P-Forms in Distributed Morphology


Material Information

P-Forms in Distributed Morphology Accounting for a Type of Semilexical Form
Physical Description:
1 online resource (85 p.)
Deacon,Robert Joel
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Master's ( M.A.)
Degree Grantor:
University of Florida
Degree Disciplines:
Committee Chair:
Henderson, Brent Mykel
Committee Members:
Haddad, Youssef A.


Subjects / Keywords:
distributedmorphology -- dm -- insertion -- insertionfilter -- littlep -- particles -- phrasal -- phrasalverbs -- prepositions -- robertdeacon
Linguistics -- Dissertations, Academic -- UF
Linguistics thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


The grammatical concept of the preposition and particle in English is both poorly defined traditionally and within Distributed Morphology, DM. This paper will show how the behavior of this historically troublesome concept, its distribution and meaning, can only be partially explained within the advancing system of DM. Thus, as an advancement to DM, this paper will propose the addition of a little p_ head to n_, v_, and a_ as a potential functional head capable of licensing ?Roots or adjoining to ?Root phrases. It will also propose that there exists (a) restriction(s) between functional heads and the licensing of certain l-morphemes. This will be viewed as a two staged filter that is capable of prohibiting and discouraging certain ?Roots from being categorized by certain functional heads, although all ?Roots by nature are a-categorical. Finally some thoughts on P-forms as particles will be discussed.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
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 Robert Joel Deacon.
Thesis (M.A.)--University of Florida, 2011.
Adviser: Henderson, Brent Mykel.

Record Information

Source Institution:
Rights Management:
Applicable rights reserved.
lcc - LD1780 2011
System ID:

This item is only available as the following downloads:

Full Text




2 2011 Robert Joel Deacon


3 To better d ays and s pirited d ebate


4 ACKNOWLEDGMENTS There are many who warrant recognition beyond the scope of this small work. However, my mother, Kathrin Diane Deacon, who devoted so much of her time to instruct me when I was young deserves more than a small remark. Despite the perple xity I expect to see from her half way through Chapter 1, I might never have arrived at such a topic if it were not f or those early debates on the nature of words. It is really inconceivable in retrospect. Also I thank my father, Robert Richard Deacon, for his support in getting me into institutions of higher learning. Moreover, I must thank my advisors Dr. Brent Henders on and Dr. Youssef Haddad. Not only would this never have happened, it would have never reached the level of analysis it rests at now without your patience and guidance. Of course infelicity, opaqueness, shortcomings, and analytic failure in general rest e ntirely upon the feigned humility of the writer himself


5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS AND TERMS ................................ ................................ ....... 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 WHAT ARE P FORMS ................................ ................................ ........................... 11 Catego rizing P ................................ ................................ ................................ ........ 12 Defining Prepositions ................................ ................................ ........................ 15 Defining Particles and Defining P forms ................................ ........................... 17 Distributed Morphology ................................ ................................ ........................... 21 The Parts of the Old Lexicon Redistributed ................................ ...................... 22 Core Components of Distributed Morphology ................................ ................... 23 P forms in Distributed Morphology ................................ ................................ .......... 25 P and the Building Blocks of Distributed Morphology ................................ ....... 25 Basic Properties of F and L Morphemes ................................ .......................... 26 Concluding Remar ks ................................ ................................ ............................... 26 2 THE ORIGIN OF P FORMS ................................ ................................ ................... 27 L or F Morphemes ................................ ................................ ................................ ... 27 P Forms as F Mor phemes ................................ ................................ ................ 28 Particles as F Morphemes ................................ ................................ ................ 32 P Forms as L Morphemes: ................................ ................................ ............... 34 The Rean alysis of P Forms in DM ................................ ................................ .......... 37 The Homonym Explanation Denied ................................ ................................ .. 38 Categorization by Little p_ ................................ ................................ ...................... 41 Concluding Remarks ................................ ................................ ............................... 43 3 THE CONTENT OF ROOTS, VOCABULARY INSERTION AND CATEGORIZATION RESTRICTIONS ................................ ................................ .... 44 Need for Restriction ................................ ................................ ................................ 45 L Morphemes and Vocabulary Insertion ................................ ........................... 49 Root Based Restriction ................................ ................................ ................... 50 Possible External Restrictions on Roots ................................ ........................... 52 Categorizing Heads and their Features ................................ ............................ 52


6 L morphemes as F morphemes in Other Languages ................................ ............. 54 Historical Derivation of English Prepositions ................................ .................... 55 Concluding Remarks ................................ ................................ ............................... 57 4 GRAMMATICAL RESTRICTIONS THE ENCYCLOPEDIA AND SEMANTIC RESTRICTIONS IN PRACTICE ................................ ................................ ............. 58 Methods for Restriction ................................ ................................ ........................... 58 Strict or Grammatical Restrictions on Roots ................................ ................... 59 VI Insertion into a Root Dominated by Little p_ ................................ .............. 59 Insertion of Forms Directly under Little p_ ................................ ........................ 62 The Encyclopedia ................................ ................................ ............................. 64 Semantic Checks on Roots ................................ ................................ ............ 66 Licens ing Restrictions on the VI, not the L morpheme ................................ ..... 67 Concluding Remarks ................................ ................................ ............................... 69 5 P FORMS AS PARTICLES AND FINAL REMARKS ................................ .............. 71 Accounting for Particles in the System ................................ ................................ .... 71 P forms in the vP Shell ................................ ................................ ..................... 72 Replacing the Particle Phrase: ................................ ................................ ......... 75 Changes to the Base Structure: ................................ ................................ ....... 76 Movement ................................ ................................ ................................ ......... 77 Final Remarks ................................ ................................ ................................ ......... 78 LIST OF REFERENCES ................................ ................................ ............................... 81 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 85


7 LIST OF TABLES Table page 1 1 Prepositional f orms ................................ ................................ ............................. 16 1 2 Particle f orms in English. ................................ ................................ .................... 19 2 1 Possible ad positional case f eatures in English. ................................ ................. 29 2 2 Phonological forms that only serve to mark prepositional c ase. ......................... 39


8 LIST OF FIGURES Figure page 2 1 Categorization by l ittle p_ ................................ ................................ ................... 42 4 1 Little p_ as relational functor. ................................ ................................ .............. 60 4 2 VI insertions for f morphemes: ................................ ................................ ............ 62 4 3 oot filled: ................................ ................................ .............. 63 5 1 Particle verb base s tructure. ................................ ................................ ............... 73 5 2 Particle appearing after verbal o bject. ................................ ................................ 73 5 3 Particle before the DP object. ................................ ................................ ............. 74 5 4 The base structure for vP s hells with a r oot. ..... 76 5 5 First movements ................................ ................................ ............................... 77 5 6 Remnant m ovement. ................................ ................................ .......................... 78


9 LIST OF ABBREVIATIONS AND TERMS DP A determiner phrase, a phrase with a determiner in the specifier position F Morpheme A functional morpheme: a bundle of morphosyntactic features that only later gets realized phonologically by a competing VI. L Morpheme Phonology is inserted based on the pre specific morphosyntactic features. L node P form The phonological forms that commonly represent both English prepositions and particles. This thesis argues they are in the l morpheme class. p_ Features [+, case] [+relational] A place holder for what we traditionally call lexical items. UG Universal Grammar: A belief that all languages work from a set of innate biological p rinciples. VI Vocabulary Item: Phonological strings used to realize f morphemes and l morphemes. vP Verb phrase with a little v_ functional head that assigns a theta role to the external agent.


10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts P FORMS IN DISTRIBUTED MORPHOLOGY ACCOUNTING FOR A TYPE OF SEMILEXICAL FORM By Robert Joel Deacon August 2011 Chair: Brent Henderson Major : Linguistics T he grammatical concept of the preposition and particle in English is both poorly defined traditionally and within Distributed Morphology DM This paper will show how the behavior of this historically troublesome concept, its distribu tion and meaning, can only be partially explained within the advancing system of DM. Thus as an advancement to DM, this paper will propose the addition of a little p_ head to n_, v_, and a_ as a potential functional head capable of licensing Roots or adj oining to Root phrases It will also propose that there exists (a) restriction(s) between functional heads and the licensing of certain l morphemes This will be viewed as a two staged filter that is capable of prohibiting oots from being categorized by certain oots by nature are a categorical. Finally some thoughts on P forms as particles will be discussed.


11 CHAPTER 1 WHAT ARE P FORMS This thesis attempts to refine the description of prepositions and particles in English using the Theory of Distributed Morphology. Moreover it shows how the behavior of p repositions also challenges the framework of Distributed Morphology. Chapter 1 is organized as follows: the i ntroduction shows that prepositions are a problem fo r t raditional categorization, t he following heading gives a simple syntactic distributional def inition for prepositions, and a conservative list of some phonological forms available for that distribu tion. In the next section particles are defined by their syntactic behavior in contrast to prepositions Moreover, in this section the number of forms used as both prepositions and particles is highlighted and the term P form is defined. After that Distributed Morphology is introduced. A reason for using the framework is given and the nature o f the system is explained. The following heading then explain s the consequences of DM and the section after lists assumptions. The next section introduces the idea of P for ms in DM and asks if they can easily be id entified in the system. Then C hapter 1 explains that within DM there are only two choices for ident ifying a morpheme. Finally it briefly argues that this appears to be a problem for DM, considering the behavior of P forms. The rest of the thesis is as follows: Chapter 2 highlights the behavior of prep ositions and particles, shows how this is a categorization problem for DM, and presents a Little p_ functional head as a possible solution. Chapter 3 examines possible controls on Vocabulary Insertion within DM and determines that controls placed external to the Root are best argued for. Chapter 4 proposes a Grammatical filter to prevent unwanted phonological strings from entering Roots dominated by a little p_ functional


12 head. It also talks about how the Encyclopedia could be responsible for further disc rimination within DM. Chapter 5 evaluates particles syntactically to account for the problems, changes, and solutions presented in the other four Chapters. It concludes the thesis by highlighting some further issues to explore and by acknowledging problem s with the present analysis. Categorizing P In English prepositions and their often homonym counterparts, particles, have consistently created problems for grammarians and linguistic theories. The nature of these forms has been the topic of recent discuss ion in work coming from different and oft en conflicting perspectives ( Botwinik Rotem 2004, Elenbaas 2007, Svenonius 2007 and 2008 and Keizer 2008 ). Often discussion stems from the fact that these forms cannot easily be categorized. This lack universal grammar is, or ought to be, the status of the basic distinctions recognized by is important to identify the basic particles from which a theory of grammar can be built and it is troubling when a piece does not cleanly fit. substantive categories, which we will functional (1995: 6). This functional category often includes prepositions. Selkirk states that n ouns, verbs and adjectives constitute the class of lexical categories in English, while determiners, prepositions, auxiliaries, modals, complementizers, conjunctions and other sorts of particles fall into the class of functional categories Howev er, as den Dikken points out, Koopman (2000)


13 category P has functional projections like the other lexical categories and thus has more structure than previously thought. Svenonius (2008) expands upon the structure of P, but it is unclear if the forms under P are lexical or functional in nature. According to of a split word class is Preposition. Some prepositions -for example, of, at, in, to, and by all qualify in have regular uses in which t a meaning and use account of prepositions). On the other hand, Celce Murcia and Larson Freeman reverberate Taylor (1993) by emphasiz and thus rich in possible meaning (1999: 404). Given this disagreement over the theoretical category, structure, and meaning of prepositions, prepositions form an interesting linguistic notion. If prepositions are not so easily partitioned, then for traditional classification (Hudson 2000:17). In short prepositions confuse some of the more common functional lexical dist inctions, and this can be seen when considering the following three notions. The Closed versus Open Class Distinction: p repositions form a relatively closed group but the category has allowed some new forms to arrive as loanwords: via qua pro circa v is a vis per and save (Hudson 2000: 19). Furthermore, according to


