Effects of instructions to produce hyperarticulated vowels on formant values

Abstract

Clear speech collected in the laboratory is typically well articulated, but speakers can produce even clearer speech (‘‘hyperarticulated’’) when so instructed. Johnson et al. [Language 69, 505–528 (1993)] found that hyperarticulations, averaged across speakers, are more extreme in both the F1 and F2 dimension. Here, 10 speakers of a Rhode Island dialect of English recorded words with 11 different vowels in (h)Vd environments (e.g., ‘‘heed,’’ ‘‘aid’’). Individuals show the effect for between 2 and 9 vowels; there was no vowel that had more extreme formant values (for F1 or F2) for all 10 speakers. (Averaged across talkers, only the low front vowel had more extreme values in both F1 and F2.) Typically, hyperarticulations that were not more extreme in formant values were equivalent to the normal production, while a few had less extreme values. The lack of consistency in the individual results calls into question the use averaged values, since speakers may be changing articulatory parameters in a more complex way. The dynamic nature of some of the vowels may have an influence, since the initial measurements were taken from a single point in the syllable. [Work supported by NIH.]