Home / Archive / 2006 / Concord and DP Structure in Bafut / dippArticle-fast.xml


Document Actions
<?xml version="1.0" encoding="UTF-8"?><document><element name="title"><value> Concord and DP Structure in Bafut This paper has been updated from an earlier version which appeared in JWAL 2000/2001). Research for the writeup was sponsored by the Alexander von Humboldt Research Fellowship. I am grateful to the managers of the foundation for all forms of assistance and to the University of Buea department of linguistsics seminar for a stimulating discussion of the major ideas in the paper.</value></element><element name="body"><value> Introduction &lt;1&gt; This article accounts for DP-internal concord phenomena in Bafut Bafut is a Grassfields Bantu language spoken in the North West Province of Cameroon. Number of speaker: about 125,000. while at the same time characterizing the internal structure of the Bafut DP in the light of various word order patterns displayed. In the Bafut DP, all arguments and satellites follow the head noun (except when focused) and agree with it in number and gender/class. The following examples illustrate the various word order and concord patterns. In these, and subsequent examples, figures in the glosses indicate noun class. (1) a. ø-lúʔú w-á b. nɨ̀-bɔ̀ʔɔ̀ ny-á c. fɨ̀-ŋkōbɨ̄ fy-á 1-spoon 1-the 5-pumpkin 5-the 19-statue 19-the the spoon the pumpkin the statue (2) a. ø - lúʔú n -tsɨ̀m b. nɨ̀ - bɔ̀ʔɔ̀ nɨ́ -tsɨ̀m c. fɨ̀ - ŋkōbɨ̄ fɨ́ -tsɨ̀m 1-spoon 1-every 5-pumpkin 5-every 19-statue 19-every (3) a. ø - lúʔú w -íì b. nɨ̀ - bɔ̀ʔɔ̀ n -íì c. fɨ̀ - ŋkōbɨ̄ f -íì 1-spoon 1-that 5-pumpkin 5-that 19-statue 19-that (4) a. ø - lúʔú ɡh -á b. nɨ̀ - bɔ̀ʔɔ̀ n -á c. fɨ̀ - ŋkōbɨ̄ f -á 1-spoon 1-my 5-pumpkin 5-my 19-statue 19-my (5) a. ø - lúʔú yím -fùʔù b. nɨ̀ - bɔ̀ʔɔ̀ ní -fùʔù 1-spoon 1-white 5-pumpkin 5-white (6) a. fɨ̀ - ŋkōbɨ́ f -ɨ́ m - fɔ̀ b. nɨ̀ - bɔ̀ʔɔ̀ n -ɨ́ m - fɔ̀ 19-spoon 19-AM 1-chief 5-pumpkin 5-AM 1-chief AM in all Bafut examples stands for Associative Morpheme; the particle which Carstens (1991, 2000) and others gloss as 'of'. See section 3.5 for a detailed treatment of this grammatical element. a chiefs statue a chiefs pumpkin &lt;2&gt; The structure of the DP in Bantu languages and concord of the type exhibited in the Bafut data above has, to the best of my knowledge, not received extensive treatment. The only treatment of DP structure in Bantu and the type of concord illustrated above that I know of is contained in Nkemnji (1995) and Carstens (1991, 2000). Nkemnji (1995) proposes to analyze agreement in Nweh DPs as involving a Spec-Head relation. He posits a new phrase: class phrase (ClassP) and projects a genitive/operator phrase (G/OP) above NumP. Three XP movement processes (NP, ClassP, NumP movement) and three head raising processes (Num 0 , G/O 0 and D 0 ) then account for observed word order facts. In Carstens analysis, based on examples from Swahili, she assumes N 0 -to-Num 0 -to-D 0 - raising for Bantu akin to the type observed in Romance languages. She then shows that the more articulated feature-checking theory developed in Chomsky (1995) provides an account of Bantu DP concord since checking relations are more numerous in this framework and are intrinsically symmetrical. &lt;3&gt; Our aim in this paper is not to evaluate either analysis. We will however compare Bafut facts to Kiswahili since Carstens N 0 -to-Num 0 -to-D 0 raising has serious implications for Bafut. We will demonstrate that although Bafut is a Bantu language (like KiSwahili), it does not allow raising of the complex head [ Num N 0 + Num 0 ] to D 0 . This because Bafut has overt determiners (which Swahili lacks) that directly license a referential feature in D 0 thus blocking the Swahili type N 0 +Num 0 -to-D 0 raising in Bafut. In Swahili, N 0 +Num 0 -to-D 0 derives the unmarked noun-initial word order generally attested in Bantu languages. Since the noun does not raise to D 0 in Bafut, in order to derive the unmarked noun-initial word order, we propose that after N 0 -to-Num 0 - raising, NumP further moves to [Spec,DP]. Despite this difference in the structures of Bafut and Swahili, the paper shows that Chomskys (1995) checking theory accounts for the concord facts in Bafut in much the same way as it does in Swahili: DP-internal concord phenomena in both languages is accounted for in terms of symmetrical checking relations between heads and their specifiers, between two adjoined heads and between an adjoined head and the specifier of its host. &lt;4&gt; The discussion is organised as follows: first we describe the internal structure of the DP in Bafut focusing on the distribution of the noun realtive to determiners and demonstratives in section 1, adjectives and genitive nouns in section 2. The second of the discussion then provides an account of concord between the noun and its various modifiers.We conclude the paper by examining the intriguing relation between a head noun a genitive modifier. DP-Internal Structure &lt;5&gt; In this section, we intend to discuss the distribution and status of various noun satellites and arguments attested in the Bafut DP. Base Structure &lt;6&gt; The simplest, most natural and common types of the DP in Bafut show the noun in initial position and the definite determiner in final position. This order is illustrated below using DPs that contain various types of satellites and arguments of the head noun. (7) a. nɨ̀-bɔ̀ʔɔ̀ 5-pumpkin b. nɨ̀-bɔ̀ʔɔ̀ ny-á 5-pumpkin 5-the the pumpkin c. nɨ̀-bɔ̀ʔɔ̀ ni-fùʼù ny-á 5-pumpkin 5-white 5-the the white pumpkin d. nɨ̀-bɔ̀ʔɔ̀ n-ɨ́ m-fɔ̄ ny-á 5-pumpkin 5-AM 5-chief 5-the the chiefs pumpkin e. nɨ̀-bɔ̀ʔɔ̀ n-ā ny-á 5-pumpkin 5-my 5-the that pumpkin of mine (which we have been talking about) f. nɨ̀-bɔ̀ʔɔ̀ n-ɨ́ m-fɔ̄ ní-fùʔù ny-á 5-pumpkin 5-AM 1-chief 5-white 5-the the chiefs white pumpkin g. fɨ̀-ŋkóbɨ́ f-ɨ́ m-fɔ̄ f-ā fi-fùʔù fy-á 19-statue 19-AM 1-chief 19-my 19-white 19-the that white statue of mine which represents a chief &lt;7&gt; In these examples, the head noun plus its satellites and arguments all precede the definite determiner. The following general word order emerges: N – (Ass DP)– (Poss. Pron.) – (AP) – (D) &lt;8&gt; Let us focus first on the structure of a simple type of DP like the example in (7b). I follow Abney (1987), Ritter (1991), Carstens (1991, 2000), Picallo (1991), Siloni (1997) and others in assuming that a mid-level functional projection - FP intervenes between DP and NP. Carstens (1991, 2000) identifies this FP for Bantu type languages as NumP. Assuming these views to be correct, the DP in (7b) would have the following structure: (8) &lt;9&gt; To account for the noun-initial word order, one must assume movement of either the noun or the determiner. Movement of the determiner to