14 Kortmann and (1992: 671). All in all, a ccepting loanwords is not co mmon wi th functional categories and prepositions break the rule. Number of Members Distinction: f unctional classes have small memberships compared to the plethora of words found in the substantive classes. Prepositions on the word 1 However, this number is not fully agreed upon as the number varies depending on your source. For example Bennett (1975: 1) lists only 37 2 while Lindstromberg (1998: 300 306) accounts for over 80 possible candidates. This kind of membership differential is not present with other functional categories such as determiners 3 and conjunctions whose number of members is more clearly established. Semantic Variation: f unctional words tend to have a smaller range of meani ng and are not often crucial in obtaining figurative meaning. However, p repositions can show variable, metaphoric meaning as can be seen with the following example. (1) John survived the whole day on just one sandwich. By means of an extension of the basic meaning of on to make surface contact or to support, one could picture John atop of a sandwich, which is metaphorically supporting him through the day. This picture is shown in Celce Murcia and Lar sen Freeman (1999: during and after which have just as 1 This is not the total number of forms used in this paper. 2 His list may not be intended to be exhaustive 3 It is possible that the number of Determiners in English is also debatable as [ ] is often proposed to be the determiner for Bare P lural NPs (Carlson 1977)


15 ( meanwhile afterwards ts do not grammatical functions. A relevant question at this point is if prepositional forms compose a natural class in English since they are so troublesome to cla ssify according to a lexical functional distinction The verdict to this question continues to be positive (Elenbaas 2007 and Svenonius 2007 ), despite different conclusions over their putative functional and lexical placement. Prepositions assumingly have a consistent syntactic distribution. Defining Prepositions Acknowledging the larger difficulty in classifying prepositional forms, pre positions can be classified as only those words that get positioned syntactically in a certain location. Thus they can b e defined through syntactic distribution as words that occupy the head position of a prepositional phrase, as noted by [ P ] in the template provided here : [P [D [N]]] preposition with embedded determiner and object noun This position requires the introduction of a new argument [N]. With this simple definition in use, problems with meaning may be at least temporarily suspended, and prepositions will be classified as all words that theoretically appear in this place. This t ype of definition is more compliant with the Distributed Morphology framework discussed later in this chapter. Moreover, s ome forms that can be found in this position are shown on Table 1 1


16 Table 1 1. Prepositional forms The forms appearing in Table 1 1 were mostly taken from prepositions listed in Moreover, t his table is conservative, as one can find prepositional lists that include many more members such as concerning (Lindstromberg 1998) or even including and given from many online ESL language learning sites (EnglishClub.com) and from unverifiable sour ces such as Wikipedia It should also be noted the some of the already mentioned forms Hudson (2000) gives as loanword prepositions are not put in the list. This merely reiterates that there is disagreement or confusion over prepositional members in Eng lish. The purpose of the Table 1 1 is not to be an authoritative list but rather to provide a base of forms f or analysis. Thus, while Table 1 1 does not reflect the full membership judgment of the author, it will serve to later show that these forms straddle the functional l exical divide. Furthermore, s everal of the forms are morphologically complex and have been so marked. Table 1 1 consists of single words, compound words, and phrases. These might not be best analyzed as a single form entering the syntax. If all of these belong under the same head, the structure of p must be complex as in Svenonious (2008) or le xicalized phrases can be inserted under p to act like single prepositions. From a 4 Consider: I traveled back home vs. I traveled home. Back does not appear to actually work as a preposition and probably should not be on the list. 5 Consider: I threw it out the window Line 1 Line 2 Line 3 Line 4 Line 5 Line 6 About A top B ack 4 In front of O ut of T o wards Above A way from Down I nside O ut side U nder Across B e fore During I nto O ver U ntil After B e hind For O ff T hrough U p Along B e low From O n T hrough out V ia Around B e yond In O n to T o At By In back of O ut 5


17 semantic perspective all these forms could be said to be relational in meaning and lack a referent. However, without a syntactic context this may not be true. Without a conte xt they could be argued to have no meaning (Acquaviva 2008 ) It should also be noted that some of these words are more commonly used as prepositions than others. The reason for this will not be fully discussed in this paper but Chapter 4 will give a sug gestion. Defining Particles and Defining P forms Particles are also a problematic form for classification. The term particle used in this paper refers to the second word in what traditionally has been termed phrasal verbs ( i.e. the second word in phrases such as turn on or think over ) Liles (1987) concisely (1998: 4). Since particles have the same phonological form as many prepositions, they ha ve been thought to be just prepositions or intransitive prepositions. They are also commonly called adverbs. Moreover, t heir role as a verbal element and secondary predicate distinguish them from prepositions, which provide Case for a subsequent noun phras Case assigning, argument controversial than prepositions. Den Dikken argues they are the head of a Small Clause (1995: 38). However, in general it is thought that they are positioned somewhere within the verb phrase and presumably they are not adjuncts to the verb phrase. With this being said, there are some syntactic tests that can help distinguish particles from prepositions (Lindstromberg 1998: 243 54). Only syntactic tests were chosen as syntax was the basis for partitioning prepositions from other words to begin with


18 The first test is the Noun Phrase Insertion Test : i f a noun phrase can be inserted between the verb and the form in question, then the form is a particle. (1) a. John started up a business b. John started a business up c. John started it up. Since (1b,c) are grammatical, up is a particle in this construction. Compare this with (2b,c) in the following example. (2) a. John climbed up the ladder b. John climbed the ladder up. c. *John climbed it up. Thus up in (1a c) is unlike the up in (2a c ) because it can be separated from the verb by a noun phrase. Also as is the case in (1c), if the object Noun Phrase is pronominalized it must move before the particle. This is due to phonological reasons with regards to stress placement 6 If a pronominalized noun phrase does not move into the between position, it can be said that the f orm in question is not a particle. The next test, the With the Ellipsis Test we can see that n oun phrases cannot undergo ellipsis if the form in question is a particle. This is because the argument after a partic le belongs to the verb and not the P form. (3) a. I turned off the road at second street. b. I turned off at second street. c. I tuned off the light in my room. 6 Particles generally rece


19 d. *I turned off in my room. By the ungrammaticality of (3d) in contrast to (3b), it can be seen again that the two forms are not operating the same way. While there a re cases of opaqueness between the results of these two tests, if a verb + form passes the first and fails the second test then the form in question can most reasonably be deemed a particle. As for this paper, the distinction will hold to these tests 7 Forms that commonly can pass these tests are found in Table 1 2 Table 1 2. Particle f orms in English Line 1 Line2 Line 3 Line 4 Line5 Across Away Down On Out Apart Back In O ver Up Around By Off T hrough Aside These forms can qualify in terms of both tests above as shown by the following examples. Forms under (a.) Insertion and (b.) Ellipsis test s : ( 1 ) a. She got across the information. She got it across. b. (2 ) a. He took apart the engine. He took it apart. b. *He took apart (the engine) in his room (3) a. I moved around the furni ture. I moved it around. b. (4) a. I set aside some money. I set it aside. b. *I set aside (the money) in a separate account. 7 25) for potential counter examples to the syntactic tests given in Chapter 1. However, there may be flaws to the counter examples as well as they represent a small number and do not concern the more general pattern scrutinized in this work.


20 (5) a. She pushed away the man. She pushed him away. b. *She pushed away (the man) on the street. (6) a. He paid back his debt. He paid it back. b. *He paid back (his debt) in January. (7) a. She pushed the man down. She pushed him down. b. *She pushed (the man) down. (8 ) a. The student handed in the assignment. He handed it in. b. *The student handed in (the assignment) on Wednesday. (9) a. The student put off the test. He put it off. b. *The student put off (the test) last Friday. (10) a. I put on a h at. I put it on. b. (11) a. He thought over the problem. He thought it over. b. *He thought over (the problem) yesterday. (12) a. She thought through the problem. She thought it through b. *She thought through (the problem) yesterday. (13) a. He carried out the box. He carried it out. b. *He carried out (the box) already. (14) a. I looked up the information. I looked it up. b. *I looked up (the information) online. The selection of the two tests which distinguish these fourteen forms is based upon the idea that the NP is not an argument of the P form. Thus movement is alright and ellipsis o mention


21 inseparable phrasal verbs where the main verb is intransitive. This paper does not intend to say that in constructions such as pass out that the P form is not a particle because the two tests above cannot confirm it. Rather it does not intend to argue for the admittance of a form that only appears with intransitive verbs, since a structural distinction between a particle attached to a intransitive verb and an intransitive preposition does not saliently exist. The combin ation of forms found in Tab les 1 1 and 1 2 are to be conveniently termed P simple observation that these two separate grammatical objects often get phonologically realized identically. Thus P forms are the phonological form common in expressing particles and many prepositions. The fact that not all prepositions are used as particles will be discussed later. In fact away and back are the only particle form s that do not get used as preposition s by themselves. However, they will be used as preposition s when joined with word s like from and to to make away from or back to Aside on the other hand could be argued to sometimes operate as a preposition in a sentence like He found her aside the road, but this is not a s common. Nonetheless what is important is observing the great phonological overlap between the realization of the two different structures and requiring that this overlap Distributed Morphology Distributed Morph ology is a theory of word structure that took root with work by Morris Halle and Alec Marantz in 1993 and 1994. In this theory the domain of morphology was radically changed from previous lexical views. The lexicon was deemed to be unnecessary and the fram ework of lexicalism was proclaimed to be


22 dead, deceased, According to Marantz, he underlying suspicion [of lexicalism] was wrong and t (1997: 2). The special meaning of deriv ed words is no greater or different than the meanings that can be assigned to phrases and the properties of words can be better explained through syntactic processes. Words are thus not impervious to syntax and the Lexicalist Hypothesis can be abandoned. However, even with the lexicon abolished, some of its essential components had to be retained. These were split up into three parts. The Parts of the Old Lexicon Redistributed model of syntax. Traditionally syntax and the lexicon operated separately. After words were formed in the lexicon, they could then be assembled into appropriate syntactic fr ames, D structure. However, now the necessary aspects of the lexicon are positioned within sentential derivation. There is initially a component called List 1 or list A In this list are two types of morphemes; bundles of morpho syntactic features called a bstract morphemes and Roots (Harley and Noyer 1999:2). morphemes while abstract morphemes get called f morphemes (Harley and Noyer 2000). List A supplies these two morpheme types with no pho nological features. DM thus proposes that l and f morphemes lexical view ( cf. Di Sciullo and Williams 1987 for a more in depth lexical account of morphology to syntax) where morphemes are built in the lexicon to form complete


23 words that are ready for syntax, l morphemes and f morphemes are now the building blocks of syntax. Thus they form the terminal nodes in syntax and their features are present at the beginning of syntactic derivation. After derivation, List 2 or List B provides every terminal node with a language specific string of phonological data called a Vocabulary Item, VI (Harley and Noyer 1999:5 6). Vocabulary Items are thus responsible for both representing f morphemes and l morphemes. VIs can also be phonologically null. Once provided with phonological material, each morpheme type is searched via a component called T he Encyclopedia The Encyclopedia is responsible for linking idiomatic meaning to individual forms and recognizing phrasal idioms (1999:8). Thus when The Encyclopedia and if The Encyclopedia encounters the phrase pa c king heat, it knows the intended With these components and a set of co re assumption, DM is able to build words and sentences and begin to account for lexical ambiguity all in one system. Core Compo nents of Distributed Morphology There are three theoretical assumptions about DM that distinguish it from other morphological t heories and make the reduction and redistribution of the lexicon possible: Late Insertion, Underspecification, and Syntactic Hierarchical Structure all the Way Down. Late i nsertion : this morphemes]


24 276) 8 Thus while syntax is impervious to phonolog y, phonology might be affected by the feature arrangements produced in syntax. Underspecification : t his is the idea that Vocabulary Items are not fully specified i n a terminal node, the identifying features of the Vocabulary Item must be a subset of Item could represent several different feature sets in a language as is evidenced by the fact that represents everything but [[+3 rd person] [+singular]] in the present tense subject verb agreement paradigm in English. However, if a VI has a feature that is not Items cannot add semantic or syntactic f eatures. These stipulations create competition specification are interesting in explaining the distribution of P forms, which will be discussed in Chapter 3 and Chapter 4. Syntactic hierarchical structure all the way d own : this is t he final cor e component of DM. structures de ramification of this is the unification of the components of syntax and morphology. However, this component will not be so important for this present analysis. 8 This gets tricky for Roots. If insertion is free for Roots, then there are no semantics at the L node or within the Root. This means semantic material can be added with the insertion of phonological informa tion into a Root.


