Effects of compound priming on listeners’ identification of a /kVC/ continuum

Abstract

A continuum was synthesized to span the words cop, cup, cob, and cub. Two cues were varied in six steps each. Vowel F2 ranged from 960 to 1160 Hz. Voice bar duration for the final consonant ranged from 10 to 60 ms. Other properties approximated the average the four words pronounced by a male speaker in the compound nouns: traffic cop, tea cup, corn cob, and bear cub. The four context words, traffic, tea, corn, and bear, were also synthesized. Thirteen listeners categorized ten replications of the stimuli in each of five contexts: as isolated words and following each of the context words. As expected, words are favored in appropriate contexts. Logistic regression indicates that about 96% of variance in listeners’ responses can be attributed to phoneme-level stimulus effects and to stimulus-independent (phonological and lexical) biases. There is also evidence for small changes in sensitivity to cues as a function of lexical context. The latter result is of interest for certain alternative models [D. Massaro, Cog. Psych. 23, 558–564 (1991)]. However, observed changes in sensitivity do not relate well to predictions from any existing theory. [Work supported by SSHRC.]