Linguistic Variation in a Border Town

MISSING IMAGE

Material Information

Title:
Linguistic Variation in a Border Town Palatalization of Dental Stops and Vowel Nasalization in Rivera.
Physical Description:
1 online resource (256 p.)
Language:
english
Creator:
Castaneda,Rosa M
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Romance Languages, Spanish and Portuguese Studies
Committee Chair:
Blondeau, Helene
Committee Co-Chair:
Pharies, David A
Committee Members:
Rothman, Jason
Burns, Allan F

Subjects

Subjects / Keywords:
contact -- portuguese -- sociolinguistics -- spanish -- variation
Spanish and Portuguese Studies -- Dissertations, Academic -- UF
Genre:
Romance Languages thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
This study focuses on the analysis of variation at the phonological level, specifically the variable realization of palatalization of dental stops before the high vowel /i/ and vowel nasalization in the speech of bilingual speakers of Uruguayan Portuguese (UP) in the city of Rivera, Uruguay. The data were collected in participant-observation and sociolinguistic interviews with fifty-four local UP speakers conducted by the researcher in the summers of 2006 and 2007. The study examines linguistic and extralinguistic factors influencing phonological variation. Variable rule analyses using GoldVarb X (Sankoff, Tagliamonte & Smith 2005) revealed that palatalization of dental stops is conditioned by linguistic and social factors. I implemented a trend methodology to compare apparent-time data from two studies at different points in time, 1995 and 2007. However, the hypothesized increase in the use of this variant over time among younger speakers is not supported in this study. Cross-sectional evidence indicates that palatalization of dental stops has stabilized at the speech community level. Results also indicate that palatalization of dental stops is a sociolinguistic marker, that is, sensitive to both linguistic and social context factors carrying both social interpretation and evaluation (Labov 1972). Vowel nasalization is a linguistic variable constrained mainly by morphophonological and sociophonetic factors. Phonological processes interact with affixation in inflectional and derivational morphology in the process of vowel nasalization. Results show that functional category of the word and preceding phonological environment play a significant role in the distribution of nasal variation in contemporary UP in Rivera.
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 Rosa M Castaneda.
Thesis:
Thesis (Ph.D.)--University of Florida, 2011.
Local:
Adviser: Blondeau, Helene.
Local:
Co-adviser: Pharies, David A.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2011
System ID:
UFE0043238:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 LINGUISTIC VARI ATION IN A BORDER TOWN : PALATALIZATION OF DENTAL STOPS AND VOWEL NASALIZATION IN RIVERA By ROSA MARIA CASTAEDA MOLLA 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 2011

PAGE 2

2 2011 Rosa Mara Castaeda Molla

PAGE 3

3 Para mi querida familia en Lima Per y en USA For all my dear friends

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank Dr. Hlne Blondeau, Dr. David Phar ies, Dr. Allan Burns, and Dr. Jason Rothman for serving on my supervisory committee. I am especially indebted to Dr. Blondeau for introducing me to the fascinating world of linguistic fieldwork and empirical soci olinguistics. I owe my gratitude also to Dr. Pharies for taking the time to serve as my co chair and for his continuous feedback. I would also like to thank Dr. Charles Perrone and Dr. Elizabeth Ginway for sharing with me their passion for Portuguese lan guage and culture. I am also very grateful for the institutional support I have had through the processes of fieldwork data collection and writing The Center for Latin American Studies awarded me the Tinker Field Research Grant, which allowed me to cond uct the preliminary field research for this study in Uruguay Brazil. The department of Spanish and Portuguese at the University of Florida granted me the Ernest G. Atkins Memorial Scholarship for Doctoral Research and Writing, which helped me return to the Uruguayan Brazilian border for the data collection phase of the study. I am most thankful for this indispensable financial support. I am indebted to my colleagues at Fort Hays State University for their continue d support and feedback. I am also grateful to Dr. Evie Toft, Dr. Amy Cummins, Dr. Pam Schaeffer, Dr. David Bov ee for editing and commenting on my work. I would also like to profoundly thank the people of Rivera for welcoming me in to their lives and for sharing with me their thoughts about life as a Riverense I would also like to extend my gratitude to the teachers, students and staff at the Centro Regional de Profesores (CERP) in Rivera for the invaluable support while conducting fieldwork. I extend my gratitude t o the faculty of the Universidad d e la Repblica in Montevideo,

PAGE 5

5 who guided me through my first trip to the border and provided me with invaluable information especially Dr. Elizaincin, Dr. Barrios, Prof. Behares and Prof. Gabbiani In Brazil I am indebted to a ll my friends in Rio de Janei ro and Bello Horizonte, especially Marcia Alvarenga for introducing me to the beautiful Bossa Nova music and teaching me my first words in Portuguese. T hanks als o to the Alvarenga family for the ir hospitality while visiting Brazil. Finally, thanks go out to my dear family in Lima, Per for their k indness and unconditional love. Special thanks go to my mother Antonieta, and my grandmother Martha, for teaching me how to be strong and determined Thanks to my fellow graduate students in Gainesville especi ally Annie, JuanPa and Alex fo r being such good friends and to my friends in Hays -Pelgy, Ti p, Lindsay, Stella, and Roberta -thanks to all of you for being there for me.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 10 LIST OF FIGURES ................................ ................................ ................................ ........ 14 LIST OF ABBREVIATIONS ................................ ................................ ........................... 15 ABSTRACT ................................ ................................ ................................ ................... 16 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 18 1.1 The Socio historical Context for Lang uage Contact ................................ ....... 19 1.2 Uruguay and the Brazilian Border ................................ ................................ .. 24 1.2.1 Economic and Social Aspects ................................ ............................... 25 1.2.2 Educational Policy in Uruguay ................................ ............................... 27 1.3 Statement of the Problem under Investigation and an Overview of Some Related Issues ................................ ................................ ............................... 32 1.3.1 Fronterizo ................................ ................................ ............................ 33 1.3.2 Research Questions ................................ ................................ ............ 39 1.4 Organization of the Study ................................ ................................ ............... 40 2 THEORETICAL FRAMEWORK ................................ ................................ .............. 41 2.1 Variation Theory ................................ ................................ ............................. 41 2.1.1 Theoretical Principles of t he Quantitative Framework ......................... 42 2.1.2 The Apparent Time Construct ................................ ............................. 44 2.2 Speech Community ................................ ................................ ........................ 46 2.3 Social Networks ................................ ................................ ............................. 49 2.4 Previous Studies of Linguistic Varieties in the Uruguayan Brazilian Border. .. 52 2.4.1 UP in Artigas City ................................ ................................ .................. 58 2.4.2 Brazilian Portuguese ................................ ................................ ............. 59 2.4.3 Uruguayan Spanish ................................ ................................ ............... 61 2.4.4 Studies on the Spanish in Rivera ................................ .......................... 62 2.5 Linguistic Variables under Investigation ................................ ......................... 63 2.5.1 Palatalizati on ................................ ................................ ......................... 6 4 2.5.1.1 The process of palatalization ................................ ................... 64 2.5.1.2 Palatalization in Portuguese ................................ .................... 65 2.5.1.3 Palatalization in UP ................................ ................................ 69 2.5.2 Nasalization ................................ ................................ ......................... 71 2.5.2.1 Vowel nasalization in Romance languages ............................. 72 2.5.2.2 Portuguese nasal vowel system ................................ .............. 72

PAGE 7

7 2.5.2.3 Phonemic interpretations of Portuguese nasal vowels ............ 76 2.5.2.4 Analysis of vowel nasalization in Portuguese .......................... 81 2.5.2.5 Variation in Brazilian Portuguese nasal vowels ....................... 89 2.5.2.6 Vowel nasalization in UP ................................ ......................... 91 3 METHODOLOGY ................................ ................................ ................................ ... 96 3.1 Quantitative Methods and Dialect Research ................................ .................. 96 3.2 The speech community of Rivera ................................ ................................ ...... 97 3.3 Fieldwork ................................ ................................ ................................ ........ 103 3.3.1 Preliminary fie ldwork ................................ ................................ ........... 103 3.3.1.1 An ethnographic approach ................................ ..................... 105 3.3.1.2 Entering the community via social networks .......................... 107 3.3.2 Data Collection Fieldwork ................................ ................................ .... 109 3.3.3 Type of Data Collection ................................ ................................ ..... 112 3.3.3.1 Individual inte rviews ................................ .............................. 114 3.3.3.2 Group interviews ................................ ................................ .... 115 3.3.3.3 Equipment ................................ ................................ ............. 116 3.4 T he Rivera Corpus ................................ ................................ .......................... 117 3.4.1 Speaker Selection Criteria ................................ ................................ 117 3.4.2 Speech Sample ................................ ................................ ................... 117 3.5 Variables under Investigation ................................ ................................ .......... 119 3.5.1 Linguistic Variables ................................ ................................ ............. 119 3.5.1.1 The variable palatalization of /ti/, /di/ ................................ ..... 120 3.5.1.2 The variable vowel nasalization ................................ ............. 121 3.5.2 Social Factors ................................ ................................ ...................... 122 3.6 Data Analysis ................................ ................................ ................................ .. 123 3.6.1 Extraction ................................ ................................ ............................ 124 3.6.2 Transcription ................................ ................................ ....................... 125 3.6.3 Coding ................................ ................................ ................................ 126 3.6.4 Quantitative Analysis ................................ ................................ ........... 127 3.6.4.1 GoldVarbX ................................ ................................ ............. 127 3.6.4.2 Binomial step up/step down analysis ................................ .... 129 4 QUANTITATIVE ANALYSES OF PALATALIZATION ................................ ........... 131 4.1 Background ................................ ................................ ................................ ..... 131 4.2 Research Questions ................................ ................................ ....................... 132 4.3 Circumscribing the Variable Context ................................ ............................... 133 4.4 Exclusions ................................ ................................ ................................ ....... 135 4.5 Linguistic Constraints ................................ ................................ ...................... 135 4.5.1 Syllable Stress ................................ ................................ ..................... 136 4.5.2 Following Phonological Environment ................................ ................... 137 4.5.3 Preceding Phonological Environment ................................ .................. 139 4.6 Social Factors ................................ ................................ ................................ 140 4.6.1 Age ................................ ................................ ................................ ...... 140 4.6.2 Socioeconomic Status ................................ ................................ ......... 142

PAGE 8

8 4.6.3 Gender ................................ ................................ ................................ 143 4.7 Results and Discussion ................................ ................................ ................... 144 4.7.1 Overall Distribution of Variants ................................ .......................... 145 4.7 .2. Factor by Factor Distributional Analysis ................................ ............. 147 4.7.3 Multivariate Analysis of the Contribution of Linguistic Factors of Palatalization of /ti/, /di/. ................................ ................................ ..... 149 4.7.3.1 Following phonological context ................................ .............. 152 4.7.3.2 Preceding phonological context ................................ ............. 157 4.7.3.3 Tonicity o f the syllable ................................ ........................... 159 4.7.4 Multivariate Analysis of the Contribution of Extralinguistic Factors of Palatalization of /ti/, /di/. ................................ ................................ ..... 163 4.7.4.1 Age ................................ ................................ ........................ 165 4.7.4.2 Socioeconomic group ................................ ............................ 170 4.7.4.3 Gender ................................ ................................ ................... 172 4.7.4.4 Cross tabulations of social factors ................................ ......... 174 4.8 Discussion ................................ ................................ ................................ ...... 180 5 QUANTITATIVE ANALYSES OF NASALIZATION ................................ ............... 184 5.1 Background ................................ ................................ ................................ ..... 184 5.2 Research Questions ................................ ................................ ....................... 185 5.3 Circumscribing the Variable Context ................................ ............................... 186 5.4 Exclusions ................................ ................................ ................................ ....... 188 5.5 Linguistic Constraints ................................ ................................ ...................... 188 5.5.1 Syllab le Stress ................................ ................................ ..................... 189 5.5.2 Following Phonological Environment ................................ ................... 190 5.5.3 Preceding Phonological Environment ................................ .................. 191 5.5.4 Syllable Structure ................................ ................................ ................ 191 5.5.5 Functional Category of the Word ................................ ......................... 192 5.6 Social Factors ................................ ................................ ................................ 193 5.7 Results and Discussion ................................ ................................ ................... 193 5.7.1 Overall Distribution of Variants ................................ ............................ 194 5.7.2 Factor by Factor Distributional Analysis ................................ .............. 194 5.7.3 Multivariate Analysis of the Contribution of Linguistic Factors of Vowel Nasalization in UP. ................................ ................................ 199 5.7.3.1 Functional category of the word ................................ ............. 203 5.7.3.2 Preceding phonological context ................................ ............. 210 5.7.4 Multi variate Analysis of the Contribution of Extralinguistic Factors of Vowel Nasalization in UP. ................................ ................................ 215 5.8 Discussion ................................ ................................ ................................ ...... 218 6 CONCLUSIONS ................................ ................................ ................................ ... 223 6.1 Research Questions ................................ ................................ ....................... 223 6.1.1 Palatalization ................................ ................................ ....................... 223 6.1.2 Vowel Nasalization ................................ ................................ .............. 231

PAGE 9

9 6.2 Methodological Conclusions ................................ ................................ ........... 236 6.3 Limitations and Recommendations for Future Research ................................ 237 6.4 Final Conclusions ................................ ................................ ............................ 2 38 APPENDIX A ................................ ............. 241 B MO DULES FOR THE SOCIOLINGUISTIC INTERVIEWS IN RIVERA ................. 243 C IRB APPROVAL OF PROTOCOL ................................ ................................ ........ 245 D IRB RENEWAL OF PROTOCOL ................................ ................................ .......... 246 LIST OF REFERENCES ................................ ................................ ............................. 247 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 256

PAGE 10

10 LIST OF TABLES Table page 1 1 Alternation between [a] and [ in Carioca and Belm dialect ............................ 37 1 2 Stressed and unstressed non low vowels in Belm and Carioca dial ects .......... 37 2 1 Palatalization of /ti/ and /di/ in EP and BP ................................ .......................... 66 2 2 Nasal vowels ................................ ................................ ................................ ...... 73 2 3 Nasalized vowels ................................ ................................ ................................ 74 2 4 Non derived nasal diphthongs in BP ................................ ................................ .. 75 2 5 Nasal d iphthongs derived by inflection in BP ................................ ...................... 75 2 6 Word internally nasal diphthongs ................................ ................................ ....... 76 2 7 Nasal vowels alternations in basic word and suffixation ................................ ..... 77 2 8 Rule 1: Stressed vowels followed by a nasal consonant ................................ .... 83 2 9 Rule 2: Stressed vowel followed by a nasal consonant in a closed syllable ...... 83 2 10 Sub rule 2.1: Deletion of nasal consonants in final position .............................. 83 2 11 ................................ ...... 84 2 12 Surface oral vs. nasalized vowels in BP ................................ ............................. 84 2 13 Nasalization of low vowels and raise to mid position ................................ .......... 85 2 14 Stressed nasal vowels ................................ ................................ ........................ 85 2 15 Pre stre ssed nasal vowels ................................ ................................ .................. 85 2 16 Rhotics after nasal and non nasal vowels ................................ .......................... 86 2 17 Base word derivations with prefix /in /, /im / before a consonant ........................ 86 2 18 Base word derivations with prefix /in / before a vowel ................................ ........ 87 2 19 Alternation bet ween nasal vowel/vowel plus nasal consonant in EP .................. 87 2 20 EP nasal diphthongs ................................ ................................ ........................... 87 3 1 Speech sample ................................ ................................ ................................ 118

PAGE 11

11 3 2 Palatalization linguistic factor groups and factors ................................ ............. 120 3 3 Nasalization linguistic factor groups and factors ................................ ............... 121 3 4 Social factor groups and factors ................................ ................................ ....... 123 4 1 Stressed /i/ ................................ ................................ ................................ ........ 133 4 2 Unstressed /i/ ................................ ................................ ................................ .... 134 4 3 Unstressed /e/ raised to [i] ................................ ................................ ................ 134 4 4 Palatalization of /t/, /d/ preceding t he oral vowel /i/ ................................ ........... 134 4 5 ................................ ........ 134 4 6 Palatalization of /t/, /d/ preceding the palatal glide [j] ................................ ........ 134 4 7 Palatalization of dental stops according to syllable stress among monolingual and bilingual speakers ................................ ................................ ...................... 137 4 8 Examples of palatalization of /ti/, /di/ accor ding to syllable stress ..................... 137 4 9 Examples of palatalization of /ti/, /di/ according to following phonological context ................................ ................................ ................................ .............. 138 4 10 Examples of palatalization of /ti/, /di/ according to preceding phonological context. ................................ ................................ ................................ ............. 139 4 11 Age distribution in each of three age groups for each speech sample ............. 142 4 12 Overall distribution of the realization of tokens of /ti/, /di/ in UP (2007 data) ..... 145 4 13 Overall distribution of the realization of tokens of /ti/, /di/ in UP (1995 data) ..... 145 4 14 Distribution of palatalization of /ti/, /di/ by syllable tonicity in UP (2007 data) ... 147 4 15 Distribution of palatalization of /ti/, /di/ by a following phonological segment in UP (2007 data) ................................ ................................ ................................ 148 4 16 Distribution of palatalizat ion of /ti/, /di/ by preceding phonological segment in UP (2007 data) ................................ ................................ ................................ 148 4 17 Variable rule analyses of the contribution of linguistic factors selected as significant to the probabil ity of palatalization of /ti/, /di/ in UP (2007 data) ........ 150 4 18 Variable rule analyses of the contribution of linguistic factors selected as significant to the probability of palatalizat ion of /ti/, /di/ in UP (1995 data) ........ 151

PAGE 12

12 4 19 Comparison of palatalization of dental stops according to syllable stress among bilingual group (Bisol 1991) and Rivera speakers (2007 data) ............. 160 4 20 Variable rule analyses of the contribution of social factors selected as significant to the probability of palatalization of /ti/, /di/ in UP (2007 data) ........ 164 4 21 Variable rule analyses of the contribution of social factors selected as significant to the probability of palatalization of /ti/, /di/ in UP (1995 data) ........ 165 5 1 Example of contrastive nasal vowels ................................ ................................ 186 5 2 Example of vowel + nasal in the same syllable ................................ ................ 187 5 3 Example of vowel + nasal in the following syllable ................................ ........... 187 5 4 Examples of vowel nasalization in UP according to syllable stress .................. 189 5 5 Examples of vowel nasalization in UP according to following segment ............ 190 5 6 Examples of vowel nasalization in UP according to preceding context ............ 191 5 7 Examples of vowel nasalization in UP according to syllable structure .............. 192 5 8 Examples of vowel nasalizatio n in UP according to functional category of the word ................................ ................................ ................................ .................. 193 5 9 Overall distribution of the realization of tokens of vowel nasalization in UP ..... 194 5 1 0 Distribution of nasalization by tonicity of the syllable in UP .............................. 194 5 11 Distribution of nasalization by following phonological segment in UP ............... 195 5 12 Distribution of nasalization by preceding segment in UP ................................ .. 196 5 13 Distribution of nasalization by syllable structure in UP ................................ ..... 197 5 14 Distribution of nasalization by functional category in UP ................................ .. 198 5 15 Variable rule analyses of the c ombination of linguistic factors selected as significant to the probability of nasalization in UP ................................ ............. 200 5 16 Overall distribution of the realization of tokens of vowel nasalization in UP ..... 202 5 17 Variable rule reanalysis of the combination of linguistic factors selected as significant to the probability of nasalization in UP ................................ ............. 202 5 18 Cyclic application of the stress rule in Portuguese ................................ ........... 207

PAGE 13

13 5 19 Variable rule analysis of the combination of extralinguistic factors selected as sign ificant to the probability of nasalization in UP. ................................ ............ 215 A 1 ................................ ............................. 241

PAGE 14

14 LIST OF FIGURES Figure page 4 1 Overall distribution of the realizations of tokens of pala talization of /ti/, /di/ in UP ................................ ................................ ................................ ................... 146 4 2 Probability of palatalization of /ti/, /di/ according to following phonological environment in UP. ................................ ................................ ........................... 153 4 3 Probability of palatalization of /ti/, /di/ according to syllable stress in UP. ......... 159 4 4 Comparison of rates of palatalization of dental stops followed by /i/ according to age in UP. ................................ ................................ ................................ ..... 167 4 5 Comparison of rates of palatalization of dental stops followed b y /i/ according to socio economic status in UP. ................................ ................................ ....... 172 4 6 Comparison of rates of palatalization of dental stops followed by /i/ according to gender in UP. ................................ ................................ ................................ 173 4 7 Cross tabulation of overall percentages of palatalization of /ti/, /di/ according to age and gender (2007 data). ................................ ................................ ........ 175 4 8 Cross tabulation of overall percentages o f palatalization of /ti/, /di/ according to gender and socioeconomic class (2007 data). ................................ ............ 178 4 9 Cross tabulation of overall percentages of palatalization of /ti/, /di/ according to socioeconom ic class and age (2007 data). ................................ .................. 179

PAGE 15

15 LIST OF ABBREVIATION S BP Brazilian Portuguese CERP Centro Regional de Profesores PBP Popular Brazilian Portuguese PP Prescriptive Portuguese SBP Standard Brazilian Portuguese UP Uruguay an Portuguese VBP Vernacular Brazilian Portuguese

PAGE 16

16 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 LINGUISTIC VARI ATION IN A BORDER TOWN: PALATALIZATION OF DENTAL STOPS AND VOWEL NASALIZATION IN RIVERA By Rosa Mar a Casta eda Molla August 2011 Chair: Hl ne Blondeau Cochair: David Pharies Major: Romance Languages This study focuses on the analysis of variation at the phonological level, specifically th e variable realization of palatalization of dental stops before the high vowel /i/ and vowel nasalization in the speech of bilingual speakers of U ruguayan P ortuguese (UP) in the city of Rivera, Uruguay. The data were collected in participant observation and sociolinguistic interviews wit h fifty four local UP speakers conducted by the researcher in the summer s of 2006 and 2007. T he study examines linguistic and extralinguistic factors influencing phonological variation Variable rule analyses using GoldVarb X (Sankoff, Tagliamonte & Smith 2005) revealed that palatalization of dental stops is conditioned by linguistic and social factors. I implemented a trend methodology to compare apparent time data from two studies at different points in time, 1995 and 2007. However, the hypothesized increase in the use of this variant over time among younger speakers is not supported in this study. Cross sectional evidence indicates that palatalization of dental stops has stabilized at the speech community level. Results also indicate that palatalization of dental stops is a

PAGE 17

17 sociolinguistic marker, that is, sensitive to both linguis tic and social context factors carrying both social interpreta tion and evaluation (Labov 1972 ). Vowel n asalization is a linguistic variable constrained mainly by morphophonological and sociophonetic factors. Phonological processes interact with affixation in inflectional and derivational morphology in the process of vowel nasalization. Results show that fu nctional category of the word and preceding phonological environment play a significa nt role in the distribution of nasal variation in contemporary UP in Rivera.

PAGE 18

18 CHAPTER 1 INTRODUCTION The present study investigates a variety of Portuguese spoken in Ri vera, Uruguay, namely Uruguayan Portuguese (UP). UP is a stigmatized minority variety spoken in several bilingual communities along the Uruguayan Brazilian frontier since colonial times (Elizaincin 1997) The research site is the city of Rivera, the larges t bilingual community along this borderland. While research on UP was initiated in the 1960s by Rona, the first sociolinguistic study on the Uruguayan Brazilian border was carried out by Hensey in 1972. The current research focuses on the analysis of varia tion at the phonological level, th e variable realization of palatalization of dental stops before the high vowel /i/ and vowel nasalization in the speech of bilingual speakers of UP For the analysis of palatalization of /ti/, /di/ I implemented a trend me thodology to compare apparent time data from two studies at differen t points in time, 1995 and 2007. Based on data collected in participant observation and sociolinguistic interviews with fifty four local UP speakers, the study examines linguistic and extr alinguistic factors influencing phonological variation. Though mainly monolingual and Spanish speaking, Uruguay contains enclaves of native Portuguese speakers. The northeast part of the country was settled by Portuguese peasants who remained in Uruguaya n territory after the borders were the kingdoms of Spain and Portugal, as far back as to the arrival of the first Portuguese and Spanish explorers in the New World. Nowad ays, the border between these two countries represents a language contact situation with intricate historical, cultural, and

PAGE 19

19 linguistic ties. Thus, a good understanding of the history geography, and social aspects of the border is necessary in order to co mprehend fully the complex social factors that have given rise to the formation of UP. 1.1 The Socio historical Context for Language Contact Spanish and Portuguese languages have had a strong presence in the Uruguayan Brazilian border since the early 1500s wh en the first Portuguese and Spanish explorers arrive d in the New World. Attempts at colonization of the region and the political treaties that ensued determined in many ways the linguistic inheritance of this disputed region. In addition, economic issues and geography played a crucial role in the struggle for the territory, the delimitation of territories, and nations (Lipski 1 994). The Treaty of Tordesillas (1494) was the major decisive factor in the initial expansion of the Spanish and Portuguese langu a ges in the region. Under the T reaty of Tordesillas, Spain and Portugal set up an imaginary north south line dividing the western hemisphere between them into separate zones of exploration, influence, and commerce. The placement of the line of demarcation 370 leagues west of the Cape Verde Islands 1 allowed the Spanish crown to acquire the title of the western part of the continent. Portugal acquired all territories to the east of the line. This demarcation was progressively pushed much further to the west in the course of subsequent events, allowing the Portuguese to colonize the region in the first decade of the XVI century 2 Moreover, in order to deter the French from competition for dyewood and to s top their 1 The islands are located off the west coast of Africa. 2 In 1494, the year in which Spain and Portugal concluded the Treaty of Tor desillas, the geography of the western Atlantic shores was unknown. Thus, at that time there was no suspicion that the demarcation line would give Portugal territory in America (Bakewell 2004).

PAGE 20

20 threat to the safety of the Portuguese fleets Portugal secure d the Brazilian coast by establishing permanent settlements (Bakewell 2004:322). Exploration of the vast new coast co ntinued throughout the XVI century. During the XVII century the bandeirantes 3 began their journeys to the interior of to For Bakewell the frontier expanding bandeirantes might be called the makers of Brazil, because it wa s their roaming to these lands territories west of the Tordesillas line Along with the exploration s inland, cattle were moved into the serto or backlands, long before the flood of gold seekers took place in the early XVII I century. As Bakewell attests, were an important instrument of the European occupation of the land in Brazil, as they were also in the pampas further to (2004: 348 ). A s colonial population grew, cattle raising spread inland from the coast to several areas of the country, including the lowest reaches of Brazil, t he Rio Grande do Sul of today. The first Spanish conqui stadores arrived in Uruguay for the first time in 1516 with the Juan Diaz de Solis expedition. The Spanish explorers did not receive a friendly welcome from the Charras Indians 4 The Charras offered strong resistance to the Spanish explorers, killing D iaz de Solis and rejecting any other attempt of conquest. The Solis expedition found neither the gold nor the interoceanic route that they were Sierra de la Consequentl y, this region remained forgotten and indifferent to the Spanish crown for almost two centuries. The Banda Oriental original denomination of the 3 The bandeirantes were members of the XVI to XVIII century South American slave hunters and gold seekers explorations (Bakewell 2004). 4 Refer to Uruguayan indigenous group who resisted Spanish conquest for several centuries and remained as a community well into the nineteenth century (Lipski 1994:339).

PAGE 21

21 present time Repblica Oriental del Uruguay tierras de 10). Neverthele ss, at the beginning of the XVII century, livestock in this region had become increasingly important due to the fertile pastures. This situation attracted Argentinean G auchos 5 to the Banda Oriental who establishe d the marketing of hides and opened business in this area (Lipski 1994:338) Suddenly, the Banda Oriental became extremely important for the Spaniards, who this time were i n search of (Elizaincin 1997). Meat and hide businesses were also very attractive resources for the Portuguese. This common commercial interest added to the already long dispute over the delimitation of this area of America created a point of conflict between the two countries. According to Elizaicin (1997) during colonial times, these borderline areas constituted an open and deserted zone where Spanish, Portuguese, and criollos 6 performed unlawful activities commercializi ng cattle goods. There did not exist a well defined borderline in the area. The passage from present day Rio Grande do Sul in Brazil to the Banda Oriental only had control to limit livestock smuggling. Portugal, recognizing the economic advantages of controlling the waterways of the Uruguay River and the estuary of th e Rio de la Plata, attempted to push its limits south and west. The Portuguese established their first permanent set tlement on the Uruguayan coast, founding the Nova Colnia do Sacramento on the Ro de La Plata estuary in 1680. However, for nearly a hund red years thereafter this t erritory was 5 Gauch o initially refers to a vagabond and cattle thief but later came to mean the farm hand and warrior normally associated with the figure of the horseman. Nowadays the term is used as a patronymic for the Rio Grande do Sul citizens. 6 n of European ancestry born in the New World.

PAGE 22

22 disputed. To repe l Portuguese advances and to defend the territory, the Spanish founded the city of San Felipe y Santiago de Montevideo in 1724 (Elizaincin 1992, 1997). Moreov er, in 1800 with the sole aim of stop p ing Portuguese immigration to the area, viceroy Lastarra allowed Castilian and Austrian immigrants to settle i n the borderline along Brazil. The primary objective of the establ ishment of cities was military defense Montevideo was considered strategic in the Spanish defense of the La Plata region. At this time, the Spanish crown decided to support the colonization efforts in this region with overseas elements: twenty five C anary I sland families and a s econd contingent that arrived in the New World in 172 9 More Spaniards arrived later from o ther regions of the Peninsula. Nevertheless, t he Banda Oriental remained almost deserted through the XVIII century (Elizaincin 1997). At the end of the XVIII century, Spain, then a French ally, went to war with Englan d, and the British invaded Buenos Aires. One of the ways to undermine Spanish power was to attack Spanish territories. Consequently, the British attempted to seize control of the Spanish colonies located around the Plata Basin and the city of Buenos Aire s was occupied in 1806. Montevideo, where the British maint ained military reinforcements, was occupied for several months in 1807. Then, Buenos Aires established a government independent from the Spanish crown, anticipating the eventual declaration of in dependence in 1816. This independ ence war was a joint effort of Argentines and inhabitants from the Banda Oriental fighting the Spanish rule. The Uruguayan national hero Jos Artigas fought to achieve Uruguayan self rule and to avoid annexation to Argent ina (Bakewell 2004).

PAGE 23

23 Lat er on, the territory of present day Uruguay was contested during t he independence period between Buenos Aires and Brazil. For a time Uruguay was further solidifi cation of Portuguese in the northern part of Uruguay (Lipski 1994: 338). During a period of uncert ainty in the 1820s, the region was administered at different times by bo th countries. A n agreement made between Brazil and the Argentine Fed eration, at British urging, c reated the sovereign nation of Uruguay in 1828 (Bakewell 2004: 392). It was not until the end of the XIX century that Uruguay finally achieved some kind of political and social stability after repelling several Argentine invasi ons ordered by the dictat or Juan Manuel de Rosas (Lipski 1994). For Barrios (1995), the independence of Uruguay was not the result of political action by a nation that was homogenous and defined linguistically and culturally. On the contrary, Uruguay was constituted as the union of two regions with different cultural and linguistic traditions; consequently the newly independent nation struggle d for cultural autonomy. On t he other hand, the government started a campaign to promote the nationalization of n orthern Uruguay. One of the ways of achieving this g oal was the foundation of t win cities adjacent to existing Brazilian towns. The objective was to limit Brazilian influence and to halt the use of the Portuguese language. On May 7 th of 1862, Villa Ceba llos, later renamed Rivera, was founded by government decree, alongside the Brazilian town of Santana do Livramento (Elizaincin 1976). Along with the creation of border towns, strong language policies were implemented Uruguayan language planning was th en based on the model of a monolingual nation establishing in 1877 the Ley de Educacin Comn that made general education obligatory and Spanish the

PAGE 24

24 language of instruction for the whole country. Despite the existence of a rich linguistic tradition in the frontier, the educational linguistic planning of the border communities followed the national monolingual model. 1.2 Urugua y and the Brazilian B order Known officially as the the country has long been called the Banda Orienta l of the River Plate. Sometimes Uruguayans refer to themselves as Orientales. Uruguay is bounded on the west by Argentina, on the north and northeast by Brazil, and on the southeast by the Atlantic Ocean. On the south it borders the Rio de la Plata, a b road estuary located between Argentina and Uruguay. The Rio de la Plata estuary provides access to the Rio de la Plata and the Rio Uruguay, the main navigable rivers in the area. Montevideo, the capital and major port, sits on the banks of the Rio de la Plata. Uruguay is the smallest Spanish speaking country in South America. Its population is 3.3 million people (2006) of whom 1.3 million live in the capital Montevideo. 7 conflict between the kingdoms of Spain and Portugal, and between two linguistic poles, one located in Buenos Aires, capital of the Spanish vice royalty, and the other in Rio Grande do Sul (Elizaincin 1996). The political boundaries between Uruguay and Brazil extend all t he way from the town of Bella Union to the confluence of the Chu y River with the Atlantic Ocean. This borderline is abo ut one thousand kilometers long and is not always defined by natural or manmade landmarks. Throughout this long stretch, 7 Source: Instituto Nacional de Estadstica, Uruguay 2006.

PAGE 25

25 estan cias, faze ndas cattle ranches alternate with urban settlements even toda y (Elizaincin 1976: 124). Uruguay is divided into departamentos departmento is divided into provincias The border departamentos are Artigas, Rivera, Cerro Larg o, and Rocha. The border population is concentrated in a set of twin cities, on each side of the Uruguaya n Brazilian border. The departmento of Rivera, roughly equivalent to a state or province, is located in northeastern Uruguay The city of Rivera is t he capital of the departamento of Rivera and is located 500 kilometers north of the n ational capital of Montevideo. southernmost state in Brazil, while Rio Grande do Sul is located dire ctly across from Rivera. Santana do Livramento is located 500 kilometers southwest from its state capital Porto Alegre The presence of F ronterizo ( § 1.3.1) has been documented in these areas but also in cities away from the borderlands, such as Tacuarembo Salto and Treinta y Tres (Elizaincin et al. 1976). Nevertheless the northern communities of Artigas and Rivera are where the influence of Portuguese language and culture is considered more profound. According to Hensey (1972), the gradual transition fr om Standard (Rio Grande) Portuguese to standard (Uruguayan) Spanish takes place wholly within parallel zones of Portuguese penetration, strongest at the border and d ecreasing southward, with a phasing 1.2.1 Economic and Social Aspects The Uruguayan economy relies heavily on trade, particularly of agricultural agricultural land. Agriculture and agro

PAGE 26

26 industry account for 23% of the gross domestic product (GDP), and for over two thirds of total exports. The main industries include meat processing, agribusiness, wood, wool, leather production, apparel, textiles, and chemicals. Uruguay is a founding member of Mercado Comn del Sur, (Southern Common Market) Mercosur 8 which is headquartered in Montevideo. Mercosur was established as an initiative of Argentina and Brazil, longstanding rivals in South America, again st protectionism and import substitution. Mercosur is also an attempt to foster cultural integration and geo linguistic dynamics in the whole region. Its purpose is not only to promote free trade and the fluid movement of goods, people, and currency but a lso to support educational integration, taking into consideration cultural and lingui stic factors (Hamel 2003:115). Rivera and Santana share the same main economic activities such as ranching as come to rely on duty free shops, where imported products are found at lower prices. Duty free shopping was established in 1986 in Rivera with the idea of attracting Brazilian tourists to the area. On the other hand, Santana has a bigger and stron ger ec onomic base than Rivera; Rio Grande do Sul is one of the most prosperous Brazilian states. Its main economic activities are grain production, ranching, and viticulture. The industrial sector is the biggest component of GPD at 42.6% followed by the servic e sector at 41.1%. Agriculture represents 16.3% of GPD (IPEA 2005). 9 8 The Southern Common Market founded in 1991 comprises Argentina, Brazil, Uruguay, and Paraguay as member states. Chile and Bolivia have the status of associated members. As the largest member, Brazil carries much of the weight in the Mercosur economy. Brazil represents 71% of the GDP, 71% of the territory and 78.7% of the population 9 Instituto de Pesquisa Economica Aplicada.

PAGE 27

27 Due to its proximity to the gigantic economic power of Brazil and its easy access, community residents cross the border to purchase groceries and household items at substantially reduced prices. Small merchants take advantage of this situation as well, acquiring considerable amounts of everyday products to stock their stores or to sell them at public markets. The flow of goods and the fluctuation of currency values have been a pattern of life in the border. Both Riverans and Santana residents take advantage of changes in economic conditions and the disparate value of the currencies of both countries. Another interesting economic activity is the presence of camels or street vendors. Street vendors line up right along the P arque Internacional (International Park ) between Rivera and Santana, offering all sorts of local and foreign goods, including movies, music, clothes, electronics, toys, etc. This situation puts street merchants at t he center of commercial exchanges and negotiations providing a potential locus for language contact. 1.2.2 Education al Policy in Uruguay Portuguese language in northern Uruguay has had a long and conflictive history that can be traced back to the time of the dominion of the Kingdoms of Spain and Portugal in the New World. While Buenos Aires and Montevideo share linguistic, historical and cultural factors, northern Uruguay maintains linguistic, historical, and cultural ties with its northern neighbor. Pri mary education in Uruguay, financed by the the highest in Latin America, and enrollment in free secondary education is also high. Nevertheless, with the downturn in the ec onomy, this situation has changed.

PAGE 28

28 Nowadays, private education for those who can afford it is favored over poorly funded public education institutions. Historically, Uruguayan language policy has consistently opposed the presence of Portuguese language in the territory. The Uruguayan government deemed the presence of Luso speakers in the country as a threat to national sovereignty and a sign of the cultural expansionism of their historical rival. It is in this context that in 1877 the Law of Common Educ ation established obligatory elementary school education in Spanish for the whole country (Behares and Gabiani 1987). As Hamel argues, the attempt was to assimilate the Portuguese speaking population using public education as the main vehicle (2003:121). Through the establishment of language policies, the government identity (Church 2007:11). As a consequence of this initiative, Uruguayan Portuguese speaking children did poorly in Spanish only schools. Thus, in 1967 a study was commissioned by the Uruguayan National Council of Primary and Secondary Educatio n to investigate the problem. The Council assumed that there were high levels of dyslexia in the school populations along th e bilingual border. After unsuccessful attempts to solve the assumed problem, the Council hired a specialist from the Inter American Children was due to the existence of a high proportion of children who did not speak Spanish as their mother tongue (Garca Etchegoyen 1975). Nevertheless, the Council and several other governmental agencies concerned about the educational problem concentrated their efforts around nationalis

PAGE 29

29 during the Uruguayan military dictatorship years 10 (1973 to 1984), when UP was deemed to be anti national (Behares 2004). T hroughout t he subsequent years, the government attempted not only to promote Spanish language monolingualism but to advance its nationalistic agenda by opposing the presence of Portuguese language along the border. Thus, the promotion of Spanish language monolingual ism was perceived as a tool in the construction of national identity. Along with the nationalistic discourse about the linguistic situation on the frontier, the military regime sponsored a number of publications, issued by educational entities, fostering the Spanish language as the language of national identity and sovereignty. Furthermore, these initiatives were supplemented by the creation of more schools and an increased number of Spanish language classes, the creation of elevised Spanish courses, and newspaper articles promoting Spanish as the state language. This nationalistic propaganda was concentrated along Uruguayan Brazilian border cities in an attempt to educate its inhabitants and eradicate their linguistic and cul tural heritage ( Academia Nacional de las Letras 1982). Nevertheless, with the establishment of Mercosur whose objectives go far beyond trade, different dynamics between Spanish and Portuguese were established 11 One of Mercosur main objectives is to a ttain regional integration and strengthen the international position of its member countries. It is in this context that the Commission 10 During the military di ctatorship the promulgation of the exclusive use of Spanish was extended to every aspect of public life (Elizaincin 1979). 11 At the beginning of the nineteenth century Spanish was present in many Brazilian areas such as in the educational system and abund ant Hispanic literature. In a sense, a Hispano American community was postulated rather than Ibero American unity. As Hamel discusses, Hispanic countries maintained a historical barrier in the face of Portuguese (2003: 118).

PAGE 30

30 of Ministers of Education approved the Triennial Plan for the Educational Sector of Mercosur in 1992. This plan conside rs educational and cultural integration to be prerequisite for all economic and political integration (Hamel 2003:129). In addition, it provides for the teaching of the two official languages of Mercosur : Spanish in Brazil and Portuguese in Argentina, Para guay and Uruguay. 12 After a long linguistic policy struggle in the country, a bilingual pilot program was implemented in the city of Rivera in 2003. Portuguese language instruction is now s implementation could not have been possible outside the Mercosur context. Under Mercosur linguistic policies, member states of the economic bloc have agreed to adopt Spanish and argues that: Argentina, Paraguay and Uruguay who have experienced the presence of Portuguese inside their borders as a threat (especially Uruguay, because of the historical presence of Portuguese dialects within its territory) now find themselves in need of promoting the Portuguese language, which they so battled with before, as a language of integration within their own territory (2003:128). It may be worth pointing out the new attitude toward the teaching of foreign languages instituted since the incursion of Mercosur Before the implementation of the economic bloc, the teaching of Portuguese language was seen as a threat, and as such was excluded from publ ic education. The rationale behind it was that Portuguese was a menace to Uruguayan sovereignty and linguistic homogeneity (Barrios 1999). 12 The plan establishes also th e development of a favorable awareness among the citizenry concerning integration, and trains human resources so as to contribute to development, and it also makes educational systems compatible and harmonious (Hamel 2003:129).

