Impressionistic transcription of long, stressed tokens of these vowels
in monosyllabic words, excepting those before /r/, results in the
range of impressionistic realizations of the above phonemes presented in
Table ![[*]](icons/crossref.png){kind=icon}
.
Some observations are worth noting. First, this speaker is not strongly participating in the widespread raising of /æ/, since he produces cardinal #4, [æ], in stressed tokens, though coarticulated tokens sound more raised, as far as [].
In fact, the opposite change is occurring in one environment: // falls towards [æ] before /l/, as in [æ]levator, Mr. B[æ]lvedere, etc. Since /æ/ does not exist in Spanish, the fall of /l/ cannot be attributed to Spanish influence. On the contrary, the fall of /l/ seems to be a purely English sound change that happens to occur in this particular ethnic group, not a contact- or contact-induced phenomenon.
/u:/ is somewhat front, as in most American and many British dialects. Anglo speech in Southern California shows even greater fronting of /u:/, to such an extent that /u:/ and /U/ overlap with /i:/ and /I/ in formant space.
According to Santa Ana (1991), even young Spanish-speaking immigrants who arrive in Southern California at 8-10 years of age, learn English with /u:/ pronounced so far to the front that it overlaps with /I/. Tokens are high and central, just as in an unpublished analysis I have done of a California Anglo speaker. For all speakers Santa Ana studied, Chicanos front /u:/ to overlap with /I/ (except one, whose /u:/ is as still as front as /I/, but is higher).
/w/ has some fronting of the nucleus and falling of the glide. In Philadelphia and many Southern dialects, the nucleus of /w/ has shifted toward [æ] and sometimes goes as far as [e], while the offglide falls from [U] to [o] and sometimes as far as [].9.16 Vince thus appears to be participating in this shift of /w/, but not to the extremes found in other dialects.
Finally, some realizations of /i:, e:, o:/ and other long vowels were transcribed as monophthongal. This may be an effect of Spanish, though other American dialects (Minnesota, and Wisconsin, for example) also show monophthongization of these vowels, which are most commonly diphthongs in English.9.17 Also, these vowels are underlyingly long monophthongs, so the general effect here is to simplify the system of phonetic implementation, as compared with the /iy, ey, ow, uw/ of many other English dialects.
These impressionistic transcriptions need some qualification: they give a rather qualitative idea of how one transcriber perceived a small number of tokens on one particular occasion. While the impressionistic transcription of a few stressed tokens can generate considerable insight, it is difficult to summarize large quantities of impressionistic data, and it may also be difficult to use such data to substantiate the finer allophonic differences which are the main focus of this research. Further, it is difficult to accurately transcribe short, coarticulated, unstressed tokens, which may behave differently from the easily transcribed long and stressed tokens. For these reasons, it is important to have more -- and more consistent -- data on vowel quality. If the distribution of a great many tokens in phonetic vowel space could be viewed at once and summarized, that would constitute a more trustworthy representation of vowel data. Such a quantitative and objective representation is possible using formant-frequency charts. As shown in Chapter 4, small but measurable differences in formant frequency can have clear, perceivable consequences. Therefore we may assume that much of the information on a chart of F1 vs. F2 frequency measurements reflects perceptible vowel quality. Thus we now move from impressions to instrumental measurements.