Complementizer-Alternation in Creole Languages: New Evidence for Spec-Head Agreement

PAPIA, São Paulo, 25(2), p. 155-176, Jul/Dez 2015. http://doi.org/10.5281/zenodo.17809940 Complementizer-Alternation in Creole Languages: New Evidence for Spec-Head Agreement Alternância de complementizadores em línguas crioulas: novas evidências para a concordância spec-núcleo Marlyse Baptista1 University of Michigan, USA [email protected] Miki Obata Tokyo University of Science, Japan [email protected] Abstract: The topic of this paper is the ki/ma alternation in Cape Verdean Creole (henceforth, CVC) with a focus on the behavior of ki within wh-questions. We argue that ki is a morphosyntactic reflex of Q-agreement between [uQ] on C and a wh-phrase, supporting spec-head agreement in Chomsky (1991), contra the probe-goal agreement system in Chomsky (2000). In so doing, we provide evidence for its position in C and show that in the absence of operator movement and Q-agreement on C, an alternative particle, ma/kuma, appears in C. We show that the symmetric patterns of extraction observed in CVC where the complementizer ki can appear when either the subject or object are extracted, may be in part reducible to a contrast in locality constraints on Agree and can be better explained within a spec-head relation framework for CVC. Keywords: Cape Verdean Creole; complementizer; spec-head; probe-goal. 1Marlyse Baptista and Miki Obata are co-first authors on this paper. e-ISSN 2316-2767 156 Marlyse Baptista / Miki Obata Resumo: O foco deste artigo é a alternância ki/ma em Crioulo de Cabo Verde (doravante, CVC), dedicando uma especial atenção ao comportamento do ki em interrogativas-wh. Defendemos que ki é o reflexo morfossintático da concordânciaQ entre [uQ] em C e um sintagma-wh, reforçando a ideia de concordância spec-head, de Chomsky (1991), contra o sistema de concordância sonda-alvo, de Chomsky (2000). Ao fazermos isto, fornecemos evidência para a posição de ki em C e mostramos que, na ausência de movimento de operador e de concordância-Q em C, uma partícula alternativa, ma/kuma, ocorre em C. Mostramos ainda que os padrões simétricos de extração observados em CVC, onde o complementador ki pode ocorrer quando quer o sujeito quer o objeto são extraídos, podem ser parcialmente reduzidos a um contraste de restrições de localidade em Agree e podem ser mais bem explicados através de um quadro de relações spec-head em CVC. Palavras-chave: Crioulo de Cabo Verde; complementizador; especificador-núcleo; sonda-alvo. 1 Introduction 1.1 Objectives This paper focuses on the particle ki in Cape Verdean Creole (henceforth, CVC) and investigates its behavior within wh-questions. Our main objective is to demonstrate that ki is a morphosyntactic reflex of Q-agreement between [uQ] on C and a wh-phrase, supporting spec-head agreement in Chomsky (1991), contra the probe-goal agreement system in Chomsky (2000). We provide evidence for its position in C and crucially show that in the absence of operator movement and Q-agreement on C, an alternative particle, ma/kuma, appears in C in CVC. We show that the symmetric patterns of extraction observed in CVC where the complementizer ki can appear when either the subject or object are extracted, may be in part reducible to a contrast in locality constraints on Agree and can be better explained within a spec-head relation framework for CVC. With this approach in mind, ki is presented as a reflex of agreement between a complementizer and a wh-phrase, an analysis that was sketched in Baptista (1993) for CVC and more recently in Takahashi & Graçanin-Yuksek (2008) for Haitian Creole2. 2 For an in-depth comparison of CVC and Haitian Creole, see Obata, Epstein & Baptista (2015). PAPIA, 25(2), e-ISSN 2316-2767 Complementizer-Alternation in Creole Languages... Alternância de complementizadores em línguas crioulas 157 1.2 Claims Based on the empirical data we introduce, we make two main claims: The first is that COMP in CVC bears an uninterpretable Q feature [uQ], which causes a COMP-alternation between ki and ma. This uninterpretable Q feature will account for the dependency between C and a wh-phrase. The second is that Q-agreement in CVC supports Chomsky’s (1991) spec-head agreement system (contra Chomsky’s (2000) probe-goal system). 