Events versus States: Empirical Correlates of Lexical Classes

Abstract

Events Versus States: Empirical Correlates of Lexical Classes Silvia Gennari (sgen@wam.umd.edu) Cognitive Neuroscience of Language Laboratory 1401 Marie Mount Hall. University of Maryland, College Park, MD 20742 David Poeppel (dpoeppel@deans.umd.edu) Cognitive Neuroscience of Language Laboratory 1401 Marie Mount Hall. University of Maryland, College Park, MD 20742 Abstract Philosophers and linguists have claimed that verb meanings are divided into semantic types or superordinate categories that differ in internal conceptual structure. In particular, eventive verbs, which have internal causal structure are distinguished from stative verbs, which have no internal causal structure. In this paper, we explore the processing consequences of assuming that the lexical representations of verb meanings differ in the complexity of their internal representations. We conducted two experiments, a lexical decision task and a self-paced reading study, that investigated how verb types of different complexity are processed. We predicted that the conceptually more complex eventive verbs would take longer to process than stative verbs. In both experiments, this prediction was confirmed. This lends support to theories of verb concepts that propose classifications based on internal representations and shows that there are discrete and abstract conceptual categories in the domain of events. Introduction An important question in cognitive science concerns how word meanings (or lexical concepts) are internally represented. Although considerable progress has been made in the domain of nominal concepts since Rosch’s studies, the nature and organization of verb concepts is less well understood. Early studies on verb meanings investigated whether verbs had internal semantic structure, as proposed in linguistic theories, but failed to find evidence supporting such a view (e.g., Fodor, Garrett, Walker, & Parkes 1980, Kintsch, 1974, Rayner & Duffy, 1986). For example, Rayner and Duffy (1986) measured the eye-fixation time on verbs during reading that were assumed to differ in internal complexity. They found no reading time differences corresponding to the semantic complexity of the verbs. This sort of finding, together with Fodor and colleagues’ theoretical arguments (Fodor, 1975, Fodor, Fodor, & Garrett, 1975, Fodor & Lepore, 1998), was taken to support the view that verb meanings are atomic and lack internal structure. However, recent psycholinguistic studies challenge this view. Several sentence processing experiments have shown that lexical semantic properties such as selectional restrictions and verb-specific thematic roles (agent vs. patient) are quickly accessed by the processor when parsing syntactic ambiguities (e.g., Trueswell, Tanenhaus & Kello, 1993, Trueswell, Tanenhaus & Garnsey, 1994). More relevant to verb concepts per se, McRae, Ferretti & Amyote (1997) have shown that thematic roles have internal conceptual structure (as object categories do) and that their feature structure is quickly accessed by the parser when resolving syntactic ambiguities. Moreover, Ferretti, McRae & Hatherell (2001) have shown that verbs prime their typical agents, patients and instruments (e.g., praying primes nun). They argue that verbs activate event schemas or generalized situation based knowledge that facilitate accessing the meaning of their typical participants. Finally, McKoon & Macfarland (2000) have found processing correlates of two types of verb meanings, those that are conceptualized as either externally caused events (e.g., break) or internally caused ones (e.g., grow). These verb types are assumed to differ in internal lexical complexity, particularly in their causal components (see also Gentner 1981). Taken together, these findings suggest that there is some internal structure in verb meanings: thematic structure and event types. The work presented here further investigates verb concepts, i.e., how verbs, which refer to events, are processed and internally represented. In particular, we ask whether there are verb-general concepts and structures beyond and above the idiosyncratic meanings of individual verbs. We follow numerous linguistic and philosophical studies (as in McKoon & Macfarland, 2000) in assuming common structural and causal properties across classes of verbs that define superordinate concepts. Thus, beyond the existence of typical agent-verb-relations (that between nuns and praying), there may be more abstract structural or conceptual properties that organize our knowledge of events stored in the lexicon. The classification of verbs and their semantic properties has been the topic of numerous philosophical and linguistic studies (Vendler, 1967). Following traditional Aristotelian classes, these studies have argued that there is a typology of events underlying verb uses. Verb types appear to be universal (Smith, 1991) and are supposed to reflect the way speakers conceptualize the domain of events, i.e., the semantic/conceptual properties they assign to a particular actual occurrence. One general distinction typically made between verb meanings is, among others, that between states and events (Vendler 1967, Dowty, 1979, Taylor 1977, Bach, 1986, Verkuyl, 1993, Jackendoff, 1990, Rappaport-Hovav & Levin 1998). The distinction seems cognitively basic because it is grounded in causal properties: eventive verbs typically denote a cause and a change from an initial state to a resulting one (e.g. write, destroy), while stative verbs simply denote properties or stable relations between participants (e.g. love, belong, contain) (Dowty, 1979, Parsons 1990). The