Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Spanish grammar wikipedia , lookup
Musical syntax wikipedia , lookup
Latin syntax wikipedia , lookup
Distributed morphology wikipedia , lookup
Lithuanian grammar wikipedia , lookup
Serbo-Croatian grammar wikipedia , lookup
Yiddish grammar wikipedia , lookup
Cognitive semantics wikipedia , lookup
Pipil grammar wikipedia , lookup
Russian grammar wikipedia , lookup
Proceedings of the Edinburgh Linguistics Department Conference '94 pp. 183{193 States, aspect and complex argument structures Martha G. Robinson [email protected] 1 Introduction The aim of this paper is to bring together two areas of investigation in Syntax and Semantics. On the one hand, the formation of Complex Predicates. This is understood to take place at the level of Argument Structure (henceforth AS) in the form of a merger process (Rosen 1989, Maranz 1984). On the other hand, Event Composition (Verkuyl 1989, Pustejovsky 1991), here interpreted as Semantic Merging. It is investigated how these two areas interact in stative Complex Predicates. Passives (with auxiliaries SER intrinsic to be and ESTAR contingent to be) and perfectives (with auxiliary TENER durative to have) in Spanish are taken to form stative complex predicates. These, however, can be distinguished from an aspectual point of view. It will be argued that these Complex Predicates undergo a Heavy Merging process. In passives it can be dened as an argument structure reduction strategy. In contrast, Heavy Merging in perfective Complex Predicates can be dened as an argument preservation strategy. In this paper, various denitions for states found in the literature, will be investigated. A fundamental claim in this paper, is the distinction between Bounded and Unbounded states. Such a position is not a novel one in the literature. In work like Carlson (1981), Bach (1986), Pustejovsky (1988) the distinction between dynamic and static states is assumed. However, the notion of dynamism will not be employed here. Instead, the Individual-Level (IL) vs. Stage-Level (SL) (Carlson 1977) distinction is applied to the verbal domain. It is claimed that this distinction is related to boundedness and unboundedness. Bounded and Unbounded states are treated as primitives for Event Composition. It will be argued that the aspectual dierences in these stative Complex Predicates can be derived from Event Composition. In order to conclude, in section 2, a discussion on the relevant data will be provided together with a brief discussion on the merger process. In section 3, the nature of states is investigated. This will lead us to nding the relevant primitives to enable the process of Event Composition which will be discussed in section 4. Finally, some consequences of this process will be discussed. 2 Syntax of Passives and Perfectives as a Merger Process In this section the syntax of Spanish perfectives and passives is discussed. It is argued that these structures can be derived in a unied manner at the level of argument structure (AS). Spanish passives and perfectives are formed periphrastically through the combination of auxiliaries and participles (with object agreement). The ASs of the lexical items involved (i.e., auxiliaries and participles) merge together to form one complex AS. This process is otherwise known as Complex Predicate formation. 2.1 Participles: Passive or Perfective? It is a common assumption in the generative literature to draw the passive/perfect distinction from dierent participial entries in the lexicon (references), namely, the passive and perfect 183 184 Martha G. Robinson participles. In contrast, here a unied account is assumed. It is assumed that participles are encoded in the F-lexicon as a single unit neither specifying perfectivity nor passive. It is claimed instead, that participles (according to the distinction drawn in the introduction between bounded and unbounded states) denote states of the bounded type. Spanish stative complexes involve the auxiliaries SER (intrinsic to be) ESTAR (contingent to be) TENER (durative to have) and HABER (perfective to have). SER, ESTAR and TENER combine with participles with object agreement. In contrast, HABER combines with a participial form lacking (overt) agreement. On the one hand, SER and ESTAR complexes form passive structures which are distinguished aspectually. Observe the contrast in (1) and (2). 