a position following the noun is ruled out on theoretical and conceptual grounds; syntactic lowering of this type is not a desirable solution in that it would not be structure preserving in the sense of Chomsky (1986a). This leaves us with movement of the noun. In her derivation of a similar noun-initial word order in another Bantu language - KiSwahili - Carstens (1991) proposes that N 0 first raises to Num 0 and the complex [ Num N 0 + Num 0 ] then raises and adjoins to D 0 . While the first step of the derivation (N 0 -to-Num 0 -raising to check the nouns number feature) is available in Bafut, the second step is obviously not. This difference between Bafut and KiSwahili is due, principally, to the presence versus absence of overt determiners in the two languages. While Bafut has overt determiners, KiSwahili does not (cf. Carstens 1991:89). Compare the Bafut example in (7f) repeated below as (9a) with its Swahili counterpart. (9) a. BAFUT b. SWAHILI nɨ̀ - bɔ̀ʔɔ̀ n-ɨ́ m - fɔ̄ ní-fùʔù ny-á Boga jeupe la mwami 5-pumpkin 5-AM 1-chief 5-white 5-the Pumpkin white of chief the chiefs white pumpkin the chiefs white pumpkin (Hassan Adam, p.c.) &lt;10&gt; Assuming the distribution in the examples above, the derivations would proceed as follows I am not specific about the internal structure of NP in the example structures since details of this internal structure will be presented in section 2. . (10) a. BAFUT b. SWAHILI &lt;11&gt; In Swahili, the head noun raises out of NP to Num 0 and subsequently to D 0 which is empty. These movement processes produce the correct word order. In Bafut, the head noun equally raises to Num 0 . D 0 however contains an overt determiner and this poses a problem for further raising the noun from Num 0 to D 0 . Notice that we cannot claim that the noun raises and adjoins to the determiner for in a structure in which the noun is modified by an adjective followed by the determiner, we would be required to left-adjoin both noun and adjective to the determiner in order to derive the correct N-Adj-Det surface word order. This would be a rather odd kind of adjunction unprecedented in the literature. The situation becomes even more complicated if a genitive NP, an adjective and a determiner modify the head noun. We would have to adjoin the head noun plus all the satellites to the determiner in violation of standard assumptions about phrase structure. &lt;12&gt; In order then to obtain the noun-initial and determiner-final word order in Bafut while preserving standard assumptions about phrase structure, we propose that subsequent to N 0 -to-Num 0 raising, NumP moves to [Spec,DP]. As will be argued in section 3, this movement is motivated by the need to check class and number features on the determiner. This movement is schematized in (11) below. (11) &lt;13&gt; The proposed analysis of the Bafut DP structure outlined above is also relevant in explaining the marked word order below in which the pre-N distribution of the determiner correlates with a contrastive focus reading. (12) a. nɨ̀bɔ̀ʔɔ̀ ny-á b. ny-áà nɨ̀bɔ̀ʔɔ̀ 5-pumpkin 5-the 5-the(Foc) 5-pumpkin the pumpkin the pumpkin(already mentioned) &lt;14&gt; In an approach which recognizes [ Num N 0 + Num 0 ] raising to D 0 , we do not expect the distribution in (12b) except we want to claim that after [ Num N 0 + Num 0 ] adjunction to D 0 , a movement which results in a complex head, the determiner further extracts and moves to pre-N position alone; a view that is not compatible with Traviss (1984) Head Movement Constraint. On the other hand, in the proposal I have outlined above, the distribution in (12b) is straightforward: after NumP raises to [Spec,DP], the determiner further raises to a position above DP. The only problem one needs to resolve in this approach is to characterize this position to which the determiner raises. &lt;15&gt; The distribution and interpretation of the determiner lead us to propose that it raises to the head position of a Focus Phrase (FocP) which contains the DP. Observe first of all, that the form of the determiner varies with its distribution. In post-N position (12a), it has a short vowel while in pre-N position (12b), it has a long vowel. Secondly, the semantic interpretation of the utterance differs depending on whether the determiner is in initial or final position. The example in (12a) for instance is understood as an expression in a context in which the existence of no other pumpkin is presupposed. Its counterpart in (12b) on the other hand can only be interpreted with contrastive reference to the existence of a pumpkin already mentioned in the discourse as opposed to others not present in the domain of discourse. The phonetic form of the determiner therefore correlates with its interpretation. It is exclusively in those contexts in which the determiner is in initial position and its vowel is long that it receives a contrastive focus reading. Our hypothesis is that lengthening of the vowel is a consequence of stress placed on the determiner when it is focused This suggests that in Bafut focusing involves a phonological process (stress assignment) and a syntactic process (movement to phrase-initial position). . In non-focus contexts, NumP moves to Spec-DP leaving the determiner in final position. In contexts involving contrastive focus, the determiner subsequently raises to the head of FocP where it surfaces in initial position and, in addition, receives phonological stress. This movement to Foc 0 , we assume following Brody (1990), is motivated by the need to check a [+FOCUS] feature. The movement of the determiner to Foc 0 is outlined in (13). (13) &lt;16&gt; This proposed analysis involving the projection of a focus phrase is not entirely new in the literature. It recalls that proposed in Rizzi (1997) for the CP domain and in Biloa (1991) for focus constructions in Tuki Tuki is a Bantu language spoken in the Centre Province of Cameroon. . Similarly, Androustopoulou (1997) proposed a FocP for modern Greek DPs. Nkemnji (1995) also proposes a FocP structure to account for the pre-N distribution of possessive pronouns in Nweh. In addition to these, the proposal adds to the structural parallels between clauses and DPs as FocP parallels the CP in clauses. The Demonstrative &lt;17&gt; Like the determiner, demonstratives follow and agree with the head noun. In the unmarked noun-initial word order, a demonstrative can co-occur with a determiner in the same DP. In this case, the demonstrative precedes the determiner as in (15b). The demonstrative can also occur in pre-N focus position as in (15c). However, although both the demonstrative and the determiner can co-occur in post-N position, both cannot co-occur in pre-N focus position as shown in (15d-e). Where focus is