Explanations for lexical exceptions of sound change

Seo-young Chae

Sat. 4:00-5:40 A

The Neogrammarian controversy has been resolved in such a way that sound change is essentially regular and phonemes are the unit of sound change, even though some lexical exceptions are allowed in the course of change. Labov (1994) lists the exceptional items in short /æ/ raising of the major cities in the Middle Atlantic States: most of them fall into phonological, morphological or semantic categories while only a few are purely unexplainable exceptions. Nevertheless, no explicit explanations have been given so far about the nature of the lexical exceptions. This paper attempts to provide a general account of the lexical exceptions of sound change, based on four vowel changes: (1) the English short /æ/ raising in Middle Atlantic States (Cohen 1970, Labov 1981, 1994), (2) the French back vowel chain shift (Eckert 1980), (3) the Persian prenasal low back vowel raising (Modaressi 1978), and (4) the Korean mid-back vowel raising (Chae, in preparation).

Lexical exceptions to sound change are primarily dependent upon the phonological characteristics or the direction of the change itself. Since English short /æ/ is raised after it is tensed, items which belong to weak positions are resistant to sound change. However, vowel raisings in French, Persian and Korean are part of a weakening process. Therefore, tense environments (such as word-initial positions, initial positions across a morpheme boundary, and mono-syllabic words) are resistant to raising. Secondly, morphological categorization of the exceptions reflects this phonological constraint. For instance, grammatical function words which are usually weak in terms of phonological stress, are resistant to English short /æ/ raising (can, am, have, said, etc.), but are primarily subject to raising in the ongoing changes of French, Persian, and Korean. The third factor is the social correlation of the lexical items. Loan words, as well as 'formal', 'archaic', and 'learned' words are generally resistant to sound change. Not all of the items in these categories can be explained in terms of low lexical frequency. In both Persian and Korean, loan words are not subject to raising since they are closely correlated with social class factors. However, a small number of old and familiar borrowings from Arabic and Chinese (respectively) are subject to sound change, so that the speakers become unaware of their foreign origin. In all four of the sound changes, words that can be categorized in terms of formality, learnedness, or archaic usage are found to be exceptional.

The actual number of lexical exceptions in each change varies enormously. Exceptions to English short /æ/ raising are treated by linguists as geographically different (e.g., Labov 1981, 1994) which may be due to the fact that the phonological environment of raising is gradually expanding. The exceptions in Persian and Korean are quite numerous since they are more restricted by phonological conditioning. Nevertheless, the lexical exceptions can be categorized and explained as described above and we can predict that these exceptions, though currently resistant to sound change may will be the target of the sound changes.