Object-sharing as Symmetric Sharing: predicate ...

Nat Lang Linguist Theory (2008) 26: 795–832 DOI 10.1007/s11049-008-9056-y O R I G I N A L PA P E R

Object-sharing as Symmetric Sharing: predicate clefting and serial verbs in Dàgáárè Ken Hiraiwa · Adams Bodomo

Received: 10 August 2007 / Accepted: 23 April 2008 / Published online: 5 November 2008 © Springer Science+Business Media B.V. 2008

Abstract Dàgáárè (a Gur language) allows various patterns of predicate clefting together with object pied-piping. This article investigates interactions of Predicate Cleft Constructions (PCCs) and object-sharing Serial Verb Constructions (SVCs) in Dàgáárè and argues that the object in object-sharing SVCs is symmetrically shared. Namely, we argue, building on Citko (2005), that it is an instance of Parallel Merge. Thus we present support for Baker’s (1989) insight of the Double-Headedness and against Collins’ (1997) VP-shell structure with a pro. This kind of empirical evidence is not available in other languages (cf. Baker 1989, Collins 1997 among others) and hence Dàgáárè provides a novel argument for a permissible structure of objectsharing SVCs and the availability of symmetric structure in UG. Keywords Predicate cleft · Serial verbs · Object-sharing · Double-Headedness · Pied-piping · Symmetry · Parallel Merge · Multi-dominance

We are grateful to Mark Baker, Seth Cable, Noam Chomsky, Chris Collins, Chizuru Ito, Jason Kandybowicz, Hironobu Kasai, Alec Marantz, David Pesetsky, Peter Sells, Akira Watanabe, three anonymous reviewers and the Associate Editor for this article for helpful comments, questions, and discussions at various stages of this project. The first author is grateful to the Grant-in-Aid for JSPS Fellows (No. 1710271). The second author gratefully acknowledges funding support from the Research Grants Council of Hong Kong (through RGC project No. HKU725104H). Dàgáárè is a Gur language mainly spoken in the northwestern parts of Ghana and in adjoining areas of Burkina Faso and Côte d’Ivoire. The data in this article come from the dialect spoken in the Jirapa district in Ghana. The abbreviations used in this article are as follows: C = Complementizer, Cnj = Conjunction marker, F = Focus marker, Fut = Future, Neg = Negation, Nml = Nominalizer, Perf = Perfective, Pl. = Plural, Pres = Present, Prog = Progressive, Pst = Past, Sg. = Singular. K. Hiraiwa () Department of Linguistics, University of Victoria, PO Box 3045, Victoria, BC V8W 3P4, Canada e-mail: [email protected] A. Bodomo Department of Linguistics, University of Hong Kong, Pokfulam Road, Hong Kong, China e-mail: [email protected]

796

K. Hiraiwa, A. Bodomo

1 Introduction How to make sense of the so-called Object Sharing phenomenon has been one of the important issues in the study of Serial Verb Constructions (see Baker 1989; Lefebvre 1991; Déchaine 1993; Collins 1997, 2002; Baker and Stewart 1999, 2002; Stewart 2001; Carstens 2002; among many others). In this article, building on novel data from interactions between a Predicate Cleft Construction (PCC) and a Serial Verb Construction (SVC) in Dàgáárè (a Gur language of the Niger-Congo Family), we argue that object-sharing SVCs in Dàgáárè must have a syntactic Symmetric Sharing structure and add to firm empirical support for Baker’s original intuition of “double-headedness” and syntactic object-sharing. The evidence comes from a particular case of object pied-piping in predicate clefting in an object-sharing SVC, which is not available in many other languages (see Ewe (Collins 1993, 1997), Bùlì (Hiraiwa 2005a, 2005b), Fngbe (Lefebvre and Brousseau 2002), Haitian Creole (Larson and Lefebvre 1991), Nupe (Kandybowicz 2006), Edo (Stewart 2001), Gungbe (Aboh 2004), among many others). An example of object-sharing SVCs in Dàgáárè is illustrated in (1). The object nńè is “shared” by the two transitive verbs s´ and ``. There are three issues that we address here. First, the object is sandwiched in between two verbs. Thus, (1a) is grammatical but (1b) is not. Second, there is only one object in the object-sharing SVC, even though there are two verbs. Finally, as we will see momentarily, even a (resumptive) pronoun is prohibited after V2 . (1)

a. ò dà s´ lá nńè ` ` . 3Sg. Pst roast F meat eat ‘He roasted meat and ate it.’ b. *ò dà s´ lá ` ` nńè. 3Sg. Pst roast F eat meat ‘He roasted meat and ate it.’

There have been various approaches to object-sharing SVCs in the literature. Baker (1989) proposes a ternary-branching structure (2a) in which the object is literally shared by the two verbs. Collins (1997), on the other hand, argues that objectsharing is mediated by a null pronoun pro. Thus, he proposes the VP-shell structure (2b) where the second verb (= V3 ) takes a pro and the first verb (= V2 ) takes the object (see also Nishiyama 1998; Carstens 2002 for similar proposals). Hale (1991) proposes yet another structure, the VP-adjunction structure (2c). This structure also posits a pro for the second verb (see also Déchaine 1993). (2)

a. Baker (1989)

Object-sharing as Symmetric Sharing

797

b. Collins (1997)

c. Hale (1991)

The curial observation that we will make in this article is that the direct object can be pied-piped either with V1 or V2 in Dàgáárè under predicate clefting. Namely, it can form a syntactic constituent with V1 or V2 . Based on this observation, we will claim that object-sharing in Dàgáárè involves syntactic symmetric sharing. It is argued that a symmetric structure is allowed in narrow syntax as long as it is converted into an asymmetric structure before Transfer, the point at which the syntactic object is handed over to PF and LF (see Chomsky 2000, 2001, 2004, 2008). The organization of this article is as follows. Section 2 first overviews objectsharing SVCs in Dàgáárè and establishes that they are SVCs and not covert coordination constructions. Then we examine predicate clefting in Dàgáárè and observe that PCCs in Dàgáárè allow object pied-piping and that PCCs interact with SVCs in interesting ways. Section 3 looks at a theory of PCCs to lay foundation for our main proposal. Section 4 proposes Symmetric Sharing in object-sharing SVCs in Dàgáárè. Section 5 discusses how Symmetric Sharing breaks and is converted into an asymmetric structure. Section 6 concludes the article.

2 Object-sharing SVCs and predicate clefting 2.1 The puzzle The puzzle that we are concerned with in this article is the following. As shown in (3b) and (3c), in Dàgáárè object SVCs, the object can form a syntactic constituent

798


both with V1 and V2 . Hence, either predicate clefting of V1 and the object or that of V2 and the object is grammatical.1 (3)

a. nńè sé´ó lá ká ó s´ ` ` . meat roast.Nml F C 3Sg. roast eat ‘It is roasting meat that he did and ate.’ (clefting of V1 + OBJ) b. nńè ´ ´ ó lá ká ó s´ ` ` . meat eat.Nml F C 3Sg. roast eat ‘It is eating meat that he roasted and did.’ (clefting of V2 + OBJ) c. nńè s´-´´ ó lá ká ó s´ ` ` . meat roast-eat.Nml F C 3Sg. roast eat ‘It is roasting meat and eating it that he did.’ (clefting of V1 + V2 + OBJ)

In other words, predicate clefting data show that the shared object forms a syntactic constituent with V1 and with V2 at the same time. As we will see, this is quite unexpected under the previous approaches to object-sharing SVCs. Given that syntactic constituency has been one of the core properties of syntax since the early days of generative grammar and continues to be important, the data that will be reported in this article will shed new light on the nature of syntactic constituency allowed in syntactic representations (see Carnie 2007 for an extensive survey of constituent structures within various frameworks). Thus the contribution of the present article would be important across frameworks. In order to appreciate the nature of the phenomenon, we first examine the syntax of object-sharing SVCs in Dàgáárè and show that they are real SVCs and depart from (c)overt coordination structures. Evidence comes from four tests: the single tense test, the negation test, the pronoun/empty category test, and the extraction test. Then we will take an in-depth look at the syntax of predicate clefting in Dàgáárè, followed by more extensive discussions of pied-piping patterns in predicate clefting of SVCs. 2.2 Object Sharing Serial Verb Constructions (SVC) are widely observed in West African languages and Creole languages (Awóbùrúyì 1973; Bamgbose 1974; Durie 1997; Déchaine 1993; Sebba 1987; Lefebvre 1991; Collins 1993, 1997; Stewart 2001; among many others). Among various types of SVC, “object-sharing” SVCs have been of particular theoretical interest, precisely because of the issue of how to formally represent objectsharing (see Baker 1989; Collins 1997, 2002 among others. For serial verbs in other languages, see Aikhenvald and Dixon 2006). Typical examples of object-sharing SVCs are illustrated below in Yorùbá and Ewe (Kwa languages). These are called the Consecutive SVC by Stewart (2001), since V1 and V2 express consecutive events. Note that there are two transitive verbs but only one shared object sandwiched between them. Crucially, no overt pronoun is allowed after the second verb. 1 PCCs are interpreted as contrastively focused and it is often difficult to translate in English. In what

follows, English translations below are not always felicitous and should be taken as more like word-byword glosses in such cases. See Sects. 2.3 and 2.4 for more discussions on semantic interpretations.


799

(4) Yorùbá (Lord 1974; Baker 1989) Blá sè ran tà (*á). cook meat Bola sell it ‘Bola cooked meat and sold it.’ (5) Ewe (Collins 1997, 480) M-a a fufu u (*-i). 1Sg.-Fut cook fufu eat it ‘I will cook fufu and eat it.’ In some languages, SVCs are sometimes hard to distinguish from Covert Coordination. For example, Collins (1997) convincingly shows that the covert coordination (6b) has two objects and is not an instance of SVC. (6) Ewe (Collins 1997) a. Me f o ka gb gba. 1Sg. hit lamp break ‘I hit the lamp and broke it.’ b. Me f o ka gb gba (yme) tsimini. 1Sg. hit lamp break its glass ‘I hit the lamp and broke its glass.’ Thus, it is very important to establish that Consecutive SVCs in Dàgáárè are indeed true SVCs. In this subsection, therefore, we present four tests to show that the SVCs in question are not coordinate constructions. A Consecutive SVC with object-sharing in Dàgáárè is illustrated in (7). (7) ò (dà) s´ lá nńè ` ` . 3Sg. Pst roast F meat eat ‘He roasted meat and ate it.’ According to Collins’ (1997) observation, a common descriptive characterization of SVCs is that they are clauses that have a single tense node, but two or more verbs, with no overt markers of coordination or subordination. The previous studies have identified several tests to distinguish between SVCs and other constructions (e.g. overt/covert coordination). In this section, some crucial hallmarks of SVCs reported in the literature will be examined for Dàgáárè and we first establish that object-sharing SVCs in Dàgáárè are identified as real SVCs (see Bodomo 1997a, 1997b, 2002 for various types of Dàgáárè SVCs). 2.2.1 The single tense-marker test First, only one tense-marker is allowed in SVCs (see Collins 1997). Thus in the SVC (8a), the past tense particle dà cannot be repeated before V2 . This contrasts with the overt coordination (8b), where the two VPs are coordinated by the overt coordination marker à and the tense particle can optionally be repeated. The same is true with the future particle nà as shown in (9a) and (9b).

