The rule of nucleus-glide differentiation can be understood as assuming a fixed glide endpoint, and shifting the nucleus away from it. However in the shift of /aw/ to [e], both nucleus and glide are shifting simultaneously, and in opposite directions. This understanding of nucleus-glide differentiation must be incorrect, if this case is an instance of it. When the glide falls to mid, and the nucleus raises to mid, the result is that both nucleus and glide are mid, whereas before they were low and high, respectively, and thus became less distinct.
The other case I know in which the glide shifts as well as the nucleus, is French ``moi'', which derived from Latin /me/, presumably by diphthongization to /mei/, nucleus-glide differentiation to /mai/ and /moi/ (the stage reflected in the current spelling). The current form is /mwa/, showing three further changes from the /moi/ form: The first segment moves to high-back; syllabicity shifts from the first to the second segment; and the second segment falls from high to low position, resulting in /moi/ 14#14 /mwa/. The ordering of the last three shifts cannot be ascertained from the two-stage picture provided by the spelling and the current pronunciation.8.18
In both the shift of Southern /aw/ to [e] and French /oi/ to [ua] (not specifying syllabicity here), the two segments in each vowel move in opposite directions, perhaps simultaneously. In the French case it is the front vowel that falls while in the Southern case it is the back vowel that falls; in both cases it is the second segment that falls and the first that rises. Why the both nucleus and glide should both shift simultaneously is a mystery.
Nucleus-glide differentiation may be related to the rules of sound change based on peripherality of Labov, Yaeger, and Steiner (1972), and to the vowel-before-vowel raising rule of Kenyon and Knott (1941:xxxviii, #92), discussed in Phonological Preliminaries. But an explanation for this shift of /aw/ to [e] has yet to be found.