25 P forms in Distrib uted Morphology The structure of Distributed Morphology is wel l suited for the fact that the phonological form of prepositions and p articles is often the same. Similar to providing well for the long standing observation that Verbs can often be Nominal or v ice versa, particle forms can double as prepositions. However, as t he ca tegorization of P forms was a problem for more traditional theories it a lso provides problem s for Distributed Morphology. Within DM, the status of prepositions and particles remains s imilarly unclear. P and the Building B locks of Distributed Morphology As introduced the theory of Distributed Morphology only divides morphemes into two In many cases their division seems easy and natural, as forms operating with traditional lexical properties such as nouns and verbs will be termed l morphemes and those with grammatical properties will be termed f morphemes, keeping the old lexical functional divide intact. Thus related to that between the functional categories and the lexical categories However, the cleanliness of the split between the two types of morphemes demands attention with regards to membership when looking at a class of forms like P forms. Similar to the problem these forms give a traditional account, P forms represent things that straddle the lexical functional divide. If we value Cro should be able to clearly define the element of analysis, a clearer distinction has to be made as to what l morphemes are in comparison to f morphemes.


26 Basic Properties of F and L Morphemes Roots such as CAT or SIT, as well as the fact t Roots are deemed to be (2005:6) 9 Based upon this description, we can begin to separate f morphemes from l morphemes by means of t heir source. Roots come from memorized sets of semantic features 10 and abstract morphemes come from Universal Grammar UG Based u pon the previous s ections, it seems that this dividing line will not be satisfactory with P forms because some P forms can represent universal grammatical features and have lexical uses. Therefore, in order to decide if P forms can be grouped into either list successfully o r if P forms have to be split up, the dividing criteria must be clearer or the existence of new category must be recognized. Thus P forms also pose a categorization problem for the theory of DM. Conclu ding Remarks P forms are forms that represent both prepositions and particles. They are difficult to classify because they have both functional and lexical qualities. Distributed Morphology is well suited for explaining the one to many relationship of form to function, however, P form classification remains uncertain within the initial criteria of DM. Thus a further analysis of how P forms can be placed within DM is needed in order for this observation to be useful. 10 This is debatable because any features that are associated with the existence of a Root before Vocabulary Insertion could be argued to influence such insertion and for Roots Vocabulary Insertion is argued to be free. This is explored mu ch more in Chapter 3 and 4.


27 CHAPTER 2 THE ORIGIN OF P FORMS This c hapt er is divided as follows: the i ntroduction presents the classification dilemma of P forms in DM, P forms as f morphemes discusses how P forms act functionally or need to b e classified as f morphemes and talks about the possible universal fea tures prepositi ons repre sent, Particles as f morphemes talks about particl es as functional elements, P forms as l morphemes presents arguments for why P forms should be l morphemes, Reanalysis of P forms concisely explicates the need to modify the P form list, The Homony m E xplanation denied argues against calling up and up homonyms Reorganization redefines P forms by separating some members, Categorization presents a Little p_ lexical functional head to catego rize l morphemes, and the conclusion notes standing problems. L or F Morphemes Examining the synchronic origination of the P form category is import ant to understanding their classification/nature Since this paper is working with DM, t he lexicon is not the point of o rigin. Origination will therefore be defined a s the features or terminal nodes in DM (Harley 2008: 3), there can only be three logical possibilities for the origination of P forms: l nodes 1 f nodes or both. L nodes corre spond to l morphemes and f nodes to f morphemes. If P forms originate solely as l morphemes, then their realization is comparable to other l morphemes that get categorized as Nouns, Verbs, and Adjectives. If they originate as f morphemes then they are 1 It is debatable if lexical words come from a single that are indexed for specific vocabulary insertion.


28 composed of m orpho syntactic features and their phonological realization or Vocabulary Insertion is comparable to other functional morphology such as determiners, complementizers, and conjunctions. If they origi nate as both, then they are either two differ ent things which happen to share the same phonological representation, homonyms, and or some of the forms have incorrectly been put into an undifferentiated list. In the latter case there would be strictly functional P for ms. Within argumentation, if the source of P forms is split between l morphemes and f morphemes, the P form list should be modified to explain this We have to say why the same form appears as both l and f morphemes and what, if any, features exist to dete rmine or guide insertion. Examining some of the d ata can narrow down the se possible options P Forms as F M orphemes The fact that P forms represent functional elements, especially when used as prepositions, suggests that they would originate as a bundle of feature(s), from an f node and not an l node or Root. In (1a,b) the forms up and down theoretically assign Case to their complement DPs as all prepositions are proposed to do under the Case Filter principle of P&P (Carnie 2007:297). (1) P repositional Usage: a. Sh e walked up the stairs She walked up them b. He walked down the stairs. He walked down them In (1a,b) they could be said to assign a directional locative case, but the label is not too important now. It is only relevant to recognize that the object pronouns in (1 a ,b) have non nomitative case markings and that this case theoretically comes from the


29 preposition or p If this traditional concept is accepted, it is then no t too controversial to say that prepositions serve a specific functional purpose ( i.e. to assign Case to a phonologically realized DP). Now verbs also assign case but it is not the lexical verb that assigns case in DM and Minimalism, but rather the little v_ functional head that licenses the l morphme in the VP. As this chapter progresses, we will see that prepositions can be the manifestation of a functional head or an l morpheme that is licensed by the functional head. Nonetheless prepositions on the sur face appear to have a grammatical role. Now it should be noted that often Prepositional Case is put in contrast to Grammatical or Structural Case. Prepositional Case is often referred to as a type of Lexical Case to account for its ostensible idiosyncrati c behavior (Anderson 1971). However, in DM there is no Lexicon to put this idiosyncratic information, so if prepositions do assign case, this case must be a feature in the syntax. In DM there is no reason to assume they have lost this functional purpose. T he phonological form of prepositions could then be argued to be the Vocabulary Insertion of prepositional case features, semantic as they may appear. Examples of possible prepositional fea tures in English are given here Table 2 1. Possible a d positional c ase f eatures in English To/of By/for With/Toward From/at +Goal +Agentive +Associative +Source +Possessive +Benefactive +Directional +Locative These are not all the features possible in human language. Calabrese (1998) gives more examples for just the Romance languages. Moreover, features as these do not have to cluster identically and some might be absent in different languages, accounting for t he well established observation that case systems often do not align


30 between languages (Mithun 1991: 510). Nonetheless, the features above can determine the P forms that represent them in English. In the case of (1a), we might first posit a simple rule su ch as insert the form / up / to represent the features [+Locative +Dir], which happen to be active in English: /up/ < -> { +Loc, +Dir}. In the phrase [{ +Loc+DIr } the stairs ], this simple rule directs the form / up / into the syntactic domain [[p_{+Loc,Dir} [n_{+Def}[ stairs]]]. Of course more will be required to partition the insertion of up versus down and other locative but in Chapter 3 we will see that this does not have to be a featural issue. Nonetheless when P forms directly expre ss the features given in Table 2 1 they would originate as f morphemes under the and that this source identifies morphemes as f morphemes ( Embick 2005:6) If these features are universally available, i t would explain why so many languages have ad positional or case marking systems that appear to realize them. For more information on case marking, see Cook (1989). Examples of some of these features represented by case and prepositional systems can be see n in Example (1). First examine the associative feature in Icelandic (Zaenen et al. 1985:464) (1) Hann t k vini s num o pnum rmum he took friend his[+REFL] open arms The instrumental marking seen with opn um and rm um is argued to be result of an ase it with accomplishes this in English as shown in the translation, and thus could be argued to also represent this associative


31 n. This locative feature in Latin can be seen in (2) (2) a Domus Home NOM b. Dom i Home LOC at Since Latin lacks ad positions like English, its nouns get case marked to show there grammatical/semantic relationship. The locational relationship presented by a locative feature in (3b) in comparison to (3a) is expressed by the case marker i in Latin an d the preposition at in English. However, English is not alone in expressi ng the features given in Table 2 1 ad positionally. This can be seen the expression of the d ative feature in Japanese. (3) Takashi ga Yumi ni hon o ageta. Takashi NOM Yumi DAT book ACC gave. to In Japanese the dative or goal feature is represented ad positionally. N i like other Japanese post position particles is limited in its distribution and purely functional. The preposition to in English also represents this feature in the same functional manner. The functionality of to is more evident when the same meaning can be achieved structurally as w ell, as is seen in English with the dative shift movem ent. (4) Takashi gave Yumi the book.


32 Takashi NOM Yumi DAT book ACC In the dative shift example of (5a), the dative interpretation remains without the preposition to, meaning the feature is present without a phonological form. This suggests that the feature is already present in the syntax and is only being realized by /to/ in certain arrangements. Nonetheless what is important to agree upon is that the P forms discussed so far are serving a functional purpose similar to case markings in Lati n and Icelandic and Japanese postpositions. Moreover, these case markings and postpositions are deemed to be the result of features present in the syntax and not the presence of an l node hosting a Root. Thus it seems reasonable that P forms that work in this way are functional and that at least some of them originate as f nodes. P articles as F M orphemes Particles on the other hand lack the structural function that prepositions do as can be seen through their use in (1) (1) a. She blew up the house. Sh e blew it up. b. The wolf blew down the house. He blew it down. Particles are not as clearly functional as prepositions in that they lack the case feature possessed by prepositions. Accusative case is given to the DPs by the verb, not the particle. T his argument is positioned on the simple observation that accusative case is given to the complement DP whether the particle is pre sent or not. This can be seen with E xample (2). (2) a. John beat him b. John beat him up.


33 Accusative case is assigned to the pronoun regardless of the particle. One might suggest that the verb and particle assign case together in (2b) but it seems better to have case consistently assigned by one thing, little v_ and not the particle. Nonethel ess, it could be argued that there is a Prepositional Head or Relatio nal Head, as we shall see in Chapter 5, in the vP that merely lacks the generic Case Feature and that this Prepositional Head is still functional in nature. The function of the P form in relation to the object of the verb is not to give it prepositional case but rather a up and down would be like other VIs in that they are underspecified and thus can be inserted in this environment as well as a prepositional environment. For example we could say that a p_ head outside a verb phrase contains {[+Case][+Loc, +Dir]} features while one inside a verb phrase merely has the {[+Loc, +Dir]} features. Therefore up would still be inserted via the rule: /up / < -> { +Loc, +Dir}. Again the {[+Loc, +Dir]} feature would need to be subdivided to predict the correct insertion of down versus up. Thus it is still conceivable that P forms are just the VI of abstract functional features positioned in the structure of syntax, but doubt is justifiable. All in all the origination of P forms as abstract feature bundles or f morphemes would be in accordance with other functional elements created by morpho syntactic features as is described by (Embick and Noyer 2005:5 7). Fe atures commonly associated with prepositions such as those listed in Table 2 1 might be able to be divided into more precise features to explain the insertion of all the prep ositional forms given in Table 1 1 in Chapter 1 Moreover, positing that P forms o riginate as f morphemes would also be helpful in explaining why l


