Selective adaptation produced by ongoing speech

Abstract

In the selective adaptation paradigm, subjects typically hear repetitions of a single syllable (e.g., [pha]) and subsequently identify syllables from a test series spanning two phonetic categories (e.g., [ba] − [pha]). Experiments have shown that the adapting syllable produces a shift in the phoneme boundary of the test series, such that fewer syllables are assigned to the phonetic category of the adapting syllable. In the present experiments we determined that selective adaptation could be observed using ongoing speech. Subjects were first presented with syllables for identification from an acoustic continuum (e.g., [tha] − [da]) in order to determine their phoneme boundary in an unadapted state. Subjects then heard three sentences containing a predominance of a particular phoneme or phonetic feature. For example, a “voiceless stop” sentence might be: “Carl tickled Tillie's toes till Tillie told Carl to quit tickling her toes.” After the third such sentence, subjects were again presented with syllables from the [tha] − [da] series for identification. The sentences produced a significant shift in the boundary of the [tha] − [da] series in the predicted direction. Similar experiments were performed for sentences containing a predominance of affricates or nasals, and similar results have been obtained using prose passages instead of sentences.