800


(8)

a. ò dà s´ lá nńè (*dà) ` ` . 3Sg. Pst roast F meat Pst eat ‘He roasted meat and ate it.’ b. ò dà s´ lá nńè, à (dà) ` ` . 3Sg. Pst roast F meat Cnj Pst eat ‘He roasted meat and then ate it.’

(9)

a. ò nà s´ lá nńè (*nà) ´ ´ . 3Sg. Fut roast F meat Fut eat ‘He will roast meat and will eat it.’ b. ò nà s´ lá nńè, à (nà) ´ ´ . 3Sg. Fut roast F meat Cnj Fut eat ‘He will roast meat and then will eat it.’

As shown above, the repetition of the tense particle is only allowed when the coordination particle exists. This indicates that SVCs and coordination have different structures in Dàgáárè. 2.2.2 The pronouns/empty category test Second, an overt pronoun cannot appear after V2 in SVCs (Baker 1989; cf. Collins 1997), whereas such a pronoun is licit in overt coordination. As (10) shows, the 3rd person plural pronoun behaves differently in the SVC and the coordination. Although Collins (1997) argues that Ewe does allow what Collins considers to be a pronoun after V2 and supports his VP-shell theory, as far as Dàgáárè is concerned, there is no evidence that a pronoun can occur after V2 , in contrast with Ewe. (10)

a. ò dà s´ lá sìngkã´ ã` ` ` (*á). 3Sg. Pst roast F groundnut.Pl eat them ‘He roasted groundnuts and ate them.’ ` b. ò dà s´ lá sìngkã´ ã, à (dà) ` ` á. 3Sg. Pst roast F groundnut.Pl Cnj Pst eat them ‘He roasted groundnuts and then ate them.’

2.2.3 The negation test As shown in (11), in SVCs, the negation particle must appear before V1 , and crucially, it cannot be placed in front of V2 .2 (11)

a. ò bá s´ nńè ` ` . 3Sg. Neg roast meat eat ‘He did not roast meat and eat it.’

2 The particle lá must be absent in A-constructions ¯ and negation in Dàgáárè (see Sect. 2.3). For either

construction, it is not possible to repeat the negation particle in front of each verb in Dàgáárè. Such a repetition of the negative particle under overt coordination as in (12c) is possible in Fngbe.


801

b.?*ò s´ nńè bá ` ` . 3Sg. roast meat Neg eat ‘He roasted meat and didn’t eat it.’ c. *ò bá s´ nńè bá ` ` . 3Sg. Neg roast meat Neg eat ‘(Lit.) He did not roast meat and not eat it.’ In overt coordination, however, the negation can marginally appear in front of V2 . Thus the robust grammaticality difference between (11b) and (12b) indicates that SVCs are distinct from (c)overt coordination. (12)

a. ò bá s´ nńè ` ` . 3Sg. Neg roast meat eat ‘He did not roast meat and eat it.’ b. ò s´ lá nńè, à bá ` ` . 3Sg. roast F meat Conj Neg eat ‘He roasted meat and didn’t eat it.’ c. *ò bá s´ nńè, à bá ` ` . 3Sg. Neg roast meat Conj Neg eat ‘(Lit) He did not roast meat and not eat it.’

2.2.4 The extraction test Finally, extraction tests also establish that SVCs are not instances of covert coordination. As shown in (13), the SVC is free from the Coordinate Structure Constraint (Ross 1967) and hence the object can be extracted (see Stahlke 1970; Baker 1989; Hale 1991). On the other hand, in the overt coordination structure, extraction of the object makes the sentence degraded.3 (13)

a. bòng lá ká ó dà s´ ` ` ? what F C 3Sg. Pst roast eat ‘What did he roast and eat?’ à (dà) ` ` ó? b. *bòng lá ká ó dà s´ what F C 3Sg. Pst roast Cnj Pst eat it ‘What did he roast and then eat?’ à (dà) ` ` ? c. *bòng lá ká ó dà s´ what F C 3Sg. Pst roast Cnj Pst eat ‘What did he roast and then eat?’

All in all, the four tests demonstrate that Dàgáárè object-sharing SVCs are not a (c)overt coordination construction and patterns with regular SVCs. 3 (13c) can be greatly improved by putting a heavy pause between V and the conjunction marker à. 1 Although it is not clear to us why such a pause makes ATB-extraction possible, the contrast between the SVC (13a) and the coordination (13b) and (13c) indicates that they are not the same constructions. It should be noted that no pause is required between V1 and V2 in the SVC (13a), and in fact such a pause is incompatible with SVCs in general.

802


2.3 Predicate Cleft Constructions Dàgáárè, as many other (West) African languages do, allows a Predicate Cleft Construction (PCC) (see Koopman 1984, 1999; Manfredi 1993; Déchaine 1993; Hiraiwa 2005a, 2005b; Kandybowicz 2006, among others for PCCs). Predicate clefting patterns with constituent clefting in Dàgáárè. So first let us examine focus constructions in this language. (14)

a. n` dà dá lá bó´. 1Sg. Pst buy F goat ‘I bought a goat.’ b. bó´ lá ká n´ dà dà (*lá). goat F C 1Sg. Pst buy F ‘It is a goat that I bought.’ c. bòng lá ká fó dà dà (*lá)? what F C 2Sg. Pst buy F ‘What did you buy?’

As shown in (14b), the focused element bó´ is moved to the left periphery position. The declarative/focus marker lá now appears with the focused element followed by the complementizer ká. The same is true with Wh-questions, as shown in (14c). In Dàgáárè, only the element focused in the left periphery receives a contrastive focus interpretation. Thus in (14b), the focus interpretation can only contrast between a goat and something else. The particle lá in Dàgáárè is multi-functional and its functions are not yet fully understood. The main function of the particle is to indicate focus and assertion (see Bodomo 1997a; Dakubu 2005; see also Dakubu 2000 for a similar particle in a closely related language Guren). lá is obligatory in a matrix declarative clause and only one instance of lá is allowed within a single clause. We assume that it is a Foc head within vP. Thus it disappears under the scope of negation. (15) n` dà bá dà (*lá) bó´. 1Sg. Pst Neg buy F goat ‘I did not buy a goat.’ As mentioned above, in Wh/Focus constructions, lá appears at the left periphery of the clause and we assume it to be a Foc head above CP. There are good reasons to think that the higher instance of lá in Wh/Focus constructions cannot be analyzed as movement (or pied-piping) of the lower/in-situ instance of lá. First, in Wh/Focus constructions, lá appears even if the underlying sentence is negative, as shown in (16). Second, as shown in (17), unlike in Wh/Focus constructions, the higher lá cannot appear in relative clauses even though the lower lá must disappear. (16)

a. n` dà bá dà (*lá) bó´. 1Sg. Pst Neg buy F goat ‘I did not buy a goat.’


803

b. bó´ lá ká n` dà bá dà (*lá). goat F C 1Sg. Pst Neg buy F ‘It is a goat that I did not buy.’ c. bòng lá ká fó dà bá dà (*lá)? what F C 2Sg. Pst Neg buy F ‘What did you not buy?’ (17)

a. Dàkóráá s´ lá nńè. Dàkóráá roast F meat ‘Dàkóráá roasted meat.’ b. à nńè (*lá) Dàkóráá náng s´ (*lá) D meat F Dàkóráá C roast F ‘the meat that Dàkóráá roasted’

In the PCC (18b), the verb dá “to buy” is contrastively focused and is moved to the left periphery. Notice that the dislocated verb is obligatorily nominalized and the original copy of the verb must be pronounced, too (see Bodomo 2004 for nominalization of predicates). (18)

a. n` dà dá lá bó´. 1Sg. Pst buy F goat ‘I bought a goat.’

b. ‘It is buying that I did to a goat (as opposed to e.g. selling it).’ PCCs in Dàgáárè receive contrastive focus interpretations. Thus in (18b), as the translation shows, the clefted predicate is contrasted with another predicate. For example, the predicate clefting sentence (18b) is used in a context where someone asked “Did you say you SOLD a goat?” and the speaker wants to deny the statement, answering “No, no, it is buying that I did to a goat”. Another felicitous context is that someone asks a Yes-No question “Did you buy or sell a goat?” and (18b) is used as an answer. These potential contexts are given below. (19)

a. (fò yèlì ká) fò dà kò`ré lá bó´ béé? 2Sg. say C 2Sg. Pst sell F goat or ‘(Did you say that) you sold a goat?’ b. fò dà dá lá bó´ béé fò dà kò`ré lá bó´? 2Sg. Pst buy F goat or 2Sg. Pst sell F goat ‘Did you buy a goat or did you sell a goat?’

Another feature worth mentioning is the fact that in Dàgáárè, clefted predicates are obligatorily nominalized (see Bodomo 2004 for details about gerunds in the language). Languages are divided into two types: in Dàgáárè, Bùlì etc., clefted predicates are morphologically nominalized, while in Fngbe, Haitian Creole, etc., they are exact copies of the original predicates.

804


The fact that long-distance predicate clefting is allowed and predicate clefting out of an island shows strong ungrammaticality indicates that the dependency between the focused predicate and the original predicate is mediated by movement. (20)

a. Dàkóráá b`ng ká n` dà dá lá bó´. Dàkóráá know C 1Sg. Pst buy F goat ‘Dàkóráá knows that I bought a goat.’

b. ‘It is buying that Dàkóráá knows that I did to a goat (as opposed to e.g. selling it).’ In Dàgáárè, predicate-clefting can apply not only to stage-level predicates, but also to individual-level predicates in Diesing’s (1992) terms (see also Bùlì in Hiraiwa 2005b), unlike languages like Fngbe and Haitian Creole (see Larson and Lefebvre 1991; Lefebvre and Brousseau 2002). Again, felicitous contexts where these predicate clefting sentences are used include answers to the questions such as “Is Dàkóráá afraid of his father?” and “Have you heard about Dàkóráá before?”, respectively, or answers to Yes-No questions “Does Dàkóráá like his father or is he afraid of him?” and “Have you just heard about Dàkóráá or do you know him?”. (21)

a. Dàkóráá nˇng lá ò bâ. Dàkóráá like F 3Sg. father ‘Dàkóráá likes his father.’ b. nńgóó lá kà Dàkóráá n`ng ò bâ. like.Nml F C Dàkóráá like 3Sg. father ‘It is liking that Dàkóráá does to his father (as opposed to e.g. being afraid of him).’

