What We Do

Spoken Syntax Lab Tools

The Spoken Syntax Lab provides resources for collaborative work on syntax using multiple sources of evidence and modern statistical models. The Lab is developing repositories of aligned phonetic, parsed, and contextualized data as well as advanced search and analysis tools. (Gabriel Recchia)

Dynamics of Higher-Level Grammatical Choices

The dynamics of language can be better understood through quantitative modeling of higher-level grammatical choices in spontaneous language use. To study the dynamics of sentence production, we are using multilevel modeling techniques from computational statistics combined with methods from psycholinguistics, corpus linguistics, and natural language processing to examine two very different kinds of higher-level grammatical choices. One involves the use of alternative word order structures conditioned on semantic classes of verbs, such as He brought the pony to my children, He brought my children the pony, while the other involves purely `functional' choices such as expressing or omitting clause-marking words that have no obvious semantic content: We know (that) it was raining, The story (that) I heard was different. (Bresnan, Wasow)

Comparative Statistical Models of Varieties of English over Time and Space

We are investigating whether different dialects or varieties of language vary in the probabilities of higher-level grammatical choices over space and time. One project involves phonetico-syntactic investigations of New Zealand and US English dative constructions in spontaneous speech over various time-depths (Hay, Bresnan). Another project involves comparative experimental and corpus studies of Australian and US English dative syntax and semantic verb classes (Ford, Bresnan). A third project studies the origins of African American Vernacular English (AAVE) by means of quantitative models of selected syntactic constructions in Caribbean and non-standard English varieties in the US (Rickford, Wasow). A fourth project studies Colloquial Singapore English (CSE), aiming to refine the available Lab tools to streamline the annotation of colloquial varieties of English such as CSE and AAVE, for subsequent research in (comparative) syntax and semantics (Fong).

The Semantics of Construction Choice

To study the semantics of construction choice, one project models the co-occurrence patterns of different sub-classes of Source/Goal prepositions with different verbs in Optimality Theory with the aim of understanding the asymmetry in Source vs Goal expressions (Fong). Another project is testing an Optimality Theoretic model of variation and ambiguity in English possessives on both written (Boston University Noun Phrase Corpus) and spoken (Switchboard) corpora (Anttila, Fong).