Katie Franich
Department of Linguistics and Cognitive Science
University of Delaware

Examining the relationship between speech timing, grammar, and predictive processes across languages

Languages differ in fundamental ways with regard to their phonetic-prosodic profiles: while some, such as English and other Indo-European languages, show acoustic evidence for rhythmic prominence asymmetries (e.g. stress or accent), other languages, such as those in the Niger-Congo family (which are largely tonal) do not. Despite these differences in phonetic patterning, languages seem to show remarkably similar behavior in other aspects of their word-level phonological structure, suggesting that metrical prominence asymmetries are present at the structural level even in the absence of canonical phonetic cues to prominence. In this talk, I explore relationships between phonetic prominence, linguistic structure, and speech perception through the lens of temporal organization of speech, focusing on speech timing patterns in English and Medʉmba, a Grassfields Bantu language of the Niger-Congo family. Using a metronome-based phrase repetition task, I show that temporal correlates of prosodic prominence are similar across languages, regardless of acoustic differences in prominence marking. Results of a subsequent speech perception experiment demonstrate that prosodic prominence plays a similar role in predictive timing in speech perception across languages. An interesting difference emerges, however, in the way in which speakers of different languages coordinate their predictions relative to an external timekeeper (a metronome). I provide some preliminary evidence that differences in phonetic-prosodic profiles may actually emerge from differences in temporal coordination strategies which are language-specific, or perhaps, culture-specific.