in general, BP allows several n-words to be present in the sentence wit hout any double negation reading, (45) (44) Ningum leu nenhum livro Nobody read.3SG none book Nobody read any books (45) Ningum nunca deu nada para ningum em lado nenhum Nobody never gave.3SG nothing PREP nobody PREP side none Nobody ever gave anything to anybody anywhere. The second part of Giannakidous defin ition deals with fragment answers. Although I give only two examples, all of the words just given are valid fragment answers in BP. (46) a. Quem fez o trabalho? who did.3S DET work who did the work b. Ningum Nobody 213

PAGE 214

(47) a. O que que voc fez? What that 2S did.2S What did you do? b. Nada Nothing BP departs from the behavior of traditional non-strict languages, such as Spanish and European Portuguese, in imperatives Recall that according to Zeijlstra, strict NC allows true negative imperatives, wh ile non-strict languages do not. Although BP clearly patterns with non-strict languages in other ways, it does allow true negative imperatives. Example (48) has an imperative, and the next example, (49) negates the imperative by simply adding the negative marker. Note however, that BP has both true negative imperatives and suppletiv e imperatives, example (50) (Scherre 2007). (48) L isso Read.2SG.IMP that Read that! (49) No l isso NEG Read.2SG.IMP that Dont read that! (50) No leia isso NEG Read.2SG.SUBJ that Dont read that! Another place where BP differs form other non-strict NC languages is if the n-word is part of a DP constituent. For other langua ges in this group, these types of sentences are not grammatical, as shown in example (13) for Italian. BP has no problem with these types of sentences, and in eac h case a NC reading is available. 214

PAGE 215

(51) Nenhum estudante leu nenhum livro no student read.3SG no book No student read any book Even though the evidence appears to leave t he status of BP as a non-strict NC language open to debate, I will assume that it is a non-strict NC language. My reason is that preverbal n-words in BP cannot co -occur with the sentent ial negative marker without yielding a DN reading. This seems to be the most important criterion and it suggests that n-words in BP behave much like those in other Romance languages (Pereira 2000, Zeijlstra 2008). 7.3.2 Feature Agreement and BP N-Words In this section I apply Zeijlstras analysis to BP. 35 Following his analysis of nonstrict NC languages, I assume that BP n-wo rds have an [uNEG] feature. BP works the same as Italian, described above in (36) and (37) In (52) the negative marker has an [iNEG] feature which checks the [uNEG] feature on the n-word ningum (53) gives a syntactic representation. (52) Joo no ligou para ningum John NEG calls PREP nobody John doesnt call anybody (53) AgrP DP NegP Joo no[iNEG] TP ligou para ningum[uNEG] 35 Zeijlstra, Giannakidou, Zanuttini all analyze Romanc e languages but in most cases their discussions of Portuguese are based on European Portuguese. As su ch, some of the conclusions that they come up with do not directly apply to Brazilian Portuguese. 215

PAGE 216

The following example show that prever bal n-words in BP behave like n-words in Italian, triggering the introduction of a nu ll negative operator t hat carries an [iNEG] feature. (54) a. Ningum veio Nobody came.3S nobody came (55) AgrP Op [iNEG] AgrP DP Agr' Ningum[uNEG] veio As in other non-strict languag es, the phenomenon of negative sp read is also captured in BP through Zeijlstras adoption of Multiple Agree. The next sentence has several nwords, all of which check their uninter pretable feature with the null operator. (56) Ningum nunca deu nada para ningum Nobody never gave.3SG nothing PREP nobody Nobody ever gave anything to anybody. (57) AgrP Op [iNEG] AgrP DP Agr' Ningum[uNEG] nunca[uNEG] deu nada[uNEG] para ningum[uNEG] 216

PAGE 217

The co-occurrence of the null operator and the sentential negativ e operator in BP causes a double negation reading. This is due to the presence of two [iNEG] features in the sentence. (58) Ningum no ajudou a Sienna Nobody NEG helped.3S DET Sienna Nobody didnt helped Sienna (i.e. everybody helped Sienna) (59) XP Op [iNEG] AgrP DP Agr' Ningum[uNEG] Agr NegP Neg TP no[iNEG] ajudou a Sienna 7.4 Negative Concord and Clause-Fi nal Negation in Brazilian Portuguese This section returns to the discussion t hat began at the end of chapter six where questions arose regarding the semantic cont ributions of CFN and the different clausefinal nos in T2 and T3. Specifically, I argued t hat the CFN marker in T2 has a [uNEG] feature, which equates it with n-words. Given Zeijlstras syst em, this [uNEG] feature is checked when AgrP bearing an interpretable [iNEG] moves to spec,TopP and ccommands the Top head. The T3 no is used as a negative element in denials which can negate a sentence without any support. Here the negative marker has an [iNEG] feature, much like the prev erbal negative marker, and does not need to be checked. My analysis of BP NC supports the claims of chapter six. Recall the syntactic environments in which CFN licenses an n-word: 217