(22)

a. n` bˇng lá Dàkóráá. 1Sg. know F Dàkóráá ‘I know Dàkóráá.’ b. bńgóó lá ká n´ b`ng Dàkóráá. know.Nml F C 1Sg. know Dàkóráá ‘It is knowing that I do to Dàkóráá (as opposed to e.g. hearing about him before).’

Now, significantly, the object can be optionally pied-piped in a PCC in Dàgáárè. Note that the object bó´ is also moved to the left periphery and the original copy of the object remains unpronounced. Note that the object comes to the left of the nominalized predicate (we will return to this in Sect. 3.2).4 4 In Dàgáárè, elements that can be pied-piped are restricted to noun phrases. This follows if the element

pied-piped needs to be assigned genitive Case, although we do not know why genitive Case cannot be assigned to the other elements. This merits future investigation. See Sect. 3.2.


805

(23) bó´ dááó lá ká n´ dà dà (*ò/*bó´). goat buy.Nml F C 1Sg. Pst buy (it/goat) ‘It is buying a goat that I did (as opposed to e.g. selling a hen).’ When the object and the predicate are focused by predicate clefting in Dàgáárè, the contrastive focus is placed on the entire clefted constituent. Thus, for example, (23) can be used as an answer to the questions “Did you buy a goat or did you sell your cow?” or “(Did you say that) you sold your cow?”. The pied-piping is not due to incorporation. As (24) shows, there is no restriction on the syntactic size of a pied-piped object: it can be a definite DP and, furthermore, it can even accompany adjectives. (24) à bó´/bó-vèlàà ná dááó lá ká n´ (dà) dà. D goat/goat-good Dem buy.Nml F C 1Sg. Pst buy ‘It is buying that (good) goat that I did.’ Furthermore, it is important to show that what looks like pied-piping is not derived from multiple focusing: fronting of the object and fronting of the predicate. That multiple focus fronting is not allowed in Dàgáárè is shown in (25) (neither is multiple Wh-fronting licit). The ungrammaticality of (25) convincingly shows that the pied-piping in (23) and (24) must be derived by moving a larger single syntactic constituent.

(25) ‘It is the knife, meat that he took and cut.’ As expected, when the object is pied-piped, the entire “object-predicate” complex is semantically contrasted with another. 2.4 PCC and SVC Now the most significant fact relevant to our discussion here is that PCCs can interact with SVCs in interesting ways (see Déchaine 1993; Hiraiwa 2005a, 2005b; Kobele 2006; among others). In the object-sharing SVC (26a), there are three possible predicate clefting patterns: either V1 or V2 is clefted or the V1 -V2 complex is clefted. (26)

a. ò dà s´ lá nńè ` ` . 3Sg. Pst roast F meat eat ‘He roasted meat and ate it.’ b. sé´ó lá ká ó s´ nńè ` ` . roast.Nml F C 3Sg. roast meat eat ‘It is roasting that he did and ate meat (as opposed to e.g. boiling it).’ (clefting of V1 ) c. ´ ´ ó lá ká ó s´ nńè ` ` . eat.Nml F C 3Sg. roast meat eat ‘It is eating that he roasted and did to meat (as opposed to e.g. throwing it away).’ (clefting of V2 )

806


d. s´-´´ ó lá ká ó dà s´ nńè ` ` . roast-eat.Nml F C 3Sg. Pst roast meat eat ‘It is roasting and eating that he did to meat (as opposed to e.g. buying something else).’ (clefting of V1 + V2 ) Again, the interpretation is contrastive in the sense that the clefted predicate is contrastively focused in the same way as the previous predicate clefting examples are interpreted. Dàgáárè is different from the Kwa language Fngbe in this respect. Lefebvre and Brousseau (2002) observe that in Fngbe, the whole serial verb receives a contrastive interpretation, irrespective of which V undergoes predicate clefting. On the other hand, in Dàgáárè, a subevent that a clefted predicate represents is always contrasted.5 Thus (26b), for example, can be used as an answer to a question “Did Dàkóráá roast meat and eat it, or did he boil and eat it?”. Now as we mentioned at the beginning of this section, the object in the objectsharing SVC can also be pied-piped either with V1 , V2 , or V1 -V2 as shown in (27a)– (27c). Crucially, pied-piping is possible with V2 and this makes Dàgáárè very unique and theoretically interesting, compared with other languages. Given our earlier observation that pied-piping is not due to multiple fronting, this means that the object can form a syntactic constituent either with V1 or V2 . (27)

a. nńè sé´ó lá ká ó s´ ` ` . meat roast.Nml F C 3Sg. roast eat ‘It is roasting meat that he did and ate (as opposed to e.g. boiling yam).’ (clefting of V1 + OBJ) b. nńè ´ ´ ó lá ká ó s´ ` ` . meat eat.Nml F C 3Sg. roast eat ‘It is eating meat that he roasted and did (as opposed to e.g. throwing away something else).’ (clefting of V2 + OBJ) c. nńè s´-´´ ó lá ká ó s´ ` ` . meat roast-eat.Nml F C 3Sg. roast eat ‘It is roasting meat and eating it that he did (as opposed to doing something else).’ (clefting of V1 + V2 + OBJ)

As the translation indicates, the pied-piped object is included in the scope of contrastive focus. For example, (27a) indicates that Dàkóráá ROASTED MEAT and ate it and that it is not the case that he boiled yam and ate it. A few more sets of examples with different lexical items are provided below. It should be noted that such patterns are completely productive and grammatical as long as proper contexts are entertained. 5 However, such a semantic difference does not seem to correlate with pied-piping patterns. For example,

another Gur language Bùlì also patterns with Dàgáárè in semantic interpretations but still the object cannot be pied-piped with V2 . We do not have any relevant information about other Kwa languages such as Yorùbá and Ewe regarding semantic interpretations. Because the main concern of this article is not a comparative study of SVCs in Dàgáárè and Fngbe and such a study is beyond the scope of this article, we will leave the source of variation open. Interested readers are referred to the detailed description of Fngbe in Lefebvre and Brousseau (2002) and references therein.


(28)

807

a. Dàkóráá póré lá kò´ nyú. Dàkóráá pour F water drink ‘Dakoraa poured water and drank it.’ b. kò´ póróó lá kà Dàkóráá póré nyú. water pour.Nml F C Dakoraa pour drink ‘It is pouring water that Dakoraa did and drank.’ c. kò´ nyúú lá kà Dàkóráá póré nyú. water drink.Nml F C Dakoraa pour drink ‘It is drinking water that Dakoraa poured and did.’ d. kò´ póré nyúú lá kà Dàkóráá póré nyú. water pour drink.Nml F C Dakoraa pour drink ‘It is pouring and drinking water that Dakoraa did.’

(29)

´ kò`rè. a. P´gdã´ ã` dà dógé lá dã´ ã-zè´ sell P´gdã´ ã` Pst boil F pito ´ ` ‘P´gdãã brewed pito and sold it.’ ´ dógóó lá kà P´gdã´ ã` dà dógé kò`rè. b. dã´ ã-zè´ pito boil.Nml F C P´gdã´ ã` Pst boil sell ‘It is boiling pito that P´gdã´ ã` did and sold.’ ´ kó´róó lá kà P´gdã´ ã` dà dógé kò`rè. c. dã´ ã-zè´ pito sell.Nml F C P´gdã´ ã` Pst boil sell ‘It is selling pito that P´gdã´ ã` boiled and did.’ ´ dógé kó´róó lá kà P´gdã´ ã` dà dógé kò`rè. d. dã´ ã-zè´ pito boil sell.Nml F C P´gdã´ ã` Pst boil sell ‘It is boiling and selling pito that P´gdã´ ã` did.’

That the object pied-piping must respect syntactic constituency is further confirmed by SVCs that do not have a shared object. In the SVC in (30), V1 is an intransitive verb and hence the object is not shared. In such cases, clefting of V1 with the object results in ungrammaticality, as shown in (30c). In the Instrumental SVC in (31), V1 and V2 are transitive verbs. However, the V1 “take” does not select the object “meat”. Again, clefting of V1 with the second object is disallowed. Neither is clefting of V2 with the first object allowed. (30)

a. n` dà wà dí lá kàpálà. 1Sg. Pst come eat F fufu ‘I came and ate fufu.’ b. kàpálà dííú lá ká n´ dà wà dì. fufu eat.Nml F C 1Sg. Pst come eat ‘It is eating fufu that I came and did to fufu.’ c. *kàpálà wááó lá ká n´ dà wà dì. fufu come.Nml F C 1Sg. Pst come eat ‘It is coming fufu that I did and ate fufu.’

808

(31)


a. ò dé lá à sò´ ngmàà nńè. 3Sg. take.Perf. F D knife cut meat ‘He cut meat with the knife.’ b. nńè ngmááó lá ká ó dé à sò´ ngmàà. meat cut.Nml F C 3Sg. take.Perf. D knife cut ‘It is cutting meat that he took the knife and did.’ c. *nńè dééó lá ká ó dé à sò´ ngmàà. meat take.Nml F C 3Sg. take.Perf. D knife cut ‘It is taking meat that he did the knife and cut.’ d. * à sò´ ngmááó lá ká ó dé ngmàà nńè. D knife cut.Nml F C 3Sg. take.Perf. cut meat ‘It is the knife cutting that he took and did meat.’

The same holds of benefactive SVCs (32). Notice that V2 cannot be pied-piped with the direct object. (32)

a. ò dà s´ lá nńè kó má. 3Sg. Pst roast F meat give 1Sg. ‘He roasted meat for me.’ b. nńè s´éó lá ká ó dà s´ kó má. meat roast.Nml F C 3Sg. Pst roast give 1Sg. ‘It is roasting meat that he did for me.’ c. *nńè kóó lá ká ó dà s´ kó má. meat give F C 3Sg. Pst roast give 1Sg. ‘It is for meat that he roasted for me.’