PAGE 31

31 The transition from language homogenization to bilingual education is being implemented, at least at the local lev 13 The program started in 2003 with two schools in Artigas and Rivera. The Uruguayan bilingual education program consists of two modalities: partial immersion in English and dual immersion in Spanish and Portuguese. In the frontier area, the implementation has been basically through the dual program in Spanish and Portuguese Under this modality, students with Spanish and UP backgrounds comprise a group and receive instruction in both languages. Each group of stud ents has two teachers, one Spanish speaking and one UP speaking. Both teachers are in charge of the school curriculum in their language specialty. In this section I have described geographic, economic, social and educational aspects pertaining to the U ruguayan Brazilian order. I have briefly discussed relevant economic features of its neighboring country, Brazil, whose superior economic power brings better employment and educational opportunities discussed above, the twin cit ies of Rivera and Santana share a longstanding historical and cultural tradition. Current dynamics of integration and globalization, at least at the local level, seem to reinforce traditional interactions and systems of communication (Hamel 2003). The es tablishment of Mercosur whose objectives go far beyond trade, has created new language dynamics on the border. One example of the new language dynamics on the Uruguayan Brazilian border is the implementation of a bilingual pilot program in the city of Riv era in 2003. The sociolinguistic implications of the teaching of 13 A dual immersion program i s a type of bilingual education. It main characteristic is that it imparts education to students with different mother tongues in the same classroom (ANEP personal communication).

PAGE 32

32 the two official languages of Mercosur, Spanish and Portuguese, and its repercussions in the speech of Riverans will be disc ussed later. In what follows, I discussed the linguistic variable s under examination and provide an overview of some of the related issues. 1.3 Statement of the Problem under Investigation and an Overview of Some Related Issues For this investigation my objective is to c onduct a classic sociolinguistic study of phon ological variation in the variety of Uruguayan Portuguese spoken i n the city of Rivera, Uruguay. The proposed investigation is based on sociolinguistic interviews conducte d in Rivera The participants are UP bilingual speakers born and raised in this city The corpus is composed of male and female speakers of different socioeconomic status and from three generations. The analysis in this study relies solely on the output of the sociolinguistic interviews with UP speakers. A specific goal in this type o f studies is to gain access t 2006 :8 ). With this goal in mind, I conducted two fieldwork trips to the city of Rivera during the summers of 2006 and 2007. During the first phase of fieldwork I performed intensive participant observation activities and assessed the feasibility of sociolinguistic research. The data collection, consisting of 106 sociolinguistic interviews, was conducted during the second fieldwork trip during the summer of 2007. The phonological variables selected for analysis are: The realization of /t/ and /d/ before / i / as a dental stop or as an affricate and the absence or presence of vowel nasalization in Uruguayan Portuguese. The two varia bles under investigation are coded for both internal (linguistic) and external (social) factors. Along with an analysis of the

PAGE 33

33 sociolinguistic distribution of palatalization of /ti/ /di/ and vowel nasalization in UP, I will compare my palatalization result s with those The linguistic variables chosen for investigation were selected on the basis of two factors: First, it was clear from the preliminary fieldwork conducted in the area that there is significant variation in these form s in local speech. Second, the variables are high in frequency and therefore likely to produce a sufficient number of tokens with which to perform multivariate analysis. The following section provides background information meant to explain the richness o f the language contact situation in the Uruguayan Brazilian frontier. It chronicles the trajectory of research conducted in the area, dating back to the earliest studies initiated in the decade of the 1960s by Rona and modern quantit ative variation analys is (e.g. Carvalho 1998, 2003). An understanding of how contemporary UP variety evolved, under complex historical, geographical, and social circumstances is necessary to achieve this goal. 1.3.1 Fronterizo As a consequence of the extensive coexistence of Spanish and Portuguese speaking populations along the border of Uruguay and B razil, a situation of cultural and language contact exists. This long standing cultural and language contact has had linguistic impact not only in the northern states of Urug uay but also in the southern cities of Rio Grande do Sul, Brazil. Although this area has also welcomed European immigrants (large communities of Germans, Italians, Poles and Ukrainians have settled cally as (luso) brasileiros and uruguayos (or castellanos) (Hensey 1972:1). The speech of southern Brazil is known

PAGE 34

34 as Portuguese Gaucho and the variety spoken along northern Uruguay is known as F ronterizo. The term F ronterizo has been used to refer to th e orthern states departamento s Rivera, Cerro Largo Rocha and Artigas, along the border with Brazil. (Elizaincin et al. 1987:12). Further, Uruguay an scholars have classified F ronterizo as a dialect of Portuguese and proposed the term dialectos portugueses del Uruguay 14 Uruguayan Portuguese has been characterized as a dialect in free variation (Elizaincin 1992) and as presenting monost ylistic variation (Behares 1984 that UP stylistic and sociolinguistic variation exist and follow systematic patterns of diffusion (1998:142). The long tradition of Portuguese languag e and culture i n Uruguay has been attested in several descriptive studies (Rona 1965, Hensey 1972, 1980, 1982, Elizaincin 1979, 1981, 1992, 1995, 1996, 19 97, 2004, and Carvalho 1998, 2004). However before Carvalho (1998), social and stylistic variation in Uruguayan Portuguese, as well as the social significance of language choice between Portuguese and Spanish, ha d not been extensively investigated (Carval ho 1998). scale sociolinguistic analysis of linguist ic and social factors that conditio n variability in the use of UP phonological forms. Her study draws from a database of 54 bilingual speakers from Rivera, Uruguay. The study analyses the sociolinguistic distribution of two phonological variables: the 14 of variability of the dialect along the Uruguayan Brazilian border.

PAGE 35

35 li nguistic realization of the palatal liquid / / and the linguistic realization of /t/ and /d/ in /i/ socioeconomic group were the two most important factors, and age was the least important, making this variable a soci olinguistic marker in the speech community of Rivera (Carvalho 1998:178). Results for p ala talization of /ti/ /di/ demonstrate that age is the most important factor followed by socioeconomic class. VARBRUL results show a sharp difference between the younges t group (16 to 29 years old), with a probability weight as high as .91, and the oldest group (59 to 70 years old), with a low probability value of .05 (Carvalho 1998:179). Previous de scriptions of UP dialect contain little or no mention of palatalization ( Rona 1965 Hensey 1972, Lipski 1994 ). Carvalho (1998) argues that the application of the palatalization rule in UP illustrates her theory of dialectal diffusion in Rivera. She further posits that dialect diffusion and the resulting linguistic change in pr ogress in UP are made possible through greater contact with the target dialect, Urban Brazilian Portuguese ( UBP ). Greater contact has been caused by social changes during the last 30 years in Rivera. Further, Carvalho argues that t he impact of television and urbanization opened the relatively isolated border rural community to a new social and linguistic model with which they had little in common ( Carvalho 1998:189). The selection criteria of palatalization of /ti/, /di/ for reexamination are based on pre vious research. Strong quantitative apparent time evidence exists showing that palatalization of /ti/, /di/ is undergoing change (Carvalho 1998, 2004) while real time data suggest little or no dental palatalization ( Rona 1965 Hensey 1972, Lipski 1994 ).

PAGE 36

36 phonological variable that is frequent, highly stratified, and easy to spot and palatalization of /ti/, /di/ is variable across varieties of Portuguese. As Hensey (1972) and Bisol (1991) furthermore argue, the conservative pronunciation of dental stops in Rio Grande do Sul is receding. These facts make palatalization of /ti/,/di/ an interesting linguistic variable to investigate. It is relevant also to determine if this is indeed a variable undergoing change and what stage of the proc ess it has currently reached. By reexamining the speech community of Rivera, it is feasible to ver ify whether the change in progress hypothesized by Carvalho (1998) has continue to advance or whether it has stabilized at the speech community level. Thus, palatalization of /ti/,/di/ is an excellent candidate for reexamination. As for the treatment of t he second linguistic variable under analysis, the absence or presence of vowel nasalization in UP, it is relevant to point out that nasal sounds are a distinctive characteristic of Portuguese. Portuguese is one of the two standard Romance varieties having nasal vowels as indepe ndent phonemes (Sampson 1999:1 75). Unlike in French, where the alternations between nasal vowels and oral vowels are clearly indicated, in most dialects of Portuguese, vowel nasalization occurs more or less generally whenever a vow el is followed by a nasal consonant, regardles s of syllable structure (Lipski 1975:67). Shaw (1986) reports variation in nasaliz ation across dialects in Brazilian Portuguese. In the varieties of Belm, Par, and of Recife, Pernambuco, low vowels (viz., /a/) (stressed and unstressed) are heterosyllabically n asalized, but in Carioca dialect there is a morphonological alternation between [a] and [ ] (examples from Shaw

PAGE 37

37 1986:83). As Shawn further points out, although there is a correspondence between stressed [ ] and [a] in the Carioca dialect, realizations of an unstressed low nasal vowel Table 1 1. Alternation between [a] and [ ] in Carioca and Belm dialect Glossary cana [ na] canavial [k can e field (Carioca variety) banana [b na] [b a na ] Carioca variety) Variation in the linguistic realization of the high vowel /u/ has also been documented, as in: fumo nico stressed an d unstressed non low vowels are not nasalized (in citation form), that is, they are not p erceived as nasal (examples from Shaw 1986:83). Table 1 2. S t ressed an d unstressed non low vowels in Belm and Carioca dialects Glossary cmulo tmulo renegado cmico minado Previous phonet ic representation accounts have dealt with vowel nasality indirectly as a subsection of some other phonological issue, s uch as stress, or development of morphophonological, ph onetic, or dialectal variation or altern ations are ignored (Shaw 1986: 6). Thus, the description of the distribution, alternation and variation of nasal vowels is of paramount importance.

PAGE 38

38 Rona (1965:35) finds nasal vowels in UP but does not provide any quantitative data. H e remarks that in the Melense variety, Portuguese words ending in o and Spanish words ending in n have yielded only one ending o (1965:43). The other three F ronterizo rtuguese diphthong o Hensey (1982:15) argues that /a/ resu lts show that Riverans nasalize /a/ as a low vowel in 70% of the cases. For Li of a given F ronterizo ( 1994: 343). Azevedo ( 1981:23) claims that there are individual and dialectal variation s of nasality in Brazilian Portugues e. However there is no cohesive descrip tion of these phenomena. Thus, the analysis of nasalization in UP dialect will shed interesting light on this linguis tic process in particular and on the nature of language variation and change in general. In addition, the sociolinguistic approach of this research will account for dialectal variations from spontaneous and informal speech, a method that is not commonly used to describe the nasaliz ation process. need for extensive and accurate data collection in this area, an d not only should this data reflect the slow, deliberate speech of citation forms, but it must also come from 1986: 87). In addition, this study r esponds to a current need for a phonolo gical analysis of vowel nasaliza tion variation in UP. Therefore, the sociolinguistic analysis of these phenomena in UP represents a challenging and exciting area of linguistic inquiry since

PAGE 39

39 this phonological process has not been fully investigated from a language in contact and variation ist perspective. In this analysis I will show how language choice and so ciolinguistic attitudes account for different meanings such as linguistic meaning and social meaning with reference to both external (social) and internal (systemic) factors in the dis cussion section of this study. Additionally, issues related to the nature of linguistic attitudes, language c hoice, and identity formation of UP speakers will also be discussed. 1.3.2 Research Questions The current investigation in the border community o f Rivera will address the following research questions: 1. What is the sociolinguistic stratification of palatalization of /ti/, /di/ in UP according to the social variables of age, socioeconomic status, and gender? 2. What are the linguistics constraints of pal atalization of /ti/, /di/ in UP? 3. What are the driving forces of the variant selection? Are these factors linguistic, social, or both? Can this linguistic process be considered a change in progress? At what stage? 4. How do my results compare with previous studies on palatalization? Vowel nasalization research questions: 1. Is variation of vowel nasalization in UP random or conditioned? If so, what factors condition its variability? 2. If vowel nasalization is conditioned, what is its sociolinguistic stratificatio n according to the social variables of age, socioeconomic status, and gender? 3. What are the linguistic constraints of vowel nasalization in UP? 4. What are the driving forces of the variant selection? Are these facto rs linguistic, social or both? 5. Can this ling uistic process be considered a change in progress? At what stage?

PAGE 40

40 1.4 Organization of the Study Th e current study has been organized as follows Chapter 1 provide s backg round information on the history, geogra phy, and a social setting of the Uruguayan B razilian border and the statement of the problem under investigation Chapter 2 presents the theoretical framework that supports this study. I explore relevant issues in a literature review of variation theory and of analytical units it posits such as spe ech community and socia l networks A literature review of the F ronterizo variety covers studies dating from the Rona, to the most updated sociolinguistic research work conducte d in Rivera, Uruguay. The l iterature review on palatalization and nasalization is presented along with the working hypotheses guiding this investigation. Chapter 3 consist s of a complete description of the methodology used, fieldwork, data collection and analysis of linguistic and extralinguistic factors. In Chapter 4 I compare data from two cross sectional studies conducted at two different points in time (1995 and 2007) to verify whether the change in progress hypothesized by Carvalho (1998) has continu ed to advance or whether it has stabilized at the speech community level. Chapter 5 contains a discussion and interpretation of vowel n asa lization results, and Chapter 6 presents the conclusions and main findings of the investigation.

PAGE 41

41 CHAPTER 2 THEORETIC AL FRAMEWORK 2.1 Variation Theory The variationist approach to sociolinguistics began in the 1960s, with the ground Vineyard (1963) and New York City (1966) demonstrated that the trajectories of linguistic changes could be inferred from the observation of patterns of variation in present traditional approaches to linguistic description, namely stru ctural linguistics, which held the diachronic stu & Gordon 2003:2). Further, Labov reacted against the methods and assumptions of contempora ry dialectological research, although more recently, dialectologists have turned to quantitative methods to deal with variable forms more effectively and to represent variability among sample populations. With the founding of sociolinguistics, social fact ors began to play a significant role in the field of linguistics, especially in the analysis and modeling of speech. Subsequent research carried out by Labov using his methods and analysis of language variation and change developed into what is nowadays kn own as variation theory (Tagliamonte 2006). This new approach to analyzing speech attempts to account for the paradoxes of language change. Tagliamonte (2006:5) argues that while formal theories of language attempt to describe language as a set of fixed rules or principles, language changes constantly. As Weinreich et theories of language points out, the ultimate goal of the

PAGE 42

42 2.1.1 Theoretical Principles of the Quantitative Fra mework The central ideas of this approach are that variable and categorical processes are both inherent components of natural languages and that the variation we witness at all levels of language is not random. Indeed, the basic principle motivating varia ble rule analysis is the hypothesis that linguistic heterogeneity, like homogeneity, is rule governed (Sankoff 1978). Hence, as Sankoff asserts, for any area of variation, it should be possible to identify categories (or factors) that statistically determ ine the relative proportions of each variant (1978:159). Several assumptions underlie the quantitative approach to the study of language we can examine closely the forms that a linguistic variable takes, and note what features of the cont ext co occur with these forms. By context is meant the surrounding linguistic environment and the social phenomena that co occur with a given variable form. With a large enough set of data, we are able to make statements about the likelihood of co occurrence of a variable form and any one of the contextual features in which we are interested (2002:118). Bailey argues that these statements allow the researcher to express quantitatively the strength of correlation between a contextual feature and a linguistic variable. On the other hand, th (2002:118). It is worth noting that the majority of studies of linguistic variation has

PAGE 43

43 shown tha t the variables examined are sensitive to several linguistic and/or extralinguistic factors. In addition to the principles of quantitative modeling and multiple causes, two other principles are fundamental to the quantitative paradigm. These are summari zed by Guy (1991): Individual speakers may differ in their basic rate of use of a variable rule, that is, in their input probability for the rule. Individuals should be similar or identical in the factor values assigned to linguistic constraints on the rul e. (This assumption is usually qualified to apply just to people who belong to the same speech community). Many sociolinguists have adopted the quantitative approach to the study of language in communities around the world, among them: Panama (Cedergren 1973), Norwich, England (Trudgill 1974), Anniston, Alabama (Feagin 1979), Guyana (Rickford 1987), Rio de Janeiro, Brazil (Guy 1981) Montreal (Sankoff et al. 1976), etc. The wealth of research in the quantitative paradigm has demonstrated the systematic n ature of the linguistic variation that was previously thought to be random. In addition, quantitative research has shown that linguistic variables are constrained by both linguistic and social factors. These factors can potentially influence a language us another variable form (Bayley 2002:123). Portuguese nasalization is a perfect example of this argument. Although the literature on this topic is extensive, there is no agreement on the appropriate analytical framework to account for this type of phenomenon. As Guy (1997:125) observes, there is opposition between what is considered the essential system of the language and the operations and products of that system, that is, the actual usage of language by speakers. He rightly points observable construct, the grammar, langue competence, and on the other hand, there

PAGE 44

44 is the concrete, observable sum of language production, parole, performance, rtheless, this opposition is undermined by two fundamental principles emerging from the study of language variation: inherent variability and orderly heterogeneity. Guy argues that the latter implies that there is variation within competence, while the f ormer implies systematicity within performance (Guy 1997:140). The concept of ordered heterogeneity was coined by Weinreich et al. (1968:100) and La bov (1982:17) This concept challenges the invariant grammar assumption, and it was proposed in order to account for linguistic elements that alternate freely. The interpretation of this position was that such items were randomly distributed. Nevertheless, several decades of quantitative research have proved that alternating linguistic forms, although non ca tegorical, are not randomly distributed, and that variants are rule governed and show strong quantitative regularities. 2.1.2 The Apparent Time C onstruct The use of apparent time differences to study language change in progress has been a useful analytic al tool in quantitative sociolinguistics for more than forty years. Labov (1963) used it in his research conducted on [ay] and [aw] diphthongs. Labov found out that that the nuclei of both diphthongs were progressivel y higher with each younger age cohort. In order to discern between two possible explanations, namely age grading and apparent time, Labov compared the Martha Vineyard data with earlier records. By comparing the older speakers interviewed in 1933 for the L inguistic Atlas of New England (LANE) to the older speakers in the 1963 interviews, Labov was able to confirmed the apparent time interpretation (Bayley et al.1991:242)

PAGE 45

45 The apparent time construct is based on the assumption that individual vernaculars re main stable throughout the course of an adult lifetime (Bayley 2002:320). In providing a mirror of real time change, apparent time forms the basis of a conceptual study of sound change has been an excellent strategy to examine the issues involved in the mechanism of change, that is, the transition, embedding, and evaluation problems (Labov 1972:161). There are two types of real time studies that can help disambiguate betw een age grading or change in progress, a panel or a trend study. A longitudinal or panel study involv es resampling the same speakers, (i.e.), individuals must be followed for an extended period of time. A cross sectional or trend study involves resampling the same age range of speakers in the same speech community at different points in time (Bayley 2002). While a panel study can answer questions about the stability or instability of individual vernaculars, a trend study is apt to answer questions about st ability or instability at the speech community level (Labov 1994 ) In the absence of adequate real time data, the apparent time construct prov ides As seen above, by adding a trend method ology to the study, it is possible to compare apparent time data from two studies at different points in time, 1995 and 2007. I explore this approach with the objective of substantiating the change in progress hypothesized by Carvalho (1998). As Labov no tes detect unstable behavior of individuals and distinguish stable from unstable

PAGE 46

46 gath ering data on language change (Labov 1994 :85). 2.2 Speech Community The speech community concept, regarded as a key component in empirical linguistics, has been seen at the intersection of several basic problems in sociolinguistic theory and method (Patrick 2002). In the early 1960s, sociolinguists elaborated definitions of the speech community, moving from structural to functional approaches along the way. It is thus necessary to examine these definitions and some of the nit ion of the speech community was t he speech community is not defined by any marked agreement in the use of language elements, as much as by participation in a set of shared norms; these norms may be observed in overt types of evaluative behavior, and by the uniformity of abstract patterns of variation which are invariant in respect, are in principle replicable (1972:120). criteria for the characterization of a speech commu nity: the nativeness of speech community members, the presence of uniform patterns of linguistic variation, the shared social evaluation of linguistic parameters, and the systematic identity of the linguistic varieties on all linguistic levels (Kerswill 1 993:36). Nonetheless, this definition has been found wanting by some linguists. Their main objection is that, according to this definition, New York City is considered a single speech community (Labov 1966:125). Thus, speakers from lower to higher statu s agree in viewing the presence of postvocalic large scale concept since all social classes are united by their common evaluation of linguistic norms. For other ling uists, members of a speech community share a sense of

PAGE 47

47 belonging to a local territory and of participating in an interacting network in side this territory (Hymes 1974, Milroy 1980). h exploring how interaction, including language, constitutes social reality. From this perspective, Gumperz challenges the notion of agreement on the social meaning of linguistic variables members of such a community typically vary with respect to certain beliefs and other aspects of behavior. Such variation, which seems irregular when observed at the level of the individual, nonetheless shows systematic regularities at the statistical level of social facts (1982:24). A later reformulation of this definit to face nations, ethnic groups or the like (Gumperz & community has served to introduce several questions still considered problematic in the speech community become restricted to certain (possibly incompatible) paradigm of sociolinguistics, or is a broad conception still originally applied to particular contexts efforts tended toward generalization (2002). One starts with a social group and consi ders the entire organization of linguistic means speech community and shares with Gumperz the shift in focus from varieties to the

PAGE 48

48 relationships among speakers. Further, for Hymes, the speech community is not a methodological prime, since one cannot know what practices are critical, or who shares them, before a study has been carried out. Moreover, for Hymes, the starting point of description is a repertoire of ways of s communicative competences and verbal repertoires (Patrick 2002:582). ue that this sociolinguistic model reflects a consensual view of society, where the community is envisaged as cohesive and self persistence of nonstandard vernacular communities uncovered by many res earchers (including Labov) are more readily interpretable as evidence of conflict and sharp society as a conflictual model of society in order to account for the phenomen on of linguistic change, with which some kind of social conflict is generally associated. importance to the diverse patterns of evaluation in the speech community under i local unity, patterns of divergence, and shared acceptance of external norms by members of different speech communities (Patrick 2002:588). The sociolinguistic model of the speech community was adopted to analyze phonological variation in Rivera. The speech community is a primary and necessary concept that starts with the social group shifting its focus from varieties to the relation among speakers (Hymes 1974). The city of Rivera can be classified as a speech

PAGE 49

49 community since its members share a sense of belonging to the local territory. Community members in Rivera also share participating in an interacting network. Please refer to the methodological section of this inve stigation (Chapter 3 ) for a detailed account of the speech community of Rivera. 2.3 Social Networks The speech community of Rivera is a close knit community, and a social n etwork s approach can account for the patterns of relationships between its member s Social network analysis was developed by social anthropologists during the 1960s and 1970s. Ricardo 1985:69). Sociolinguistic research carried out by L. Milroy (1987 ) and Milroy and Milroy (1992) espoused the impact of studying the networks of a speech community. The Milroys based their studies on previous sociological research (Mitchell 1969, Boissevai n & Mitchell 1973 ), paying special attention not only to intergroup interactions but also to the personal relationships of the speakers with other individuals. The Milroys established a number of indicators as a way of facilitating network analysis, base plexity is a measure of the range of different types of transactions people are involved in with other individuals. On the other hand, a uniplex relationship is one in which the link with another person occurs in only one area, whereas multiplex relations involve interactions with others along several dimensions ( e.g. a workmate may also be a ne ighbor and member of the sam e church, etc .).

PAGE 50

50 These personal communities are constituted by individuals who engage in a variety of endeavors on a daily basis and create interpersonal ties of different types and strengths. Thus, a fundamental premise of net personal communities to provide a meaningful framework for solving the problems of content differences between networks imp act individuals critically (Milroy & Gordon 2003:117). Consequently, if a personal network consists of strong ties that are multiplex, and if the network is relatively dense, then such a network has the capacity to sustain its members in practical and sym bolic ways. Conversely, such a network can also impose unwanted constr aints on its members. As Milroy and Gordon argue, a strong network, that is, one containing dense and multiplex ties, supports localized linguistic norms, resisting pressures from compe ting external norms. On the other hand, a weakening of these ties produces conditions favoring contexts of language change. Thus, network analysis is a useful analytical tool in variationist analysis since it can help explain why speakers support a partic ular linguistic system that stands in opposition to forces of standardization, and why another system might be more sensitive to extralinguistic influences (Milroy & Gordon 2003:118). Network analysis has been implemented widely by scholars in several di sciplines. Although no standard procedure for analyzing social networks can be identified, variationists have adopted it for several reasons. First, it provides a set of procedures for studying small groups where speakers belong to a rather homogenized so cial group, as for example, the southeastern United States island communities investigated by

PAGE 51

51 Wolfram, Hazen, and Schilling Estes (1999). It has been used in studying minority ethnic groups, migrants, rural populations, or populations in non industrialize d societies. Another advantage is that, since social network is a concept that relates to local practices and norms, it can reveal the social dynamics within the speech community that are driving language variation and change. Lastly, network analysis is an analytical tool capable of dealing with variation among individual speakers, rather than among groups constructed with reference to predetermined social categories (Milroy & Gordon 2003:120). The idea of using social networks as a speaker variable was developed in the Belfast study (Milroy 1987) as a way of investigating the capacity of close knit networks to function as a mechanism to enforce norms (Milroy 1987:106). A network approach was employed in many studies in variationist sociolinguistics duri ng the decade of the olescents; Bortoni (1990) study of language differences among Melanesians in New Caled onia; Lippi among others. Labov and his colleagues in their studies of Philadelphia neighborhoods used the network concept at the fieldwork stage (Labov & Harris 1986). A network approach to f ieldwork of Rivera, providing the set of procedures for analyzing relationships where speakers

PAGE 52

52 belong to a relatively small an d homogeneous group such as in Rivera Another advantag e of a network based approach to fieldwork is that, since social network is a concept that relates to local practices and norms, it can reveal the social dynamics driving language variation and change (Milroy & Gordon 2003). Please refer to the methodolog y section (Chapte r 3 ) for specifics about entering the speech community of Rivera via social networks 2.4 Previous Studies of Linguistic Varieties i n the Uruguayan Brazilian Border. Earlier studies of F ronterizo (Rona 1965, Elizaincin 1979, Elizaincin et al. 1987) have focused on this variety, which has been classified as a tercera lengua (1965) F ronterizo classific ations, stated that this dialect was more variabl defined regional varieties would suggest. Elizaincin et al. (1987) classified F ronterizo as a dialect of Portuguese and proposed the name Dialectos portugueses del Uruguay (DPU) F ronterizo being one of them. One of the first scholars to document this dialect was Jos Pedro Rona Rona (1958, 1965) distinguishes two different F ronterizo dialects: the first a Portuguese based F ronterizo with Spanish influence, in other words one that has a phonological and lexi cal system that is essentially Portuguese; the second a Spanish based F ronterizo with Portuguese influence in resto del Uruguay y las influencias lxicas, morfolgicas y sintcticas, portuguesas, aunque nume rosas of Uruguay and the lexical, morphological and syntactic Portuguese influences, 8 my translation ). In addition, Rona argues t hat the delimitation of isog losses are evidence of four sub

PAGE 53

53 dialectal F ronterizo varieties denominated as: artiguense, tacuaremboense 1 melense y yaguaronense. The first li nguistic stu dy on the Uruguayan Brazilian border was carried out by F. Hensey (1972). Linguisti c Interview Materials), Hensey (1972:34) gathered samples of the segmental phonology and of lexical item s in order to test the frequency with which supposed interference phenomena occurred. Participants were asked to describe the sketches in the PLIM and to comment on the contact situation, particularly the linguistic aspect. Hensey examine s the influence of Spanish phonology on the Portuguese of bili nguals living in the border F ronterizos developed from Standard Portuguese by under differentiations which led to reinterpretations of distinctive features and eventually to restructurin g. Phone substitutions, whether traceable to Spanish or to areal traits of pronunciation, may have assisted in bringing Hensey describes the community of Rivera as bilingual, where Portuguese is acquired in childhood without the benefi t of later formal study. He also argues that such bilinguals learn both languages simultaneously, Portuguese being the language preferred by children to communicate with parents, siblings, an d peers. H e notes that for Uruguayans, bilingualism i s less the acquisition of Portuguese than an indication that the i ndividual has acquired Spanish. Hensey also notes that Spanish, whi ch is spread by an effective public education system, is forced to compete at a disadvantage with sub standa rd Portuguese ( Hensey 1972:137 42 ) 1 esta variedad cubre los departamentos de Rivera y Tacuaremb Rivera and Tacuaremb 1965:14 my translation )

PAGE 54

54 Although most of the population in Rivera speaks a variety of Portuguese, Uruguayans themselves judge their speech as unsatisfactory, and to most Brazilians this is an unacceptable form (Hensey 1972:142). Hensey (1972:55 61) provid es several examples of Spanish phonological interference in the speech of bilingua ls speakers: 1. f inal unstressed /e/ and /o/ are frequently realized as mid vowels [e] and [o] instead of high vowels [i] and [u], (30.8%) 2 ; 2. n asalized /a/ ( N), realized as a low central vowel [] rather than a mid central vowel in tonic position [ ], (70%). 3. s yllable final /l/ realized as an alveolar lateral with firm apical occlusion rather than a posterior semivowel, (60.5%); 4. i nitial and intervocalic /rr/ realized as an alv eolar trill rather than a voiceless velar fricative, (83.3%); 5. p alatal nasal ) realized as an alveopalatal nasal with firm occlusion rather than a nasalized yod, (58.0%); 6. i ntervocalic /b/, /d/, and /g/ realized as simple fricatives rath er than occlusives, (58.1%); 7. /ti/ and /di/ realized as simple stops rather than palatalized or affricate consonants, (80%). Hensey (1972) adds to these seven types of phone substitution the following potential (my emphasis) structural interferences in th e form of underdifferentiation: 1. u nderdifferentiation of Portuguese open and closed /e and /o (9.1%) ; 2. u in favor of the voiceless member, (15%); 3. u nderdifferentiation of Portuguese b/v (16.6%). 2 Hensey (1972) only provid es overall rates of interference.

PAGE 55

55 per group 3 ; therefore, as he suggests, results are presented without assumption of significance. Nevertheless, the phonic interference in the form of phone substitutions points to potential structural conseq uences which are not clear: that is, the preference for a simple dental stop rather than a palatalized allophone for /t/ /d/ bef ore /i/ (Hensey 1972:77). He determines that Spanish is the first language of these participants based on their nationality, that is, it is assumed that if a participant is born in Rivera or Rio Branco, the two Uruguayan cities surveyed, their dominant l anguage is Spanish. Hensey does not take into consideration social factors, deemed to be crucial in contact induced lang uage change situations. I n a long and sustained language contact situation such as the one found along the Uruguayan Brazilian border language interference and other languag e contact phenomena can be expected I t is a challenging task to establish potential structural interference since Spanish and Portuguese are languages typol ogically close However, the phonological system represents the most salient difference between these two grammars In this light, Cabrelli and Rothman (2010 ) examine the linguistic implications of non native phonological acquisition of language pairings such as Spanish and Portuguese 4 I n the present study the speech of bilinguals in Rivera is conceptualized as varying ac 3 Hensey (1972) surveyed two sets of twin cities along the border area. The Uruguayan cities of Rivera and Rio Branco, and the Brazilian cities of Santana do Livramento and Yaguaro (17). 4 Cabrelli and Rothman (2010:278) propose the Phonological Permeability Hypothesis predicts that if native and seemingly successful non native phonological systems are learned/constructed in the same manner, then an addition of a third language will affect these systems (s omewhat) equally within the same timeframe. Alternatively, if they are constructed in a different manner (and are thus mentally configured differently), then the successive system will undergo much more rapid and pervasive cross linguistic interference fr

PAGE 56

56 characteristics In this sense, p articipants are perceived as memb ers of a speech community who share a sense of belong ing to a local territory and participate in an interacting network inside this territory ( § 2.2 ). T he approach to language variation following modern sociolinguistics, is the speech community not the isolated individual. the linguistic outcomes of language contact are determined in large part by the history and social relations among populations, including Elizaincin et al. (1987) F ronterizo classific ations stated su ggest. They further point out that although DPU presents an invariant linguistic system, it also shows a tendency toward linguistic simplifications. However this study does not discuss simplification processes or attribute them to language contact between Spanish and Portuguese. Unfortunately, the study relies on impressionistic and qualitatively data to describe the speech of monolingual speakers of Portug uese in different towns along the Uruguayan Brazilian border from Rivera, and lacks treatment of quantitative data. In 1994 analysis, F ronterizo is classified as a tercera lengua Lipski argues that the reasons for the formatio n of a F ronterizo variety which the rural residents of an isolated and marginalized zone were pulled linguistically in two di rections, but where neither was strong enough to completely coalesce into a 1994: 3 42). In addition, Lipski posits that the F ronterizo variety is not just confined to the immediate Uruguayan Brazilian border, but penetrates deep into

PAGE 57

57 Uruguay. He suggests that the reasons for the heavy influence of Portuguese lexical, phonological and syntactic forms into Urugua yan speech are many, but the most important is the fact than in the Northern regions, better economic opportunities are to be found in Brazilian territory. Lips ki (1994:343) identifies the following p honological characteristics of the F ronterizo variety : 1. Portuguese has, in addition to the five vowels of Spanish, two mid lax vowels, a centralized unrounded schwa like vowel (found only in unstressed contexts and derivable phonologically from /a/), and five distinctive nasal vowels. In unstressed contexts, particularly word finally, /e/ usually raises to [i] and /o/ raises to [u]. One F ronterizo sp ecimen is the degree of approximation to the Portuguese vowel system. It is more frequent for a five vowel system to emerge among oral vowels (even many monolingual Brazilians are moving in this direction), while distinctive nasal vowels are retained. Un stressed vowel raising is variable in F ronterizo. 2. ]/[ ] before [i] is rare in F ronterizo speech, nor is it categorical in regional dialects of southern Brazil. 3. Portuguese distinguishes the pairs /s/ /z/ and F ronterizo this distinction is also variable, especially since Uruguayan Spanish /y/ has the same groove fricativ Carvalho (1998) represents the first large scale sociolinguistic analysis of the speech of Riverans. Drawing from a corpus of 54 bilingual participants from Rivera, Carvalho identifies the social and linguistic factors that condition variability of two phonological variables : realization of /ti/ and /di/ and realization of the palatal liquid / /. Carvalho points out that recent urbanization of border communities has allowed greater acceptan ce of urban Brazilian Portuguese, which has caused local Uruguayan Portuguese to be pulled in the direction of the more prestigious variety. Carvalho opposes the use of the term F ronterizo to characterize the Portuguese of Uruguay.

PAGE 58

58 Instead a dialectal c ontinuum is proposed where local Uruguayan Portuguese can be situated at one end and urban Brazilian Portuguese at the other. As Carvalho posits rural and hybrid origin, toward an assimilation of linguistic features that are stereotypically Brazilian, as the result of a desire to emulate speakers of larger urban (2004:128). 2.4.1 UP in Artigas City Douglas ( 2004) examines language shift, displacement and loss in the city of Artigas, Uruguay. She maintains that UP in Artigas does not fit into any current theory or standard model of language change, such as language shift and death, dialect leveling, or decreo lization. Comparing data from speakers of the town of Artigas and the villages of Bernab Rivera and Siqueira, Douglas finds that UP is undergoing both functional and structural transformations as it is displaced, leading to its eventual loss as a native t ongue. Further, she argues that language change in UP is socially stratified as UP speakers abandon the variety and adopt new features fro m the competing standards ( 322). To account for the complex language environment in Artigas, Douglas proposes two alt ernative models to examine the transformation of UP: the proficiency continuum, to demonstrate functional displacement, and the post creole continuum, as an example of structural transformation during the process of displacement (2004:111). She furthermo re argues that the proficiency continuum is the model of language change within the area of language shift and death that accounts for the progressive decline of

PAGE 59

59 displace d variety. In other words, this model assumes that bilingual speakers, having attained a certain degree of proficiency in both varieties, gradually lose proficiency i n the subordinate variety ( 112). Moreover, for Douglas the mixed characteristic of UP b ears significant resemblance to the intermediate varieties of a decreolizing creole that emerges as creole speakers incorporate features in their appr oximation of the standard ( 127). Douglas maintains that the decreolization approach is a model that can be applied to varieties confronting language displacement in the face of a dominant standard (2004:126). She argues that the repertoire of a UP speaker is composed of features drawing from the contact of all three varieties, standard Uruguayan Spanish, stan dard their language variety are affected by the sociopolitical and psychosocial constraints linguistic realizations of their variety are determined by their social and ideological characteristics. That is, there is no cohesion at the speech community level. Data results show that speakers have a variety of phonological forms from whic h to choose. 2.4.2 Brazilian Portuguese The Portuguese variety spoken in Brazil, like many other varieties spoken over a large territory, presents overlapping regional and social linguistic variation (Azevedo 2005:211). Azevedo asserts that such variation educational level, which in turn is linked to their socioeconomic status. In fact, as he further points out, the most salient contrasts within Brazilian Portuguese (BP) are not regional but social.

PAGE 60

60 There is divergence b etween the vernacular, the speech of the educated minority, and the normative language dictated by prescriptive grammars which is based on the formal written usage found in Portuguese literary works. However, there is a third variety spoken by educated sp eakers in casual speech and informal types of writing which do not require Prescriptive Portuguese known as Standard Brazilian Portuguese (SBP). In order to help understand this intricate situation, Azevedo proposes to visualize it as a triangle, where on e of the vertices represents Vernacular Brazilian Portuguese (VBP), another vertex corresponds to normative or Prescriptive Portuguese (PP), and the last one to SBP. Although, it is common for languages to have a vernacular and a distinctive standard, in the case of Brazil the presence of vernacular features in the unmonitored speech of educated speakers is common, suggesting that VBP and SBP are a fundamental part of Brazilian Portuguese as a whole and not mere discrete varieties (2005:212 13). Traditiona lly, the literary variety has been considered to be the only valid variety. 5 unsystematic selection of examples have led normative grammars to contradict each 11). Nevertheless, current linguistic research on Brazilian linguistics has invalidated this view, and it has been shown that vernacular features are an integral part of the language. Another possible source for the unique features of Brazilian Portuguese is contact with other languages. Portuguese has long had contact with speakers of African 5 This attitude toward the actual language use has been a major challenge for low social strata Brazilian students who are not exposed to the educated variety varieties of speech. This situation ha s been

PAGE 61

61 languages spoken by the 3.5 million slaves taken to Brazil between 1500 to1852. The Atlantic slave trade has had huge linguistic consequences not only for Brazilia n Portuguese but also for other major slave holding centers, such as Jamaica, Haiti, the southern states of the USA, the Lesser Antilles, and Guyana (Guy 1981:4). 2.4.3 Uruguayan Spanish The Spanish variety spoken in Montevideo, Uruguay is an extension o f the porteo speech of Buenos Aires, Argentina. No striking differences can be found between these two varieties, although dwellers of each city maintain the contrary. Yet, linguistically, Uruguay can be divided, once social factors are considered, along an urban rural axis with bilingualism along the Brazilian border. While the Montevideo dialect is the main model, rural speakers from the interior exhibit speech patterns not found elsewhere (Lipski 1994:338). Along the Uruguayan Brazilian border, UP dialect and different degrees of bilingualism are found. Phonologically, Uruguayan Spanish exhibits the following characteristics (Lipski 1994:340 41): 1. known as or sheismo. 2. Preconsonantal /s/ is aspirated except in careful speech. Deletion is strongly favored at the lowest sociolinguistic levels. Phrase final /s/ is ma intained in educated speech, but it is deleted among the lower social classes. Word final prevocalic /s/ is realized as [s] in prestigious speech, whereas aspiration [h] is stigmatized. 3. Word final /n/ is realized as alveolar. 4. Intervocalic /d/ is elided i n all registers of speech. 5. Posterior fricative /x/ becomes palatal before /i/.

PAGE 62

62 Although the Uruguayan lexicon has been extensively documented, there is as yet no comprehensive account. In general, Uruguayan Spanish shares most of its lexicon with Buenos Ai res, including lunfardo 6 slang. While Uruguayan Spanish does not exhibit notable differences from patterns found in other Spanish speaking varieties, in its syntax, the use of voseo as the preferred form of address in Montevideo is striking. 7 T is favored in several northern and southeastern areas of the country, while vos and t compete in the intervening areas (Lipski 1994:341). In the border area the forms of address of choice are t, vos, and o/a senhor/a, el/la seor/a depending on the sociolinguist ic characteristics of the speaker. 2.4.4 Studies on the Spanish in Rivera Waltermire (2006) examines the sociolinguistic conditioning of two phonological variables in the Spanish of Rivera, Uruguay: the realization of intervocalic /d/ as either occlusive [ d] or as phonetic zero [] and the realization of syllable and word final /s/ as aspirated [h] or a phonetic zero []. Waltermire finds that occlusive realizations of intervocalic /d/ are favored in word initial, stressed syllables whereas /d/ deletion is favored in word medial, unstressed syllables. Members of the community with non professional occupations tend to prefer occlusive articulations, while /d/ deletion is favored by members of the younger generation. 6 Lunfardo (sometimes shortened to lunfa ) developed among the socially marginalized classes of Buenos Aires. Although considered to originate as a criminal argot, Lipski argues t hat in its broader sense, lunfardo is the vernacular speech of the Buenos Aires working classes, similar to the Cockney of London (1994:175). 7 Voseo paradigm in Uruguay includes combinations such as t with verb forms corresponding to vos (e.g., t volv s (you return); an analogical final / s/ is often added to preterit forms, e.g., dijistes etc.). Verbal forms corresponding to vos alternate with those belonging to t in the present subjunctive (Lipski 1994:341).

PAGE 63

63 As for /s/ aspiration, Waltermire argues that women and students favor it in pre consonantal environments, while deletion is the choice of preference by men and non professionals. Furthermore, Waltermire suggests that participants who have positive attitudes towards Brazil and language mixing t end to incorporate loan articulations from Portuguese into their Spanish more often than speakers who prefer Spanish and have negative attitudes toward Brazil. This section concludes the o utline of linguistic varieties o n the Uruguayan Brazilian border. As shown above, there exist several linguistic varieties in contact in the standard and dialectal form. While the Spanish spoken in Rivera can be classified as a variety of Uruguayan Spanish, Uruguayan Portuguese can be characterized as a language contac t variety combining southern Brazilian Portuguese and Uruguayan Spanish. Due to complex historical, socio economic, and political factors, these language varieties have coexisted along the Uruguayan Brazilian border for centuries. These competing varieti es are part of the language admixture available to speakers according to their social characteristics. Social factors compel a speaker to adopt or resist a linguistic variant. Bilingualism occurs in the border populations varying from speaker to speaker. This bilingualism can be characterized as scalar rather than discrete. In order to account for phonological variation in UP in Rivera, I will examine two linguistic variables. 2.5 Linguistic Variables under Investigation The study investigates the soc iolinguistic distribution of two phonological variables in the speech of bilingual speakers of UP in Rivera, Uruguay: palatalization or dental realization of /ti/, /di/, and the presence or absence of vowel nasalization.

