Extended Argument Structure : Progressive as Unaccusative

Abstract Our lexical relational analysis of the progressive construction is based on Mateu’s (1999) theory of Relational Semantics. Drawing empirical evidence from cross-linguistic studies, we argue that the progressive involves a nominalization process that joins a locative unaccusative structure and the argument structure corresponding to the lexical verb. As a result, the unspecified Ground of the unaccusative structure in which the Figure is «centrally located» turns out to be the entire event coded by the lexical verb. Furthermore, a structurally-based solution of the so-called «Imperfective Paradox» is then pursued. Empirical evidence in favor of our analysis of the progressive is also shown to come from «thematically transparent» languages like Basque, where a different argument structure involves a different case assignment. Key words: progressive aspect, unaccusativity, argument structure, Relational Semantics. Resum. L’estructura argumental ampliada: la construcció progressiva com a inacusativa La nostra anàlisi lèxico-relacional de la construcció progressiva es basa en la teoria de la Semàntica Relacional de Mateu (1999). A partir de dades d’estudis interlingüístics, defensem que la construcció progressiva implica un procés de nominalització que uneix una estructura inacusativa locativa a l’estructura argumental que correspon al verb lèxic. Com a conseqüència d’això, el fons de l’estructura inacusativa en què «se situa centralment» la figura passa a ser tot l’esdeveniment codificat pel verb lèxic. Proposem, a més, una solució estructural a l’anomenada «paradoxa de l’imperfectiu». L’evidència empírica a favor de la nostra anàlisi de la progressiva prové també de llengües que són «temàticament transparents» com el basc, en què una estructura argumental diferent implica una assignació de cas diferent. Paraules clau: aspecte progressiu, inacusativitat, estructura argumental, Semàntica Relacional. *We are grateful to the audience of the IX Colloquium on Generative Grammar for their comments and suggestions. Research for this paper has been supported by the Ministerio de Educación y Cultura through project PB96-1199-C04-02 and the Generalitat de Catalunya through projects 1997 SGR 00033 and 1998 XT 00065. Laia Amadas also acknowledges a FI-FIAP grant from the Generalitat de Catalunya. CatWPL 7, 1999 159-174 Extended Argument Structure: Progressive as Unaccusative* Jaume Mateu Fontanals Universitat Autònoma de Barcelona. Departament de Filologia Catalana 08193 Bellaterra (Barcelona). Spain [email protected].es Laia Amadas Simon Universitat Autònoma de Barcelona. Departament de Filologia Catalana 08193 Bellaterra (Barcelona). Spain [email protected].es Received: December 13th 1998 Accepted: March 17th 1999 CatWPL 7 159-174 13/6/00 12:29 Página 159 1. Introduction: Progressive as Locative In this paper, we will analyze the progressive construction on the basis of Mateu’s (1999) theory of Relational Semantics. Our proposal is that the progressive construction must be regarded as implying an unaccusative structure over that structure assigned to the verb in the lexicon. That is, we will make it clear that the argument structure relations involved in sentences like those in (1) are those associated with the unaccusative structure into which beis integrated, plus those associated with the argument structure corresponding to the lexical verb (breakand laughin (1)). (1)a.John is breaking the window. b.John is laughing. Quite interestingly, our analysis of the progressive construction as involving an unaccusative structure can be argued to be empirically motivated by cross-linguistic studies like Bybee’s et al. (1994), where it is clearly shown that the progressive construction corresponds with or originates as a locative construction in most languages of the world: (2)«The majority of progressive forms in our database derive from expressions involving locative elements (cf. Blansitt 1975; Comrie 1976; Traugott 1978; Heine and Reh 1984). (...) The locative notion may be expressed either in the verbal auxiliary employed or in the use of postpositions or prepositions indicating location —‘at’, ‘in’, or ‘on’. The verbal auxiliary may derive from a specific postural verb, such as ‘sit’, ‘stand’, or ‘lie’, or it may express the notion of being in a location without reference to a specific posture but meaning only ‘be at’, ‘stay’, or, more specifically, ‘live’ or ‘reside’. (...) The form of the main verb is usually nominal (cited as a verbal noun or a gerund), although serial constructions are attested.» Bybee et al.