necessary, only the demonstrative can occur in pre-N position as the difference between (15f) and (15g) shows. (15) a. fɨ̀ŋkóbɨ́ fìí 19-statue 19-that that statue b. fɨ̀ŋkóbɨ́ fìí fyâ 19-statue 19-that 19-the that statue (which we have mentioned) c. fìí fɨ̀ŋkóbɨ́ 19-that(Foc) 19-statue that statue (opposed to any other) d. *fìí fyâ fɨ̀ŋkóbɨ́ 19-that(Foc) 19-the 19-statue e. *fyáà fìí fɨ̀ŋkóbɨ́ 19-the (Foc) that 19-statue f. fíí fɨ̀ŋkóbɨ ́ fyâ 19-that 19-statue 19-the that statue over there (mentioned) opposed to any other g. *fyáà fɨ̀ŋkóbɨ́ fìí 19-the 19-statue 19-that &lt;18&gt; These facts concerning the distribution of the demonstrative relative to the head noun and the determiner raise two interesting questions: what exactly is the categorial status of the demonstrative and how should it be represented structurally? &lt;19&gt; A priori, one would expect demonstratives to be a subclass of determiners. However, facts from the data suggest that treating demonstratives as a subclass of determiners misses out a number of minor but significant differences. First of all, the fact that a demonstrative can co-occur with a determiner in the same post-N position as in (15b) suggests that they do not belong in the same syntactic category. Secondly, the demonstrative is distinguishable from the determiner in that the former (i.e. the demonstrative) may stand alone (i.e. without a head noun), while the latter cannot. The examples in (16) illustrate this point. (16) a. Mə́ lɔ̀ɔ̀ nîi b. *Mə́ lɔ̀ɔ̀ nyâ I want that I want the I want that one &lt;20&gt; In addition to the above arguments, the form of the concord morpheme on the two categories is different: in the examples in (15), fy. for the determiner and f. for the demonstrative. The distinction between the determiner and the demonstrative is obvious Related studies (e.g. Carstens (1991), Giusti (1993a,b), Brugè (1996) Bernstein (1997) and Dryer (1992) observe that in many languages the demonstrative is adjectival in nature, exhibiting the full range of adjectival inflection and often occupying the position typical of adjectives The demonstrative in Bafut does not exhibit adjectival inflection, does not occur in the same position as adjectives and, rather, regularly co-occurs with adjectives in the same construction. . At the same time, the intuitive relationship between these categories is also clear: both the determiner and the demonstrative specifier a particular object (noun) in the domain of discourse. So, although they behave differently, there is no denying the fact that their semantic/pragmatic roles are very similar. Based on these, we propose to split the DP into an article phrase (ArtP) which bears the definite determiner and a demonstrative phrase (DemP) which contains the demonstrative. However, in order not to multiply phrase types unnecessarily, we will continue to use DP to refer to ArtP. In this proposal, the examples in (15) are derivable from a structure such as (17) below. (17) &lt;21&gt; Deriving (15a) and (15b) is straightforward: N raises to Num 0 and NumP moves to [Spec,DemP] through [Spec,DP]. In (15c) the demonstrative further raises to Foc 0 . The examples in (15d) and (15e) are ungrammatical because for reasons of economy: there is only one focus feature to check (as we will outline in greater detail in section 3) and so movement of both the definite article and demonstrative is ruled out. In (15f), the demonstrative raises to Foc 0 without violating any locality conditions on movement but in (15g), the determiner cannot raise across the demonstrative because this would violate the HMC. The Structure of NP &lt;22&gt; The NP in Bafut generally comprises the noun, its arguments (genitive NP modifiers) and adjectives. In this section, we discuss the distribution of each of these vis-à-vis the head noun. The Position of the genitive noun &lt;23&gt; In the surface word order, the head noun precedes all genitive modifiers. I illustrate this distribution in (18). In these examples, AM means Associative Morpheme. For a discussion on the status of this element, see section 3.5. (18) a. bɨ̀lúʼú bɨ́ máŋɡyè [N - Poss] 2-spoons 2-AM woman a woman's spoons b. mɨ̀fàʼà mɨ́ tsítsà [N - Agent] 6-products 6-AM 1-teacher a teacher's products c. fɨ̀ŋkóbɨ́ fɨ́ nɨ́ɡhɨ́ɡhɨ̀ [N - Theme] 19-statue 19-AM 5-insect a statue of an insect (a statue representing an insect) &lt;24&gt; Based on the surface word order alone, it is plausible to posit a structure in which the head noun is base generated directly in its surface position to the left of the modifying genitives. Evidence from bound pronoun interpretation and the binding of reflexives however shows that in these examples, the head noun is in a derived position. &lt;24&gt; Consider once more the example in (18c) in which the modifier of insect can be interpreted in three different ways. In one reading, the one in (18c), insect is interpreted as the theme of statue. In this case, the statue represents the insect. In another reading, insect modifies statue. Insect is the possessor; the person to whom the statue belongs. The statue, in this case, does not necessarily represent the insect. In the third reading, insect is interpreted as the agent; the person who made the statue. In a framework (e.g. Giorgi and Longobardi 1991) that seeks to preserve some sort of isomorphism between form and meaning, the PS system would be required to distinguish between these interpretations in some way. &lt;26&gt; One idea which suggests how the PS system could handle this intuitive contrast in the interpretation of the genitive modifier in (18c) is the Thematic Hierarchy . In the thematic hierarchy, possessors are more prominent than agents which are in turn more prominent than themes. Prominence here is to be understood in sense of C-command. Projecting the thematic hierarchy into the PS representation, we propose a structure like the one in (19) in which the arguments of the head noun are generated in a shell above NP core. In this structure, the possessor occupies the highest structural position and the agent (external argument) occupies a position intermediate between the theme (internal argument) and possessor. (19) &lt;27&gt; In this proposal, the head noun necessarily has to raise to a position above NP in order to account for the noun-initial order especially in cases like (18a) and (18b) where the modifying noun (possessor and agent) is structurally to the left of the head noun. This movement of the noun out of NP has been independently motivated to account for the distribution of determiners and demostratives relative to the head noun. Noun movement always applies but in the case of (18c), the results of movement are invisible since the modifying noun is structurally located to the right