PAGE 218

Table 7-1. N-Words and CFN Environment T1 T2 T3 Allowed with Subject N-words X X Allowed with Object N-words X Compatible with NPI Idiomatic Expressions X The table above indicates that CFN has no effect on the licensing of n-words. The licensing picture is the same whether CFN is present or not. If the n-word is postverbal, then preverbal negation must be present and CFN can be present or not, i.e. an object n-word is licensed by T1 or T2 but not T3 When the n-word is the subject, preverbal negation cannot be present and CFN can be present or not, i.e. a subj ect n-word cannot appear with T1 and T2, only with T3. These pa tterns are illustrated by the following examples: (60) A Sienna no ajudou ningum (no) DET Sienna NEG helped.3S nobody NEG Sienna didnt help anyone (61) Ningum ajudou o Ablio (no) Nobody helped.3S DET Abllio NEG Nobody helped Abllio In the structure in (62) corresponding to (60) the [iNEG] feature of preverbal no in Neg checks the [uNEG] feature on the postv erbal n-word. The [iNEG] feature that percolated to AgrP checks the [uNEG] on the CFN in Top. 218

PAGE 219

(62) TopP AgrP[iNEG,iTOP]i Top DP Agr Top AgrP no[uNEG,uTOP*] ti A Sienna Agr NegP Neg[iNEG] TP no T vP ajudou ningum[uNEG] In (63) corresponding to (61) the subject n-words [uNEG] feature is checked by the null operator. The CFNs [uNEG] feature is again checked by t he [iNEG] feature on AgrP. (63) TopP AgrP[iNEG,iTOP]i Top Op [iNEG] AgrP Top AgrP no[uNEG,uTOP*] ti DP Agr Ningum[uNEG] Agr TP T vP ajudou o Abllio This analysis equates n-words and T2 CFN by suggesting that both have an [uNEG] feature that must be checked by an [ iNEG] feature elsewhere in the structure. As such T2 is a case of negative concord. 219

PAGE 220

The theory also correctly predicts that T3 with [iNEG] in Top, will not license an object n-word. After AgrP mo ves, the Top head, even though it has an [iNEG] feature, does not c-command any n-words inside AgrP Consequently, their [uNEG] features will remain unchecked. I demonstrate this claim by showing how it blocks the ungrammatical example in (64) As can be seen in the structure below, Top does not ccommand and therefore does not check the n-words [uNEG] feature. (64) *O Z ajudou ningum no DET Z helped.3S nobody NEG Z helped nobody (65) *TopP AgrP[uTOP]i Top DP Agr Top t i no[iNEG,uTOP*] O Z Agr TP ajudou ningum[uNEG] The ungrammaticality of (64) thus has the same explanation as T3s inability to license an NPI idiom that we saw in chapter six. Unfortunately, the same DN pr oblem from chapter six arises in these structures. Because there are two CFN mark ers, either one could be insert ed into the derivation. However, if the T3 negative marker, which is [iNEG], were inserted into the derivations in (62) and (63) then a double negative read ing would incorrectly result as the clause would have two [iNEG] features. This is the same problem from chapter six which I have not solved. Neverless, it seems necessary to keep T2 and T3 distinct analytically because we saw that T3 is subject to a stricter discourse restriction than T2. 220

PAGE 221

7.5 Conclusion This chapter has discussed NC in BP. T he data show that BP is a non-strict NC language similar to Spanish and Italian (Zeijlstra 2004). I followed Zeijlstra in analyzing the licensing of n-words as an instance of syntactic Agree in which an [iNEG] must ccommand and check any n-words [uNEG] featur e. Three interpretable elements are in principle available to do this checking: preverbal negation, a null negative operator which is present just in case an n-word is preverbal, and CFN no [iNEG] in Top. In practice however, CFN cannot license an n-word on its own because it does not ccommand into AgrP once AgrP has moved to spec,TopP. The analysis of CFN from chapter six thus correctly accounts for the interaction of CFN and n-word licensing. CFN has no affect on n-word licensing. 36 36 Of particular interest to future work is question of diachronic change of n-words. Some BP dialects vary on the level of Negative Concord. Consider the fo llowing examples from Cavalcante (2007) where the nword is not licensed by any c-commanding negative element: (i) Fui cobrar nada dele Went.1S charge. INF nothing PREP.3S.M I went to get nothing (ii) Veio ningum Came.3S nobody Nobody came (iii) Tou sabendo de nada Am knowing PREP nothing I know nothing Since some languages change from strict negative concord languages to double negation languages, it is interesting to understand this proc ess in more detail and make some cross-dialectal comparisons within BP. Also of interest is the fr equency of clause-final negation in dialects with different types of n-words. This work claims that clause-final negation is found in all sectors of Brazil; however, there are definite regional differe nces concerning n-words that should be examined. Specifically, the relationship between n-words and clause-final negation in northern and northeastern dialects appear to suggest a relationship betwe en frequency of clause-final negation and [iNEG] carrying n-words. 221

PAGE 222

CHAPTER 8 CONCLUSION In this work, I sought to understand ClauseFinal Negation in Brazilian Portuguese within the Minimalist Program. Clause-Final Negation (CFN) is a somewhat rare form of negation, and this is especially so withi n the Indo-European family of languages. Brazilian Portuguese has three types of negat ion and two of these include clause-final negative markers. CFN is thus a part of Brazilian Portuguese that is in need of study. The work began in chapter one by laying out the data and setting the stage for a discussion that involved both syntactic as well as semantic/pragmatic elements. Chapters two, three, and four furthered the discussion by implementing a theoretical framework and clause structur e for BP. In these chapt ers I gave evidence for the clause structure of BP that is needed to understand negation generally and CFN in particular. Chapter five discussed the semantic and pragm atic restrictions on the use of CFN. I showed, following Schwenter, that CFN can only be fully understood through a proper delineation of the discourse factors that influence the fe licity of CFN sentences. Chapter six offered a syntactic analysis of CFN. I argued that the negative marker no comes to be clause-final when the entire clause, AgrP, fronts to the left no, which is a topic head in the left periphery. The move ment of AgrP yields a topic interpretation for the clause. The analysis opens the door to further examination of topicalization in BP. In particular, this work provides ev idence that large phrases can be topicalized. This discussion of clause topicalization prom ises to be an area which will show an even tighter link between discourse and syntax. 222

