Chilcotin vowel flattening and sibilant harmony: Diachronic cues to a synchronic puzzle, Gunnar Olafur Hansson

Among the Athapaskan languages, Chilcotin is unique in that pharyngealization plays a central and very active role in its phonological system (Krauss 1975; Cook 1983, 1993; Andrews 1988). Chilcotin has two separate processes that involve pharyngealization or "flattening": Vowel Flattening (VF) and Sibilant Harmony (SH). The latter is similar to the harmony systems found in many other Athapaskan languages, such as Navajo or Tahltan, but unusual in that it involves a pharyngealization opposition ("flat" vs. "sharp") rather than place of articulation (alveolar vs. alveopalatal, dental vs. alveolar, etc.). VF, on the other hand, consists in "flat" consonants (pharyngealized sibilants, uvulars) causing the lowering and/or backing of nearby vowels according to a highly complex pattern.

Unlike SH, Chilcotin VF is unique within the Athapaskan language family. Interestingly, though, pharyngealized consonants are found in many of the neighbouring Salish languages, and these too cause flattening of nearby vowels to some extent. For this reason, Chilcotin flattening is often taken to be "an areal feature being shared by Interior Salish, particularly Shuswap, Lillooet, and Thompson" (Cook 1993:150). Nonetheless, simply attributing Chilcotin flattening to areal influence would be insufficient as an explanation, since Salish does not display any of the complexities of Chilcotin VF. For example, there is a striking asymmetry between leftward and rightward flattening, and a further asymmetry between flattening triggered by sibilants (S-flattening) and that triggered by uvulars (Q-flattening). In short, leftward S-flattening is special in being unbounded, and in being unaffected by any intervening consonants. By contrast, all the other subtypes of VF are limited in that they rarely reach beyond the immediately preceding or following vowel, and that they can be blocked by an intervening "sharp" consonant (velar or non-pharyngealized sibilant). These asymmetries have proven particularly challenging for synchronic phonological analyses (see Cook 1993), and cannot be reduced to areal influence from Salish.

In this paper I argue for a historical scenario for the development of VF in Chilcotin, and suggest that its emergence is directly dependent on the presence of an inherited SH system in the language. Cook (1993) already observed the similarities between SH and leftward S-flattening: (i) both proceed from right to left; (ii) both are unbounded, applying to the entire word; and (iii) both are unaffected by the quality of any intervening consonants and vowels. In the scenario proposed here, these similarities are due to a direct historical relationship between the two phenomena.

First of all, I assume that the contrast between the two series involved in SH, the PA *dz and *dZ/*dzw series, developed into a (post)dental vs. alveolar contrast in pre-Chilcotin, as in many other Northern Athapaskan languages. In Chilcotin, the dental vs. alveolar contrast was substituted for by a pharyngealized vs. non-pharyngealized contrast, and it is precisely this development that may well be due to areal influence from Salish languages (which lack a dental/alveolar contrast). As a result, SH came to involve pharyngealization rather than a place-of-articulation feature. As in other languages with pharyngealized consonants, the Chilcotin flat consonants have profound effects on the quality of vowels in neighboring syllables; in fact, the pharyngealization contrast is perceived not on the sibilants themselves but on the quality of the surrounding vowels (Krauss 1975; Cook 1993). This local effect can still be seen in rightward S-flattening, and in both types of Q-flattening, and it is safe to assume that at an earlier stage, even leftward S-flattening was of this strictly-local character. The reason why leftward S-flattening is special in modern-day Chilcotin, I propose, is due to its interaction with SH. In forms with more than one sibilant, the combined effect of SH and local S-flattening – from the point of view of the listener – was that a flat sibilant typically gave rise to flattening of all preceding vowels in the word. I suggest that this triggered an analogical restructuring of the process, whereby the flattening was attributed directly to the rightmost (flat) sibilant, rather than indirectly (through local flattening superimposed onto a right-to-left sibilant harmony). As a result of this reinterpretation, flat sibilants came to have an unbounded leftward flattening effect even in forms with no additional sibilants, i.e. in forms where SH did not apply.

The proposed analysis has several advantages. First, it explains the asymmetric (and synchronically problematic) behaviour of leftward S-flattening, and the similarities it shares with SH. Second, the analysis explains why SH and leftward S-flattening coexist in the language, in spite of the considerable overlap in their effects (cf. also the fact that SH seems to be disappearing altogether, judging by the findings of Andrews 1988). Third, it sheds some light on how SH in Chilcotin fits into the greater picture of SH elsewhere in Athapaskan. Finally, the analysis gives a more concrete and precise definition of the sense in which the pharyngealization phenomena of Chilcotin can be attributed to contact with Salish.

Cook, Eung-Do. 1983. Chilcotin flattening. Canadian Journal of Linguistics 28:123-32

Cook, Eung-Do. 1993. Chilcotin flattening and autosegmental phonology. Lingua 91:149-74.