PAGE 64

64 2.5.1 Palatalization One of the dependent variables of the study is palatalization of /t/ /d/ before /i/. The actual phonemic articulation in some varieties of Brazilian Portuguese (BP) of /t/ as [ ] and /d/ as [ ] before [i] results from t he phonological process called palatalization. The conservative articulation, where the process of palatalization is not operative, is characteristic of other Portuguese varieties, namely European Portuguese, and varieties from northern Brazil; in which /t/ and /d/ before /i/ are realized as the dent o or dento alveolar (Azevedo 2005). 2.5 .1.1 The p rocess of p alatalization Palatalization can be described as a process in which the primary articulation is palatalized i f the point of articulation moves toward the palatal region in some particular palatalization two conditions g it must be a front vowel, a palatal semivowel, or a palatal or palatalized consonant), and [that] the sound that results must be palatal or have a o a single phenomenon, but it is used as a cover term for three different diachronic processes that can occur alone or in different combinations. These processes include tongue fronting, tongue rising, and spirantization 8 (Bhat 1978:50). 8 Spirantization is defined as

PAGE 65

65 Mattoso Cmara argues that phonological changes which occurred in the informal style as well as intrusions from sub dialects 9 have had repercussions in the modern consonantal system of Portuguese. These changes can be traced historically through the evolution of the Por tuguese consonantal system from the Latin language. Further, an important development in the received language, 10 at least in certain areas of Brazil, was the creation of an affricative positional variant of /t/ /d/ before i. In this environment the stop becomes a mid palate fricative with a slight bushing sound as in tia [ a] d ia [ a] 2.5.1.2 Palatalization in Portuguese One of the most salient characteristics of the Carioca dialect and several other Brazilian dialects 11 is the alternation of the dento alveolar plosives /t/ and /d/, with the affricates [ ] [ ]. Palatalization is also one of the main characteristics that differentiate European from Brazilian Portuguese. In Brazilian Portuguese pairs like tipo and digo wo rd. This phonetic articulation in Brazilian Portuguese of /t/ as [ ] or /d/ as [ before [i] results from the palatalization process. Azevedo (198 1:31) ascertains that the apico dental realization of /ti/,/di/ is the conservative realization of the phonem e, as in tio, dia. However, these phonemes 9 Mattoso Cmara states that sub dialects of this type have different phonological systems than those of the received dialect. Further, he offers certain Brazilian dialects (1972:43). 10 For Mattoso C mara in both Brazil and Portugal, the re is a tendency toward the cre ation of a slight diff vulgar ) speech and the normal, accepted pronu nciation of educated speakers. This fact has inevitably impacted the consonantal system, developing several new positional variants that modify the interrelations of the individual consonants (1972:44). 11 This phenom ena varies according to sociolinguistics factors in the dialect of Sao Paulo. Giangiola (2001:138) reports that in the Northeastern states of Sergipe and Pernambuco /d/ and /t/ are pronounced as stops [d] and [t] in all cases, which sounds provincial to those Brazilians who do palatalize.

PAGE 66

66 undergo several degrees of palatalization. He posits that in Paulista (So Paulo dialect) and in Mineiro (dialect from Minas Gerais, Brazil), palatalization varies, when it occurs, from a slightly affricated art iculated in the pre palatal region to a full affricate, which is its normal rea lization in Carioca dialect (198 1:31). In other Portuguese varieties, such as European Portuguese (EP), the phonological process of palatalization is not operative, hence, /ti/ and /di/ are always pronounced as dentals [ti], [di]. Table 2 1 show examples rendering of /ti/ /di/ in European and Brazilian Portuguese (Mateus & dAndrade 2000:17): Table 2 1. Palatalization of /ti/ and /di/ in EP and BP E P BP G los sary t ia [ta] [ a] d ia [ da] [ a] p ote o ti] o ] p ode i] The occurrence of the affricates is context dependent. It is determined by the presence of a following /i/. This phenomenon is described as a palatalization unde rgone by the plosives. As shown in Table 2 1, this is a variable feature in BP; it is categorical in EP. In fact, for many scholars the standard for BP is partially different from the standard for EP. Although much variation exists between these two varieties, differences in speech are more intensely felt, and familiarity with one dialect does not ensu re immediate comprehension of the other (Azevedo 2005:20). Many scholars have argued about the prestige of the palatal realization of /t/ /d/ be fore /i/ ( Giangiola 2001 ). The author

PAGE 67

67 since the mid 1990s is said to coincide with the advent of radio and television broadcasting in Brazil 12 which popularized certain aspects of the speech of Rio de Janeiro 13 In the prestige dialect /d/ is pronounced as the palatal affricate [ ] before [i] ]. Furthermore, / d/ and / t/ a d vogado a d mitir ri t mo t nico 1:138). While in the popular Portuguese spoken in southern Brazil the rendering of /ti, /di/ is historically the conservative pronunciation, the application of the palatalization rule continues to advance. In fact, data show that palatalization is becom ing mainstream in Rio Grande do Sul, the state adjacent to Rivera, mostly in large metropolitan cities Santana do Livramento and Santa Maria, Rio Grande do Sul, analyzed p alatalization of dental stops. His results show that speakers from the interior city of Santa Maria palatalized 79% of the time, whereas speakers in Santana applied the palatalization rule at a lower rate, 47% of the time. More recently, in subsequent r esearch conducted in Rio Grande do Sul Brazil, Bisol (1991 b ) analyzes palatalization of dental stops followed by /i/ The study covers speech samples from 15 monolingual speakers from Porto Alegre, the state capital, 15 from the border city of Santana do Livramento, 15 bilinguals from a German settlement, and 15 bilinguals from an Italian settlement. All of these participants reached only 12 The influence of TV as a source of prestigious speech has been discussed also by Carvalho (1998, 2004). 13 It has been argued that palatalization of dental stops /ti/ and /di/ originated in Rio de Janeiro and has sprea d as a prestige feature across the country (Bortoni Ricardo 1985:14).

PAGE 68

68 primary school education. In her study she uses a control group of 15 graduate students living in the capital. The juncture, preceding context, and following context. As for social factors, ethnicity and age are analyzed. For the syllable factor group, which tests the influence of stress, s he found that the application of the palatalization rule in the capital favors the relative stronger positions, in the following order: stressed, pre tonic, and post tonic syllable. The reverse order was found in the bilingual groups (Bisol 1991:107). Wh ile preceding phonological context highlights the role of the sibilants /s/, /z/ as a strong inhibitor of the palatalization rule, laterals and nasals show to favor it. On the other hand, following phonological context shows that the lateral and the nasal palatal favor the application of the palatal rule. While trills have an irregular behavior, nasals seem to play an unfavorable role. Yet, vowel, other consonants and pause are not relevant (Bisol 1991:120). As for the social factors the results for ethni city showed that capital dwellers palatalize the most (.88), followed by speakers from Livramento (.72), and bilingual groups (Germans at .40, and Italians at .08) palatalize the least. As for age, the younger speakers recorded the highest figures of rule application in all the communities except time the palatalization rule is in process of expansion. She furthermore argues that although this is not confirmed by the sta ndard speech sample, it is supported by the general evidence (Bisol 1991:111). Furthermore, Bisol concludes that the contact of the

PAGE 69

69 Gaucho dialect with languages which do not palatalize dental stops (Spanish, Italian, and German) inhibits the application o f the palatalization rule (1991:120). As Hensey (1972) and Bisol (1991) studies show, the conservative pronunciation of dental stops in the Rio de Grande do Sul state is receding. Palatalization of /ti/, /di/ is becoming the new trend among young speaker s willing to align themselves with standard Brazilian Portuguese. In this way, the progression of the change cannot be separated from the influence of social and affective facto rs toward the prestige feature. 2.5.1.3 Palatalization in UP Rona (1965), one o f the first scholars describing F ronterizo (see § 2.4 for further discussion), documents the existence of a slight degree of palatalization in this variety. Rona (1965:40) points out that alveolares se palatalizan ante vo cales anteriores, como en portugu s, adoptando las realizaciones [ y [ ] En la variedad tacuaremboense 14 sin embargo, esta palatalizacin es mucho menos perceptible y es ms prxima a [t] y [d] alveolar phonemes are palatalized before front vowels, as in Portuguese, taking the realizations [ ] and [ ]. In Tacuaremboense v ariety, however, this palatalization is much less noticeable and is closer to [t] and [d]" (my translation). ) research in several communities along the Uruguayan and Brazilian border attempts to describe the languag (1972) approach to the study of interference applied research techniques for general 14 As seen previously, Rona (1965:14) distinguishes four subdialectal varieties of fronterizo.artiguense tacuaroembense, melense and Yaguaronense. The Tacuaremboense variety covers Rive ra and Tacuaremb and the oriental part of Artigas and Salto.

PAGE 70

70 questionnaire was implemented to gather speech s amples from three groups: (1) six native speakers of Portuguese from the interior of Brazil, (2) six speakers from Santana, and (3) six bilingual speakers from Rivera. His data on application of the palatalization rule in the speech of bilingual speakers in Rivera show 20% of affricate realizations, UP as portuol, relegating its status as a sign of linguistic interference of Spanish in Portuguese. Thus, the high frequency of dental realizations is attributable to Spanish classes favored the palatal realization of dental stops, lower socio economic speakers and men from the upper and middle group s disfavored it. Lipski (1994) in a s urvey of Latin American varieties argues that t to [ ]/[ ] before [i] is atypical in F ronterizo speech, nor is it categorical in regional varietie s of southern Braz il. has investigated both social and linguistic factors that condition variation in UP. Carvalho investigates the sociolinguistic distribution of two phonological varia bles: the palatal liquid / / and its variants the alveolar lateral followed by a high semivowel [lj], or the palatal glide [j], and the dental or palatal realization of / ti/, /di/ Data were collected during five months of fieldwork in Rivera. Participants were all residents of R ivera and were selected on the basis of availability. The speech sample consists of 54 participants and an output of 2,248 tokens for the palatal linguistic variable. Participants were grouped according to their socioeconomic status (working class, lower middle class, and mid middle class), age (15 29, 30 49, 50 70 years old), and gender. The

PAGE 71

71 linguistic factors considered were: tonicity of the environment, preceding and following segment, and juncture (Carvalho 1998:171). Statistical results show that p alatal realization occurred 32%, and dentals were stratification in rendering the palatal pronunciation of dental stops. For Carvalho palatalization in UP is a variable related mainly to age. In the youngest group its application is very frequent with a probability weight as high as .91, whereas in the oldest group palatalization is very rare, with a probability value of .05. The second most important factor is socioeco nomic status. The mid middle class group palatalized more than members of the other groups, showing a value of .81. Lower middle class and working class members show very low factor values at .34 and .30 respectively. As expected, women favor the palat al rendering of dental stops more than men, showing a probability of .60 in the application of the rule, whereas men show a probability factor weight of .39 (1998:150 59). Carvalho (1998) argues that the tendency to palatalize among the mid middle socioeco nomic group shows that speakers are using the symbolic value of the standard variable to signal urban orientation, different from the traditions of the rural border community. Speakers from the working socioeconomic group, men, and the elderly, who normal ly rely on local connections and resources, favor a more conservative form of UP. 2.5.2 Nasalization The second dependent variable under investigation is vowel nasalization in UP. As has been already mentioned, in most varietie s of Brazilian Portuguese v owel nasalization occurs generally whenever a vowel is followed by a nasal consonant,

PAGE 72

72 regardless of syllable structure. Further, Lipski notes that this vowel nasalization process is not addressed in current descriptions of the Portuguese language, althoug h it is an accepted phenomenon in Brazilian Portuguese (Lipski 1975:67). Thus, as Lipski (1975) observes, in order to account for what appears to be a basic and ever spreading fact of vowel nasalization in Portuguese, the criteria selection was extended to cover every instance of nasalization phonemic or not. 2.5.2 .1 Vowel nasalization in Romance l anguages Nasal vowels can be found across Romance languages, although two varieties in particular have been widely accounted for, namely French and Portuguese. Sampson and Portuguese have nasal vowel phonemes, while other Romance varietie s present high levels of allophonic nasality. Nasalization phenomena affect several other Romance varieties, among them Galician Portuguese, Andalusian Spanish, Gallo Romance, some varieties of Rheto Romance, North Italian varieties (including Gallo Ital ian varietie s transplanted to Sicily), varieties of Corsican (especially in the north west), varieties of Sardinian (especially in the center and south), and Romanian. Historically, it seems that vowel nasalization might have been found in other Romance va rieties as well, but due to the process of subsequent denasalization, its traces have been camouflaged (Sampson 1999: 32). 2.5 2 .2 Portuguese nasal vowel s ystem There are two ways of indicating nasalization in modern Portuguese orthography: a consonant l etter / m or n (Mattoso Cmara 1972:51). The phonetic inventory of Portuguese nasal vowels is a

PAGE 73

73 controversial issue among linguists in that they disagree on the number of nasal vowels. On the other hand, there are reports attesting the presence of open nasalized [ ] and [ ] in some varietie s, raising the possibility that there can be distinctive contrasts among them (Shaw 1986:21). For Wetzels (1997), Brazilian Portuguese has a system of sev en oral vowels /i,u,e,o, ,a/ that is fully exploited in stressed syllables only (204). In her study on the speech of Brazilian Portuguese, Shawn (1986) maintains that there are five nasal vowels, fourteen nasal diphthongs, and eight tripthongs in the pho netic inventory of Brazilian Portuguese. Moreover, the literature on Portuguese nasal vowel system distinguishes between nasal vowels, nasalized vowels, and nasal diphthongs. Traditionally phonologies of Portuguese distinguish two types of nasal vowels: one which is said to be (surface) contrastive; the other allophonic. Conventionally the (Wetzels 1997:205). In Brazilian Portuguese, nasal vowels occur in stressed as we ll as in unstressed syllables, both word internally and word finally, as shown in Table 2 2 and 2 3 (examples from Wetzels 1997:205). Table 2 2. Nasal vowels a. word internal stressed Glossary ka] nunca b. pretonic pente interesse [bg] c. word final stressed bong ] alem [a d. word final unstressed armazm [wb ] lbum [m ] mrmom

PAGE 74

74 Nasal vowels are usually represented in the orthography by the presence of nasal consonant in the syllable coda. 15 Along with nasal vowels, Brazilian Portuguese consonant from which the nasal feature spreads to th e vocalic element(s) of the representation of contrastive nasality, both types of nasalization are treated as the result of a single rule of leftward spread (Saciuk 1970, Bra sington 1971, Almeida 1976, Quicoli 1990). However, different analyzes of the nasalization processes argue against a unified spreading rule. Table 2 3. Nasalized vowels a. stressed before /n,m/ Glossary ma] cama mu] b. unstressed befo re /n,m/ fumo [tens] tenaz boneca cozinha cegonha [k canhoto [k cunhado in We tzels (1997) has proposed that contrastive nasalization is obligatory at any point in the sequence in which it occurs. He further argues that allophonic nasalization is the result of a variable rule, constrained by both linguistic and non linguistic varia tion 16 Wetzels maintains that in all varietie s of Brazilian Portuguese, allophonic 15 In the specific case of /a/, whether word final or part of a nasal diphthong, the orthographic diacritic (~) is used to mark contrastive nasality: l intern ally [j] orthographic practice allows both (i) and (aim/n), as in cibra and caimbra 16 In the Carioca dialect, allophonic nasalization is obligatory for stressed vowels (Wetzels 1997), whereas allophonic nasalization of post t onic vowels was relatively rare in all dialects, including the Carioca dialect (Abaurre & Pagotto 1996).

PAGE 75

75 nasalization is most systematic in stressed vowels, whereas contrastive nasalization is realized independently of stress. In the present study, stress is deemed to be a fac tor potentially influencing variation in UP vowel nasalization. In Brazilian Portuguese there are derived and non derived nasal diphthongs. Only three occur in the non can also arise a s the result of affixation in inflectional morphology. In verb forms, the third person plural always ends in [w] or w] falam falem falar The diphthong [j] occurs as the irregular plural of words whose singular forms ends in [w], as in [kw] co ~ [k js] ces alternations such as: [pox] pr u] ponho p e (Wetz els 1997:222). Tables 2 4, 2 5 and 2 6, provide an overview of derived and non derived nasal diphthongs (examples from Wetzels 1997). Table 2 4. Non derived nasal diphthongs in BP Diphthong Glossary [w] canho [j] me Table 2 5. Nasal diphthongs derived by inflection in BP Diphthong Glossary [w] falam from falar [j] ces from co [j] pe from pr canhes c annons from falem from the verb falar As shown in Tables 2 4 and 2 5, nasal diphthongs occur predominantly in word final position. Words which end in a nasal diphthong usually are stre ssed in the final

PAGE 76

76 syllable. Thus, nasal diphthongs do not occur in syllables closed by a consonant. Wetzels also argues that nasal diphthongs occur word internally only in derived words, as shown in Table 2 6: Table 2 6. Word internally nasal diphthongs Diphthong Glossary [w] m ozada [j] coraezinhos [j] cezinhos 2.5 2 3 Phonemic interpretations of Portuguese nasal v owels Vowel nasality in Portuguese has not been a peaceful field of enquiry fo r linguists. The thorniest problem of all has been establishing an adequate phonemic description and analysis of nasal vowels (Head 1970). However, it has been shown that the disagreement is due more to differences in the theoretical approach adopted by sc holars than to differences in the phonetic facts (Vandressen 1975). Thus, this complexity has led to numerous studies dealing with the Portuguese vowel nasalization process mainly from a generative theoretical framework. Almeida (1976:369) points out the two main reasons for the controversy surrounding the phonemic interpretation of phenomena and a rigid taxonomical analytical procedure. Analyses of Portuguese nasal vowels have been basically within the generative theoretical framework. The main hypothesis has been that nasality is predictable in Portuguese. Nevertheless, this hypothesis has been theoretically and empirically challenged. The basic generative hypothesis po stulates that the surface nasality of Portuguese vowels and diphthongs is derived by assimilation. Thus, all nasality is derived from an underlying tautosyllabic nasal consonant and its emergence as a full

PAGE 77

77 consonant in derivation provides the main supporti ng evidence of this hypothesis. To illustrate this statement, we will consider the following pairs in Table 2 7 (examples from Shawn 1986:116): Table 2 7. Nasal vowels alternations in basic word and suffixation Glossary um nico bem benfico Table 2 7 shows a nasal vowel in the basic word alternates with an oral realization followed by a heterosyllabic nasal consonant. Yet, the sequence [ ] an d [VnV] is also found in prefixation, such as the prefix of negation /in /, as in impossvel possvel inofensivo ofensivo 17 Consequently, as Shawn argues, t he generative approach does not account adequately for [VnV] sequences and dialectal variations. Nevertheless, the analysis of Portuguese nasal vowels has been handled differently by structuralists who are of the opinion that both oral and nasal vowels are phonemic. The following represents a systematic review of former interpretations of the phonemic interpretatio n of Portuguese nasal vowels. They are divided into three fundamental views: the biphonemic view, the archiphonemic view, and the monophonemic v iew (adapted from Almeida 1976 :369 72). The and Reed & Leite 1947) : The biphonemic overlap with vowels and which occurs after the nasalized vow els, before open juncture 17 Shawn (1996:117) maintains that the generativist treatment of the [VnV] surface sequences range from marking words with a diacritic feature, such as learned, popular, foreign, native or nonnative, which accounts for their failure to undergo certain rules.

PAGE 78

78 as [ ) analysis constitutes the first study on Brazilian pronunciation. This description is based on the speech of a native from Vitria capital of Espritu Santo in southeastern Brazil, supplemented by material from a Portuguese grammar book. With regard to nasal vowels, Hall proposes a (~) as a phoneme of nasalization. This phoneme would be superimposed on an oral vowel phoneme in the same way suprasegmental phonemes such as stress or pitch are superimposed on various vocalic segments (Lipski 1975:63). Reed and Leite (1947) represent the second analysis of the segmental phonemes of Brazilian Portuguese. The study is based on the speech of So Paulo, Brazil. This analysis represents a seven vowel system and treats nasal vowels as allophones of their respective oral vowels followed in the same syllable by a nasal consonant phoneme (Reed & Leite 1947:197 98). orais Barbosa 1962 and Mattoso Cmara 1953 1970) : an analysis of Portuguese nasal vowels as a sequence of vowel plus nasal archiphoneme. Morais Barbosa (1962) considers each nasalized vowel to be followed in the same syl lable by an archiphoneme of nasality /N/. Morais Barbosa argues that this treatment is a more economical description because it gives one single nasal archiphoneme instead of five nasal vowels. In addition, Morais Barbosa brings forward another argument vowel nasalization. He notes that the /b,d,g/ phonemes when intervocalic, show ] but not so after closed syllables or nasal vowels, in which case the variants are the plosives [b,d,g].

PAGE 79

79 Cmara (1953 1970) provides the first analysis of the segmental phonology of Brazilian Port uguese carried out by a native speaker. He interprets nasalized vowels as a vowel phoneme followed by an archiphoneme of nasality. In addition, Mattoso vowels in Brazilian Portuguese: (a) nasality occurs in all vowels which are followed by a nasal consonant, whether the consonant is in the same or in the following syllable (both in cama campo nasalized); (b) nasal vowels have the same effect as a closed syllable. Evidence of this fact is the presence of the intermittent [r] which occurs word medial after closed as well as open syllables, while the flap [ ] occurs word medially only after an open syllable; after a nasal vowel the only possible R sound is the intermittent [r]; therefore, as Mattoso Cmara argues, a syllable with nasal vowels is to be considered as closed; (c) word final nasal vowels cannot go into crasis with a following word initial vowel; (d) the existence of an epenthetic nasal consonant between nasal vowels and following vowels ( uma nio ninho (e) the existence of a diphthongized vowel word finally before a pause, in which the second element is a homorganic nasa lized vowel (Mattoso Cmara 1953 :89 97). m 1954, 1962 and Ld tke 1953) : Hammarstr m (1954, 1962 ) and Ldtke (1953) propose treating nasal vowels as single phonemic units. Further, t argues for the communicative irrelevance of the nasal consonantal segment after nasal vowels. For Hammarstrm th existent except before plosives and even here only as

PAGE 80

80 a reduced transition segment; moreover, it is said not to be present in the 1962: 25). Azevedo (2005) in his introductory overview o f the Portuguese language states that Portuguese is one of the two standard Romance varieties in which nasal vowels appear as independent phonemes, the other case being standard French. In his treatment of Portuguese vowels, Azevedo posits that nasal vowe ls contrast with their oral counterparts as in: l l [l cito cinto Moreover, Azevedo also discusses the ways in which the phonological nature of nasal vowels has been interpreted. The first interpretation postulates five independent nasal /, which yields the phonetic nasal vowels [ second interpretation postulates phonological sequences of vowel plus a nasal consonant that may be interpreted as /n/, thus: /an, en, in, on, un/. As Azevedo argues, each sequence undergoe s two phonological processes: vowel nasalization followed by ]) (2005:37). As these differing approaches demonstrate, the phonemic interpretation of Portuguese nasal vowels is a controversial issue among lingui sts. The studies cited have sought to determine whether Brazilian Portuguese nasal vowels are monophonemic, biphonemic or present a variant form. Moreover, these conflicting proposals show two problems: the number of vowel phonemes posited and the interp retation of the nasal vowels. Further, each analysis refers to a different Brazilian Portuguese variety and uses a different application of lin guistic theory. As Almeida (1976 ) argues, the difficulty in finding an adequate phonemic interpretation of the P ortuguese nasal vowels does not lie in the facts but in how theory is applied to them

PAGE 81

81 On the other hand, for Lipski all these classification attempts are quite arbitrary since they depend on formal patterns. He furthermore discusses that the different s considered, and also on the data ignored, present a suitable case for either possible 1975: 63). In this respect, Lipski proposes a mo re realistic analysis that would take into account the structure of the entire language as well as different aspects of the nasalization process itself. 18 Thus, in order to account for vowel nasalization phenomena in UP in Rivera, it is indispensable to tak e into consideration the variation in aspects of the nasalization process present within the same variety and the divergence of this process among Portuguese varietie s. Further, it is imperative to avoid the tendency to look for language specific data to justify universal hypothesis of vowel nasalization. 2.5 2 .4 Analysis of vowel n asalization in Portuguese Mattoso Cmara (1972) provides a diachronic and synchronic analysis of Portuguese nasal vowels. He notes that although optional nasalization of stres sed vowels in position before a syllable beginning with a nasal consonant constituted a new datum in the phonological system of European Portuguese, this process was always the general rule in Brazilian Portuguese, where it was associated with a concurrent change from /a/ to / / and from an open mid vowel to the corresponding closed mid vowel. Mattoso Cmara observes that in Latin any consonant, nasal or non nasal, could close a syllable. But, early in the history of the language, syllable final nasals we re 18 For Lipski, the difference between these interpretations is a matter of terminology and does not result in a substantially different analysis from the standpoint of the overall phonological process (1975:63).

PAGE 82

82 dropped in position immediately before /s/. As a consequence of this change, the first syllable was opened (i.e. consul > /kosul/, mensis > /mesis/, the remaining /s/, which became intervocalic, underwent the normal lenition to /z/ (ex.: mesa /meza/, from mexa) (1972:50) Mattoso Cmara further emphasizes that syllable final nasals were reduced in the environment of any other following consonant, and the phonological closing was completed by strongly nasalizing the vowel. Mattoso Cmara points out tha t this process occurred only in the Lusitanian Romance; in Castilian the nasal consonant alone, as in Latin, persisted (1972:50). In addition, Mattoso Cmara posits that eventually the way to indicate nasalization in modern Portuguese was to adopt the pr actice of writing / m/ or / n/ according to the Latin usage, in syllable internal position and in syllable final position after / e/, /i/, /o/ and /u/ a/ (i.e. lana > la > l, orphnu > rfo ). Regar ding final nasal vowels in Portuguese, Mattoso Cmara considers them as phonetically diphthongized and a characteristic especially noticeable in the So Paulo variety. Further, this phenomenon has received special attention, particularly in the case of the pair /e(n)/, which is a (phonetic) diphthong in both Brazil and Portugal. Yet, diphthongized nasal vowel with which it can contrast in order to create a distinctive opp /u(n)/, /o(n)/, and even /a(n)/ when the diphthong is conditioned by a nasal closing. In t

PAGE 83

83 Quicoli (1990) proposes an analysis for the phenomenon of vowel alternation and nasalization in Brazilian Portuguese. In this analysis he does not account for nasalized lized vowels is based on two major rules, Nasalization and Nasalized Vowel Raising (NV Raising), and the way these two rules interact with cyclic stress to explain some unusual effects (1990:329). Quicoli presents the rules that account for vowel nasaliza tion and some alternations affecting nasalized from a common dialect spoken in Rio de Janeiro and So Paulo, Brazil, where all that is required for vowel nasaliz ation is the presence of a following nasal consonant. In the dialect of Portuguese under investigation, there are essentially three basic environments in which vowels can be nasalized. Table 2 8. Rule 1: Stressed vowels followed by a nasal consonant G lossary af[i]ndo (fine sharpened) p[ ]na p[e]nacho (feather headpiece). f[u]ma (tobacco smoke) Table 2 9. Rule 2: Stressed vowel followed by a nasal consonant in a closed syllable Glossary (drop dropped) p[ ]nte p[ ]nteado (comb hairdo) (bottom sunk) Quicoli argues that this rule, plus a sub rule that deletes nasal consonants in final position, accounts for nasalization in forms such as l wool, ma apple. Table 2 10 sub rule 2 .1 shows deletion of nasal consonants in final position. Table 2 10. Sub rule 2.1: Deletion of nasal consonants in final position /# lan #/ /# masan #/ Nasalization Nasal deletion [ l ] [ mas ]

PAGE 84

84 Table 2 11. Glossary p[ ]nha p[ ]nhsco (rock cliff) verg[]nha verg[]nhso (shame shameful) (fist fistful) Table 2 11 shows examples where a vowel is nasalized before the palatal nasal nh ), regardless of stress The three cases, presented in Tables 2 8 through 2 consonant; but if the vowel is unstressed, then it will be nasalized only if the following nasal consonant is either high (i.e. the palatal nasal) or is followed by another He argues that t his rule subsumes the core cases of nasalization in Portuguese standard dialect. The diffe rences between nasal/non nasal are assumed to be categorical in the formulation of the nasalization rule. Moreover, to account for the change of [] to [ ], it is argued that in the official dialect of Brazilian Portuguese there are no nasalized low vowe ls in surface representation. The contrasts between surface oral vs. nasalized vowels are shown in Table 2 12: Table 2 12. Surface oral vs. nasalized vowels in BP (a) Oral vowels (b) Nasalized vowels +hi, lo: i u hi, lo: e o hi, +lo: E a O The characterization of Brazilian Portuguese [E, O, a] as [+low], seems to be supported by the view that these vowels behave as a class wh en affected by

PAGE 85

85 nasalization. Thus, to account for this phenomenon, Quicoli proposes a rule that raises low vowels to mid position when th ey are nasalized, as shown in Table 2 13: Table 2 13. Nasalization of low vowels and raise to mid position /kama/ /fO nte/ /tEmpo/ Stress Nasalization NV Raising [k ma] [f nte] [t mpo] The empirical claim made by the rule of NV Raising is that it applies only to vowels that have undergone Nasali zation, in other words, NV Raising is fed by Nasalization (Quicoli 1990:324). phonetic level the two varieti es have the same nasal vowels and these vowels are always [ low]. Further, the authors present an inventory of nasal vowels following word stress position (examples from Mateus & Table 2 14. Stressed nasal vowels Non final Glosse F inal Glosse [ ] canto tu] [ ] irm ] entre cinco fim [] ponto [] som [s] fundo comum Table 2 15. Pre stressed nasal vowels Glosse [ ] cantar entrar findar [] apontar [aptr] untar

PAGE 86

86 Portuguese, and that it is advisable to treat them as sequences of oral vowel plus nasal segment. In addition, they discuss the following arguments a gainst the evidence of lexical nasal vowels (2000:21): final consonant, ( e.g. desregrado Israel ) The pronunciation of /r/ after a nasal vowel only a and non nasal vowels to illustrate their hypothesis. Table 2 16. Rhotics after nasal and non nasal vowels a. Nasal vowels b. Non nasal vowels [] r onronar [o] coral [u] curral t enro [] pera [] perra ) [] h onra [] coro [] corro The examples show n above illustrate the diverse behavior of nasal and oral (2000:21). Table 2 17 and 2 18 show examples of base words which have the prefix /in /, /im/. (ii) Derivation from base words which have the same prefix /in /, /im /. Table 2 17. Base word derivations with prefix /in /, /im / before a consonant Glosse inteno [ ] incapaz imposto

PAGE 87

87 Table 2 18. Base word derivations with prefix /in / before a vowel Glosse inacabado inoporturno inaceitvel [in These words are derivations of base words: teno / inteno capaz incapaz posto imposto acabado inacabado oportuno inoportuno aceitvel inaceitvel vowel and a nasal consonant [in]. In order to acc ount for this phenomena, Mateus and but is phonetically realized as a nasal vowel before a consonant, or as a vowel followed by a nasal consonant, when the following co The alternation nasal vowel/vowel plus nasal consonant is further shown with word final nasal vowels and EP nasal diphthongs: Table 2 19. Alternation between nasal vowel/vowel plus nasal consonant in EP a. Word final nasal v owels b. Vowel plus nasal consonant [ ] irm ] irmanar fim final [fin w ] [] som [s] sonoro ru] The sequence of vowel plus nasal consonant (b) corresponds t o the word final nasal vowels (a). Table 2 20 shows nasal diphthongs in European Portuguese. Table 2 20. EP nasal diphthongs Po [p panito Le o [lj leonino [liunnu] The examples show n above reflect a distribution gap wit h respect to nasal consonants: they occur syllable initially (e.g. [p ntu] but they do not occur syllable

PAGE 88

88 final ztu] ). allowed in Po rtuguese (with a few exceptions ending in [ n] like abdmen [ m n] or smen m n] In BP these words have a final nasal diphthong (e.g. [ m ]) (Mateus & evidence to support thei r hypothesis that, underlyingly, Portuguese nasal vowels receive their nasality from a nasal segment that is deleted at the phonetic level (2000:23). Azevedo (2005) discusses that in Portuguese there are nasal diphthongs and triphthongs, as in: me [m j Moreover, Azevedo asserts that muito although in BP rui (Azeved o 2005:32). Furthermore, he posits that in Brazilian Portuguese a nasal consonant in syllable initial position tends to nasalize the preceding vowel. A word like cama tenho ng feature of of phonetic nasalization by assimilation to be responsible for regional variation and differences between from one speaker to another (Azevedo 2005:38) In Brazilian Portuguese, the lexical nasal consonant spreads the nasal feature over the preceding vowel when stressed, ( e.g. cama ) Moreover, in many Brazilian varietie s,

PAGE 89

89 the dorsal nasal consonant loses its [+consonant] feature and is r ealized as a glide, ( e.g. unha ). 2.5 2 5 Variation in Brazilian Portuguese nasal v owels Linguistic variation is rarely referred to in the literature on Portuguese nasal vowels although different types of linguistic variation have been briefly mentioned in the literature on vowel nasality. To my knowledge, there has not been an adequate description of Portuguese vowel nasal phenomena from a variationist perspective. However, Shaw (1986) provides evidence of the existence of dialectal variations conditioned by stylistic and sociolinguistic factors such as social class (80). Vandresen (1975) remarks that the tendency for stressed vowels before a nasal consonant in the next syllable to be nasalized is supported by the assumption that the degree of nasality is intensified by s tress, which in general intensifies other parameters as well. 19 Further, Vandresen also points out that the alternation of unstressed nasalized [] and unstressed [a]; and stressed non ariation with the oral vowels corresponding in quality (that is, stressed [e], [o], [i], [u]). Thus, it is relevant at this point to ask if we can account for the conditioning factors deriving these alternations, although different phonetic degrees of nasa lity proposed by Vandresen (1975), Almeida (1976) and other researchers are not analyzed in the current investigation. Degrees of nasalization do not appear to be linguistically (i.e. psychologically) relevant since there appears to be no language in 19 While it is normally agreed that there are different phonetic degrees of nasality, linguists differ as to the possible number of degrees and the context in which they occur. However, it is g enerally assumed that vowels nasalized tautosyllabically are stronger than those nasalized heterosyllabically; the latter are also believed to be allophonic while the former can be either allophonic or phonemic (Shaw 1986:75). It has been argued that degre e of nasality is generally stronger in Brazilian Portuguese than in European Portuguese. It is also believed that vowel nasality is stronger in French than in Portuguese (Morais Barbosa 1962).

PAGE 90

90 wh ich degrees of nasalization are employed to differentiate meaning (Quicoli 1990). Further, distinctions between degrees of nasalization are compelling for acoustic or instrumental treatments of nasalization, but are not a fruitful distinction for our purp oses in the present study. variety of Brazilian Portuguese, spoken in Rio de Janeiro, reports that vowel nasalization before a nasal consonant in the following syllable is ignored in many descriptions of the language, a lthough it is an accepted phen omenon in Brazilian Portuguese. Lipski further points out: Although this latter sort of nasalization in the Carioca dialect is much more marked in some words than in others, it appears to be a basic and ever spreading facet of the vowel nasalization process, in fact its more general manifestations, and therefore the basic nasalization rule must be fur ther extended to cover every instance of vowel nasalization, ph onemic or otherwise (1975:67). Lipski (1975) strongly points out the need to account for every instance of vowel nasalization phenomena in Brazilian Portuguese although even though this type of vowel nasalization does not result in a phonemic contrast with its oral counterparts. Further, Lipski remarks that these phono logical manifestations show that BP is moving in the direction of adopting the maximally general universal schema of vowel nasalization: V /___ N. Azevedo (1981) documents the existence of some variation among dialects with regard to the possibility of stressed nasalized / /. In Carioca speech [o] is the norm (Mattoso Cmara 1953:77, Head 1964:179), whereas for speakers from So Paulo and Minas Gerais, both options exist. Azevedo argues that in this case nasalization is less frequent and rather slight (1981:23).

PAGE 91

91 As the literature review indicates, dialectal variation in vowel nasalization in Portuguese has been somehow neglected. Althou gh vowel nasalization before a nasal consonant in the following syllable has been attested in EP and BP the linguistic constraints conditioning its variation has not been investigated. One of the challenges seems to be the disagreement on the adequate inc orporation of linguistic variation into grammatical theory. An additional aspect to take into consideration is the hypothesis then data approach that favors implied assumptions or generalizations. Thus, the quantitative analysis of vowel nasalization pheno mena in contemporary UP is of paramount importance since it would provide a sociolinguistic distribution of a neg lected linguistic feature in this variety 2.5 2 .6 Vowel n asalization in UP Although the literature on UP contains incidental references to the presen ce of vowel nasalization in this variety this phenomenon has not been investigated yet. A review of the literature on UP indicates that vowel nasali zation is a variable feature However, one of the challenges for the description of UP vowel nasali ty remains the lack of previous research. Rona (1965) attests to the presence of the Portuguese distinction of oral and nasal vowels in all the F ronterizo varieties but does not provide any quantitative data. Rona observes that in Portuguese words, nasa l vowels are maintained, whereas in Spanish words closed word finally by /N/, the nasal is lost and the vowel is nasalized, thus, creating lexical items that do not exist in either language ( i.e. ), [ bot ] botn [ bot ] verbo botar ). Further more Rona remarks that in the Melense variety, Portuguese words ending in o and Spanish words ending in n, have yielded only one ending o (1965:43).

PAGE 92

92 The other three F ronterizo ation are said to have kept the Portuguese diphthong o. This has caused an opposition between a nasal and an oral diphthong in F ronterizo : [bot u ] botado bo tu ] botn my translation). Moreover, Rona states that vowel nasalization, before a nasal consonant in the following syllable, is not characteristic of F ronterizo. Rona analyzed the writings of F ronterizo schoolchildren and found the /n/ deletion in words such as: *domigo auqu e pieso these examples are evidence that nasal vowels are unique phonemes in F ronterizo. ronterizo las vocal e s nasal e s son fonemas nicos 20 (Rona 1965:36 my translation) Hensey (1982 :15) argues that /a/ results show that Riverans nasalized /a/ as a low vowel in 70% of the cases. On the other hand, Lipski (1994) in a study of Latin American dialectology posits that vowel nasalization in F ronterizo is variab le and can be measured according to its degree of approximation to the Portuguese vowel system (1994:343). As it can be observed, the existence of Portuguese vowel nasalization studies is sparse, although they do indicate the presence of the phenomena in UP variety. Thus, the quantitative analysis of vowel nasalization in UP would provide empirical evidence of the alternations available for speakers in this variety It is therefore of paramount importance to base analysis on observable data to account appr opriately for the linguistic and social variation of vowel nasalization in UP. 20 los nios haban aprendido a escribir con forme a children have learned to write according to Spanish rules (1965:36 my translation).

PAGE 93

93 In this chapter I have outlined the theoretical framework that supports the present study. I have explored the main aspects of the quantitative variationist approach to sociol inguistics to account for the phonological variation in UP. One of the central tenets behind quantitative sociolinguistics is that variable and categorical processes are not random but rather rule governed (Sankoff 1978). A quantitative combined with a q ualitative approach, which provides a better understanding of the intersection of the linguistic and social factors, is implemented in the present investigation. Indeed, quantitative research has shown that linguistic variables are constrained by both inte rnal another variable form. The speech community and social networks analytical units are regarded as key components in empirical linguistics. The city of Rivera i s a close knit speech community and thus a social network approach can account for the patterns of relationships analysis has been employed in many studies in variatio nist studies from the 1980s to the present. The idea of using social network as a speaker variable was developed in the Belfast study (Milroy 1987), as a way to investigate the capacity of close knit networks to function as a norm enforcement system. Thu s, speech community and social networks are adopted in this project as analytical tools. In addition, I have outlined the linguistic varieties coexisting along the Uruguayan Brazilian border. Due to intricate socio political and economic circumstances, Sp anish and Portuguese and its regional varieties, have coexisted for many centuries in the

PAGE 94

94 frontier. The competing varieties provide the speaker with a selection of linguistic variables to choose from according to their social and ideological characterist ics. The literature reviewed has described the main aspects of the two linguistic variables under investigation, linguistic realization of /ti/ and /di/, as a dental stop or as an affricate and the absence or presence of vowel nasalization in Uruguayan Po rtuguese. Palatalization of dental stops is one of the most salient features of Carioca variety and several other Brazilian varieties Previous descriptions of UP mentio n none or little palatalization of dental stops (Rona 1965, Hensey 1972, Lipski 1994) In studies conducted in the Rio de Grande do Sul state, Hensey (1972) and Bisol (1991) show that the conservative pronunciation of dental stops is receding. In fact, palatalization of /ti/, /di/ is becoming the new trend among young speakers willing to align themselves with standard Brazilian Portuguese. This new trend seems to be spreading across national stratification in rendering the palatal realization of dental stops in Rivera. Carvalho points out that the application of the palatalization rule in UP illustrates a theory of dialectal diffusion in Rivera. Indeed, Carvalho argues that palatalization of dental stops in UP is a variable related mainly to age. Further, mid middle socioeconomic class women are using the symbolic value of the standard variable to signal urban orientation, different from the traditions of the rural border community. The literature reviewed on Portuguese vowel nasalization has shown two importa nt aspects that motivates the analysis of vowel nasalization in UP. On the one hand, linguistic variation is rarely referred to in the literature on Portuguese nasal vowels. Although, different types of linguistic variation across individuals and varietie s in

PAGE 95

95 BP has been briefly mentioned in the literature on vowel nasality. However, to my knowledge, there has not been an adequate description of Portuguese vowel nasal phenomena from a variationist perspective. On the other hand, analyses of Portuguese na sal vowels have been basically within the generative theoretical framework. The main hypothesis has been that vowel nasalization is predictable in Portuguese. Nevertheless, this hypothesis has been theoretically and empirically challenged. Thus, in order to account for vowel nasalization phenomena in UP in Rivera, it is indispensable to take into consideration the variation of aspects of the nasalization process present within a same variety and the divergence of this process among Portuguese varietie s. Further, it is crucial to avoid the tendency to look for language specific data to justify universal hypothesis of vowel nasalization. Therefore, the quantitative analysis of vowel nasalization in UP, would provide empirical evidence of the alternations av ailable for speakers in this variety. The present study, in sum, represents an attempt to fill this research gap. The next chapter describes the research methodology. It includes a description of the speech communit y of Rivera with emphasis on social an d economic aspects An account of the fieldwork, both preliminary and data collection phase of the research, is described as well as the sociolinguistic interview techniques to collect optimal data for on criteria and the composition of the final speech sample. The linguistic and social variables under consideration are circumscribed, and data analysis including the extraction, transcription, coding, and quantitative analysis of the final sample is desc ribed.