PAGE 223

Chapter six posited two CFN markers. One is interpretable, bearing an [iNEG] feature, and the other is uninterpretable, being specified as [uNEG]. The need for two heads motivated three observations: 1) partially distinct discourse licensing restrictions on T2 versus T3, 2) the need to have a sem antically negative interpretation in T3 sentences, and 3) the need to avoid a Double Negation reading in T2 sentences. An interpretable head with [iNEG] is used in T3 environments in order to give such sentences a negative interpretation. On t he other hand, the uninterpretable head with [uNEG] was required in order to prevent double negation readings with T2. Unfortunately, this picture could still not pr event the T3 head from also being used in combination with preverbal negat ion, resulting in a double negation reading. I left this problem for future work. Chapter seven closed the discussion by test ing the theory from chapter six in the domain of Negative Concord (NC). I applied Zeijlstra s theory of NC to BP and the theory in conjunction with the analysis of CFN from chapter six was able to explain why CFN does not have any influence on t he licensing of n-words in BP. 223

PAGE 224

LIST OF REFERENCES Adger, David. 2003. Core syntax: A Minimalist approach New York: Oxford University Press. Adrago, Maria do Mar and Joo Costa. 2004. On the Status of Preverbal Subjects in Null Subject Languages: Evidence from Acquisition. In Proceedings of GALA 2003, ed. Jacqueline van Kampen and Sergio Baauw, 690 The Hague: Utrecht University, Lot Publications. Aissen, Judith. 1992. Topics and Mayan. Language 68, 1:4380. Alexiadou, Artemis and Elena Anagnostopoulou. 1998. Parametrizing Agr: Word order, Verb-Movement and EPP-checking. Natural Language and Linguistic Theory 16:491539. Alkmim, Monica. 2001. As negativas sentenciales no dialeto mineiro: uma abordagem varacionista. Doctoral dissertation, Universidade Federal de Minas Gerais. Ambar, Manuela. 1999. Aspec ts of the Syntax of Focus in Portuguese. In The Grammar of Focus eds. Georges Rebuschi and Laurice Tuller, 2353. Amsterdam/Philadelphia: John Benjamins. Aoun, Joseph and Audrey Li .2000. Sc ope, structure, and expert systems: A reply to Kuno et al. Language 76:13355. Armstrong, Meghan E. 2008. Pragmatic Rest rictions on Affirmative Response Choice in Brazilian Portuguese. In Selected Proceedings of the 10th Hispanic Linguistics Symposium ed. Joyce Bruhn de Garavito and Elena Valenzuela, 288299. Somerville, MA: Cascadilla Proceedings Project. Azevedo, Milton. 2005. Portuguese. A linguistic introduction Cambridge, United Kingdom: Cambridge University Press. Barbosa, Pilar. 1995. Null Subjects. Do ctoral dissertation, Massachusetts Institute of Technology. Barbosa, Pilar. 2004. Two Kinds of Subject pro Paper presented at the 14 th Colloquium of Generativ e Linguistics, University of Oporto. Barme, Stefan. 2000. Negao (e afirma o) no portugus brasileiro falado informal. ms.,University of Mainz. Barss, Andrew. 1986. Chai ns and anaphoric dependence. Doctoral dissertation, Massachusetts Institute of Technology. 224

PAGE 225

Bastos, Ana. 2001. Fazer, eu fao! T opicalizao de Constituintes Verbais em Portugus Brasileiro. Master Th esis, Universidade de Campinas. Bastos-Gee, Ana. 2009. Topicalization of Verbal Projections in Brazilian Portuguese. In Minimalist Essays on Brazilian Syntax ed. Jairo Nunes, 161190. Amsterdam: John Benjamins. Belletti, Adriana. 1990. Generalized Verb Movement Turin: Rosenberg e Sellier. Belletti, Adriana. 2004. Aspects of the Low IP Area. In The Structure of CP and IP. The Cartography of Synt actic Structures, Vol. 2 ed. Luigi Rizzi, 1651. Oxford: Oxford University Press. Beninc, Paola, 2001. The position of t opic and focus in the left periphery. In Current Studies in Italian Syntax. Essays Offered to Lorenzo Renzi ed. Cinque, Guglielmo and Giampaolo Salvi, 3964. Amsterdam: Elsevier. Beninc, Paola and Cecilia Poletto. 2004. Topic, Focus and V2. Defining the CP sublayers, In The Structure of CP and IP. The Cartography of Syntactic Structures, Vol. 2 ed. Luigi Rizzi, 5275. Oxford: Oxford University Press. Bernini, Giuliano and Paolo Ramat. 1996. Negative Sentences in the Languages of Europe: A Typologic al Approach (Empirical Approaches to Language Typology 16 ). Berlin: Mouton de Gruyter. Besten, Hans den. 1986, D ouble Negation and the Genesis of Afrikaans. In Substrata versus Universals in Creole Languages, ed. Pieter Muysken and Norval Smith, 185230. Amsterdam: John Benjamins. Biberauer, Theresa. 2009. Jespersen off course? The case of contemporary Afrikaans negation. In Cyclical Change, ed. Elly Van Gelderen, 91132. Amsterdam: John Benjamins. Biberauer, Theresa and S onia Cyrino. 2009. Negative developments in Afrikaans and Brazilian Portuguese, paper presented at the 19th Colloquium on Generative Grammar, University of the Basque Country, Vitoria-Gasteiz. Bobaljik, Jonathan. 2003. Floating Q uantifiers: Handle With Care. In The Second Glot International State-of-The-Article Book eds. Lisa Cheng and Rint Sybesma, 107148. Berlin: Mouton de Gruyter. Bobaljik, Jonathan and Hskuldur Thrinsson. 1998. Two Heads Arent Always Better than One, Syntax 1.1:37. 225

