Date: Tue, 11 May 2004 22:21:30 +0200 (CEST) From: Hana Skrabalova Subject: The Structure of Coordination
Camacho, José (2003) The Structure of Coordination: Conjunction and Agreement Phenomena in Spanish and Other Languages, Kluwer Academic Publishers, Studies in Natural Language and Linguistic Theory 57.
Hana Skrabalova, University of Paris 3-Sorbonne Nouvelle.
INTRODUCTION This book analyzes the structure of coordination with respect to the properties of conjuncts and the way conjuncts interact with other categories outside coordination (agreement and other grammatical phenomena). A substantial amount of data are taken from varieties of Spanish.
SYNOPSIS The book contains an introduction (chapter 1) and four other chapters (chapter 2 to 5). In the introduction, the author announces his proposals concerning the structure of coordination: (i) one conjunct must c- command the others (c-command asymmetry), (ii) each conjunct should reflect the same structural properties as if it were in a simplex sentence (licensing symmetry), and (iii) coordination always involves a set of sentential functional projections(cf. Chomsky 1957, ''Conjunction Reduction'').
In chapter 2, the authors deals with properties of coordination and reviews different proposals concerning the structure of coordination (Goodall 1987, Johannessen 1998, Höhle 1990, Munn 1993, etc.). First, he claims that each conjunct is licensed in the same way as it would be if it were not conjoined. Consequently, all conjuncts must have the same syntactic features (cf. Pullum & Zwicky's (1986) 'Law of Coordination of Likes') and the same syntactic status. Therefore, the author argues against proposals involving Conjunction Phrase, because they assume that conjuncts are structurally asymmetric. There follows a brief discussion of asymmetric coordinations which are claimed not to constitute evidence against the general symmetry of conjuncts. The main discussion of asymmetries concerning partial agreement and coordination of different categories are postponed to chapters 4 and 5 respectively. Second, the author argues, following Munn (1993), that one of the conjuncts should be structurally higher and c-command the other conjunct. Finally, he claims that all coordinations are interpreted sententially. This claim is based on the following observations: (1) Coordination and plural DPs differ in their distribution and in their entailment: (i) contrary to non conjoined bare NPs, conjoined bare NPs can appear in preverbal subject position, and (ii) only conjoined NPs, not plural DPs entail that each individual participed in the event. (2) Sentential adverbs can modifiy conjoined DPs, but not plural DPs. (3) Plural- like categories corresponding to conjoined verbs, adjectives, or adverbs are cross-linguistically absent.
In chapter 3, the author develops his proposal concerning the structure of coordination. First, he assumes that conjunctions are heads (cf. Johannessen 1998). Assuming also that coordinate structures are as underspecified as possible (cf. Gazdar et al. 1985) and that coordination is sentential in nature (cf. above), he suggests ''conjunction be a sentential functional head that has a propositional content'', represented by a single feature (PROP) on the conjunction. To account for the fact that coordination has the distribution of its conjuncts, the author argues that conjunction copy features from another functional category. Depending on the position of the conjunction, a different functional head will license the conjunction by giving it its content: subject coordination will be licensed by Inflection, cf.(1), while object coordination will be license by Agr-o. The fact that each conjunct in (1) is in the same structural position (Spec-IP) as it would be in a simplex sentence is supposed to reflect the assumption that a coordination of subjects is sentential.
(1) [IP DP1 [I' Co(tns,f...)[IP DP2[I' I (tns,f...) [VP...]]]]]
Coordination of other categories than DPs is assumed to have the same underlying structure, see (2a), but differs with respect to the head X licensing the conjunction (y).
(2) [XP Conjunct1 [X' y [XP Conjunct2 [X' X [...]]]]]