of the head noun. The Position of Adjectives &lt;28&gt; Adjectives always follow the arguments of the head noun. In this unique position, the adjective can modify either the head noun or its argument; the difference being made by the agreement morphology on the adjective. The examples in (20-22) illustrate this distribution of the adjective. (20) bɨ̀lúʼú bɨ́ máŋɡyè bísɨ̄ɡɨ̄nɨ̀ [Adjective modifies head noun] 2-spoons 2-AM 1-woman 2-nice a woman's nice spoons (21) fɨ̀ŋkóbɨ́ fɨ́ nɨ́ɡhɨ́ɡhɨ̀ nísɨ̄ɡɨ̄nɨ [Adjective modifies theme argument] 19-statue 19-AM 5-insect 5-nice a statue of a nice insect (a statue representing a nice insect) (22) bɨ̀lúʼú bɨ́ máŋɡyè yínsɨ̄ɡɨ̄nɨ̀ [Adjective modifies possessor argument] 2-spoons 2-AM 1-woman 1-nice a nice woman's spoons &lt;29&gt; Valois (1991), Cinque (1994) and Crisma (1996) treat adjectives as parallel to adverbs, attached to various projections in DP; in Chomskys (1995) terms, they are specifiers. The unique position of the adjective in the examples above could lead one to the expectation that adjectives have a unique attachment site in Bafut; most likely, they are attached to the matrix NP. Their behavior, with regard to concord, however casts doubts on this highly plausible view. Let us take for example the case in (21) where the adjective modifies the theme argument. The number and gender/class features of this adjective match those of the theme noun. Were the adjective in this case to be attached to the matrix NP, ordinary intuition would lead us to expect it to bear concord morphology governed by matrix N (i.e. the head noun). Since concord morphology on the adjective is governed by the theme noun, we propose that each argument projects a separate NumP and an adjective modifying a noun (head noun or argument noun) is right-adjoined to the NumP containing the noun it modifies. We illustrate this structural representation in (23). For ease of exposition (space), we represent only an agent argument noun. (23) &lt;30&gt; AP 1 modifies the head noun while AP 2 modifies the agent noun. Since each adjective is structurally located within a specific NumP, the concord morphology on the adjective is governed by the noun within that NumP. This government of concord morphology will be accounted for in terms of feature checking in section 3. To derive the surface word order [N - Agent - Adj. – Det.], after the head and agent nouns raise to their respective Num 0 positions, the matrix NumP raises to the Spec position of the matrix DP as outlined in section 1.1. Possessive Pronouns &lt;31&gt; Within the DP, possessive pronouns always follow the possessed noun but precede the determiner in the surface word order. In a genitive DP, the pronoun follows the genitive noun and, depending on agreement morphology, the pronoun can be construed as modifying either the head noun or the genitive noun. The example in (24a) illustrates the distribution of a possessive pronoun relative to a single noun and a determiner. (24b-c) illustrate the occurrence of a possessive pronoun and a genitive noun. In (24b), the possessive pronoun modifies the head noun while in (24c) it modifies the genitive noun. (24) a. fɨ̀ - ŋkōbɨ̄ f -ā fy -ā 19-statue 19-my 19-the the statue of mine (the one we have been talking about) b. fɨ̀ - ŋkōbɨ̄ f -ɨ́ mfɔ́ f -ā fy -ā 19-statue 19-AM 1-chief 19-my 19-the my statue of a chief (the statue we have been talking about) c. fɨ̀ - ŋkōbɨ̄ f -ɨ́ mfɔ̀ ɡh -à fy -ā 19-statue 19-AM 1-chief 1-my 19-the the statue of my chief (the chief we have been talking about) &lt;32&gt; In a context where the speaker makes contrastive reference to a salient possessed noun, different from any other in the domain of discourse, the possessive pronoun occurs in pre-N focus position. However, where the possessive pronoun modifies the genitive noun, in contexts of contrastive focus, the possessive pronoun is in a position between the associative morpheme and the genitive noun. In (25a), the possessive pronoun, which modifies the head noun, is in pre-N position. In (26b), it modifies the genitive noun and is in a position following the associative morpheme but preceding the genitive noun. (25) a. f -áà fɨ̀ - ŋkōbɨ̄ f -ɨ́ mfɔ́ fy -ā b. fɨ̀ - ŋkōbɨ̄ f -ɨ́ ɡh -àà mfɔ̀ fy -ā my(Foc) statue AM chief the statue AM my(Foc) chief the my own statue of a chief a statue of my own chief &lt;33&gt; One last interesting characteristic of possessive pronouns is that, although multiple lexical genitives are allowed in the same DP, more than one possessive pronoun cannot co-occur in the same DP as shown in (26) These constraints apply to all persons and not just the 3rd person. Thus we cannot have the Bafut equivalent of the English DP ‘My statue of your chief’ for instance. . (26) a. nɨ - ̀līʔī n-ɨ́ fɨ́ - bwē f-ɨ́ bɨ́ - fɔ̄ b-ɨ́ yàmndè b. * nɨ - ̀līʔī n-í fy-áá by-áá eye AM fish AM chiefs AM Yaounde eye its their their the eye of the fish of the chiefs of Yaounde &lt;34&gt; The questions we would like to answer here are where the possessive pronoun is generated, and how it gets to the different surface positions it occupies in (24-26). We will propose that possessive pronouns are actually definite determiners generated as head of DP. In a manner to be made clear as the discussion progresses, their status as definite determiners bars the occurrence of more than one of them in a genitive construction such as (26). Regarding their structural positions, we will propose that the DP they head is embedded inside a (Genitive Phrase) GenP like lexical genitives and that the pre-N focus distribution of the possessive pronoun results from movement to Foc 0 akin to movement of determiners discussed in section 1.1. Let us begin with their structural positions in the DP. &lt;35&gt; A fact which makes it reasonable to claim that possessive pronouns are contained in a genitive phrase is that, in addition to the fact that they stand in the same genitive relation to the head noun as lexical genitives, they trigger tonal changes on the head noun identical to ones triggered by lexical genitives. In the examples in (27a), the nouns bear low tones in citation form. When they occur as heads in a genitive construction, the tone on the final syllable changes to high as illustrated in (27b-c). (27) a. ɨ̀ - bɔ̀ʔɔ̀ mushroom à - bàà bag. b. ɨ̀ - bɔ̀ʔɔ́ Sue bó Bill c. à - báá tsítsà mushroom Sue and Bill bag teacher Sue and Bill's mushroom a teachers bag &lt;36&gt; Such tonal changes in Bantu are generally accounted for by the presence of a floating genitive (associative) tone in the underlying representation. In the surface realization, this floating tone associates to the final syllable of the head noun delinking its underlying tone. Now observe that the same tonal changes occur when a possessive