1 2 3 (1) El camino fue bloqueado (por el ejercito). The-masc.sg path be-past.3.sg block-part.masc.sg by the-masc.sg army. The path was blocked by the army. (2) El camino esta bloqueado (por el ejercito). The-masc.sg path be-past.3.sg block-part.masc.sg by the-masc.sg army. The path is being blocked by the army. Traditionally (1) is considered to be the process passive, since the reference is to the event of the path. being blocked. In contrast, (2) is considered to be the perfective passive. The reference is to the resultant state of the path having been blocked. This aspectual dierence is brought about by the alternation of the auxiliaries SER and ESTAR. Further discussion of this aspectual alternation will be provided later. On the other hand, TENER forms what is considered to be an idiosyncratic perfective with durative connotations. Consider (3) and (4). (3) Tiene perdidos muchos kilos. Have-3sg loose-part.masc.pl many-masc.pl kilos. He/She has (already) lost many kilos. (4) Tengo vistas muchas de sus pelculas. Have-1sg see-part.fem.pl many-fem.pl of his/her- lms-fem.sg. I have (already) seen many of his lms. In both (3) and (4) the reference is to the continuous performance of the action of loosing weight and the viewing of lms. As mentioned above, these auxiliaries combine with participles with object agreement. This type of agreement is assumed to be the marker of syntactic selectional restrictions, since it is a necessary condition for these participles to be transitive. 4 1 Cann (1994) assumes a version of binary grammatical lexicon. Functional and cognitive elements are stored in two separate lexica which do not interact with one another, i.e., the Functional Lexicon and the Cognitive Lexicon. 2 From a semantic point of view if Event Composition is applied (in the form of adding the aspectual values of the lexical items involved), the complexes formed with participles can be subdivided into accomplishment (auxiliaries IR to go and LLEVAR to carry) and stative ( auxiliaries SER, ESTAR, TENER, HABER). 3 Complex predicates involving HABER will only be commented upon briey since this is the auxiliary used in compound (perfective) tense formation and here it is assumed that this auxiliary is involved in a dierent type of complex predicate discussed briey later on in this paper. 4 Rather than being a passive marker (Kayne 1987). 185 States, aspect and complex argument structures Finally, another stative complex is the one involving the auxiliary HABER (perfective to have) used in compound (past) tense formation in Spanish. The participle it combines with does not display any agreement . In Robinson (1993) it is assumed that this is an instance of grammaticalized agreement induced by lack of selectional restrictions. 2.2 Argument Structure Merger In the literature, the structures generally categorised as Complex Predicates have been Causatives, Restructuring verb constructions (Rosen 1989) and Light verb constructions (Grimshaw and Mester 1988). In this paper, the view taken into account concerning Complex Predicate formation, is the one put forth in Rosen (1989). There, Complex Predicate formation, is dened as taking place at the level of argument structure in the form of a merger process. Rosen denes the AS merger as the process where two verbs provide two ASs and the merger merges them together to form one. Every verb provides an <e> (event) role and this role will be of prime importance in the merger process, since the crucial operation in the merger is the identication of <e> roles. As mentioned above, in this paper we assume that Spanish periphrastic passives and perfectives are also formed by a merger process taking place at the level of AS. However, instead of assuming dierentiated lexical entries for both auxiliaries and participles (as is the general tradition), we assume dierentiated merging processes. In other words, through dierent merger processes we create dierent Complex Predicates. These merging processes are to be interpreted as language specic. It is under this assumption that it is possible to preclude a unied account of periphrastic passives and perfectives in Spanish. The view is taken that Complex Predicates can be subsumed under two categories: Heavy Merging and Light Merging. On the one hand, passives and perfectives with object agreeing participles undergo Heavy Merger. On the other hand, compound tense formation can be derived as Light merger. The type of participial form (with or without object agreement) is understood to be a determinant factor for classifying the type of Complex Predicate these auxiliaries are involved in. This is related to the amount of participants that are allowed to percolate onto the complex AS. In other words, the lack or presence of selectional restrictions (related