Speaker Variation and Vocal‐Tract Size

The hypothesis that differences in the formant‐frequency value sets among speakers of the same dialect are owing chiefly to variation in individual vocal‐tract size was tested on the Peterson‐Barney formant‐frequency data. For three speaker classes—men, women, children—the distribution of values for each of three formants of each of ten vowels was correlated with every other such distribution. The hypothesis implies that the correlation scores should be high. Actually, most scores were low, though interesting subsets (F2 with F3 for each front vowel, F1 with F2 for each back vowel, F3's with one another for children) were high. Furthermore, the separation between male and female distributions for some vowel formants is much sharper than variation in individual vocal‐tract size can reasonably explain. The variation within class must be stylistic, not physical; and the difference between male and female formant values, though doubtless related to typical male and female vocal‐tract size, is probably a lingu...