pronoun is used in the place of the lexical genitive. (28) a. ɨ̀ - bɔ̀ʔɔ́ w-áá b. à - bàá y-í mushroom their bag his their mushroom his bag &lt;37&gt; This suggests that (27b-c) and (28) have an identical structure as represented below. (29) &lt;38&gt; The AM is marked by a floating tone which, in the surface realization, associates to the final syllable of the head noun. We can therefore conclude that the possessive pronoun is in a genitive phrase generated in the same position as other genitive phrases containing full lexical genitives. What we have to account for now is the precise position it occupies within the genitive phrase. &lt;39&gt; We propose that inside the genitive phrase, the possessive pronoun is located in head position of DP, that is, the possessive pronoun itself is actually a determiner. One fact which makes it plausible to treat the possessive as a determiner is that possessive pronouns in Bafut, like definite articles/determiners, mark a DP for definiteness. Generally, the definite or indefinite character of a DP is determined by the determiner. Assuming then that definiteness is encoded in D 0 position, consider the contrast below in which the genitive DP is definite in (30a) but indefinite in (30b). (30) a. fɨ̀ - ŋkōbɨ̄ f-ɨ́ m - fɔ̄ w-á b. fɨ̀ - ŋkōbɨ̄̀ f-ɨ́ m - fɔ̄ statue AM chief the statue AM chief a statue of the chief a statue of a chief &lt;40&gt; The definite genitive [of the chief] in (30a) and the indefinite [of a chief] in (30b) can, however, both be replaced by a possessive pronoun. Once the possessive pronoun is inserted, both DPs (30a-b) have a definite interpretation as shown in (31). (31) a. fɨ̀ - ŋkōbɨ̄ f-í b. fɨ̀ - ŋkōbɨ̄ f-í statue his statue his his statue his statue &lt;41&gt; We can account for why both (31a) and (31b) have a definiteness interpretation if we assume that the possessive pronoun is itself a definite article/determiner generated in D 0 . In this view, the structure of the genitive constructions in (31) in which chief is a theme argument would be as represented in (32). (32) &lt;42&gt; In the derivation, the possessive determiner í cliticizes onto the genitive morpheme n-ɨ́ . Cliticization results in the fusion of the possessive and the genitive vowels or in the deletion of one of them such that in the end, what surfaces as a possessive determiner is a complex unit comprised of a number/gender concord consonant, a genitive tone and a possessive vowel. &lt;43&gt; The analysis of the possessive pronoun in this section accounts for the tonal changes and the definite versus indefinite character of DPs observed above. First of all, analyzing the possessive pronoun as projecting a GenP allows us to account for the fact that the possessive pronoun, just like a full lexical genitive, triggers a high tone on the final syllable of certain head nouns. Secondly, analyzing the possessive pronoun as a subclass of determiners explains why a DP in which it occurs always has a definite reading. &lt;44&gt; This completes the broad picture of the internal structure of the DP in Bafut. To summarize, we have shown that although Bafut is a Bantu language, it does not allow [ Num N 0 + Num 0 ] raising to D 0 like Swahili. As such, to derive the noun-initial word order, after N 0 -to-Num 0 raising, NumP further moves to [Spec,DP]. This analysis accounts for various DP types in Bafut containing the head noun alongside different satellites and arguments: Demonstratives project a DemP above DP. Theme arguments are generated as complements to the head noun, agent arguments as specifiers and possessors as right adjuncts. Each argument projects a GenP and an adjective modifying a specific noun is adjoined to the NumP above the NP containing that noun. Possessive pronouns are a subclass of determiners generated in D 0 within GenP. Raising of the different nouns to their respective Num 0 positions and subsequent movement of matrix NumP to [Spec,DP] yields the correct surface word order; N-Argument-Pron.-Adjective-Demonstrative-Def. Art. After this characterization of the structure of DP, we turn to DP-internal concord phenomena in the next section. DP-Internal Concord &lt;45&gt; We begin the analysis of DP-internal concord in Bafut by presenting an overview of the checking theory outlined in Chomsky (1995) along with modifications introduced in Carstens (2000). Then we briefly discuss the agreement morphology and finally show how the related features are checked in the structure of the DP outlined in the preceding sections. Overview of Checking Theory &lt;46&gt; In Chomskys (1995) checking theory, items enter the syntax bearing Case, agreement, and categorial subcategorization features of heads. These features may be + or - interpretable . Only interpretable features reach LF where they are required for semantic interpretation. Uninterpretable features are erased in the syntax via checking, that is, pairing with a matching feature in an appropriate structural relation. The checking relation holds in three configurations: (a) between specifier and head, (b) between two heads α and β, α adjoined to β, (c) between the adjoined α and γP in [Spec, β]. Chomsky hypothesizes that movement occurs when an uninterpretable feature attracts some feature into its checking domain. &lt;47&gt; A targets feature may be strong or weak . When strong, it must be checked immediately via overt movement either in a [Spec,X] or y-raising to adjoin to X configuration. A weak feature needs not be checked in overt syntax but it may be if it is carried along as a free rider . In addition to the free rider phenomenon, economy requires a category that checks a weak feature to remain in situ until LF. &lt;48&gt; In Chomskys system, because interpretable features do not erase, they are available throughout the derivation and can therefore check more than one uninterpretable instantiation of F, (F 1 …F n ). In the same vein, specifiers may iterate enabling a target feature that does not erase to check multiple items. Concord Morphology &lt;49&gt; Before we discuss the mechanism of checking concord features in the Bafut DP, a brief comment on the concord morphology is in place. Our first comment on concord morphology relates to the constituency of the concord morpheme borne by noun-dependent words. Each noun class in Bafut is associated with a distinctive pattern of agreement borne by noun-dependent words (determiners, adjectives, arguments, etc.). The noun-dependent words agree with the noun in number and class. Class here can be equated roughly to gender in Romance. Agreement in number is often easy to identify as the dependent words bear distinct morphology. Thus an adjective modifying a noun in the singular form bears a concord morpheme which is different from one borne