Cross-language investigation of context effects

Allard Jongman, Yuwen Lai, Corinne Moore

            In the perception of spoken language, the context of an utterance is influential. Two aspects of this context are highly relevant: fundamental frequency and duration. It has been shown that fundamental frequency (which gives rise to our percept of pitch) and temporal parameters can affect the recognition of a following word or phrase. For normal speech perception, the contextual influences are critical. Changes in fundamental frequency (F0) are the result of changes in speaker, with females typically having a higher F0 than males. Changes in duration are the result of changes in speaking rate. Both aspects of context have a major impact on speech perception processes.

            The purpose of this project is to examine if and how listeners compensate for contextual changes in F0 and speaking rate. A first set of experiments explored the extent to which different acoustic cues (i.e., spectral and temporal) contribute to the specification of a given linguistic distinction. Specifically, the roles of F0 range and Turning Point were explored as cues to the distinction between Mandarin Chinese Tones 2 and 3. These findings were published as Moore and Jongman (1997).

            A second set of experiments compared performance by native and non-native speakers to address whether compensation for contextual influences is a language-specific or automatic, language-universal process. A first report on this cross-linguistic comparison can be found in Jongman and Moore (2000).  The most recent results were presented at the June meeting of the Acoustical Society of America (Jongman, Lai, and Moore (2002): Language-specific effects in speaker and rate normalization).

[Research supported by The University of Kansas]