The Meaning of Verbal Roots Across Languages

University Of Texas At Austin, Austin TX

Investigators

Abstract

Language serves not just as a means of communication but also a window into how humans perceive the world. Verbs have been a rich source of information about how humans categorize events, and one tantalizing result of work on verb meaning is that the possible meanings a verb can have are not arbitrary. Rather, there are recurring meaning components of high generality that define classes of verbs according to the broad situation type being described, including cause and effect, bodily action, and possession, and verbs within a class tend to have shared grammatical behaviors. Individual verbs within a class are distinguished by idiosyncratic details that fill in the specifics of the broader situation, e.g. the name of a specific action, usually with few grammatical ramifications. However, how general and idiosyncratic components combine in the meaning of a verb is still not well understood, nor is the space of what idiosyncratic meanings an individual verb within a class can have, including how distinct general and verb-specific meanings even are. Understanding these issues has consequences for our understanding of human languages, and can serve practical purposes, particularly in relation to search engine technology and machine translation. This project addresses these lacunae by examining the idiosyncratic meanings of verbs describing change-of-state (e.g. break, redden) and caused possession (e.g. give, send), which have been claimed to exemplify possible limits on idiosyncratic meanings. The methodology involves (i) in depth studies of these verbs in English and Kinyarwanda (chosen for its rich verbal morphology), the latter through field work, and (ii) a broad survey of change-of-state verbs in a genealogically balanced 100 language sample drawn from existing published reference materials. The goal is to isolate general and idiosyncratic meaning in verbs through inferential and morphological patterns. The results will be integrated into a theory of how verb meanings are composed using modern techniques in compositional semantics. The expectation based on pilot data is that while idiosyncratic meanings are not limited in terms of their literal content, they fall into limited types regarding their relation to the rest of the verb's meaning, making some predictions about possible and impossible verbs.