by the same adjective when it modifies the same noun in the plural form. (33) a. fɨ̀ - ndzɔ́ɔ̀́ fì -fùʼù b. mɨ̀ - ndzɔ́ɔ̀́ mì -fùʼù 19-frog 19-white 6-frog 6-white a white frog white frogs &lt;50&gt; Class concord, unlike number concord, is not very easy to locate since there is no overt morpheme that marks class. There is nonetheless good evidence that the dependent words also agree with the noun in class. Consider the following examples in which the adjective white modifies different plural nouns. (34) a. mɨ̀ - ndzɔ́ɔ̀́ mì -fùʼù b. bɨ̀ - lúʼú bí -fùʼù c. ɨ̀ - kâŋ dʒì -fùʼù 6-frog 6-white 2-spoon 2-white 8-dish 8-white white frogs white spoons white dishes d. * mɨ̀ - ndzɔ́ɔ̀́ bì̀ -fùʼù e. * bɨ̀ - lúʼú dʒí -fùʼù f. * ɨ̀ - kâŋ mí -fùʼù 6-frog 2-white 2-spoon 8-white 8-dish 6-white &lt;51&gt; Although the morphemes mi , bi , and dʒi mark agreement between the adjective and a plural noun (cf. a-c), there are strict restrictions on which morpheme is used with which noun (cf. d-f). Noun stems that belong in the same class as frogs for instance are associated only with the morpheme mi on adjectives and so on. This indicates that, in addition to agreeing in number as discussed above, the dependent word and the noun also agree in class. Since the agreement morpheme in the examples in (34) is determined by the class of the noun, it is reasonable to assume that class is an intrinsic property of nouns while the noun prefixes are morphological correlates of number features. &lt;52&gt; Our second comment concerns the noun prefixes. Every noun stem in Bafut belongs to one of a total set of 10 classes. Each class is associated with a particular prefix which marks the stem as plural or singular and there is a rigid one-to-one relation between class and prefix. Consider the following stems and prefixes: (35) A B Prefixes - ŋɡɔ̀ɔ̀ - lwìn Sing . Pl. - bɔ̀ʼɔ̀ - fɔ̀ N- bɨ̀ - kàʼà - súkàʼà nɨ̀ - m ɨ̀ - (36) a. nɨ̀ - ŋɡɔ̀ɔ̀ plantain b. m ɨ̀ - ŋɡɔ̀ɔ̀ plantains c. n - lwìn old person d. bɨ̀ - lwìn old persons (37) a. *n-ŋɡɔ̀ɔ̀ plantain b. * bɨ̀ - ŋɡɔ̀ɔ̀ plantains c. * nɨ̀ - lwìn old person d. *m ɨ̀ - lwìn old persons &lt;53&gt; To designate singular and plural entities, stems in A take the prefixes nɨ̀ - and mɨ̀ - while those in B take N- and bɨ̀ - (cf. 36a-d). Although N- and nɨ̀ - both mark nouns for singular, N- is restricted to occurring with only group B nouns and nɨ̀ - with group A nouns (cf. 37a-d). We attribute this restriction in the association of a prefix to a particular group of stems to subcategorisation restrictions. That is, a prefix morphologically subcategorises for a specific set of nouns. Since number morphology projects a number phrase (NumP), we further propose that in Bafut, Num 0 bears an N-categorial feature given the observation that noun prefixes (the number morphology) are restricted to occurring with specific nouns. In addition to this N-categorial feature, Num 0 , as expected, bears a ±singular number feature. To summarize then, the heads of the different constituents of DP bear the following features: N 0 bears: class feature Number feature N-categorial feature Num 0 bears: Number feature N-categorial feature &lt;54&gt; The noun-dependent words (determiners, demonstratives, genitives, adjectives, etc.) each bear a number plus class feature. While the features on the noun (N 0 ) are +interpretable, those on the noun-dependent words as well as those on Num 0 are –interpretable. In Chomsky´s checking theory, the –interpretable features in Num 0 and in the noun-dependent words require the noun to raise to various positions for checking purposes as outlined in the paragraphs that follow. Concord on Determiners &lt;55&gt; As outlined in the preceding section, the determiner bears uninterpretable number and class features which must be checked and erased before the derivation is submitted for interpretation. The process that results in the checking of these features is the following: First, the head noun raises to Num 0 motivated by the presence of the uninterpretable Number and N-categorial features in Num 0 . When the noun gets to Num 0 , it checks and erases the uninterpretable N-categorial and number features in Num 0 . However, since the parallel features on the noun are interpretable, they do not erase. Therefore, the complex head [Num 0 + N 0 ] now bears the uninterpretable class, number and categorial features of the noun. The first two features are needed to check the uninterpretable class and number features of D 0 . These can be checked in one of two ways: raising [Num 0 + N 0 ] to D 0 where the features will be checked in a head-head configuration or Moving NumP to Spec-DP where checking will be accomplished in a Spec-Head configuration. As outlined in section 1.1.1, the presence of overt determiners in D 0 blocks movement of the complex head into this position. In this case, NumP moves to Spec-DP and the class and number features of the determiner are then checked in a Spec-Head configuration. The relevant derivation is schematized below. (38) &lt;56&gt; In a structure containing a demonstrative, the DemP is located above DP and so after checking the features of the determiner, NumP further moves to [Spec,DemP] where it also checks the same class and number features of the demonstrative in a similar Spec-Head configuration. In Swahili-type languages that lack overt determiners, the complex head [Num 0 + N 0 ] can, and must, raise to D 0 in order to license the referential feature and also check other relevant features. When the complex head raises to D 0 the class and number features of D 0 are then checked in a head-head configuration. Adjective Concord &lt;57&gt; In the discussion in section 2.2, we proposed that Bafut adjectives are right-adjoined to NumP above the NP containing the noun that the adjective modifies. In a DP containing a head noun, an agent genitive noun and an adjective, if the adjective modifies the head noun, it is right adjoined to core NumP (i.e.the higher NumP). On the other hand, if the adjective modifies the genitive noun, it will be right adjoined to the lower NumP. These structural positions of the adjective are shown in (39) where AP 1 modifies the head noun while AP 2 modifies the agent genitive noun. (39) &lt;58&gt; Following Bernsteins (1993) idea about adjective agreement in Romance that the relation between N and an adjective adjoined to XP is analogous to the Spec-Head configuration, we assume that the adjectives adjoined to NumP are in a quasi Spec-head configuration with Num 0 . Along these lines, if there is an adjective in AP 1 position modifying the head noun, when the latter raises to Num 0 , the class plus number features of the adjective are checked against those of the noun in this quasi Spec-Head