It is also worth noting that this kind of sharing in pied-piping is also found in resultative SVCs, as shown in (33). In (33), both verbs are transitive verbs and the object Dàkóráá is shared. Dàgáárè lacks SVCs in which the object of a transitive V1 is shared by the subject of an unaccusative V2 . (33)

a. n´ dáá lá Dàkóráá l´´ . 1Sg. push F Dàkóráá fell ‘I pushed Dàkóráá and got him down.’ b. Dàkóráá dááó lá ká n´ dáá l´´ . Dàkóórá push.Nml F C 1Sg. push fell ‘It is pushing Dàkóráá that I did and got him down.’ (clefting of V1 ) c. Dàkóráá l´´ ó lá ká n´ dáá l´´ . Dàkóórá fall.Nml F C 1Sg. push fell ‘It is getting down Dàkóráá that I pushed and did.’ (clefting of V2 )

One might wonder if the pied-piping patterns may be due to some linear adjacency condition rather than syntactic constituency. Thus, for example, one could argue that the reason why examples like (30c) and (31c) are ungrammatical is because they are not linearly adjacent to each other in the underlying surface form. Notice, however, that the ungrammaticality of (31d) and (32c) dismisses this possibility because the


809

objects and the verbs that are clefted in these examples are indeed adjacent in their pre-movement positions.6 The irrelevance of linear adjacency can be shown with double-object constructions. The word order between an indirect object and a direct object in this language is rigidly fixed—V IO DO.7 (34)

a. Dàkóráá kò lá D´rè à gánè. Dàkóráá give F D´ré D book ‘Dàkóráá gave D´ré the book.’ b. *Dàkóráá kò lá à gánè D´ré. Dàkóráá give F D book D´ré ‘Dàkóráá gave D´ré the book.’ c. à gánè kóó lá kà Dàkóráá kò D´ré. D book give.Nml F C Dàkóráá give D´ré ‘It is giving the book that Dàkóráá did to D´ré.’

The pied-piping patterns in Dàgáárè predicate clefting of object-sharing SVCs are typologically quite rare and significant. As we have observed, they provide direct evidence that the shared object is syntactically shared. The patterns in which serial verbs behave under predicate clefts have been only sporadically and briefly described in some languages (Manfredi and Láníran 1988; Déchaine 1993; Manfredi n.d. for Yorùbá; Stewart 2001 for Edo; Lefebvre and Brousseau 2002 for Fngbe; Kandybowicz 2006 for Nupe; Bodomo and Lam 2003 for Dàgáárè; and Lee 2003 and Hiraiwa 2005a, 2005b for Bùlì). One reason why the Dàgáárè-type patterns are rare is that many languages do not allow clefting of V2 in the first place and even if some do, they often do not allow pied-piping itself. For example, Bùlì, another Gur language, does not allow clefting of V2 in objectsharing SVCs. (35) Bùlì (Hiraiwa 2005a, 2005b) a. Àtìm sè l¯am `bì. Àtìm roasted meat eat ‘Atim roasted meat and ate it.’ b. (ká) s¯e-k¯a àl¯ı/àtì Àtìm sè l¯am `bì. F roast–Nml C Àtìm roasted meat eat ‘It is roasting that Àtìm did and ate meat.’ 6 We are grateful to an anonymous reviewer for pointing this out. 7 The status of the pied-piping of the indirect object is not grammatical, as shown in (i), which follows if we

assume the standard structure of double-object constructions in which DO is merged as the complement of V and IO is merged in its specifier or is introduced by a higher head (perhaps an applicative head). (i) *D´ré kóó lá kà Dàkóráá kò à gánè. D´ré give.Nml F C Dàkóráá give D book ‘It is giving D´ré that Dàkóráá did to the book.’

810


c. (ká) s¯e-¯b¯ı-k¯a àl¯ı/àtì Àtìm sè l¯am `bì. F roast-eat-Nml C Àtìm roasted meat eat ‘It is roasting and eating that Àtìm did to meat.’ d. *(ká) ¯b¯ı-k¯a àl¯ı/àtì Àtìm sè l¯am `bì. F eat-Nml C roasted meat eat Àtìm ‘It is eating that Àtìm roasted and did to meat.’ e. *(ká) l¯am ¯b¯ı-k¯a àl¯ı/àtì Àtìm sè `bì. F meat eat-Nml C Àtìm roasted eat ‘It is eating meat that Àtìm roasted and did.’ Similarly, Manfredi (n.d.) describes predicate clefting patterns in Yorùbá, in which clefting of V1 or V1 + V2 is grammatical, while clefting V2 alone results in ungrammaticality. (36) Yorùbá (Manfredi n.d.) a. Mo se eran tà. animal 1Sg. boil sell ‘I boiled [some/the] meat and [then] sold it.’ b. sí-se(-tà) ni mo se eran tà. ED -boil(-sell) C 1Sg. boil animal R sell ‘It is boiling and selling that I did to [some/the] meat and sold it.’ c. sí-se eran ni mo se eran tà. ED -boil meat animal R C 1Sg. boil sell ‘It is boiling [some/the] meat that I boiled and sold.’ d. *tí-tà (eran) ni mo se eran tà. R ED-sell meat C 1Sg. boil animal sell ‘It is selling [some/the] meat that I boiled and sold.’ Stewart (2001), on the other hand, observes that either V1 or V2 can be clefted as shown in (37b) and (37c), while clefting of V1 + V2 is ungrammatical (for morphological reasons). Crucially, however, Edo PPCs do not allow pied-piping of objects. (37) Edo (Stewart 2001) a. Òzó lé èvbàré ré. Ozo cook food eat ‘Ozo cooked the food and ate it.’ b. ùlémwn ré Òzó lé èvbàré ré. Nml-cook-Nml F Ozp cook food eat ‘It is cooking that Ozo cooked the food and he ate it, (not shred it).’ c. ùrémwn ré Òzó lé èvbàré ré. Nml-eat-Nml F Ozp cook food eat ‘It is eating that Ozo cooked the food (and ate it), (not shred it).’


811

Kandybowicz (2006) notes that Nupe behaves differently from Edo in that only the initial verb of a serial verb construction can be clefted. However, pied-piping is not allowed, either. (38) Nupe (Kandybowicz 2006) a. Musa du eci kun o. Musa cook yam sell F ‘Musa did to the yam before selling.’ b. Du-du Musa du eci kun o. R ED-cook Musa cook yam sell F ‘It was cooking that Musa did to the yam before selling.’ c. *Ku-kun Musa du eci kun o. R ED-sell Musa cook yam sell F ‘It was selling that Musa did after cooking.’ Lefebvre and Brousseau (2002) also briefly describe the behavior of SVCs under PCCs in Fngbe. First, in no case can the copy of both verbs occur simultaneously in clause-initial position. Second, the copy of any of the first verbs is allowed to be clefted, whereas the copy of the second verb is also possible in most cases (with some variation). But Fngbe crucially disallows pied-piping in PCCs (and in fact, pied-piping of objects is generally impossible in this language, whether in PCCs or in normal transitive sentences). (39) Fngbe (Lefebvre and Brousseau 2002) a. K`kú s´ àsń ´ yì àxì m`. Koku take crab D go market in ‘Koku brought the crab to the market.’ b. S´ w`, K`kú s´ àsń ´ yì àxì m`. take it.is Koku take crab D go market in ‘It is bringing the crab to the market that Koku did. (as opposed to e.g. selling it)’ c. o.k./* Yì w`, K`kú s´ àsń ´ yì àxì m`. go it.is Koku take crab D go market in ‘It is bringing the crab to the market that Koku did. (as opposed to e.g. selling it)’ d. *S´ yì w`, K`kú s´ àsń ´ yì àxì m`. take go it.is Koku take crab D go market in ‘It is bringing the crab to the market that Koku did. (as opposed to e.g. selling it)’ The chart (40) summarizes the patterns of PCCs and SVCs.8 8 The “multi-event” type of SVC does allow predicate clefting of V . See Déchaine (1993), Manfredi 2

(n.d.). Also, Jason Kandybowicz (p.c.) notes that predicate clefting of V1 + V2 in Nupe is rather marked and only optionally allowed in certain resultative SVCs.

812

(40) Cross-linguistic Variation of PCCs and SVCs Bùlì Yorùbá Edo Nupe Fngbe V1 o.k. o.k. o.k. o.k. o.k. V1 + Obj o.k. o.k. * * * V2 * * o.k. * o.k. V2 + Obj * * * * * V1 + V2 o.k. o.k. * (*) *


Dàgáárè o.k. o.k. o.k. o.k. o.k.

Despite the variation, some interesting generalizations emerge. While it seems to be cross-linguistically true that no language disallows clefting of V1 (unless V1 is reanalyzed as an aspectual auxiliary) and those languages that allow pied-piping also allow predicate clefting of entire serial verbs (with a confound in Nupe), the patterns observed above show that all the languages fail to provide the kind of data that Dàgáárè can. Namely, among these languages, only Dàgáárè allows clefting of V2 alone and pied-piping of the object. This makes Dàgááreè outstanding and unique.9 The Dàgáárè data lead us to the following descriptive generalization: (41) In Dàgáárè, not only V1 and the object can form a syntactic constituent excluding V2 , but also V2 and the object can form a syntactic constituent excluding V1 . With this in mind, let us consider the syntactic structure of the object-sharing SVCs in Dàgáárè. However, before doing so, it is worth spending some time in explicating a theory of PCCs that we adopt in this article.

3 A theory of PCC 3.1 Hiraiwa (2005a, 2005b) There have been a number of proposals about the syntactic derivation of PCCs (Koopman 1984, 1999; Manfredi 1993; Harbour 2008; Hiraiwa 2005a, 2005b; Kandybowicz 2006). In this article, we adopt Hiraiwa’s (2005a, 2005b) theory of PCCs and review important aspects of his proposals. A number of important studies on the parallelism between DPs and CPs (Chomsky 1970; Abney 1987; Lefebvre and Muysken 1988; Szabolcsi 1994; Collins 2001; among many others) have revealed that the DP domain and the CP domain exhibit a large degree of parallelism. Building on Rizzi’s articulated left periphery, Hiraiwa (2005a, 2005b) proposes the following symmetric structure between CP and DP.10 9 An anonymous reviewer suggests that the cross-linguistic variation might be correlated with whether

focused predicates are morphologically nominalized or they are exact copies in PCCs. Although the suggestion is interesting, among those languages in the chart, only Fngbe belongs to the latter type of language, while the other languages use morphological nominalization, and Edo and Fngbe exhibit the same patterns despite such a difference. 10 See Bernstein (2001) for arguments for NumP and Travis (1991) for VP-internal AspP.