34 never appear as prepositions or particle but are inconceivable operating as such. This observation and issue will also be further explored and reevaluated in Chapter 3. Nonetheless presently there is some motivation to assume a non Root origination for prepositions and particles. This would theoretically account for the fact that they share the same phonological form without accidence and if this were the end of the story, then P forms would simply be a case of underspecified VIs, capable of being inserted in either functional domain. Particles would be bleached prepositions. P Forms as L M orphemes : In contrast to the above section, t he e vidence that P forms are l morphemes is fairly strong. Assuming non homonym origins, if there are no P form VIs for l nodes, then we would expect all P form VI s to correspond with identifiable feature bundles held under ( a ) functional node (s) That is those feature bundles co nstructed from the features given in Table 2 1. This would suggest that these forms would not appear in other places because other places would not possess the feature(s) necessary for P form Vocabulary Insertion. However E xamples (3 11) present a different story. It can be seen that the same P forms up and down that were used to represent the proposed f morphemes in the previous section have a much wider syntactic distribution, a distribution that lacks identifiable functional features. Examples ( 3 11) show P forms appearing in different syntactic domains. In (3a,b) a verbal usage is seen. (3) a. The doctor upped the dosage. b. The doctor downed a whiskey. It should be noted that these examples share the argument structure of related verb particle constructions: The doctor pushed the dosage up and the doctor drank a whisky


35 down. Whether the forms / up/ or / down/ are directly inserted into a Root under little_v or get there derivationally will not be argued. Regardless, other f morphemes do not work their way into such positions in English. The sentence, I *ored her, with the or the sentence, I *thated with an unknown meaning is also ungrammatical. Moreover, P forms that do not get used as particles such a behind can take a nominative interpretation. (4) In (4 sition cannot be a particle behind you not you behind and if it is from a prepositional position, then this preposition has to be lexical as functional elements should not be able to move under other lexical functional heads, in th is case n_. Again this paper will not argue if the P form is directly inserted under n_ or moved there. However, that discussion could be valuable in the future. Moreover, we also see P forms acting like adjectives. (5) a. The uppity teenager was rebuked b. ?She is down In (5a) it would be difficult to see where the form /up/ comes from other than being directly inserted under the lexical node a_. (5b) is less clear. Nonetheless the form /up/ main tains its core meaning. Someone who is uppity might need to be slapped down The relational contrast to down is maintained while the form is under a_. There are also cases where P forms get used as nouns. (6) a. That basketball player has some ups b. How many downs in football do we get for every possession?


36 Again in (6a,b) the P forms are found in a nominal position. This time it is possible that they are derived from particle usage. (ex. He really jumped (himself) up high. Ye got some ups. The football team has been taken down four times. The possession is finished. The other team took them down.) Nonetheless what is interesting is that the P forms have acquired nominal characteristics such as countability downs and collectiveness ups Moreover, P forms are also subject to other morphological processes common to l morphemes and not f morphemes. Examples (7 11) show P forms creating deverbal nouns, compounded prepositions, compound nouns with left adjoined P fo rm, compound verbs with left adjoined P form, and compound nouns with right adjoined P form: (7 ) a. The upper the doctor gave was great. b. The downer made her drowsy. (8) a. She moved upstage b. He moved downstage (9) a. They upgraded my seat. b. They downgraded the threat status. (10) a. Last night s victory by the inferior team was an upset b. Rain is frequent but that was quite a downpour (11) a. The setup of the game requires time. b. Did you see that amazing touchdown From examples (3 11), it is clear that the forms up and down can be used in many different syntactic contexts and that they can obtain many different traditional syntactic


37 phonological forms used for other f morphemes. Determiners and conjunctions do not display any such distribution, suggesting a clear difference between what can elicit a P fo rm and what can elicit other functional elements. Thus having a heads n_,v_,a_, is useful in order to explain such behavior. This requires that at lea st some P forms such as up and down be considered l morphemes like dog and walk. Moreover, this proposal would be in line with the assumption Svenonius makes when However, such a proposal will still require a resolution with the res ults t hat came from the previous two sections, which indicated that P forms originate as f morphemes. The fun ctional aspect of P forms is evident too. The Reanalysis of P F orms in DM Due to the data and arguments of the previous three sections, the list of P forms if this is right, some prepositions are content words and some are FWs [function wo This will be done instead of giving a homophonous explanation because a homophonous explanation says using the same form functionally and lexically is an accident, that there is no semantic motivation, and that all functional forms have a n equal shot at having a lexical twin.


38 The Homonym Explanation Denied Homonyms are two different words with identical phonological forms. What is meant by different words is that they have clearly separate meanings. This could be from two different histor ical words achieving the same phonology by more or less accidence such as bat bat undergoing severe semantic drift such as bank the financial institution and bank have perhaps done. Either way the two words do not share a semantic core that can be easily explained. Most speakers would say they are separate words 2 However, the up used in examples (3 11) above has arguably the same core relational meaning as the up s in John walked up the ladd er and John blew it up There is the concept of movement from a low to high position. In this way up has a semantic field similar to high. Up like many P forms is certainly polysemous but not homonymous as the other examples are. Some of the P forms might turn out to have a homonym functional partner, but as with up this is not always the case. Thus all P form instantiations of both functional and lexical representation cannot be writt en off as mere homonymy. Thus to maintain some elegance, a homophonous form analysis will be rejected as much as possible. Since P forms carry similar semantic content regardless of their functional or lexical use, homophony will not be the best analysis. Reorganization and True Prepositions Since homonymy is not at least always the case, there has to be a way to have l morphemes represent f nodes to support the observation that P forms are the phonological representation of a functional distributional cla ss in English and to explain 2 This is the judgment of the author who has no on hand scientific evidence to justify such a claim.


39 their wider nonfunctional be havior. However, the nine forms f ound in Table 2 2 will be partitioned from the rest because they lack syntactic and morphological flexibility to j ustify ever being l morphemes. Moreover doing this r prepositional inventory and better aligns it with the smaller ad positional inventories found in languages such as Japanese and many of the Bantu languages. If reason for this partition holds, we can then have a clear division b etween true functiona l prepositions and P forms. The forms separated can be seen in Tab le 2 2 where the itions has been reduced to nine members 3 Table 2 2 Phonological forms that only serve to mark prepositional c ase Line 1 Lin e 2 Line 3 Line 4 At Of From For By 4 To With 5 a 6 Via The first interesting thing to po int out from Table 2 2 Historically a As a functional item it lost its full pronunciation and has been reduced to schwa in modern English. This is why so many prepositions start with a in English. Moreover many of the forms from Table 1 1 in Chapter 1 have been reanalyzed as single forms. In fact it could be argued that all the forms that have thi s historical derivation have been reanalyzed as single forms. However, this paper will entertain the notion that a is active as a possible representation for the functional head p_, which will be discussed later. 3 It is the belief of this paper that a similar thing is happening in any language that appears to have a bloated list of ad positions 4 The form by refers only to its agentive use. 5 The form with only refers to its instrumental/associative usage. 6 [ a ] is the result of a historical process and could be viewed as an allomorph of at. It could be the functional representation of a [+Locative] feature


40 These eight forms + a will no longer be co nsidered P forms because they only look at or do not pass the par ticle tests given in Chapter 1 such as the Noun Phrase Insertion Test: (11) a. John looked at the store. *John looked it at. b. John ran by the store. *John ran it by. This means that particles will not be considered f morphemes like prepositions in that they have no case assignment function. This also means that some adverbial elements that never get used as prepositions sho uld be able to function as particles such as away and other similar terms. Moreover, another behavior al distinctions between Table 2 2 and P forms is that the forms that represent true prepositions do not attach to other Roots productively. Ostensible cou nter examples are found with the form with in withdraw and withhold, but these words are fossils of when with One might also consider by because of the words like bylaw, but its etymology would seem to preclude such a propos bi form bi had a different meaning (Online Etymological Dictionary 2010). Moreover, none of the seven forms appear normally as nouns or verbs: at s atted, bys byed, ofs ofed, tos toed, fr oms fromed, withs withed, fors fored. Now it should be noted that not all P forms are used in all lexical positions as up is. Membership in the P form class requires that the phonological form maintain some lexical and functional duality while maintaining an arguably similar meaning. Therefore, the proposal is that the forms in Table 2 2 will only be inserted to mark


41 P forms are a categorica l, a s the data from P form s as l morphemes suggests. They can thus appear in many syntactic domains including that of a preposition. T raditionally this multi categorical nature could also be explained by having a lexicon with several copies of words such as up and down or by implementing conversion rules, including affixation. However, within DM all that is r being inserted near a categorizing head. Since generally the only listed functional heads capable of giving a syntactic category are n_,v and a_ (Trias 2009:1), this paper proposes that a four th le xical functional head called little p_ be added to the list This seems to accord with the category p natural locus o f relational notions of containment, attachment, and support which are f little p_ can be included with the other established heads, all multi categorical P forms can b e elegantl y classified as l morphemes Categorization by L ittle p_ To review, DM postulates that all l morphemes are a categorical by nature. To achieve a syntactic identity, Roots or l morphemes must be positioned in a structural relation to a functional head or f morpheme capable of assigning a category. L morphemes cannot exist without a categorizing functional head (Embick a nd Noyer 2005: 5 ). This relationship is usually described as an f morpheme immediately c commanding an l morpheme. Furthermore, it should for sentence elements, such as noun, verb, and adjective, have no universal Nonetheless, DM still uses the traditional labels to discriminate between the three commonly accepted category


42 assigning functional heads: n_,v_, and a_. In this way DM can recognize that such labels are not necessarily universal but that l morphemes do o btain similar identities given certain placement. It can be inferred that while these labels are just labels and kept for convenient communication purposes there must be universal features distinct enough for us to want to retain the la bels, as Baker (2003 ) argues. P forms get categorized as other l morphemes get categorized. However, when they operate as prepositions, the features under the p_ terminal node check with the selected Root that p_ immediately c commands. This means that like other categorizi ng functional nodes, p_ has selectional requirements as to which Roots can be merged and categorized with it. The inclusion of selectional requirements is not salient in DM and the general spirit of the theory seems to avoid them. However, they do seem to creep into the discussion because of necessity (Harley and Noyer 2000, Siddiqi 2005, and Harley 2008 for licensing possible VIs). Selection requirements are further explored and developed in Chapter 3 and 4, but for now we must note they have to be stricter in the case of prepositions. This is why we do not see very many new Roots being used as prepositions. The basics of this are shown here Figure 2 1. Categorization by l ittle p_ in dog,* run, friend


43 be limited to the appropriate VI form depending on the context and feature checking. The nature of this will be investigated in Chapter 3. Conclu ding Remarks The struggle to classify prepositions and particle neatly within older grammatical frameworks is understood. Since many of the forms that mark prepositions and particles in English are so lexically flexible, a new way to organize them was needed. With this bei ng so, the theory of Distributed Morphology provides the framework necessary to explain why one form can be synchronically used to represent different functions. In doing so, the list of P forms needed to be revised to account for the different lexical fle xibility of its so called members. The results are good in that now there is a grammatical explanation for the old observation that prepositions in English are semi lexical and that the class is relatively closed. We now have a small functional list of for ms that only behave as prepositions and a larger list of forms that can realize prepositions as well as other functional heads. Thus prepositions remain functional and l morphemes remain relatively free to appear in many syntactic domains. However, now it must be shown clearer why certain l morphemes fail to get categorized by little p_.


44 CHAPTER 3 THE CONTENT OF ROOTS, VOCABULARY INSERTION AND CATEGORIZATION RESTRICTIONS This chapter is organized as follows: the first section, Need for Restriction, introduces the notion of different levels of grammaticality. There are well formed sentences, there are ungrammatical sentences and there are semantically deviant/odd sentences. It also talks about form flexibility and that not all lexemes hav e equal distribution across grammatical categories and that this distribution might have to do with semantic structure and a general con servative property of langu age. Following this, the next section reexamines the vocabulary insertion of L morphemes to s ee if restrictions ca n be feasibly added. The section after examines the possibility of having restrictions based upon properties of different Roots. Then there is a discussion on how features external to the Root might influence insertion. The section f ollowing looks at the features of categorizing functional heads to see if they can reasonably supply features to the Root. T hen the chapter looks at other possible cases of l morphemes acting like f morphemes to justify this as a part of th e general DM th eory. The section after that gives a small historical account of a P form used as both a lexical and functional item before becoming fully grammaticalized. This relates the notion of semi lexical to grammaticalization and the fact that the grammar cannot e asily explain forms that operate in the middle. Finally the conclusion calls for a better description of the mechanism or mechanism(s) responsible for Vocabulary Insertion control.


