Accounting for Speaker Variation in the Production of Prominence using the Bayesian Information Criterion

This study investigated speaker variation in the production of various acoustic cues of prominence, including durational and intensity measures. This study stems from our prior work where we used the Bayesian Information Criterion to determine whether each cue were gradiently or discretely associated with prominence. In our prior work, we found that features vary as to whether they are gradient or discrete. In the present study, we found a similar result. For all speakers, some features are gradient and some features are discrete in the manner in which they cue prominence. Under a metrical stress notion of hierarchically layered prominence, our result would suggest that some speakers do not exploit the full range of prominence distinctions