2 Extraction patterns and the behavior of ki in Cape Verdean Creole This section elaborates on some preliminary observations that were made in Baptista (1993) where she explored the behavior of ki in subject-extraction, object-extraction, short-distance movement and long-distance movement in CVC wh-questions. She stipulated that ki appears in Comp whenever a whphrase moves through Spec-CP (within a GB framework) and consequently analyzed ki as an indicator of spec-head agreement (Baptista 1993: 9) between the wh-word and Comp, following a similar analysis in Chung and McCloskey (1987) for Modern Irish. The relevant examples are laid out below. 2.1 Wh-Questions in CVC: Subject and object extraction In wh-questions 3 , a Comp ki in CVC can be or must be phonetically realized when argument extraction is involved 4 . Consider the following examples illustrating subject extraction (1) and object extraction (2) in simple matrix questions: Subject extraction in matrix question: 3 For an in-depth overview and analysis of wh-questions and extraction patterns in CVC, see Alexandre (2009; 2012), which remain to date the most comprehensive treatment on this topic. 4 One should note, however that there is considerable dialectal and idiolectal variation in the language with regard to the obligatory use of ki.Ki tends to be obligatory in the case of subject extraction involving the monosyllabic wh-phrase ken in contrast to kenhi. Furthermore, in Veiga (2000: 181), there are attestations of dialectal variation with regard to object extraction. He contrasts the Santiago dialect where ki is not required to its São Vicente counterpart, in which ki must appear: (i) Kuze Øbu kre? (Veiga 2000: 181) Santiago variety what you want “What do you want?’ PAPIA, 25(2), e-ISSN 2316-2767 158 Marlyse Baptista / Miki Obata (1) a. Kenhi who ki ki odja see João? João “Who saw João?” b. Kuze what ki ki kontise happen ki that po-bu put-you ben come kaza home es this ora hour li? here (Brüser et al. 2002: 370) “What happened that made you come home at this time?” Object extraction in matrix question: (2) a. Kuze what ki ki João João odja? see “What did John see?” b. Kuze what ki ki nhos you fla? say (Brüser et al. 2002: 370) “What did you say?” The same pattern obtains in the case of embedded questions involving subject extraction, as in (3) or object extraction, as in (4): Subject extraction in embedded question: (3) Bu you sabe know kenhi who ki ki ta MOOD bai go merkadu market manhan? tomorrow “Do you know who will go to the market tomorrow?” Object extraction in embedded question: (4) Bu you sabe know kuze what ki ki João João kunpra? bought “Do you know what John bought?” (ii) Keze ke bo kre? (Veiga 2000: 181) São Vicente variety what ke you want “What do you want?” For the purpose at hand, suffice to say that ki can, and in some dialects, must appear in matrix simple questions as well as embedded questions involving both subject and object extraction. PAPIA, 25(2), e-ISSN 2316-2767 Complementizer-Alternation in Creole Languages... Alternância de complementizadores em línguas crioulas 159 Relatives involving wh-questions also display ki: (5) João João ka NEG sabe know kenhi who ki ki furta steal bolsa. bag “João does not know who stole the bag.” (6) Mudjer woman ki ki dja COMP ka NEG sabeba know+ANT kuze what k’ ki el she al would fase... do (Brüser et al. 2002: 371) “The woman who did not know what she would do...” Furthermore, it is worth noting that there is an asymmetry between argument extraction and adjunct extraction in CVC. While ki can occur in both subject and object argument extraction, it is often the case that ki is preferably absent with adjunct extraction (7a) though it can also be expressed (7b). (7) a. Maria Maria ka NEG sabe know undi where Ø João João bai. go “Maria doesn’t know where João went.” b. Maria Maria ka NEG sabe know undi where ki COMP João João bai. go “Maria doesn’t know where João went.” We will not linger on this asymmetry in this paper but refer the reader to Alexandre (2009; 2012) for an in-depth and comprehensive coverage of whquestions involving argument and adjunct extraction, in addition to restrictive relative clauses in CVC. Her analyses also tackle the thorny issues of gap and resumptive strategies. Having laid out the behavior of ki in both