to object agreement) is the crucial factor for determining the complex AS. Heavy Merger occurs in the case where selectional restrictions apply. The participial form in this type of dependency displays object agreement. It is this agreement that blocks the percolation of participants of the participial complement onto the complex AS. Passives and perfectives undergo a Heavy Merger process. In passives this can be dened as an Argument Structure reduction strategy and in perfectives as an AS preservation strategy. Following, is a discussion on how this merger operates. In the table below are the representations of the ASs of the lexical items involved in the merger. A distinction between external and internal (direct and indirect) arguments is assumed (Williams 1980). In this sense, SER and ESTAR license one internal argument. In contrast, 5 6 7 8 5 This is based on the davidsonian event argument Davidson (1967), Higginbotham (1985), Kratzer (1988). In this paper, however, this <e> role is not assumed. For further discussion on this matter, the reader is referred to Robinson (forthcoming) 6 This terminology has been borrowed from Grimshaw & Mester (1988) and Rosen (1989). Here this terminology is applied to dierentiated merger processes rather than to dierentiate lexical items (the assumption in these papers). 7 This could also be interpreted as a detransivizing operation (Dowty 1994). 8 It is assumed here that SER and ESTAR are unaccusatives (Burzio 1986). 186 Martha G. Robinson TENER licenses two arguments, an external and an internal one. The transitive participle, also licenses two arguments (one internal, one external). This is a necessary condition, hence, the presence of the object agreement marking this restriction. SER ((x)) ESTAR TENER (y(x)) Participle (Object agreement) (y(x)) Remember how the merger process was dened earlier. The merger process takes two ASs of two independent lexical items and combines them together to form one. The result, in a sense, is that arguments of both lexical items are licensed in the complex AS. Heavy Merging complexes can be divided into intransitive and transitive complexes. a) Intransitive Merger: The auxiliaries involved are the intransitives SER and ESTAR ((x)) (y(x)) > ((x)) Merging succeeds in a process of identication, where the sole internal argument of the auxiliary is licensed through the direct internal argument of the participle. The result is an intransitive complex argument structure. It is in this sense, that Heavy intransitive Merger is dened as an argument reduction strategy. This characterises an identication structure (the agreement of the participle and the auxiliary coincide). The remaining external argument licensed in the grid of the participle is licensed in the form of an (optional) adjunct-argument, in the sense of Grimshaw (1990). b) Transitive Merger: The auxiliary involved is the transitive TENER (y(x)) (y(x)) > (y(x)) In similar fashion, the internal argument of the transitive auxiliary merges with the internal argument structure of the participle. In contrast, the external argument of the auxiliary is licensed through the external argument of the participle. The result is a transitive (asymmetric) structure where the agreement of the subject and the agreement of the object belong, in a sense, to independent domains. Light Merger, in contrast is dened where no syntactic selectional restrictions apply. The participial complement allowed in this type of dependency can be of any type (i.e., transitive, States, aspect and complex argument structures 187 intransitive, etc.). This is marked in the complex through lack of over agreement on the participle (we assumed earlier that this was a case of grammaticalized agreement). As a consequence, any argument licensed through the AS of the participle is allowed to percolate onto the complex AS. The stative auxiliary undergoing Light Merger is HABER. However, this type of merger will not be discussed here further, since it is not the object of this paper. In order to sum up, it was argued that the syntax of passives and perfectives in Spanish can be derived at the level of AS. It has also been argued in this section that the mechanism for Complex Predicate formation is merger of ASs. The existence of two types of merger (Heavy and Light Merger) has been noted. Passives and perfectives (participles with object agreement) in Spanish undergo Heavy Merger. In contrast, compound tense formation would be categorised as undergoing Light Merger. Section 3 will explore the nature of states and will investigate the connection between states and the concepts of boundedness and unboundedness. It will also be illustrated how these concepts relate to other related ideas such as genericity and existentiality. 