RYSLING COLLOQIUM AT UCLA
Amanda Rysling gave a colloquium talk at UCLA on October 19. The topic touched on her dissertation work, along with some newer developments of those and related ideas. The abstract is below:
Regressive spectral assimilation is a default in speech perception
The vast majority of work on segmental perception focuses on how listeners differentiate adjacent speech sounds from each other, or compensate for coarticulation. Much of this work assumes that listeners are so successful at differentiating successive speech sounds that their failures, cases when adjacent speech sounds are heard as similar, are vanishingly rare (e.g., Ohala, 1981, 1993, i.a.). The first part of the talk demonstrates that such failures are not as rare as they have been assumed to be: listeners in clear listening conditions productively hear the first of two sounds as similar to the second, effectively failing to use available context information to compensate for coarticulation. Four claims are then advanced: (1) that segmental perception is better understood as a process of incremental evidence accumulation and evaluation, which is neutral as to whether acoustics are attributed to their source segments or gestures, (2) that the auditory system defaults to treating different types of acoustic properties (e.g. spectral weight, duration, intensity) differently, (3) that perceptual spectral assimilation on clear speech, which previous accounts of compensation for coarticulation have predicted should not occur, has a domain-general basis, and (4) that this tendency represents a default in speech perception that is responsible for the overwhelming typological prevalence of regressive major place assimilation in the phonologies of the world’s languages.
