| Abstract | The acoustic realization of speech sounds depends greatly on the vocal tract that produces them. Adult speakers differ in their vocal anatomy, especially across the sexes, and every listener must take such differences into account. Additionally, every adult begins life as a child (“I started out as a child,” as Bill Cosby used to say), and producitve language begins as early as 1 year of age, well before the head has reached its full, adult size. So each speaker must be able to normalize his or her parents’ speech in order to know what they are supposed to learn. And one’s own speech must be normalized for the current size of the vocal tract, not to the size during the early stages of acquisition. Given such variation, humans have had to make use of a communicative system that allows listeners to make compensations for changes in speakers. Listeners clearly do make adjustments, most clearly in the case of vowels (Nearey, 1989), but also with stops (Rand, 1971) and fricatives (May, 1976; Strand & Johnson, 1996). Thus speaker normalization must be considered a feature of speech perception. |
|---|