3 States and boundedness The previous section dealt with the syntactic aspects of the formation of Complex Predicates. In this and the following section (section 4) the semantics of Complex Predicate formation will be explored. First, in section 3.1 the nature of states will be discussed, since states are the constitutive parts of the Complex Predicates object of this paper. This will be followed, in section 3.2, by a brief discussion on the types of states. It is a fundamental claim in this paper that states can be subsumed under two categories namely bounded and unbounded states. 9 3.1 The nature of states In this section the denitions for states found in the literature will be taken under scrutiny, since in the literature on aspectual typology, the stative/non-stative distinction has been the subject of much research . Great emphasis has been placed in separating states from the remaining aspectual classes often dened under the generic term of EVENTS. (these include processes, accomplishments and achievements.) . However, the status of states has never been dened clearly. States are commonly dened in the literature, in terms of interval semantics. For instance, in Allen (1984) the world is divided into both static and dynamic aspects (otherwise the stative/ non-stative distinction). Static aspects refer to properties, while dynamic aspects refer to occurrences. He denes stative and dynamic aspects in relation to intervals of time. Thus, in terms of interval semantics, properties are claimed to hold over intervals of time (IT), while occurrences change some form of the situation over intervals of time. A similar characterisation is oered in Parsons (1990) States are characterised as holding over IT in contrast, events are claimed to culminate over IT. At rst sight, it seems dicult to dene states in terms of IT like events. IT are dened as a set of moments of time ordered by the precedence relation Cann (1993). The notion of time progression in the denition of states and events is a central one (although, intuitively, states cannot involve the same time concept than events): States do not imply progression of time: States are in time, but they do not take time (Taylor 1977). In contrast, events seem to imply the progression of time; events take 9 In the literature generally considered unbounded. 188 Martha G. Robinson time. According to Tenny (1993) states can have temporal duration, temporal location, but they lack internal temporal structure. And this lack of internal temporal structure has lead to the idea that states lack event structure (Tenny 1993, van Voorst 1989, inter alia). But what is this internal temporal structure if not location and duration ? Unfortunately, this is still an open question. In connection with this idea, states have often been claimed not to take place, in contrast with events that happen (van Voorst 1989, Smith 1990). Pustejovsky (1988) oers one of the more reasonable characterisations of lacking internal temporal structure: the internal structure of a state is homogeneous, in other words, states consist of a period of undierentiated moments. However, going back to the interval theses, why could we not be able to dene states in terms of MT, the constitutive parts of IT ? It is claimed, that the dierence between events and states can be dened in terms of IT and MT. On the one hand, events can be dened in terms of IT, since they are themselves composed out of smaller units. Events are divisible, i.e., they can be divided into substages of dierent types. They take time (there is a natural progression inherent to time). In contrast, stative properties correlate with those of a MT. If states are dened as indivisible units we can only predicate indivisibility over MT not over IT. Therefore, they are constitutive parts of greater units. States are primitives. Since they are the smallest unit of time, they cannot be a period of undierentiated moments because as soon as MT are iterated we have IT. States lack temporal structure, this is they are in time but no progression is implied. Together with Smith (1990) I assume states to be the simplest of the situation types. Therefore, although they themselves cannot be decomposed into parts, the can be part of other events. In sum, states are interpreted as the primitives par excellence, of the situation types. It was claimed that states due to their indivisibility (i.e., states cannot be decomposed into substages like events) and homogeneity showed properties parallel to those of MT rather than IT. At the same time, it was concluded that states are to be interpreted as the simplest of the situation types. In the following section, it will be contended that although states are interpreted as the simplest of the situation types, they do not always show the same properties. It will be claimed that states can be either bounded or unbounded. 