Advances in Sociophonetics

26 William Labov

Table 2. Comparison of two vowel variables in second and third generations

in New York City.

Upper middle class Jewish Working class Jewish

Younger men [21–30 years] Older women [40–65 years]

/æh/ scores /oh/ scores /æh/ scores /oh/ scores

2nd gen3rd gen 2nd gen 3rd gen 2nd gen 3rd gen 2nd gen3rd gen

N 3 6 3 6 6 3 6 3

Mean 31.0 30.3 27.0 24.5 28.6 28.6 19.9 20.6

Std dev 7.8 5.8 1.6 4.8 3.9 10.9 3.5 1.1

t-test 0.15 0.86 0.47 0.33

There are no significant differences between generations for either group in the rais-

ing of (æh). Nor are there any significant differences between the 2nd and 3rd gen-

erations in the raising of (oh). We see that the eventual result for young and middle

aged adults is the same for children whose parents were speakers of Yiddish and

those whose parents were native speakers of the New York City dialect.

2.7 The effect of ethnicity on sound change in Philadelphia

In the 1970s, the Philadelphia Neighborhood Study created a stratified sample of

the white mainstream areas of Philadelphia in ten neighborhoods that embod-

ied a full range of social classes and ethnicities (Labov 1980, 2001). The circles

on Figure 7 shows the mean values of the vowels of 116 speakers. The arrows

indicate the direction of change as determined by the age coefficients of step-

wise multiple regression. The head of the arrow represents the expected values

for speakers 25 years younger than the mean age for the sample; the tails of the

arrow show expected values for speakers 25 years older than the mean.^2 The

three largest arrows indicate the new and vigorous changes in the system; here

we are focusing on the raising and fronting of /ey/ in checked syllables: main,

paid, late, etc. This change in progress was first recognized in the course of the

acoustic analysis of the 116 speakers but is found consistently across all social

groups. These results from apparent time studies were confirmed by a re-study

in real time by Conn (2005).

Table 3 shows the full output of the regression analysis for the fronting of

/ ey/ in checked syllables. Age is negative and significant at < .0001 probability,

as Figure 7 indicated: that is, the younger the speaker, the higher the value of the

2. Expected values for F1 and F2 are calculated by multiplying the age coefficients in the output
   of the regression times 25 or −25 and adding this to the regression constant.