| Early language research has revealed important insights into the building blocks of language, such as morphosyntactic features and rules and truth-conditions of sentences. Once we situate language in real-life use, however, a wide range of factors come into play. Language interacts not only with the surrounding linguistic context but also with the situational context, our mental representation of content, and our background knowledge. The discourse-level interaction among linguistic and extralinguistic factors is relevant to both sides of the communication -- the speaker, in choosing and organizing linguistic expressions, and the listener, in selecting among different possible structures and meanings for the linguistic input.;The question I address in this dissertation is 'how we keep track of time when we use language.' My specific interests are (1) whether story time in narrative discourse is one of the critical dimensions that are dynamically updated as discourse progresses, and (2) how fine-grained our time representation is for discourse -- whether it is simply an ordering of temporal points and intervals for the events and states described in the discourse, or a timeline where duration is preserved in greater detail.;In order to elaborate on these issues, I discuss results from my narrative production experiment and my narrative comprehension experiment. In the production study, based on wordless picture books, two kinds of linguistic expressions were found much more frequently after longer intervals in story time compared to shorter intervals: (1) explicit temporal marking with lexical or phrasal markers of topic time (e.g., when, the next morning , etc.); and (2) proper names in referring back to previously mentioned characters. In the comprehension study, based on short "two-minute mysteries," longer duration in temporal adverbials in the stories tended to lead to longer reading times.;I conclude that magnitudes such as duration in story content are preserved in our linguistic encoding and have observable impact on our linguistic decoding, and extend the situation-model framework of discourse comprehension (van Dijk & Kintsch, 1983; Zwaan, 1999) to discourse production. My findings thus support an account of communication as alignment of situation models (Pickering & Garrod, 2006). |