PAGE 96

96 CHAPTER 3 METHODOLOGY 3.1 Quantitative Methods and Dialect Research William Labov (1972:99) discussing linguistic methodology, classified the different subfields of linguistics according to whether their practitioners were primarily to be found Schilling Estes furthermore states, sociolinguistics steps beyond the laboratory and out into the street, to gather data on language as people use it in everyday life (20 02:17). linguistic research, inasmuch as my main objective was to gain access to the vernacular, the uncensored style in the speech of the people from Rivera in their dai ly lives. In conducting sociolinguistic research, it is important to study the community in situ a nd to overcome what Labov (1972 1 archers like me because of the influence of audience in determining language choice (Milroy & Gordon 2003:71). ethnography of communication research in Belfast 2 as well as Blom and G (1972) ethnographic work in the Norwegian town of Hemnes. One of the first challenges I encountered was how to enter and present myself to a community about which I had 1 2 The Mil rigid ethnic and religious divisions make it necessary that the fieldwork be conducted by a woman, Lesley Milroy, rather than a man, and important that she enters each community not as an outsider but as a research within a social network framework (Chambers 2003:134).

PAGE 97

97 acquired knowledge only through readings and library research. It was importan t for me to day experience and the socio geographic unit that is the s peech community In this chapter I will describe the research site that is the speech community of Rivera. I will also describe the two fie ldwork trips I carried out in Rivera: the preliminary fieldwork, in which much of the ethnographic work was done, and the data gathering fieldwork, during which most of sociolinguistic interviews were conducted. I will also describe the Rivera corpus, sp variables under examination. I will provide as well a description of the data analysis, including the data extraction techniques used, data transcription, and coding. Lastly, I discuss the main sta tistical analysis performed in standard quantitative sociolinguistics. 3.2 The speech community of Rivera The city of Rivera and its Brazilian twin city Santana do Livramento form one urban complex; in a way, it is one city with strong cultural, linguist ic and ethnic ties. The population of the city of Rivera is approximately 100,000 inhabitants. It is considered to be the largest urban center on the border. Santana do Livramento has a slightly smaller population with 90,000 inhabitants. Currently, Rive ra and Santana do Livramento are economically and demographically the largest and most important cities on the Uruguayan Brazilian frontier. The long sustained interaction and contact between both cities allow citizens from these two countries to frequentl y cross borders to carry out everyday life activities. Lack of immigration or customs checkpoints along this border allows community members of both cities to interact freely. It is common to see community residents from both cities crossing the Parque In ternacional to carry out daily activities, such as working,

PAGE 98

98 attending school, grocery shopping, playing sports, etc., without any customs control or checkpoint There are no geographic barriers separating Rivera from Santana do Livramento; thus it is com monly known as la frontera de la paz The P arque Internacional which is located between the two cities celebrates the unity and integration of the two countries. In the center of the P arque I nternacional is an obelisk with two clocks, s howing Uruguayan and Brazil ian time. There are also plaques that narrate the foundation of the cities along with Uruguayan and Brazilian flags waving side by side. At night y oun g people from both cities gather downtown, along the main avenue, Avenida Sar and to socialize and drink mate 3 an infusion prepared with dried leaves and one of the main G aucho culture social practices. Both ethnic groups, Brazilians and Uruguayans, share the G aucho traditions of the Pampas 4 As a consequence of the geographica l, cultural and social closeness, mixed marriages and domestic partnerships are a frequent fact of life. It is common practice and view as an advantage to have one child born on Uruguayan soil and another in Brazil. The term doble chapa 5 who have dual nationality eithe r by birth or mixed marriage. Being a doble chapa is perceived as an advantage in this community, 3 Mate is traditionally drunk in a social setting among family and friends. It is considered to be the national drink of Argentina, Paraguay and Uruguay. It is also a common social practice in the Rio Grande do Sul area. 4 The common Gaucho tradition is reflected in a body of regionalis t literature in Spanish or Portuguese, Martn Fierro (Hensey 1972:11). 5 The term originated during the 60s when Uruguay and Brazil established a car registration system in which a vehicle could circulate within Rivera or Santana city limits.

PAGE 99

99 since it entails better education and job opportunities not only in Santana but anywhere in Braz il. The main economic activities are duty free shopping, sales, and services Informal commerce is also an important source of revenue for the two cities. Along the imaginary line 6 separating Uruguay and Brazil, right next to the P arque Internacional it is common to find food stands, informal money exchangers and an enormous array of stands selling Paraguayan and Argentinean contraba nd goods. C ontraband, on both a small and large scale is part of Rivera S nce of a customs office right along the border, in my own opinion, does not seem to present an obstacle for this lucrative activity. When asked about this fact, community dwellers seem to perceive it as an ordinary and natural commercial activity character istic of their unique historical circumstances. Avenida Sarandi, offer a wide variety of free shops attract Brazilians from cities as far away as Porto Alegre, mai nly duri ng long weekends. The presence of Brazilian dwellers in Rivera increases the opportunities to communicate in Portuguese language. It is interesting to note that people from Santana are not as keen to speak Spanish a s Riverans are to speak Portugue se As a result, Riverans working in the downtown area, where the majority of economic opportunities exist, feel the need to communicate in Portuguese on a daily basis 6 ory of the nation as an imagined community in relation to identity construction in borderlands (Church 2007:18)

PAGE 100

100 Portuguese is the language commonly used along the border, since all the commercial activities are located in this busy area especially along A venida Sarandi On the other hand, S panish language is not used in Santana. When crossing to Santana to shop or to work, Riverans are expected to communicate in Portuguese. During my fieldwork on the border, I witnessed many verbal exchanges at duty free shops, department stores, and informal commercial booths. Store employees at the duty Portuguese, enabling them to carry out a flawless communicative interaction with the Portuguese monolingual buyers. Uruguayans a but these businesses provide an important source of employment for t hem. Self reported data from participants revealed their opportunities to engage in daily interactions with Santana teachers, mainly females, revealed previous employment as duty free employees. It is in these busi nesses, where face to face interactions take place on a daily basis, that Riverans derive their Portuguese linguistic model. During the month of July 2007, when I conducted the first fieldwork, Brazilian tourists in Rivera were so numerous that it became a problem due to a lack of hospitality services available in the city. The increase in tourism was accompanied with a bit of resentment from Rivera residents, who saw their city invaded by a large number of Brazilian shoppers with more buying power. Mix ed opinions can be found among residents, about the influx of Brazilian tourism. It is common to hear Riverans complaining about the fact that Brazilian tourists are buying up much need ed basic

PAGE 101

101 products. However, for duty free shop owners and small store keepers, it is always good news to see an increase of tourists with buying power. A similar situation is reported by Labov (1963, 1972) in conducting his first notes that the island is divided into two distinctive sections: up island, where the summer homes are located, is strictly rural and contains few villages and farms. Down island is where three fourth of the population lives (Labov 1972:4). Labov found a social cor relation between centralization of the diphthongs (ay) and (aw) with expressions of strong resistance to the incursions of summer visitors on the island. In addition, Labov reports that while people in close knit groups opposed seasonal visitors on the is land, residents in the up island section welcomed them. Labov argues that Vineyard, although the constant pressure from this direction, and the growing dependence of the isla nd upon a vacation economy, has had powerful indirect effects Opportunities for communicative interactions with Portuguese speakers are vast in Rivera. Not only do commercial interactions provid e a platform for linguistic exposure, Riverans are also eager consumers of Portuguese media. Brazilian radio stations broadcasting popular music are favored over Uruguayan stations, many times played using loudspeakers to adve rtise sales or special events Community dwellers access the powerful Brazilian television network Globo for free, while access to Uruguayan national television is accessed through paid cable. It was my experience while conducting fieldwork, that community members from Rivera were m ore aware of

PAGE 102

102 Brazilian social and political issues than Uruguayan politics. The impact of Portuguese As d iscussed above (§1.2.1), Rivera and Santana do Livramento maintain strong historical and cultural ties. The twin cities also share community traditions, values, and language usage norms. Both cities celebrate internat ional festivities such as Carnaval de la Integracin, Cielo sin Fronteras Fiesta de la Hermandad y el Patriotis mo and Desfile Internacional de la Farroupilha 7 These celebrations bring dwellers from both cities together, where t he municipal band plays traditional Gaucho music in the P arque Internacional followed many times by a patriotic speech, where the twin fraternal conviviality is reiterated. Although national borders can be thought as a place where sharp dividing lines are drawn, the dynamics of integration, globalization and transformation on the Uruguayan Brazilian border gives evidence to the contrary. As Hamel (2003) posits, (131). Nevertheless, this regional i ntegration at the local level, where trade, cultural contact, and exchange are reinforced, contrasts with the slow action of the government (Hamel 2003:132). 8 7 Revoluo Farroupilha was a Republican uprising that began in the Brazilian state of Rio Grande do Sul in 1835 due to difference between the economies of this state and the rest of the country. Source: Intendencia Municipal de Rivera. 8 Mercosur initiatives show a bit of resistance to integration itself (2003:131).

PAGE 103

103 3.3 Fieldwork The fieldwork was conducted during two separate periods of time, during the sum mers of 2006 and 2007. The first period lasted six and the second eight weeks. The first trip was basically preliminary fieldwork in order to assess the feasibility of the project and to establish social contacts in order to gain access to the community. The second period of fieldwork was devoted to the data collection phase of the project. Gaining access to the community, and more importantly, to the vernacular is one of the most challenging tasks for a sociolinguist. In order to achieve this goal, I ad opted throughout both fieldwork trips a participant observation role, which entails participant involvement with the community in their natural environment. These social contacts proved extremely useful especially during the second fieldwork trip when the data collection phase of the project took place. 3.3.1 Preliminary fieldwork For the preliminary fieldwork, one of my objectives was to get acquainted with the linguis tic behavior, to acquaint myself with their ways of life and to obtain a good sense observat opposition between local and non local values in the Rivera speech community. Prior to arriving in Montevideo, I had established preliminary contact with faculty members of Univer sidad de la Repblica, in Montevideo. It is at this institution that Drs. Elizaincin, Barrios, and Behares have been conducting research on the Dialectos Uruguayos del Portuguese, or DPU (Uruguayan Portuguese Dialects), as it is known by

PAGE 104

104 linguists. These scholars gave me their advice, support and all the necessary information on the linguistic situation in the cities along the Uruguayan Brazilian border. In addition, they met with me on several occasions to discuss general issues regarding their research on the border and provided me with excellent contact information in Rivera. While in Montevideo, I visited university and public libraries, and governmental offices ( Instituto Nacional de Estadistica, Mercosur office, Superintendencia de Montevideo) to g ather information related to the research. I also conducted informal the UP variety spoken in Rivera or baiano as it is commonly known among the capital city dweller s. The information about attitudes toward UP gathered in Montevideo confirmed the thoughts and feelings shared by the speakers of the speech community of Rivera. UP is also known as or (Brazilian style speech) by community m embers in Montevideo. The local variety has been stereotyped and mocked, considered to be a sign of lack of education and patriotism. Community members of Rivera are aware of the stigma attached to their speech and prefer not to reveal it to outsiders. F aculty at the Universidad de la Repblica in Montevideo put me in contact with the director of the Centro Regional de Profesores CERP in Rivera. 9 CERP is a three year educational institution where students from different small towns in Rivera and surroundi ng departamentos attend classes in order to become school teachers. Students willing to become school teachers must have graduated from the liceo, or high 9 The Centro Regional d e Profesores CERP is part of the Asociacin Nacional de Educacin ANEP. ANEP is the institution in charge of all major educational decisions in Uruguay.

PAGE 105

105 school, and pass an entrance exam designed and evaluated in Montevideo. This center offers school tea cher training in several areas such as math, social sciences, language and literature, physics and English as a foreign language. The center does not train teachers for the teaching of Portuguese. Many of the students at CERP, especially those coming from Rivera departamento are UP speakers. During my fieldtrips in Rivera, I stayed at the female residence building at CERP. This unique setting allowed me numerous opportunities to observe, learn, and interpret rules of language choice, language use, and lin guistic residence are located on the outskirts of the city and at the entrance to Rivera city, on the main interstate and just 15 miles away from the city downtown. This loc ation in the outskirts of the city put me close to the peripheral barrios or neighborhoods where most of the inhabitants speak UP. Furthermore, being part of the setting that I was studying gave me an advantage in gaining access to the vernacular speech. 3 .3.1.1 An ethnographic approach One of the challenges in sociolinguistic research is obtaining insider knowledge about the communities under investigation. This local knowledge is necessary to discover what is important for community members in order to be able to use this knowledge to interpret speech data. It is through local knowledge that a researcher can come to understand that the social value of phonological variants may differ from community to community. The ethnographic approach to sociolinguis t inquiry enables the researcher to achieve this goal, and it is especially effective when the researcher is (1972) work in Norway by adopting the ethnographic approach for th is research. This

PAGE 106

106 local cultural categories before any structured elicitation of language was even Although I did not conduct as many interv iews in UP as I had planned during the preliminary fieldwork, I was able to gather a good amount of qualitative data. Indeed, the period of time in which I engaged in participant observation activities in Montevideo and Rivera, Uruguay and in Santana do L ivramento, Brazil, was fundamental to In fact as Feagin (2002:33) observes, participant observation is a qualitative method of data collection that supplements and aids in the interpretation of quantitative data. Further, as Milroy and Gordon (2003:68) emphasize, the benefits of participant observation are the amount and quality of the data collected and the familiarity with community practices gained b y the investigato r. During this phase of the research, I was given permission by the local school authorities to visit public schools. I visited most of the public and private schools in Rivera, volunteering as teacher aid and substitute. At public schools, I could hear students communicating with each other in UP mainly during lunch break and recess. At private institutions, students were quick to distance themselves from UP speakers by of the Portuguese variety they speak. At the same time, pri vate Parallel to my work with the students, I established contacts with teachers, administrators and parents, who would share their impressions about the recent implementati on of the teaching of Portuguese in public schools.

PAGE 107

107 In my attempt to experience different spheres of social life in Rivera, I attended several community gatherings such as soccer games, open air markets, religious services, free shops, baking classes at the Universidad del Trabajo del Uruguay UTU 10 talks at Mundo Afr o, 11 cultural events organized by the Intendencia Municipal de Rivera, and Escuela Agraria classes, and I volunteered occasionally as a tutor at the residence. Thus, by adopting an ethnographic approach, I was able to become knowledgeable about the local norms in the speech community and establish social relationships with several community members who were in a position to introduce me to potential participants for the study. 3.3.1.2 Entering the c ommunity via social networks The director at CERP was my initial contact with the speech community. She kindly provided me with the first contacts with teachers and students in the area. This decision proved not to be very satisfactory. People woul situation constituted an obstacle for the research, and it also taught me to distance myself from any kind of authority in the communi ty. As the authors argue, referring to area schools, any association with the institution might rankly with her (Milroy & Gordon 2003: 68). My way of overcoming this situation was to distance myself from any affiliation with the authorities and to avoid any official role in the schools. I continued volunteering in the 10 UTU offers technical education to young adults. 11 Uruguay has a black population that few people ac knowledge. From the middle of the eighteen century, large numbers of African slaves were imported into the River Plate, a substantial number of which ended up in the Uruguayan cattle raising interior (Lipski 1994:339).

PAGE 108

108 classrooms, especially in the d ual immersion program, but introduced myself as a volunteer. By volunteering in the dual immersion program, geared initially toward participants younger than 15 ye ars old eligible to take part in the project. An additional challenge to overcome was being perceived as a community outsider. Many times people did not feel quite at ease talking UP when I was around them. At times people would quickly code switch from UP to Spanish upon noticing my presence. Most of the time outsider status represents a challenge to carry out research (Milroy & Gordon 2003:68). It was interesting to observe the code switching and the reasons the speakers offered for it nd carry on with the conversation. Nevertheless, with time I was able to overcome this difficulty and was successful in establishing and developing links with insiders in the community. Students at the residence started to show interest in my research when they realized that I could speak my own variety of Portuguese, enjoy Brazilian music, novela da tarde volunteered to be interviewed once they got to know me better. Students even Rosinha. They were a great source of practical local information about schools, ev ents, institutions, etc. As Milroy (1987) points out, these initial relationships are crucial to the design of the more structured part of the investigation, and it proved to work for the data collection phase of the fieldwork.

PAGE 109

109 3.3. 2 Data Collection Field work The data collection phase of the current project was conducted during the summer of 2007. An advantage of having performed preliminary fieldwork in the Rivera community was that I had established social links with insiders and had interacted with them in several community events. In addition, these relationships and the knowledge I had gained from them were very important for the design of the interview modules and for the structure of the whole data collecting phase of the research. I reentered the in the local history of Rivera. My approach was to contact the person I had kept in touch with. I will refer to her as Adita, a working class single parent of four children, with whom I communicated occasionally via email upon my return to the United States. Being interested in the local history and having a friend as a link to the community was extremely helpful in getting participants for the project. The convenient location of t he residence was suitable for gaining access to the barrios. area is surrounded with peripheral barrios such as b arrio Manduv, La Pedrera, b arrio Bicio, Rivera Chico, Santa Teresa, Santa Isabel, Cerro de Marco, etc. Among the more popul ated neighborhoods in Rivera are barrio La Pedrera across the street from the CERP, and barrio Manduv within walking distance of it. Most of the interviews with working class participants were conducted in these neighborhoods Adita my contact in barr io La Pedrera, introduced me to some of her friends from church and community organizations. Befriending a popular community member was a successful way to get access to the close knit working class community of Rivera. Adita was not only an insider in he r neighborhood, she was also well liked, well known to

PAGE 110

110 community members and regarded as one of the more active community members in the neighborhood. My experience was that when I approached potential speakers and mentioned referral, people felt at ease and showed a willingness to participate in the study. Community members started to become familiar with me, offering me local recipes and advice on how to go around Rivera. A similar situation is reported in the Belfast study (Milroy 1980) where fieldworkers were received with warmth, friendliness and trust, city areas. The school setting proved to be an excellent source of young informants. The Uruguayan public school system maintains an extensive schedule to cater to several school populations throughout the day. A typical l iceo starts class at around 7.30 am to 12 pm, followed by another session starting at 1 pm till 6pm, and there is also an escuela nocturna that ca ters to working young adults that starts at 8pm and ends at midnight. At the escuela nocturna I interviewed participants who had suffered from the At high schools and evening schools, my approach was similar to that of Eckert 12 : I tried to maintain a low profile and stayed outside of classrooms, preferring public areas such as cafete rias, hallways, and open areas in the school. This was an opportunity for them and eventually ask them for an informal interview. 12 Eckert (2000) spent two years stud Gordon 2003:68).

PAGE 111

111 M ost of the working and lower middle class participants were interviewed in the barrios while performing daily activities such as cooking, harvesting, building houses, fixing cars, cleaning, looking after children, cattle ranching, etc. Mid middle class speakers were interviewed at their wo rkplaces or at home. Working and lower middle class participants were not interested nor did they have the time to sit with me for an fact that the focus was not th e interview event but rather the task at hand reduced the pressure of the interview. occupation. As Labov (1972) argues, this factor correlates best with linguistic variation. neighborhood. The speech community of Rivera is a highly stratified society The wealthy and middle class inhabitants who identify with Montevideo culture, live in th e downtown area. The poor and working class inhabitants, who identify with local border culture, live in the barrios or neighborhoods located away from the downtown center While contacting and interviewing working and lower middle class participants was a fairly uncomplicated task (as speakers were easily identified by the neighborhood where they reside and the type of close knit networks to which they belonged), reaching s that people in the upper socioeconomic strata do not acknowledge the use of UP in their lives. However, in my experience it is quite difficult to get by in Rivera without some understanding of UP. One way to overcome this situation was to visit private institutions such as private schools and professional and governmental organizations. Younger

PAGE 112

112 speakers of the mid middle socioeconomic group interviewed for the present study attend private schools and either travel to Brazil for pleasure, or take Portugu ese classes at the Instituto de Idiomas repertoire falls short of that of the close knit network speakers. Through these professional and governmental institutions I was able to secure mid middle class sp eakers for interviews. Among the professionals I interviewed were journalists, lawyers, writers, business owners, land owners, government and educational authorities, etc. I must add that methodologically it was a sound decision to start out with the inte rviews in the barrios Middle class speakers in government and educational institutions were more interested in knowing the reasons why I was concerned with local culture and the speech of Rivera. I was even asked to participate in televised and radio int negative evaluations of the culture and speech of the barrios, clearly distancing themselves from it 3.3.3 Type of Data Collection In order to analyze the phonological v ariation in UP systematically, it was important to design an interview schedule that would provide many instances of the variation identified two types of conversational style, casual and formal. This procedure

PAGE 113

113 Hence, casual style, or the style closest to the so call ed vernacular is said to be the product of the minimum amount of conscious self monitoring (Milroy & Gordon 2003:200). As is common practice in sociolinguistic studies, the interviews were structured around pre determined conversational modules. The conv ersational modules included a group of questions around particular topics. Based on insider knowledge acquired during the preliminary fieldwork for this research, I selected relevant conversational topics for the speech community of Rivera, such as footb all festivities celebrated in Rivera, carnival, soap operas, local stereotypes, the teaching of Additional modules were modified from the ones used fo r a class project at the University of Florida, directed by Professor Blondeau investigating linguistic variation and change in the speech community of Gainesville, FL. The modules were structured around more general topics such as family, celebrations, ex periences from childhood, etc. Questions concerning attitudinal and value judgments about UP and BP were also fruitful topics. Demographic data such as age, occupation, residence, education, language history, family relations, etc. necessary for sociolin guistic analysis were gathered at the beginning of the interview. Participants were asked to give me their oral consent to participate and to allow me to record the interview. Due to the nature of the informal interviews and objectives of the project, wri tten consent was not required. 1 The length of the interviews varies between 45 and 65 minutes each. The high frequency of 1 Please refer to the Appendix section for Institutional Review Bo ard (IRB) forms.

PAGE 114

114 the variables selected namely, palatalization of dental stops and vowel nasalization provided ample tokens from which to select. A total of one hundred and six interviews were conducted in situ by myself. Fifty four of them were selected for analysis. The output of these interviews constitutes the prime focus of the present study. 3.3.3.1 Individual interviews The starting point f or individual interviews was the social networks found in the barrios to one of an apprentice, decreasing the degree of authority between the researcher and the speaker. Acco rding to Labov, the counter strategy of the sociolinguistic interview is to emphasize the position of the interviewer as a learner (Labov 1972:40). In addition, the favorable location of the residence where I stayed contributed to gaining access to potent ial participants. I took care to dress according to local customs so as not to create social distance with participants. When interviewing adolescents, I wore casual clothing, drank mate with them, many times joining them sitting on the floor to chat and as a way to show solidarity. Thus, I immediately created a connection with the speakers of the working class group. In fact, many interviews were conducted in more than one ult to get speakers involved in spontaneous conversations. In this respect, Riverans are regarded as being as friendly and outgoing as their Brazilian neighbors. class community gave me the advantage of observing and ide ntifying the quality of the relationships that exist in the community. The structural characteristics of the local networks in the barrios are close knit, dense and multiplex (Milroy 1980). Speakers are related by several indicators such as

PAGE 115

115 neighborhood, kin, religious and recreational associations, friendship and work ties. Indeed, several participants indicated that they were born and raised in the neighborhoods and that most of their family still remains there. These indicators create close knit ties w here solidarity and local identity are precious values in the community. 3.3.3.2 Group interviews Most of the interviews with adolescents were pair or group interviews. The use of pair interviewees creates a three way conversation, which eliminates so me of the awkwardness of two strangers having to speak one on one. Using this technique, the conversation seems to flow smoothly giving the fieldworker time to assess the flow of the interview event and determine whether to keep going with the same topic or introduce a new one. Another technique used to break down the interview structure was to study groups. The procedure used to elicit the vernacular from adolescents was modeled on the work of Labov and his colleagues (1972 ) in the African American comm unity in Harlem. An advantage of resorting to group interviews is that it steers the dynamic of the interview away from the one on one format facilitating the production of casual speech. This may be accomplished by having either two or more interviewers o r two or more interviewees (Milroy & Gordon 2003:66). Having an additional interviewer was not an option for me, but interviewing participants in a group setting was feasible for the present study. This technique proved to be very useful for my data coll ection in Rivera, especially when interviewing teenage boys and girls both in public and private educational institutions. Many times, during my visits to schools, I would approach a small group of adolescents who were either talking or playing on schools grounds, and I would start up a conversation with them. Normally I would start out with a group of four

PAGE 116

116 or five speakers and would continue the interview with the most talkative participants. These conversations took place away from classrooms in public s chool areas. An additional source of participants was found at the residence. Female adolescents were interviewed in groups and then individually. The group interviews were conducted in dormitories, while female participants were relaxing or preparing t o go out. The use of peer interviewing was a way to identify local norms, attitudes toward existent linguistic varieties, and gave me direct access to the vernacular. A simi lar situation is reported by Blo m and Gumperz (1972), Milroy and Milroy (1978) am ong other linguists working with group sessions to obtain linguistic data. Indeed, as Labov (1966) rightly points out, the close supervision exercised by the group over vernacular norms is so strict that a speaker must adhere to them or risk being taunted 3.3.3.3 Equipment All interviews were digitally recorded using a SONY ICD SX68 Digital Voice Recorder. I selected the recording setting mode STQH high quality (stereo recording mode) to obtain the best sound quality possible. Digital interviews were s aved on digital folders, then downloaded into a computer and converted into MP3 files. Sound files were then classified into PC folders according to the social characteristics of the speaker. In addition, PC folders were saved on a USB drive. An additio nal piece of equipment used was a foot pedal control to complete the transcription process. This discussion of fieldwork has served to illustrate the methodology adopted for both the preliminary and data collection phase of the present study. I have desc ribed the way I entered the speech community, gathered participants for the project, and the two types of sociolinguistic interviews conducted. The next section presents the speech

PAGE 117

117 sample, participant selection criteria, and linguistic and extralinguistic variables under investigation. 3.4 The Rivera Corpus The Uruguayan Portuguese/Spanish computerized corpus consists of one hundred and six sociolinguistic interviews with bilingual UP and Spanish speakers. Participants are divided by gender, by three age groups ranging from 15 to 70 years old, and by socioeconomic group, working, lower middle and mid middle. Fifty four participants out of the computerized corpus were selected for phonological analysis in the present study. 3.4.1 Speaker Selection Crit eria For the speaker selection criteria it was determined that the speaker had to be born or raised in Rivera and be a life long resident of the city or its immediate rural hinterlands. However, as already pointed out, a s a consequence of the geographical, cultural and social closeness, mixed marriag es and domestic partnerships are common It is a common practice and viewed as an advantage to have one child born in Uruguaya n soil and another one in Santana This is an ordinary fact of life in Rivera. Many o f the participants in the current project are doble chapa a colloquial term normally used to designate inhabitants who have dual nationality either by birth or mixed marriage. These participants have either married a Brazilian partner or have been born in Santana and reside in Rivera. The criterion for inclusion of doble chapa speakers in the sample was to be a life long resident of Rivera. 3.4.2 Speech Sample The final sample is shown in Table 3 1. The data presented here were taken from the total River a corpus sample of 106 speakers. In order to provide a sample of the

PAGE 118

118 speech community comparable with other studies, I included in the final sample fifty four participants. Age, gender, and socioeconomic class were taken as independent or extralinguistic variables. Participants are grouped according to their social characteristics. AGE. T hree age groups (1) 15 29, (2) 30 49, and (3) 50 70 years old. GENDER. 27 females (F) and 27 males (M) SOCIECONOMIC CLASS. T hree groups (1) mid middle, (2) lower middle, and (3) working socioeconomic group. As Table 3 1 indicates, participants were assigned a speaker identification number and grouped by age at the time of interview, socio economic status and gender. Additional social information, such as occupation and date of birth, etc. can be found in the appendix section. Table 3 1. Speech sample Age groups Working C. Low Mid. C. Mid Middle C Gender N 15 29 years old CL003 16 CL004 19 KL002 15 F MY012 23 CH003 21 SB008 15 F MX023 23 PA005 15 DR010 15 F PF024 19 WR014 16 LY021 15 M TN025 16 DD026 19 XNN13 16 M AW022 18 JG016 19 YY055 26 M 18 30 49 years old AQ074 41 AA064 46 RM039 40 F MD043 37 FF062 48 MA040 38 F IC034 30 MS028 35 ME07 0 49 F HJ049 34 CL051 36 PP058 33 M AI082 42 HF044 35 ET053 31 M AS046 40 HV050 30 WW057 40 M 18 50 70 years old MB072 50 JJ061 52 MA067 54 F SW093 61 GW085 62 DS066 50 F MN063 50 TY087 64 SS088 55 F GE106 63 HA099 58 VV103 60 M NN081 52 TS097 57 FW101 70 M TK083 50 DV077 54 AE080 54 M 18 N = 54

PAGE 119

119 This section concludes the description of the Rivera corpus. As shown above, the final speech sample consists o f fifty four participants grouped by age, gender, and socio economic status. All the participants selected for the final speech sample are bilingual UP speakers. Furthermore, additional speaker selection criteria were discussed. The next section describ es the dependent and independent variables analyzed in the present study. In addition, I outline in detail the linguistic and extralinguistic factor groups and factors considered to condition the two linguistic variables under examination in the present p roject. 3.5 Variables under Investigation This study focuses on the analysis of variation at the phonological level, th e variable realization of palatalization of /ti/, /di/ and vowel nasalization in the speech of bilingual speakers of UP. Based on data collected in participant observation and sociolinguistic interviews with fifty four local UP speakers, the study examines linguistic and extralinguistic factors influencing phonological variation. In what follows, I describe the linguistic and social facto r groups included in the analysis of the two linguistic variables under examination. 3.5.1 Linguistic Variables The study investigates the sociolinguistic distribution of two phonological variables in the speech of bilingual speakers in Rivera, Uruguay: palatalization of /t/, /d/ in /i/, and vowel nasalization. The linguistic variables selected for the current investigation were chosen based on their frequency in unstructured interviews, their saliency, and the asymmetric social stratification they show ed in preliminary explorations.

PAGE 120

120 3.5.1.1 The variable p alatalization of /ti/, /di/ One of the dependent variables in the study is the realization of /t/, /d/ in /i/ as a dental stop or as an affricate The non standard pronunciation is represented by the although this is a variable feature in several Brazilian varieties and in UP. This variable can occur in pairs like tipo and digo i The linguistics factors considered in the present study are shown in Table 3 2. Table 3 2. Palatalization linguistic factor groups and factors Dependent variable p palatal realization of /t/, /d/ in _i d dental realization of /t/, /d/ in _i Independent variables Linguistic factors 1. Tonicity s stressed syllable p pre tonic syllable x post tonic syllable 2. Following environment n nasal consonant p stop l lateral s sibilant v vowel r vibrant o pause, zero phonolo gical environment 3. Preceding environment n nasal consonant l lateral s sibilant v vowel r vibrant o pause, zero phonological environment

PAGE 121

121 The total number of tokens coded to analyze palatalization of dental stops in UP is 2,526. Each to ken was coded for the operation of internal, or linguistic, and external, or extralinguistic, constraints. Tokens are distributed as dental realizations 1,783 (71%) and palatal realizations 743 (29%). The three linguistic factors commonly found in the lite rature that condition the variable pronunciation of dental stops in BP (Bisol 1991) and UP (Carvalho 1998) are: tonicity, following, and preceding linguistic environment. 3.5.1.2 The variable vowel nasalization The second dependent variable under analysis is the absence or presence of vowel nasalization in UP. The total number of tokens coded to analyze vowel nasalization in UP is 2,121. Each token was coded for the operation of internal or linguistic and external or extralinguistic constraints. Tokens a re distributed as oral realizations 1,674 (79%) and nasal realizations 447 (21 %). Each token was coded for a series of phonological and morphological factors that could have an effect on vowel nasalization. Due to the lack of quantitative research on vowe l nasalization in varieties of Portuguese, linguistic factors were extrapolated from the historical and contemporary literature and from formal, prescriptive, and sociolinguistic studies that have been reported to have an effect on vowel nasalization. The linguistics factors considered in the present study are shown in Table 3 3. Table 3 3. Nasalization linguistic factor groups and factors Dependent variable n nasal vowel o oral vowel Independent variables Linguistic factors 1. Tonicity s s tressed position (non final/word internal) f final stressed (word final stressed) p pre tonic position

PAGE 122

122 Table 3 3. Continued Independent variables 2. Following environment n nasal consonant (non palatal) a palatal nasal p stop f fricative (non sibilant) s sibilant d glide o pause, zero phonological environment 3. Preceding environment n nasal consonant (non palatal) a palatal nasal p stop f fricative (non sibilant) s sibilant d glide l lateral r vibrant o pause, zero phonological environment 4. Functional category n noun t noun + diminutive v verb (no gerund or progressive forms) g gerund, progressive forms j adjective d adverb e preposition p pronoun c conjunction 5. Syllable structure m monosy llabic p polysyllabic 3.5.2 Social Factors The extralinguistic or independent variables considered in this study for both palatalization and nasalization linguistic variables are the following: participants, age, gender, and socio economic status. Ea ch individual correspond to a factor in the factor group.

PAGE 123

123 Table 3 4 Social factor groups and factors Social factors 1. Participants 54 speakers 2. Age 1 15 29 years old 2 30 49 years old 3 50 70 years old 3. Gender F fem ale M male 4. Socio economic group 1 Mid Middle 2 Low Middle 3 Working Palatalization linguistic and extralinguistic quantitativ e results will be discussed in Chapter 4 Nasalization linguistic and extralinguistic quantitative results will be discus sed in Chapter 5 3.6 Data Analysis Variationist analysis takes variation to be inherent to language. The mathematical and statistical principles of variationist analysis were developed by Sankoff, in collaboration with Cedergren, L abov, analysis program called VARBRUL is a computer program developed to handle statistical methods used in variationist analysis of linguistic phenomena. It handles both linguistic and social distribution of variant linguistic fo rms. The preferred statistical model of variationist sociolinguistics is logistic regression. As Paolillo (2002) points out, logistic regression is closely related to linear regression, which is another type of statistical modeling. These

PAGE 124

124 techniques pro amo For the data analysis of the corpus, I used the traditional protocol for phonological analysis. My first task was to lis ten to all the interviews. All speakers were assigned a unique pseudonym that correlates with the interview report 2 the transcription file and the recorded data. While I listened to the interviews, my primary objective was to identify linguistic variation across speakers of different social backgrounds, ages, and gender. At the phonological level, this is a straightforward task, that is, the variants may differ by an extra phonological feature or two (Tagliamonte 2006:70). For the present study, several alternations were considered, but either they did not fulfill the requirements to be considered a linguistic variable or did not show social distribution across speakers. Upon selection of the linguistic variables under investigation, palatalization of /t/ /d/ in / i/, and vowel nasalization, the next step was to circumscribe the linguistic environments. Palatalization and vowel nasalization linguistic and extralinguistic factor groups will be discussed on Chapter 4 and Chapter 5 respectively. The final s ection of this chapter describes the data handling procedure and the techniques implemented to represent the speech data faithfully in writing. All steps of the data handling of natural speech, extraction, transcription, coding, and GoldVarb analyses were performed by the researcher. 3.6.1 Extraction In order to provide a sample of the community comparable with other studies, I included in this analysis the interviews of fifty four speakers. I listened to each entire 2

PAGE 125

125 interview in order to capture the mo st animated and unmonitored section of it, in which the vernacular is most likely to be found. This procedure was repeated several times in order to represent the recorded data faithfully and consistently, and to achieve greater reliability (Tagliamonte 2 006:55). 3 The two phonological variables selected for analysis, palatalization of /ti/,/di/ and vowel nasalization, are very high in frequency of o ccurrence. Thus, ample data were collected from a rather small amount of recorded speech from each partici pant. With plentiful data available, the extraction process was limited to approximately fifty tokens per speaker. Additionally, because some lexical items are more frequent than others, I followed the standard approach of type token sampling (Wolfram 19 69:58, 1993:214), restricting the number of tokens per speaker to five. The type token question is token sampling is appropriate for 3.6.2 Transcription The transcription was performed using a transcription foot pedal to control the computerized digital recorder. All tokens were written out on a coding sheet, prepared individually for each speaker, with its surrounding phonological context. Additionally, I phonetically transcribed all the words that contained t he linguistic variable under investigation. Since the study deals with quantitative analysis of the data, I transcribed 3 Hansen (2001:213) in the study of lexical diffusion in French nasal vow els, reports the unsuccessful

PAGE 126

126 features relevant to the study; it was not considered necessary to transcribe all the interviews in their entirety. Upon completion of the transcription process, the number of occurrences was calculated. 3.6.3 Coding Coding is probably the most time consuming and important task in VARBRUL analysis. Each token that is, every realization of the linguistic variable under investigation is coded according to the linguistic environment presumed to exert effect each token of the variable in the data, and then assigning values to all the hypothesized indep Coding of palatalization of /ti/ /di/ was performed first. I included a phonetic transcription of the interval of speech containing the token under investigation and its surrounding linguistic e nvironment. In coding the data, I classified the tokens as palatal realizations or application of the palatalization rule (p); and dental realizations or non application of the palatalization rule (d). As it is standard procedure in variationist linguisti cs methodology, preceding and following phonological context were coded for both the underlying phonological and the surface phonetic environment. For the coding of the linguistic factor syllable stress, no distinction was made between primary, secondary o r tertiary stress. Coding of vowel nasalization was quite an intricate and lengthy task. As already discussed in the literature review, one of the main challenges was the appropriate approach to account for all instances of vowel nasalization in UP variet y regardless of preconceived models of the system of Portuguese. After many theoretical reviews and considerations, tokens were analyzed as either showing the absence or presence of

PAGE 127

127 vowel nasalization. Thus, tokens were classified as either oral realiza tions (o); or nasal realizations (n). The objective was to count the absence or presence of every instance of vowel nasalization, phonemic or not, regardless of syllable structure (Lipski 1975:67) that is, to account for the diverse aspects of the nasali zation process in UP, regardless of whether it follows the so called rules of vowel nasalization of standard Portuguese. The tokens selected for analysis were individually coded for all the linguistic and extralinguistic factors considered in the study. Coding was first completed manually and then transferred to an Excel data sheet. In addition to the pseudonym assigned during the transcription process, each participant was given a speaker number. For the purpose of the analysis that follows and consist ent with current sociolinguistic research procedure, I coded the data for linguistic and extralinguistic factors. All the tokens analyzed for the present study were coded impressionistically by the researcher. Instrumental analysis was not performed Once all the data were coded on Excel, it was then transferred to GoldVarb X in order to perform the quantitative analysis. 3.6.4 Quantitative A na lysis Nowadays, most linguistic variation can be carried out by statistical software programs, usually under the n ame of logistic regression. VARBRUL, which has been used widely in quantitative linguistics analysis, is an application program designed for performing variable rule analysis (Bayley & Preston 1996:256). One of the most extensively used versions is GoldVar b. To conduct the quantitative analysis of the present study, GoldVarb X version was used. 3.6.4.1 GoldVarbX GoldVarbX is a multivariate analysis application and a key methodological tool for conducting variationist sociolinguistics (Sankoff, Tagliamont e, & Smith 2005).

PAGE 128

128 GoldVarbX variable rule application for Windows was used to handle the quantitative analysis of the two variables under investigation for the current project. Upon completion of the extraction, transcription, coding, and transfer of the tokens into an Excel spread sheet, the data was converted into a machine readable corpus. Data was then transferred as a text file into the GoldVarb X program. The next step was to perform the first distributional analysis of the data. The output of this initial analysis provides a quick overview of how the data is distributed in each factor group. The information included contains the raw numbers and percentages of rule application involving each factor, the input probability and the weights assigned by the program to each factor (Bayley & Preston 1996:269). Once the first GoldVarbX run was obtained, it ates that there is a 0 per cent or a 100 per cent value in one of the cells in the variationist analysis, thus signaling that the data is not variable (Tagliamonte 2006:152 3). This is a normal procedure when performing quantitative variationist analysis. This fact just indicates the need to run subsequent analyses. With respect to this point, Bayley and Preston argue that to gradually refine the model of variation until is reliable (i.e. it contains no interactive factors, no knockout factors, and no sin gleton factor group s) and it is parsimonious (i.e. several factors that are theoretically similar and that seem to be exerting a similar influence are collapsed into a single factor (1996:278). Upon eliminating or recoding factor groups, and once the data is free of bad distributed cells, it is time to run the binomial analysis. This analysis reveals whether or not a particular factor or factor group contributes significantly to the model of variation under investigation.