PAGE 226

Bokovi eljko. 1999. On multiple feature c hecking: multiple Wh-fronting and multiple head-movement. In Working Minimalism ed. Samuel Epstein and Norbert Hornstein, 15987. Cambridge: MIT Press. Bokovi eljko and Howard. Lasnik. 2007. Minimalist Syntax: The Essential Readings Oxford: Blackwell. Bosque, Ignacio. 1980. Sobre la Negacin. Madrid : Catedra. Bouquiaux, Luc. 1970. La Langue Birom Paris: Bibliothque de la Facult de Philosophie et Lettr es de l'Universit de Lige. Brito, Ana Maria. 2001. Clause Structure, Subject Positions and Verb Movement. In Romance Languages and Linguistic Theory ed. D'Hulst et al., 6386. Amsterdam: John Benjamins. Camargos, Marcelo Luiz de. 2002. Teoria de Cpia e Negao. Masters thesis, Universidade Federal de Minas Gerais. http://www.ufop.br/ichs/c onifes/anais/LCA/clca03.htm Cardinaletti, Anna. 1997. Subjec ts and Clause Structure. In The New Comparative Syntax ed, Liliane Haegeman London: Longman. Cardinaletti, Anna. 2004. Towards a cart ography of subject positions. In The Structure of CP and IP. The Cart ography of Syntactic Structures vol. 2, ed. Luigi Rizzi. Oxford and New York: Oxford University Press Cardinaletti, Anna & Michal Starke. 1999. The Typology of Structural Deficiency: A Case Study of the Three Classes of Pronouns. In Clitics in the Languages of Europe, ed. Henk van Riemsdijk, 145234.Berlin: Walter de Gruyter. Carnie, Andrew. 2006. Syntax: A Generative Introduction Oxford: Wiley Blackwell Publishers. Casielles, Eugenia. 1997. Topic, Focus, and Bare Nominals in Spanish Doctoral dissertation. University of Massachusetts, Amherst. Cavalcante, Rerisson. 2007. A negao ps-verbal no portugus brasileiro: anlise descritiva e teric a de dialetos rurais de afro-descendentes. Master thesis, Universidade Federal da Bahia. Cinque, Guglielmo, 1999. Adverbs and Functional Heads. A Cross-Linguistic Perspective. New York: Oxford University Press. Cinque, Guglielmo (Ed.), 2002 Functional Structure in DP and IP. The Cartography of Syntactic Structures, Vol. 1 New York: Oxford University Press. 226

PAGE 227

Cheng, Lisa L.-S. 1991. On the Typology of Wh-questions, Doctoral dissertation, Massachusetts Institute of Technology. Cheng, Lisa L-S.1997. On the Typology of Wh-Questions New York: Garland Publishing. Chomsky, Noam. 1976. Essays on Form and Interpretation Amsterdam:NorthHolland. Chomsky, Noam. 1981. Lectures on Government and Binding Dordrecht: Mouton de Gruyter. Chomsky, Noam. 1995. The minimalist program Cambridge, MA: The MIT Press. Chomsky, Noam. 2000. Minimalist Inquiries: the Framework. In Step by Step. Essays in Honor of Howard Lasnik ed. Roger Martin, David Michaels and Juan Uriagereka, 89155. Cambridge, Mass: MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Ken Hale: A life in language ed. Michael Kenstowicz. Cambridge, Mass.: MIT Press. Cinque, Guglielmo. 1990. Types of A-Dependencies. Cambridge: MIT Press. Costa, Joo. 1998. Word order variation : a constraint-based approach The Hague: Holland Academic Graphics. Costa, Joo. 2000. (ed.). Portuguese syntax New comparative studies Oxford: Oxford University Press. Costa, Joo. 2001. Spec, IP ou Deslocado? Prs e contras das duas anlises dos sujeitos pr-verbais. DELTA, 17:283304. Costa, Joo. 2004. Subject Positions and Interfaces : The Case of European Portuguese The Hague: Mouton de Gruyter. Costa, Joo and Charlotte Galves. 2002. Exte rnal subjects in two varieties of Portuguese: evidence for a non-unified analysis. In Romance Languages and Linguistic Theory 2000, ed. Claire Beyssade et al., 109125. Amsterdam: Johns Benjamins. Da Cunha, Maria. 1996. Gramatica lizao dos Mecanismos de Negao em Natal. In Gramaticalizao no Portugus do Brasil, ed. Maria Martelotta et al., 167189. Rio de Janeiro:Tempo Brasileiro. Da Cunha, Maria. 2001. O Modelo das Motivaes Competidoras no Domnio Funcional da Negao. DELTA 17:130. 227

