Third European Conference on Speech Communication and Technology

Berlin, Germany
September 22-25, 1993


Factors Affecting Adaptation to Time-Compressed Speech

Gerry T. M. Altmann, Duncan Young

Laboratory of Experimental Psychology, University of Sussex, Brighton, UK

Speech inputs vary widely in amount of background noise, speaking rate, and accent; yet listeners can quickly adapt to such variation and recognize the content of an utterance. The present study is aimed at understanding the mechanisms underlying this adaptation, and the processing unit which the human recognition system attempts to recover during adaptation. We examined adaptation to speech under various conditions of time compression. Specifically, we explored how the intelligibility of a target set of compressed English sentences varied as a function of prior exposure to three kinds of compressed stimuli: compressed English, compressed French, and in a separate study, compressed "nonsense" sentences. We also explored the degree to which the adapting effects of prior exposure to compressed speech persist over time. We conclude on the basis of the results that lexical level word recognition does not drive the adaptation mechanism. Instead, recognition of sub-lexical units, or suprasegmental regularities in rhythm, may form the basis for successful, and persistent, adaptation.

Full Paper

Bibliographic reference.  Altmann, Gerry T. M. / Young, Duncan (1993): "Factors affecting adaptation to time-compressed speech", In EUROSPEECH'93, 333-336.