The author's main argument for a sentential analysis of coordination comes from switch-reference phenomena. Switch reference systems in languages like Hopi, Ono and Mojave involve morphological markers that signal coreference relation between arguments (usually subjects) of a main clause and a subordinate clause. Several researchers had observed parallelism between coordination and switch reference (Hale & Jeanne 1976, Haiman 1983). Hale & Jeanne (1976) note, for instance, that switch reference (SR) markers in Hopi always adjoin to inflected verbs, and that these markers are identical to those used for coordination. The author proposes an analysis of SR in Hopi and of comitative coordination in Mojave where the fact that SR markers adjoin both to inflected verbs and to conjunctions follows naturally from his proposal that coordination is a sentential functional projection. The author next endeavors to corroborate his hypothesis: (i) heads cannot be conjoined, and (ii) conjoined elements do not form a constituent. Given coordination is generally assumed to behave as a constituent, the last part of the chapter deals with deriving the constituency effects (with respect to movement, agreement and binding) without coordination forming a constituent. In order to do that, the author suggests that conjuncts are generated in their surface position and that they are linked to a silent phrase (little pro), which is the constituent that undergoes movement, cf.(3):
(3) John(i)and Mary(j)seem pro(i+j)to have been called t(i+j)
The author argues that an argument in favor of a silent phrase comes from the distribution of subject bare NPs in Spanish. According to him, the fact that non conjoined bare NPs are only licensed in postverbal position, whereas conjoined bare NPs may appear in preverbal explains if they are linked to a pro in postverbal position (a position licensing bare NPs). The subsequent analysis of coordination assumes that coordination involves a chain between conjuncts and a silent category. It also assumes that lexical categories are categorial matrices, i.e. bundles of features identified by a categorial label. In the typical case, all the features appear together under the same matrix. But the features can also be inserted in a different syntactic position than the matrix (in checking projections). Consequently, movement is viewed as an operation by which a matrix fills its unspecified values. In coordination, part of the features of the chain are inserted in the lowest chain position and move to the conjuncts. In the case of conjoined subjects, for instance, agreement features of the conjoined DPs are generated in the Spec-IP, while case features and theta- role are inserted in the matrix of the silent category (DPx), generated in Spec-VP. This category matches the feature specifications of each conjunct: an important assumption is that plurality is a sum of singularities (SG, SG). The DPx moves to the Spec-IP where it fuses with the matrix of the second conjunct (DP2). Fusing is a partial copying since the categories are not identical. The resulting matrix (DP2/x) will have its theta-role and case specified. It will check the features of the I and then move up to the specifier of the conjunction and merge with the first conjunct (DP1). The resulting matrix will check the features of the conjunction, identical to I.
(4) a. Lucía y Yesi corren. Lucia and Yesi run. b. [yP [DP1 Lucia][y' y [IP [DP2 Yesi][I' I [VP [DPx]corren ]]]]] c. [yP [DP1(THETA,CASE,SG,3P)][y' y(TNS,NOM,SG,SG) [IP [DP2(THETA,CASE,SG,3P)][I' I(TNS,NOM,SG,SG) [VP [DPx(AGENT,NOM,SG,SG)]...]]]]]
In chapter 4, the author returns to agreement asymmetries. First, he claims that partial agreement (i.e. agreement with one conjunct only) shows in the non canonical word orders. Second, he claims that conjoined structures with partial agreement can be divided into two types: those that behave as if agreement had no interpretative consequences (PF agreement structures) and those where agreement patterns do have interpretative consequences (LF agreement structures). Partial V-S agreement in Irish and Czech, as well as clitic and adjectival agreement in Spanish are claimed to be cases of PF agreement. On the contrary, partial V-S agreement in Moroccan and Lebanese Arabic and in Spanish, as well as copular and adjectival agreement in Brazilian Portuguese are claimed to be cases of LF agreement. The analysis of partial agreement (PA) is based on the theory of categories and feature insertion (cf. above). The author argues that LF PA involves gapping in the second VP conjunct (cf. ABS's (1994) clausal analysis of PA in Arabic) and that it is covert agreement restricted to one of the specifiers in the structure of coordination. Agreement features of the verb of the first conjunct are generated in a functional head F above XP. Therefore, the verb must move up to F to have it matrix filled, cf. (5a). After Spell-Out, the higher (i.e. closest) conjunct moves to the Spec-FP to check agreement with the verb in F, cf. (5b). (The analysis of gapping is dealt with in chapter 5).
(5) a. [FP [F' Vi [XP [VP DP1 ti ][X' and [XP [VP DP2 ei [PP ...]][X' ...]]]]]] b. [FP DP1 [F' Vi [XP [VP tDP1 ti ][X' and [XP [VP DP2 ei [PP ...]][X' ..]]]]]]