45 Need for Restriction The a categorical nature of l morphemes is a very appealing aspect of DM, as so many base forms can appear in different syntactic frames in many languages 1 Moreover, syntactic labels do not have to be equal cross linguistically because these Japanese grammarians can insist something is an adjective even though it functions more like an English verb. The existence of a categorical forms works well with the observation from Chapter 1 that the phonological form of most particles is the same as prepositions. However, it is also true that form flexibility, the acceptability or use of a phonological form in different syntactic positions, is not the same for every form. In other words a form, a VI, does not always naturally or easily coincide with t he perspective given by a licensing functional head such as a_, v_, n_, or p_ This means one cannot expect an l morpheme to occur in all syntactic positions with equal probability. There seems to be rules and or semantic inhibitors that influence the pro bable distribution of l morphemes. Otherwise distribution inequalities are mere accidents. In some cases what might seem like a distribution inequality is just an accident, but to deny any connection between universally shared concepts, even if the edges o f such concepts are fuzzy, and grammatical categorization seems like an unnecessarily harsh rebuke of every grade school grammar book that teaches nouns are people, places, or things. We know this is not the full story but there is a pattern. All people, p laces and things to have existed inside the English language have had an attested nominal usage, 100%. Can we say the same as for things and attested verbal usage? 1 This observation is of course not unique to DM alone.


46 Categorization inequality is perhaps related to the observation that not all forms as sociated with certain syntactic categories are borrowed with equal ease (Haugen: 1950: 224). As far back as Whitney (1881:19 20), loanword flexibility could be scaled as Noun>Verb, Adjective> Adverb>Preposition, meaning nouns are most often borrowed and pr epositions least. One reason of course has to do with need, new inventions and novel items require new words to be coined, but given our theory, DM, these new forms should be able to find their way under other functional heads as well. The trend seems to b e for loanwords to stay in the categorical frame they get introduced in or imported s if the borrowing language lacks the source languages grammatical distinction, a loan words grammatical category will have to adjust, but if both languages have the same d as that keeps the form as it was and not what it could be. There is a type of natural categorical loyalty to the input. The reason for this conservation should be c onsidered as it might be similar to the reason why lexical flexibility surfaces unequally. However, this will not be explored here. This is only a generalization as there can be exceptions and the form over time may become more flexible. In a language such as English where range of usage (1881:20). One might be tempted to say they work t heir way into the


47 Native American words have been long borrowed without attested changes in Karate does not get used as a verb in English 2 The point is l morphemes do not always appear in all locations governed by a lexical functional head and borrowed words tend to exemplify this pattern. Loan word categorization tendencies might not be rela ted to native word categorization tendencies, but its use in a verbal or adjectival frame is abnormal in English. If you go to ( http://www.verbix.com/webver bix/English/apple.html ), you will find it conjugated, but to find any examples of this use via search engines. Its abnormality can be seen by the novelty of the sent ences in (1b) and (1c) in comparison to (1a) where a normal placement of (1) a. Apples are food. The reason for this lack of distribution may involve two things: grammatical categorization or semantic probability. In this case it seems to be semantic probability that is to blame for the awkwardness of (1b,c). In other words the meaning of apple has been contained wi thin a small conceptual frame (Fillmore (1976) for more on semantic 2 Note the compound form karate chop though.


48 frames) 3 An argument will be that generally the hyponym term is more restricted than its hypernym, because the hyponym is a more specific instantiation 4 Thus, time gets used as a verb an d a noun but hour and minute do not chair can but not armchair or stool dog can but not Dalmatian or Terrier, clothe s can but not shirt or pants 5 It is important to note that as for the grammar all of the above forms might get used as both nouns and ve rbs, but that the probability of this happening is lower than hypernyms of high frequency. Now consider the two incorrect sentences in E xample (2). (2) a. ***I got a bad grade apple my test. b. *I got a bad grade in my test. There is something d rastically more incorrect with E xample (2a) than (2b). This may accord well with the notion of Levels of Ungrammaticality. Example (2a) is inconceivable as a single sentence. It makes no grammatical sense. Everything we know about apples forces us not to use it as a preposition. (2b) on the other hand is easier to understand difference in the level of ungrammaticality is (2b) only has a semantic or association problem while (2b) has a real grammatical error, a phonological form that cannot be relational or the insertion of an illegal VI. Now consider the word in (3). (3) *? I found it bottom of the lake. 3 great for the theory of DM as it predicts this possibility. However, it does not predict why it is not common or why we have apple pickers and not applers. 4 This could be comparable to the chicken and egg causality dilemma. 5 owever, this is not widely used or accepted.


49 This will appear ungrammatical because of the omission o f which normally appears under p in this construction. Bottom is normally used as a noun but the semantics of bottom have a relational implication. The reason that this is ungrammatical is different than both (2a) and (2b). In theory it should be comparable to (1b,c). The meaning should be alright but the form is not used under little p_ and this makes its initial acc eptance strange, a general conservative property of language. However, there is no reason why is. Thus it is a bottom abottom is a candidate for little_ p However, in cases suc h as (2a) there has to be a harsher restriction as apple is not a candidate. This requires a more in depth examination of Vocabulary Insertion at L nodes in order to find possible ways to account for this information. L Mor phemes and Vocabulary Insertion The idea that l morphemes are completely free for vocabulary insertion is morpheme is defined as one for which there is a choice in spell insertion of any Vocabulary Item capable of being placed under an l node or a subset of ostensibly b e brought on by the presence of features. However, the determination of what if any features are present in l morphemes before Vocabulary Insertion is also controversial. Some claim that l morphemes have phonetic features and semantic features throughout t he derivation (Embick and Noyer 2005 ), which would seem to precisely determine their representation. Others think they should be completely underspecified (Acquaviva 2008), allowing for many choices at Late Insertion. There


50 also may be middle ground wher e the realization of l morphemes is only partially determined. This seems intuitive because there seems to be levels of ungrammaticality, meaning some insertions seem worse than others. This will be looked at further. Nonetheless, if the middle ground is t aken, Roots or l morphemes would not contain fully deterministic features. Therefore, not all VIs associated with Roots would be available for insertion. Instead a subset of VIs could be inserted. Root Based Restriction If indexing is present in Roots this would prevent the insertion of Dog in a noun phrase where Cat is desired. This does not seem desirable because there would then be no choice at Spell Out. This indexing would make the Vocabulary Insertion of Roots competition free. If Roots or l morphemes contain semantic features as is indicated by Halle and Marantz (1994) then their realization is not arbitrary and may still involve choice. In principle if there are any features that differentiate one l morpheme from another, then Vocabular y Insertion can begin to be controlled. The question is then how many features can be put into a Root in List A before it begins to appear like a lexicon. These features would need to be few and well established. If one Root contains the feature [+locati ve] while another contains {[+animate] [+count] }, then it is { [Root] { [Root] [+animate] [+count]}. With these rules in place, it would hav e to be stipulated that the VI / dog / could never realize an l morpheme c commanded by a little p_ head because only l morphemes with prepositional features such as [+locative] could be hosted there to begin with. On the other hand, /up/ would still need to be able to be dominated by a littl e n_ head. Thus a little p_ head would need to be more selective in the type of l morpheme it can dominate


51 than little n_. Another rule or list might be needed and this would amount to two versions of up Of course this is not desirable. Nonetheless this t ype of restriction placed on the realization of l morphemes would be due to a property of the l morpheme itself and thus be internally motivated. Without further analysis, it is unclear how many semantic features would be needed to make everything work if it could all work. If positing semantic features in an l morpheme is rejected, there is also the idea at encodes reasonable to posit that some l morphemes might contain a category feature if this cannot already be determined by class features. The solution would be to say all P forms possess a diacritic and thus a Root under a little p_ would inherently be different than other Roots in the derivation by means of this diacritic feature: /in, on, { [Root]p}. This also does not seem like it would work. This reasoning runs counter to the general lexical decomposition hypothesis core to DM. In the general spirit of DM, less is more. The proposal that there are features or diacritics in a Root would be increasing the bulk of List A, making it appear more like a lexicon. Acquaviva category l morphemes can be deemed to be featureless, meaningless place holders. Anything that could differentiate Roots for insertion restrictions should start Root externally.


52 Possible External Restrictions on Roots If l morphemes are proposed to be featureless (Acquaviva 2008), meaning there is only one instantiation of Root in List A, then perhaps VI insertion is restricted by features present under the categorizing nodes. This proposes that the features that give a functional categorizing head its identity can restrict Vocabulary Item insertion. In other words there can be stipulations placed on VIs in c commanding relationships to certain functional heads. The feature(s) that makes little p_ a unique functional head will be used to dictate what l morphemes can be c com manded by it. In this case rules for the insertion of VIs for l morphemes could be written with environmental considerations. A very simple rule might look like this /P form/ / [ p_ ] (where p is a functional head that c commands). Remember p_ is just a label to mark the presence of a prepositional case feature. While in the case of little n_ v_ or even a_, the rule would be even more expansive: /P forms + non P forms [v_, n_, a_ ] ( where the c commanding functional head is listed) Now obv iously these rules do not stipulate what form exactly the final solution either. However, this is better than the internal constraint as it still predicts there to b e choice for VI while clearly stipulating that P forms can become both nouns and prepositions while non P forms cannot. However, it does seem problematic that rules such as these would limit certain forms from being inserted under a functional head without clear reason. How does the grammar know the identity of the functional head or what forms are P forms? Categorizing Heads and t heir Features It would seem reasonable to question the need for postulating the existence of categorizing heads when as mention ed in Chapter 2, DM does not assume absolute


53 syntactic categories. In other words, why do l morphemes have to have a category when we do not universally recognize them? The reason for this is that l morphemes would be meaningless without categorization. Ac quaviva (2008) argues that meaning is dependent on perspective and without categorization l morphemes acquire no perspective. a comparative or quality reading, little v_ an even t, and little p_ a relational reading. Thus the Categorization Assumption of Embick and Noyer (2005: 5) remains central: l morphemes need a functional head to be realized and moreover to be meaningful. However, it is not fully evident what features are pre sent in the functional heads that are proposed to categorize l defense that nouns, verbs, and adjectives are universal categories, the labels n, v, and a are almost always listed as examples of categoriz ing heads, but it is unclear if there should be a [+nominal] feature in the UG inventory. Moreover, out of these potential studied head of this type is the verb as BECOME, CAUSE, and DO (5). However, it is uncertain whether these varieties of v are caused by identifiable features as is suggested by Harley (2006) or if a [+become] is the actual feature that can define th is categorizing head. In other words, does a uggests that categorizing features are semantic. For this paper, the claim will be that these heads must have features to exist and that these features could be either semantically or grammatically understood. Furthermore,


54 before Vocabulary Insertion. How this exactly works is the focus of Chapter 4. However, if this is only a problem in English then we can be doubtful that such restrictions exist in UG. However, there are other cases of apparent non homonymous forms being used in both functional and lexical positions. L morphemes as F morphemes in Other Languages Other than the P forms, previous work has recognized that Vocabulary Items used to normally represent l morphemes sometimes get used in other languages to represent f morphemes. Look at the nouns being used as classifiers in Vietnamese data provided by Acquaviva (2008: 12). (4) a. hai ci bao two thing bag 'two bags' b. hai bao cam two bag orange 'two bagfuls of oranges' In this example the form bao is used as a noun in (3a) and then gets put in the classifier position in (3b). This supposedly would be a case where a VI associated with l morphemes is representing an f morpheme, as classifiers are generally considered a type of f morpheme. Acquaviva ( 2008: 13 ) also gives two example lexical items from German that can be used as prepositions. L aut the German word for 'sound', can be kraft the word for 'force' can be used to mean 'by means of' This ca n be seen in Example (5) 6 6 My thanks to Dr. Jules Gliesche for the examples and the assurance that Germanalogically the use of