PAGE 129

129 3.6.4.2 Binomial step up/step do wn analysis This analysis is also known as a stepwise regression procedure. Paolillo argues (2002:85). This analysis is conducted in two phases; the first one is the step up. In the step up analysis, the regression starts at level 0 and adds one factor group at a time, comparing each new model with the last using the likelihood ratio statistic Once a model is found that cannot be improved upon by adding any new parameters, the results for the analysis are reported. The second phase of the analysis is called step down. In this phase the program odel when all the factor groups are included considers the less complex models, dropping the factor groups that do not have a significant effect on the likelihood of the mod el (Paolillo 2002:85). Factor groups are excluded one at a time at each level until the best model is found, that is, when none of the factors can be excluded without affecting significantly the likelihood of the model. At this point, step down stops and the results are reported. Under normal circumstances both results from the stepping up/stepping down analysis are the same, that is, they contain the same factor groups. Several analyses of this type were performed until reaching the best model for the data. A full account of this and other data analyses performed will be included in the next chapters and in the discussion section. In this chapter I have outlined the methodological approach to phonological variation implemented for the current project. I have described relevant social aspects of the speech community of Rivera, and described in detail both the preliminary and the

PAGE 130

130 data collection fieldwork conducted for the present study. I have also described the individual and group sociolinguistic in terviews along with the interview modules designed to elicit the unmonitored speech of bilingual speakers of UP in Rivera. I have phonological variation of the two linguistic variables under investigation. A detailed account of the linguistic and extralinguistic factor groups and factors coded to analyze the two linguistic variables under examination was provided along with an overview of quantitative linguistic analysis comm only used in sociolinguistic research was discussed. T he following chapter (Chapter 4 ) presents the quantitative analysis of palatalization of /t/, /d/ in / i/. I will discuss the factor groups and factors considered in the analysis. The results of the distributional and of the binomial step up/step down (1998) work conducted on palatalization in the speech community of Rivera

PAGE 131

131 CHAPTER 4 QUANTITATIVE ANALYSE S OF PALATALIZATION 4.1 Background In this chapter, I present quantitative results for the first linguistic variable under investigation, that is, palatalization of dental stops before /i/ in Rivera. As seen above (§ 2.5.1), the actual articulation in some va rieties of BP of /t/ as [ ] and /d/ as [ ] before [i] results from the phonological process called palatalization. The conservative articulation, in which the process of palatalization is not operative, is characteristic of other Portuguese varieties namely, EP and varieties fro m northern Brazil; in these varieties /t/ and /d/ before /i/ is realized as dental or dento alveolar. Further, the palatal articulation of dental stops before [i] is variable across varieties of BP (Azevedo 2005). While in the popular Portuguese, spoken i n southern Brazil, /ti/ and /di/ are historically rendered with the conservative pronunciation, the application of the palatalization rule continues to advance. In fact, palatalization of dental stops is becoming mainstream in Rio Grande do Sul, the Brazi lian state adjacent to Rivera variant is favored by young speakers, mostly in large metropolitan cities. These findings are confirmed by a recent study conducted in Antnio Prado, a small southern city in Rio Grande do Sul. In this bilingual city, founded by Italian immigrants, it was found that the overall rate of palatalization is 29% (Battisti & Hermans 2009). Further, along the Uruguayan Brazilian border, palatalizatio n of /ti/, /di/ has been attested. Carvalho (1998) sociolinguistic research in Rivera constitutes the first study that has investigated both linguistic and extralinguistic factors that condition linguistic variation in UP.

PAGE 132

132 Carvalho (1998:223) provides quantitative apparent time evidence suggesting that palatalization of dental stops is undergoing linguistic change. In what follows, I present quantitative analyses of the palatalization of dental stops. I add a trend component to the analysis by comparin g apparent time data from two studies collected at different points in time, 1995 and 2007. By reexamining the same speech community, it is feasible to verify whether the change in progress hypothesized by Carvalho has continued to advance or whether it ha s stabilized at the speech community level. As Labov (1972) notes, the only way to test the validity of apparent time data and predictions is to make observations of the same community at least twice, at two different points in time. 4.2 Research Questi ons The dental or palatal realization of /ti/, /di/ has been a topic of interest within the domain of Portuguese linguistics in Brazil and in the frontier city of Rivera, Uruguay. Previous descriptions of UP variety mention li ttle or no palatalization ( Ro na 1965 Hensey 1972, Lipski 1994 ). To my knowledge, early explorations of the phonetic realizations of dental stops before /i/ in Rivera have applied interpretive, qualitative techniques to analyze this phenomenon. Carvalho (1998) in contrast, examines both linguistic and extralinguistic factors that condition the palatalization of dental stops in Rivera. In fact, she argues that the young, women, and middle class speakers are the groups leading diffusion in Rivera, which has caused the rise of this li nguistic change (223). In the present investigation, I followed variationist methodology by using quantitative and qualitative analysis to describe and explain the patterns followed by the speakers of the speech community of Rivera, with regard to palatali zation of dental

PAGE 133

133 stops before /i/. I investigate whether the hypothesized change in progress has continued to advance or has stabilized at the speech community level. If the hypothesis proposed by Carvalho is corroborated, then an increment in the palat alization rate would be expected. Thus, the prediction is that there will be an increase in palatalization among the young, women and mid middle class speakers. On other hand, if no increase in the palatalization rate is found, then I would argue that the language change in progress hypothesized has stabilized in Rivera. I will also identify the driving forces responsible for the variant selection and the linguistic factors conditioning palatalization of dental stops in UP. 4.3 Circumscribing the Variable Context In the data of a variationist study, the researcher must identify all the linguistic contexts in which variation of the feature in question may occur. Since there may be situations where no variation is possible, the inclusion of environments whe re the use of the linguistic variant or feature is categorical obscures the actual constraints that may apply to the data (Labov 1969:728). In order to examine the behavior of palatalization of /t/, /d/ before / i/, it is necessary to delimit its context of occurrence. In most varieties of Brazilian Portuguese, both an underlying high vowel /i/ in stressed or unstressed positions and a phonetic [i] rose from an underlying /e/ in unstressed positions may palatal ize the preceding dental stop. Tables 4 1 to 4 3 show examples (Battisti & Hermans 2009:236). Table 4 1. Stressed /i/ Stressed /i/ Glossary t io medida [me ativo [a

PAGE 134

134 Table 4 2. Unstressed /i/ Unstressed /i/ Glossary d iretor ] d if cil timo Table 4 3. Unstressed /e/ raised to [i ] Unstressed /e/ raised to [i] Glossary p onte vinte ] sede ] For the present study, the envelope of l inguistic variation (Labov 1972 ), that is to say, the variable context, included the dental or affricate realization of /t/ or /d/ before an underlying high vowel /i/ in stressed or unstressed position or a phonetic [i] raised from an underlying central vowel /e/ in unstressed position. Hence, I incorporated tokens as a dental or affricate realization of /t/ or /d/ preceding the vowel /i/, this vowel being oral, nasal or glide. Tables 4 4 to 4 6 show s examples extracted from the data. Table 4 4. Palatalization of / t/, /d/ preceding the oral vowel /i/ G los sary t ia [ a] d ia [ a] p ote o ] Table 4 G los sary t i nt a [ a] paint tinha a] st 3 rd ma ndinga [ ga] Table 4 6. Palatalization of /t/, /d/ preceding the palatal glide [j] G los sary diurno [ ] daytime dio [ ju ] hate mandioca j ka] As discussed above (§2.5.1.2), rates of palatalization of d ental stops differ significantly among speech varieties in Brazilian Portuguese (Bisol 1991). Previous

PAGE 135

135 studies suggest that variation correlates also with social factors such as age, gender, and socio economic status. Thus, in the present investigation t hree linguistic and three extralinguistic factors have been analyzed to examine palatalization of dental stops in contemporary UP in Rivera. 4.4 Exclusions I included in the analysis all words containing underlying /t/ or /d/ followed by an underlying h igh vowel /i/ in stressed or unstressed position or a phonetic [i] raised from an underlying central vowel /e/ in unstressed position. As seen above (§4.3), in most varieties of BP, pretonic mid vowels /e/ and /o/ raise to [i] and [u] respectively and may palatalize the preceding dental stop (Bisol 1989, Battisti & Hermans 2009:236). While vowel raising is a known phenomenon in the Portuguese language and in other Romance languages, such as Spanish ( Hual de 1989, Holmquist 2001; among others) and Italian (M aiden 1991), further explorations of vowel raising variation will not be included in this analysis since its inclusion goes beyond the scope of the present investigation. As it is standard in the methodology of variationist sociolinguists, I excluded fro m the analysis the following: false starts or truncated utterances, repetitions, and utterances that were not understandable or not audible. 4.5 Linguistic Constraints Deciding which factor groups and factors to test in any variationist study entails a deep understanding of the linguistic and or social influences that may condition the choice of a linguistic variable. Such factor groups are a mutually exclusive set of factors that may influence the phonetic realization of the linguistic variable. As t he literature review reveals, there are three linguistic factors most commonly found to condition the

PAGE 136

136 variable pronunciation of palatalization of dental stops: stress and following and preceding linguistic environment (Bisol 1991, Carvalho 1998). These l inguistic factor groups were included in the present study in direct response to the body of literature of palatalizat ion of /ti/, /di/ in BP and UP. 4.5.1 Syllable Stress Because of the fundamental relevance of stress to the entire Portuguese phonological system, it is worthwhile to look at stress as a potential constraint on variation. Mattoso stressed by increasing the relative strength with which the breath is released, it is necessari Furthermore, in Portuguese, stress may be placed on the last, the penultimate, or the antepenultimate syllable. Penultimate stress is dominant and antepenultimate stress can be consid ered as marginal in Portuguese (Mattoso Cmara 1972:25). Based on previous research (Bisol 1991, Carvalho 1998), it can be inferred that the southern state of Ri o Grande do Sul shows that stress is a conditioning factor in the application of the palatalization rule. Bisol (1991) sampled two groups: monolinguals (speakers from the capital city, Porto Alegre, and from the border city of Santana) and bilinguals (spea kers from Italian and Ger man settlements). Table 4 7 sho ws the distribution of palatalization of dental stops according to syllable stress among As Table 4 7 show s, Bisol fou nd out that the monolingual group from Porto Alegre, favor the relative stronger positions, in the following decreasing order: stressed (.60), pre tonic (.48), and post tonic (.42). The bilingual group favored the weakest positions;

PAGE 137

137 post tonic (.57), stre ssed (.50), and pre decreasing order of statistical results: stressed syllables a (.57) weight value, followed by pre tonic pos itions (.54), and an inhibiting value of (.41) for post tonic positions (1998:171). Table 4 7. Palatalization of dental stops according to syllable stres s among monolingual and bilingual speakers Monolingual speakers Bilingua l speakers Porto Alegre Santana Carvalho Italian Stressed .60 .64 .57 Post tonic .57 Pre tonic .48 .51 .54 Stressed .50 Post tonic .42 .35 .41 Pre tonic .43 Following previous studies, three degrees of stress were distinguished: stress ed, pre tonic, and post tonic syllable. In order to consider a syllable as stressed, any degree of primary or secondary stress was considered. It was hypothesized that syllabic stress plays a pivotal role on the conditioning of the linguistic variable. S tressed syllables were predicted to have the greates t influence. Table 4 8. Examples of palatalization of /ti/, /di/ according to syllable stress Syllable Stress Glossary Stressed syllable dica Pre tonic syllable tirar ] Post tonic syllable dio 4. 5.2 Follow ing Phonological Environment The nature of the phonological segment that follows /ti/ and /di/ has proven to be a strong linguistic c nasals (.76) tend to favor the palatalization rule, while nasals have an unfavorable effect (.50), and finally zero (.66), vowels (.40), and other consonants (.55) are not

PAGE 138

138 relevant. It is worthwhile to point out the role of sibilants, which clearly serve as an inhibiting factor (.05) in the application of the palatalization rule palatalization: nasal consonants tend to favor the palatal variant (.63), while sibilants disfavor it (.13). Carvalho reports that after subsequent analysis of the data, she found that most of the nasals in this environment were palatal nasals. 1 She argues that palatalization in this en vironment may be triggered by regressive assimilation. The (Bisol 191:113). T he presence of a following sibilant in this environment produces forms like: disculpe [dsklpi], [ di s k quente form. As results reveal, the dental variant is the most frequent realization in this environme nt. Table 4 9 shows examples of palataliation of dental stops according to following phonological environment. Table 4 9. Examples of palatalization of /ti/, /di/ according to following phonological context Following context Glossary Vowel teatro Lateral diluir ] Vibrant director ret ] Stop medida [me Nasal t mido Zero de Sibilant desconhecido 1 Carvalho (1998:174) reports that the palatal nasal is high in frequency in her corpus due to the constant presence of the word tinha Tinha st and 3 rd person singular of the past imperfect of the verb ter dinheiro diminutive suffix nho ( zapatinho festinha gatinho etc.). However, it is tokens extracted for certain frequent lexical items.

PAGE 139

139 It was hypothesized that the following phonological environment would be a strong linguistic constraint that conditions the linguistic variable under study. It was also predicted that the effect of the following environment would be stronger than that of the preceding phonologic al environment. 4.5.3 Preced ing Phonological Environment The preceding phonological context has been shown to be a relatively weak linguistic constraint in previous studies of dental stop palatalization (Carvalho 1998). Carvalho reported that the preced ing phonological context was eliminated in the final statistical run (173). Bisol (1991) reported the inhibiting role of sibilants with a .28 factor weight. Other factors show no relevancy either because of statistic irregularity or lack of data (109). In order to test the role of the preceding phonological context in conditioning the palatalization rule and specifically the role of the sibilant, I coded the preceding environment in the present study. It was predicted that the sibilant would an importan t role in conditioning the linguistic variable under investigation. Table 4 10. Examples of palatalization of /ti/, /di/ according to preceding phonological context. Preceding context Glossary Vowel timo Liquid perdido [ pe du ] Nasal vinte Zero te nd pers. Sibilant festive As mentioned above, three linguistics factors were coded to account for the variable realization of palatalization of /ti/, /di/ in Rivera. These linguistic factors are: tonicity of the syllable and following and preceding phonological environment. In what follows, I describe the extralinguistic or social factors included in the present study.

PAGE 140

140 4.6 S ocial Factors Previous studies have shown the importance of social or extralinguistic factors when analyzing language variation and sound change. Social factors are essential in modern sociolinguistic research since they allow researchers to predict the c hoice of a linguistic variable, and to determine if the variable under consideration is undergoing change or whether there are other factors at play. The many aspects of the interplay of social factors and language have been widely discussed. A more criti cal issue to be extralinguistic factors most commonly found to constrain the variable pronu nciation of dental stops are the following: age, socioeconomic status, and gender. Thus, these social factors were included in the present investigation. 4.6.1 Age It is of special importance to incorporate age in a variationist analysis since it can help the researcher determine if the speakers of different generations demonstrate similar or different patterns of use of the linguistic variable under investigation. If so, these different rates may point toward age grading or a change in progress (Bailey 199 1:241). time evidence suggesting that palatalization of dental stops in Rivera is undergoing linguistic change. The basic principle underlying the apparent tions of adults mirror actual diachronic developments in a language when other factors, such as social class, are held constant. The speech of each generation is assumed to reflect the language as it existed at the time when that generation learned the lan

PAGE 141

141 (Bailey1991:242). The apparent time method assumes that individual vernaculars remain basically stable after the formative period of language acquisition A s apparent tries to reach from the present to the future, real diachrony entails a link fr o m the p resent to the past (Bailey et al.1991 ). Through the comparison of two sets of apparent time data, I will be able to test This task is performed by comparing speech own sample; grouped in comparable age cohorts, that is, Generation 1, formed by younger speakers aged between 15 and 29 years old; Generation 2, formed by young adults aged between 30 49 years old; and Generation 3 older speakers formed by speakers aged between 50 70 years old. It should be noted that the 1995 sample actual interviews referred to as such. Table 4 11 outlines the age limits of the younger, young adult, and older generations for each of the two samples under study. It s hould be noted that each age group is formed by an equal number of male and female speakers, that is, nine for each gender, representing the three socioeconomic groups established for the study (§3.4.2). The two samples shown above in Table 4 11 are compa rable in size, age distribution among age groups, socioeconomic class, and gender. Thus, the sample is adequate for comparability purposes in real time sociolinguistic research (Bailey et al. 1991).

PAGE 142

142 Table 4 11. Age distribution in each of three age gro ups for each speech sample 1995 sample 2007 sample Younger speakers (Generation 1) N=18 16 29 years old (born ~ 1979 1966) 15 29 years old (born ~ 1992 1978) Young adult speakers (Generation 2) N=18 30 49 y ears old (born ~ 1965 1946) 30 49 years old (born ~ 1977 1958) Older speakers (Generation 3) N=18 50 70 years old (born ~ 1945 1925) 50 70 years old (born ~ 1957 1937) Total N= 54 Total N= 54 Note: The 1995 sample correspon Based on previous research (Bisol 1991, Carvalho 1998, Battisti & Hermans 2009), I can infer that age is the strongest extralinguistic constraint conditioning palatalization of dental stops and that younger speakers tend to prefer t he innovative palatalized variant. Thus, the hypothesis is that younger speakers will show an increase in the use of the palatal variant. 4.6.2 Socioeconomic Status When sociolinguists first approached the study of language variation and change, speaker change. It was Labov (1966) who advanced our understanding of the interrelation between social class and level of consciousness. That is, when a sound change is above the leve l of awareness, speakers located in the upper levels of society tend to favor the innovative variant more often, whereas when a sound change is below the level of awareness, speakers situated in the lower classes prefer to use the variant more frequently ( Labov 1972).

PAGE 143

143 The speech community of Rivera is a highly stratified society (§3.3.2). Previous research shows that wealthy and middle class inhabitants who identify with Montevideo cul ture, favor palatalization of dental stops, whereas poor and working cl ass inhabitants, who identify with local border culture, disfavor it (Carvalho 1998). Therefore, the prediction is that middle class speakers would use the palatalized variant more often than working and low middle class speakers. 4.6.3 Gender sex has become one of the most important social factors in the quantitative study of phonological variation. This notion has been advanced so as to encompass the social construction of sex, that i s, gender. For Eckert (1997 ), sex is a biological categor y that constitutes the foundation for the differentiation of roles, norms, and expectations in all societies. These roles, norms, and expectations form the basis of gender. r eflection of the effects on linguistic behavior of gender the complex social construction of sex and it is in this construction that one must seek explanations for such so ciolinguistics is a critical issue for researchers studying language change in progress. In fact, gender differences are extremely complex, principally in modern societies where women are adopting new roles and moving towards the marketplace. Gender role s and ideologies allow men and women to create different ways to experience life, culture, and society (Eckert 1989). Carvalho (1998) found that gender differences between working and mid middle class speakers show opposite tendencies

PAGE 144

144 women and men in the working class focus their dialect around the local variety, while in the mid middle class they diffuse towards UBP. On the other hand, the lower middle class is where men and women differ the most, women being the group that assimilates the mid middle linguistic behavior, while men maintain a more conservative dialect, similar to that of the working class (183). For Carvalho (1998), women appear to be leading the process of linguistic change in Rivera. If the palatalization of dental stops is in deed undergoing linguistic change in Rivera, then it is expected that women will use the palatalized variant more frequently than men. As Eckert (1989) proposes, the interpretation of gender roles and ideologies will be analyzed within the social context of the speech community in Rivera in order to better understand the sociolinguistic processes that differentiate linguistic behavior between women and men. As mentioned above, three extralinguistic or social factors were examined to account for the variabl e realization of /ti/, /di/ in Rivera. The most commonly analyzed social factors in quantitative sociolinguistic research are gender, socioeconomic status, and age. Thus these social factors were coded to examine the social stratification of the palatali zation of /ti/, /di/ in the speech community of Rivera. Both linguistic and extralinguistic factors are instrumental in the interpretation of sociolinguistic phenomena. In what follows, I present and discuss results of multivariate analyses for both inter nal and social factors. 4. 7 Results and Discussion The data set amounted to 2526 tokens, which I analyzed using GoldVarbX (Sankoff, Tagliamonte, & Smith 2005). The following section presents an overall

PAGE 145

145 distribution of the variants, a factor by factor di stributional analysis, and multivariate analysis of the contribution of factors in the palatalization of /ti/, /di/ in UP. 4.7.1 Overall Distribution of Variants An overall distribution of variants refers to the relative frequency of each variant of the linguistic variable without considering any linguistic or extralinguistic factor (Tagliamonte 2006:135). Table 4 12 shows the overall distribution of the data grouped according to whether there was a dental or a palatalized variant. In addition, Table 4 12 shows that the overall rate of palatalization of /ti/, /di/ is 29% (N=743) and that dental realization is 71% (N=1783) for a total of 2526 tokens analyzed. Table 4 12. Overall distribution of the realization of tokens of /ti/, /di/ in UP (2007 data) P alatal realization Dental realization % N % N 29 743 71 1783 Total N 2526 As Table 4 12 shows, 29% of dental stops are palatal realizations whereas dental variants make up 71% of the data. This distribution corroborates empirical reports in other extent that of UP research by Carvalho (1998). It is noteworthy to point out that the same frequency rate of rule application (29%) is found in the bilingual southern cit y of Antnio Prado in Rio Grande do Sul, Brazil. I will return to this point in the discussion section. Table 4 13 replicates, as far as I can tell, the overall distribution of variants presented in Carvalho (1998:171). Table 4 13. Overall distribution o f the realization of tokens of /ti/, /di/ in UP (1995 data) Palatal realization Dental realization % N % N 32 7 19 68 1 529 Total N 2 248

PAGE 146

146 As Table 4 13 shows, the overall distribution of linguistic variants identified in Carvalho (1998) are palatal realizations, 32% (N=719), and dental realization or non these overall frequencies compare between the two studies. Figure 4 1 compares the results of the overall distrib ution of the realizations of Figure 4 1. Overall distribution of the realizations of tokens of palatalization of /ti/, /di/ in UP. RMC= Rosa M ar a Figure 4 1 show s that the overall frequency rate of rule application in the 2007 study is slightly lower than the one reported in 1995. The comparison of overall frequencies found in the 1995 and 2007 data (32% and 29% respe ctively) points to a relative stability of linguistic variation at the speech community level. Nevertheless, the difference in overall rate is not in itself indicative of a difference in the grammar of 32% 68% 29% 71% 0% 10% 20% 30% 40% 50% 60% 70% 80% Palatalizations Non-palatalizations Percentage AMC (1995) RMC (2007)

PAGE 147

147 palatalization of dental stops (Tagliamonte 2006). Th e next step is to examine the independent factors conditioning the linguistic variable under investigation. 4.7.2. Factor by Factor Distributional Analysis A factor by the use of the (de by factor analysis considers each independent variable one at the time; it also provides factor weights, that is to say, the values assigned by the variable rule program indicating the probability of rule application (Tagliamonte 2006). Table 4 14 show s the distribution of palatalization of dental stops according to syllable stress in UP. While palatalization in stressed and pre tonic syllables reveals comparable rates of rule application, 30% (N=899) an d (N=342) respectively, the palatalized variant can be seen to occur slightly less frequently in post tonic syllables 28% (N=1285). Table 4 14 Distrib ution of palatalization of /ti/, /di/ by syllable tonicity in UP (2007 data) Syllable tonicity % N St ressed 30 899 Pre tonic 30 342 Post tonic 28 1285 Table 4 15 reports the effect of following phonological environment on palatalization of /ti/, /di/. Sibilants have a relatively low rate of palatalized variant, only 18% (N=135). The highest rate of palatalization is with following vowels 42% (N=174), while the effect of following liquids (laterals and vibrants were collapsed in a final statistical run) is 37% (N=67). The effect of stops is 36% (N=501), and of nasals 34% (N=191). All three s how comparable palatalization proportions. Apparently, UP is a variety where following liquids pattern with following stops and nasals.

PAGE 148

148 On the other hand, a following pause or zero phonological environment stands out amongst the factors analyzed with an intermediate rate of palatalization, 26% (N=1458). Table 4 15. Distribution of palatalization of /ti/, /di/ by a following phonological segment in UP (2007 data) Following phonological segmen t % N Vowel 42 174 Liquids (laterals and vibrants) 37 67 Stop 36 501 Nasal 34 191 Zero 26 1458 Sibilant 18 135 Table 4 16 outlines the effect of the preceding phonological environment on /ti/, /di/ palatalization. As seen in following phonological environment, sibilants show a re latively low rate of palatalized variants, 22% (N=147). This rate is comparable to the proportion of palatalized variants for preceding liquids (laterals and vibrants collapsed in a single factor): 22% (N=254). On the other hand, palatalized variants can be seen to occur more frequently with preceding vowels, 31% (N=474), nasals 31% (N= 872), and pause or zero phonological environment 30% (N=779). These last three factors show comparable rates of palatalization. Table 4 16. Distrib ution of palatalizati on of /ti/, /di/ by preceding phonological segment in UP (2007 data) P receding phonological segment % N Vowel 31 474 Nasal 31 872 Zero 30 779 Liquids (laterals and vibrants) 22 254 Sibilant 22 147 In sum, the tabulations of effects just present ed suggest that the three internal factors included in the present study condition the occurrence of /ti/, /di/ palatalization in UP. Although a factor by factor analysis offers many insights into the factors conditioning the linguistic variable under inv estigation, it does not provide the combined

PAGE 149

149 impact of all the factors together. In the next section, I analyze the data using the variable rule application. This analysis reveals whether a par ticular factor or factor group contributes statistically signi ficant effects to variant choice when all are considered simultaneously, as well as the relative magnitude of effects vis v is each other ( Tagliamonte 2006 ). It must be noted that factor weights can be values anywhere from 0 to 1. While a factor weight favoring disfavoring showing values around .50 are interpreted as having a neutral effect on the application v alue. However, as Tagliamonte (2006) further points out, what represents an accurate criterion for interpreting results is the relative position of factor weights in relation to each other. 4.7.3 Multivariate Analysis of the Contribution of Linguistic Fa ctors of Palatalization of /ti/, /di/. Table 4 17 show s the results of the multivariate analysis of the contribution of linguistic factors of palatalization of dental stops. The factors are selected by the stepwise multiple regression procedure incorporate d in the variable rule program as significant to the probability of palatalization of /ti/, /di/ in UP. Table 4 17 reports the binomial step up and step down analysis of the data. The linguistic factor groups selected as presenting statistically significa nt effects on the palatalization of /ti/, /di/ in UP are the following and preceding phonological segment and tonicity of the syllable. Table 4 17 also shows that following and preceding phonological context s distinguish segments based on manner of articul ation

PAGE 150

150 Table 4 17. Variable rule analyses of the contribution of linguistic factors selected as significant to the probability of palatalization of /ti/, /di/ in UP (2007 data) Corrected mean .29 Log likelihood 1494. 993 Total N 2526 Following phonological segment Vowel /glide Factor weight .66 % 42 N 174 Liquids (vibrant + l aterals) .65 37 67 Stop .62 36 501 Nasal .59 34 191 Zero /Pause .43 26 1458 Sibilant .38 19 135 Range 28 Preceding phonological segment Vowel /glide .53 31 474 Nasal .53 31 872 Zero /pause .51 30 779 Sibilant .39 22 147 Liquids (vibrant + laterals) .3 9 22 254 Range 14 Tonicity Post tonic position .54 29 1285 Stressed position .46 30 899 Pre tonic position .44 30 342 Range 10 Note: All factor groups selected as statistically significant. Table 4 18 replicates, as far as I can tell, the variable rule analysis of application of the palatalization according to linguistic factors presented in Carvalho (1998:171). Table 4 18 indicates that the following phonological context and tonicity of the syllable contribute a statistically significant effects on palatalization of /ti/, /di/ in UP. Preceding phonological environment and juncture factors were eliminated in the final s tatistical run (Carvalho 1998:171). As Tables 4.17 and 4.18 indicate, the following

PAGE 151

151 phonological context is the strongest factor group conditioning palatalization of dental stops in Rivera. While tonicity of the syllable is also selected as statistically significant, its position in the ranking differs in both sets of data. The next section provides a detail description and comparison of every factor group selected as significant in both studies. Table 4 18. Variable rule analyses of the contribution of linguistic factors selected as significant to the probability of palatalization of /ti/, /di/ in UP (1995 data) Total N 2248 Following phonological segment Nasal Factor weight 63 % 39 N 315 Other consonants 53 33 440 Zero/pause 52 33 918 Vowel 50 31 341 Lateral .48 36 11 Vibrant 47 28 67 Sibilant 13 9 148 Range 50 Tonicity Str essed 57 33 889 Pre tonic 54 30 461 Post tonic 41 32 898 Range 16 Preceding phonological segment Lateral [.60] 38 24 Nasal [.56] 38 581 Vowel [.55] 30 540 Vibrant [.49] 27 151 Sibilant [.44] 27 85 Zero/pause [.43] 3 1 867 Juncture Initial [.54] 29 840 Medial [.54] 35 502 Final [.44] 31 866 Clitic [.44] 57 40 Note: Factor groups not selected as significant appear in square brackets. (adapted from Carvalho 1998).

PAGE 152

152 4.7.3.1 Following phonological cont ext First, the greatest effect is presented by the following phonological context, with a relative magnitude of its effect at 28. This finding is consistent with previous studies of UP and across dialects of BP. Carvalho (1998) and Bisol (1991), found tha t the following environment applies the strongest effect with a 50 and 53 of relative magnitude respectively. Vowels and liquids (laterals were collapsed with vibrants in a final run), exert the highest effect on favoring the palatalization rule with a 66, and .65 probability, respectively. This finding provides evidence to suggest that palatalization of /ti/, /di/ is more likely to occur when followed by another vowel or glide, forming a diphthong, and a lateral or a vibrant. In BP, laterals in syllabl e final position are usually realized as a velar glide [w], which forms a diphthong with the preceding vowel (Azevedo 2005:49). Hence, it appears that diphthongs create a favorable linguistic constraint promoting palatalization of dental stops in this env ironment. On the other hand, stops at (.62), and nasals at (.59) favor the palatalization rule, whereas a following pause or zero phonological context (.43) and sibilants (.38) disfavor it. Table 4 18 shows that in the 1995 data the following context se parates nasals (including the palatal nasal) from other consonants (including stops) and distinguishes laterals from vibrants, with pause, vowels, and sibilants treated separately. The two types of following environments contributing to exert the most not able effect on palatalization of /ti/, /di/ are nasals favoring palatalization of dental stops (.63) and sibilants disfavoring it (.13). One of the most significant aspects resides in the behavior of a following sibilant. Carvalho (1998) reports a low .1 3 probability, which indicates an inhibitor effect on the application of the palatalization rule.

PAGE 153

153 The next question to ask is how consistent the findings are in both data sets. With that question in mind, I compare the linguistic factors selected as sig nificant in both studies. I will start with following phonological context selected as significant in both samples Figure 4 2 compares both data sets. Figure 4 2. Probability of palatalization of /ti/, /di/ according to following phonological environmen t in UP. As Figure 4 2 show s, while in the 1995 data, vowels exert a rather neutral effect on the application of the palatalization rule (.50), in the current study they exert the strongest effect (.66). However, in another variety of Portuguese, namely Po rtuguese Gaucho, show (.55) for the bilingual and a disfavoring (.40) for the metropolitan speakers. 0.5 0.47 0.63 0.13 0.52 0.53 0.66 0.65 0.59 0.38 0.43 0.62 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Vowel Liquids Nasal consonant Sibilant Pause/zero Others GoldVarb weights for palatalization AMC (1995) RMC (2007)

PAGE 154

154 Laterals and vibrants, collapsed into a single factor group in the present study as liquids, proved to be a strong linguistic factor favoring palatalization (.65), while Carvalho (1998) found that laterals (.48) and vibrants (.47) disfavor it. However, the shows that laterals exert the most significant effect on the application of the palatalization rule (.73) for the bilingual and (.79) for the metropolitan group. Although irregul ar behavior due to their low number of occurrence in the data (1991:110). analysis (.63), while in the 2007 study nasals show a slightly lower effect, at (.59). In the present analysis, palatal nasals were collapsed with the bilabial and alveolar nasals in a final run to better account for the data. Although collapsed into a single factor group, nasal consonants exert a slightly favoring effect of the application of the palatali zation rule. Carvalho argues that the effect of a following nasal favoring palatalization could be of regressive assimilation, triggered anticipation of the point of articulation of [ ] palatal nasals (.60 and .76) for the bilingual and metropolitan group respectively, exert a strong effect on rule application. It is worth noting that the difference in probabi lities between the As Figure 4 2 show s, and consistent with previous research, a following /s/ promotes dental realizations of /ti/, /di/ in UP. The 1995 and 2007 samples e xhibit a

PAGE 155

155 very low probability for the sibilant (.13 and .43) respectively. These results are also comparable cross dialectally. for the bilingual and metropolitan groups, respectively, signaling the tendency of a sibilant to hinder the application of the palatalization rule (111). On the other hand, in the 2007 data, following pause strongly disfavors palatalization of dental stops (.43), while in the 1995 sample, following pause exerts a rather neutral effect of rule application (.52). The present study shows that stops favor palatalization (.62), while in the 1995 data stops were collapsed with other consonants, providing a combined factor weight of .53. This fact obviously obscures the indivi dual potential role of stops constraining the linguistic variable under investigation in this environment, hence impeding statistical comparisons as well. The position of vowels at the highest position in the hierarchy requires some comment. As mention ed above, the present study examines palatalization of dental stops before /i/, where unstressed mid vowel /e/ raised to a high vowel [i] and the palatal glide /j/ are taken into account. The use of high vowels in place of mid vowels results from a phenom enon known as vowel raising (Mattoso Cmara 1953, Bisol 1989) 2 Vowel raising is pervasive, and its use is variable in colloquial Portuguese language, and seemingly at play in UP as well, as data results suggest. However, the variable use of high vowels i n place of mid vowels will not be examined in the present study. 3 2 In Brazilian Portuguese there is a vowel raising rule which raises the pretonic /e/ and /o/ vowels to /i/ and /u/ respectively. This rule is conditioned by linguistic and social factors and thus treated as a variable rule (Bisol 1989). 3 The role of the high vowel feeding the palatal ization rule in Portuguese has been attested (Bisol 1991, Battisti & Hermans 2009). Battisti and Hermans found in the community of Antonio Prado that the amount of unstressed mid vowels raised and palatalized were only 13%, most of them in final post toni c

PAGE 156

156 The point of articulation of the palatal nasal provides a favorable environment for palatalization. The extensive use of the suffix nho, nha in UP must be taken into account. The freque nt use of diminutives in BP has already been reported ( Giangola 2001 ). Th e author points out that both young and adult Brazilians are given to the use of diminutives. The use of diminutives appears to be more frequent in conversational speech, expressing many subtle shades in meaning. Diminutives are formed by the indicate a greater or lesser degree, dimension, or intensity of a derivative noun or adjective with r As Mattoso Cmara further argues, the principal diminutive suffix in Portuguese is inh (o, a), which is considered to be unrestricted in use (e.g. casinha little house, gatinho little cat). Carvalho (1998:174) also reports a high frequency of words with the diminutive suffix nho in her data. It can be argued that UP speakers have incorporated the productive diminutive suffix in their speech. However, the semantic and pragmatic funct ions of diminutives in UP remained to be examined. As shown by previous studies, sibilants have an inhibitory effect on palatalization of /ti/, /di/. The 2007 results show the unfavorable role of the sibilant, restricting the palatal realization of d ental stops in UP. Carvalho (1998) and Bisol (1991) report similar probability weight as low as .13 (1998:174). Bisol reports (.20 and .05) for the bilingual positions. Battisti and Hermans argue that this linguistic environment is disfavorable to palatalization (.23), due to low rates of mid vowel raising and the influence of Italian dialects spoken in the city (2009:238).

PAGE 157

157 and metropolitan groups respectively, signaling the tendency of a sibilant to hinder the application of the palatalization rule (1991:111). The behavior of a following sibilant, at the bottom of the hierarchy in both varieties, suggests that across dialects t he underlying grammar producing the surface form is basically the same, at least for this linguistic factor. As Tagliamonte (2006:241) argues, if two varieties share the same constraint rankings, it is an indication of the similarity of their grammars. A s expected, 2007 data confirm that the following context conditions palatalization of dental stops in UP. Furthermore, the following phonological segment has been shown to be a robust linguistic constraint not only in UP but cross dialectally as well. The following environment also confirms that /s/ promotes dental realizations of /ti/, /di/ in UP and across different dialects of BP. On the other hand, nasals favor palatalization of dental stops in this linguistic environment. The present analysis highlight s the fact that palatalization of dental stops is more likely to occur when followed by a vowel, glide or lateral (realized as a velar glide [w] in coda position). 4.7.3.2 Preceding phonological context Consistent with previous studies, it was found that the effect of the preceding phonological environment was not nearly as strong as that of the following segment. This factor group exerts the second highest magnitude of effect in the analysis, obtaining a range of 14. While in the present study and in Bi of a preceding segment proved to be statistically significant, in the 1995 sample, this factor group was eliminated in the final statistical run. As in the following phonological context, vowels exert the most significant effe ct on applying the palatalization rule in this environment, with a probability of (.53). It must be noted that this is a rather neutral effect. This result is consistent across dialects as well.

PAGE 158

158 or weight (.53) for the bilingual participants. Preceding nasal consonants apply also a rather neutral effect of rule for the bilingual group. In the present i nvestigation, it was found that a preceding pause exerts a neutral effect of rule application (.51). This finding is consistent cross dialectally. In fact, that pause lacks statistical significance in this environment. On the oth er hand, liquids (laterals and vibrants) disfavor palataliza tion of dental stops (at .39). In line with previous research, the disfavoring role of the sibilant shows a similar pattern in UP and in BP. Bisol reports a low probability (.24) for bilingual s peakers, while in the present analysis, sibilants show a disfavoring role (.39). Bisol (1991) proposes a phonetically based explanation to account for the blocking force of the sibilant in the application of the palatalization rule There is a physically de finable connection between the stop [t,d] and the fricative [s,z] which motivates the retention of the alveolar. It is the neutral behavior of the full alveolars [s,z] which opposes the raising and fronting of the body of the tongue necessary for the arti culation of the palatal, thus eliminating the physical effort involved in the articulation of the affricates [ ] (1991:113). Bisol furthermore argues that this mutual attraction, caused by the features that the stops [t,d] and the fricatives [s,z] have i n common [ high] in distinctive feature theory represents a mere process of minimization of articulatory effect that tends to prevent the assimilation of the high vowel that causes palatalization (1991:113). In sum, multivariate analysis confirms that th e preceding phonological environment constrains palatalization of dental stops, but does not exert nearly the effect that the following segment does. The present study shows that vowels and nasals slightly favor

PAGE 159

159 palatalization of dental stops, while pause have a neutral effect of rule application. Preceding liquids and sibilants disfavor palatalization of /ti/, /di/. The role of the sibilant disfavoring palatalization in this linguistic context patterns across dialects. 4.7.3.3 Tonicity of the syllable T he last linguistic group to be selected as significant in the current investigation is the syllable was the second most significant linguistic factor group. As in previous studies (Carvalho 1998, Bisol 1991), the lesser effect is contributed by tonicity of the syllable. In the present study the range (10) (14) analysis of the bilingual participants. Figure 4.3 compares both data sets. Figure 4 3. Probability of palatalization of /ti/, /di/ according to syllable stress in UP. Multivariate analysis of the 2007 sample shows that post tonic syllable stress favors palatalization (.54) while stressed (.46) and pre tonic syl lable stress (.44) disfavor 0.41 0.57 0.54 0.54 0.46 0.44 0 0.1 0.2 0.3 0.4 0.5 0.6 Post-tonic syllable Stressed syllable Pre-tonic syllable GoldVarb weights for palatalization AMC (1995) RMC (2007)

PAGE 160

160 application of the palatal rule. However, in the 1995 sample, results show that palatalization is more frequent in stressed syllables (.57) followed by pre tonic (.54) and post tonic stressed syllables (.41). Figure 4 3 shows the probability of dental stops according to syllable stress across studies. The position of the post tonic syllable at a higher point in the hierarchy in the 2007 study, however, requires some comment. Interestingly, the constraint hierarchy found in th e present investigation that is, the post tonic, stressed and pre tonic syllable mirrors Bisol reports that palatalization frequencies show an inverse behavior in the diale cts of the settlement area, in which the more frequently occurring environment for rule application is the unstressed syllable (Table 4 19). Table 4 19. Comparison of palatalization of dental stops according to syllable stress among bilingual group (Biso l 1991) and Rivera speakers (2007 data ) Bilingual speakers Italian settlement Rivera speakers (2007) Post tonic syllable .57 .54 Stressed syllable .50 .46 Pre tonic syllable .43 .44 As Table 4 19 show s, the Italian group exhibits the same syllable stress pattern as the 2007 data that is, post tonic, stressed, and pre tonic syllables. Metropolitan and onger position in the decreasing order: stressed, pre tonic and post tonic syllable stress (Table 4 7). This Bisol (1991) notes that this contradictory pattern is based on the principle of

PAGE 161

161 affricate realization of dental stops is a new rule, which is not existent in the Italian or German language, palatalization is fulfilled in weak syllables first, where it is less salient, and gradually spreads to other contexts (118). The favored position of post tonic syllable stress at the highest point in the constraint hierarc hy and its parallel language contact. Affricate realization of dental stops is not operative in the Spanish language, and thus it can be argued that it is an innovation for UP speakers analyzing them through unstressed syllables. Summing up, tonicity of the syllable has shown to be the weakest factor group conditioning palatalization of /ti/, /di/ in UP. Multivariate results show that the more frequent environment for rul e application is the unstressed syllable. This is an unexpected result. It may be recalled that, based on the literature review, it was hypothesized that stressed syllables would exert the highest influence on rule application. However, it appears that c ontact with the Spanish language, where the affricate realization of dental stops is not operative, is a contributing factor favoring a (1991) study, where bilingual p articipants favor the weakest positions (1991:108). The analysis of this finding may be interpreted as a consequence of UP and Spanish language contact in the speech community of Rivera. In sum, as far as the linguistic constraints are concerned, the 2 007 data are generally consistent with previous research both in UP and across varieties of BP. As multivariate analysis results indicate, following and preceding phonological environment

PAGE 162

162 and tonicity of the syllable play a significant role in conditioning the variable realization of palatalization of /ti/, /di/ in contemporary UP. Following phonological environment proved to exert the highest effect of rule application not only in UP but across dialects of BP as well. The most outstanding finding is the r ole of the sibilant in preventing the application of the palatal rule in both the following and preceding segments in UP and other varieties of BP. As Tagliamonte comparison. environmental constraints are critical in language change but the constraint hierarchy as well. Hence, there seems to be a language specific constraint preventing clusters where the fol lowing/preceding alveolar sibilant blocks the affricate pronunciation of dental stops before /i/. The position of /s/ at the bottom of the constraint hierarchy suggests that the underlying grammar producing the dental variant is basically the same in both UP and BP. As expected, the preceding phonological constraint does not have nearly the effect that the following segment does. This finding patterns cross dialectally but does not correlate with the 1995 sample. The present study shows that vowels and n asals slightly favor palatalization of dental stops, while liquids and sibilants disfavor it. The role of the sibilant disfavoring palatalization in this linguistic context patterns across dialects. Tonicity of the syllable exerts the lesser effect constr aining palatal stops in UP. The constraint hierarchy shows a contradictory order. It was expected that stressed syllables would favor rule application the most; however, the only favorable environment

PAGE 163

163 for rule application is post tonic syllables. The une xpected results indicate a potential influence of the Spanish language, where palatalization of dental stops is not operative. In the next section, data will be analyzed using the variable rule application in order to examine the effect of extralinguisti c factors conditioning palatalization of dental stops in UP. As already discussed, this analysis reveals whether a particular factor or factor group contributes statistically significant effects to variant choice when all the factors are taken into accoun t at the same time. The analysis also reveals the strength or relative magnitude of the extralinguistic factors when calculated against each other (Tagliamonte 2006). 4.7.4 Multivariate Analysis of the Contribution of Extralinguistic Factors of Palataliza tion of /ti/, /di/. Table 4 20 show s the results of the multivariate analysis of the contribution of extralinguistic factors selected as significant to the probability of palatalization of /ti/, /di/ in contemporary UP. The factors are selected by the ste pwise multiple regression procedure incorporated in the variable rule program as significant to the probability of palatalization of /ti/, /di/ in UP. As shown in Table 4 20, all the extralinguistic factors are selected as presenting statistically signifi cant effects on palatalization of dental stops before /i/. Table 4 20 also reports that the greatest and most significant effect is exerted by the factor age group with a range of .49. This is a notable finding in the study since age is the primary social correlate of language change (Chambers 2002:349). The cross sectional analysis among the two samples will shed light on the process of generational preference of the palatal variant in Rivera. Thus, language change and age grading will be analyzed as pos sible linguistic outcomes in the discussion section.

