Among the clearest of ``soft coarticulation'' (linguistic-phonetic) effects between vowels and following consonants is that between // and following //.
The effect of following // is phonetically quite strong in certain cases, and in combination with certain vowels. This is reasonable in a number of ways. First, the body of the tongue must raise in order to produce a velar closure for //; this movement is relatively slow, compared to the movements of the tongue tip, and this makes it more likely that the // gesture will be coarticulated with surrounding segments. Partial nasalization of a vowel in the context of a following nasal consonant occurs frequently in English; nasalization has an obscuring effect on the formant frequencies of vowels. Finally, // represents a lengthening environment in English, since vowels before voiced segments are generally phonetically longer than in other environments. In an environment where lengthening has applied, there is extra time for gestures to reach their targets (relative to a following non-lengthened environment, such as before /k/), so that the restrictions of hard coarticulation are weakened, and there is an opportunity for variation in the temporal form of the gesture. Thus this environment seems a priori to be one that may well be exploited to produce stylistic or dialect-specific phonetic effects that are audible but not lexically distinctive. Consider, then, the effects of following // in different dialects, as shown in Figure .
Insofar as this data is capable of displaying, dialects are perfectly self-consistent, but quite different from one another. I will concentrate on the effects of // on preceding //. The opportunity that occurs in the environment of //, to have dialect-particular patterns of soft coarticulation, appears to be exploited in the Alabama dialect, as opposed to the other dialects. In Chicago and elsewhere, the effect on the short high-front vowel // (written /i/ in Jamaican) is uniformly one of fronting, and usually also raising (except in L.A. Chicano). This fronting effect cannot be due to a phonological alternation between /i:n/ or /n/ and // in the form of the present participial suffix -ing (the transcription distinguished -in' and -ing, so cases with following /n/ are not included here). Thus there is a general high-fronting effect on // of following //, which may be understood as a natural phonetic effect, since front-articulated velars might reasonably have a fronting effect on preceding vowels. (This is a complex point, however, since velars get their front-back articulation from the adjacent vowels to begin with. For this effect to actually work this way, there must be two steps, in which the consonant gets its relatively front place of articulation within the range of ``velar'' places of articulation from the front-back features of the vowel, and then the vowel is itself fronted by the fronted velar consonant.)
However, in the Alabama pattern of James H., in Figure , the effect on // is entirely different. Here, instead of the apparently natural high-fronting effect found in other dialects, including Jamaican as well as LA Chicago and Chicago White English, the effect is an opposite, unnatural, anticoarticulatory effect. Instead of raising //, following // lowers it. Impressionistically, in this speech I find that // sounds like [æ]. This occurs not just in the suffix /-I/, but also in spring, finger, thing, etc.
The effect of // in Alabama is perfectly general within this idiolect, insofar as this data shows it: vowels preceding // undergo lowering. Thus // is realized as , /w/ is realized as [o] (quite different from [o] which is typical in other environments). F1 rises in frequency for // by 118Hz (p19#190.0002), for // by 108Hz (p19#190.00001) and for /w/ by 54Hz (p19#190.005). This consistent and significant pattern suggests a phonetic lowering rule for the nuclei of vowels in the context of following // in James H.'s speech. The pattern is consistent with the generally greater amount of gliding and diphthongization occurring in Southern speech (cf. Feagin 1991).
This is further clear evidence for dialect differences in the system of phonetic implementation. I know of no proposals that // itself might different across dialects, so I assume that the consonant which triggers the effect is the same as in dialects where it does not have this effect. The effect cannot instead be attributed to the phonological features of any particular subset of the vowels, since all three of these vowels show the same effect. It must be a phonetic process, characteristic of this speaker and, one may assume, of other speakers of his dialect.