As for PF PA, the author assumes Aoun & Benmamoun's (1999) postsyntactic merger proposal and suggests that before the derivation of coordination branches off to the interpretative component the structural configurations allow for plural antecedents. Postsyntactically, the first conjunct will raise to the specifier of the projection headed by the V, yielding the PF agreement paradigm. The postsyntactic merger analysis is applied also to partial dative agreement in clitic doubling structures in Spanish.
Finally, the author proposes an analysis of PA with nominal modifiers in Spanish. In Spanish, determiners and prenominal adjectives cannot have full agreement, though they can have scope over both conjuncts. The author argues (against Longobardi 1994) that cases of apparent NP coordination inside DP involve full DPs coordinations with null structure licensed under identity, cf.(6). More generally, the author claims that the lowest category that can be conjoined inside DP is a predicative phrase (PredP), as predicted by his proposal that coordination affects functional categories with predicational content. PA with prenominal elements is therefore a regular agreement within the first conjunct.
(6) [DP1 la supuesta imagen]y[DP2(D) (ADJ) reflejo] the alleged image and (the)(alleged)reflexion
As for postnominal adjectives, the author suggests that they agree in a higher position that coordination since conjoined DPs and PPs displaying full number agreement and partial gender agreement still may have scope over both conjuncts, cf.(7). He thus introduces an agreement projection above conjoined DPs/PPs which can also be a scopal projection. Given the feature-insertion analysis assumed for coordination, PA takes place between the features of the lower head and the second conjunct, whereas full agreement (FA) takes place where agreement features are inserted in the higher head X (Agreement head).
(7) [[NP-M.SG] and [NP-F.SG]] ADJ-F.PL
The author concludes that coordination involves a functional projection both in the DP and in the IP. The difference is that the agreeing head in the IP case is a verb, cf.(8a), and that it the DP case, it is an adjective, cf.(8b).
(8) a. [IP DP1 [I' [and [IP DP2 [I' [Verb]]]]]]] b. [XP X [YP DP1 [Y' [and [YP DP2 [Y' [ADJ]]]]]]]
Finally, the author extends his analysis with Agr Phrase (XP) to double conjunction structures. He argues that exclusive / distributive reading of these double conjunctions structures is linked to the first conjunction and these operators occupy the head Agr.
(9) [XP ni[YP DP1 [Y'[ni[YP DP2 [Y'[...]]]]]]]
In the last chapter, the author deals with asymmetric coordinations involving two conjuncts of apparently different categories (cf. Höhle (1990) for German). Given his assumption that conjuncts must belong to like- categories (cf. chapter 2), he argues that these coordinations involve gapping. Following Zoerner (1995), the author proposes that gapping and Right-Node-Raising (RNR) involve conjunction of predicational projections. Gapping is claimed to differ from DP coordination in that it relates VPs in the specifiers of conjunction nodes. As for RNR constructions, the authors suggest that the shared object is not raised, but rather it is complement of the second conjunct, since there are constructions where only the structurally lower verb has its selectional restrictions satisfied by the ''shared'' object.
EVALUATION The book provides the reader with a wide range of interesting data concerning clitics, nominal modifiers, and PP coordination in Spanish. The idea of applying the feature-insertion theory and matrix movement to coordination is original. The proposal to introduce an agreeing head above coordination allows in particular an elegant account of cases of mixed agreement with postnominal modifiers. An extended use of this head also allows for a nice account of double conjunction structures: despite of their formal identity, the conjunctions are not structurally identical, since each of them is a different head. This accounts for their different syntactic distribution (before and between conjuncts) and different semantic properties (distributive and exclusive readings are due to the first conjunction).
However, the whole proposal is based on two claims which seem to me to be problematic. First, the claim that conjuncts are licensed in the same way as if they were not conjoined predicts for instance that collective predicates are compatible with singular NPs, which is incorrect. The fact that conjoined bare NPs in Spanish can function as preverbal subjects contrary to non conjoined bare NPs (cf. chapter 2) also shows that the distribution of conjuncts and that of coordination are not always parallel. Second, the claim that all coordination it sentential is mostly based on analysis of coordination and switch reference in Hopi and in Mojave. Though convincing, it is not clear why this analysis should be equally assumed for languages without any SR system. Other arguments (chapter 2) do not show either that coordination must be sentential. For instance, nominal coordinations where sentential adverbs are infelicitous, as in (7b), are not mentioned and remain unexplained.
(10) a. Perhaps John and maybe Mary will come. b.*Perhaps John and maybe Mary will meet. c. John and Mary will meet.