(1994: 129-130) The fact that a locative structure is involved in the progressive construction can be easily shown with examples like those in (3)-(6). A locative structure is apparent not only in the Basque progressive construction (cf. (3)), where the nominalized form of the verb has locative case, but also in Celtic languages (cf. (4)), in Germanic languages like Dutch (cf. (5)), in French (cf. (6)), or in Middle English (cf. (7)). Quite importantly, the spatial preposition/particle represented in bold characters 160 CatWPL 7, 1999 Jaume Mateu Fontanals; Laia Amadas Simon Table of Contents CatWPL 7 159-174 13/6/00 12:29 Página 160 1. Introduction: Progressive as Locative 2. On the Relational Semantics of Argument Structure 3. Extended Argument Structure: the Progressive Construction 4. Case Theory and the Progressive Construction 5. Conclusions References in (3) through (7) will be argued below to provide clear evidence for considering the progressive construction as containing a stative unaccusative structure.1 (3) a. Miren liburua irakur-tze-ndago. (Basque) Miren-ABS book-ABS read-NOM-LOC be-3-SG.ABS ‘Miren is reading the book’. b. Amaia leihoa apur-tze-nari da. Amaia-ABS window-ABS break-NOM-LOC engage be-3-SG.ABS ‘Amaia is breaking the window’. Demirdache & Uribe-Etxebarria (1997: 10-11) (4) a. Mae Rhiannon yn cysgu. (Welsh) is Rhiannon in sleep ‘Rhiannon is sleeping’. b. Eman va breur ovont a-hed an hent. (Breton) is 1SG brother PRT go along the road ‘My brother is going along the road’. c. Tá mé ag tógáil teach úr. (Irish) are I at build house new ‘I am building a new house’. d. Bha an tuathanach ag gearradh na craoibhe. (Scottish Gaelic) be-PAST the farmer-DIR ag cut-VN the tree-GEN ‘The farmer was cutting the tree’. ex. (4a)-(4c) from Borsley & Roberts (1996: 35); ex. (4d) from Ramchand (1997: 31) (5) a. Ik ben aan het/’t werken. (Dutch) I am on the working ‘I am working’. b. Ik ben een boek aan het/’t lezen. I am a book on the reading ‘I am reading a book’. c. Ik was die film aan het/’t bekijken. I was that movie on the regarding ‘I was regarding that movie’. Extended Argument Structure: Progressive as Unaccusative CatWPL 7, 1999 161 1. According to Bybee et al. (1994: 132), «the conclusion concerning stative sources for progressives, then, strongly points to location as a necessary semantic element, and no clear cases of progressives formed with a copula without a locative element have been found in our data.» CatWPL 7 159-174 13/6/00 12:29 Página 161 d. Ik ben het huis aan het bouwen. I am the house at the build ‘I am building the house’. ex. (5a)-(5c) from van Gelderen (1993: 180-182); (5d) from Demirdache & Uribe-Etxebarria (1998: 25) (6) a. Zazie est en train de miauler. (French) Zazie is in along of miaowing ‘Zazie is miaowing’. b. Zazie est en train de jouer. Zazie is in along of play ‘Zazie is playing’. Demirdache & Uribe-Etxebarria (1997: 9; 1998: 25) (7) a. He is on hunting. (Middle English) b. He was a-coming home. Jespersen (1949: 168), apud Bybee et al (1994: 132) On the other hand, it is interesting to point out that the auxiliary selected in the Spanish progressive construction is estar, which diachronically derives from a Latin locative verb stare (‘stand’ or ‘stay’): cf. (8). The auxiliary estar is typically used with either locative or stage-level predicates (cf. (9)). Once again note that this provides empirical evidence for the analysis of the progressive as involving a locative structure. (8) a. Juan está estudiando. (Spanish) Juan is studying ‘Juan is studying’. b. Estoy escribiendo un libro. am writing a book ‘I am writing a book’. (9) a. Juan está en la habitación. (Spanish) Juan is in the room b. Juan está cansado. John is tired This also seems the appropriate place to comment on an interesting observation found in Bybee et al.