813

(42) CP/DP Parallelism (Hiraiwa 2005a, 2005b)

One important property that is shared by PCC languages cross-linguistically is that a predicate is doubled in PCCs: the predicate is pronounced in the focused position as well as in the original position. Abels (2001) proposes that this follows from a morphological principle. In Russian PCCs, the predicate is obligatorily pronounced in its original position as in (43a), but it cannot be pronounced when there is an auxiliary as shown in (43b). (43) Russian (Abels 2001) ˇ (-to) Ivan eë *(ˇcitaet), no niˇcego ne ponimaet. a. Citat’ but nothing not understand read(I NF) (TO) Ivan it(F EM .ACC) read ‘Ivan does read it, but he doesn’t understand a thing.’ ˇ (-to) on budet cˇ itaet. b. Citat’ read(I NF) (TO) he will read ‘He will read.’ Abels (2001) proposes to attribute the difference to the Stranded Affix Filter (see Lasnik 1981, 1995): (44) A morphologically realized affix must be a syntactic dependent of a morphologically realized category, at surface structure. Now according to this principle, the difference above is due to the fact that in (43a), T’s inflectional features are otherwise stranded in the absence of verb movement or an overt particle in T, but in (43b), they are taken care of by the overt future particle budet. In other words, if predicate clefting leaves T’s morphosyntactic features stranded, the copy of the predicate is pronounced to save the stranded features.11 11 Interestingly, Aboh (2004) has some similar data where predicate clefting does not require the pronun-

ciation of the original copy in Gbe languages among Kwa languages. Crucially, those cases have an overt aspectual auxiliary particle, as expected.

814


This analysis cannot be straightforwardly extended to PCCs in Gur and Kwa languages because the presence of a preverbal tense particle does not allow the original copy to be unpronounced. Example (45) illustrates the point in Dàgáárè. (45)

a. ò nà s´ lá nńè. 3Sg. Fut roast F meat ‘He will roast meat.’ b. sé´ó lá ká ó nà *(s´) nńè. roast.Nml F C 3Sg. Fut roast meat ‘It is roasting that he will do to meat.’

Hiraiwa (2005a, 2005b), however, argues that this is due to a difference in what is moved. He proposes, following Marantz (1997, 2007), Arad (2005), and Embick and Marantz (2008) among others, √ that syntactic categories such as verbs and nouns are decomposed into a root ( r) plus a category-determining head v/n. Notice that in the verb (46a), the root is dominated by v, whereas in the noun (46b), the same root is dominated by n (for a similar idea of root movement, see Harbour 2008). (46)

a. Verb: v + Root

b. Noun: n + Root

Hiraiwa (2005a, 2005b) argues, extending Abels’ insight,√that PCCs in Bùlì (and Kwa √ languages) target the minimal category—namely, Asp- r—rather than v-Aspr). This is supported by evidence from Case-marking. Dàgáárè does not have overt Case-marking but 1st person pronouns show different forms depending on Case: má for the accusative and n` for the nominative/genitive. In PCCs, the fronted predicate is not capable of assigning accusative Case and hence the pronoun must take the nominative/genitive form, as observed by Bodomo (2004). (47)

a. Dàkóráá ngm´ má/*n` lá. Dàkóráá beat 1Sg.Acc/1Sg.Nom F ‘Dàkóráá beat me.’


815

` b. n/*má ngm´ó lá kà Dàkóráá ngm`. 1Sg.Nom/Gen/1Sg.Acc beat.Nml F C Dàkóráá beat ‘It is beating me that Dàkóráá did.’ Hiraiwa (2005a, 2005b) proposes that the burden of the interface conditions naturally carries over to v√ in the sense that the “verbalizer”—being affixal—cannot be stranded without a r at Spell-Out. Reframing it under the proposed theory of CP/DP Symmetry, at Spell-Out, the v head must determine the category of its complement. Therefore, the original copy of the predicate (i.e. the complement of v) must be phonologically realized. The remaining question is how the focused predicate is nominalized at the left periphery. Hiraiwa (2005a, 2005b) claims that the key is Lefebvre’s Generalization (48). (48) Lefebvre’s Generalization (Lefebvre 1992) The availability of predicate clefting in a given grammar is correlated with the presence of a syntactic position for clausal determiners within S (=IP). Lefebvre (1992) observes that a dialect of Fngbe, Haitian Creole, and Yorùbá allow both clausal determiners and PCCs while another dialect of Fngbe, which lacks the former, also lacks the latter. The Fngbe examples below show that the language allows a PCC (49a) and the determiner ´ also occurs as a sentence-final clausal determiner, as shown in (49b) (we gloss w` as F(ocus) not “it-is” as originally done in Lefebvre 1992). (49) Fngbe a. Lń w` súnù ´ lń. jump F man D jump ‘It is jump that the man jumped.’ (Lefebvre 1992) b. Súnù ´ gbà m´tò ´ ´ . man D destroy car D D ‘The man destroyed the car.’ (Larson 2003) (50) PCCs and Clausal Determiners (cf. Lefebvre 1992) PCC Clausal D √ √ Bùlì, Yorùbá, Fngbe (A), Gungbe, Haitian Creole, Vata, (dialects of) Ewe, Ga * * Standard Ewe, Fngbe (B) √ * Russian, Yiddish, Hebrew, Polish Hiraiwa (2005a, 2005b) argue that while, as DeGraff (1994) observes, the generalization seems to be too strong given the fact that there are languages that allow PCCs but that disallow clausal determiners (e.g. Russian, Hebrew, Yiddish), it captures an important fraction of the truth about PCCs. That is, PCCs in those languages that have clausal determiners obligatorily nominalize the focused predicates in one way or another. (51) In languages that allow a clausal determiner, focused predicates in PCCs are nominalized (Hiraiwa 2005a, 2005b).

816


He argues that nominalization takes place in the course of the narrow syntactic derivation. As we have seen, what is moving in PCCs in Bùlì and other languages is not an element whose categorial status is established, but rather an element that awaits categorial determination. The clausal determiner D/C(=Force) in our CP/DP parallel geometry in (42) functions as a categorial determiner under the CP/DP Symmetry and causes nominalization. Note that Lefebvre’s insightful observation and our proposal (51) focus on the correlation between a clausal determiner and predicate clefting/nominalization in a language. Hence it does not mean that predicate clefting realizes an overt clausal determiner. An example of a clausal determiner in Dàgáárè is given below, where the definite determiner à heads the entire factive clause. (52) à Dàkóráá náng ngmàà à gánè dèyâng y´l´ p`lé lá ò D Dàkóráá C write.PERF D book last.year matter whiten.PERF F 3Sg. p´g(´) pó´ yágà. wife stomach a.lot ‘The fact that Dàkóráá wrote the book last year pleased his wife a lot.’ Now let us see how the derivation works in Dàgáárè. 3.2 PCCs in Dàgáárè We assume that lá is a focus head (see Sect. 5 for more detailed discussion on lá). In Dàgáárè, what moves in PCCs is the AspP: namely, the complement of v. First, the AspP undergoes a successive-cyclic focus movement to [Spec, FocP] as in (53). (53)

Then, the highest C(=Force) is merged with FocP as in (54). By (51) and the CP/DP parallelism (42), it follows that C functions as a clausal determiner and hence a categorial determiner. Thus, the root in the focused position is realized as a noun.


817

In contrast, notice that the lowest copy of AspP must be realized as well, because otherwise v is stranded. Since v is a categorial determiner, the root in the original position is realized as a verb. (54)

We propose that object pied-piping is a result of higher pronounciation of the copy of the object. If, on the other hand, the original copy is pronounced, no pied-piping surfaces. (55)

a. [AspP . . . OBJ ] C . . . v [AspP . . . OBJ ] (object pied-piping) b. [AspP . . . OBJ ] C . . . v [AspP . . . OBJ ] (no pied-piping)

A caveat is in order here. As we know, the pied-piped object obligatorily precedes the nominalized predicate in PCCs. (56)

a. bó´ dááó lá ká n´ dà dà (*ò/*bó´). goat buy.Nml F C 1Sg. Pst buy (it/goat) ‘It is buying a goat that I did.’ b. *dááó bó´ lá ká n´ dà dà (*ò/*bó´). buy.Nml goat F C 1Sg. Pst buy (it/goat) ‘It is buying a goat that I did.’

Thus in (54), something must require the object to come before the root. We argue that this is done by movement. It is very important to point out that DPs in Dàgáárè are mostly head-final and only the definite determiner à and a possessor precede the root. Recall that the pied-piped object pronouns are obligatorily marked in the nominative/genitive Case rather than in the accusative Case in this language.

818


(57) à Dàkóráá gámà nˇa. D Dàkóráá’s book.Pl. Dem ‘those books of Dàkóráá’s’ Thus in Dàgáárè, within a DP, a licensing position for a noun phrase is its specifier position and a noun phrase moves there from its original position. The surface ordering is derived as follows: The root first undergoes head-movement to adjoin to Num and then, the whole DP undergoes movement to [Spec, DemP].12 (58)

a. [DemP [Dem ná ] [DP [D à ] [nP Dàkóráá [NumP -ma [√r gan- ]]]]] b. [DemP [Dem ná ] [DP [D à ] [nP Dàkóráá [NumP [√r gan- ]-ma t√r ]]]] c. [DemP [DP [D à ] [nP Dàkóráá [NumP [√r gan- ]-ma t√r ]]] [Dem [Dem ná ] tDP ]]

In gerunds, it is possible to have two arguments (see also Bodomo 2004). In such cases, the subject must precede the object. (59) Dàkóráá nńè sé´ò. Dàkóráá’s meat roast.Nml ‘Dàkóráá’s roasting of meat’ Given this, it is natural to think that the pied-piped object in PCCs is located in a lower position. We assume this position to be [Spec, AspP/NumP] and the movement of the object takes place at the CP phase level—simultaneously with the focus movement of AspP to [Spec, FocP]. Recall that the fronted AspP is still categorially neutral in the sense that it contains a root and a category-determining head is missing (it is no longer c-commanded by the phase head v). Thus, the closest categorial determiner is the higher phase head C/D. Under the framework adopted in this article (Chomsky 2004, 2008), at this phase level, categorial determination and the AspPinternal movement apply at once.13 A DP within a DP or a gerund can only receive the nominative/genitive Case-marking, as we have already seen, because they lack v.