10 3.2 Types of states In the previous section the nature of states has been dened. Hinging on the idea that states can be interpreted as nite and innite MT here it will be argued that states can be subdivided into bounded and unbounded types. This distinction will be a crucial one for the illustration of aspectual contrasts in periphrastic passives and perfectives. Following Smith's (1990) train of thought, we know intuitively that in examples like being tired, being beautiful, being on the table the stative property is not equivalent in all three. Although the cluster of properties (indivisibility and homogeneity) dened earlier hold for these states it is still possible to feel that they dier. It is evident that being tired and being on the table describe the moment where the stative property is more salient. On the other hand, being beautiful although also a state, seems to dene a more permanent property. A similar consideration applies to the distinction between Snow is beautiful and The snow is beautiful. As a result, in this investigation, a distinction between states is assumed. In the 11 10 Pustejovsky (1988, 1991) is assumed here: accomplishments are formed by two substages: a process and a state. 11 In these examples, the semantic contrast (permanent vs. accidental) is brought about through the bare States, aspect and complex argument structures 189 literature this position is not a novel one (although it is not discussed in great depth). Carlson (1981), Pustejovsky (1988) and Bach (1986) distinguish between dynamic and static states. However, it does seem bizarre to dene states in terms of dynamics, since events would be dened in this manner (as seen before in Allen 1984). This notion, therefore, will be dispensed of. Instead, the Individual-level (IL) and Stage-level (SL) distinction (Carlson 1977, Kratzer 1988) is applied to the verbal domain. This distinction refers to predicates expressing more inherent and essential sorts of properties and to predicates expressing highly temporary and accidental properties. This distinction is not used to separate events (commonly assumed to be SL) from states (commonly assumed to be IL). It is claimed that this distinction applies at a more universal level. It is related to boundedness and unboundedness. Verbal States, hence, are not only IL, they can be subsumed under both categories (this being a lexical property), i.e., IL and SL. In other words, states can either be permanent (unbounded) or accidental (bounded). This lexical property might not be an obvious one in English, but it is easily recognised in Spanish. A number of stative verbs in Spanish partition according to this distinction. Consider the examples below: 12 IL (generic) SL (existential) saber (to know) conocer (to know) entender (to understand) comprender (to understand) ver (to see) mirar (to see) 13 Other clear examples of this bounded/unbounded stative contrast are also SER - ESTAR (to be) and TENER - HABER (to have) . On the one hand, SER and TENER would be interpreted as the unbounded pair. This is based on the fact that they have durative connotations. They are used to refer to permanent properties (possession, origin, inherent properties of individuals). On the other hand, ESTAR and HABER would be interpreted as bounded. An interesting fact about this bounded interpretation is that this pair (as opposed to SER and TENER) appears with locative expressions. An extension of this aspect is that ESTAR on the one hand, appears to denote existential properties. HABER in contrast, appears as the perfective (past tense) auxiliary. In this section it was argued that verbal states partition according to boundedness and unboundedness. This was seen to be an extension of the generic / existential interpretation of properties (also related to permanent and accidental properties). In the following section it will be shown how this distinction is relevant to the stative complex predicates under study. 