configuration. Spelling out of the checked features therefore results in the adjective agreeing with the head noun only. Similarly, if the adjective is in AP 2 , when the agent noun raises to lower Num 0 the features of the adjective are checked aginst those of the gentive noun in the same quasi Spec-Head configuration resulting in agreement with the genitive noun only. The different structural positions reserved for adjectives as well as the checking mechanism outlined here accounts for the phenomenon which we encountered in section 2.2 where a single adjective occurring in the same construction with two nouns can modify either one of the nouns depending on the concord morphology which it (i.e. the adjective) bears. Genitive Concord &lt;59&gt; The agreement relation between the head noun and its arguments is mediated by the Associative Morpheme; that is, it is the Associative Morpheme (and not the modifying noun itself) that bears the concord morphology. The Associative Morpheme (AM) is the Bafut equivalent of what Carstens (2000) analyzes as of. We have however decided to maintain the traditional Bantuist expression - Associative Morpheme The Associative Morpheme (AM) has been referred to variously in the literature as a relational element (Carstens 1991), an associative morpheme (following Bantuist tradition) (Mfonyam 1989, Ambe 1989, Nkemnji 1995), a genitival connective (Vitale 1981), etc. Most often, it is referred to simply as an agreement/concord element (see Hyman 1977, 1981 and Hedinger 1980). for a variety of reasons. First is that this particle establishes a variety of semantic associations (possession, agent, theme, material make-up, content, time/place of use, kinship, etc.) between the head noun and its object which cannot all be expressed by of. Secondly, treating the AM as of amounts to treating it as a preposition. However, prepositions in Bafut do not bear concord morphology. Thus, in a construction like (40), for instance, the preposition nɨ̄ with relates a head noun to a following noun just like an ordinary associative morpheme, but it does not agree with the head noun. (40) a. m - àŋɡyè nɨ̄ mɨ̀ - tʃè b. bɨ̀ - fɔ̀ nɨ̄ mɨ̀ - tʃè 1-women with 6-wisdom 2-chiefs with 6-wisdom wise women wise chiefs (village sages) &lt;60&gt; Concord on prepositions is, however, not altogether unprecedented since something like it is attested in Celtic and Berber languages (Higgins, p.c). There is, however, a significant difference in that in these languages, the concord-like property holds between the preposition and its object. In Bafut genitive phrases, concord is between the AM and the preceding head noun and not between the AM and its object. Treating the associative morpheme in Bafut as a preposition therefore seems untenable first because ordinary prepositions in Bafut do not bear agreement and second because where Bafut appears to pattern with Celtic and Berber languages, the associative morpheme agrees with the head noun and not with its object. &lt;61&gt; We will therefore treat the associative morpheme simply as relational element which links the genitive noun to the head noun and also establishes the associative relation between the two. In the syntax, this relational element projects a functional projection which, for ease of exposition, we continue to refer to as a genitive phrase (GenP). &lt;62&gt; However, whether we treat this particle as of or AM does not really matter for the present discussion. What is important is how to account for the concord features of this morpheme within Chomskys checking theory and to explain why this morpheme agrees with the head noun and not with its object. First the concord features. &lt;63&gt; In section 2.1, we proposed that arguments of the head noun (i.e. genitive phrases) are generated in a shell above NP core. In this configuration, for the head noun to get to Num 0 , it has to transit through the head positions of these NP shells. When it gets to each of these head positions, the class and number features of the argument which is in the Spec position of the NP shell are checked against those of the head noun in the usual Spec-Head configuration. A Problem &lt;64&gt; In section 3.5, we observed that the concord relation between the head noun and the genitive noun (the argument) is mediated by the AM and not the genitive noun itself, i.e. it is the AM and not the genitive noun itself that agrees with the head noun. There are actually two problems here; first why the genitive noun does not exhibit overt concord with the head noun and second why the AM agrees with the head noun and not its object. The facts of Bafut are very similar to those of Swahili. Thus, in both languages, we can have (41a) but not (41b) and (41c). (41) BAFUT SWAHILI a. nɨ̀ - kàʼà n - ɨ́ fɨ́ - njɔ́ɔ̀ a. kiti cha mtoto 5-leg 5-AM 19-toad 7-chair 7-of 1-child A toads leg The childs chair [AM/of agree with head N] b. * nɨ̀ - kàʼà f - ɨ́ fɨ́ - njɔ́ɔ̀ b. * kiti wa mtoto 5-leg 19-AM 19-toad 7-chair 1-of 1-child [*AM/of and object agree] c. * nɨ̀ - kàʼà ɨ́ nɨ - fɨ́ - njɔ́ɔ̀ c. *kiti a ch-m-toto 5-leg AM 5/19-toad 7-chair of 7/1-child [*Head N and object agree] &lt;65&gt; Considering the second problem first, the fact that the genitive noun does not exhibit overt concord with the head noun supports an analyzes whereby the genitive noun should not be in a relevant (head-head or Spec-head) relation with the head noun. In Carstens (2000) proposal for Swahili, and the one we have outline in this paper for Bafut, the genitive noun is quite low down in the structure. Raising it into a head-head or Spec-head relation with the head noun (so that it can check the relevant features) will violate a number of locality conditions on movement. Observe also that each noun has intrinsic number and class features and the concord morphology is also specified for the number and class features of the head noun. Were the genitive noun to exhibit overt concord with the head, it would end up with two class features and two number features. One can imagine what this will entail for the checking theory assumed in this paper. The number + class features borne by the concord morpheme require the genitive noun to raise to matrix Num 0 (through Num 0 embedded inside GenP) in violation of locality conditions on movement as pointed out immediately above. &lt;66&gt; Since the genitive noun does not exhibit overt concord with the head noun, one could imagine a condition like the one in (42) which militates against a single noun bearing two distinct class or number features. (42) Feature Uniqueness Condition *X f1,f2 where f 1 ,f 2 are two instantiations of the same feature, F. Under (42), the genitive noun cannot bear concord morphology and so the concord morphology controlled by the