Here I discuss the set of phonological contrasts in Chicago White English (henceforth CWE). A summary of the next few paragraphs is given in tables of phonemes and the lexical sets that constitute them below. Again, we will use the lexical sets of Wells (1982) as a convenient way of summarizing the phonological differences between this dialect and others.
CWE, like LA Chicano English and most American dialects, does not distinguish the pairs of Wellsian lexical sets NORTH and FORCE; CLOTH and THOUGHT; BATH and TRAP; PALM and LOT. But CWE is not entirely like LACE in inventory: the merger of //(CLOTH, THOUGHT) with // (LOT, PALM) has not overtaken Chicago, though it has reached other cities in the Great Lakes area including Cleveland and Pittsburgh.
Some confusion exists about the ``broad A'' class of words (Wells' PALM set). Bloomfield (1934) reported that ``the great majority [of educated Chicago speakers] do not distinguish the vowel of hot, sod, bomb from that of father, far, balm''(p. 97), that is, LOT is not distinguished from PALM (so-called ``short O'', and ``broad A''). However, he claimed that ``some speakers, like B[ritish RP speakers], distinguish [these vowels]''(p. 97). Of the Americans that Bloomfield had contact with at the University of Chicago, it is quite likely that many had exposure to RP, which held great prestige in the U.S. at that time, and further that some of these speakers modified their pronunciations of some items in the PALM class in the formal elicitation style which linguists of that era commonly used for data. Thus it cannot be assumed that broad A has ever been distinct in the vernacular English dialect of Chicago or the other Northern Cities.
Further evidence can be found to support this position. An examination of the Northern Cities speakers7.7 in Labov, Yaeger, and Steiner (1972, Figures 10-23 and 28) shows that where broad A (PALM, symbolized as /ah/ in those figures, following Trager & Bloch 1941) was classified separately from short O (LOT) (that is, in Figures 11, 14, 16, 18, 28), they always overlap each other, more or less completely, excepting only a single token in a reading-list passage by one Chicago speaker (Fig 28). Thus there appears to be no good evidence that the PALM set is distinguished from the LOT set in Chicago.
The unstressed high-front vowel of the lexical set HAPPY is in this dialect higher and fronter than /I/ as in KIT, and furthermore, upgliding, so that it is phonetically most similar to the vowel /iy/ as in FLEECE. Phonetic similarity and complementary distribution7.8 suggests that HAPPY and FLEECE contain the same phoneme. This is different from the situation in British RP, for example, where HAPPY ends in a high-front lax vowel [I], making it phonetically most similar to the phoneme in KIT. Thus it is identified phonologically as /I/ in RP. By a similar argument, unstressed // as in LETTER is the unstressed allophone of the // phoneme in NURSE in this dialect. This differs from partially rhotic dialects such as Jamaican, where NURSE retains some phonetic rhoticity, but where the final vowel in LETTER is not distinguished from the final vowel in COMMA. Thus the following phonemes are contained in words of the indicated lexical sets:
|//||LOT = PALM|
|/:/||THOUGHT = CLOTH|
|/æ/||TRAP = BATH|
|/or/||NORTH = FORCE|
|/iy/||FLEECE = HAPPY|
|//||NURSE = LETTER|
I assume that the phonological inventory, exclusive of // and the unstressed vowels, is that represented in Table .7.9
As in Reference American, the long vowels are formally represented nuclei with phonologically unspecified glides. The phonetic glides that occur on /iy, ey, ow, uw/ are attributed to rules of phonetic implementation. The -y, -w transcriptions, used in this chapter, represent the more fully specified phonetic forms which are derived from the abstract classes.
Two further comparisons to Reference American should be pointed out. First, Mary, merry, and marry have merged in Chicago (though Bloomfield's (1934) discussion suggests otherwise).
Second, despite Bloomfield's conjectures about broad A in Chicago, it is apparently not distinguished there. The gap opened up by the merger of broad A with short O (PALM and LOT, respectively) in the long, low-front position would therefore seem to be part of the explanation for the lengthening, peripheralizing, and raising of the /æ/ phoneme (TRAP, BATH), which is the first step in the Northern Cities Chain Shift discussed below. The raising of /æ/ is one of the most well-understood American sound changes, both in its phonetic and social conditioning (cf. LYS:Chapter 3, Labov 1990, Callary 1978, Feagin 1991, among others). If this raising is analysed phonologically as first lengthening and then peripheralizing, raising, and breaking, then the first step is the shift of this sound class to the long vowel (V:) subsystem. But this shift itself has as a precondition a gap in the /a:/ slot, which is occupied in Reference American by the broad A class. Thus the merger of broad A and short O (PALM and LOT) is a precondition of the raising of /æ/. Further research on English dialects is necessary to confirm this prediction.
Third, the phonological identification of // with // may be impossible in this dialect, since // (but not //) is moving to the low-back corner, as discussed below.