PAGE 164

164 Table 4 20. Variable rule analyses of the contribution of social factors selected as significant to the probability of palatalization of /ti/, /di/ in UP (2007 data) Corrected mean 24 Log likelihood 1206.903 Total N 2526 Age 15 29 Factor weight .70 % 45 N 829 30 49 .62 35 874 50 70 .21 7 823 Range 49 Socio economic group Mid Middle class 76 53 728 Low Middle class .39 25 7415 Working class .39 16 1083 Range 37 Gender Female .62 35 1303 Male 38 23 1223 Range 24 Note: All extralinguistic factor groups selected as significant. The last two factor groups selected as significant are s ocio economic status and gender with a relative strength of 37 and 24 respectively In orde r to find out how these results compare with the 1995 data, I now turn to present the multivariate analysis of Table 4 21 replicates, as far as I can tell, the variable rule analysis of the application of pala talization of /ti/, /di/, according to extralinguistic or social factors presented in Carvalho (1998:176) As seen below, all social factors are selected as sta tistically significant. A ge, socioeconomic status and gender contribute statistically significa nt effects to palatalization of /ti/, /di/ in contemporary UP. These findings correlate with the 2007 data. As shown in Table 4 21, the most important social factor conditioning palatalization in UP is age with a relative strength of 86. The other social

PAGE 165

165 factors selected as significant are socio economic status and gender with a strength of 56 and 31, respectively. Table 4 21. Variable rule analyses of the contribution of social factors selected as significant to the probability of palatalization of /ti/, /di/ in UP (1995 data) Total N 2248 Age (16 29) Factor weight .91 % 61 N 1017 (30 49) .29 13 637 (50 70) .05 2 594 Range 86 Socio economic group Mid Middle .81 52 764 Low Middle .34 24 693 Working .30 19 791 Range 51 Gender Female .60 35 1166 Male .39 29 1082 Range 21 Note: All extralinguistic factor group s selected as statistically significant. (adapted from Carvalho 1998). As Table s 4 20 and 4 21 show there is an identical pattern between the two data sets. Not only are all the factor groups selected as significant but the constraint ranking in each f actor group patterns similarly. As expected, a ge is clearly the dominant variable, applying the strongest effect conditioning palatalization of /ti/, /di/ In what follows, I discuss each individual extralinguistic factor comparing both samples. 4.7.4 .1 Age First, the greatest effect is contributed by the factor group age, with a relative magnitude of 49. This is the strongest social constraint conditioning palatalization of stops in contemporary UP in Rivera. As expected and consistent with previous studies, the youngest speakers (15 29 years old) tend to prefer the palatal variant at (.70).

PAGE 166

166 This finding mirrors previous studies of UP and different varieties of BP. In fact, 1995 data show that the youngest participants (15 29 years old), strongly fav or the study, the youngest group, formed by speakers aged 25 35 years old, used the innovative variant more often, presenting a factor weight of .63. In the Antni o Prado study (§4.1), Battisti and Hermans (2009) report that the palatal variant is favored by Figure 4 4 show s the distribution across a ges in the 1995 and 2007 data. Clearly, younger speakers strongly fa vor the innovative variant of /ti/, /di/ in contemporary UP, while older speakers disfavor it. These findings are evidence that palatalization of dental conservative variant has a proportionately higher ratio of occurrences among older speakers, and when the innovative variant finds higher ratio of preference among younger speakers, age stratification of this sort may be interpreted as an index of change, ( i.e. ), parent differences found in the present study point also to the possibility of age grading, thus; for the time being, I will consider both interpretations. As Figure 4 4 shows, there is a repeated age gradient distribution among the groups. While in the 2007 sample young and young adults favor the innovative variant, older speakers disfavor it. The 1995 sample shows that while young people prefer the palatal variant, young adults and older speakers disfavor it. However, a cross sectional analysis shows that the decreasing frequencies among the 1995 and 2007 Generation 1

PAGE 167

1 67 group points out that the latter generation has adopted the innovative variant but at a slower pace than the former generation did (Table 4 11) Figure 4 4. Comparison of rates of palatalization of dental stops followed by /i/ according to age in UP. On the other hand, the present study shows that speakers in the 2007 Generation 2 group (30 49 years old) tend to favor the innovative varian t at (.62), while the 1995 data show s a different scenario: speakers in this age bracket tend to favor the dental use of the palatal variant over time by this genera tion. That is, speakers in the 2007 Generation 2 group (30 49 years old) were 18 and 37 years old, respectively, twelve years earlier, which is the time depth of the study. Carvalho reports an almost categorical (.91) probability of use of the palatal var iant among speakers in the 1995 Generation 1 group and (.29) for speakers in the Generation 2 group. The generational differences across studies indicates that the 1995 Generation 1 group (15 29 years old) started out at (.91) probability of rule applicat ion. These speakers have continued 0.05 0.29 0.91 0.21 0.62 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 50-70 year old (3rd. Gen.) 30-49 year old (2nd. Gen.) 15-29 year old (1st. Gen.) GoldVarb weights for palatalization AMC (1995) RMC (2007)

PAGE 168

168 favoring the palatal variant but at a much slower pace as indicated by the decrease (.62) in the 2007 Generation 2 group. In direction that is similar to the findings of previous research; 2007 data show that Generatio n 3 speakers (50 70 years old) prefer the conservative variant at (.21). Older participants in the 1995 sample show a mere (.05) of preference for the palatal variant. Cross dialectally, the same pattern can be found. Bisol (1991) reports that the older group, speakers aged 36 55 years old, favors the conservative pronunciation of dental stops, showing a probability of rule application of (.37) Comparing across studies, speakers in the 2007 Generation 3 group (50 70 years old) were 38 and 58 years old, respectively, in 1995. The probability of palatalization for speakers in this age cohort in 1995 falls between Generation 2 (.29) and Generation 3 (.05). These results indicate that for speakers in this age group, the dental variant is still the norm. Ho wever, these speakers, also, slightly favor the palatal variant at very sparse rates, reaching in 2007 an unfavorable (.21) of rule application. It is worth noting that linguistic variables undergoing change might reach a point of relative stability. In f act, this is not an unusual finding. As Chambers (2002:364) points show str ong rates of palatalization distributed across age groups ranging from 0.05 to 0.91. Thus, the time window 4 captured in the 1995 data signaled a synchronic pattern 4 e window captured by synchronic studies of linguistic variation is limited by many factors, among which is the average life span of individuals of the particular time of the

PAGE 169

169 of a vigorous change. However, twelve years later, comparative data indicate that the ling uistic change has come to a pause. In line with previous research, Battisti and Hermans (2009:235) report that palatalization of dental stops shows signs of stabilizing in the speech community of Antnio Prado due to strong social conditioning. The fact t hat apparent time data from a neighboring state in Brazil shows an identical distribution further strengthens the results. This is a worthwhile finding since it indicates that UP in the speech community of Rivera is marching in line with southern BP var ieties in response to regional changes. In sum, as expected, age is the strongest social factor conditioning palatalization of dental stops in Rivera. Apparent time evidence of both data sets corroborates that palatalization of dental stops is age relat ed, since the frequencies of palatalization are strongly associated with age differences. Data results confirmed the hypothesis that younger speakers tend to prefer the innovative variant. However, the hypothesized increase in the use of the palatal vari ant over time among the younger generation is not confirmed in the present study. As seen above, cross sectional evidence indicates that palatalization of dental stops in the speech community of Rivera has stabilized. As Chambers (2002:360) argues, chang e in progress shows incremental increases in the use of a particular variant in the speech of younger people. Although such an increment in the use of palatalization of dental stops is found across age groups in apparent time; cross sectional comparisons point toward a state of relative stability at the speech community level. On the other hand, the age distributions indicate that speakers of all ages participate in palatalization of dental stops in Rivera, but different age groups represent

PAGE 170

170 different pa latalization rates. This finding point out what (Labov 1972) would call a economic status, will clarify community of Rivera. 4.7.4.2 Socioeconomic group Consistent with previous studies, the socio economic factor group does not have nearly as great an effect as the factor group age. The socio economic group demonstrates the second highest magnitude of effect in the analysis, obtaining a range of 37. This finding is consistent with previous studies of UP and across varieties of BP. Not only is this factor selected as significant but the constraint ranking mirrors that of the 1995 data. Results show that part icipants in the higher socio economic class bracket, mid middle class, strongly favor the palatal variant (.76), whereas low middle and working class speakers exhibit the same low probability of palatalization of dental stops (.39). It is worth noting tha t speakers in the low middle and working class groups show the same tendency to use the innovative variant. This finding may indicate that palatalization of dental stops is moving along social spheres over time. Once more, these findings resemble 1995 da ta results. Mid middle class speakers produced more palatalized variants (.81), whereas low middle and working class speakers tended to favor the conservative pronunciation of dental stops at (.34 and .30), respectively. As in previous research, palatal ization of dental stops shows a clear social stratification. T he wealthy and middle class inhabitants, who identify with Montevideo cul ture and distance themselves from UP speakers, favor palatalization of dental stops, whereas t he poor and working class inhabitants, who identify with local border culture,

PAGE 171

171 disfavor it. Furthermore, these results are consistent across dialects of BP. Battisti and Hermans (2009) found that speakers living in the urban area give room to the innovative palatalized variants, while informants living in rural areas prefer the more conservative non palatalized variant. Figure 4 5 shows a clear pattern of social stratification in which the wealthiest speakers tend to prefer the incoming palatalized variant while the low middle a nd working class speakers favor the co nservative form. The extralinguistic factor socio economic status provides an unambiguous picture. The results of this analysis indicate that palatalization of dental stops in Rivera can be interpreted as a sociolingui stic marker since it seems to signal social class, due to the clear break among the social groups. In fact, Labov (1972:220) argues that for a prestige marker, the higher a stigma tized markers, the reverse is true. This result points to the fact that palatalization As expected, socioeconomic status proves to be a strong social constraint conditioning palatali zation of dental stops. As hypothesized, mid middle class speakers tend to use the innovative variant more often than low middle and working class speakers. However, speakers in the latter groups present a very similar probability of rule application. Th ese results are in line with previous research in UP and across dialects as well.

PAGE 172

172 Figure 4 5. Comparison of rates of palatalization of dental stops followed by /i/ according to socio economic status in UP. 4.7.4.3 Gender The last extralinguistic factor group tested was gender. This social factor showed the lesser magnitude of effect in the analysis, obtaining a range of 24. Once again, the two sets of data show consistency. As expected, statistical results clearly reveal that women tend to produce more palatalized variants (.62) than men (.38). These results also pattern with the 1995 data, where women favored the innovative variant (.60) whereas men at (.39), showed a clear preference for the conservative realization of /ti/, /di/. Figure 4 6 compare s the rate of palatal variants in both 1995 and 2007 data sets according to gender. Figure 4 6 shows the findings in both studies As seen above, the picture is unambiguous showing that females are clearly ahead of men in using the innovative variant. Not only do genders pattern similarly across studies, they also exhibit almost identical probability of preference for the innovative variant. Noticeably women lead men by a considerable margin in palatalization of dental stops. These results indicate

PAGE 173

173 tha t females are leading the adoption of the innovative variant in the speech community of Rivera Figure 4 6. Comparison of rates of palatalization of dental stops followed by /i/ according to gender in UP The results of this analysis pattern across sp eech communities where sociolinguistic research has identified the important role of women in advancing language change. It is well known that women use fewer non standard variants than men of the same social group. In fact, as Chambers (2002) argues, th e linguistic correlations with gender are consistent and partly predictable (352). Thus, unsurprisingly, women proved to be advancing the process of adoption of the palatal variant in contemporary UP in Rivera In sum, as far as the social factors are con cerned, the 2007 data are generally consistent with previous research both in UP and across dialects of BP. As expected, young people, females, and mid middle class speakers favor the innovative variant, 0.39 0.6 0.38 0.62 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Male Female GoldVarb weights for palatalization AMC (1995) RMC (2007)

PAGE 174

174 while older people, males, and low middle and worki ng class speakers tend to disfavor it. social constraint conditioning palatalization of /ti/, /di/ in Rivera. Data results confirmed the hypothesis that younger speakers tend t o favor the palatal variant. However, the hypothesized increase in the use of this variant over time among the younger speakers is not confirmed in the present investigation. Cross sectional evidence indicates that palatalization of dental stops has reac hed a relatively stable mode at the level of the speech community. In line with previous research, stability is also found across varieties of BP. This finding signals that UP is marching alongside southern varieties of BP. Furthermore, here we have an example of a linguistic variable that is stabilizing as a sociolinguistic marker. As expected, socioeconomic status has been found to be a strong social constraint conditioning palatalization of dental stops. As multivariate analysis results indicate, m id middle class speakers tend to use the innovative variant more often than low middle and working class speakers. As predicted, women are ahead of men by a considerable margin in palatalization of dental stops. These results indicate that females are adv ancing the adoption of the innovative variant in the speech community of Rivera 4.7.4.4 Cross tabulations of s ocial f actors In order to further explore the interactions between social factors in the speech community of Rivera, it was necessary to run c ross tabulation analyses Cross tabulation analyses allow us to find out how the data are distributed for each intersection of factors (Tagliamonte 2006:182). In what follows, I present GoldVarb

PAGE 175

175 results of cross tabulations of each social factor group. I will start by cross tabulating age and gender social factors. Figure 4 7 show s the cross tabulation results of age and gender. As Figure 4 7 shows, there is a clear age and gender based pattern. Young females exhibit a considerably higher ratio of p alatalization of dental stops than older males. The youngest group presents the biggest difference between females and males in the adoption of the innovative variant. However, both males and females in the Generation 2 group show a similar pattern, with women slightly in the lead. It is also interesting to note that males in the Generation 1 and 2 groups pattern similarly in the adoption of the palatal variant. While older speakers are the most linguistically conservative group altogether, older males p resent almost categorical use of the dental realization of /ti/, /di/. Figure 4 7 Cross tabulation of overall percentages of palatalization of /ti/, /di/ according to age and gender (2007 data). The linguistic behavior of speakers in the Generation 2 group (30 49 years old) requires some comment. It can be argued that these speakers are entering a 0 10 20 30 40 50 60 15-29 years old (1st. Gen.) 30-49 years old (2nd. Gen.) 50-70 years old (3rd. Gen.) Female Male

PAGE 176

176 professional stage in their lives and as such are more sensitive to the social advantages of the incorporation of standard features into their speech. Th e linguistic marketplace concept (Sankoff & Laberge 1978) has been applied in sociolinguistics research to activity or occupation. We cannot underestimate the potentia l role of professional pressures operating on UP speakers wishing to advance in society. The job market in the area of public school education looks promising for community members with a ivalent alternative. Public school teachers, mainly females, revealed previous employment as duty free employees. Self reported data from participants revealed their opportunities to engage in daily interactions with Santana shoppers and their willingn their variety to the standard. Above all, younger generations in Rivera are aware of the social capital attached to the standard lang uage. In this respect, participants expressed their desire to learn standard Portuguese for practical reasons, in order to become more marketable in their professional life. As one speaker express ed ( 4.1 ) (4.1) A gente tm que olhar para o Brasil la t m mais oportunidades de emprego temos que pensar no futuro e estar preparado, no ? (YY055 m26) we have The perception of language as a tool in order to ascend higher in society is evident among the young and young adults. Young Riverans are quite aware of the linguistic value of standard Portuguese and its potential value not only in the job market but also

PAGE 177

177 for educational opportunitie s. However, Rivera dwellers are considerably aware of the stigma attached to the variety spoken in the city. In this respect, Riverans of all social classes share a community norm of linguistic evaluation: the social evaluation of UP as languag Riverans even refer to UP speakers as rompe idioma stigma attached to Rivera speech as part of their mixed identity and as a byproduct of conviviality with neighboring Brazil, mid middle class speakers attach negative evaluations to the variety and express their dissociation with UP non standard linguistic features. Figure 4 8 show s cross tabulations by gender and socioeconomic class. The Figure reveals a strong interaction between gender and social status. As seen below, there is a clear gender and socioeconomic based pattern. Women in the highest socioeconomic group in Rivera present the highest frequency of the innovative variant, followed by females i n the low middle and in the working class groups The social stratification of men is less dramatic. This finding conforms to the stance of sociolinguistic markers commonly found across speech communities. Labov (1981) stic markers, the mean values for women are shifted

PAGE 178

178 Figure 4 8 Cross tabulation of overall percentages of palatalization of /ti/, /di/ according to gender and socioeconomic class (2007 data). The very sim ilar linguistic behavior of males and females in the working class group can be interpreted within the social networks framework. Working class males and females share the same linguistic norms, reinforced through close knit social networks. As previously discussed (§2.3), Rivera is a close knit community, and thus barrios engage in a variety of activities on a daily basis. These activities many times involve other community members in different capacities such as neighbors, kin, coworkers, etc. This type of social interactions reinforces local linguistic norms and may also diminish the impact of external linguistic models. Feelings of are common in both working and low middl e class Rivera residents. Women are especially sensitive to external higher standards of correctness in language associated with upward social mobility (Labov 0 10 20 30 40 50 60 70 80 90 Females Males GoldVarb percentages for palatalization Mid-Middle class Low-Middle class Working class

PAGE 179

179 1990:214). Women in Rivera are sensitive to the social pressures imposed on them. Due to thei r weaker s ocial standing in the community, women tend to pay attention to This situation is exacerbated with the implementation of the teaching of Portuguese in public schools, which has brought a change in linguist ic attitudes among Rivera residents. Parents, mainly women, initially opposed the idea of having Portuguese classes at school. Parents feared that children will eventually switch to UP, defeating the whole purpose of attending school. Figure 4 9 show s cross tabulation of socioeconomic class and age. The results shown in Figure 4 9 indicate that palatalization of dental stops correlates both with socioeconomic status and age. As seen below the stratification decreases from left to right down the socia l domain from mid middle to lower middle and working class. Figure 4 9. Cross tabulation of overall percentages of palatalization of /ti/, /di/ according to socioeconomic class and age (2007 data). 0 10 20 30 40 50 60 70 80 Mid Middle class Low Middle class Working class GoldVarb percentages for palatalization 15-29 years old (1st. Gen.) 30-49 years old (2nd. Gen.) 50-70 years old (3rd. Gen.)

PAGE 180

180 Figure 4 9 also shows that mid middle class females a re the leaders of the adoption of the innovative variant in Rivera. It must be noted that mid middle class speakers view the incorporation of standard Portuguese as a way to increase their advancement in society. Mid middle class adolescents in Rivera at tend private schools where English and Portuguese languages have long been an essential part of the school curriculum. They also attend the Instituto de Lenguas where standard Portuguese is taught. Language classes are complemented with trips to popular Brazilian destinations, such as, Porto Alegre, Santa Catarina, So Paulo and Rio de Janeiro. Young people in Rivera of all social classes are avid followers of Brazilian media. Brazilian music, dances, movies, soap operas, shows i n general is preferred. The powerful Brazilian television network O Globo public events. Riverans can easily access Brazilian television and radio stations for free, while the paid national networks do not seem to appeal to Rivera inhabitants. As discussed above, young adults in the low middle and working classes present an interesting pattern. These speakers, just as their coetaneous in the mid middle class, seem to have incorporated the standard variant into their li nguistic repertoire in order to attain social mobility. Young adult speakers in the working class group have incorporated the prestige variant into their speech at much higher rates than young participants in the same social group. As discussed earlier, s ocial and professional pressures may be responsible for this linguistic behavior. 4.8 Discussion Making use of standard variationist methodology, I have put to empirical test a number of hypotheses about the nature of linguistic and extralinguistic facto rs

PAGE 181

181 conditioning the choice of the phonetic realizations of dental stops in UP. In the present investigation, I have compared data from two cross sectional studies conducted at different points in time (1995 and 2007) with a time depth of twelve years. Cr oss sectional studies are well suited to answer questions about language change or stability at the speech community level. Labov (1981) advises that the most straightforward approach to the study of change in progress is to combine studies of age distri bution with points in real time. By adding a longitudinal perspective to the study, it was possible to obtain a clear picture of the sociolinguistic evolution of palatalization of dental stops in the speech community of Rivera. The trend methodology imp lemented in the study allowed me to compare apparent time data from two studies at different points in time, 1995 and 2007. I have explored this approach with the objective of substantiating the change in progress hypothesized by Carvalho (1998). Based on this hypothesis, an increment in palatalization rates was expected if palatalization was indeed undergoing change; however, statistical analyses do not support this prediction. As seen above, the linguistic variable under investigation has not increased over time; instead, palatalization of dental stops has reached a state of relative stability at the speech community level. The first confirmation of this conclusion is obtained from the overall distribution of the variants. In the present investigation the overall distribution of variants shows that 29% (N=743) of the tokens analyzed are palatal realizations, while 71% (N=1783) correspond to the dental realizations of the variant. On the other hand, Carvalho reports 32% (N=719) palatal realization and 68% (N=1529) of non palatalizations of the variant. As mentioned above, an increase of palatalization over

PAGE 182

182 time was expected in order to confirm the hypothesized change in progress. The overall distribution of the palatal variant points to a relative st ability at the speech community level when compared at two discrete points in time 12 years apart. been detected in a given speech community, we will need evidence to show that s ome variation within the community is a direct result of the fact that in the recent past language learners acquired a different form of the language than they are acquiring sed change has moved further in the same direction. In fact, results show that palatalization of /ti/, /di/ has not increased over time but rather has reached a relative stability. The time window captured in the 1995 data signaled a synchronic pattern o f a vigorous language change in apparent time. Regression analyses showed that the most important social factor conditioning palatalization is age, clearly signaling change in apparent time. Additional evidence contributing to the language in change hyp othesis was the existence of sharp differences across age groups in the data (Carvalho 1998:179). Unfortunately, one of the challenges in quantitative sociolinguistics is the scarcity of real time data. In the specific case of Rivera, the only type of re al time data available are early qualitative descriptions of UP, in which the dental variant is characterized as the norm among UP speakers. Therefore, taking this account into consideration as real time evidence, it is reasonable to assert that at least in the late 1950s dental stops before /i/ in Rivera had only one linguistic variant. The linguistic variable palatalization of dental stops in the speech community of Rivera is heavily conditioned by extralinguistic factors, but it is also linguisticall y

PAGE 183

183 significant role than linguistic factors. These results indicate that palatalization of dental stops is a sociolinguistic marker, that is, sensitive to both linguistic an d social context factors carrying both social interpreta tion and evaluation (Labov 1972 ). In this chapter I have discussed quantitative analyses of palatalization of dental stops in Rivera. I have stated the rationale behind the hypothesis and reformulat ed the research questions. Linguistic and extralinguistic constraints were also presented along with distributional and multivariate analyses. In addition, I conducted a cross sectional analysis of the data to further explore the linguistic variable und er examination. The following chapter (Chapter 5 ) presents the quantitative analysis of vowel nasalization in UP. I will discuss the linguistic and extralinguistic factor groups and factors considered in the analysis. The results of both distributional and multivariate analyses will be discussed.

PAGE 184

184 CHAPTER 5 QUANTITATIVE ANALYSE S OF NASALIZATION 5.1 Background In this chapter I present quantitative results for the second linguistic variable under examination in the present study that is, the presenc e or absence of vowel nasalization in UP. Nasal vowels can be found across Romance languages, although they are found particularly in two varieties, French and Portuguese. Portuguese language is one of the two standard Romance varieties having nasal vowel s as independent phonemes (Sampson 1999:175). Unlike in French, where the a lternations between nasal and oral vowels are clearly indicated, in most varieties of Portuguese, vowel nasalization occurs generally whenever a vowel is followed by a nasal conson ant, regardles s of syllable structure (Lipski 1975:67). Lipski (1975) reports that vowel nasalization before a nasal consonant in the following syllable is ignored in many descriptions of BP although it is an accepted phenomenon in this variety Indivi dual and dialectal variations on BP vowel nasalization have been attested by Azevedo (1981:23) and Shaw (1986) Previous phonetic accounts have dealt with vowel nasality indirectly as a subsection of some other pho nological issue, such as stress o r develop ment of diphthongs However, vowel nasalization has not been fully accounted for. As seen above ( § 2.6.1.5), the tendency has been to look for language specific data to justify universal hypotheses of nasalization failing to account for a general descript ion of Por tuguese nasal vowels. T he hy pothesis then data approach favored implicit assumptions about phonetic representations. T

PAGE 185

185 morphophonological, phonetic, or dialectal variation or alternations has b een ignored (Shaw 1986:6). A lthough the literature on UP contains incidental reference s to the prese nce of vowel nasalization, this phenomenon has not been yet examined (§2.6.1.6) The sociolinguistic app roach of this research will account for dialecta l variations from spontaneous and informal speech, a method that is not commonly used to describe the vowel nasalization process. T he quantitative analysis o f vowel nasalization in UP will provide empirical evidence of the alternations available for UP sp eakers in contemporary Rivera. 5.2 Research Questions Linguistic variation is rarely referred to in the literature on Portuguese nasal vowels although there is evidence of the existence of dialectal variations conditioned by stylistic and sociolinguisti c factors such as social class (Shaw 1986:80). As indicated in the literature review ( §1.3.1), e arly descriptions of UP have applied qualitative and interpretative techniques to analyze vowel nasalization in UP, but there has been no a quantitative descrip tion of UP vowel nasalization from a variationist perspective. Rona (1965:35) remarks that in the Melense variety, Portuguese words ending in o and Spanish words ending in n have yielded only one ending o (1965:43). The other three Fronterizo variet diphthong o. Hensey (1982:15), in a study analyzing the influence of Spanish phonology on the Portuguese of bili nguals living in the Uruguayan Brazilian border, argues that /a/ rises to / / when preceded by a nasal consonant. Lipski (1994:343) points out that vowel nasalization in UP is variable and can be measured according to its degree of approximation to the Portuguese vowel system

PAGE 186

186 The sociolinguistic analysis of vowel nasalization i n UP represents a challenging and exciting area of linguistic inquiry since this phonological process has not been investigated from a variationist perspective. My variationist approach to account for the distribution, alternation, and variation of vowel nasalization in UP will fill a research gap in this field of inquiry. In this study I have included phonetic, morphological and social factors in order to determine which internal and external mechanisms have a significant effect on the distribution of th e variation. I first discuss the linguistic factors conditioning the variant selection. I will then discuss the social factors selected as presenting statistically significant effects on vowel nasalization in contemporary UP in Rivera. 5.3 Circumscribing the Variable Context Traditional phonologies of Portuguese distinguish two types of nasal vowels: one contrastive or phonological, the other allophonic or phonetic. Conventionally the former ( i.e. ), those capable of ent ering into a phonemic opposition with a corresponding oral vowel; the latter class is considered to be ( Mattoso Cmara 1970, 1972; Wetzels 1997:205). In order to decide what would constitute an instance of vowel nasal ization, it was necessary to choose from among competing analyses of Portuguese vowel nasalization. A main concern was whether to cons ider only contrastive nasal vowels (e.g., l, there ) or to include also vowel + nasal (VN) followed in the s ame or in the following syllable (e.g., campo cam a Table 5 1. Example of contrastive nasal vowels [ l ] l [ la ] la um [u] o

PAGE 187

187 Table 5 2. Example of vowel + nasal in the same syllable pu] campo mundo Table 5 3. Example of vowel + nasal in the following syllable ma] cama [k cunhado in A more critical point to take into consideration is the one raised by Lipski (1975) where he claims that restr nasal vowels appear to be analyzable as being followed in the same syllable by a nasal phenomenon of vowel nasalization in its 1 (Lips ki 1975:67). As mentioned above, in most dialects of Portuguese vowel nasalization occurs generally whenever a vowel is followed by a nasal consonant regardless of syllable structure. In order to account for what appears to be a basic and consistent fac t of Portuguese vowel nasalization, the criteria selection were extended to cover every instance of nasalization phonemic or not. For the present study the envelope of linguistic va riation (Labov 1972 ), that is to say, the variable context, included the p resence or absence of a fully nasal vowel or a nasalized vowel followed by a nasal closure (VN). No distinction was made to differentiate between nasal or nasalized vowels; instead they were merged into a single linguistic variable. The selection criterio for a more realistic analysis of vowel nasalization in BP, in order to account for every realization as it occurs in naturally occurring speech. The presence of vowel nasalization was coded as (n) or application o f the nasalization rule; the absence of vowel nasalization was coded as (o) or non application of the nasalization rule. T he 1 Lipski (1975:67) asserts that it is only in these positions where phone mic nasal vowels occur in Portuguese; vowel nasalization in the Carioca dialect is not confined to these environments.

PAGE 188

188 dependent variable was defined as binary, that is, as the presence or absence of vowel nasalization in UP. As indicated in the data analysis section (§3.6) instrumental analysis was not performed. 2 5.4 Exclusions I included in the analysis all words containing the presence of a fully nasal vowel or a sequence of a perceived nasal quality of the vowel followed by a nasal closure (VN). As it is standard in the methodology of variationist sociolinguists, I excluded from the analysis false starts or truncated utterances, repetitions and not understandable or not audible utterances. Distinctions between degrees of nasalization were not included in the analysis. As discussed above (§2.6.5), d egrees of nasalization do not appear to be lingui stically relevant since there appears to be no language in which these gradations are employed to differentiate meaning (Quicoli 1990). Further, distinctions between degrees of nasalization are compelling for acoustic or instrumental treatments of nasalization, but their treatment goes beyond the scope of the present study. 5.5 Linguistic Constraints Deciding which factor groups and factors to t est in any variationist study involves a deep understanding of the linguistic and or social influences that may condition speaker choice of a linguistic variable. In order to carry out the quantitative analysis of vowel nasalization in UP, I conducted an e xhaustive literature review on the process of BP nasalization (§2.6.1). The criteria for the selection of the linguistic factor groups were extrapolated from the literature review on nasalization and denasalization across 2 A similar procedure was used by Hansen (2001) in the study of lexical diffusion in French nasal vowels lization in Brazilian Portuguese.

PAGE 189

189 languages and prescriptive Portug uese language grammars. I integrated factor groups and factors in direct response to the most relevant issues raised in the body of literature on the nasalization process in BP. Previous studies reveal that stress, following and preceding phonological e nvironments, syllable structure, and functional category of the word are among the most common factor groups conditioning the variable pronunciation of vowel nasalization across languages. These factor groups were included in the analysis of vowel nasaliz ation in UP. 5.5.1 Syllable Stress As discussed above (§4.5.1), stress is funda mental to the entire Portuguese phonological system. It is a sound decision to test the effects of stress conditioning vowel nasalization in UP. Nasal vowels occur in stresse d as well as in unstressed syllables, word initial, medial, and final. Wetzels (1997) maintains that in all dialects of BP, allophonic nasalization is most systematic in stressed vowels, whereas contrastive nasalization is realized independently of stress. In order to consider a syllable as stressed any degree of primary or secondary stress was taken into account. It was hypothesized that vowel nasalization in UP would be more frequent in stressed syllables Table 5 4. Examples of vowel nasalization in UP according to syllable stress Syllable Stress Glossary Pre tonic untar ] Final stressed irm ] Word internally stressed canto tu]

PAGE 190

190 5.5.2 Following Phonological Environment The nature of the following phonological context has proven to be a strong linguistic constraint. In his study on the denasalization process in PBP 3 Guy (1981) discusses the importance of testing the effect of the segments following and preceding to whether they themselves were so argues that historically the Portuguese language has had vast documented cases of both progressive ( e.g., regressive vowel nasalization ( e.g., cegonha the need to test the hypothesis that being adjacent to another nasal segment might favor the nasalization process. In order to discr iminate as finely as possible the nature of vowel nasalization in UP, I coded seven factors to test their impact on the nasal vowels. As the literature review suggests ( § 2.6.14), I considered it necessary to code nasals separately, that is, alveolar and de ntal nasals in one group and palatal nasals in other. This procedure was followed for both following and preceding phonological contexts. It was predicted that following phonological environment is a strong linguistic constraint that would condition the nasalized variant. Table 5 5. Examples of vowel nasalization in UP according to following segment Following context Glossary Nasal consonant cama ma] Palatal nasal cegonha Glide muito Stop bone ca F ricative enfim Sibilant uns [ Pause alem ] 3 PBP stands for Popular Brazilian Portuguese.

PAGE 191

191 5.5.3 Preceding Phonological Environment As stated above (§5.5.2), preceding phonological context has been show n to influence the denasalization process in BP (Guy 1981). In order to examine the nature of the preceding phonological segment and the impact of progressive nasal assimilation, if any, I coded nine linguistic factors. I followed the same procedure as i n following phonological context on coding the nasal consonants separately. The prediction is that preceding phonological context constrains vowel nasalization in UP. Table 5 6. Examples of vowel nasalization in UP according to preceding context Precedi ng context Glossary Nasal consonant me [m j] mother Palatal nasal amanh [am ] Stop punho ] Fricative (non sibilant) fim Sibilant som [s] Lateral malandro dru] vibrant roncar [ Glide quo w] Pause um 5.5.4 Syllable Structure In the present investigation I have also included two non phonetic constraints. These factors are syllable structure and functional category of the word. Within this factor group, tokens were classified on the basis of the number of syllables of the lexical item: monosyllabic or polysyllabic. The rationale to include this factor group in the analysis was my initial observations and intuitions from the fieldwork conducted in Rivera. I considered relevant to test whether vowel nasalization was mainly found in monosyllabic words. The prediction is that monosyllabic words are nasalized more frequently than polysyllabic ones.

PAGE 192

192 Table 5 7. Examples of vowel nasalizatio n in UP according to syllable structure Syllable structure Glossary Monosyllables me [m j] mother Polysyllables cafezinho 5.5.5 Functional Category of the Word The second non phonetic linguistic constraint tested is functional category of the word The functional category of the word has been found to condition a reverse phonological process, weakening and deletion of verbal /n/. Poplack (1979) did not find significant phonological constraints on the deletion rule in Puerto Rican Spanish; however, she reports the significant conditioning of morphological category. As she ization in BP did not support this finding. Guy found that denasalization applied equally to nasal vowels that were plural markers and to those that were not (1981:331); ( e.g., they know abram homem jov em In order to test the potential role of morphological factors conditioning the variation of vowel nasalization in UP, I included in the analysis the functional category of the word. In this factor group nine factors were examined. I coded nouns and nouns with diminutive suffixation in two separate groups. It was hypothesized that vowel nasalization would affect vowels equally regardless of their morphological function. Summing up, five linguistic factor groups, three phonetic and two non phoneti c, were coded to account for the variable realization of vowel nasalization in Rivera. These groups were extrapolated from the literature review on nasalization and denasalization in Portuguese and across languages and early intuitions from the fieldwork. The following factor groups were examined: tonicity of the syllable, following and preceding

PAGE 193

193 phonological environment, syllable structure, and functional category of the word. It was hypothesized that vowel nasalization in UP would be more frequent in stressed syllables and monosyllabic words, and that it would affect all words regardless of their morphological status. An additional prediction was that following and preceding phonological environment would condition the selection of the nasalized varia nt in the speech of bilingual speakers in Rivera. In what follows, I describe the extralinguistic or social factors included in the present study. Table 5 8. Examples of vowel nasalization in UP according to functional category of the word Functional cate gory of the word Glossary Noun cozinha Noun and diminutive cafezinho Adjective malandro dru] Adverb muito Progressive form of the verb morando du] Non progressive tenh o ] 1 st pers. sing Conjunction embora ] Pronouns quem Preposition sem 5.6 Social Factors In examining vowel nasalization in UP, I analyzed the same social constraints considered i n the analysis of palatalization of /ti/, /di/ (§4.6). These factors include age, gender, and socioeconomic status. It was predicted that these factors would condition the linguistic variable under examination. In the next section I present and discuss th e results of multivariate analyses for both internal and social factors. 5.7 Results and Discussion The data set amounted to 2121 tokens, which I analyzed using GoldVarbX (Sankoff, Tagliamonte, and Smith 2005). The following section presents an overall

PAGE 194

194 distribution of the variants, a factor by factor distributional analysis and multivariate analysis of the contribution of factors to condition vowel nasalization in UP. 5.7.1 Overall Distribution of Variants Table 5 9 shows the overall distribution of the data grouped according to whether there was a nasal or an oral variant. The data show that the overall rate of vowel nasalization in contemporary UP is 21% (N= 447) and the oral realization is 79% (N= 1674) for a total of 2121 tokens coded. Table 5 9. O verall distribution of the realization of tokens of vowel nasalization in UP Vowel nasalization Oral realizations % N % N 21 447 79 1 674 Total N 2121 As shown in Table 5 9, 21% of variants are nasal vowels whereas oral vowels make up 79% of the data. In what follows, I present the results of the factor by factor distributional analysis. This analysis will allow us to determine the role of the independent factors conditioning the linguistic variable under examination. 5.7.2 Factor by Factor Distributional Analysis As already discussed (§4.6.2), a factor by factor analysis considers each independent variable one at the time (Tagliamonte 2006). Table 5 10 displays the distribution of vowel nasalization according to syllable stress in UP. Table 5 10. Distribution of nasalization by tonicity of the syllable in UP Syllable tonicity % N W ord internal ly stress ed 23 1397 Final stressed 17 724 Three linguistic factors were originally considered for the analysis, that is, pre t onic, word internal and final stressed syllable. However, the linguistic factor pre tonic stressed syllable was excluded in the final analysis because of its low rate of

PAGE 195

195 nasalization (2%). As seen above, nasalization in word internally stressed syllables is realized 23% (N= 1397) of the time. Note that this context represents twice the number of tokens in comparison with final stress. Final stressed syllables show 17% (N=724) of rate of nasalization. As Table 5 10 shows, word internal and final stresse d syllables account for all instances of vowel nasalization, while the context pre tonic stressed syllable presented insufficient data. This result indicates that vowel nasalization occurs categorically in stressed syllables in UP. This affirmation is supp reports that stressed vowels and diphthongs are inhibitors of denasalization in BP (231). Table 5 11 outlines the effect of following phonological environment in vowel nasalization in UP. Table 5 11. Distribution of nas alization by following phonological segment in UP Following phonological segmen t % N Glide 32 490 Nasal consonant ( palatal nasals ) 27 286 Stop 21 725 Fricative 16 217 Pause 7 403 Seven linguistic factors were considered to test the effect of following phonological segment: nasal consonants (non palatal), palatal nasals, stops, fricatives (non sibilant fricatives), sibilants, glides, and pause. However, in the last statistical run all nasal consonants and all fricatives were collapsed into two separate groups. T able 5 11 shows that glides and nasal consonants present the highest rates of vowel nasalization, 32% (N=490) and 27% (N=286), respectively. The high rate of glides points to a potentially favorable conditioning of a following glide or a vowel raised in this environment producing nasal diphthongs or triphthongs. As mentioned earlier (§4.4)

PAGE 196

196 vowel raising is pervasive in Brazilian Portuguese, raising pretonic mid vowels /e/ and /o/ to [i] and [u], respectively, in syllable final position. An additional factor to consider for the high incidence of glides in this position is that in BP final nasal vowels are phonetically diphthongized, ( i.e. ), /e(n)/ is realized as /ei(n)/ (Mattoso Cmara 1972:52). This result also reveals the potential role of the diphthongization process in UP. The high rate of vowel nasalization following nasals may appear to be an expected result; however, it is also a revealing finding in light of the phonological processes operating in the la nguage. As mentioned previously, both progressive and regressive vowel nasalization can be found in Brazilian Portuguese. Vowel nasalization in the Carioca dialect is found more frequently when any vowel is followed by the palatal nasal tenho punho the potential role of regressive nasal assimilation effects operating in UP. While stops 21% (N=725) and fricatives 16% (N=217) show relatively moderate rates of effect on nasal ization, following pause reveals a very low effect 7% (N=403). Table 5 12 outlines the effect of preceding phonological segment. Table 5 12. Distribution of nasalization by preceding segment in UP P receding phonological segment % N Pause 3 8 133 Liq uid (vibrant, laterals) 26 341 Stop 2 0 867 Fricative ( sibilant s ) 19 7 04 Glide 3 76 To examine the effect of preceding environment conditioning vowel nasalization in UP, I coded nine linguistic factors, namely, non palatal nasal consonants, palatal nasals, stops, non sibilant fricatives, sibilants, laterals, vibrants, pause, and glides. After several statistical runs it was necessary to go back to the data to recode, exclude,

PAGE 197

197 and collapse some of the linguistic factors yielding a final count of fiv e factors. Nasals were combined into a single group; however, they were excluded in the final statistical run due to interacting effects with other factors. The interaction will be clarified with the multivariate analysis and cross tabulation of the data. Four linguistic factors were collapsed into two groups: fricatives and sibilants; and vibrants and laterals were joined As seen above, pause 38% (N=133) and liquids 26% (N=341) present the highest rate of vowel nasaliz ation in this phonological environment. While stops 20% (N=394) and fricatives 19% (N=704) show comparable rates of nasalization, glides reveal a very low rate 3% (N=76). The presence of preceding pause or zero phonetic environment at the top of this grou p may indicate that vowel nasalization is analyzed through word initial syllables in UP. This assumption will be clarified by the multivariate analysis. The first non phonetic constraint included in the analysis was syllable structure of the word. I code d the data into two groups, whether a lexical item was a polysyllable or a monosyllable. Table 5 13 shows the marginal results of this analysis. Table 5 13. Distribution of nasalization by syllable structure in UP Syllable structure % N Polysyllabl e 23 1837 Monosyllable 9 284 Table 5 13 reports that polysyllabic lexical items outnumbered monosyllab ics 23% (N=1837) to 9% (N=284). The last group tested was functional category of the word. I coded nine factors in this factor group: nouns, nouns with diminutives, verbs, progressive forms of the verb, adjectives, adverbs, prepositions, pronouns, and conjunctions. Three factors were excluded in the final analysis due to insufficient data in the cells: pronouns, prepositions

PAGE 198

198 and conjunctions. Noun s and nouns containing diminutive suffixation were tested separately but collapsed together into a single group. The final statistical output is shown in Table 5 14. Table 5 14. Distribution of nasalization by functional category in UP F unctional categor y of the word % N P rogressive forms of the verb 32 184 Noun (noun and diminutive ) 3 2 904 Adjective 16 323 Adverb 6 306 Verb (non progressive forms) 6 404 As Table 5 14 shows, progressive form of the verb and nouns show the highest rates of nasalization 32% (N=184 and 904), respectively. While adverbs and non progressive forms of the verb have a relatively low rate of nasalization, only 6% (N=306 and 404), respectively; adjectives are found in the middle with a 16% rate of nasalizat ion (N=323). The distributional analysis indicates that vowel nasalization is a linguistic variable that implicates both phonology and morphology. Final /n/ functions as suffixal inflection in Portuguese, thus affecting both levels of grammar. This result signals the potential interaction between phonological and morphological rules operating in the process of vowel nasalization in UP. In sum, the tabulation of effects discussed above suggests that the five linguistic factors examined in the present study condition the occurrence of vowel nasalization in UP. As discussed earlier (§4.7.2), a factor by factor analysis, although informative, does not show the combined impact of all factors tested simultaneously (Tagliamonte 2006). In the next section, I anal yzed the data using the variable rule application of the program. This analysis will reveal which factor or factor groups contributes statistically significant effects to the probability of vowel nasalization in UP.