(1994: 133), which is originally due to Dwight Bolinger. It is the case that the original function of the progressive periphrasis is to give the 162 CatWPL 7, 1999 Jaume Mateu Fontanals; Laia Amadas Simon CatWPL 7 159-174 13/6/00 12:29 Página 162 location of an agent as in the middle of an activity. For example, note that this function can be shown clearly in the following English dialogue in (10): (10) A: Where’s Lou? B: He’s taking a bath (having a nap, etc.) Bybee et al. (1994: 133) Let us summarize the main points that we will be dealing with here. Our analysis of the progressive as unaccusative will be shown to be empirically motivated on the basis of two facts: on the one hand, we will claim that there is a nominalization process joining the unaccusative structure into which be is integrated, and that structure corresponding to the lexical verb. On the other hand, the different argument structures assigned to two sentences like John was breaking the window and John broke the window will be seen to be on a par with their different Case properties in «thematically transparent» languages like Basque. Before analyzing the progressive construction, it will be useful to sketch out the model of argument structure which our analysis of the progressive will be argued to depend on (cf. Mateu (1999)). 2. On the Relational Semantics of Argument Structure The basic tenet of our theory is that there is a syntactically relevant semantic structure, which can be represented in a tree structure (cf. Pinker (1989), Bouchard (1995) or Juffs (1996) for the same proposal). In particular, we claim that there is an inherent semantics to the formal relations one can establish among the elements of tree structures such as those in (11) (i.e., the R(elational) S(emantic) S(tructure) of transitive structures), (12) (i.e., the RSS of unergative structures), or (13) (i.e., the RSS of unaccusative structures). For example, let us consider the more complex RSS, i.e., that in (11): there are two complete semantic relations (Rand r). Given that each complete relational element must always involve two related elements, each of them will project up to a third level by virtue of binary branching: the related elements are situated in a different position with respect to the relational element, one occupying the specifier position and the other one occupying the complement position. rcan be viewed as a spatial relation in the sense that it purely relates two non-relational elements into our cognitive space: say, «Figure» (specifier of r) and «Ground» (complement of r), to use Talmy’s (1985) terminology. In (11) there is only one «relation» that has no specifier but only a complement. This «semi-relation» is T. T can be regarded as a transition towards the spatial relation r. R can be considered to be a source relation in the sense that its complement can come into existence by virtue of being immediately related with the superior relational node R, whose specifier can be interpreted as the «Originator» (in accordance with van Voorst’s (1988) or Borer’s (1994) terminology). Extended Argument Structure: Progressive as Unaccusative CatWPL 7, 1999 163 CatWPL 7 159-174 13/6/00 12:29 Página 163 We want to argue that the formal relations one can find in structures such as those in (11), (12), and (13) turn out to be meaningful if they represent basic relations of our cognitive space: basically, source relations (R), transitional relations (T), and spatial relations (r). Given our present purposes, we will assume that R, T, and rcan be argued to take two different values according to their corresponding conceptual semantics (cf. (14)). This notwithstanding, it should be kept in mind that these binary values are irrelevant to the syntactic projection of arguments. For example, in spite of their different conceptual content corresponding to their RSS elements, two sentences such as John is in the store and John went to the store are identical with respect to the syntactic projection of their arguments. 