PAGE 228

Da Cunha, Maria. 2007. Gr ammaticalization of the st rategies of negation in Brazilian Portuguese. Journal of Pragmatics 39:16381653. Den Besten, H. 1986. Double negation and the genesis of Afrikaans. In Substrata versus universals in Creole languages. Papers from the Amsterdam Creole workshop, April 1985 ed. Pieter Muysken and Norval Smith, 185. Amsterdam: John Benjamins. De Swart, Henriette and Ivan Sag. 2002. Negative Concord in France. Linguistics & Philosophy 25: 373417. Dryer, Matthew S. 2007. Verb-ObjectNegative Order in Central Africa. In Negation Patterns in West Africa ed. Norbert Cyffer, Erwin Ebermann, and Georg Ziegelmeyer. Ms. Duarte, Maria. 1995. A Perda do Prin cpio Evite pronome no Portugus Brasileiro. Doctoral dissertation, Universidade de Campinas. Duarte, Maria. 2000. The loss of the a void pronoun principle in Brazilian Portuguese. In Brazilian Portuguese and the Null Subject Parameter, ed. Mary Kato et. al.,1736. Frankfurt-Madrid: Ve rvuert-Iberoamericana. Emonds, Joseph. 1978. The verbal complex V'V in French. Linguistic Inquiry 9:15175. Ferraz, Luiz Ivens. 1979. The Creole of So Tom Johannesburg: Witwatersrand University Press. Fonseca, Helena. 2004. Marcador negat ivo final no Portugus Brasileiro. Cadernos de estudos lingsticos 46: 519. Galves, Charlotte. 1989. O objeto nulo no portugus brasileiro: percurso de uma pesquisa. Cadernos de Estudos Lingsticos 17:6590. Galves, Charlotte. 1998. Tpicos e su jeitos, Pronomes e Concordncia no Portugus Brasileiro. Cadernos de Estudos Lingsticos 34:1931. Giannakidou, Anastasia. 1998. Polarity sensitivity as (non)veridical dependency Amsterdam: John Benjamins. Giannakidou, Anastasia. 2000. Negative ... Concord? Natural Language and Linguistic Theory 18:457523. 228

PAGE 229

Giannakidou, Anastasia. 2006. N-words and negative concord. In The Linguistics Companion vol. III, ed. Martin Everaert and Henk van Riemsdijk. 327391. Oxford: Blackwell. Goodall, Grant. 2001. The EPP in Spanish. Objects and Other Subjects eds. William Davies and St anley Dubinsky, 193223. Amsterdam:Kluwer. Grohmann, Kleanthes. 2003. Prolific Domains: On the An ti-Locality of Movement Dependencies Amsterdam: John Benjamins. Gnther, Wilfried. 1973. Das Portugiesische Kreolisch der Ilha do Prncipe Marburg an der Lahn: Marburger Studien zur Afrika und Asienkunde. Haddad, Youssef. 2007. Adjunct Control in Telegu and Assamese. Doctoral dissertation. University of Florida. Haegeman, Liliane. 1995. The syntax of negation. Cambridge Studies in Linguistics 75 Cambridge: Cambridge University Press. Haegeman, Liliane and Raffaella Zanuttini 1991. Negative Heads and the NegCriterion. The Linguistic Review 8:233251. Haegeman, Liliane and Raffaella Zanuttini 1996. Negative Concord in West Flemish. In Parameters and Functional Heads, eds. Belleti Adriana and Luigi Rizzi,117-180. Oxford: Oxfo rd University Press. Han, C-H. 2001. Force, negation and imperatives. The Linguistic Review 18, 289325. Hansen, Quinn. 2005. Basic Word Order: Kaqchikel Mayan. Masters thesis, Universidade Nova de Lisboa. Hiraiwa, Ken. 2001. Multiple Agreement and the Defectiv e Intervention Effect, Ms. MIT. Heim, Irene. 1982. The semantics of de nite and inde nite noun phrases. Doctoral dissertation, University of Massachusetts, Amherst. Herburger, Elena. 2001. The negative concord puzzle revisited. Natural Language Semantics 9:289333. Higginbotham, James and Robert May. 1981. Questions, Quantifiers, and Crossing. The Linguistic Review 1, 4180. Holmberg, Anders. 2005. Is there a little pro? Evidence from Finnish. Linguistic Inquiry 33, 53364. 229

PAGE 230

Horn, Laurence. 1989. A Natural History of Negation Chicago: The University of Chicago Press. Horn, Laurence. 2002. Assertoric inertia and NPI licensing. CLS 38 : 5582. Hornstein, Norbert. 2001. Move! A Minimalist theory of construal: Generative syntax. United Kingdom: Blackwell. Hornstein, Norbert; Jairo Nunes and Kleanthes Grohmann. 2005. Understanding Minimalism Cambrigde: Cambrigde University Press. Huang, C.-T. J. 1982. Logical Relations in Chinese and the Theory of Grammar. Doctoral dissertation, Massachus etts Institute of Technology. Iatridou, Sabine. 1990. About Agr(P). Linguistic Inquiry 21: 551. Israel, Michael. 1998. The rhetoric of grammar: scalar reasoning and polarity sensitivity. Doctoral di ssertation, University of California, San Diego. Jespersen, Otto. 1917. Negation in English and other Languages Copenhagen: A.F.Hst. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge: MIT Press. Kato, Mary. 1999. Strong and Weak pronominal s in the Null Subject Parameter. Probus 11: 137. Kato, Mary. 2000. The partial pro-drop nature and the restricted VS order in Brazilian Portuguese. In Brazilian Portuguese and the Null Subject Parameter, ed. Mary Kato et. al., 207240. Frankfurt-Madrid: Vervuert-Iberoamericana. Kato, Mary. 2003. Null objects and VP ellipsis in European and Brazilian Portguese. In Romance Languages and Linguistic Theory 2001, ed. Josep Quer et al.,131154. Amsterdam: John Benjamins. Kato, Mary. 2009 Mudana de ordem e gramaticalizao na evoluo das estruturas de foco no Portugus Brasileiro. Estudos Lingsticos 38: 375385. Kato, Mary and Carlos Mioto. 2005. A mu lti-evidence analysi s of European and Brazilian Portuguese Wh-questions. In Linguistic evidence: empirical, theoretical and computational perspectives ed. S Kepser and Marcelo Reis, 307328. Berlin: Mouton de Gruyter. 230