PAGE 199

199 5.7.3 Multivariate Analysis of the Con tribution of Linguistic Factors of Vowel Nasalization in UP. Table 5 15 shows the results of the multivariate analysis of the contribution of linguistic factors detected by the stepwise multiple regression procedure incorporated in the variable rule progra m as significant to the probability of vowel nasalization in UP. Table 5 15 shows the binomial step up and step down analysis of the data. Three conditioning factors stand out as the most salient predictors of vowel nasalization: preceding context, funct ional category of the word, and following phonological segment. These three groups contribute statistically significant effects to vowel nasalization. Two linguistic factor groups were eliminated by the multiple regression analysis: tonicity of the syll able and syllable structure of the word. The highest constraint is exerted by preceding environment. Pause or zero phonetic environment favors vowel nasalization (.64) whereas glides and fricatives disfavor it (.15 and .44), respectively. Liquids (.57) fo llowed by stops (.53) slightly promote vowel nasalization in UP. The presence of a pause at the highest point in the constraint hierarchy might indicate that vowel nasalization in UP is more likely to occur in syllable initial positions. This interpreta tion seems contradictory with the characterization of pause as neutral in phonological rules (Guy 1981:330). These results do not seem to provide a plausible explanation of the linguistic variable under analysis. When we look at the following phonological context, the constraint hierarchy shows a reverse order of the preceding context. The odd distribution of the constraint hierarchy in this environment needed to be corrected. Upon noticing this configuration, I ran cross tabulations of all the linguistic factors and noticed

PAGE 200

200 badly distributed cells and interacting factors. In view of these facts and in order to correct this irrational distribution, I recoded the data and run a new multivariate analysis. Table 5 15. Variable rule analyses of the combinati on of linguistic factors selected as significant to the probability of nasalization in UP Corrected mean .16 Log likelihood 955.512 Total N 2121 Preceding phonological segment Pause Factor weight 6 4 % 3 8 N 133 Liquids (laterals and vibrants) .57 26 341 Stops .53 20 867 Fricatives (and sibilants) .44 19 704 Glides .15 3 76 Range 49 Functional category of the word Progressive forms, gerunds .70 32 184 Nouns .66 3 2 904 Adjectives .47 16 323 Adverbs .27 6 306 Verbs (non progressive forms) .26 6 404 Range 44 Following phonological segment Glides .62 32 490 Nasal consonants .59 27 286 Stops .49 21 725 Fricatives (and sibilants) 42 16 2 1 7 Pause .35 7 403 Range 27 Syllable structure Monosyllable [.58] 9 284 Polysyllables [.49] 23 1837 Tonicity of the syllable Final stressed [.52] 17 724 Word internally stressed [.49] 23 1397 Note: Facto r groups not selected as significant are shown in square brackets. I must note at this point that the reanalysis of data in methodological sociolinguistics is standard practice. Sometimes it is necessary to go back to the data

PAGE 201

201 and recode and collapse gro ups in order to account faithfully for the linguistic variable under investigation. For the reanalysis of the data, I excluded two factor groups, tonicity of the syllable and syllable structure of the word. These factor groups were already eliminated in the first multivariate analysis and their factor weights approximate toward the mean (.5), indicating rather neutral effects conditioning the dependent variable. As the factor by factor analysis shows, vowel nasalization in UP occurs categorically in stre ssed syllables. As for syllable structure of the word, we noted that nasalization was more frequent in polysyllabic than monosyllabic words. Following nasals were excluded from the analysis due to interaction effects with adjectives and adverbs. Followi ng glides were also excluded due to their low percentage (3%). Guy (1988) notes that any context over 95% or fewer than 5% should be removed from variable rule analysis, since they are not variable. As discussed above, preceding pause was excluded from th e analysis and adjectives and adverbs were collapsed into a single group due to interacting effects. In what follows, I present the results of the reanalysis of the data of the contribution of linguistic factors detected by the stepwise multiple regressio n procedure incorporated in the variable rule program as significant to the probability of vowel nasalization in UP. Table 5 16 shows the overall distribution of the data grouped according to whether there was a nasal or an oral variant. The data show the overall rate of vowel nasalization in contemporary UP is 14% (N= 222) and the oral realization is 86% (N= 1412) for a total of 1634 tokens. These results represent a lower overall rate of vowel nasalization t han the first analysis showed.

PAGE 202

202 Table 5 16. Overall distribution of the realization of tokens of vowel nasalization in UP Vowel nasalization Oral realizations % N % N 14 222 86 1 412 Total N 1634 The next step is to run the binomial step up and step down analysis of the data. T he log likelihood improved notably, from 955.12 to 596.211, revealing a better fit of the analysis. 4 Table 5 17 displays the results of the multivariate reanalysis of the contribution of linguistic factors of vowel nasalization in UP. Table 5 17. Vari able rule reanalysis of the combination of linguistic factors selected as significant to the probability of nasalization in UP Corrected mean 11 Log likelihood 596.211 Total N 1634 Functional category of the word Progressive form of the verb Factor weight .77 % 32 N 181 Nouns .62 17 587 Adjectives and adverbs .46 10 544 Verbs (non progressive form) .22 4 322 Range 55 Preceding phonological segment Stops .59 16 581 Liquids (latera ls and vibrants) .52 16 376 Fricatives (and sibilants) .47 12 442 Nasals .42 10 165 Glides .11 2 70 Range 48 Following phonological segment Stops [.53] 17 957 Fricatives (and sibilants) [.51] 13 255 Pause [.43] 7 422 N ote: Factor groups not selected as significant are in square brackets. 4 of the analyses. Figures closer to 0

PAGE 203

203 Table 5 17 reveals that functional category of the word and preceding phonological environment contribute statistically significant effects to vowel nasalization in UP. The factor g roup following phonological context was eliminated in the step down analysis. In the first multivariate analysis (Table 5 15) following context presented the lowest relative strength (27) of the three factor groups selected. Unsurprisingly, in the reanal ysis of the data this factor group was eliminated As seen above, functional category of the wor d is the strongest factor group conditioning vowel nasalization in contemporary UP, followed by preceding phonological environment. This is the reverse order of the first multivariate analysis, where preceding environment was stronger than functional category of the word conditioning the nasalized variant, presenting a range of 49 and 44, respectively (Table 5 15, 5 17). The next section provides a detailed d escription of each factor selected as significant in the reanalysis of the data. 5.7.3.1 Functional category of the word First, the greatest effect is contributed by functional category of the word. In this analysis, the constraint hierarchy of the first multivariate analysis is maintained, that is, progressive form of the verb, nouns, adjectives and adverbs, and non progressive form of the verb (Table 5 15, 5 17). Besides being selected as statistically significant, this factor group shows a relative mag nitude of its effect with a range of 55. Not only is this factor group selected as the highest constraint conditioning vowel nasalization in UP, but the relative strength of its effect is quite significant. This finding is consistent with previous studie weakening and deletion of final /n/ in Puerto Rican Spanish,

PAGE 204

204 (1996) study of phonetic nasalization in BP 5 report the favorable effect of morphological status of th e word conditioning vowel nasalization. In the present analysis, verbal categories such as the progressive or gerund forms, favor the nasalized variant at (.77) probability, while non progressive forms of the verb strongly disfavor it (.22). Nouns, incl uding nouns with diminutives, provide a favorable context for vowel nasalization at (.62). As seen above (Table 5 17), the two favorable contexts, progressive forms of the verb and nouns, pattern with a clear division between them. Adjectives and adverbs (.46) do not constitute a favoring environment for vowel nasalization in UP. These results suggest that vowel nasalization in UP distinguishes major categories of the grammar. The constraint hierarchy presented in the present study patterns similarly acr oss nasalization the most followed by nouns (.55), while adjectives (.49) and adverbs (.21) disfavor it. Results suggest that the mechanisms that produce variation from the underlying grammars of UP and BP are very similar. The results also reveal that vowel nasalization in UP patterns across dialects. Progressive forms of the verb exert the highest effect on favoring the nasalization rule with a .77 probability. This fi nding provides evidence to suggest that vowel nasalization is more likely to occur in sequences such as ndo found in gerunds and progressive forms. The position of progressive form of the verb at the top of the constraint hierarchy requires some comment. In contemporary BP, the use of a gerund 5 I n their study of vowel nasalization in BP, Abaurre and Pagotto (1996) did not account for phonological or contrastive nasalization since this is a categorical phenomenon in BP. The scope of their analysis dealt with phonetic or contextual nasalization exc lusively (1996:496).

PAGE 205

205 has been extended to indicate a temporary state that would be absent if the simple present were used; as in, c num tem cerveja preta? c sabe que eu no estou tendo? can be argued that UP speakers have grammaticalized the use of the gerund to other functions. However, a discourse pragmatic analysis would be necessary in order to determine the specialized meaning of the gerund as it is used in the speech community of Rivera. In informal speech the sequence nd of the present progressive suffix, ( e.g. falando comendo dormindo ) may be pronounced simply as [n], (e.g., falando .nu], wi thout the /d/) (Lipski 1975, Giangiola 2001:138). For Lipski (1975:64) this realization indicates that illiterate BP speakers analyze nasal vowels as a sequence of oral vowel plus nasal consonant. However, Rona (1965) argues that UP speakers analyze nasa l vowels as unique phonemes (§2.6.1.6). Rona bases this assumption on the analysis of writings of schoolchildren in Rivera where he found /n/ deletion (i.e. domigo As Table (5 17) shows, nouns proved to be a strong linguistic factor favoring the nasalization rule with a .62 probability. Lipski (1975:67) points out that vowel nasalization before a nasal consonant in the following syllable is an accepted phenomenon in the Carioca dialect and that this type of nasalization is more mark ed in some words than in others. In view of this argument and observing the grammatical conditioning of vowel nasalization, we may be looking at lexical effects. However, vowel nasalization before a nasal consonant in the next syllable is an on going de velopment in modern BP. This development suggests that BP is moving in the direction of adopting

PAGE 206

206 the maximally general universal schema of vowel nasalization: V /___ N (Lipski 1975). It remains to be seen how this type of vowel nasalization will be an alyzed by speakers of UP in the speech community of Rivera. I will discuss this point further in connection with the results obtained in preceding phonological environment. The role of grammatical conditioning has been interpreted as evidence of lexical d iffusion. Hansen (2001) examines the notion of lexical diffusion in a study of change in Modern French nasal vowels which appear to be undergoing a counterclockwise chain shift, ( i.e. ), a vowel change in place of articulation. 6 Hansen points out that alth ough prosodic and phonetic factors condition the variation of nasal vowels, independent lexical differences exist (2001:248). In this study, Hansen reports that word class seems to constrain the innovative pronunciation of the vowel where adverbs (en ding in ment particularly) lead the new pronunciation followed by verbs and nouns (243). In order to assess the potential role of lexical effects conditioning vowel nasalization in UP, a word frequency and lexical analysis would be necessary. However, t his task will be the left to future investigation. Table 5 17 shows that adjectives and adverbs, collapsed into a single group, do not favor the nasalized variant (.46). Previous studies have suggested a correlation between adjectives and derivational suf fixation. As discussed above Hansen (2001:243) found that word class seems to constrain the new pronunciation of the vowel in Modern French. Abaurre and Pagotto (1996:509) report that adverbs block the nasalization rule (.21). Cross tabulation analys is confirmed the assumption that all adverbial instances contained the suffix mente. 6 / approaches / /, / approaches / /, and / / becomes very rounded and closed ( bain [b banc ], bon

PAGE 207

207 In the present study, the low probability rate presented by adverbs and adjectives may be interpreted as a result of derivational suffixation, most likely by the diminu tive suffix (z)inho/a Upon looking at the tokens, I noticed this type of adverb and adjective in the data (e.g., baratinho gordinha devagarinho on. Recall that following nasals were excluded from the analysis due to interaction effects. Cross tabulation of the data reveals that following nasal interacts with adjectives and adverbs (§5.7.3). There seem to be morphophonological internal constraints guiding this pattern. As discussed above (§2.6.4), Quicoli (1990:322) proposes an analysis for the phenomenon of vowel alternation and nasalization in Brazilian Portuguese. Quicoli argues that the diminutive suffix (z)inho/a sense that it is (1990:322). In addition to the diminutive suffix (z)inho/a, the superlative issimo/a and the adverbial suffix mente make up the class of cyclic su ffixes in Portuguese. He notes that these suffixes behave differently from others suffixes, since they affect the application of the nasalization rule in Portuguese Consider the following example s caminha caminha [k (examples from Quicoli 1990:322). Table 5 18. Cyclic application of the stress rule in Portuguese (a) [kami + a] (b) [[kam ] + i a # ] primary stress -------------------------------- secondary stress [kam a] [km a ] [k a] nasalization

PAGE 208

208 As Table 5 18 (a) illustrates, the unstressed vowel in the first sy llable is not nasalized as predicted by the nasalization rule (§2.6.4 ). Table 5 18 (b) shows that the vowel in the first syllable has been marked for secondary stress; it would now undergo the nasalization rule As stated above, morphological processes s uch as derivational and inflectional alternations and progressive verb forms might be interfering with the application of the nasalization rule. However, this correlation needs to be further examined (i.e. code the data for root morphemes and suffixal an d derivational morphemes) in order to discover other potential morphological phenomena acting upon the process. Non progressive forms of the verb (i.e. present, preterit, subjunctive, etc.) disfavor the nasalized variant in UP (.22 ). This result indicate s that UP speakers must resort to some other linguistic strategies to convey verbal plurality in order to avoid grammatical ambiguity. In Standard Portuguese language, nasalization of a final vowel is the principal marker of plurality in the verb, where th e verb must agree with its subject in person and number. This type of nasalization is regularly followed by diphthongization. The diphthongs [ w], [ j], and [ j] arise as the result of affixation in inflectional morphology. In verb forms, the third per son plural ends in [ w] or [ j], as in falam [fl w] PRES IND or falem [fl unstressed, so that there are often other phoneti c differences between singular and An additional phenomenon affecting final nasal vowels in popular BP is monophthongization or diphthong contraction. Monophthongization reduces an

PAGE 209

209 unstressed diphthong by el imina ting the nucleus creating alternations such as / (e.g., falam [fl and sabem [sb j] to [sb ] ). For Wetzels (1997:221) all nasal vowels may show some degree of phonetic diphthongization especially when stressed and word finally. Thus, unstressed [ ] in word final position. In this respect Lipski argues that: what appears to be in operation is in effect a surface phonotactic constraint which attaches a nasalized glide at the appropriate place in the word, in order to satisfy the surface structure of the language, a phonotactic constraint learned by each native speaker as part of his a utomatic competence (1975:71). Poplack (1979) examined the effect of regular verbs ( e.g., e.g., and their effect on loss of final /n/ in Spanish Results show that regular verbs favor deletion of verbal /n / at (.73), while irregular verbs disfavor it (.27). Poplack argues that although verbal /n/ deletion on regular verbs could lead to ambiguity, languages have adapted to such processes through changes in morphology and higher levels of grammar. (1979:371) A context for ambiguity in UP would be the absence of nasalization in pairs like come PRES IND., and come m disfavor vowel nasalization. We can infer that UP speakers mark the plurality in the personal pronoun or use some other mechanism to convey verbal plurali ty. Nevertheless, this is just a speculation. A more detailed account of this phenomenon would be necessary to reach conclusive arguments, the scope of which goes beyond the present investigation.

PAGE 210

210 Summing up, functional category of the word exerts the highest effect conditioning vowel nasalization in UP. This finding is consistent with previous studies across final /n/ in Puerto Rican Spanish, (1996) study of nasalization in BP, report that the morphological status of the word conditions vowel nasalization. The main finding is the presence of interacting effects between the morphological process, derivational and inflectional alternations (i.e. progressive forms of the verb), and the application of the nasalization rule in UP. There seems to be an internal grammatical constraint guiding this pattern. However, this correlation needs to be further examined in order to discover other potential m orphological phenomena acting upon the process. 5.7.3.2 Preceding phonological context The factor group preceding phonological context exerts the second highest effect conditioning vowel nasalization in UP, obtaining a range of 48. This finding is consi stent stronger than the following segment conditioning nasal vowels in French (2001:225). When comparing across multivariate analyses, it can be noted that the constraint hi erarchy was slightly altered (Table 5 15, 5 17). Preceding phonological environment distinguishes between glides and consonants and classifies them by manner of articulation. Stops favor vowel nasalization (.59), whereas glides (.11), fricatives (.47), and nasal consonants disfavor it (.42). Liquids, collapsed with vibrants, (.52) slightly promote vowel nasalization in UP. A preceding stop exerts a rather weak effect on vowel nasalization (.59). Guy (1998:210) found that preceding velars favor denasali

PAGE 211

211 finding, it would be reasonable to assume that a preceding palatovelar is not a favorable environment for nasalization. Both vowel raising and palatalization of dental stops can interfere with the application of the nasaliza tion rule in this environment. In words such as ontem i] ordem i] underlying /e/ in unstressed position palatalizes the preceding dental stop, the nasalization rule is blocked. As discussed before (§2.6.4), vowel raising does not affect nasal vowels, but rather monophthongal, unstressed, oral mid vowels, that is, /e/ and /o/. We can see in this phonological context the interaction of the two linguistic variables under study, palatalization of dental stops and vowel nasalization. Table (5 17) reveals that liquids exert a neutral effect conditioning vowel nasalization at (.52). The neutral effect is indicated by the proximity to (.5). This result is nasal consonants condition ing the nasalized variant at (.5). In the present study, nasals disfavor the application of the nasalization rule at (.42). Data analysis shows that a preceding nasal does not favor vowel nasalization, thus blocking its effect and the potential result of progressive nasal assimilation. The disfavoring role of a preceding nasal in this context may be interpreted as an indication that progressive nasal assimilation is not a productive phenomenon in the vowel nasalization process in UP. As discussed above ( §2.6.1.2), BP is sensitive to both progressive and regressive nasal assimilation. However, as Lipski (1975:75) argues, progressive nasal assimilation seems to be an on going process in modern Carioca dialect. On the other hand, regressive nasalization is the predominant historical process that accounts for most nasalized vowels and diphthongs in Portuguese (Shaw 1986:96). As noted above (§2.6.1.4), regressive nasal

PAGE 212

212 assimilation is the most common manifestation of vowel nasalization in BP. That is nasal vowel is followed in the same syllable by an orthographically indicated The factor by factor analysis shows that following nasals obtained a high rate of vowel nasalization (Table 5 11). Cross tabulation of the data indicates that most of the nasals in this environment are palatal and the vowels present a categorical status (i.e. they are nasal). This result is supported by claims in the literature review stating that nasalization before palatal nasals occurs independently of the position of primary stress, just like contrastive nasalization does (Wetzels 1997:218). A large sc ale survey carried out in Brazil confirmed the robustness of these phenomena. Abaurre and Pagotto (1996) found that nasalization is categorical when the vowel precedes a palatal nasal consonant, regardless of syllable stress. As data results suggest, a f ollowing palatal nasal is shown to be a categorical context, that is, when the application of the nasalization rule always applies in UP. This finding is consistent with language universal tendencies attesting that most nasalization occurs through regress ive nasal assimilation rather than progressive assimilation. Glides disfavor the application of the nasalization rule at (.11). Portuguese has both rising or crescent diphthongs and falling or decrescent diphthongs, depending on whether the salience is o n the rightmost or the leftmost vocalic element, respectively (Shaw 1986:15). A preceding glide followed by a nasal vowel can be found in rising diphthongs in Portuguese, as in quando environment does not promote nasalization in UP. Cross tabulated data show that

PAGE 213

213 glides interact with adverbs and adjectives, and verbs (i.e. progressive and non progressive forms). This result indicates, again, the i ntersection of different levels of grammar conditioning vowel nasalization in UP. As expected, preceding phonological environment has been found to have a significant effect on the distribution of the variation. This finding is consistent with studies ex amining vowel nasalization across languages (Hansen 2001). Stops favor vowel nasalization, while glides, fricatives and nasal consonants disfavor it. Liquids exert a rather neutral effect conditioning the application of the rule. The role of the nasal im peding nasalization in this linguistic context suggests that progressive nasal assimilation is not a productive process operating in contemporary UP. As Lipski (1975:75) argues, progressive nasal assimilation is an on going development in modern BP. It r emains to be seen how this process will develop in contemporary UP. In sum, as far as the linguistic constraints are concerned, the results are generally consistent with previous research across languages and varieties of BP. As multivariate analysis res ults indicate, functional category of the word and preceding phonological environment play a significant role in the distribution of the variation. The role of functional category of the word conditioning vowel nasalization in UP comes as an unexpected res ult. I hypothesized that the nasalization rule would affect equally all words despite of their morphological status. As the results show, vowel nasalization is a linguistic variable, constrained by factors of morphophonological and sociophonetic nature. However, a more detailed analysis modeling the grammatical constraints operating on the variation is needed in order to reach conclusive arguments.

PAGE 214

214 Cross tabulation analysis of the data shows that preceding nasals and glides interact (i.e. overlap) with adjectives and adverbs, verbs, and non progressive forms of the verb This is a natural result when examining naturally occurring speech. Further examination of the data, controlling for detailed morphological aspects, would shed more light on the interp lay of the different levels of grammar involved in the variable realization of vowel nasalization in the speech community of Rivera. Preceding phonological environment constrains vowel nasalization in UP. Results show that stops and liquids favored the n asalized variant while fricatives, nasals and glides disfavor it. The most outstanding finding is the role of a preceding nasal disfavoring the application of the nasalization rule in this phonological environment. This result suggests that progressive n asal assimilation in UP is not a productive phenomenon in the vowel nasalization process in UP. This finding is in line with the contemporary nasalization process development in BP. As data results suggest, nasalization when the vowel precedes a palatal nasal is categorical in UP. This type of nasalization is a result of regressive nasal assimilation. These findings are consistent with language universal tendencies attesting that most nasalization occurs through regressive nasal assimilation than progre ssive assimilation. In the next section, data will be analyzed using the variable rule application to examine the effect of extralinguistic factors conditioning vowel nasalization in UP. As already noted, this analysis reveals whether a particular factor or factor group contributes statistically significant effects to variant choice when all the factors are taken into account at the same time (Tagliamonte 2006).

PAGE 215

215 5.7.4 Multivariate Analysis of the Contribution of Extralinguistic Factors of Vowel Nasalizati on in UP. Table 5 19 displays the results of the multivariate analysis of the contribution of extralinguistic factors selected as significant to the probability of vowel nasalization in contemporary UP. The factors are selected by the stepwise multiple re gression procedure incorporated in the variable rule program as significant to the probability of vowel nasalization in UP. As seen in Table 5 19, the multivariate analysis selected the factor group age as significant. The other two social factors examin ed, socio economic class and gender, were eliminated in the final statistical run. Table 5 19 reports that the social factor age has a significant effect on the distribution of the variation of vowel nasalization in contemporary UP. Table 5 19 Variable rule analysis of the combination of extralinguistic factors selected as significant to the probability of nasalization in UP. Corrected mean .20 Log likelihoo d 1344.428 Total N 2688 Age Factor weight % N 30 4 9 56 24 933 50 70 48 19 8 82 15 29 46 1 7 87 3 Range 10 Socio economic group Mid Middle class [.52] 21 869 Low Middle class [.51] 20 893 Working class [.47] 19 926 G ender M ale [.5 2 ] 22 1479 Fem ale [.47] 18 1 209 Note: Factor groups not selected as significant are in square brackets. Table 5 19 shows vowel nasalization in UP, obtain ing a range of 10. This group was the only social factor

PAGE 216

216 ranked as statistically significant, while socioeconomic status and gender were eliminated in the binomial step up and step down analysis of the data. These results indicate that vowel nasalization i s constrained mainly by linguistic factors and thus it may be considered neither a sociolinguistic marker nor an indicator. low (10). In addition, the ratios of pre ference among age groups do not represent substantial age stratifications. This result is an indication that vowel nasalization is not a linguistic variable undergoing change in UP. Unlike palatalization of dental stops, discussed in Chapter 4 here the age distributions across generations do not point to any direction of change. Abaurre and Pagotto (1996:505) report a similar result. In their study the factor group age was eliminated, signaling that phonetic vowel nasalization in BP is a stable variabl e. However geographic region do play a significant role conditioning the variability. Abaurre and Pagotto found more vowel nasalization in northern varieties of BP, Recife (.66) and Salvador (.57), less in southern varieties, So Paulo (.43) and Porto Al egre (.34), with Rio de Janeiro (.52) in the middle. While data results indicate that there are no significant differences between the age groups, speakers in the Generation 2 group (30 49 years old), slightly favor the nasalized variant at (.56). Speak ers in the Generation 1 group (15 29 years old) and in Generation 3 (50 70 years old) pattern similarly; they disfavor the nasalized variant at (.46 and .48), respectively. This finding provides us with an interesting scenario. So far we know that the lin guistic variable vowel nasalization is not totally constrained by social factors, but age does seem to be a considerable factor, especially between speakers in the Generation 2 group.

PAGE 217

217 This result is consistent across languages as a similar result was found on a study on vowel nasalization in Southeastern France. Although limited by a small number of old, as well as those with a higher level of education and mobility produce more forms that are similar to Reference French, showing 91.9% of vowel nasalization more than the other participants. Violin argues that participants in this age bracket are working age people and expected to project a certain image of respons ibility and respectability, achieved through language approximation to Reference French. 7 As discussed in Chapter 4 speakers in the Generation 2 group (30 49 years old) exhibit a distinctive linguistic behavior. These speakers are entering a professional stage in their lives and therefore are more sensitive to the social advantages of the incorporation of standard features into their speech. We cannot underestimate the potential role of professional pressures operating on UP speakers wishing to advance i n society. As discussed above (§1.2) Santana has a bigger and stron ger economic base than Rivera; Rio Grande do Sul is one of the most prosperous Brazilian states. Brazil possesses a stronger economy which provides mor e jobs and educational opportunities. Summing up, as far as social factors are concerned, results are generally consistent with previous studies across languages and varieties of BP. The only social factor shown to have a significant effect on the distri bution of the variation is age, although its effect is relatively moderate. This result is supported by previous studies 7 Reference French is used as a neutral label to describe a variety of French that is regarded as a point of

PAGE 218

218 examining vowel nasalization across languages (Violin 2001) and other varieties of Portuguese (Abaurre & Pagotto 1996). The fact that extralinguistic factors did not play a key role on the distribution of the variation is a somewhat e xpected result. As Labov (1972 :251) argues indicate that vowel nasalization is constrained mainly by linguistic factors and thus it may not be considered a sociolinguistic marker or an indicator. 5.8 Discussion The following conclusions can be drawn from the anal ysis of vowel nasalization in UP. Vowel nasaliz ation is a linguistic variable constrained mainly by morphophonological and sociophonetic factors. As seen above, phonological processes interact with affixation in inflectional and derivational morphology in the process of vowel nasalization. As data res ults show, these morphophonological processes highly constrain vowel nasalization in UP. As for the first two hypotheses postulated in the present study, it was predicted that vowel nasalization would be more frequent in stressed syllables and that follo wing phonological environment would constrain vowel nasalization in UP. The factor by factor analysis reveals that vowel nasalization occurs categorically in contexts of stressed syllables. The distributional analysis also indicates that vowel nasalization is near categorical when a vowel precedes a palatal nasal. This result is supported by claims in the literature review stating that nasalization than before /m/ or /n/. Multivariate analysis shows that functional category of the word has the greatest effect on rule application, followed by preceding phonological environment. This was an

PAGE 219

219 unexpected result. It was h ypothesized that vowel nasalization would apply to all words despite their morphological status; however, phonological and morphological effects interact with the application of the nasalization rule. Nouns and progressive forms of verbs promote nasalizat ion while non progressive forms of verbs, adjectives and adverbs disfavor it. The constraint hierarchy presented by functional category of the word patterns across varieties of BP. The similarity of environmental constraints suggests that the mechanisms th at produce variation in the underlying grammars in both UP and BP are basically the same. As Tagliamonte (2006:241) argues, if two varieties share the same constraint rankings, it is an indication of the similarity of their grammars. Progressive forms o f verbs exert the highest effect favoring the nasalization rule. This finding provides evidence to suggest that vowel nasalization is more likely to occur in sequences such as ndo found in gerunds and progressive forms. It can be argued that UP speakers have grammaticalized the use of the gerund to other functions. However, a narrative structure examination would be necessary in order to determine the discourse and pragmatic uses of the gerund in the speech community of Rivera. In the present study, the low probability rate presented by adverbs and adjectives may be interpreted as a result of derivational suffixation, most likely by the diminutive suffix inho/a, ( e.g bonit + a diminutive suffix inha bonitiha pretty ). Cross tabulation of the data reveals that following nasal interacts with adjectives and adverbs (§5.7.3). There seems to be grammatical internal constraints guiding this pattern. This result is supported by claims in the liter ature arguing that word

PAGE 220

220 class, particularly adverbs, seems to condition the nasalized variant across languages and varieties of BP. Non progressive forms of verbs disfavor the nasalized variant in UP. The absence of nasalization of a final vowel may lead t o ambiguity, since in standard Portuguese this feature constitutes the principal marker of verbal plurality. It can be inferred that UP speakers mark the plurality in the personal pronoun or use some other mechanism to convey verbal plurality in order to avoid grammatical ambiguity. However, a more detailed account of this phenomenon would be necessary to reach conclusive arguments. Preceding phonological context conditions the nasalized variant in UP. This finding is consistent with studies examining v owel nasalization across languages. Stops favor vowel nasalization, while glides, fricatives and nasal consonants disfavor it. Liquids exert a rather neutral effect conditioning the application of the rule. Data analysis shows that a preceding nasal does not favor vowel nasalization, thus blocking its effect and the potential result of progressive nasal assimilation. The role of the nasal impeding nasalization in this linguistic context, suggests that progressive nasal assimilation is not a productive pro cess operating in contemporary UP. This finding is consistent with language universal tendencies attesting that most nasalization occurs through regressive assimilation than progressive assimilation. Guy (1981:203) discusses that throughout its history Po rtuguese has had numerous documented cases of both progressive and regressive nasal assimilation. For Lipski (1975), progressive nasal assimilation is an on going process in modern Carioca dialect. Regressive nasal assimilation is the predominant historic al process that

PAGE 221

221 accounts for most nasalized vowels and diphthongs in Portuguese (Shaw 1986:96). As noted above (§2.6.1.4), regressive nasal assimilation is the most common manifestation of vowel nasalization in BP. The Portuguese nasal vowel system has b een in almost constant flux throughout its history and it is still in flux today (Guy 1981:201). It remains to be seen how progressive nasal assimilation will be analyzed by speakers of UP in the speech community of Rivera. Concerning the external factors, age was the only social factor ranked as statistically significant; with the Generation 2 group (30 49 year old) speakers favoring the nasalized variant. Two extralinguistic factors were eliminated in the binomial step up and step down analysis of the da ta, socioeconomic status and gender. These results indicate that vowel nasalization is constrained mainly by linguistic factors and thus it is selected as significant, the strength of its effect is rather low (10). In addition, the ratios of preference among age groups do not represent substantial age stratifications. This result is an indication that vowel nasalization is not a linguistic variable undergoing change i n UP. In this chapter I have discussed quantitative analyses of vowel nasalization in Rivera. I have stated the rationale behind the hypotheses and reformulated the research questions. The linguistic and extralinguistic constraints were also presented along with distributional and multivariate analyses. In the final c hapter of this study (Chapter 6 ), I will answer the research questions and summarize the main findings found to condition the variables under investigation, that is, palalalization of den tal stops before /i/ and vowel nasalization in the speech of bilingual speakers in Rivera. I will

PAGE 222

222 discuss the results in light of the quantitative an d qualitative data discussed in Chapter 4 and Chapter 5 I will also present the limitations of the study and offer recommendations for future research. Lastly, I will formulate some general conclusions.

PAGE 223

223 CHAPTER 6 CONCLUSIONS In the final chapter of this study, I present the main results found in the analysis of the two linguistic variables under examinati on, that is, the variable realization of palatalization of dental stops before /i/ and vowel nasalization in contemporary UP in Rivera. I answer the research questions set forth in Chapter 1 and discuss the limitations of the study with regard to the two l inguistic variables under investigation. I present methodological conclusions and offer recommendations for future research. Lastly, I formulate some general conclusions. 6.1 Research Questions In this section I answer each research question, based on t he quantitative and qualitative results for each linguistic variable under examination and discuss the main findings and the limitations of the study. I begin with the first linguistic variable analyzed, palatalization of dental stops before /i/ and contin ue with vowel nasalization in UP. 6.1 .1 Palatalization The first question guiding the present study concerned the sociolinguistic stratification of the palatalization of /ti/, /di/ in UP according to the social variables of age, socioeconomic status, and g ender. I also investigated whether the change in progress hypothesized by Carvalho (1998) has continued to advance or whether it has stabilized at the speech community level. The examination of age in variationist analysis is crucial sin ce it can help t he researcher determine if speakers of different generations demonstrate similar or different patterns in the use of the linguistic variable under investigation. If so, these different rates may point toward age grading or change in progress (Bayley 1991:2 41).

