ASA 125th Meeting Ottawa 1993 May

4pSP21. Alternative scoring systems for the modified distinctive features difference test [DFD(m)].

Penny A. C. Anderson

Donald G. Jamieson

Hear. Health Care Res. Unit, Elborn College, Univ. of Western Ontario, London, ON N6G 1H1, Canada

The modified distinctive features difference test [Cheesman et al., Proc. Int. Conf. on Spoken Lang. Processing (1992)] measures the ability to identify the 21 dominant English consonants in an /(inverted vee)C / context. The test contains 84 stimuli (21 sounds, each spoken by four speakers of Central Canadian English). Target sounds are normally presented in a noise background. The possibility that the sensitivity of the test increases when listener's DFD(m) confusion errors are scored was examined in terms of phonetic features rather than in the more traditional, whole-item terms. Several alternative scoring approaches were examined: the classical place, manner, and voice distinction and those systems used by Chomsky and Halle (1968), Miller and Nicely (1955), and Feeney and Franks (1982). Scoring using a phonetic feature system does offer certain advantages over whole-item scoring. [Work supported by NSERC and the Ontario Ministry of Health.]