Haskins Labs., 270 Crown St., New Haven, CT 06511
Univ. of Connecticut
Catherine T. Best
A spoken word's phonetic form provides information not only about the speaker's intended message but also about the speaker's language and/or dialect background. The form [dak] may reflect southern U.S. dike, Boston dark, or Chicago dock. Someone who makes an incorrect dialect attribution may misinterpret statements about the [dak]. Anecdotal reports of such misunderstandings abound (e.g., Labov, 1994). In an experimental investigation, it was assessed whether listeners' labeling of isolated words is systematically influenced by correct versus incorrect attribution of speaker dialect. Words that are incorrectly assigned to their own dialect should be mislabeled, in ways predictable from the relative positions of vowel nuclei in the vowel spaces of the two dialects. Stimuli were tokens of heed, hayed, head, hood, hoed, HUD, produced by Utah and Connecticut women. Discriminant analyses trained on the Connecticut productions correctly classified Utah heed, hayed, and head; Utah hoed was often classified as Connecticut hood, and Utah hood and HUD were classified as Connecticut head. Connection listeners answered two questions about each token: i. What word is the speaker saying? and, ii. Does she speak the same dialect you do? Their judgments were generally consistent with the discriminant analysis classifications.