164 CatWPL 7, 1999 Jaume Mateu Fontanals; Laia Amadas Simon (11) The RSS of a transitive structure = unergative structure + unaccusative structure R X R R T T r X r r X (13) The RSS of unaccusative structures T T r X r r X (12) The RSS of unergative structures R X R R X CatWPL 7 159-174 13/6/00 12:29 Página 164 Indeed, what is relevant at the syntax-semantics interface is that both sentences involve the same RSS, that in (13). (14) Relational R: (+) CAUSE/DO / / (-) HAVE elements:T:(+) BECOME/GO / / (-) BE r: (+) DIRECTIONAL RELATION (r1) (-) LOCATIVE RELATION (r2) Non-relational element: X Assuming the validity of our approach to RSS’s, note that Hale’s (1995: 6) statement in (15) must be reconsidered in the sense that the combinations in (16) cannot be argued to be only relevant to a purely (lexico-)syntactic domain. According to Hale, lexical heads can enter into four possible combinations: that is, in (16a) x only takes a complement; in (16b) x takes a specifier and a complement; in (16c) x only takes a specifier, «alfa» being a host category that provides the necessary configuration for x to have a specifier; finally, in (16d) xtakes no specifier and no complement either. The prototypical morphosyntactic realizations in English of the lexical syntactic heads in (16) (i.e., x’s) are the following: Vin (16a), Pin (16b), Ain (16c), and Nin (16d). (15) «For the lexical projections, just the relations head-complement and subjectpredicate are relevant —this is part and parcel of the very notion of “lexical head”» Hale (1995: 6) (16) Head (x); complement (yof x), predicate (xof z) Hale (1995: 1) Quite interestingly, the combinations in (16) can also be argued to be relevant to our meaningful RSS’s. As a result, note that the relations head-complement and specifier-head are always dependent on the semantics of the relational element. Actually, note that we are trying to give explanatory power to the well-known descriptive claim that «s(emantic)-selection» has epistemological priority over «c(ategorial)-selection». Unlike Hale (1995), we postulate that «A(djective)» is not a primitive element of grammatical theory (cf. the xin (16c)). Our claim is that «A» can be decomposed into two elements: a morphologically unspecified relational syntactic element corresponding to r, which is represented by the @ symbol, plus a non-relational Extended Argument Structure: Progressive as Unaccusative CatWPL 7, 1999 165 (a) x (b) x (c) α(d) x x y z x z α x y αx CatWPL 7 159-174 13/6/00 12:29 Página 165 syntactic element, the latter being incorporated into the former. Accordingly, two sentences like those in (17a,b) are assigned the same RSS, that in (17d): both sentences are stative unaccusative structures. Note that the same analysis is valid for adverbs in predicative contexts (cf. (17c)): The «prototypical» correspondences that can be drawn between morphosyntactic elements and RS elements are the following: Ncorresponds with X, Vcorresponds with {R/T}, Pcorresponds with r, and Adj and Adv correspond with r+X (X being incorporated into r). In non-predicative contexts, Adjs modify non-relational elements, whereas Advs modify relational elements. Quite interestingly, a principled explanation of the meaningfulness of argument structure relations turns out to be valid in accounting for both the very limited number of lexical syntactic categories and the very limited number of the (syntactically relevant (Baker (1997)) «theta-roles». We will not enter here into discussing Hale & Keyser’s (1993) syntactocentric explanation of both of these facts. With this very sketchy review of the Relational Semantics of argument structure in mind, let us now comment on in some detail our RS analysis of the progressive construction. 3. Extended Argument Structure: the Progressive Construction Consider the example in (1a), repeated in (18). Our claim is that two RSS’s are always involved in the progressive construction (cf. (19)): the unaccusative RSS depicted in (19a), and the transitive RSS depicted in (19b). As noted before, we want to argue that any progressive construction involves an unaccusative structure plus that structure corresponding to the verb lexically chosen. Concerning the unaccusative RSS involved in any progressive construction, it is the case that the complement of the spatial relation (r) always corresponds to an unspecified «Ground»: accordingly, the interpretation associated with (19a) is «John is centrally located in some unspecified ground». On the other hand, the interpretation associated with the transitive RSS corresponding to break is «John caused the window to go into break», that is, «John caused the window to become broken». 