PAGE 231

Kato, Mary and Eduardo Raposo. 1996. European and Brazilian Portuguese word order: questions, focus and topic constructions. In Aspects of Romance Linguistics eds. C. Parodi et. al., 26777. Washington: Georgetown University Press. Kato, Mary and Eduardo Raposo. 2007. Topi calization in European and Brazilian Portuguese. In Romance Linguistics 2006 ed. Jose Camacho et. al., 205-218. Amsterdam: John Benjamins. Kayne, Richard. 1989. Notes on English Agreement. CIEFL 1: 40-67. Kayne, Richard. 1994: The Antisymmetry of Syntax. Cambridge, MA: MIT Press. Koopman, Hilda. And Dominique Sportiche. 1991. The position of subjects. Lingua 85.1:211258. Ladusaw, William A. 1992. Ex pressing negation. In Proceedings of the Second Conference on Semanti cs and Linguistic Theory ed. Chris Barker and David Dowty, 237259. Columbus: Department of Linguistics, Ohio State University. Laka, Itziar. 1990. Negation in Syntax: on the Nature of Functional Categories and Projections. Doctoral dissertation, Massachusetts Institute of Technology Larson, Richard .1988. On the double object construction. Linguistic Inquiry 19:335391 Lasnik, Howard. 1972. Analyses of negation in English Doctoral dissertation, Massachusetts Institute of Technology. Lasnik, Howard. 1999. Minimalist Analysis Oxford: Blackwell. Lasnik, Howard and Mamoru Saito. 1992. Move : Conditions on its application and output Cambridge: The MIT Press. Lessa de Oliveira, Ana. 2008. Oraes Relativas do Portugus: questes tericas e fatos de aquisio. Doctor al Dissertation, Universidade de Campinas. Lobato, Lcia Maria Pinheiro. 1986. Sintaxe gerativa do portugus. Da teoria padro teoria da regncia e ligao. Belo Horizonte:Viglia. Lobeck, Anne, 1995. Ellipsis: Functional Heads, Licensing, and Identification. New York: Oxford University Press. Martins, Eneida. 1997. Sentential negation in spok en Brazilian Portuguese Washington: Georgetown University. 231

PAGE 232

Martins, Ana. 1994. Clticos na histria do portugus. Doctoral dissertation, Universidade de Lisboa. Martins, Ana and Jairo Nunes. 2006. Raising Issues in Brazilian and European Portuguese. Journal of Portuguese Linguistics 4: 5377. Martins, Ana and Jairo Nunes. 2008. Apparent Hyper-raising in Brazilian Portuguese: Base-Generation of Topi cs and Long Distance Agreement. In Selected papers from Edges in Syntax, Cyprus 2006 ed. Phoevos Panagiotidis. Oxford: Oxford University Press. Martnez, Cristina. 2005. Concordancia de negacin en el portugus minero. Pterodctilo: Revista de Arte, Literatura y Lingstica 3:2635. Matos, Gabriela. 2003. Aspectos sintcticos da negao. In Gramtica da lngua portuguesa ed. Maria Helena Mira Mateus, 767793. Lisboa: Caminho. May, Robert. 1985. Logical Form: Its Structure and Derivation Cambridge: MIT Press. McCloskey, James. 1991. Clause structure, ellipsis and proper government in Irish. Lingua 85:259. Merchant, Jason. 2001. The Syntax of Silence: Sluici ng, Islands, and the Theory of Ellipsis Oxford: Oxford University Press. Mioto, Carlos. 1992. Negao Sentential no Portugus Brasileir o. Doctoral disseration, Universidade de Campinas. Mioto, Carlos. 1998. Preposies: Ncl eos lexicais ou funcionais? In Lingstica e Ensino eds. Luis Cabral et al., 165179. Florianpolis, Editora Insular. Moscati, Vincenzo. 2006. The Scope of N egation. Doctoral dissertation, University of Siena. Muller, Ana, et. al. 2007. The Braz ilian Portuguese Quantifier "Todo" in Distributive and in Collective Contexts. DELTA 23:7195. Namiuti, Cristiane. 2008. Aspectos da histria gramatical do portugus: Interpolao, negao e mudana. Doct oral disseration, Universidade de Campinas. Nunes, Jairo (ed). 2009. Minimalist Essays on Brazilian Portuguese Syntax Amsterdam: John Benjamins. 232

