Abstract:

Phonological speech errors are a traditional source of evidence for phonological units, such as features and segments, in both linguistic and psycholinguistic theory. Assumptions about the unitary nature of speech errors have been brought into question by research that found evidence for subfeatural errors in speech production [R. Mowrey and I. MacKay, J. Acoust. Soc. Am. 88, 1299--1312 (1990)]. This poster presents data from an ongoing acoustic study of speech errors in /s/ and /z/ productions. Previously reported findings indicate that the errors involve either single or multiple changes in periodicity, amplitude and duration of frication, and vowel rise time [S. Frisch and R. Wright, J. Acoust. Soc. Am. 101, 3178(A) (1997)]. The data are tape recordings of an error-inducing experiment using nonsense tongue twisters. Recordings of six talkers producing four different tongue twisters targeting /s/ and /z/, for examples, ``sit zap zoo sip,'' were digitized and analyzed. This study examines the phonetic variation in featural and subfeatural errors across talkers and segmental contexts. Results suggest that details of individual talker's production strategies and the phonetic context influence the distribution of featural and subfeatural errors. [Work supported by NIH Grant No. DC00012.]