Statistical information and coarticulation as cues to word boundaries: a matter of signal quality

Abstract

We investigated how statistical information in the form of transitional probabilities (TPs) interacts with coarticulation, another sublexical cue to word boundaries, and examined the impact of signal quality on the weighting of these cues. In an artificial-language-learning setting, with phonetically intact speech, coarticulation overruled TPs, suggesting the predominance of subsegmental, low-level information. However, whereas the role of coarticulation in segmentation was highly modulated by signal quality, TPs were very resilient to noise. When coarticulation was rendered unreliable by strongly degrading the input with a 10-dB signal-to-noise ratio (SNR), only statistical information drove segmentation. In a more mildly degraded 22-dB SNR condition, in which some acoustic properties were still available, coarticulation was exploited, although with less reliability than in optimal conditions. These results can be interpreted according to a hierarchical approach (Mattys, White, & Melhom, 2005) in which both the available segmentation cues and the listening conditions have an important role in speech segmentation.