PAGE 233

Nunes, Jairo and Mary Kato. 2009. A Uniform Raising Analysis for Standard and Non-Standard Relative Clauses. In Minimalist Essays on Brazilian Syntax ed. Jairo Nunes, 93120. Amsterdam: John Benjamins. Ordez, Francisco. 1997. Word Order and Clause Structure in Spanish and other Romance Languages Doctoral dissertation CUNY. Ordez, Francisco and Eduardo Trevio. 1999. Left-dislocated subjects and the pro-drop parameter : A case study of Spanish. Lingua 107: 39. Ouhalla, Jamal. 1990. Sent ential negation, Relativised minimality, and the aspectual status of auxiliaries. The Linguistic Review 7: 183231. Pereira, Iris. 2000. Para uma defini o do processo de concordncia negativa em portugus europeu. In XV Encontro Nacional da Associao Portuguesa de Lingustica : actas, ed. Rui Castro and Pilar Barbosa, 125149. Lisboa: Associao Portuguesa de Lingustica. Peres, Joo. 2000. Towards a Comprehensiv e View of Negative Concord. In Meaning Through Language Contrast ed. K.M Jaszczolt and Ken Turner, 2944. Philadelphia: John Benjamins. Pires, Acrisio. 2007. The subject, it is her e! The Varying Struct ural Positions of Preverbal Subjects. DELTA 23:113147. Pires, Acrisio and Heather Taylor. 2007. T he Syntax of Wh-in-situ and Common Ground. Paper presentation Penn Linguistics Colloquiu m 31, University of Pennsylvania, Feb 23-25. Poletto, Cecilia, 2000. The Higher Functional Field. Evidence from Northern Italian Dialects New York: Oxford University Press. Pollock, Jean-Yves. 1989. Verb Move ment, Universal Grammar, and the Structure of IP. Linguistic Inquiry 20:365424. Postma, Gertjan, and Wim van der Wurf f. 2007. How to say No and Don't: Negative Imperatives in Romance and Germanic. In Imperative Clauses in Generative Grammar: Studies Offered to Frits Beukema, ed. Wim van der Wurff 73110. Amsterdam: John Benjamins Prince, Ellen. 1981. Toward a tax onomy of given-new information. In Radical Pragmatics ed. Peter Cole, 223. Ne w York: Academic Press. Prince, Ellen. 1992. The ZPG letter: s ubjects, definiteness, and informationStatus. In Discourse Description: Diverse Analyses of a Fundraising Text eds. Sally Thompson and W. Mann, 295. Amsterdam: Benjamins. 233

PAGE 234

Radford, Andrew. 2004. Minimalist Syntax: Exploring th e structure of English. Cambridge: Cambrige University Press. Rezac, Milan. 2003. The Fine St ructure of Cyclic Agree. Syntax 6: 15682. Rivero, Mara-Luisa and Arhonto Terzi. 1995. Imperatives, V-movement, and Logical Mood. Journal of Linguistics 31:301. Rizzi, Luigi. 1982. Issues in Italian Syntax Dordrecht: Kluwer. Rizzi, Luigi. 1997. The Fine Structure of the Left Periphery. In Elements of Grammar ed. Liliane Haegeman, 281338. Dordrecht: Kluwer. Rizzi, Luigi, 2001. On the position Int(err ogative) in the left periphery of the clause. In Current Studies in Italian Syntax. Essays Offered to Lorenzo Renzi. ed. Guglielmo Cinque and Giampaolo Salvi, 287296. Amsterdam: Elsevier. Rizzi, Luigi (Ed). 2004. The Structure of CP and IP. The Cartography of Syntactic Structures, Vol. 2 New York: Oxford University Press. Roberts, Ian. 2004. The C-System in Brythonic Celtic Languages, V2, and the EPP. In The Structure of IP and CP, ed. Luigi Rizzi, 297328. Oxford/New York: Oxford University Press. Rodrigues, Cilene. 2004. Loss of Morphology and A-Mo vement out of Case Domains. Doctoral dissertati on. University of Maryland. Roncarti, Cludia. 1996. A Negao no Portugus Falado. In Variao e Discurso ed. Alzira Tavares de Macedo et al., 97112. Rio de Janeiro: Tempo Brasileiro. Ross, John. 1967. Constraints on Variables in Syntax. Doctoral dissertation, Massachusetts Institute of Technology. Rowlett, Paul. 1998. Sentential Negation in French New York/Oxford: Oxford University Press. Saeed, John. 1997. Semantics Oxford: Blackwell Publishers. Santos, Ana Lcia. 2004. How do children say Yes in European Portuguese?. In Proceedings of GALA 2003J acqueline, eds, van Kampen and Sergio Baauw. Utrecht: Utrecht University. Scherre, Maria Marta Pereira. 2007. As pectos sincrnicos e diacrnicos do imperativo gramatical no portugus brasileiro. Alfa 51:189222. 234