14 4 Event Composition as Semantic Merger This section is devoted to the semantics of Complex Predicates. It was assumed earlier that Complex Predicates undergo Semantic Merging. This process is understood to be equivalent to Event Composition, where the aspectual class of a verb may be modied by other elements in 15 and denite NP alternation. 12 It is important to note that here boundedness does not necessarily imply culmination. In a sense, it relates to temporary and accidental properties. 14 In modern Spanish this property is not easily distinguished, because the ussage of HABER is only as an auxiliary and TENER can be used both as auxiliary and lexical verb. However, in Old Spanish both TENER and HABER alternate between both functions with semantic subtleties: TENER had durative connotations and HABER had terminative and inchoative connotations. 15 In the literature this is also called event-type shifting. 190 Martha G. Robinson the sentence (Mourelatos 1981, Dowty 1979, Verkuyl 1972, 1989). Although, in the literature Event Composition has been applied to derive aspectual sentential properties, in this paper the aspectual properties of a complex predicate are derived in similar fashion (i.e., compositionally) from the interaction of aspectual primitives. It is in this sense, that the aspectual contrasts between stative passives and perfectives can be derived from Event Composition. 4.1 Complex Predicates as Complex Event Representations The previous section concluded by assuming that states can be either bounded or unbounded. Boundedness and unboundedness will constitute the primitives for Event Composition. However, it is assumed that this distinction also correlates with the open and closed intervals theses (Bennett 1981). Open and closed intervals are dened as the contrast between intervals with endpoints and intervals without endpoints. Open and closed intervals are represented graphically by open and closed square brackets (][, []). It was seen earlier that Event Composition refers to how basic event structures interact with other syntactic constituents to form derived event representations. These derived event representations could also be called Complex Event Representations. Complex Event Representations are represented in the form of aspectual templates. Open an closed square brackets constitute the (graphic) primitives standing for boundedness and unboundedness. Complex Event Representations of stative Complex Predicates are of two types: Heterogeneous and Homogeneous. 16 a) Heterogeneous Complexes: They are formed by unbounded SER and TENER combining with a bounded participle (this is represented in the template below by the combination of an open and a closed bracket representing the substages of the complex). 17 ]SER[ ]ser[ ((x)) [part.] ]TENER[ (y(x)) ]tener[ [part.] A consequence of the heterogeneous internal structure is that passive with SER can be modied by in (en in Spanish) adverbials like in (5), the complex with SER and (6), the complex with TENER. 18 (5) 16 Remember El camino fue bloqueado por el ejercito en una hora. The-masc.sg path was block-part-masc.sg by the-masc.sg army in one hour. The path was blocked in one hour. that the last time homogeneity was addressed it was in relation to the indivisible nature of states. However, homogeneous Event Complexes are dened when the substages (primitives) of the complex are of the same type, i.e., represented as [[], []] or [][, ][]. In contrast, heterogeneous complexes would be dened when the substages are of dierent types, i.e., either [][, []] or [[], ][]. 17 Note that in section 2.1 it was assumed that participles denote bounded states. 18 In section 2.1 it is mentioned that SER passive refers to the event and that TENER perfective retains durative connotations, this is reinterpreted reference to the event. 191 States, aspect and complex argument structures (6) Tengo vistas muchas de sus pelculas en dos a~nos. Have-1sg see-part.fem.pl many-fem.pl of his/her lms in two years. I have (already) seen many of his lms in two years. b) Homogeneous Complexes: They are formed by the coocurrence of bounded auxiliaries ESTAR and HABER and the bounded participles (in the template below this is represented graphically by the combination of two closed brackets representing the substages of the complex). 