12 As an anonymous reviewer points out, object pied-piping in Dàgáárè and Bùlì differs from Yorùbá in

that in the latter, the object follows the clefted predicate and seems to receive accusative Case (see Hiraiwa 2005a, 2005b). The same reviewer suggests an insightful possibility that the difference might come from the fact that the nominalizing morpheme is a suffix whereas it looks like a prefix and a circumfix in Yorùbá and Edo, respectively. He/she proposes that there may be a principle that requires that the pied-piped object must be adjacent to the verb root. Then it follows that the pied-piped object must precede the nominalized predicate in Dàgáárè (because if it follows, the object is separated from the root by the nominalizing suffix), while it must follow in Yorùbá (because if it precedes, the object is separated from the root by the reduplicant prefix). In Edo, in contrast, the nominalizing morpheme being a circumfix, the object has no way to be adjacent to the verb root and hence no pied-piping is allowed in this language. We have to leave further investigation for future research, but we believe that this is one important direction to seek an answer to the variation in (40). We are very grateful to the anonymous reviewer. 13 We will argue in Sect. 5 that a full DP object shifts to [Spec, AspP] in Dàgáárè.


819

(60)

Summarizing this section, we have introduced a theory of PCCs and object piedpiping therein. Then we have illustrated how the proposed theory accounts for the basic facts of PCCs in Dàgáárè. However, it cannot straightforwardly explain the pied-piping patterns in object-sharing SVCs in Dàgáárè. In the next section, we will propose a symmetric-sharing structure for object-sharing SVCs and show that all the observed patterns follow from it under the theory of PCC introduced in this section.

4 Object Sharing as Symmetric Sharing 4.1 The issue Our main empirical observation is summarized as follows: (61) In Dàgáárè, not only V1 and the object can form a syntactic constituent excluding V2 , but also V2 and the object can form a syntactic constituent excluding V1 . This fact is entirely unexpected under the previous analyses of object-sharing SVCs. On Baker’s approach, it is correctly expected that the object can be piedpiped together with V1 and V2 . However, because of the ternary structure, V1 -Object and V2 -Object constituencies are not straightforwardly explained. Collins’ analysis also suffers a problem. It is not immediately clear how the constituencies in (61) are formed.14 Finally, Hale’s VP-adjunction structure also fails to explain the constituency of V2 and the object excluding V1 , due to the null pronoun.15

14 We should note here that Collins’ analysis is motivated by his Ewe data. It is possible that constraints of

UG make available more than one structure for object-sharing SVCs and thus we leave for future research a parametric investigation of object-sharing SVCs. In fact, pied-piping of the object with V2 has not been well attested in other languages, to the best of our knowledge. 15 There is another possible analysis in which the object undergoes sideward movement from the comple-

ment of V2 to the complement of V1 in (62c). This could explain (61), but it raises a question of whether sideward movement should be allowed in UG. As far as we know, no such analysis of object-sharing SVCs has been proposed in the literature.

820

(62)


a. Baker (1989)

b. Collins (1997)

c. Hale (1991)

As an anonymous reviewer points out, one could still entertain a possibility that predicate clefting patterns might be explained under Hale’s theory with the assumption that each verb underlyingly has a full DP object and one of them is “pronominalized”. Although it is not easy to find direct evidence against such a hypothesis, we have some indirect evidence that object-sharing SVCs in Dàgáárè could not involve a pro(nominalization). First, Collins (1997) observes interesting data in Ewe, in which the shared object surfaces as two distinct but related elements. (63a) is an example of object-sharing SVC. In contrast, in (63b), note that there are two different objects. (63) Ewe (Collins 1997) a. Me f o ka gb gba. 1Sg. hit lamp break ‘I hit the lamp and broke it.’


821

b. Me f o da gb gba (yme) tsimini. 1Sg. hit lamp break its glass ‘I hit the lamp and broke its glass.’ Although Collins argues that this is only possible in covert coordination constructions in Ewe, it is plausible that examples like (63b) are only possible if there are two distinct instances of the shared object syntactically in SVCs, as in Collins’ theory or Hale’s theory. But significantly, the counterpart of (63b) is not grammatical in Dàgáárè. The ungrammaticality would remain a mystery if object-sharing SVCs involved a pro(nominalization) with Hale’s structure. (64)

a. Dàkóráá dà ngm´ lá à kyáánáá ngmárè. Dàkóráá Pst beat F D mirror break ‘Dàkóráá hit and broke the glass.’ b. *Dàkóráá ngm´ lá à lámpè ngmárè ò kyáánáá. Dàkóráá beat F D lamp break 3Sg. mirror ‘Dàkóráá hit the lamp and broke its glass.’

Second, Cable (2004) observes that Yiddish and Brazilian Portuguese allow what he calls Genus-Species PCCs. In these languages, a pied-piped object can be doubled by an object in-situ only when the former and the latter are in a whole-part relation. The examples below are from Cable (2004). (65)

a. Yiddish ?Essen fish est Maks hekht. to-eat fish eats Max pike ‘As for eating fish, Max eats pike.’ b. Brazilian Portuguese Comer peixe, eu normalmente como samão. to-eat fish 1Sg. usually eat salmon ‘As for eating fish, I usually eat salmon.’

Again, as expected under our theory of PCCs, this is not possible in Dàgáárè. (66)

a. zóm` ´ nnóó lá kà Dàkóráá ` nnè ` ` . fish.Pl. grill-Nml F C Dàkóráá grill eat ‘It is grilling fish that Dàkóráá did and ate.’ b. *zóm` ´ nnòò lá kà Dàkóráá ` nnè p´´ ré ` ` . fish.Pl. grill-Nml F C Dàkóráá grill herring.Pl. eat ‘It is grilling fish that Dàkóráá did and ate herrings.’

Assuming that this is only possible with a non-movement chain (i.e. base-generation and a pro(nominalization)), the ungrammaticality of such sentences in Dàgáárè follows from our theory, which does not involve a pro(nominalization), contra Collins (1997) or Hale (1991). As we will see below, all the facts naturally make sense if the structure contains only one shared object.

822


4.2 Symmetric Sharing Based on the observation in (61), we propose that the object in object-sharing SVCs (in Dàgáárè) is structurally shared. We call this Symmetric Sharing as shown in (67). (67) The object is symmetrically shared by both verbs in SVCs in Dàgáárè. By symmetric sharing, we mean that an element Y is merged with X and Z simultaneously as shown in (68). This is called Parallel Merge by Citko (2005). In other words, it constitutes a multi-dominance structure (see Citko 2005 and references therein). (68)

Applying symmetric sharing to object-sharing SVCs, we propose that objectsharing SVCs have the following syntactic structure in (69). Under this structure, the object is merged with V √1 and V2 at the same time (just for expository purposes, we use “V” rather than “ r here). Both V1 and V2 have the argument structure agent, theme and the Theme θ -role is assigned to the shared object by the two verbs. The structure crucially differs from Baker’s (1989) theory in that V1 and V2 are not merged. Thus in (69), the pied-piping patterns reduce to the movement of the syntactic constituent AspP1 or the movement of another constituent AspP2 . If the higher constituent AspP1+2 is moved, (27c) results. (69)

The proposed structure has two important aspects to note. First, it is “doubleheaded” in the sense of Baker (1989) and Baker and Stewart (1999). Namely, the


823

entire Asp Phrase consists of two distinct Asp projections, each of which dominates V1 and V2 , respectively. We assume (70), following Baker and Stewart (1999). (70) Double-Headedness Double-headed structures are allowed when features of each head are identical or nearly so (Baker and Stewart 1999). Second, the proposed symmetric sharing structure raises the question: to what extent is such Parallel Merge allowed in narrow syntax? We adopt the position that Parallel Merge is licit in narrow syntax and interface conditions require that such a structure be rendered linearizable before Spell-Out. (71) Parallel Merge and Linearization Parallel Merge (or Ternary Branching) is allowed in narrow syntax as long as the structure is made linearizable before Spell-Out. (see Yang 1999; Wilder 1999; Moro 2000; Citko 2005) That AspP is double-headed in object-sharing SVCs in Dàgáárè is empirically supported by the fact that imperfective aspectual markers must be realized on each V as shown in the intransitive SVCs and object-sharing SVCs. If VP were double-headed, this would be unexpected. (72)

a. ò dà zò-ró g`-r´ wùò-ró lá hááné. 3Sg. Pst run-Imperf go-Imperf collect-Imperf F berries ‘(Lit.) S/he was running, going there, and collecting berries.’ b. *ò dà zò-∅ g`-r´ wùò-ró lá hááné. 3Sg. Pst run go-Imperf collect-Imperf F berries ‘(Lit.) S/he was running, going there, and collecting berries.’ c. *ò dà zò-ró gàà-∅ wùò-ró lá hááné. 3Sg. Pst run-Imperf go collect-Imperf F berries ‘(Lit.) S/he was running, going there, and collecting berries.’

(73)

a. ò dà s´ lá nńè ` ` . 3Sg. Pst roast F meat eat ‘He roasted meat and ate it.’ b. ò dà sè´-r´ lá nńè ` ` -r´. 3Sg. Pst roast-Imperf F meat eat-Imperf ‘He was roasting meat and eating it.’

5 Symmetry breaking and object shift In this final section, we address two important challenges that our theory of Symmetric Sharing faces. First, as we have stated, symmetric sharing is not linearizable by itself (Kayne 1994; Chomsky 1995; Citko 2005). Thus, it is necessary to break the symmetry before Spell-Out. Citko (2005) proposes that ATB Wh-movement in English has a parallel merge structure underlyingly. She argues that the shared element must undergo movement because otherwise the structure is not linearizable. Thus Wh-movement of what results in the licit asymmetric structure.

824


(74) English ATB-Movement and Parallel Merge (Citko 2005) a. I wonder what Gretel recommended and Hansel read. b.

Second, symmetric sharing does not explain why the object is sandwiched on the surface in object-sharing SVCs. To see this, let us consider (75). In earlier discussions, we have assumed, without any argument, that symmetric sharing takes the form as shown in (75a) (where X = V1 , Y = Object, Z = V2 ). However, nothing inherent to symmetric sharing forces (75a); in fact, nothing prevents us from imagining (75b). The crucial difference between these options is the surface order: the structure in (75a) has a sandwiched order, whereas (75b) has a “V1 V2 Object” order. Thus some mechanism must exist to ensure the sandwiched surface order. (75)

a.

b.