simple matrix and embedded subject and object extraction, the next question to address is the exact position of ki in the sentence structure. If ki is assumed to be in C, what are the diagnostics for its position in C? This is the topic of the next subsection. PAPIA, 25(2), e-ISSN 2316-2767 160 Marlyse Baptista / Miki Obata 2.2 Diagnostics of ki as a complementizer in CVC Given that ki systematically appears with wh-questions, the first candidate that comes to mind as a possible diagnostic for the position of ki in C would be yes/no questions. We must, however, discard these cases because in contrast to wh-questions, ki is not allowed to occur in yes/no-questions, as shown by the ungrammaticality of (8). (8) *Ki C João João kunpra bought libru? book “Did John buy the book?” [cf. English: *(Did) John buy the book?] Instead, yes/no questions in CVC are only expressed through raising intonation: (9) João João kunpra bought libru? book “Did John buy the book?” [cf. English: John bought the book?] As no reliable diagnostic of ki as a complementizer can be found in yes/no questions, we turn to a more promising diagnostic, that of subject-auxiliary inversion. 2.2.1 Diagnostic for ki as a Comp from subject-auxiliary inversion In CVC, subject-auxiliary inversion can take place in the case of wh-questions involving copulas. (10) a. Undi where Ø sta is bu your libru? book “Where is your book?” (Veiga 1995: 367) b. Undi where ki ki bu your libru book sta? is “Where is your book?” If in (10a), the auxiliary sta can be assumed to move to C to derive the subject-auxiliary word order, then the piece of data in (10b) naturally ensues from the fact that ki appears in C while sta remains in situ. It is worth noting that while copulas can fill C in CVC, main verbs cannot move to such positions, as witnessed by the ungrammaticality of (11): PAPIA, 25(2), e-ISSN 2316-2767 Complementizer-Alternation in Creole Languages... Alternância de complementizadores em línguas crioulas 161 (11) a. *Undi where ki ki kume eat João? João b. Undi where (ki) João João kume? eat “Where did João eat?” The fact that inversion is only possible with copulas provides us with a first diagnostic for the position of ki in C (see Baptista 1997; 2002; Alexandre 2009). Sluicing provides us with another piece of evidence, as examined in the next subsection. 2.2.2 Diagnostic for ki as a Comp from sluicing In languages which involve overt Comps, Comps cannot survive after sluicing (Merchant 2001), which is interpreted as TP deletion after wh-movement. Given this basic assumption, our prediction is that if CVC is indeed a language which involves overt Comps, then, ki cannot survive after TP is deleted by sluicing operations. Consider the data in (12) illustrating a sluicing operation. (12a) shows that after sluicing applies, ki cannot occur on C; our assumption that ki is Comp is empirically supported. (12) a. *João João kunpra bought algun some kuza, thing ma but N I ka NEG sabe know kuze what ki. C “John bought something but I don”t know what.” b. João João kunpra bought algun some kuza, thing ma but N I ka NEG sabe know kuze. what “John bought something but I don”t know what.” We bring further empirical evidence to ki being in C when we consider in the next section the ki/ma alternation in CVC where we show that the complementizer ma is in complementary distribution with ki, hence they occupy the same position in C. Table 1 summarizes the basic facts regarding the behavior of CVC ki in the presence of wh-phrase in Spec-CP. PAPIA, 25(2), e-ISSN 2316-2767 162 Marlyse Baptista / Miki Obata Distribution of ki in CVC Subject extraction Object extraction ki can appear in matrix C + + ki can appear in embedded C + + Tab. 1: ki in CVC. 2.3 Some basic assumptions We follow partially Takahashi & Graçanin-Yuksek’s proposal for Haitian Creole (see Obata, Epstein and Baptista 2015 for an extensive comparison between CVC, Haitian Creole and other languages) in making the following set of assumptions: we start out by assuming that ki is a phonological reflex of agreement between the complementizer C and a wh-phrase (see Baptista (1993: 9) for a similar analysis). Crucially, we consider Takahashi & GraçaninYuksek’s hypothesis that C, in the relevant context, contains an uninterpretable wh-feature (uwh) and