- American English "Short a" Revisited:A Phonological Puzzle
In this paper I revisit a well-attested and much-discussed phenomenon in the American English vowel system, the dialect variation associated with so-called "short a," as well as the phonological analysis of this variation. In particular, this vowel presents a phonological puzzle in my dialect, though, judging from other discussions of "short a," it appears that this is a puzzle shared by many speakers of one or another variety of mid-Atlantic American English. The data suggest a number of interesting questions for both segmental and prosodic analysis, not all of which can be dealt with here. Several solutions for the issues I will raise have been offered (e.g., Trager 1930, 1940; Cohen 1970; Kiparsky 1989; Labov 1994), though obviously I believe there is still more to be said about the data and how they might be analyzed. Thus, I see this paper as a contribution to a continuing discussion, the resolution of which has still not been achieved. "Short a" is the central concern, but its interaction with other low front vowels in monosyllabic and polysyllabic words must be considered as well.
My native dialect is that of metropolitan New York City, presumably modified somewhat by 44 years of expatriate living, mainly in Wisconsin—a good place, incidentally, to observe a different treatment of "short a."
The vowels which concern me are those that contrast in can 'a container' and can 'modal auxiliary of ability or permission'. This contrast is well known, having an almost hallowed recognition at least as early as Trager's (1930) account. In Trager and Smith (1951), the vowel of can 'auxiliary verb' is represented phonemically as /æ/ and the vowel of can 'a container' as /æh/. More recently, Labov uses the same symbols to represent these different pronunciations in his dialect-normalized notation. In Trager and Smith, /æ/ is categorized as 1 of 9 "simple nuclei"; /æh/ is categorized as a "complex nucleus" consisting of the vowel /æ/ + the postvocalic glide /h/ (i.e., a sequence of vowel plus centralizing inglide). In the Trager and Smith "overall pattern" of American English syllabic nuclei, /h/ is one of three semivowels (along with /y/ and /w/), all of which can combine with any of the 9 "simple nuclei" to express a dialectal grid of 27 possible syllabic [End Page 358] nuclei. Added to these 27 nuclei are the 9 "simple nuclei," thus yielding an inventory of 36 possible syllabic nuclei, only some of which are realized in any particular regional dialect.
I prefer to forgo the elegant symmetry of the Trager and Smith analysis in favor of the view that vowel contrasts are located in the vowels themselves and not in the presence or absence of postvocalic glides. Indeed, I take the view that the postvocalic glides of American English are largely predictable from properties of the vowels and their phonetic environments, though this is not an analytic issue that I will pursue in this paper. Therefore, I will adopt unitary symbols to represent phonetically the vowel contrast under consideration: [æ] for the vowel of can 'modal auxiliary'; [ε] for the vowel can 'a container'. By way of orientation to the symbols being used here, it may be helpful to suggest that my use of [ε] for the vowel of can 'a container' represents the same vowel which, in studies of historical English phonology, is represented as , i.e., "long open e" (e.g., ME < OE stelan); for almost all speakers of present-day standard English, it is the vowel heard in care, bear, and stair. Further, to put these two vowels in the larger context of front vowel contrasts in my dialect, I cite the following minimally distinctive forms, including their feature categorization in terms of vowel height and tense/lax:
The [+tns] vowels are predictably followed by glides: [y] after [i, e], after [ε], that is, a high palatal glide after the high and mid tense front vowels, a low centralizing glide after the low tense front vowel. Rule (2) describes this predictable glide insertion.
2. ø → [–cons, αhi]/[+tns, –αlo]___ (where α ranges over + or – values, and – α expresses the obverse of the value of α)
Thus, (1) and...