5.1 Short verb movement Significantly, there is evidence that the verb in Dàgáárè undergoes short movement. As described in Sect. 2 and Bodomo (1997a), almost every main (declarative) clause in Dàgáárè has the particle lá, as shown in (76). (76)

a. Dàkóráá dà s´ lá nńè. Dàkóráá Pst roast F meat ‘Dàkóráá roasted meat.’ b. *Dàkóráá dà s´ nńè. Dàkóráá Pst roast meat ‘Dàkóráá roasted meat.’

As we have seen, this particle is the same focus particle as what we find in WhQuestions and (Predicate) Celfting. The difference is its position. In normal declarative sentences, the focus particle must follow the verb. But in Wh/Focus constructions, the particle lá appears with a Wh/Focus element at the left periphery. (77)

a. bòng lá ká ó dà s´? what F C 3Sg. Pst roast ‘What did he roast?’ b. nńé lá ká ó dà s´. meat F C 3Sg. Pst roast ‘It is meat that he roasted.’


825

Thus we propose that lá is a focus head and there are two focus projections: a higher FocP and a lower FocP. The clausal architecture of Dàgáárè is represented as follows. Specifically, the lower FocP is located “VP-internally”, below vP and above AspP. (78)

Now, significantly, the main verb must precede the focus particle lá in normal declarative sentences, as shown in (79a). Thus, (79b) is ungrammatical. This indicates that the verb moves to v in Dàgáárè. (79)

a. n` ngm´ lá Dàkóráá. 1Sg. hit F Dàkóráá ‘I hit Dàkóráá.’ b. *`n lá ngm´ Dàkóráá. 1Sg. F hit Dàkóráá ‘I hit Dàkóráá.’ c.

826


5.2 Object shift It is also important to note that weak pronominal objects always shift to the left of the particle lá. As shown in (80), the weak pronoun ó must front to the left of lá (but to the right of the moved main verb). Assuming that the verb moves to v, a natural hypothesis is that the weak pronoun cliticizes onto the particle lá. This derivation is represented as in (81). (80)

a. n` ngm´ ó lá. 1Sg. hit 3Sg. F ‘I hit him.’ b. *`n ngm´ lá ó. 1Sg. hit F 3Sg. ‘I hit him.’

(81)

In contrast, a full DP object cannot undergo the same movement. Rather, it must remain to the right of the focus particle lá. (82)

a. n` ngm´ lá Dàkóráá. 1Sg. hit F Dàkóráá ‘I hit Dàkóráá.’ b. *`n ngm´ Dàkóráá lá. 1Sg. hit Dàkóráá F ‘I hit Dàkóráá.’

However, let us formulate the following hypothesis that even though a full DP object doesn’t undergo movement past the Foc head, it does undergo “short” object shift (see Chomsky 2008; Hiraiwa 2005b for short object shift). (83) A Hypothesis (Object Shift) Full DP objects always shift to the edge of AspP (but not to the edge of vP) in Dàgáárè.


827

Combining (83) with the V-to-v movement, the derivation of SVCs is represented as in (84). V1 moves to v and the shared object undergoes short object shift to the specifier of AspP1+2 . Importantly, the verb movement and the object shift change a symmetric structure to an antisymmetric structure. Furthermore, these two operations correctly derive the sandwiched word order for object-sharing SVCs. Thus the hypothesis (83) gives us a simple answer to both challenges. Assuming that operations apply simultaneously (Chomsky 2004, 2008; Hiraiwa 2005b), predicate clefting, verb movement, and object shift all apply simultaneously at the vP phase level. As a result, the symmetric structure is converted to an asymmetric structure in which V1 asymmetrically c-commands the object and the object asymmetrically c-commands V2 . (84)

A question arises as to what principle chooses V1 , but not V2 as the target for Vto-v movement.16 Apparently, our symmetric structure predicts that either verb could be moved to v, contrary to the fact. This is because either verb is equally close to v. (85)

a. ò dà s´ lá nńè ` ` . 3Sg. Pst roast F meat eat ‘He roasted meat and ate it.’ b. *ò dà ` ` lá nńè s´. 3Sg. Pst eat F roast meat ‘He ate meat and roasted it.’

However, we adopt the strong minimalist thesis here and assume that Merge comes free in principle. Thus, in narrow syntax, either V1 or V2 can indeed move to v. We propose that the derivation with the illicit surface order is excluded at LF. 16 Thanks to Jason Kandybowicz and an anonymous reviewer for helpful comments and/or discussions.

Another possibility is to abandon the double-headedness and to adjoin AspP2 to AspP1 . This singles out V as the target for V-to-v movement (while it faces some adjunct island issues).

828


Li (1993) for example proposes the Temporal Iconicity Condition which stipulates an ordering between the verbs (see also Muysken 1988; Bodomo 1997b; Nishiyama 1998 for related proposals). (86) Temporal Iconicity Condition (Li 1993, 499) Let A and B be two subevents (activities, states, changes of states, etc.) and let A’ and B’ be two verbal constituents denoting A and B, respectively; then the temporal relation between A and B must be directly reflected in the surface linear order of A’ and B’ unless A’ is an argument of B’ or vice versa. We assume that the Temporal Iconicity Condition is an LF constraint, requiring the temporally preceding verb to take a scope over the other. The scope relation is determined based on the asymmetric c-command relation between the two verbs. Now, as we mentioned, Merge comes free without any constraint. And the PF requirement forces one of the verbs to move to v in narrow syntax to break the symmetry. The two possible derivations are illustrated below (suppose V1 = “roast” and V2 = “eat”). There is no problem with any of these derivations as far as narrow syntax is concerned. At LF, however, the structure (88) is ruled out due to the iconicity requirement (86), because V2 asymmetrically c-commands V1 . (87)


829

(88)

Needless to say, it is an important remaining issue to investigate the nature of such a condition, which goes beyond the scope of this article. It should be noted, however, that a similar assumption is implicitly adopted and required in any of the previous approaches. All the three approaches in (62) implicitly encode some information about which verb is merged as V1 (and hence closer to v) on the Merge operation. Otherwise, there is no way to prevent the other verb from being merged as V1 (or they might also implicitly assume a version of the LF iconicity principle that we assume here). Thus, the only difference lies in where the principle like (86) comes in during the derivation—at the time of Merge or at LF. While we do not see a difference in empirical predictions between the two approaches, it is expected that V1 and V2 can freely swap in SVCs if there is no temporal precedence pragmatically. Unfortunately, we do not find any example where verbs can swap in object-sharing SVCs in Dàgáárè. But examples can be found with intransitive SVCs as shown in (89). (89)

a. Dàkóráá wà gàà` lá. Dàkóráá come go F ‘Dàkóráá came here and went away.’ b. Dàkóráá gàà wà` lá. Dàkóráá go come F ‘Dàkóráá went somewhere and came back.’

6 Conclusion and implications In this article, we have presented novel data from Dàgáárè about interactions between predicate clefting and object-sharing SVCs and shown that in Dàgáárè, V2 as well as V1 can form a syntactic constituent with the object in object-sharing SVCs. Thus, building on Baker (1989) and Citko (2005), we have proposed that object-sharing

830


SVCs must have a Symmetric Sharing structure. It has been claimed that the syntactic symmetry breaks down before Spell-Out because of V-to-v movement and short object shift. Let us mention some theoretical implications that our proposed theory of objectsharing has. First, as far as we know, adverb-sharing or PP-sharing in the form [V1 Adv/PP V2 ] has not been attested cross-linguistically. This is in fact exactly what our theory expects. As we have shown, to break the symmetry, object shift is crucially required. Because adverbs and PPs cannot undergo object shift, the observation follows. This in turn brings another prediction to light. As stated in footnote 3, (61) is rarely observed in other languages. Our theory links those languages without (61) to the absence of object shift and verb movement. Without them, symmetric sharing crashes at PF and hence is not allowed. Although a quick look suggests this prediction seems to be borne out, further cross-linguistic investigation remains to be done. The data presented in this article and the proposed theory have a broader theoretical implication for a theory of phrase structure. Syntactic constituency has been one of the core properties of syntax and will continue to be significant unless the relevance of syntactic constituency is entirely eliminated from syntax, together with a number of generalizations built on it. The proposed symmetric sharing theory thus presents another argument for the relevance of syntactic constituency. The proposed theory also contributes to the “old” but never solved question about permissible multidominance structures (or crossing branches) (cf. McCawley 1982; Goodall 1987; Wilder 1999; Moro 2000; Citko 2005; among others).

References Abels, Klaus. 2001. The predicate cleft constructions in Russian. In Proceedings of the FASL 9, eds. Steven Franks, T.H. King, and M. Yadroff, 1–19. Ann Arbor: Michigan Slavic Publications. Abney, Steven. 1987. The English noun phrase in its sentential aspect. PhD dissertation, MIT. Aboh, Enoch Oladé. 2004. The morphosyntax of complement-head sequences: Clause structure and word order patterns in Kwa. New York: Oxford University Press. Aikhenvald, Alexandra Y., and R.M.W. Dixon. 2006. Serial verb constructions: A cross-linguistic typology. New York: Oxford. Arad, Maya. 2005. Roots and patterns: Hebrew morphosyntax. Berlin: Springer. Awóbùrúyì, Oládélé. 1973. The modifying serial construction: A critique. Studies in African Linguistics 4: 87–110. Baker, Mark. 1989. Object sharing and projection in serial verb constructions. Linguistic Inquiry 20: 513– 553. Baker, Mark, and Osamuyimen T. Stewart. 1999. On the double-headedness and the anatomy of the clause. Manuscript, Rutgers University. Baker, Mark, and Osamuyimen T. Stewart. 2002. A serial verb construction without constructions. Manuscript, Rutgers University. Bamgbose, Ayo. 1974. On serial verbs and verbal status. Journal of West African Linguistics 9: 17–48. Bernstein, Judy B. 2001. The dp hypothesis: Identifying clausal properties in the nominal domain. In The handbook of contemporary syntactic theory, eds. Mark Baltin and Chris Collins, 536–561. Oxford: Blackwell. Bodomo, Adams. 1997a. The structure of Dagaare. Stanford: CSLI. Bodomo, Adams. 1997b. Paths and pathfinders: exploring the syntax and semantics of complex verbal predicates in Dagaare and other languages. PhD dissertation, Trondheim: Norwegian University of Science and Technology. Bodomo, Adams. 2002. The syntax of serial verbs in Dagaare. In New directions in Ghanaian linguistics, eds. Felix Ameka and Kweku Osam, 27–54. Accra: Advent Press.