uninterpretable ϕ -features (u ϕ ). The uwh-feature allows for a dependency between C and a wh-phrase and the u ϕ features on C are motivated by cross-linguistic empirical evidence 5 . The uwh and u ϕ features allow Takahashi & Graçanin-Yuksek to derive the extraction asymmetry between subject and object wh-phrases in Haitian, as subject extraction requires the presence of the complementizer ki in Haitian whereas object extraction forbids it but such an asymmetry does not obtain in CVC, as already shown in examples (1) and (2). At the core of Takahashi & GraçaninYuksek’s analysis is the argument that in the case of subject extraction, C enters an Agree relation with the wh-phrase in Spec-TP and both uwh and u ϕ features on C are checked and valued by a single goal, the subject wh-phrase. Crucially, Takahashi & Graçanin-Yuksek assume that the complementizer is spelled out as ki if all its features are checked off by a single goal. They assume that the EPP property is a subfeature of the uwh, following Pesetsky & Torrego (2001) (Takahashi & Graçanin-Yuksek 2008: 229). While subject extraction requires the presence of ki in Haitian, ki cannot appear when the object undergoes wh-movement. Takahashi & GraçaninYuksek account for the asymmetry between subject and object extraction by holding the following reasoning: they argue that uwh and u ϕ on C are 5 See Takahashi & Graçanin-Yuksek (2008) for empirical evidence from Bantu languages, based on Carstens (2003). PAPIA, 25(2), e-ISSN 2316-2767 Complementizer-Alternation in Creole Languages... Alternância de complementizadores em línguas crioulas 163 able to search the structure independently to look for their own matching features and agree with the closest goal. Since the subject in Spec-TP is the closest constituent involving ϕ-features to C, uϕon C are checked off by the subject. On the other hand, uwh on C is checked off by the wh-phrase that carries the wh-feature. Therefore the two types of features on C are checked off by different goals in contrast to subject extraction where both uwh and uϕare checked and valued by a single goal. Takahashi & Graçanin-Yuksek’s main claim is therefore that the subject-object asymmetry in the distribution of Haitian ki is reducible to the locality constraint on Agree (Takahashi & Graçanin-Yuksek 2008: 231). While this analysis elegantly accounts for the Haitian facts, as well as subject extraction in CVC, it falls short of accounting for object extraction in CVC. Indeed, if we adopt the same approach to subject extraction in CVC, the above analysis follows through. Consider the derivation in (14) featuring subject extraction represented in the example (1) above and repeated here for convenience as (13): (13) Kenhi who ki ki odja see João? João “Who saw João?” (14) [CP ki [uwh, EPP] [uϕ][TP kenhi [wh] [ϕ]T [vP <kenhi>odja João]]] agree In (14), following a similar reasoning to Takahashi & Graçanin-Yuksek (2008: 228) for Haitian (as discussed in Obata, Epstein and Baptista 2015), we propose that in the case of CVC subject extraction, C enters into an Agree relation with the wh-phrase in Spec-TP and both uwh and u ϕ features on C are checked and valued by a single goal, in this case the subject wh-phrase. The derivation illustrates our point that it is C and no other element that is spelled out as ki. The complementary distribution between the auxiliary sta and ki, as well as the sluicing tests presented in (10) and (12) respectively, will be further corroborated below by the ki/ma alternation, examined in section 4. While the probe-goal framework provides a satisfactory analysis for subject extraction based on local dependency relation, the question is how one can account for object extraction in CVC. Recall the example of object extraction in (2), repeated here as (15) for convenience: PAPIA, 25(2), e-ISSN 2316-2767 170 Marlyse Baptista / Miki Obata (29) Kenhi who ki C fra-m told+me ma/*ki/*Ø C Maria Mary kunpra bought kuze? what ‘Who told me Mary bought what?’ Furthermore, if wh-phrases move through the embedded Spec-CP but does not land, the selected comp is ma, not ki, as shown in (30). (30) Kuze what ki C João João fra-m told+me ma /*ki/*Ø C Maria Maria kunpra bought ’What did John tell me Mary bought?’ So the preceding data lead us to formulate the following generalizations, the topic of the next section. 4.2 Some basic generalizations So far, we have been able to make the following observations: • If an illocutionary verb is used and no wh-word appears in Spec-CP, ma is obligatorily selected, not ki. • If a wh-word appears in Spec-CP after the illocutionary verb, then ki must appear in C, not ma. •If a wh-phrase remains in situ and no movement occurs, ma is selected by the illocutionary verb. No ki can appear. • If a wh-word moves through spec-CP and crucially does not stay in C, ma is selected by the illocutionary verb, not ki. This summary is schematized in Table 2; note that <WH> stands for an unpronounced copy. One can find an answer to the observed comp-alternation in CVC by formulating the generalization in (31): (31) The complementizer ma changes to ki iif and only if a wh-phrase is interpreted at its Spec position. This generalization can be tested against all four cases laid out in Table 2. In Case 1, repeated as (32), ma is selected instead of ki because no wh-phrase is interpreted at the embedded Spec-CP. CASE 1: no WH (=27) PAPIA, 25(2), e-ISSN 2316-2767 Complementizer-Alternation in Creole Languages... Alternância de complementizadores em línguas crioulas 171 CASE 1: no WH (=27) tell [CP ma/*ki [TP . . . . . ]] CASE 2: WH at Spec-CP (=28) tell [CP WH [ki/*ma [TP . . . <WH> ]] CASE 3: WH in situ (=29) tell [CP ma/*ki [TP . . . WH ]] CASE 4: WH moving through Spec-CP (=30) tell [CP <WH> [ma/*ki [TP . . . <WH> ]] Tab. 2: Complementizer-alternation with ma/ki. (32) João told fra-m +me ma/*ki/*Ø C Maria Maria kunpra bought libru. book ‘John told me Mary bought the book.’ In Case 2 repeated here as (33), ki is selected as C instead of ma because the wh-phrase kuze is interpreted as the embedded question at the embedded Spec-CP. CASE 2: WH at Spec-CP (=28) (33) Kenhi who ki C fra-m told+me kuze what ki/*ma/*Ø C Maria Maria kunpra? bought ‘Who told me what Mary bought?’ In Case 3, repeated here as (34), ma is selected instead of ki because the wh-phrase kuze is NOT interpreted as the embedded question at the embedded Spec-CP but as the matrix question by interacting with the matrix C. CASE 3: WH in situ (=29) (34) Kenhi who ki C fra-m told+me ma/*ki/*Ø C Maria Mary kunpra bought kuze? what ‘Who told me Mary bought what?’ PAPIA, 25(2), e-ISSN 2316-2767 172 Marlyse Baptista / Miki Obata Finally, in Case 4 repeated here as (35), ma is selected instead of ki because the wh-phrase kuze only moves through the embedded Spec-CP, so that it is NOT interpreted as the embedded question but interpreted as the matrix question. CASE 4: WH moving through Spec-CP (=30) (35) Kuze what ki C João João fra-m told+me ma/*ki/*Ø C Maria Maria kunpra? bought ‘What did John tell me Mary bought?’ Based on the empirical data above, we are able to formulate several important generalizations regarding the CVC comp-alternation: The complementizer ma changes to ki iff a wh-phrase is interpreted at its Spec position; in other words, if a wh-phrase is interpreted in the embedded Spec-CP, then ki must appear. This is shown in (33). A correlated generalization is that ki does not appear in the embedded Spec-CP if no wh-phrase is interpreted at the embedded Spec-CP, as seen in (32). If as in (34) illustrating an in-situ case, the wh-phrase kuze is NOT interpreted as the embedded question at the embedded Spec-CP but as the matrix question by interacting with the matrix C, then ki cannot appear. Finally, in overt movement cases, as shown in (35), if the wh-phrase kuze only moves through the embedded Spec-CP, so that it is NOT interpreted as the embedded question but interpreted as the matrix question, ki cannot appear either. Based on these data and the previous generalizations, we argue in this paper that the best way to explain the observed comp-alternation is to postulate an uninterpretable Q feature on C or [uQ] on C, which is equivalent to [uwh] on C in Takahashi and Graçanin-Yuksek (2008). More precisely, we propose that given that the complementizer ma changes to ki if a wh-phrase is interpreted at its Spec position, the trigger of the comp-alternation is Q-agreement between C and a wh-phrase. The same type of agreement has been observed in other languages such as Irish (Chung and McCloskey 1987) and Kilega (Carstens 2005), also see Baker (2008). 