19 [ESTAR] ((x)) [estar] [part.] [HABER] (y(x)) [haber] [part.] A consequence of this homogeneous internal structure is that complexes with ESTAR are naturally modied by for (durante in Spanish) adverbials. (7) 20 El camino esta bloqueado por el ejercito durante una hora. The-masc.sg path is block-part-masc.sg by the-masc.sg army for one hour. The path is blocked by the army for one hour. In this section it was contended that it was possible to derive Complex Event Structures via the application of Event Composition. These complex structures were represented in the form of templates. The result has been that by combining open and closed intervals (related to boundedness and unboundedness) it was possible to derive two types of semantic structures, i.e., homogeneous and heterogeneous. These semantic structures correspond to the aspectual structures of the stative complex predicates. As a consequence, homogeneous complexes are modied by for adverbials while heterogeneous complexes are naturally modied by in adverbials. 5 Conclusion This paper explored the interaction of syntax and semantics in the formation of Complex Predicates in Spanish. It was contended that the syntax of these predicated could be derived at the level of AS in the form of a Merger process. Complex Predicates involving participles can be subsumed under two categories depending on how the Merger operates, i.e., Heavy and Light Merger. It was claimed that this mechanism derived complex argument structures. Simultaneously to the syntactic Merger, it was claimed that semantic merger also was in operation in the formation of Complex Predicates. Semantic Merger results in Complex Event 19 In the case of HABER the homogeneous internal structure would prove to be a speculation for the fact that this auxiliary (as opposed to TENER) is the one used in compound (past) tense formation in Spanish. However, since the category Aspect becomes intertwined with the category of Tense HABER will not be discussed further due to the complexity of the matter and the lack of space. 20 Likewise, in section 2.1 it was claimed that ESTAR passive referred to the resultant state rather than the event. 192 Martha G. Robinson Structures which are represented in the form of Complex Event Representations (templates). These Representations were derived from primitives (boundedness and unboundedness) and subdivided in two categories Homogeneous and Heterogeneous Complexes. References Bach, E. 1986: The Algebra of Events. Linguistics and Philosophy 9 (1). Bennett, M. 1981: Of Tense and Aspect: One Analysis. Syntax and Semantics, Tense and Aspect, vol. 14. Burzio, L. 1986: Italian Syntax. D. Reidel, Dordrecht. Cann, R. 1993: Formal Semantics. An Introduction. Cambridge. Cann, R. 1994: Homonymy in the Functional Lexicon. ms. University of Edinburgh. Carlson, G. 1977: Reference to Kinds in English. PhD Thesis, University of Massachusetts, Amherst. Carlson J. 1981: Aspect and Quantication. Syntax and Semantics, Tense and Aspect, vol. 14. Davidson, D 1967: The Logical Form of Action Sentences. in Nicholas Rescher, ed., The Logic of Decision nad Action, University of Pittsburgh Press. Dowty, D. 1994: Adjunct-to Argument Reanalysis in a Model of Grammar Growth. Paper presented at Centre for Cognitive Science. University of Edinburgh. Dowty, D. 1979: Word Meaning and Montague Grammar. D. Reidel, Dordrecht. Grimshaw, J. 1990: Argument Structure. MIT Press. Grimshaw, J. & Mester, A. 1988: Light Verbs and -Marking. Linguistic Inquiry. 19 Higginbotham, J. 1985: On Semantics. Linguistic Inquiry. 16 Kayne, R. (undated): Thematic and Case-Assigning Properties of Past Participles. m.s., MIT. Kratzer, A. 1988: Stage-Level and Individual-Level Predicates. In E. Bach, A.Kratzer and B. Partee, eds., Papers on Quantication, NSF Report, grant BNS 8719999, University of Massachusetts at Amherst. Maranz, A. 1984: On the Nature of Grammatical Relations. MIT Press, Cambridge. Mourelatos 1981: Events, Processes and States. Syntax and Semantics, Tense and Aspect, vol. 14. Pustejovsky, J. 1988: Event Semantic Structure. ms., Brandeis University. Pustejovsky, J. 1991: The Syntax of Event Structure. Cognition. States, aspect and complex argument structures 193 Robinson, M.G. 1993a: Participles in Complex Predicates. Paper presented at the Para- metric Variation Workshop. Centre for Cognitive Science. University of Edinburgh, 25 May 1993. Rosen, S.T. 1989c: Argument Structure and Complex Predicates. PhD Thesis Brandeis University. Smith, C. : 1990: The Parameter of Aspect. Dordrecht, Kluwer. Taylor, B. 1977: Tense and Continuity. Linguistics and Philosophy, 1. Tenny, C. 1992: The Aspectual Interface Hypothesis. Lexical Matters. Ivan A. Sag and Anna Szabolcsi, eds. Verkuyl, H. 1972: On the Compositional Nature of the Aspects. Dordrecht: Reidel. Verkuyl, H. 1989: Aspectual Classes and Aspectual Composition. Linguistics and Philosophy. Voorst, J. van : 1989: Event Structure. Current Issues in Linguistics. Williams, E. 1980: Predication Linguistic Inquiry 11.