head noun is obliged to associate to the AM/of The Feature Uniqueness Condition is reminiscent of other modules of grammar (e.g. Theta Criterion, Case theory, etc.) which require that an item be specified for a unique syntactic role. . &lt;67&gt; Turning now to the other problem why the AM does not agree with the genitive noun, Carstens (2000) exploits Chomskys (1995, 1998) ideas about Merge to account for the failure of concord between of and its object in Swahili. According to her proposal, checking between of and its object fails because the two items are co-terms of merge, i.e. the two items are adjacent to each other as a result of merging and not movement. We adopt her proposal here and add to it the fact that in addition to being co-terms of merge, the syntactic relation between of and its object is not like the one between other heads (e.g. verbs and prepositions) and their objects. According to Chomsky (1998), Case is a manifestation of concord. Assuming this to be true, we expect the AM to agree with the genitive DP which it case-marks. An explanation as to why the AM does not agree with the genitive DP concerns the status of AM as a Case assignor. Recall that the issue of genitive Case had been debated upon for long and one of the conclusions arrived at was that genitive Case is inherent. If we return to this assumption, we can provide an explanation for why the AM does not agree with the genitive noun. Genitive Case is inherent implying that it is not assigned by any particular category. Assuming this to be correct, and holding on to the idea that Case is a manifestation of Concord, we do not therefore expect the AM to agree with the genitive noun. Conclusion &lt;68&gt; This article has demonstrated that despite significant differences in the structures of Bafut and Swahili concerning the presence versus absence of overt determiners, the checking theory outlined in Chomsky (1995) can account for Bafut DP-internal concord phenomenon in much the same way as it does for Swahili. Checking concord features on agreeing determiners, demonstratives, adjectives and associative morphemes is accomplished in terms of symmetrical checking relations between heads and specifiers and between two adjoined heads. &lt;69&gt; The paper also suggests that the genitive noun does not exhibit overt concord with the head noun because of the Feature Uniqueness Condition which allows an element to bear only a single set of a particular feature. In a related problem, it was proposed that the Associative Morpheme does not agree with its object because of the syntactic relation between it and the object. The AM and its object are co-terms of Merge (see Carstens 2000) and in addition, the AM does not Case-mark its object; thus concord between the two is not expected in a system where concord is a manifestation of Case. References Abney, S.P. 1987 The English Noun Phrase in its Sentential Aspect. Ph.D. dissertation, MIT Ambe, S.A. 1989 The Structure of Bafut. Ph.D. dissertation, Georgetown University Androutsopoulou, A. 1997 On remnant DP-movement in modern Greek, unpublished M.A. thesis, University of Carlifornia, Los Angeles Bernstein, J. 1993 Topics in the Syntax of Nominal Structure Across Romance. Ph.D. dissertation, City University of New York Bernstein, J 1997 'Demonstratives and Reinforcers in Romance and Germanic Languages'. Lingua 102:87-113 Biloa, E. 1991 Operator Constructions in Tuki. Ph.D. dissertation, University of Southern California Brody, M. 1990 'Remarks on the Order of elements in the Hungarian Focus Field.' In: Approaches to Hungarian 3 , ed. by Istvan Kenesei, pp.95-122. Szeged: Joszef Attila Tudomanyegyetem Bruge, L.1996 'Demonstrative Movement in Spanish: A Comparative Approach.' University of Venice Working Papers in Linguisitics 6.1 Carstens, V.M. 1991 The Morphology and Syntax of Determiner Phrases in Kiswahili. Ph.D. dissertation, UCLA Carstens, V.M. 2000 'Concord in Minimalist Theory'. Linguistic Inquiry 31:319-355 Chomsky, N.1986 Barriers. Cambridge, Mass.: MIT Press Chomsky, N.1995 The Minimalist Program. Cambridge, Mass.: MIT Press Chomsky, N. 1998 [2000] 'Minimalist Inquiries: The framework'. MIT Occasional Papers in Linguistics 15. Cambridge, Mass: MIT [reissue in: Step by step , ed. by Roger Martin, David Michaels and Juan Uriagereka. Cambridge: MIT Press] Cinque, G. 1994 'On the evidence for partial N-movement in the Romance DP'. In: Paths towards Universal Grammar: Studies in Honor of Richard S. Kyne , ed. by Guglielmo Cinque, Jan Koster, Jean Yves Pollock, Luigi Rizzi, and Raffaella Zanuttini, pp.85-110. Washington D.C.: Georgetown University Press Crisma, P. 1996 'On the configurational nature of adjectival modification'. In: Grammatical Theory and Romance Linguistics , ed. by Karen Zagona, pp.59-71. Amsterdam: John Benjamins Dryer, M. 1992 'The Greenbergian Word Order Correlations'. Language 68:81-138 Giusti, G. 1993 La Sintassi dei Determinanti. Padova: Unipress Giorgi, A. and G. Longobardi 1991 The Syntax of Noun Phrases: Configuration, Parameters and Empty Categories. Cambridge University Press Hedinger, R. 1980 'The Noun Classes in Akoose (Bakossi)'. In: (ed.) Noun Classes in the Grassfields Bantu Borderland. SCOPIL 8 , ed. by L.M. Hyman. University of Southern California, Los Angeles Hyman, L. M. 1977 'Noni (Misaje Group).' In: Noun Classes in Grassfields Bantu, ed. By Larry M. Hyman and Jan Voorhoeve, Paris: Centre National de la Researche Scientifique Hyman, L. M. 1981. Noni Grammatical Structure. SCOPIL 9: University of Southern California, Los Angeles Mfonyam, J.N. 1989 Tone in Orthography: The Case of Bafut and Related Languages. Ph.D. dissertation, Universite de Yaounde, Cameroun Nkemnji, M. 1995 Heavy Pied-Piping in Nweh. Ph.D. dissertation, UCLA Picallo, C. 1991 'Nominals and Nominalization in Catalan'. Probus 3:279-316 Postal, P. M. 1969 'On So-Called Pronouns in English'. In: Modern Studies in English: Readings in Transformational Grammar , ed. by David Reibel and Sanford Schane, pp.201-224. Englewood Cliffs, New Jersey: Prentice-Hall Ritter, E. 1991 'Two Functional Categories in Noun Phrases. Evidence from Modern Hebrew'. In: Syntax and Semantics 25: Perspectives on Phrase Structure , ed. By Susan Rothstein, pp.37-62. New York: Academic Press Rizzi, L. 1997 'The Fine Structure of the Left Periphery'. In: Elements of Grammar , ed. By Liliane Haegeman, pp.281–337. Kluwer, Dordecht Siloni, T.1997 Noun Phrases and Nominalizations: The Syntax of DPs. Dordrecht, Boston, London: Kluwer Academic Publishers Travis, L. 1984 Parameters and Effects of Word Order Variation. Ph.D. dissertation, MIT Valois, D. 1991 The Internal Syntax of DP. Ph.D. dissertation, UCLA Vitale, A. 1981 Swahili Syntax. Foris: Dordrecht</value></element></document>