4.3 Towards a spec-head agreement analysis Our basic analysis can be summarized as follows: The ki/ma-alternation is a morphosyntactic reflex of Q-agreement between [uQ] on C and a wh-phrase (contra Chomsky 2007). PAPIA, 25(2), e-ISSN 2316-2767 Complementizer-Alternation in Creole Languages... Alternância de complementizadores em línguas crioulas 173 4.3.1 Some evidence from wh at Spec-CP and wh in-situ If the current analysis of the CVC comp alternation as Q-agreement is on the right track, it implies that the agreement is carried out under the spec-head Agreement, not through the probe-goal system. In order to strengthen our position, let us revisit Cases 2 and 3 laid out above and repeated here for convenience as (36) and (37): CASE 2: WH at Spec-CP (=28) (36) Kenhi who ki C fra-m told+me kuze what ki/*ma/*Ø C Maria Maria kunpra? bought ‘Who told me what Mary bought?’ CASE 3: WH in situ (=29) (37) Kenhi who ki C fra-m told+me ma/*ki/*Ø C Maria Maria kunpra bought kuze what ‘Who told me Mary bought what?’ Both (36) and (37) show that only when Spec-CP hosts an overt wh-phrase (=spec-head), does the comp alternation occur. The implication of this is that the probe-goal system wrongly predicts that Q-agreement with an in-situ wh-phrase is possible. The CVC comp alternation supports the spec-head agreement (Chomsky 1991) but runs against the probe-goal system (Chomsky 2000). Our analysis finds further empirical support from yes/no and echo questions. 4.3.2 Some further evidence from yes/no and echo questions Recall (section 2.2.) that yes/no questions in CVC do not allow the occurrence of ki in C, as shown in (8), repeated here for convenience. (38) *Ki C João João kunpra bought libru? book ‘Did John buy the book?’ [cf. English: *(Did) John buy the book?] It is also the case that ki is not allowed to appear with echo questions, as shown in (39): PAPIA, 25(2), e-ISSN 2316-2767 174 Marlyse Baptista / Miki Obata (39) *Ki ki Maria Maria kunpra buy kuze? what (40) Maria Maria kunpra buy kuze? what “Maria bought what?” Given our spec-head analysis, we can account for the two pieces of data in (38) and (39) by postulating that in the case of yes/no questions in (38), there is no wh-phrase at the Spec of C which has [uQ], so that unvalued [uQ] causes crash. In similar ways, in the case of the echo question in (39), the wh-phrase is in-situ, hence the spec-head configuration is not established. Consequently, the [uQ] on C is not valued and causes crash. 5 Findings and conclusion There are a few generalizations to be made about the behavior of CVC ki. In CVC, ki appears in cases of subject extraction. The derivation in (14) shows that, as there is a spec-head configuration between C and the wh-word, ki is expected to appear and does so. In the case of object extraction, the wh-word does not start out in a spec-head relation with C but as we have proposed earlier in this paper, if we assume that Comp in CVC bears an uninterpretable Q feature on C [uQ], this then causes the wh-phrase to check that feature and by doing so moves into Spec-CP and establishes the spec-head relation necessary to the phonological realization of ki. The same uninterpretable Q-feature also accounts for an independent set of data in the language, the observed COMP-alternation between ki and ma, as ki appears in embedded C only if a wh-phrase is interpreted at its Spec. This set of empirical data allows us to argue that spec-head agreement is a necessary condition to the occurrence of ki in CVC. CVC ki can readily appear in a spec-head configuration with a wh-phrase or be triggered by movement of awh-phrase to Spec-CP by virtue of the uninterpretable Q-feature on C. This makes CVC a spec-head agreement language. In summary, this paper has highlighted three main points: • COMP in CVC bears [uQ] triggering Q-agreement and is the single feature accounting for the distributional behavior of CVC ki. • The ki/ma-alternation in CVC is a morphosyntactic reflex of Q agreement between [uQ] on C and a wh-phrase. Crucially, our analysis relies on morpho-syntactic features with consequences for learnability. • Q-agreement in CVC supports Chomsky’s (1995) and Koopman (2006) spec-head Agreement system (contra Chomsky’s (2000) probe-goal system) for languages like CVC. PAPIA, 25(2), e-ISSN 2316-2767 Complementizer-Alternation in Creole Languages... Alternância de complementizadores em línguas crioulas 175 Abbreviations C: Complementizer; COMP: Complementizer; CVC: Cape Verdean Creole; GB: Government and Binding; LF: Logical Form; TP: Tense Phrase; EPP: Extended Projection Principle; PP: Prepositional Phrase. References Alexandre, Nélia. 