Small-Town Values and Big-City Vowels

Matthew J. Gordon tackles a much needed area of research. Previous work on the Northern Cities Shift (see fig. 1) has focused on the larger urban areas involved, including Buffalo, Detroit, and Cleveland. This emphasis is based on the idea, put forth by Labov (1994), Callary (1975), and others, that urban areas are the innovators in sound change. However, the Northern Cities Shift is not found exclusively in those areas. Gordon demonstrates its appearance in small rural communities as well.

Figure 1.

The Northern Cities Vowel Shift

(after Labov 1994, 191; cited in Gordon 2001, 2, fig. 1.1)

Gordon focuses on Paw Paw and Chelsea, Michigan. Both have populations of roughly 4,000 people; both are located 20 miles from a medium-sized city. Paw Paw is close to Kalamazoo (population ~80,000) in the state's southwest corner, while Chelsea is close to Ann Arbor (population ~100,000) in the southeast. Neither town is highly isolated as both are located on Interstate 94, which connects Chicago and Detroit.

Sixteen speakers from each community made up the sample. Age was controlled with two equal groups, one representing adolescents (ages 16–18) and the other adults (ages 39–51). The sample group covered both sexes. By constructing the sample in this way, Gordon was able to look for the well-known social factors of geography, age, and sex that commonly affect language use. Data were collected using word lists and conversational interviews, though the quantitative analysis which followed used only the conversational data.

In any sound change, it is important to account for the contributions of both linguistic and social factors. The author considers each of the six vowel shifts separately along these grounds. For any possible linguistic effect, he looks at several factors, starting with the potential influence of neighboring phonological elements. Preceding and following phones are categorized in terms of place, manner, and voicing. It is also noted whether the adjacent elements form consonant clusters. For example, Gordon's phonological analysis of elements following (æ) shifting to [e] or [ε] is reproduced in table 1. The statistical testing indicates that all factors except clustering are statistically significant; however, the data also establish that while some following (and preceding) environments may favor or disfavor shifting, there is no absolute phonological conditioning from adjacent elements, and as can be seen in table 1, the shifting occurs only roughly half the time, not making a strong case for linguistic conditioning.

The last consideration is of word-level factors, specifically the number of syllables in the attested word and the position of the syllable in which a vowel has shifted. Table 2 shows that final syllables do favor (æ)-shifting, while words containing more than four syllables tend to disfavor it. Once again, however, the results are not compelling enough to make a case for linguistic conditioning.

Table 1.

Effects of Following Phonological Factors on (æ)-Shifting: General Indices, Frequency of Shifting, and Chi-Square Results (from Gordon 2001, 128, table 4.3)

One drawback to the phonological analysis presented is that it does not give a thorough account of the effect of stress reduction. In terms of English vowel quality, stress is a far more important conditioning factor than syllable position or syllable count. While not a factor in the case of (æ) shifting to [ε] or [e], the Northern Cities Shift is notable for the number of other affected vowels shifting to schwa. For example, while (ε) has been traditionally marked as the only vowel shifting to [ə], Gordon also confirms Labov's (1997) finding that (I) may also shift to that target. Given that English has a normal phonological pattern of reduction to schwa in unstressed position, it is important to separate what is Northern Cities Shift from what is stress reduction...