A Methodological Suggestion on /aj/ Ungliding

The /aj/ vowel, one of the most studied in the English language, is noted especially for its involvement in the Great Vowel Shift (Labov 1994, 145) and its role in traditional variation patterns of the U.S. South. Although this vowel and the Southern system were closely considered by earlier work (Sledd 1966), quantitative sociolinguistic studies (Thomas 1995; Schilling-Estes 1996; Wolfram, Hazen, and Schilling-Estes 1999) have discovered linguistic constraints for unglided /aj/ and have established a specific methodology for studying it. Within the linguistic constraints, the normal descending order of phonetic environments that favor unglided /aj/ is the following: liquids (e.g., /l/ as in mile and rhotic liquids as in tire), nasals (e.g., /n/ in mine), voiced obstruents (e.g., /d/ in bide), and voiceless obstruents (e.g., /t/ in bite). Using these categories helps considerably in assessing the degrees of linguistic and social variation within a speech community.

The field of language variation is now old enough to have commonly accepted methodologies for variables such as copula absence (Blake 1997) and /aj/ ungliding (Wolfram and Schilling-Estes 1995). But complete descriptions of language, and thereby better theories of both social and linguistic constraints on language, must exhaustively determine the range of linguistic variation. All possible effects of the linguistic environment need to be assessed for possible correlations with the dependent variable. Otherwise, patterned linguistic variation may go unrecognized. Following environment for /aj/ ungliding is a case in point.

For this environment, the syllabic position of the following segment may be important. ¹ As for nasalization in English, preceding vowels are nasalized when the following nasal is in the coda of the syllable (e.g., bane) but not when the following nasal is in the onset of the following syllable (e.g., banal). With liquid vocalization, the postvocalic /r/ or /l/ has to be in the coda of the syllable and not in the following onset (e.g., Par, Paris; Hell, Hello). With both nasalization and liquid vocalization, the syllabic position of the sonorant is crucial: the vowel must share the rime with the following sonorant for the process to take effect. In the traditional model of the syllable, the rime comprises a possibly branching nucleus and a possibly branching coda in English (Carr 1999; cf. Blevins 1995). Simple linear order is not sufficient to determine what, theoretically, the effects of the following segment might be. Hierarchical order must be considered. ²

Phonetically, /aj/ ungliding takes place when the diphthong's offglide is severely shortened, reduced, or eliminated (all three I will group under the term NEUTRALIZE). The traditional hierarchy of following environments moves along the sonorancy scale (Blevins 1995, 210; Carr 1999, 72), presumably because more-sonorant following elements neutralize the offglide to a greater extent than less-sonorant following elements. Since the sonorancy effects on the offglide are the crucial linguistic mechanism in /aj/ ungliding, the syllabic position of the following segment must be considered. The degree to which sonorancy affects the vowel depends on whether the consonant falls in the same syllable (e.g., mining versus mine). If the sonorant does not share the rime with the vowel, there is less opportunity for the sonorant to spread to the preceding vowel and extensively neutralize the offglide (see figure 1). This effect is not deterministic; variation of unglided /aj/ still occurs in both tauto- and nontautosyllabic environments (see Guy 1997).

In a study of West Virginia and North Carolina families (Hazen 1999; Hazen and Hall 1999), /aj/ ungliding was influenced by the syllabic position of the following sonorant, in addition to simply the type of following segment. For example, one speaker had a 92% rate of ungliding (24/26) before tautosyllabic nasals but 27% (3/11) before nontautosyllabic nasals. If only the type of following segment had been assessed, then his rate of ungliding before nasals would have been 88% (22/25), but an important distinction in the data would have been missed. This variation...