831

Bodomo, Adams. 2004. The syntax of nominalized complex verbal predicates in Dagaare. Studia Linguistica 58(1): 1–22. Bodomo, Adams, and Olivia Lam. 2003. Modelling verb order in complex multi-verbal predicate constructions. Paper presented at the 17th Pacific Asia conference on language, information and computation, Singapore. Manuscript, University of Hong Kong. Cable, Seth. 2004. Predicate clefts and base-generation: Evidence from Yiddish and Brazilian Portuguese. Manuscript, MIT. Carnie, Andrew. 2007. Constituent structures. New York: Oxford University Press. Carstens, Vicki. 2002. Antisymmetry and word order in serial constructions. Language 78(1): 3–50. Chomsky, Noam.. 1970. Remarks on nominalization. In Studies on semantics in generative grammar, 11–61. The Hague: Mouton. Chomsky, Noam. 1995. The minimalist program. Cambridge: MIT Press. Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Step by step: Essays on minimalist syntax in honor of Howard Lasnik, eds. Roger Martin, David Michaels, and Juan Uriagereka, 89–155. Cambridge: MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Ken Hale: A life in language, ed. Michael Kenstowicz, 1–52. Cambridge: MIT Press. Chomsky, Noam. 2004. Beyond explanatory adequacy. In Structures and beyond: The cartography of syntactic structures, vol. 3, ed. A. Belletti, 104–131. New York: Oxford University Press. Chomsky, Noam. 2008. On phases. In Foundational issues in linguistic theory, eds. Robert Freidin, Carlos P. Otero, and Maria Luisa Zubizarreta, 133–166. Cambridge: MIT Press. Citko, Barbara. 2005. On the nature of merge: External merge, internal merge, and parallel merge. Linguistic Inquiry 36(4): 475–496. Collins, Chris. 1993. Topics in Ewe Syntax. PhD dissertation, MIT. Collins, Chris. 1997. Argument sharing in serial verb constructions. Linguistic Inquiry 28(3): 461–497. Collins, Chris. 2001. Aspects of plurality in Hoan. Language 77: 456–476. Collins, Chris. 2002. Multiple verb movement in Hoan. Linguistic Inquiry 33(1): 1–29. Dakubu, Kropp M.E. 2000. The particle la in Guren. Gur Papers 5: 59–65. Dakubu, Kropp M.E. 2005. Dagaare: Collected language notes, vol. 26. Legon: Institute of African Studies. Déchaine, Rose-Marie. 1993. Predicates across categories: towards a category-neutral syntax. PhD dissertation, University of Massachusetts, Amherst. DeGraff, Michel. 1994. A review of Travaux de recherche sur le créole haïtien, no. 8. The clausal determiners of Haitian and Fon. By Claire Lefebvre. Université du Québec à Montréal. Journal of Pidgin and Creole Languages 9(2): 370–376. Diesing, Molly. 1992. Indefinites, Cambridge: MIT Press. Durie, Mark. 1997. Grammatical structures in verb serialization. In Complex predicates, eds. Alex Alsina, Joan Bresnan, and Peter Sells, 289–354. Stanford: CSLI. Embick, David, and Alec Marantz. 2008. Architecture and blocking. Linguistic Inquiry 39(1): 1–53. Goodall, Grant. 1987. Parallel structures in syntax, Cambridge: Cambridge University Press. Hale, Ken. 1991. Misumalpan verb sequencing. In Serial verbs: Grammatical, comparative and cognitive approaches, ed. C. Lefebvre, 1–36. Amsterdam/Philadelphia: John Benjamins. Harbour, Daniel. 2008. Klivaj predika, or predicate clefts in Haitian. Lingua 118: 853–871. Hiraiwa, Ken. 2005a. Predicate clefts in Bùlì: Phase and category. Linguistic Analysis: a special volume: African linguistics in the new millennium 32(3/4): 544–583. eds. R. Letsholo and N.C. Kula. Hiraiwa, Ken. 2005b. Dimensions of symmetry in syntax: agreement and clausal architecture. PhD dissertation, MIT. Kandybowicz, Jason. 2006. Conditions on multiple copy spell-out and the syntax phonology interface. PhD dissertation, University of California, Los Angeles. Kayne, Richard. 1994. The antisymmetry of syntax, Cambridge: MIT Press. Kobele, Greg. 2006. Generating copies: An investigation into structural identity in language and grammar. PhD dissertation, UCLA. Koopman, Hilda. 1984. The syntax of verbs: from verb movement rules in the Kru languages to universal grammar, Dordrecht: Foris. Koopman, Hilda. 1999. Unifying predicate cleft constructions. In The syntax of specifiers and heads, 366– 382. London and New York: Routledge. Larson, Richard. 2003. Event descriptions in Fon and Haitian Creole. In Creole languages from a generative perspective, ed. D. Adone. Tubingen: Niemeyer.

832


Larson, Richard, and Claire Lefebvre. 1991. Predicate clefting in Haitian Creole. In The proceedings of the NELS 21, 247–261. Amherst: GLSA. Lasnik, Howard. 1981. Restricting the theory of transformations: A case study. In Explanations in linguistics, eds. Norvert Hornstein and David Lightfoot, 152–173. London: Longman. Lasnik, Howard. 1995. Verbal morphology: Syntactic structures meets the minimalist program. In Evolution and revolution in linguistic theory: Essays in honor of Carlos Otero, eds. Héctor Campos and Paula Kempchinsky, 251–275. Washington: Georgetown University Press [Reprinted in Minimalist analysis. Oxford: Blackwell. 1999]. Lee, Youngjoo. 2003. Serial verb constructions in Buli. In Studies in Buli grammar, eds. George AkanligPare and Michael Kenstowicz. Vol. 4 of MIT working papers on endangered and less familiar languages, 97–116. Cambridge: MITWPL. Lefebvre, Claire. 1991. Serial verbs: Grammatical, comparative and cognitive approaches. Amsterdam/Philadelphia: John Benjamins. Lefebvre, Claire. 1992. Towards a typology of predicate cleft languages. In Travaux de recherche sur le créole haïtien: Numéro 8–Avril 1992, 69–80. Montréal: Départment de linguistique, Université du Québec à Montréal: Groupe de rechercher sur le créole haïtien. Lefebvre, Claire, and Anne-Marie Brousseau. 2002. A grammar of Fongbe. Berlin: Mouton de Gruyter. Lefebvre, Claire, and Pieter Muysken. 1988. Mixed categories: nominalization in Quechua. Dordrecht: Kluwer Academic. Li, Yafei. 1993. Structural head and aspectuality. Language 69(3): 480–504. Lord, Carol. 1974. Causative constructions in Yoruba. Studies in African Linguistics Supplement 5: 195– 204. Manfredi, Victor. 1993. Verb focus in the typology of Kwa/Kru and Haitian. In Focus and grammatical relations in Creole languages, eds. F. Byrne and D. Winford, 3–51. Amsterdam: John Benjamins. Manfredi, Victor. n.d. Observations about “predicate clef”/verb focus in Yòrùbá serializations. Notes, Boston University. Manfredi, Victor, and Y. Láníran. 1988. Extraction from Yòrùbá serializations. Manuscript, Paper presented at the 2nd Niger-Congo Syntax and Semantics Workshop at MIT. Marantz, Alec. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. UPenn Working Papers in Linguistics 4(2): 201–225. Marantz, Alec. 2007. Phases and words. In Phases in the theory of grammar, ed. Sook-Hee Choe, 191–222. Seoul: Dong In. McCawley, James. 1982. Parentheticals and discontinuous constituent structure. Linguistic Inquiry 13: 91–106. Moro, Andrea. 2000. Dynamic antisymmetry. Cambridge: MIT Press. Muysken, Pieter. 1988. Parameters for serial verbs. In Niger-Congo syntax and semantics 1, ed. Victor Manfredi, 65–75. Boston: Boston University African Studies Center. Nishiyama, Kunio. 1998. V-V compounds as serialization. Journal of East Asian Linguistics 7: 175–217. Ross, John Robert. 1967. Constraints on variables in syntax. PhD dissertation, MIT. Sebba, M. 1987. The syntax of serial verbs: an investigation into serialization in Sraman and other languages, Amsterdam: John Benjamins. Stahlke, Herbert. 1970. Serial verbs. Studies in African Linguistics 1: 60–99. Stewart, Osamuyimen T. 2001. The serial verb construction parameter. New York: Garland. Szabolcsi, Anna. 1994. The noun phrase. In The syntax of Hungarian. Vol. 27 of Syntax and semantics, 179–274. San Diego: Academic Press. Travis, Lisa. 1991. Derived objects, inner aspect, and the structure of vp. A paper presented at NELS 22. Wilder, Chris. 1999. Right node raising and the LCA. In The proceedings of WCCFL 18, eds. Sonya Bird, et al., 586–598. Somerville: Cascadilla Press. Yang, Charles D. 1999. Unordered merge and its linearization. Syntax 2(1): 38–64.

Object-sharing as Symmetric Sharing: predicate ...

Object-sharing as Symmetric Sharing: predicate ...

Suggest Documents

Object-Sharing as Symmetric Sharing: Evidence from DÃ gÃ¡Ã¡rÃ¨

Multibaryons as Symmetric Multiskyrmions

Cooperads as Symmetric Sequences

Predicate Interpretations as Intensions restricted along Dimensions

Airbus A320 Braking as Predicate-Action Diagrams

Ranking Abstraction as Companion to Predicate Abstraction*

Mycosis fungoides presenting as symmetric

Predicate-based Test Coverage and Generation Predicate-complete ...

Illative Combinatory Logic Without Equality as a Primitive Predicate

Predicate 'which'-Appositives

SHARING BY YOUNG CHILDREN' Sharing, as a unique ... - NCBI

Sharing talk, sharing cognition: philosophy with children as the ... - cejsh

The Semantics of Predicate Logic as a Programming Language

The Semantics of Predicate Logic as a Programming Language

Online Distributed Predicate Evaluation

Multiple Symmetric Lipomatosis (Madelung's Disease) Presenting as ...

Predicate Liftings Versus Nabla Modalities - CyberLeninkawww.researchgate.net › publication › fulltext › Predicate-

Ranking Abstraction as Companion to Predicate Abstraction - CiteSeerX

Predicate Transformation as a Proof Strategy - Semantic Scholar

Predicate composition and argument extension as general options - a ...

Ranking Abstraction as Companion to Predicate ... - Semantic Scholar

5. PREDICATE DE INTERACŢIUNE

Predicate dan Quantifier

MATRICES FOR PREDICATE LOGICS