PAGE 224

224 In this study I entertained both possibilities at several stages of the analysis of palatalization of dental stops. Multivariate analysis shows that the strongest social factor conditioning the variation is age. Apparent time evidence of both data sets corroborates that palatalization of dental stops is age related, since the frequencies of palatalization are strongly associated with age differences. time evidence suggest ing that palatalization of dental stops in Rivera is undergoing linguistic change. The basic principle underlying the apparent time construct is that language when other f actors, such as social class, are held constant. The speech of each generation is assumed to reflect the language as it existed at the time when that time method assumes that individual verna culars remain basically stable after the formative period of language acquisition. As Labov argues, an apparent diachrony tries to reach from the present to the future, real diachrony entails a link from the present to the past (1995:45 70). In order to s ubstantiate the change in progress hypothesized by Carvalho (1998), I compared data from two cross sectional studies conducted at two points in time, 1995 and 2007. Based on this hypothesis, an increment in palatalization rates was expected if palatalizat ion was indeed undergoing change; however, the hypothesized increase in the use of the palatal variant over time among the younger generation was not confirmed in this study. Although an increment in the use of palatalization of dental

PAGE 225

225 stops is found acr o ss age groups in apparent time, cross sectional comparisons point toward a state of relative stability at the speech community level. The first piece of evidence that leads to this conclusion comes from the comparison of overall frequencies found in the 1995 and 2007 data (Figure 4 1). In this study, the overall distribution of variants shows that 29% (N=743) of the tokens analyzed are palatal realizations, while 71% (N=1783) correspond to the dental realizations of the variant. Carvalho (1998) reports 32% (N=719) palatal realization and 68% (N=1529) non palatalization realization of the variant. The overall distribution of the variants suggests that palatalization of dental stops in the speech community of Rivera has not increased over time. Multivaria te analysis demonstrates a repeated age gradient distribution among the groups (Figure 4 4). However, a cross sectional analysis (Table 4 11) shows that the decreasing frequencies among the 1995 and 2007 Generation 1 group (15 29 years old) indicates that the latter generation has adopted the innovative variant at a slower pace than the former generation did (.91 and .70), respectively. Participants in the 2007 Generation 2 group (30 49 years old) tend to favor the innovative variant at (.62), while the 1995 data depicts a different scenario: speakers in this age bracket tend to favor the dental variant showing a low probability (.29). This generation. That is, Gen eration 2 group participants were 18 and 37 years old, respectively, twelve years earlier, which is the time depth of the study. Carvalho reports an almost categorical (.91) probability of use of the palatal variant among participants in the 1995 Generatio n 1 gr oup as against only (.29) for participants in the Generation 2

PAGE 226

226 group. The generational differences across studies indicates that the 1995 Generation 1 group started out at (.91) probability of rule application. These participants have continued fav oring the palatal variant but at a much slower pace as indicated by the decrease (.62) in the 2007 Generation 2 group. As for the last group, 2007 Generation 3 (50 70 years old), these participants were 38 and 58 years old, respectively, in 1995. The pro bability of palatalization for speakers in this age cohort in 1995 falls between Generation 2 (.29) and Generation 3 (.05). These results point out that for speakers in this age cohort, the dental variant is still the norm. The ag e distributions discussed above show that speakers of all ages participate in palatalization of dental stops in Rivera, but different age groups represent d ifferent palatalization rates. This finding indicates One of the challenges i n quantitative sociolinguistics is the scarcity of real time data. In the specific case of Rivera, the only type of real time data available are early qualitative descriptions of UP, in which the dental variant is characterized as the norm among UP speake rs. Therefore, taking these descriptions as real time evidence, it is reasonable to assert that at least in the late 1950s dental stops before /i/ in Rivera had only one linguistic variant. The time window captured in the 1995 data signaled a synchronic p attern of a vigorous language change in apparent time. However, twelve years later, cross sectional analysis shows that this vigorous linguistic change has reached a relatively stable mode at the level of the speech community. Linguistic variables u ndergo ing change can reac h a point of relative stability, but this stability can be followed by periods of considerably flux (Chambers 2002:364). In line with previous

PAGE 227

227 research, stability was also found in the speech community of Antnio Prado, Rio Grande do Sul This finding signals that UP is a paralelling southern varieties of BP in this respect Socioeconomic status proved to be a strong social constraint conditioning palatalization of dental stops. As hypothesized, mid middle class speakers tend to use the innovative variant (.76) more often than do low middle (.34) and working class speakers (.30). Speakers in the low middle and working class groups show the same tendency to use the conservative variant. This finding may indicate that palatalization of de ntal stops is moving along social spheres over time. The results of this analysis indicate that palatalization of dental stops in Rivera can be interpreted as a sociolinguistic marker since it signals social class, due to the clear break among the social g roups. Here we have an example of a linguistic variable that is stabilizing as a sociolinguistic marker in the speech community of Rivera. The last social factor examined was gender. As expected, statistical results clearly reveal that women tend to produ ce more palatalized variants (.62) than men (.38). Unsurprisingly, women proved to be advancing the process of adoption of the palatal variant in contemporary UP in Rivera Young females exhibit a considerably higher ratio of palatalization of dental stop s than do older males. The youngest group presents the biggest difference between females and males in the adoption of the innovative variant (Figure 4 7). Cross tabulation of the data shows a strong interaction between gender and social status. As seen above ( Figure 4 8), there is a clear gender and socioeconomic based pattern. Women in the highest socioeconomic group in Rivera present the highest frequency of the innovative variant, followed by females in the low

PAGE 228

228 middle and in the working class groups The social stratification in the speech of men is less dramatic. This finding conforms to the stance of sociolinguistic markers commonly found across speech communities. The second question concerned the linguistic factors conditioning palatalization of dental stops before /i/ in UP. The linguistic factors shown to have a significant effect on the distribution of the variation are: following and preceding phonological segment and tonicity of the syllable. The greatest effect is presented by the following phonological context, with a relative magnitude of its effect at 28. This finding is consistent with previous studies of UP and across dialects of BP. This study shows that palatalization of dental stops is more likely to occur when fol lowed by a vowel o r glide (.66) or a lateral (.65), or its realization as a velar glide [w] in coda position. In BP, laterals in syllable final position are usually realized as a velar glide [w], which forms a diphthong with the preceding vowel. It appears that diphthongs c reate a favorable linguistic constraint promoting palatalization of dental stops in this environment. Stops at (.62) and nasals at (.59) favor the palatalization rule, whereas a following pause or zero phonological context (.43) and sibilants (.38) disfavo r it. Following environment also confirms that /s/ promotes dental realizations of /ti/, /di/ in UP and across dialects of BP. Vowels are found at the highest position in the constraint hierarchy. As mentioned above, the present study examines palataliza tion of dental stops before /i/, where the unstressed mid vowel /e/ raised to a high vowel [i] and the palatal glide /j/ are taken into account. The use of high vowels in place of mid vowels results from a phenomenon known as vowel raising (Mattoso Cmara 1953, Bisol 1989). Vowel raising is pervasive,

PAGE 229

229 and its use is variable in colloquial Portuguese language, and is seemingly at play in UP as well, as data results suggest. Preceding phonological environment exerts the second highest magnitude of effect in the analysis, obtaining a range of 14. Results show that vowels or glides (.57) and nasals (.53) slightly favor palatalization of dental stops, while stops (.52) and pause (.51) have a neutral effect of rule application. As in previous studies, pause lacks statistical significance in this environment. Preceding liquids (.39) and sibilants (.39) disfavor palatalization of /ti/, /di/. The role of the sibilant disfavoring palatalization in this linguistic context patterns across southern varieties of BP. The position of /s/ at the bottom of the constraint hierarchy suggests that the underlying grammar producing the dental variant is basically the same in both UP and BP. Tonicity of the syllable was the weakest linguistic factor conditioning palatalization of /ti/, /di/ in UP, presenting a relative magnitude of 10. Multivariate analysis shows that post tonic syllable stress favors palatalization (.54) while stressed (.46) and pre tonic syllable stress (.44) disfavor application of the palatalization rule. T his was an unexpected result. It was hypothesized that stressed syllables would exert the highest influence on rule application. However, the constraint hierarchy found in this study that is, the post tonic, stressed and pre tonic syllable mirrors the constraint hierarchy of the The favoring pattern found is based on the 8). It appears that contact with the Spanish language, where the affricate realization of dental stops is not operative, is a contributing factor favoring a less salient environment for rule application. The

PAGE 230

230 analysis of this finding may be interpreted as a consequence of UP and Spanish language contact in the speech community of Rivera. The third question was to identify the driving forces behind the variant selection. It can be argued that the social characteristics of speakers determine the selection o f the variant. Young community members in Rivera are aware of the social capital attached to the standard language and are quite aware of the linguistic value of standard Portuguese and its potential value not only in the job market but also for education al opportunities. In this respect, participants expressed their desire to learn standard Portuguese for practical reasons, in order to become more marketable in their professional life. At the same time, members of the younger generation of speakers in Riv era, mainly from the working and low middle class, try to maintain their border identity. The transition from language homogenization to bilingual education has brought language standardization forces to the speech community of Rivera. UP speakers are co nfronted with conflicting language loyalties as language attitudes and evaluation of the standard and localized varieties are being reexamined not only by educational authorities but by community members as well. The linguistic behavior of working age spe akers requires some comment. It can be argued that these speakers are entering a professional stage in their lives and as such are more sensitive to the social advantages of the incorporation of standard features into their speech. The linguistic marketp lace concept (Sankoff and Laberge 1978) has been applied in sociolinguistic research to examine the relationship between

PAGE 231

231 Rivera dwellers are considerably aware of the stigma at tached to the variety spoken in the barrios Feelings of are common in both working and low middle class Rivera residents. In this respect, Riverans of all social classes share a community norm of linguistic evaluation : the social evaluation of Riverans even refer to UP speakers as rompe idioma the stigma attached to Rivera speech as part of their mixed identity and as a byproduct of conviviality with neighboring Brazil, mid middle class speakers attach negative evaluations to the variety and express their dissociation with UP non standard linguistic features. Women are especially sensitive to externa l hig her standards of correctness in language associated with upward social mobility (Labov 1990:214). Women in Rivera are sensitive to the social pressures imposed on them. Due to their weaker s ocial standing in the community, women tend to pay attentio n to external standards of One of the limitations of the analysis of palatalization of /ti/, /di/ was the absence of preadolescents in the sample. Labov (1999:49) advises that we must take into account data from preadolescents from 8 to 11 years old in tracing a variable though apparent time. 6.1 .2 Vowel N asalization The first question guiding the study involved identifying which internal mechanisms contribute to the variability of vowel nasalization in UP. I examined three phonetic and two non phonetic factors. Two factors, one phonetic (preceding phonological context) and one non phonetic (functional category of the word) were selected as significant to

PAGE 232

232 the probability of vowel nasalization. The greatest effect was presented by funct ional category of the word, with a relative magnitude of its effect at 55. The role of functional category of the word conditioning vowel nasalization in UP came as an unexpected result. It was hypothesized that the nasalization rule would affect vowels e qually regardless of their morphological status. As the results show, vowel nasalization is a linguistic variable, constrained by factors of morphophonological and sociophonetic nature. As seen above, phonological processes interact with affixation in in flectional and derivational morphology in the process of vowel nasalization. These morphophonological processes highly constrain vowel nasalization in UP. Verbal categories such as the progressive or gerund forms favor the nasalized variant at (.77) prob ability, while non progressive forms of the verb strongly disfavor it (.22). Nouns provide a favorable context for vowel nasalization at (.62), while a djectives and adverbs (.46) do not These results suggest that vowel nasalization in UP is conditioned by major categories of the grammar. The constraint hierarchy presented in the present study patterns similarly across varieties of BP. The similarity of environmental constraints suggests that the mechanisms that produce variation in the underlying gramma rs in both UP and BP are very similar. If two varieties share the same constraint rankings, it is an indication of the similarity of their grammars (Tagliamonte 2006:241). Progressive forms of the verb exert the highest effect on favoring the nasalization rule. This finding provides evidence to suggest that vowel nasalization is more likely to occur in sequences such as ndo found in gerunds and progressive forms. In contemporary BP, the use of a gerund has been extended to indicate a temporary state

PAGE 233

233 that would be absent if the simple present were used. It can be argued that UP speakers have grammaticalized the use of the gerund to other functions. A djectives and adverbs do not f avor the nasalized variant Previous studies have suggested a correlation b etween adjectives and derivational suffixation. Abaurre and Pagotto (1996:509) report that adverbs block the nasalization rule (.21). Cross tabulation analysis confirmed the assumption that all adverbial instances contained the suffix mente. T he low prob ability rate presented by adverbs and adjectives may be interpreted as a result of derivational suffixation, most likely by the diminutive suffix (z)inho/a Cross tabulation of the data reveals that following nasal interacts with adjectives and adverbs (§ 5.7.3). There seem to be grammatical internal constraints guiding this pattern. This result is supported by claims in the literature arguing that word class, particularly adverbs, seems to condition the nasalized variant across languages and varieties of B P. Non progressive forms of the verb disfavor the nasalized variant in UP The absence of nasalization in a final vowel may lead to ambiguity, since in standard Portuguese this feature constitutes the principal marker of verbal plurality. It can be inferr ed that UP speakers mark the plurality in the personal pronoun or use some other mechanism to convey verbal plurality in order to avoid grammatical ambiguity. Preceding phonological context exerts the second highest effect conditioning vowel nasalization in UP, obtaining a range of 48. Preceding phonological environment distinguishes between glides and consonants and classifies them by manner of articulation. Multivariate results show that stops favor vowel nasalization (.59), whereas glides (.11), frica tives (.47), and nasal consonants disfavor it (.42). Liquids exert a

PAGE 234

234 neutral effect condit to (.5). This result is consistent with Aba non na sal consonants conditioning the nasalized variant at (.5). Data analysis shows that a preceding nasal does not favor vowel nasalization blocking its effect and the potential result of progressive nasal assimilation. The role of the nasal impeding nasalizat ion in this linguistic context, suggests that progressive nasal assimilation is not a productive phonological process operating in contemporary UP. Progressive nasal assimilation is an on going development in modern Carioca BP (Lipski 1975:75). On the ot her hand, regressive nasalization is the predominant historical process that accounts for most nasalized vowels and diphthongs in Portuguese (Shaw 1986:96). Throughout its history Portuguese has had numerous documented cases of both progressive and regres sive nasal assimilation (Guy 1981:203). The Portuguese nasal vowel system has been in almost constant flux throughout its history and it is still in flux today (Guy 1981:201). It remains to be seen how progressive nasal assimilation would be analyzed by s peakers of UP in the speech community of Rivera. Preceding nasals and glides interact (i.e. overlap) with adjectives and adverbs, verbs, and non progressive forms of the verb. This is a natural result when examining naturally occurring speech. Further ex amination of the data, controlling for detailed morphological aspects, would shed more light on the interplay of the different levels of grammar involved in the variable realization of vowel nasalization in the speech community of Rivera.

PAGE 235

235 As for the two hypotheses postulated in this study, it was predicted that vowel nasalization would be more frequent in stressed syllables and that following phonological environment would constrain vowel nasalization in UP. The factor by factor analysis reveals that vowe l nasalization occurs categorically in contexts of stressed syllables (Table 5 10). The factor by factor analysis also shows that following nasals obtained a high rate of vowel nasalization (Table 5 11). Cross tabulation of the data indicates that most of the nasals in this environment are palatal and the vowels present a categorical status (i.e. they are nasal). This result is supported by claims in the liter /n/. Allophonic nasalization before palatal nasals occurs independently of the position of primary stress, just like contrastive nasalization does (Wetzels 1997:218). This finding is consistent across varieties of BP. This finding is also consistent with language universal tendencies attesting that most nasalization occurs through regressive nasal assimilation rather than progressive assimilation. The second question was wh ether external mechanisms contribute to the variability of vowel nasalization in UP. Multivariate analysis selected only the factor group age as significant. The other two social factors examined, socio economic class and gender, were eliminated in the bi nomial step up and step down analysis of the data. While data results indicate that there are no significant differences between the age groups, participants in the Generation 2 group (30 49 years old), slightly favor the nasalized variant at (.56). Part icipants in the Generation 1 group (15 29 years old) and in the Generation 3 (50 70 ye ars old) pattern similarly, disfavor ing the nasalized variant at (.46 and .48), respectively. The ratios of preference among age groups do not

PAGE 236

236 represent substantial age s tratifications. This finding is an indication that vowel nasalization is not a sociolinguistic variable undergoing change in UP. Unlike palatalization of dental stops, here the age distributions across generations do not point to any direction of change. This result is supported by previous studies examining vowel nasalization across languages (Violin 2001) and other varieties of Portuguese (Abaurre & Pagotto 1996). The fact that extralinguistic factors did not play a key role on the distribution of the v ariation is somehow an expected result. These findings indicate that vowel nasalization is constrained mainly by linguistic factors and thus it may not be considered a sociolinguistic marker or an indicator. One of the limitations of the analysis of vowel nasalization was the lack of instrumental treatments or acoustic measurements to analyze the nasal vowels. Another limitation was the treatment of the nasal vowels as a phonetic variable. As stated above, I attempted to analyze vowel nasalization from a variationist perspective; however, this analysis needs to be extended in order to further explore the interaction between phonology and morphology. My research is opening a new window on this linguistic phenomenon. 6. 2 Methodological C onclusions Concernin g the methodology, the ethnographic approach to sociolinguist inquiry used in the preliminary and data collection fieldwork proved to be crucial in the implementation of this study. Through participant observation I was able to obtain local cultural knowle knowledge was fundamental to discover what is important for community members in order to be able to use this knowledge to interpret quantitative data. The ethnographic

PAGE 237

237 approach is especial ly effective when the researcher is not a member of the community, as in my case. The network concept at the fieldwork stage was implemented in this study to enter the community via social networks. I must add that methodologically it was a sound decision to start out with the interviews in the barrios of the speech community of Rivera. The trend component implemented in the study allowed me to compare apparent time data from two studies at different points in time, 1995 and 2007. By adding a longitudina l perspective to the study, it was possible to obtain a clear picture of the sociolinguistic evolution of palatalization of dental stops in the speech community of Rivera. In this study, it has been shown that insights can be gained from both distributio nal and multivariate analyses of the data. Numerous attempts were made in order to assess the effect of some of the interacting linguistic factors in the vowel nasalization data from a multivariate analysis alone. However, the examination of each of the variable rule analyses performed in this study (Table 5 12 5 14) as well as the cross tabulation of the factors tested, revealed the location and character of the relationship between the factors conditioning vowel nasalization in UP. 6. 3 Limitations and R ecommendations for Future Research The analysis of palatalization of /ti/, /di/ and vowel nasalization in UP provides a fruitful area for future research. A possible direction would be to extend the examination of palatalization of dental stops to differen t Portuguese speaking areas to determine its variability across speech communities. Another exciting area of research would be to determine the interaction between vowel raising and palatalization of /ti/, /di/. As discussed above (Chapter 4), in most v arieties of Brazilian Portuguese, both an

PAGE 238

238 underlying high vowel /i/ in stressed or unstressed positions and a phonetic [i] derived from an underlying central vowel /e/ in unstressed positions may palatalize the preceding dental stop. On methodological gro unds, adding a panel component to the study would evaluate language change at the individual level in the speech community of Rivera As Sankoff (2006:12) claims, in most panel studies researchers have found that when a trend study signals change in progr ess, grouped d ata from the panelists indicate a modest increase in the direction of the change. In the study of [r] [R] change in Montreal French (Sanko ff & Blondeau 2007 ) the authors investigate the relationship between language change in the historic al sense, and language change at the individual level thorugh the combination of a trend and a panel study. Vowel nasalization in UP is another productive field of enquiry. One area is the examination of the potential role of lexical effects conditioning vowel nasalization in UP, which would include a word frequency and lexical analysis. Since vowel nasalization variation is part of a more abstract agreement system, an analysis of subject verb relationship including discourse or pragmatic aspects would de termine the mechanisms used by bilingual UP speakers to convey plurality and avoid ambiguity. 6.4 Final Conclusions The Portuguese language in northern Uruguay has had a long and conflictive history. However, with the establishment of Mercosur different dynamics between the Spanish and Portuguese language have been established along the Uruguayan Brazilian border. It is in this context that after a long linguistic policy struggle in the country, a bilingual pilot program was implemented in the city of Riv era in 2003. New socioeconomic and cultural initiatives have brought language standardization forces to

PAGE 239

239 the Uruguayan Brazilian border. Along with these standardization forces, language attitudes and evaluations of the standard and localized varieties are being reexamined by educational authorities and community members. The coexistence of linguistic varieties brings about conflicting linguistic loyalties among speakers in bilingual communities, particularly in minority speech communities such as Rivera. It remains to be seen what effect language standardization forces will have o n the speech community of Rivera. The compound Rivera Santana do Livramento can be thought as one socio geographic center, since people from both Rivera and Santana do Livramento cross borders frequently to carry out everyday activities without any control or immigr ant inspection. B order inhabitants who are eager to p oint out their rivalry, particularly when discussing soccer, a re at the same time quick to acknowledge their uniq ue and fraternal linkage to their neighbor ing twin city Although national borders can be thought of as a place where sharp dividing lines are drawn, the dynamics of integration, globalization and transformation on the Uruguayan Brazilian border give evid ence to the contrary. As Hamel notes constituting strips of fluid contact and developing hybrid cultures and systems of Due to complex historical, socio economic, and political factors, Spanish and Portuguese varieties have coexisted along the Uru guayan Brazilian border for cent uries. These competing varieties are part of the linguistic options available to inhabitants of the speech community of Rivera a ccording to their social and ideological characte ristics. As data result reveal social factors compel a speaker to adopt or resist

PAGE 240

240 a linguistic variant. As discussed above, palatalization of /ti/, /di/ in Rivera shows a clear pattern of social stratific ation in which the wealthiest speakers tend to prefer the incoming palatalized variant while the low middle and wor king cla ss speakers favor the local dental form. In this respect, palatalization of dental stops is considered a sociolinguistic marker sinc networks reinforce and maintain these norms (Milroy 1999:10). It can be argued that localized variants are tied to allegiance to the barrio and serve as linguistic marker of social identity.

PAGE 241

241 APPENDIX A APHIC INFORMATIO N Table A 1. d emogra phic i nformation Spk No. Pseud. Age Date of birth Age No. Gender. Soc. class Soc. C No. 1 CL003 16 1991 1 F Work C 3 2 MY012 23 1984 1 F Work C 3 3 MX023 23 1984 1 F Work C 3 4 CL004 19 1988 1 F Low Mid 2 5 CH003 21 1986 1 F Low Mid 2 6 PA005 15 1992 1 F Low Mid 2 7 KL002 15 1992 1 F Mid Midl. 1 8 SB008 15 1992 1 F Mid Midl. 1 9 DR010 15 1992 1 F Mid Midl. 1 10 PF024 19 1988 1 M Work C 3 11 TN025 16 1991 1 M Work C 3 12 AW022 18 1989 1 M Work C 3 13 WR014 16 1991 1 M Low Mid 2 14 DD026 19 1988 1 M Low Mid 2 15 JG016 19 1988 1 M Low Mid 2 16 LY021 15 1992 1 M Mid Midl. 1 17 XX107 21 1986 1 M Mid Midl. 1 18 YY055 26 1981 1 M Mid Midl. 1 19 AQ074 41 1966 2 F Work C 3 20 MD043 37 1970 2 F Work C 3 21 IC034 30 1977 2 F Work C 3 22 AA064 46 1961 2 F Low Mid 2 23 FF062 48 1959 2 F Low Mid 2 24 MS028 35 1972 2 F Low Mid 2 25 RM039 40 1966 2 F Mid Midl. 1 26 MA040 38 1969 2 F Mid Midl. 1 27 ME070 49 1958 2 F Mid Midl. 1 28 HJ049 34 1973 2 M Work C 3 29 AS046 40 1966 2 M Work C 3 30 AI082 42 1965 2 M Work C 3 31 CL051 36 1971 2 M Low Mid 2 32 HF044 35 1972 2 M Low Mid 2 33 HV050 30 1977 2 M Low Mid 2 34 PP058 33 1974 2 M Mid Midl. 1 35 ET053 31 197 6 2 M Mid Midl. 1 36 WW05 40 1967 2 M Mid Midl. 1 37 MB072 50 1957 3 F Work C 3 38 SW093 61 1946 3 F Work C 3 39 MN063 50 1957 3 F Work C 3 40 JJ061 52 1955 3 F Low Mid 2 41 GW085 62 1945 3 F Low Mid 2 42 TY087 64 1943 3 F Low Mid 2 43 MA067 54 195 3 3 F Mid Midl. 1 44 DS066 50 1957 3 F Mid Midl. 1

PAGE 242

242 Table A 1. Continued. Spk No. Pseud. Age Date of birth Age No. Gender. Soc. class Soc. C No. 45 SS088 55 1952 3 F Mid Midl. 1 47 NN081 52 1955 3 M Work C 3 48 TK083 50 1957 3 M Work C 3 49 HA099 58 1949 3 M Low Mid 2 50 TS097 57 1950 3 M Low Mid 2 51 DV077 54 1953 3 M Low Mid 2 52 VV103 60 1947 3 M Mid Midl. 1 53 FW101 70 1937 3 M Mid Midl. 1 54 AE080 57 1950 3 M Mid Midl. 1

PAGE 243

243 APPENDIX B MODULES FOR THE SOCI OLINGUISTIC INTERVIE WS IN RIVERA The following topics were chosen because they can be adapted to any speech community. For the speech community I am interested in, football is a very popular Fronterizo/Uruguayan Portuguese (UP) speaking people and Brazilians. Q GEN Module 1: Football 1. Have you ever watched a football match at the stadium? 1.1 When was that? 1.2 Who was playing? 1.3 How did you describe the game? 1.4 How often do you go watch a game/football? 2. Who do you think play the best football in the region? 2.1 Why do you think that? 2.2 Do you think Brazilians are good at football? 3. How often did you play football when you were a c hild? 3.1 Have you ever had any fights when playing football/a game? 3.2 What happened? 3.3 Were you scared? Q GEN Module II: Television People in the border watches Brazilian TV and according to Carvalho (1998) TV serves as a linguistic model for groups who are seeking one (mostly teenagers). Many times people would stop what they are doing to watch a novela da tarde (soap opera). 1. What do you think about television? 1.1 Do you watch TV a lot? 2. What are your favorite shows/channels? 2 .1 Why do you prefer them? 3. Do you watch Brazilian TV? 3.1 Do you think is better than Uruguayan TV? 3.2 Why? Q G EN Module III: Language Choice

PAGE 244

244 This is a very sensitive topic since people in R ivera often think of Uruguayan Portuguese (UP) i 1. So where were you born? Do you consider yourself Uruguayan, Brazilian, or Fronterizo ? Why? 2. Who do you speak UP with? Do your parents punish you for that? 3. Do you prefer to speak Spanish or Portuguese? Why? Do y ou think one language is more important/useful than the other? Can you explain?

PAGE 245

245 APPENDIX C IRB APPROVAL OF PROTOCOL

PAGE 246

246 APPENDIX D IRB RENEWAL OF PROTOCOL

PAGE 247

247 LIST OF REFERENCES Abaurre Maria Bernadette, and Emilio Pagotto. 1996. Nasalizao do Portugus d o Brasil. Gramtica do Portugus Falado 6. 495 522. Academia Nacional de Letras Comisin para el Estudio del Espaol en la Zona Fronteriza 1982. Estudio sobre el problema idiomtico fr onterizo Montevideo: La Comisin. Almeida, Antnio. 1976. The Portuguese nasal vowels: Phonetics and phonemics. Readings in Portuguese linguistics, ed by J. Schmidt Radefelt, 349 96. New York: North Holland. Azevedo, Milton. 1981. A contrastive phonol ogy of Portuguese and English. Washington, D.C.: Georgetown University Press. Azevedo, Milton. 2005. Portuguese: A linguistic introduction Cambridge: Cambridge University Press. Barrios, Graciela. 1995. Planificacin lingstica y Mercosur: El caso ur uguayo. Associao de Universidades Grupo Montevideo Anais do Encontro sobre Polticas Lingsticas 41 49. Curitiba: Universidade Federal do Paran. Barrios, Graciela. 1996. Planificacin lingstica e integracin regional: El Uruguay y la zona de front era. Fronteiras, educao e integrao ed by Almeda Menice Trindade and Luis Alberto Behares, 83 110. Santa Maria: Pallotti. Barrios, Graciela. 1999. Minoras lingsticas e integracin regional: La regin fronteriza uruguaya brasilea. Polticas ling sticas para Amrica Latina. Actas del Congreso Internacional 1997, 85 92. Buenos Aires: UBA. Barrios, Graciela; Beatriz Gabbiani; Luis Ernesto Behares; Adolfo Elizaincin; and Susana Mazzolini.1993. Planificacin y polticas lingsticas en Uruguay. Polt icas del lenguaje en Amrica Latina ed. by Rainer Enrique Hamel, 177 90. Mexico: UAM. Bakewell, P.J. 2004. A history of Latin America: 1450 to the present Malden, MA: Blackwell. Battisti, Elisa, and Ben Hermans. 2009. Fixed and variable properties of the palatalization of dental stops in Brazilian Portuguese in an immigrant community. Phonetics and phonology: Interactions and interrelations ed. by Marina Vigrio, Sonia Frota and J. Freitas, 235 46. Amsterdam: John Benjamins. Bayley, Guy. 2002. Real and apparent time. In Chambers et al. 2002, 312 32.

PAGE 248

248 Bayley, Guy; Tom Wikle; Jan Tillery; and Lori Sand. 1991. The apparent time construct. Language Variation and Change 3.241 64 Bayley, Robert, and Denis R. Preston (eds.) 1996. Second language acquisiti on and linguistic variation Amsterdam: John Benjamins. Behares, Luis Ernesto. 1984. Planificacin lingstica y educacin en la frontera uruguaya con el Brazil. Montevideo: Instituto Interamericano del Nio. Behares, Luis Ernesto; Carlos Ernesto Diaz; and Gerardo Halzmann. 2004. Na frontera nos fizemo assim Montevideo : Universidad de la Repblica. Behares, Luis Ernesto, and Graciela Gabbiani. 1987. Educacin y lengua en la frontera. Relaciones 36. 6 10. Bhat, D.N.S. 1978. A general study of palatali zation. Universals of human language ed. by Joseph Greenberg, 47 92. Stanford : Stanford University. Bisol, Leda. 1989. Vowel harmony: A variable rule in Brazilian Portuguese. Language Variation and Change 2.185 98. Bisol, Leda. 1991. Palatalization and its variable restriction. International Journal of the Sociology of Language 89.107 24. Blom, J., and John Gumperz. 1972. Social meaning in linguistic structures: Code switching in northern Norway Directions in sociolinguistics, ed. by John Gumperz and D. Hymes, 407 34. New York: Holt, Rinehart, and Winston. Bortoni Ricardo, Stella Maris. 1985. The urbanization of rural dialect speakers: a sociolinguistic study in Brazil. Cambridge: Cambridge University Press. Boissevain, J., and J. Mitchell. 1973. N etwork analysis: Studies in human interaction The Hague: Mouton. Brasington, R.W.P. 1971. Noun pluralization in Brazilian Portuguese. Journal of Linguistics 7.151 177. Cabrelli Amaro, Jennifer, and Jason Rothman. 2010. On L3 acquisition and phonologic al permeability: A new test case for debates on the me ntal representation of non native phonological systems. International Review of Applied Linguistics in Language Teaching 48. 275 96. Carvalho, Ana Maria. 1998. The social distribution of Uruguayan Por tuguese in a bilingual border town Berkeley: University of California dissertation.

PAGE 249

249 Carvalho, Ana Maria. 2003. Rumo a uma definio do portugus uruguaio. Revista Internacional de Lingstica Iberoamericana 1.125 49. Carvalho, Ana Maria. 2004. I sp eak like the guys on TV: Palatalization and the urbanization of Uruguayan Portuguese. Language Variation and Change 16.127 51. Cedergren, Henrietta. 1973. The interplay of social and linguistic factors in Panama Ithaca, NY: Cornell University dissertatio n. Cedergren, Henrietta. 1987. The spread of language change: Verifying inferences of linguistic diffusion. Language spread and language policy: Issues implications and case studies ed by Peter Lowenberg, 44 60. Washington, DC: Georgetown University Pre ss. Chambers, J.K. 2003. Sociolinguistic theory: Linguistic variation and its social significance. Malden and Oxford: Blackwell Publishers. Chambers, J.K.; Peter Trudgill; and Natalie Schilling Estes (eds.) 2002. The handbook of language variation and change. Oxford: Blackwell. Church, Meredith. 2007. Portuol and border identity: Reconciling identities and claiming space in the national imaginary Lawrence: University of Kansas MA thesis. Douglas, Kendra L. 2004. Uruguayan Portuguese in Artigas: T ri dimensionality of transitional local varieties in contact with Spanish and Portuguese standards Madison: University of Wisconsin dissertation. Eckert, Penlope. 1997. Gender and sociolinguistic variation. Readings in language and gender ed. by Je nnifer Coates, 64 76. Oxford: Blackwell. Eckert, Penlope. 1989. Jocks and burnouts: Social categories and identity in the high school. New York: Teachers College Press. Eckert, Penlope. 2000. Linguistic variation as social practice: The linguistic construction of identity in Belten High. Malden: Blackwell. Elizaincin, Adolfo. 1976. The emergence of bilingual dialects on the Brazilian Uruguayan border. International Journal of the Sociology of Language 9.122 34. Elizaincin, Adolfo. 1979. Algunas p recisiones acerca de los dialectos portugueses en el Uruguay. Foro literario 3.50 1. Elizaincin, Adolfo (ed.) 1981. Estudios sobre el espaol del Uruguay Montevideo: Universidad de la Repblica.

PAGE 250

250 Elizaincin, Adolfo. 1992. Dialectos en contacto: Espao l y portugus en Espaa y Amrica Montevideo: Arca. Elizaincin, Adolfo. 1995. La interpretacin en la lingstica histrica: La banda oriental del siglo XVIII. Cuadernos Americanos 4.213 21. Elizaincin, Adolfo. 1996. La sociolingstica en Argentina, Uruguay y Paraguay. Internacional Journal of the Sociology of Language 117.1 9. Elizaincin, Adolfo. 2004. Las fronteras del espaol con el portugus en Amrica. Revista Internacional de Lingstica Iberoamericana 2.105 118. Elizaincin, Adolfo; Luis Behar es; and Graciela Barrios. 1987. Nos falemos brasileiro: Dialectos Portugueses del Uruguay. Montevideo: Amesur. Elizaincin, Adolfo; Marisa Malcuori; and Virginia Bertolotti. 1997. El espaol en la banda oriental en el siglo XVIII Montevideo: Universidad de la Repblica. Feagin, Crawford. 1979. Variation and change in Alabama English: A sociolinguistic study of the whi te community. Washington: Georgetown University Press. Feagin, Crawford. 2002. Entering the community: Fieldwork. In Chambers et al. 2002, 20 39. Gal, Susan. 1979. Language shift: social determinants of linguistic change in bilingual Austria New York: Academic Press. Garcia, Etchegoyen de Lorenzo, Mara Elosa. 1975. Dialecto fronterizo: Un desafo a la educacin M ontevideo: Instituto Americano del Nio. Giangola, James. P. 2001. The pronunciation of Brazilian Portuguese Muenchen: Lincom Europa. Gumperz, John, and D. Hymes. 1972. Directions in sociolinguistics New York: Holt Rinehart and Winston. Gumperz, J ohn. 1982. Language and social identity Cambridge: Cambridge University Press. Gumperz, John, and Stephen C. Levinson. 1996. Rethinking linguistic relativity Cambridge: Cambridge University Press. Guy, Gregory. 1981. Linguistic variation in Brazilian Portuguese: Aspects of the phonology, syntax, and language history Pennsylvania: University of Pennsylvania dissertation.

PAGE 251

251 Guy, Gregory. 1991. Explanation in variable phonology: An exponential model of morphological constraints. Language Variation and Ch ange 3.1 22. Guy, Gregory. 1997. Competence, performance, and the generative grammar of variation. Variation, change, and phonological theory ed. by Frans Hinskens, Roeland van Hout, and W. Leo Wetzels, 125 44. Amsterdam John Benjamins. Hall, R.A. Jr. 1943. The unit phonemes of Brazilian Portuguese. Studies in Linguistics 1.1 15. Hamel, Rainer Enrique. 2003. Regional blocs as a barrier against English hegemony? The language policy of Mercosur in South America. Languages in a globalizing world, ed. b y Jacques Mauris and Michael A. Morris,111 141. Cambridge: Cambridge University Press. Hammarstrm, G. 1954. Review of Sten 1944. Revista do Laboratorio de Fontica Experimental 2.158 62. Hammarstrm, G. 1962. Review of Carvalho 1958. Romance Philolo gy 15.350 6. Hansen, Anita. 2001. Lexical diffusion as a factor of phonetic change: The case of Modern French nasal vowels. Language Variation and Change 13.209 52. Hensey, Frederick. 1972. The sociolinguistics of the Brazilian Uruguayan border The Hag ue: Mouton. Hensey, Frederick.1980. El fronterizo del norte del Uruguay: Interlingua e interlecto. Foro Literario 4.54 9. Hensey, Frederick.1982. Uruguayan fronterizo : A Linguistic sampler. Word 33.193 97. Holmquist, Jonathan C. 2001. Variacin vocl ica en el habla masculina de Castaer, Puerto Rico. Cuaderno Internacional de Estudios Hispnicos y Lingsticos 1.96 103. Hualde, Jos Ignacio. 1989. Autosegmental and metrical spreading in the vowel harmony systems of northwestern Spain. Studies in R omance Linguistics 27.773 805. Hymes, Dell H. 1974. Studies in the history of linguistics: Traditions and paradigms Bloomington: Indiana University Press. Kerswill, Paul. 1993. Rural dialect speakers in an urban speech community: the role of dialect co ntact in defining a sociolinguistic concept. International Journal of Applied Linguistics 3.33 56. Labov, William. 1963. The social motivation of a sound change. Word 19.273 309.

PAGE 252

25 2 Labov, William. 1966. The social stratification of English in New York Cit y Washington, D.C: Center for Applied Linguistics. Labov, William. 1969. Contraction, deletion and inherent variability of the English copula. Language 45.715 62. Labov, William. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Pr ess. Labov, William. 1981. What can be inferred about change in progress from synchronic descriptions? Variation Omnibus, ed. by David Sankoff and Henrietta Cedergren, 177 199. Edmonton, Alberta: Linguistic Research. Labov, William.1982. Building on empir ical foundations. Perspectives in historical linguistics, ed. by Winfred P. Lehmann and Yakow Malkiel,17 92. Amsterdam and Philadelphia: John Benjamins. Labov, William.1990. The interaction of sex and social class in the course of linguistic change. Lang uage Variation and Change 2 205 254. Labov, William. 1994 P rinciples of linguistic change v olume I : I nternal factors London, New York: Blackwell. Labov, William, and Wendel A. Harris. 1986. De facto segregation of black and white vernaculars. Diversi ty and diacrony ed. by D. Sankoff,1 24. Philadelphia: John Benjamins. Ladefoged, Peter. 2001. Vowels and consonants: An introduction to the sounds of languages. Malden, Massachusetts: Blackwell. Lippi Green, Rosina L. 1989. Social network integration and language change in progress in a rural alpine village. Language in Society 18.213 34. Lipski, John. 1975. Brazilia n Portuguese vowel nasalization: Secondary aspects. Canadian Journal of Linguistics 20.59 77. Lipski, John. 1994. Latin American Spanish London: Longman. Ldtke, H. 1953. Fonemtica Portuguesa. Boletim de Filologia 13.273 88. Maiden, M. 1991. Interacti ve morphonology. Metaphony in Italy London: Routledge. Mateus, Maria Helena, and Ernesto d'Andrade. 2000. The phonology of Portuguese Oxford: University Press. Mattoso Cmara, Joaquim.1953. Para o estudo da fonmica Portugusa. Rio de Janeiro: Editora Simes.

PAGE 253

253 Mattoso Cmara, Joaquim.1970. Estrutura da lngua Portugusa Petrpolis: Editra Vozes. Mattoso Cmara, Joaquim.1972. The Portuguese language Chicago: University of Chicago Press. Milroy, Lesley, and Matthew J. Gordon. 2003. Sociolinguistics: Method and interpretation. Malden, MA: Blackwell. Milroy, James, and Lesley Milroy. 1978. Belfast: Change and variation in an urban vernacular. Sociolinguistic patterns in British English, ed. by Peter Trudgill, 19 36. Lond on: Edward Arnold. Milroy, James, and Lesley Milroy. 1985. Linguistic change, social network and speaker innovation. Journal of Linguistics 21.339 84. Milroy, James, and Lesley Milroy. 1992. Social network and social class: Toward an integrated sociolin guistic model. Language in Society 21.1 26. Milroy, Lesley. 1980. Language and social networks Oxford: Basil Blackwell. Milroy, Lesley. 1987. Observing and analyzing natural language: A critical account of sociolinguistic method. Oxford: Basil Blackwel l. Mitchell, J. Clyde. 1969. Social networks in urban situations. Manchester: Manchester University Press. Mitchell, J. Clyde. 1986. Network procedures. The quality of urban life ed by D. Frick, 73 92. Berlin: de Gruyter. Morais Barbosa, Jorge.1962. Les voyelles nasales portugaises: interprtation phonologique. Proceedings of the 4th International Congress of Phonetic Sciences, 691 708. The Hague: Mouton Patrick, Peter. 2002. The speech community. In Chambers et al. 2002, 573 97 Paolillo, John C 2002. Analyzing linguistic variation: Statistical models and methods. Stanford: Center for the Study of Language an d Information. Poplack, Shana. 1979. Deletion and disambiguation in Puerto Rican Spanish. Language 56.371 85 Qucoli, A. C. 1990. Harmony, lowering and nasalization in Brazilian Portuguese. Lingua 80.295 331.

PAGE 254

254 Rickford, John R. 1987. Dimensions of a creole continuum: History, texts and linguistic analysis of Guyanese Creole Stanford: Stanford University Press Reed D.W. and Yolanda Leite. 1947. The segmental phonemes of Brazilian Portuguese: Standard Paulista dialect. Phonemics ed. by Kenneth Pike, 190 202. Ann Arbor: University of Michigan Press. Rona, Jos Pedro. 1958. Aspectos metodolgicos de la dialectologa hispanoamericana. Montevideo: Universidad de la Republica. Rona, Jos Pedro. 1965. El dialecto fronterizo del nort e del Uruguay Montevideo: Universidad de la Repblica. Saciuk, Bodan. 1970. Some basic rules of Portuguese phonology Studies Presented to Robert B. Lees by his Students Papers in Linguistics, Monograph Series I, ed. by Jerrold M. Saddok and Anthony L. Vanek, 197 222. Edmonton: Linguistic Research. Sampson, Rodney. 1999. Nasal vowel evolution in Romance New York: Oxford University Press. Sankoff, David. 1978. Linguistic variation: Models and methods New York: Academic Press. Sankoff, Gillian. 2001 Linguistic Outcomes of Language Contact In Trudgill et al. 2001, 638 48. Sankoff, Gillian. 2006. Age: Apparent time and real time. Encyclopedia of Language and Linguistics ed by A. Anderson et al., 110 16. Oxford : Elsevier. Sankoff, Gillian, and H lne Blondeau 2007. Language change across the lifespan: /r/ in Montreal French. Language 83.560 88. Sankoff David, and Henrietta Cedergren. 1976. The dimensionality of grammatical variation. Language 52.63 78. Sankoff, David, and Suzanne Laberge. 1978 The linguistic market and the statistical explanation of variability. Linguistic variation: models and methods ed. by David Sankoff, 239 50. New York: Academic Press. Sankoff, David; Sali A. Tagliamonte; and Eric Smith. 2005. GoldVarb X. Toronto: Dep artment of Linguistics, University of Toronto. Online: http://individual.utoronto.ca/tagliamonte/GoldVarb/GV_index.htm Schilling Estes, Natalie. 2002. Investigating stylisti c variation. In Chambers, et al. 2002, 375 401.

PAGE 255

255 Schooling, S.J. 1990. Language maintenance in Melanesia: Sociolinguistics and social networks in New Caledonia Summer Institute of Linguistics and the University of Texas at Arlington Publications in Lingui stics, 91. Shaw, Ines Senna. 1986. Vowel nasality in Brazilian Portuguese: An experimental approach with focus on derivational and inflectional alternations Lawrence: University of Kansas dissertation. Tagliamonte, Sally. 2006. Analyzing sociolinguis tic variation Cambridge, UK; New York: Cambridge University Press. Tagliamonte, Sally. 2002. Comparative sociolinguistics. In Chambers et al. 2002, 729 63. Trudgill, Peter. 1974. The social differentiation of English in Norwich. Cambridge: Cambridge University Press. Vandresen, Paulino. 1975. O vocalismo Portugus: Implicaes tericas. Revista Brasileira de Lingstica 2.80 103. Violin, Anne. 2001. Variation in southeastern French nasal vowels and optimality theory Indiana: Purdue University diss ertation. Waltermire, Mark. 2006. Social and linguistic correlates of Spanish Portuguese bilingualism on the Uruguayan Brazilian border Albuquerque: University of New Mexico dissertation. Weinreich, Uriel; William Labov; and Marvin Herzog. 1968. Empir ical foundations for a theory of language change Austin: University of Texas Press. Wetzels, Leo. 1997. The lexical representation of nasality in Brazilian Portuguese Prob us 9.202 32. Wolfram, Walt. 1969. A sociolinguistic description of Detroit negro speech Washington DC: Center for Applied Linguistics. Wolfram, Walt. 1993. Identifying and interpreting variables. American dialect research ed. by Dennis Preston, 193 221. Amsterdam and Philadelphia: John Benjamins. Wolfram, Walt; Kirk Hazen; and Natalie Schilling Estes. 1999. D ialect change and maintenance on the outer banks. Tuscaloosa : University of Alabama Press. Young, Robert, and Robert Bayley. 1996. VARBRUL analysis for second language acquisition research. Second language acquisition and linguistic variation, ed. by Robert Bayley and Dennis R. Preston, Amsterdam: John Benjamins.

PAGE 256

256 BIOGRAPHICAL SKETCH Rosa Mara Castaeda Molla was born in Lima, Peru. She received a B A in education from the Women University of the Sacred Heart in Lima. S he taught high school history, b efore moving to London, where she volunteered at the World Association of Girl Guides and Girl Scouts In London she taught at Chelsea and Westminster Hospital School. She received a n MA in a pplied l inguistics from Florida Atlantic University In 201, sh e received a Ph.D. from the University of Florida. Rosa Mara holds a tenure track position at Fort Hays State University.