55 (5) a. Laut dem Bundesfinanzminister ist der Staat pleite. According to the finance minister is the state broke b. Kraft des Gesetzes ist finanzielle Unabhngigkeit By act of the law is financial independence ab sofort verboten. Immediately forbidden. However in Kortmann and : 672 ) account of deverbal and denominal prepositions in German the nouns laut and kraft among others are listed as denominal prepositions Kortmann and Knig show that the process of verbs and nouns becoming prepositions is historically quite attested via grammaticalization. Thus it might be expected that the use of the same form in both functional and lexical frames i s a precursor to grammaticalization, a phase in the evolution of language. Moreover, it is uncertain how long a form usually straddles the semi lexical position in the grammiticalization process. Furthermore, there is nothing that says it must eventually b e adopted into the greater l and f morpheme categories. Nonetheless, if these forms are synchronically being used lexically and functionally like P forms, we can conclude that l morphemes get used as f morphemes in more than just English and that this is a common process. Comparing present day P forms to the general greater process of grammaticalization seems to flow well with the history of English. Historical Der ivation of English Prepositions The English language historically was more synthetic, relying on case markings to show the relationships between noun phrases. As it changed into an analytic language,


56 losing some inflectional marking, lexical words took over necessary functional roles. Day English de rive ultimately from today had both adverbial and prepositional uses. Also there was a stage where both case markings and the prepositions existed side by side (1996:155) This stage of the language can be seen with an example from Old English where t can operate either as an adjunct or as a necessary preposition. Smith (1996:15 4) shows this with this example. (6) a. H cw mannum He spoke the men to b. H cw mannum He spoke to the men adverb lexical function, has in Present Day English b 154). This agrees with the analysis that P forms are l morphemes working in functional positions because there was a time where t was doing the same. The implications might suggest that P forms will continue to be grammaticalized. However, the opposite might also be argued (Ramat (1992) for arguments for degrammaticalization and


57 Conclu d ing Remarks The probability of a l morpheme being represented by a particular VI depends on the functional head that licenses it. This probability is not accidence but the result of some sort of conceptual organization, witnessed by observable patterns in language. If insertion into an L node needs to be controlled then it is better within the theory of Moreover, there is evidence that other languages use the same form in both lexical and functional positions. If this is so, they too would require some sort of control placed into the grammar because not all lexical forms can work as classifiers in Vietnamese nor all nouns as prepositions in German. This control mechanism ne eds to be better defined in theories such as Distributed Morphology. Moreover, if this mechanism is understood, it can help us understand grammaticalization from a synchronic perspective. It seems unreasonable to believe that grammaticalization suddenly re quires a form one day to be forever functional. Rather it seems natural that the form will go through a period of straddling the fence.


58 CHAPTER 4 GRAMMATICAL RESTRICTIONS THE ENCYCLOPEDIA AND SEMANTIC RESTRICTIONS IN PRACTICE This chapter is divided as f ollows: the Methods for Restriction returns to the idea that there needs to be inhibitors to Vocabulary Insertion. It suggests that this can be done with two filters, a grammatical filter an d a semantic filter. The next section introduces the idea of the s trict or Grammatical filter. H ow this works is then shown with the use of the proposed lexical functional head p_. The features of p_ restrict insertion as Chapter 3 found external based restrictions best. The section after shows how insertion works for p when no Root is present. Then the Encyclopedia is discussed as a component that might influence wh ich Vocabulary Item is chosen. The next section discusses the idea of a semantic filter that is powered by the Encyclopedia. This filter is responsible fo r deciding what form is best from the context given th at there is still choice for insertion at a lexical node. This is applied to the i nsertion of P forms. The next section then briefly looks at a slightly alternative view for restricting the insertion of Roots and notes that others have noticed the need to explain ostensible functional forms operati ng in lexical domains Methods for Restriction The DM model has no present way to explain why only some VIs can be inserted in Roots dominated by certain f nodes, the reason apple cannot be inserted under little p_ as was shown in Chapter 3 (2a). To explain this, specific rules must exist in the grammar. Restrictions places on Vocabulary Insertion seem best motivated external to t he Root of insertion. A proposal of this thesis is that the Vocabulary Insertion of Roots must have restrictions and that this involves two filters within DM: a grammatical filter and a semantic filter. The grammatical filter is formulated on features ex ternal to


59 the Root but the semantic filter is not as clear. Vocabulary Insertion will add meaning to the grammar in the case of l morphemes, but this meaning should not interfere with the syntax. Strict or Grammatical Restrictions on Roots If it is acce pted that l morphemes cannot operate freely under all proposed functional heads, then there must be restrictions and we should account for them. The first type of restriction is a grammatical restriction. This is the restriction that absolutely prohibits t he insertion of a particular VI into a Root node c commanded by a certain categorizing head. The Encyclopedia cannot forbid a Vocabulary Item from being inserted in any particular Root node. To explain the inability for certain Vocabulary Items to be grammatically inserted in L nodes dominated by a certain functional head, exclusionary rules must be a part of the grammar. The other way to do this would be to have homophonous forms, one as a representation of the f morpheme and the other as a VI for a l morpheme. However, as argued in Chapter 2 this is not desirable. With that being said, VI insertion into a Root is guided by rules. VI Insertion into a Root Dominated by Little p_ rules which in essence form a list of eligible Vocabulary Items for insertion. Siddiqi (2005: 111) lists example phonological licensing environments which achieve a similar purpose to what needs to be done here. For example Vocabulary Items that are intransitive cannot appear under a [+cause] little v_ head. Siddiqi licensing conditions are probably what ultimately need to be elaborated on for the overall the ory to be able to explain all cases of ungrammatical lexical VI insertion.


60 However, not as many stipulations are needed to explain P form insertion into little p_ and thus a simpler approach to VI restriction is taken here. Functional heads contain featur es as explained in 3.5, even if their identity is unclear. VIs in List B are listed with feature environments. VIs cannot be inserted in a contrary to the particular VI the functional head that licenses them, their external environment. In the case of little p_, this functional head has a [+ relational, +case] feature. This is justified by the fact that the pr imary job of prepositions is to relate two phrases together and to provide case. This is shown b y Figure 4 1 As will be explored in Chapter 5 the [+relational] feature may be the best feature to define the little p_ head. Figure 4 1. Little p_ as relational functor Given this [+relational] feature, Vocabulary Items can be relegated. This is seen by the following rule.


61 P form Insertion Rule: /in/, /on/, /up/, /off/ /P forms/ ( Root: [ + r elational ] [+, case] ) 7 features. However, they are not limited to environments with only these feature and thus can be inserted under little n_, v_,and a_ which have a [ relational] feature. The pu rpose of the n_, v_,and a_ functional heads is not to show how one phrase connects to another. It should be noted that the presence of the [+case] feature is not necessary for the core meaning of the word to be associated with a relation. This is seen by t he evidence given in Chapter 2 If I say that I am pissed off off does not grammatically in Great Britain. However, the off in piss off maintains a relational meani ng similar to away from as the command piss off means get away from me. Moreover, a P form does not have to have a [+relational] feature to keep its core meaning. The word up in the e P form can be similar, despite it appearing in a different location. If we accept Harley and Noyer (2000), the same type of rule is needed to explain the grammaticality of /sink/ cause]), which can be found in both transitive and intransitiv e environments, in comparison to /fall/ cause]), which can only be found in intransitive environments (Siddiqi 2005 for an alternate analysis). These environmental licensing rules allow for The ship sank and The ship sank the other ship but proh ibit John fell Sally Environmental licensing rules are by nature both exclusionary and underspecified. They are specified only enough to forbid the insertion of certain forms while ultimately permitting choice. This is different than Vocabulary 7 Remember the feature is in the syntax, not List B. P forms do not need this feature to be inserted and the insertion of P forms does not add this feature.


62 Insertion with f morphemes because even though Vocabulary Items are underspecified for abstract morphemes too, only one form can win and be inserted at any given f node. This is seen with the cases of true prepositions, where there is no justification for the form to correspond with an l morpheme. Insertion of Forms Directly u nder L ittle p_ Distributed Morphology makes crucial use of root out insertion, that is Vocabulary insertion must affect the verbal or nominal root before it affects any functional head adjoine Legate 1999:15 ): commanded by little p_, then vocabulary insertion occurs like any other f morpheme. The form specified for the feature(s) under the little p_ gets ins erted. This is shown by here with Figure 4 2 Figure 4 2. VI insertions for f morphemes With the structure given in Figure 4 2, t hese following rules could dictate the distribution of f morpheme prepositional forms in English. (1) a. [of] {+ Genitive} b. [at] {+Locative} [[a] {Locative} \ 8 c. [by] {+Agentive} 8 The insertion of [a] would occur when a Root is directly c commanded by p_, so figure 4 2 is not accurate for this realization of p_.


63 d. [from] {+Source} e. [to] {+Goal} f. [with] {+Associative } g. [for] {+Benefactive} h. [via] {+Path, +Source} 9 With these eight forms, we can account for the forms that represent purely grammatical prepositions in English. In the cases of the other meanings for by and with, which we see expressed in examples like bystander and withdraw we can now say this is homophony. The by in bystander is not the Agentive by but the Locational by and the with in withdraw or withhold could be argued to be a fossilization of the old more lexical with, which has the meaning al Dictionary). Such radically different meanings suggest homophony as discussed in Chapter 2 directly c commanded by a p_ functional head, then insertion will be like this. Figure 4 3. oot filled up Insertion of a null item could be said to happen in most cases. 10 Moreover, the 9 It is uncertain if via constitutes an l morpheme in English. It appears in many Latinate borrowings such not been established.


64 insertion of a schwa in the case of l morphemes such as long board and side could be said to occur in this relationship when there is a [+locational] feature under p_ Another that forms such as home can be said to fill this spot. Thus the sentence I went home has a // insertion under little p_ while if we want home to be in the DP, little p_ will be realized: I went to my home. The inclusion of home in the P form list is possi ble and gives more ward which attaches to P forms and is a possible insertion inside the expanded structure of P given by Svenonius (2008), can of course also attach to home Ho me can also be used Relational] [ Case]). All in all it is an interesting thought given the pre dic tions of this present analysis. morphemes. Moreover, the presence of a Grammatical Filter explains why some VI insertions at L nodes violate the grammar and others do no t. It does not, however, explain why some insertions are preferred over others. This requires the final component of DM, the Encyclopedia, to perform a kind of semantic check. This creates a second filter, the Semantic Filter. The Encyclopedia The E ncyclopedia is not a well developed idea in Distributed Morphology. Harley top ). 10 DM does not req uire a null insertion because not every node needs to be filled by a VI.


65 phonological expres t remains an open question how the Ency clopedia effects [grammatical well formedness] on semantic interpretation Since DM has done away with lexical semantics, the Encyclopedia has been responsible for associating non linguistic knowledge and certain meanings with exponents (Harley and Noyer 1999: 2). It is also responsible for storing the idiomatic meaning a morpheme, word, or phrase might have (Harley and Noyer 1999: 2) We might be a ble to say that the Encyclopedia [has an] entry for the Root (or Roots) which are involved Thus every Root must have an entry in the Encyclopedia as morphemes are typically not idioms, but l morphemes are alway s idioms This association could also have the power to influence the insertion or permanence of certain Vocabulary Items under lexical functional heads. However, according to Siddiqi (2005: 8): The Encyclopedia plays no part in the construction of a derivation, nor does it serve any role in determining whether the derivation is well formed. Rather, it simply assesses the interpretability of the sentence. rless green thoughts sleep furiously to be possible grammatically and yet senseless according to eems to agree make sentences we would never use (2000: 2). As Chapter 3 argued, in the case of P forms, it is much harder to imagine sentences such as, *I got a bad grade location my test. This sentence does not seem grammatical as the one given by Chomsky is. Also it


66 is arguably harder to produce. Thus it seems we need the grammatical restricti on argued for in the previous section if P forms are to be viewed as l morphemes. However, this grammatical restriction cannot predict degrees of grammaticality. There must be some freedom for different forms to appear in the same place. Semantic Checks on Roots A semantic check is something that conceptually happens during/after Vocabulary Insertion. This happens as phonological information gets interpreted by the Encyclopedia. consu lted subsequent to the output of PF/LF, which we abbreviate simply as meaningless could be noticed and in theory reevaluated because the Encyclopedia cannot meaningfully inte rpret them. In other words what would prevent the Encyclopedia from initiating a redo at the Vocabulary Insertion stage. The information at LF would not change. This view like the grammatical filter of the previous section also goes against the idea that t because while all available candidates could fill a L node slot, insertions are practically screened after the fact. However, an Encyclopedic semantic check does not support the notion that the via insertion rules can be inserted. Thus many forms can be sensibly inserted in the same place, allowing for competition. There is a choice (Harley and Noyer 1999:5). Thus the examples in (1) are grammatical but (1c) is not normal. (1) a. Nine passengers aboard flight 987 became ill. b. Nine passengers on flight 987 became ill. c. *Nine passengers in flight 987 became ill.