2012. The defective copy theory of movement: Evidence from wh-constructions in Cape Verdean Creole. Amsterdam/Philadelphia: John Benjamins. Alexandre, Nélia. 2009. Wh-constructions in Cape Verdean Creole: Extensions of the copy theory of movement. Ph.D. dissertation, University of Lisbon. Baker, Mark. 2008. The syntax of agreement and concord. Cambridge: Cambridge University Press. Baptista, Marlyse. 1993. A comparative analysis of the particle ki in Cape Verdean and Haitian. In Hskuldur Thráinsson, Samuel D. Epstein, and Susumu Kuno (eds.), Harvard Working Papers in Linguistics, vol. 3, 3-12. Cambridge: Harvard Department of Linguistics. Baptista, Marlyse. 1997. The Morpho-Syntax of Nominal and Verbal Categories in Cape-Verdean Creole. Ph.D. dissertation, Harvard University. Baptista, Marlyse. 2002. The syntax of Cape Verdean Creole: The Sotavento varieties. Philadelphia/Amsterdam: John Benjamins. Brüser, Martina, André dos Reis Santos, Ekkehard Dengler & Andreas Blum, under the direction of Lang, Jürgen. 2002. Dicionário do Crioulo da Ilha de Santiago (Cabo Verde). Tübingen: Gunter Narr Verlag. Carstens, Vicky. 2003. Agree and EPP in Bantu. Natural Language and Linguistic Theory 23: 219-279. Carstens, Vicky. 2005. Rethinking complementizer agreement: Agree with a Case-checked goal. Linguistic Inquiry 34: 393-412. Chomsky, Noam. 1991. Some notes on economy of derivation and representation. In Freidin, Robert (ed). Principles and Parameters of Comparative Syntax, 417-454. Cambridge: MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge: The MIT Press. PAPIA, 25(2), e-ISSN 2316-2767 176 Marlyse Baptista / Miki Obata Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Roger Martin, David Michaels, and Juan Uriagereka (eds.) Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 89-155. Cambridge: MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Kenstowicz, Michael (ed.) Ken Hale: A life in language, 1-52. Cambridge: MIT Press. Chomsky, Noam. 2007. Approaching UG from below. In Sauerland, Uli and Hans-Martin Gärtner (eds.) Interfaces + Recursion = Language? 1-29, Berlin, Mouton de Gruyter. Chung, Sandra & James McCloskey. 1987. Government, barriers and small clauses in Modern Irish. Linguistic Inquiry 18: 173-237. DeGraff, Michel. 1992. The syntax of predication in Haitian. The North East Linguistic Society (NELS) 22: 103-117. Koopman, Hilda. 1985. The genesis of Haitian: Implications of a comparison of some features of the syntax of Haitian, French and West African languages. In Peter Muysken and Norval Smith (eds.), Substrata versus Universals in Creole Genesis, 279-30. Amsterdam/Philadelphia: John Benjamins. Koopman, Hilda. 2006. Agreement configurations: In defense of “spechead”. Ms., UCLA. Merchant, Jason. 2001. The syntax of silence: Sluicing, islands and the theory of ellipsis. Oxford: Oxford University Press. Moro, Andrea. 1997. The raising of predicates: predicative noun phrases and the theory of clause structure. Cambridge: Cambridge University Press. Obata, Miki, Samuel Epstein & Marlyse Baptista. 2015. Can crosslinguistically variant grammars be formally identical? Third factor under specification and the possible elimination of parameters of UG. Lingua 156: 1-16. Pesetsky, David & Esther Torrego. 2001. T-to-C movement: Causes and consequences. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 355-426. Cambridge: MIT Press. Rizzi, Luigi. 1990. Relativized minimality. Cambridge: MIT Press. Takahashi, Shoichi & Martina Graçanin-Yuksek. 2008. Morphosyntax of movement dependencies in Haitian Creole. Syntax 11(2): 223-250. Veiga, Manuel. 1995. O Crioulo de Cabo Verde: Introdução a Gramática. Praia: Instituto Nacional do Livro e do Disco/Instituto Nacional da Cultura. Veiga, Manuel. 2000. Le créole du Cap-Vert: Etude grammaticale descriptive et contrastive. Paris: Karthala. Received: 02/01/2015 Accepted: 07/31/2015 PAPIA, 25(2), e-ISSN 2316-2767