166 CatWPL 7, 1999 Jaume Mateu Fontanals; Laia Amadas Simon (17) a. The cat is on the table. b. The cat is happy. c. The cat is here. d. T0V0 T0r1V0{P1/@1} Xar1corresponds with Na {P1/@1} r1Xb{P1/@1} Nb CatWPL 7 159-174 13/6/00 12:29 Página 166 (18) John is breaking the window. Basing our analysis on the data in (3) through (7), we claim that the progressive construction involves the integration of two RSS’s (e.g., those in (19)) into one lexical syntactic structure (e.g., that in (20)) by means of a nominalization process. Our proposal is that the noun that corresponds to the unspecified Ground of the unaccusative structure (that is, Nb) turns out to c-select the verbal structure associated with the argument structure corresponding to the verb lexically chosen. Following Grimshaw (1990) and Picallo (1991), we assume that, as a result of such a nominalization process, the external argument corresponding to the specifier of Ris not projected into the lexical syntactic structure in (20). We also assume that the transitive lexical verb in (20) enters the syntax as a morphological atom (but cf. Hale & Keyser (1993)): a lexical operation has fused the abstract ground, the complex spatial relation, the transitional relation, and the causal relation into the transitive verb. Without entering into discussing the advantages of a syntactic approach to incorporation, we also assume that the transitive verb incorporates into the noun heading the nominalization (i.e., Nb), the resulting complex being then incorporated into the relational element P2/@2. Extended Argument Structure: Progressive as Unaccusative CatWPL 7, 1999 167 (19) a. b. R3 XaR3 JOHN T1R3T4 CAUSE T1r2T4r5 BE GO Xar2Xcr5 JOHN WINDOW r2Xbr5Xd IN GROUND INTOBREAK (20) V1 V1P2/@2 NaP2/@2 P2/@2Nb NbV3-4-5-d V3-4-5-d Nc CatWPL 7 159-174 13/6/00 12:29 Página 167 Chomsky, N. (1993). «A Minimalist Program for Linguistic theory». In Hale, K.; Keyser, S. J. (eds.). The View from Building 20. Cambridge, MA: The MIT Press. Demirdache, H.; Uribe-Etxebarria, M. (1997). «The Primitives of Temporal Relations». University of British Columbia / University of California at Irvine: Unpublished manuscript. —(1998). «The Primitives of Temporal Relations». University of British Columbia / University of California at Irvine: Manuscript. Dowty, D. (1979). Word Meaning and Montague Grammar. Boston: Reidel. Gelderen, E. van (1993). The Rise of Functional Projections. Amsterdam/Philadelphia: John Benjamins. Grimshaw, J. (1990). Argument Structure. Cambridge, MA: The MIT Press. Hale, K. (1985). «Notes on World View and Semantic Categories: Some Warlpiri Examples». In Muysken, P.; van Riemsdijk, H. (eds.). Features and Projections. Dordrecht: Foris. —(1995). «Universal Grammar and the Necessity of Linguistic Diversity». MIT: Unpublished manuscript. Hale, K.; Keyser, S. J. (1993). «On Argument Structure and the Lexical Expression of Syntactic Relations». In Hale, K.; Keyser, S. J. (eds.). The View from Building 20. Cambridge, MA: MIT Press. Juffs, A. (1996). Learnability and the Lexicon. Amsterdam/Philadelphia: John Benjamins. Klein, W. (1995). «A Time Relational Analysis of Russian Aspect». Language, 71.4: 669-695. Laka, I. (1993). «Unergative that Assign Ergative, Unaccusatives that Assign Accusative». MITWPiL 18: Papers on Case and Agreement I, Cambridge, MA: 149-172. Landman, F. (1992). «The Progressive». Natural Language Semantics, 1: 1-32. Mateu, J. (1999). «Universals of Semantic Construal for Lexical Syntactic Relations». Paper presented at the GLOW Workshop on Sources of Universals, University of Potsdam. Distributed as a research report (Report GGT-99-4), Grup de Gramàtica Teòrica, Universitat Autònoma de Barcelona. Parsons, T. (1989). «The Progressive in English: Events, States and Processes». Linguistics and Philosophy, 12: 213-241. Picallo, C. (1991). «Nominals and Nominalization in Catalan». Probus, 3.3: 279-316. Pinker, S. (1989) Learnability and Cognition: The Acquisition of Argument Structure. Cambridge, MA: The MIT Press. Ramchand, G. C. (1997). Aspect and Predication: The Semantics of Argument Structure. Oxford: Clarendon Press. Smith, C. S. (1991). The Parameter of Aspect. Dordrecht: Kluwer. Talmy, L. (1985). «Lexicalization Patterns: Semantic Structures in Lexical Forms». In Shopen, T. (ed.). Language Typology and Syntactic Description III: Grammatical Categories and the Lexicon. Cambridge, MA: CUP. Voorst, J. G. van (1988). Event Structure. Amsterdam/Philadelphia: John Benjamins. 174 CatWPL 7, 1999 Jaume Mateu Fontanals; Laia Amadas Simon CatWPL 7 159-174 13/6/00 12:29 Página 174