Phonetic categorization shifts for four continua embedded in fluent semantic contexts

Abstract

Initial studies on nonacoustic factors influencing categorical perception of naturally produced speech (Spencer and Halwes, APA, 1978) demonstrated that a labeling crossover between hearing “d” or “t” in a crossed spliced VOT continuum remained approximately the same for different acoustic contexts. But when the continuum steps were inserted into contrasting biasing meaningful sentence carriers, 35 of 42 listeners exhibited crossover shifts. To examine the generalizability of semantic influences on perceived phonetic content, three more series, another “tense‐dense” continuum with changed sentence carriers and two “slit‐split” continuua with similar sentence carriers, were prepared with the same computer editing procedures and presented in listening conditions as before. Only six of 18 listeners exhibited crossover shifts appropriate for semantic contexts in the “tense‐dense” series. The results of one “slit‐split” series were consistent with a semantic effect interpretation but the results of the other were not. Variability across continuua and listeners' responses suggest procedural limitations for revealing influences on phonetic categorization from listeners' attention, acoustic support, and semantic factors. [Research supported by NICHD.]