67 Many would argue that (1c) is ungrammatical but according to our account so far this is not due to features under little p_. The meaning of these three P forms should be alright with a [+locative] feature. Thus the problem has to do with the something else the greater context and word associations. A similar context and a different object NP yield a more acceptable reading as seen in (2). (2) a. Nine passengers aboard the plane became ill. b. Nine passenger on the plane became ill. c. ? Nine passengers in the plane became ill. Pragmatics might be responsible for making in acceptable in certain cases. Also P forms can be inserted under functional heads where the reading will be senseless. (3) a. The blue downs down the orange ups were o uted. b. The blue ups up the orange downs were offed. c. The blue downs along the orange ups were outed. anything. All in all these examples are meant to show that there is something outside the formal grammar that regulates the interpretation of Vocabulary Items in sentences and renders things meaningful. This way there can be many options for which P form will be inserted but perhaps only a few will make sense. Thus we do not need a huge list of grammatical features to predict the grammatical insertion of every P form Licensing Restriction s on the VI, not the L morpheme Taking the externally motivated restrictions idea further, we can say that licensing restrictions have nothing to do with features in the Root or under the categorizing head per se. First we must partition the idea of an l morpheme. 11) distinction that as node Root as exponent Root as category free d


68 we can attribute all licensing restrictions to the process of Vocabulary Insertion. We can L node node realized by a root that non deterministic Vocabulary insertion interpretation can take the greater phrase context into account (2000:4). In other words, when the phonology is inserted, the information goes to the Encyclopedia where real world knowledge is stored and meaningful choic es can be made. If a phonological string is placed under a little p_ node that has no semantic relation to little p_, the result will be a serious conceptual error, prompting a different string to be placed under the L node. They use this form of checking to differentiate from awkward sentences to fully ungrammatical. If grammar were purely feature checking and or lists of stipulations, it would be more difficult to say why one occurrence is more ungrammatical than another. However if it is the encyclopedi a or real world knowledge that is informing our l morpheme choice, then the level of ungrammaticality can be better differentiated or looked at as a matter of degree. compatible with the generated structu re may be inserted at the l he consequences of this are desirable As Acquaviva (2008: 12) predicts : [This] allows us to envisage cases where the same Vocabulary item is inserted in both L and F nodes. This seems the correct characteri zation for what are often called 'semi lexical' categories, which arise when a 'lexical' open class morpheme has an additional use as a grammatical morpheme. Therefore this thesis does not claim or need to claim that there are different


69 different heads. The restriction will come on the realization of that Root. It could be argued that it is possible for the meaning of most P forms to conceptually receive more perspectiv es than non P forms. In other words, there is more to the understanding of words such as house, walk and green than a sole relational understanding. This is not accidence as it mirrors what we know about our conceptual abilities. We are able to parse obj ects and actions into many items while the way we can conceive a relationship between them is more limited. Thus, while the exact content of Roots is currently a matter of debate within the theory of DM, this paper only wishes to provide possible justific ations for restricting the list of l morphemes that can possibly be c commanded by a little p_. Concluding Remarks The probability of an l morpheme being represented by a particular VI depends on the functional head that licenses it. This probability is not accidence but the result of some sort of organization, witnessed by observable patterns in language. This organiz ation could theoretically be the result of restrictions. This means that lexical VIs are not simply a disorganized mass of phonological strings kept in an abstract list. Rather they can be organized via their licensing environments. A cause for organizatio n in List B does not seem to be a negative consequence of this proposal. Forms often get put into groups that get discovered as language specific categories or classes. This is based on the grammatical structure of the language and does not depend on encyc lopedic or semantic knowledge per se. The proposal is that there are at least two types of restrictions placed on the


70 particular VI from getting considered under a certain functional head and a Semantic Filter helps decide appropriate via the greater context. The Grammatical restriction is language specific and explains how P forms can land under little p_ and other functional heads while other VIs cannot. Moreover if no commanded by little p_ then VIs for f morphemes will compete for insertion. If two P forms both pass the grammatical filter, then the appropriateness of insertion is left to the Encyclopedia. This creates a kind of Semantic Filter which in theory can evaluate the meaningfulness of Vocabulary Items in a phrase compared to the greater context of the sentence and discourse. This looks to be enough to explain the distribution of all prepositions in English.


71 C HAPTER 5 P FORMS AS PARTICLES AND FINAL REMARKS This chapter is organized accordingly: it begins by proposing that particles are a p_ head in a vP shell, then it talks about the role of the particle in the vP shell and gives an established structural basis for its pos itioning and movement, after that it argues that the label Particle Phrase can be abandoned in favor of a functional head t hat has an identifiable feature and i t also further di scusses the nature of this head, the next section proposes modifications to the established vP st ructure and particle shift, then the section after explains how movement might be accounted for after the se modifications have been made, and the conclusion summariz es the main points of this work and explicates some of the uncertainti es remaining. Accounting for Particles in the System According to this account of prepositions and particles, p articles are an instantiation of a P form inside a verb phrase, vP shell. This paper wi ll not get into what type s of possible little v_ heads would render the appropriate structure, as there probably is more than one variety. Rather with the knowledge of licensing environments, it will briefly discuss how insertion could work via a generic v_ head dominating a c asele ss p_ head. This Caseless p_ head will be viewed as a type of relater. This is because this head adds a relational aspect/aktionsart between the verb and its object. Caseless p_ will be the head proposed to be responsible for categorizing a P form inside a vP shell. This analysis does not intend to trump any previously argued for structures that could host a P form in the vP shell. Rather this head is merely necessary to represent a previously proposed feature, unify the present analysis, and suggest ways t hat modifications to the existing structure may be helpful.


72 P forms in the vP S hell When P forms get inserted into a vP shell, they direct or relate the action of the verb to an endpoint Observe E xample (1a). (1) a. Mary drank the water down In this s entence down result to the action ( Embick 2004 for more ). Co mpare this with the sentence in (2). (2) a. Mary drank up the water Mary drank the water gone As in (1a), the particle in (2a) also provides a result but without a literal direction In this case up which is often described as a perfective particle ( Lindstromberg 1998 specifically and Brinton 1985 for more generally on this), gives us the interpretation water is no mor e The drinking process is finished. The endpoint results in the water being inaccessible. Even though in this case up does not literally mean the water had to travel from a low to high place to reach a new location, i t is hard to say up has lost its core meaning in this context. Conceptually, up makes sense as the direction to a location that is out of reach Things that move up get out of reach. In other words one ble of making Likewise, the water is now out of reach. the result of the object of the action. Up does not have to have this role though as the sentence She drank the water up from the stream does not emphasi ze that the water is no more but rather shows the direction the water fro m which the water entered the drinkers body. What is similar is that the particle relates the object of the verb to some result or direction. Moreover, while we can accept that there may be different types of particles, it seems evident from the previous chapters that they operate similarly structurally. If we accept the Extended


73 Verb Phrase Hypothesis, EVPH, (Nicol 2002), then before movement the vP plus particle structure might look like Figure 5 1 which is based on Harley reasoning (2006 ) Figure 5 1. Particle verb base s tructure The hierarchical structure in Figure 5 1 is based on the tree Harley gives in (2006:10) Moreover, Harley includes a focus functional head In Figure 5 1 P a rtP stands for Particle Phrase and P a rt stands for the particle head. The l morpheme dri nk will need to move to v_ to get categorized as the categorization assumption of Embick and Noyer (2005) demands Harley has a Focus Phrase above the Root Phrase and this is where Harley (2006) has the DP Object move to get focus Figure 5 2. Particle appearing after verbal o bject.


74 In Figure 5 2 the Root dri nk moves through F, picks up the features there, and then moves to the sister position of v . The DP then moves to Spec FP to check accusative case. In this way we can derive the surface order Mary drank the water up. In cases where up precedes the Object DP, Harley (2006) has the Particle incorporate with the dri nk and then move. Th is is seen with Figure 5 3 where we see up merge with dri nk before moving to get focus The motivation for this movement is all based upon feature checking to accord with minimal syntax and the realization of strong features. This same motivation will be maintained but this paper does not intend to support or deny this motivation for movement. Figure 5 3. Particle before the DP object The structure of Figure 5 3 then gives you the surface order Mary drank up the water. It should be noted that tense is not accounted for in this model. Thus after the Particle to be categorized by little v_. This paper does not attend to argue for or against the


75 types of movement or the motivation for them. Rather this structure gives us an established place to insert the findings of this thesis. Replacing the Particle Phrase: As Harley is not concerned with the relationship betw een particles and prepositions, the existence of the Particle Phras e can be thought of as a c onvenient label There is no reason for it to exist other than the fact that Eng lish has things we like to call particles. There is no prop osed feature responsible for the creation of a particle head. This paper will call the structure responsible for the particles as a relational little p_ or a Caseless little p_. This head is created by a [+ relational] feature. This feature which represents several varieties of direction and aspect, helps explain the connection betwe en the meaning of the P form, the ver b and object, and the result of the verb and object The relational feature is generic for the types of relationships the particle and verb can have, Aspectual or Aktionsart (Brinton 1985 for arguments between the two). The difference between this head and the p_ head that creates prepositions is that this head is not a case assigner. The same rules that prohibited [ relational] forms from being categorized by little p_ would naturally apply here as well. Furthermore, it should be explicated that it is was the presence o f [+case] features that allowed for the insertion of the f morphemes on Table 2 2 in Chapter 2 those forms that do not get used as particles. Moreover, this relational p_ head could be called something else, but it is not a prepositional hea d and the term Particle head tells us nothing of about the function of the form inserted under it.