This claim also has undesirable consequences for both the representation and derivation of coordinate structures. The structure of non nominal coordinations (chapter 3) contains functional projections which do not seem independently motivated, and whose identity is rather vague (Event Phrase, Predicational Phrase). The derivation of clausal coordination (p. 59-62) involves several movements apparently motivated only by the necessity to derive a correct word order.
The feature-insertion analysis, though attractive, assumes that plurality is a sum of singularities, cf.(3) above. But, if two singular features are assumed to yield a semantic plural, what would the representation of plural in generic DPs or universally quantified DPs be ?
Although partial agreement receives a lot of attention, the author does not say how he would derive full agreement with postverbal conjoined subjects. This leaves the picture of agreement phenomena incomplete. Note also that it is not clear why the projection above coordination allows to obtain PA in case of V-S agreement (p.120), but full agreement in case of modifier agreement (p.135).
More crucially, the structure of DP coordination with postnominal modifiers assumes that conjunction is licensed by a predicational functional head, hosting the adjective, cf.(9a). This structure parallels the structure of subject DP coordination where conjunction is licensed by the verbal functional head, cf.(11b). Then, it is not clear at all what the structure of a subject coordination modified by a postnominal adjective should be: if the head licensing the conjunction were I, as in (11a), where would the ADJ be, and vice versa ?
(11) a. [IP DP1 [I'[and [IP DP2 [I'[I...]]]]]] b. [XP X [YP DP1 [Y'[and [YP DP2 [Y'[ADJ...]]]]]]
Finally, the argumentation is somewhat incoherent. In the chapter 3, conjoined bare NPs are claimed to be allowed in preverbal position, since they are linked to a pro in postverbal position. The author stipulates that conjoined bare NPs cannot be linked to a pro because ''coordination would force a kind of movement that ordinary NPs cannot undergo''. This is not consistent with the assumption that conjuncts do not move. In the chapter 4, the prenominal adjective agreeing only with the first conjunct is said to be able to have scope over both conjunct, but the author claims that the ungrammaticality of full agreement on prenominal adjectives relates to the fact that they cannot have scope over both conjuncts. Finally, the data analyzed with respect to gapping in the chapter 5 do not involve gapping, but V(P) coordination (''John came and left'').
REFERENCES Aoun, J., Benmamoun, E. & Sportiche, D., 1994. ''Agreement, Word Order and Conjunction''. Linguistic Inquiry, 25: 195-220. Aoun, J., & Benmamoun, E., 1999. ''Agreement, Coordination, and Gapping''. Ms. USC and SOAS. Chomsky, N., 1957. Syntactic Structures. The Hague: Mouton. Gazdar et al., 1985. ''Coordination and How to Distinguish Categories''. Natural Language and Linguistic Theory, 3: 117-171. Goodall, G., 1987. Parallel Structures in Syntax. Cambridge: CUP. Haiman, J., 1983. ''On Some Origins of Switch Reference Marking''. In J. Haiman & P. Munro (eds.), Switch- Reference and Universal Grammar, 105-128. Amsterdam: John Benjamins. Hale, K. & Jeanne, L., 1976. Hopi workshop notes. Ms. University of Arizona. Höhle, T., 1990. ''Assumptions about Asymmetric Coordination''. In J. Mascaró & M. Nespor (eds.), GLOW Essays for Henk van Riemsdijk. Dordrecht: Foris. Johannessen, J., 1996. ''Partial Agreement and Coordination''. Linguistic Inquiry, 27: 661-676. Johannessen, J., 1998. Coordination. Oxford: OUP. Longobardi, G., 1994. ''Reference and Proper Names''. Linguistic Inquiry 25 :609-666. Munn, A., 1993. Topics in the Syntax and Semantics of Coordinate Structures. Ph.D. dissertation, University of Maryland. Pullum, G.,& Zwicky, A., 1986. ''Phonological Resolution of Syntactic Features Conflict''. Language 62: 751-773. Zoerner, E., 1995. Coordination: The Syntax of &P. Ph.D. dissertation, University of California.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Hana Skrabalova is a teaching and research assistant at the University of Paris 3. Her field of research (Ph.D. dissertation in progress) includes syntax and semantics of coordinate expressions and related phenomena (agreement, DP structure, plurality). She is mostly working on Slavic (Czech) and Romance (French) languages.