|
Plenary talk 2 Consonant Epenthesis in Austronesian: Natural and Unnatural History
Juliette Blevins University of California, Berkeley
Phonological rules of consonant epenthesis occur in many of the world’s languages, and typically involve insertion of a glide adjacent to a vowel. Both the feature content of the inserted segment, and its position have been treated as consequences of synchronic markedness constraints. Under these accounts, epenthetic consonants of a certain quality are inserted because these are the least marked non-syllabic segments. These consonants are inserted adjacent to vowels in order to satisfy syllabic markedness constraints, for example, the requirement that syllables have onsets.
Four general problems arise for synchronic analyses of consonant epenthesis which are rooted in markedness constraints like those just mentioned. First, segmental markedness constraints are unable to account for a striking cross-linguistic generalization: in the majority of cases where the historical phonology can be reconstructed, and where segments are not phonetically predictable, epenthetic consonants are precisely those for which earlier consonant-loss is evidenced. Second, segmental markedness constraints are unable to account for rare but attested cases of highly marked epenthetic consonants; in these cases also, historical rules of consonant loss are attested. A third problem for such accounts is that in some languages, the inserted epenthetic consonant is not a contrastive segment, and hence, cannot be a direct consequence of phonological segmental markedness constraints. A final problem for markedness accounts involves general phonotactics: consonant epenthesis is often restricted to either word-initial or intervocalic position, though, as a general onset-filling mechanism it is expected in both positions.
In this paper, I suggest that the four problems noted above arise from misguided analyses of consonant epenthesis grounded in universal constraints and markedness principles. Here, I present alternative diachronic explanations for a range of sound patterns involving consonant epenthesis, with a focus on Austronesian languages, illustrating how reference to both segmental markedness constraints and syllabic markedness constraints can be eliminated. Like many other common sound patterns, regular consonant epenthesis may have a natural history, reflecting the phonologization of earlier phonetically conditioned sound change, or an unnatural history, involving rule inversion, rule telescoping, analogy, or language contact. The failure of universalist models to distinguish these diachronic origins results in the range of local problems for treatments of consonant epenthesis noted above, as well as non-local problems in phenomena in which the same set of universal constraints are claimed to play a role.
|