76 Changes to the Base Structure: Figure 5 4 specifies where particle(s) can be inserted to be better account for the data presented in the previous chapters. In other words, it posits a feature as being responsible for insertion. Figure 5 4. The b ase s tructure for vP Shells with a relational p_ head select oot. This structure adds a pP as a replacement to the Particle Phrase. This structure commanded by a [+relational] feature. Thus, in this way Particle insertion complies with this papers main proposal. This is a good incorporating after Moreover, if you have a compound particle construction, one does not need to conjoin two functional heads, as you would if you just had a Particle Phrase to w ork with. It does not seem common to compound the phonological representation of functional heads, determiners, number, tense etc. On the other hand, if the particle phrase is compounded then we would expect something like *Mary turned on the water and off


77 that gets categorized as a verb under the premise that this movement is optional. Another analysis would be to say that the conjunction occurs higher and the most of the sec ond part is ellipsed. (1) a. Mary turned the lights on and (Mary turned the lights) off. The first problems with (1a ) is that most of the time ellipsis occurs with complete constituents and Mary turned the lights is not in this case. The second is that on and off do seem to form a constituent in the sense that John turned the lights on and off and the sprinklers too He turned them both on and off. Movement In order for the tree in Figure 5 4 to work, the movement prop osed by Harley can be modified slightly for what seems to be better results. The first stages of movement under the proposed model is shown in Figure 5 5 Figure 5 5. First m ovements Starting from the bottom up with head movement happening first the head of the the DP object moves to check accusative case. This


78 will get the surface structure Mary drank the water up. Drink get categorized as a verb and up by relational little p_. Now if the surface structure for Mary drank up the water is desired, then a second Remnant movement is proposed. Figure 5 6. Remnant m ovement. In Figure 5 6 instead of having the particle incorporate with drink and then move through F to the complement of v to get categorized, the pP waits until its DP object moves out and then moves to the complement of drink. Thus head movement happens first, drink to v ; then phrase movement, the object DP to Spec F and then pP to the complement of drink. This allows the verbal Root to later obtain tense without it appearing on the P form which would be ungrammatical drink upped The verb particle construction do es get categorized as a compound verb. All in all the structure prese nted here allows for the integration of the analysis of Chapter 2. Final Remarks The purpose of this paper is to highlight an old problem within the literature of English grammar, the dis tinction between Prepositions and Particles and their greater categorization. It also pointed out that many phonological forms responsible for


79 representing both can appear in non functional places. It argued against homonymy, as this divorces the clear mea ning relationship between up as a noun, verb, adjective, preposition and particle. To solve these problems it shows that the architecture of Distributed Morphology is suited to elegantly account for the different syntactic behavior of words such as up. We can say it is the same form with the same meaning being inserted under two different functional heads, or in two different syntactic arrangements. However, the fact that up can also be used in Nominal or Verbal frame posed a problem for DM as it would othe r theories Since Distributed Morphology only has two classes of morphemes, there had to be a way to get Vocabulary Insertions of the same form in both lexical and functional places if a homonymous explanation is rejected To do this a little p_ functional head was proposed to exist with the base features [+case] and [+relational]. This corresponded with instantiations of prepositions. This head was so forms like up could classified as Vocabulary Items that are inse rted under L nodes. If the head did not categorize a the following forms: at, by, for, from, of, to, via, and with. These forms are not deemed to be l morphemes becau se they do not appear non prepositional slots and they do not behave like l morphemes Once the list of P forms was refined and classified as a type of l morpheme, another problem arose. How do we keep other l morphemes from being categorized by little p_? To prevent unwanted Vocabulary Insertion at little p_, two filters were proposed, a grammatical filter and a semantic filter. Together they are able to explain different degrees of wellformedness and prevent unwanted forms from entering under a little p_ head. The main proposal of this work is not the actual


80 existence of these two filters, but rather that P forms are viewed as l morphemes. If insertion can be controlled in a way deemed more friendly to the main spirit of DM, then by all means it should be developed and cla rified. Furthermore, we can place a caseless p_ head in the vP shell Prepositions and Particles can be seen as the realization of the same basic functional head containing either [+, case] and different possible relational features. This head is responsible for the insertions that surface as particles in English. With all of this in place, the preposition/particle category can be better understood and so can their behavior Further research should focus on the types of relational feat ures present under little p_ in the verbal domain and how to more elegantly control for insertion restrictions within the theory of DM.


81 LIST OF REFERENCES Acquaviva, Paolo. 2008 Roots and lexicality in Distributed Morphology. Ms., University Col lege Dublin and Universitt Konstanz. Anderson, John. 1971 The G rammar of C ase: T owards a L ocalistic T heory. Cambridge: Cambridge University Press. Anderson S tephon R. 1992 A Morphous Morphology. Cambridge: Cambridge U niversity P ress Baker, Mark C 2003 Lexical categories : Verbs nouns and adjectives Cambridge: Cambridge U niversity P ress B aker, C. M, and Bobaljik D. J. 2008 Introduction to Morphology Unpublished: Rutgers and McGill. Bennett David C. 1975 Spatial and Temporal Uses of English Prepositions: An Essay in Stratificational Semantics. London: Longman Group Limited. Boe rtien, H. 1997 Left Headed Compound Prepositions. Linguistic Inquiry 4. 689 697 Cambridge: MIT Press. Borer, H. 2009 Roots a nd categories Talk given at the 19th Colloquium on Generative Grammar, University of the Basque Country, Vitoria Gasteiz. Botwinik Rotem, I. 2004. The Category P. Features, Projections, Interpretation Ph.D. dissertation, University of Tel Aviv. Brinton, Laurel J. 1985 Verb parti cles in English: Aspect or Aktionsart?. Studia Linguistica 39. 175 168. Calabrese, Andrea. 1998. Some remarks on the Latin case system and its development in Romance. In Jos Lema and Esthela Trevio ( ed s.) Theoretical analyses on Romance languages 71 126 Amsterdam Philadelphia: John Benjamins. Carlson, Gregory. 1977. Reference to Kinds in English Doctoral dissertation, University of Massachusetts Carnie, Andrew. 2007 Syntax A Generative Introduction 2 nd ed. MA : Blackwell Publishing Celce Mur cia, M., and Larsen Freeman D. 1999 The Grammar Book: An ESL/EFL course 2 nd ed Boston: Heinle and Heinle.


82 Chomsky, N oam 1970 Remarks on nominalization. In R. Jacobs and P. Rosenbaum Waltham ( ed s.) Readings in English transformational grammar 184 221. MA : Ginn. Chomsky, Noam. 1995 The minimalist program. Cambridge, MA: MIT Press. Cook, Walter 1989 Case Grammar Theory. Georgetown University Press. Croft W. 1991 Syntactic Categories and Grammatical Relations. Chicago: The Univers ity of Chicago Press. den Dikken, Marcel. 1995 Particles: On the Syntax of Verb Particle, Triadic and Causative Constructions Oxford U niversity Press: New York den Dikken, M arcel 2006 On the functional structure of locative and directional PPs. Ms: CUNY. Di Sciullo, A. M, and Williams. 1987 On the Definition of Word. Cambridge: MIT Press. Elenbaas, M. 2007 The synchronic and diachronic syntax of the English verb particle combination. LOT Dissertations in Linguistics. Embick, D avid 2004 On the Structure of Res ultative Participles in English. Linguistic Inquiry 35( 3 ) 355 92. Embick, David and Rol f Noyer 2005 Distributed Morphology and the Syntax Morphology Interface. To Appear in Gillian Ramchand & Charles Reiss (e ds.) The Oxford Handbook of Linguistic Interfaces 1 28 Oxford: O xford University Press Fillmore, Charles J. 1976. Frame semant ics and the nature of language. In Annals of the New York Academy of Sciences: Conference on the Origin and Development of Language and Speech 280 20 32 New York Academy of Sciences: New York. H alle, Morris and Alec Marantz. 1994 Some Key Features of Distributed Morphology. MIT Working Papers in Linguistics 21 275 288. Harley, H eidi and R olf Noyer. 1999. State of the Article: D istributed Morphology Glot International 4( 4 ) 3 9. Harley, Heidi and Rolf Noyer. 2000 Formal versus Encyclopedic Properties of V ocabulary: Evidence from n ominalizations. In B. Peeters (e d.) The lexicon encyclopedia interface Elsevier: Amsterdam.


83 Harley, H. 2006. The morphology of nominalizations and the syntax of vP. Ms. University of Arizona Harley, Heidi. 2008 Compounding in Distributed Morphology. To appear in R. Lieber and Stekauer, P. (eds.). The Oxford Handbook of Compounding. Haspelmath, Martin. 1999 Why is g rammaticalization irreversible? Linguistics 37( 6 ). 1043 1068. Haugen, E. 1 950 The analysis of linguistic borrowing. Language 26.2, 210 31 Hudson, R. 2000. Grammar without Functional Categories. In R. Borsley (ed.), The Nature and Function of Syntactic Categories 7 35 Academic Press. Keizer, M. E. 2008. English prepositions in Functional Discourse Grammar Functions of Language 15( 2 ). 216 256 Koopman, Hilda. 2000. Prepositions, postpositions, circumpositions and particles: The structure o f Dutch PPs. In Hilda Koopman (e d.). The syntax of specifiers and heads 204 60 London: Routledge. Kortmann, Bernd & Knig, Ekkehard 1992 Categorial reanalysis: the case of deverbal prepositions. Linguistics 30 671 697. Legate, J. 1999 The morphosyntax of Irish agreement. In K. Arregi, B. Bruening, C. Krause, & V. Lin (eds.). Papers on Morphology and Syntax, Cycle One: MIT Working Papers in Linguistics 33 219 240. Lieber, R. 1992 Deconstructing Morphology. Chicago: University of Chicago Press. L iles Bruce L. 1987 A Basic Grammar of Modern English Englewood Cliffs, NJ: Prentice Hall Lindstromberg Seth. 1998 English Prepositions Explained. Philadelphia: John Benjamins Publishing Company. Marantz, A. 1 997 Privacy of Your Own Lexicon. UPenn Working Papers in Linguistics 4 201 225. Mithun M. 199 1 Active/agentive case marking and its motivations Language 67(3). 510 46 Nicol, F. 2002. Extended VP shells and the verb particle construction. In: Deh, N. et al (eds.). Verb Particle Explorations Berlin/New York: Mouton de Gruyter.


84 Prepositions and Particles in English: A Discourse Functional Account New York: Oxford University Press. O nline Etymological Dictionary. 2010 Douglas Harper: http://www.etymonline.com/ Roeper, T. 1999 Leftward Movement in Morphology. In V. Lin, C. Krause,B Bruening and K. Arregi (e ds), Morphology and Syntax C ycle Two, MIT W orking P aper in L inguistics 34 35 67. Ramat, Paolo 1992 Thoughts on degrammaticalization. Linguistics 30 549 560. Selkirk, E. O. 1995 The prosodic structure of function words. In K. Demuth & J. Morgan (e ds.). Signal to syntax. Hillsdale, NJ: Erlbaum. Siddiqi Daniel. 2005 Distributed Morphology Without Secondary Exponence: a Local Account of Licensing Thematic Licensing of Vocabulary Items and Strong Verb Alternations Coyote Papers 14 108 143 Smith, Jeremy 1996 An Histori cal Study of English: Function, Form and Change London: Routledge. Svenonius, Peter. 2007 Adpositions, particles and the arguments they introduce. In: E. Reuland T.Bhattacharya & G. Spathas (eds.). Argument Structure 63 103 Amsterdam: Benjamins Svenonius, P eter. 2008 Spatial P in English CASTL, University of Troms Taylor, J. 1993 Prepositions: Patterns or Polysemization and Strategies of Disambigu ation. In C. Zelinsky Wibbelt (ed.). Natural Language Processing: The semantics of Preposition s 151 175. The Hague: Mouton de Gruyter. Trias S, P. 2009 Compounds in Distributed Morphology: pros and cons Seminario LyCC, CSIC. Whitney, W. D 1881. On mixture in language. Transactions of the American Philological Association 1 26. Zaenen, A., J. Maling & H. Thrinsson 1985 Case and gramm atical functions: The Icelandic passive. Natural Language and Linguistic Theory 3(4). 441 483.


85 BIOGRAPHICAL SKETCH Being from Melbourne Florida, I always preferred to be outside. Woods, palmetto bushes, canals, lakes, the Indian River and the sea, school was quite an inconvenience compared to these. I can remember set ting fire to my fifth grade English book. It is dream like remembering just how one really gets from A to B. I ended up m ajoring in English and m in oring in Psychology to the surprise of a few Moreover, a fter I graduated from the Univ ersity of North Florida in 2005, I traveled again to Europe Perhaps some things had changed but I still needed to explore, to understand and to render even if it be just a fraction of what lies upon the surface. Upon returning, I worked both in a home improvement store and as an educator in the Jacksonville Public schools. Manual labor is not all that bad, but it can be a powerful motivator. After quitting the retail job, I worked as tutor in a learning center based in Fruit Cove outside of Jacksonville There I got to work with all sort s of kids and with begin ning college students. With kids and disabilities there are many challenges but o verall i t was a great experience and my boss was wonderful However, I knew the job was transient. I had taken an introductory class in Linguistics my senior year of college and cont inued to be interested in it, and I had read some books about teaching English abroa d. However, I felt unqualified and uncertain Thus when I received my acceptance letter to UF, it seemed like the next p hase of my life had been given clear direction. There is a joy t o knowing y ou have a purpose and that that purpose is mildly fugacious Likewise as this work comes to fruition a new directio n and phase will hopefully begi n T hank you for taking the time to consider these small things