Event Structure: Empirical Data and Linguistic Theory Considerable recent research on the syntax-lexical semantics interface hypothesizes that sentence structure is largely predictable from the lexical properties of a sentence's predicate. Verb meaning has been pushed to explain as much of argument -- and hence sentence -- structure as possible (e.g., Jackendoff 1996; Levin & Rappaport Hovav 1995; Pustejovsky 1991; Tenny 1994; Van Valin & LaPolla 1997). Syntactically-relevant verb meaning is frequently said to take the form of semantic decompositions, or "event structure templates," with linking rules determining the realization of arguments in the syntax. We explore event structure representations empirically with corpus studies and with real-time language processing experiments, and show that data from language production and language comprehension converge to give evidence of the psychological reality of event structure templates. Our investigation focuses on two classes of causal verbs (L&RH 1995; Smith 1970): Internally caused change of state verbs, "bloom" verbs, which express a change of state for which the cause resides in the entity undergoing the change (e.g., flowers bloom because of something internal to flowers); and externally caused change of state verbs, "break" verbs, which express a change of state that comes about because of something external to the entity undergoing the change (e.g., the responsibility for windows breaking lies with some external force, not the windows). Internally caused change of state verbs have the event template (x (BECOME IN STATE)), and externally caused change of state verbs have the more complex template (alpha CAUSE (x (BECOME IN STATE))), where alpha represents the causing event. Our central hypothesis is that sentences are produced and understood through processes that make use of the verbs' underlying lexical semantic event structures. To study language production, we collected large numbers of sentences with internally and externally caused change of state verbs from a 280-million-word corpus of naturally produced English texts (newspaper and magazine articles, adult and juvenile fiction and nonfiction, spoken dialog). Sentences from the corpus followed constraints predictable from the verbs' event structures. Here, we give one example: The event structure for an externally caused verb has two subevents, so the causing subevent and its participants can be relatively separate from the change of state subevent. In contrast, the internally caused template allows for participants in only its one, change of state, event. Thus, when an internally caused verb appears in a transitive sentence (e.g., "the waves eroded the beach"), the entity in subject position is constrained to be an inherent participant in the change of state. But when an externally caused verb appears in a transitive sentence, the entity in subject position is not so constrained because it is a participant in a separate subevent from the change of state. This distinction between verbs of the two classes was confirmed by a statistically significant difference in the distributions of the kinds of entities appearing in corpus sentences as transitive subjects of the verbs. To study language comprehension, we performed experiments in which subjects were asked to read simple sentences such as "the flowers bloomed," "the concrete crumbled," "the flood water rotted the carpet," and "the fire alarm awoke the residents." If verbs' event structures are used in sentence processing, then sentences with externally caused verbs should have significantly longer reading times than sentences with internally caused verbs because the externally caused verbs' event structures are more complex. This prediction was confirmed for both transitive and intransitive sentences. We also tested verbs of the two classes in lexical decision where subjects must decide whether a given string of letters is or is not a word. Since differences in the complexity of verbs' event structures affect sentence comprehension time, it might also be expected that they affect processing at the word level, and they did: response times were significantly slower for the externally than the internally caused change of state verbs. (In all the experiments, verbs and sentences for the two classes were matched on a number of dimensions including length, frequency, and imageability.) In sum, production and comprehension data from language in actual use converge to present evidence for the psychological reality of event structure templates, not only for the internally and externally caused verbs we discuss in this abstract, but also for other verb classes that will be presented (McKoon & Ratcliff 2001). The research embodies the interplay of corpus data that show differences in verbs' event structures reflected in naturally produced sentences, psycholinguistic data that show the real-time processing consequences of event structures, and theoretical lexical semantics. Furthermore, as will be discussed, the data provide a testing ground for psycholinguistic theories of sentence processing, including theories based on modular syntactic processing systems and theories based on statistical regularities in language use. Jackendoff, R.S. (1996). Conceptual semantics and cognitive linguistics. Cognitive Linguistics, 7, 93-129. Levin, B. & Rappaport Hovav, M. (1995). Unaccusativity: At the Syntax-Lexical Semantics Interface, Linguistic Inquiry Monograph 26, MIT Press, Cambridge, MA. McKoon, G. & Ratcliff, R. (2001). Meaning through syntax: Language comprehension and the reduced relative clause. Pustejovsky, J. (1991). The syntax of event structure. Cognition, 41, 47-81. Smith, C.S. (1970). Jespersen's 'Move and Change' class and causative verbs in English. In M.A. Jazayery, E.C. Polome, and W. Winter (Eds.), Linguistic and literary studies in honor of Archibald A. Hill. Vol. 2: Descriptive Linguistics (pp. 101-109). The Hague: Mouton. Tenny, C.L. (1994). Aspectual Roles and the Syntax-Semantics Interface. Dordrecht: Kluwer. Van Valin, R.D. & LaPolla, R.J. (1997). Syntax: Structure, meaning and function. Cambridge: Cambridge University Press.