Prosodic Evidence for Syntactic Structure?

Monday, 14 February 2011, Colloquium

Michael Wagner, McGill University.

Abstract

What can prosody tell us about the grammatical structure of a sentence? The answer to the question whether and how reliably prosody reflects grammatical properties such as its syntactic constituent structure will decide to what extend we can use prosodic evidence to inform or even falsify syntactic analyses. This paper looks at the question whether speakers disambiguate different surface constituent structures only when aware of an ambiguity. Previous studies found conflicting results: Snedeker & Trueswell (2003) found clear evidence showing that indeed, disambiguating cues are only reliably present when speakers are aware of the ambiguity; Schafer et al. (2000) and Kraljic & Brennan (2005), on the other hand, found evidence that syntactic ambiguities are prosodically distinguished even when speakers were unaware. Possible explanations considered in the literature are differences between the studies in the length of the stimuli and in the degree to which the task was communicative. This paper pursues an alternative hypothesis: Maybe there are syntactic reasons why some ambiguities are systematically prosodically distinguished and others are not. This idea is supported by several production experiments which replicate the apparently conflicting patterns of earlier studies within a single experiment, depending on the type of syntactic ambiguity involved.