PAGE 235

Schwegler, Armin. 1988.Word-order Changes in Predicate Negation Strategies in Romance Languages. Diachronica 2:21. Schwegler, Armin. 1991a. Predicate N egation in Contemporary Brazilian Portuguesea Change in Progress. Orbis 34:187. Schwegler, Armin. 1991b. Negatio n in Palenquero. Synchrony. Journal of Pidgin and Creole Languages 6:165214. Schwenter, Scott. 2005. The pragmatics of negation in Brazilian Portuguese. Lingua 1 15:14271456. Schwenter, Scott. 2006. Fine-Tuning Jespersens Cycle. In Drawing the Boundaries of Meaning: NeoGricean St udies in Pragmatics and Semantics in Honor of Laurence R. Horn ed. Betty J. Birner and Gregory Ward. Amsterdam: John Benjamins. Schwenter, Scott and Glucia Silva. 2002. Overt vs. null direct objects in spoken Brazilian Portuguese: a se mantic/pragmatic account. Hispania, 85:577586. Silva, Glucia. 2001. Word order in Brazilian Portuguese : a minimalist analysis The Hague: Mouton de Gruyter. Silva, Claudia. 2004. A nat ureza de AGR e suas implicaes na ordem VS: um estudo comparativo entre o portugus bras ileiro e o portugus europeu. Doctoral dissertation, Universid ade Federal de Alagoas. Sportiche, Dominique. 1988. A theory of fl oating quantifiers and its corollaries for constituent structure. Linguistic Inquiry 19: 425449. Sousa, Arivaldo Sacramento. 2004. As estruturas de negao em uma comunidade rural afro-brasileira: Helvcia BA. http://www.vertentes.ufba.br/souza.doc Sousa, Liliana. 2007. Formas Reduzidas de Itens Negativos no Portugus Brasiliero. Masters thesis. Un iversidade Federal de Minas Gerais. Stalnaker, Robert. 1978. Assertion. In Pragmatics: Syntax and Semantics Volume 9 ed. P.Cole. New York: Academic Press. Stalnaker, Robert. 2002. Common Ground. Linguistics and Philosophy 25: 701 21. 235

PAGE 236

Szabolcsi, Anna. 2006. Strong vs. weak islands. In The Blackwell Companion to Syntax ed. Everaert and v an Riemsdijk. 4:479532. Thrinsson, Hskuldur 1996. On the (non)-universalit y of functional projections. In Minimal ideas ed. W. Abraham et. al., 25381. Amsterdam: John Benjamins. Travis, Lisa. 1984. Parameters and Effects of Word Order Variation. Doctoral dissertation, MIT Uriagereka, Juan. 1995. Aspects of the Syntax of Clitic Placement in Western Romance. Linguistic Inquiry 26: 79123. Uribe-Echevarria, Mara. 1992. On the St ructural Positions of Subjects in Spanish, and their Consequences for Quantification. In Syntactic Theory and Basque Syntax eds. Jos Lakarra and Julio de Ur bina, 447, San Sebastian: ASJU. van Riemsdijk, Henk and Edwin Williams. 1986. Introduction to the Theory of Grammar. Current Studies in Linguistics Cambridge: MIT Press. Vicente, Helena. 2006. O Quantificador Flutuante Todos no Portugus Brasileiro e no Ingls: Uma Abordagem Gerativa. Universidade de Brasilia. Vitral, Lorenzo. 1999. A negao: teoria da checagem e mudana lingstica. DELTA, 15:1. Wouden, Ton, van der and Frans Zwarts. 1993. A Semantic Analysis of Negative Concord. In Semantics and Li nguistic Theory III, 202219. Ithaca:Cornell University Press. Wouden, Ton, van der. 1997. Negative Contexts London: Routledge. Zanuttini, Raffaella. 1991. Syntactic pr operties of sentential negation. A comparative study of Rom ance languages. Doctoral disse rtation, University of Pennsylvania. Zanuttini, Raffaela. 1997. Negation and clausal structur e. A Comparative Study of Romance languages. New York, Oxford: Oxford University Press. Zanuttini, Raffaela. 2001. Sentential Negation. In The Handbook of Contemporary Syntactic Theory ed. Mark Baltin and Chris Collins, 511535. Malden, MA: Blackwell Publishers. Zeijlstra, Hedde. 2004. Sentential Negati on and Negative Concord. Doctoral dissertation, University of Amsterdam. 236

PAGE 237

Zeijlstra, Hedde. 2006. The B an on True Negative Imperatives Empirical Issues in Formal Syntax and Semantics 6 : 405424 Zeijlstra, Hedde. 2008. Negative Concord in Syantactic Agreement Ms. Zocca, Cynthia. 2007. And the wh goes where. Ms. University of Connecticutt. Zubizarreta, Maria Luisa. 1998. Prosody, Focus, and Word Order Cambridge, MA: MIT Press. 237

PAGE 238

BIOGRAPHICAL SKETCH Quinn McCoy Hansen was born in Salt Lake City, Utah. Before graduating from the University of Texas at Austin, he studied at universitie s in Hawaii, Mexico, Portugal and Spain. In May 2003, he graduated with honors with a degree in Spanish and Portuguese languages and literatures from the Univ ersity of Texas at Austin, College of Liberal Arts. He subsequently studied at t he Universidade Nova de Lisboa, Portugal where in 2005 he received the Grau de Me stre in linguistics for his work Basic Word Order in Kaqchikel which received the highest mark of muito bom After studying in Lisbon, Quinn moved to Aleppo, Syria wher e he was employed as a first grade teacher at the National School of Aleppo. He began his journey towards earning a Doctorate degree in linguistics at the University of Flor ida in August 2006. Since arriving at the University of Florida, he has taught Po rtuguese in the Department of Spanish and Portuguese, Quinn has also taught linguistics for the Program in Li nguistics. In 2009, he received the award for outstanding Portugues e Teaching Assistan t. In 2010, Quinn received a prestigious Graduate Student Teaching Award from the Univ ersity of Florida and the College of Liberal Arts and Sciences In the Fall of 2010, he will teach Portuguese at the University of Florida. 238