Athapaskan Linguistics: Proto-Athapaskan Phonology

Abstract

ly more tg cw kw abstractly tyw kyw With this scheme, in which the newly established obstruents fit symmetrically, Krauss (17) showed how these four sets of obstruents merge in ways in which other obstruents, including the velars, do not. The y-feature (whatever phonetic characteristics it may represent), which is shared by the four sets of obstruents, suggests that the obstruents constitute a natural class. Tharp (32) first suggested that the *kw series that Krauss postulated for PA is coronal, i.e. not a pre-velar, but dorsopalatal. Krauss now believes that *kw was pre-velar in PAE, but changed to palatal (i.e. coronal) in PA. I shall return to this question shortly. The interesting point is that the four sets of PA obstruents have maintained a remarkable structural relationship as shown by the pattern of historical developments. The coronality of the *k series (and *kw series) is also indicated by internal reconstruction. In 1978 I (6) postulated a voiced dorsopalatal fricative (Y = y) for Sarcee purely on the basis of internal phonological alternations. It has turned out that this dorsopalatal which derives from *y (i.e. ,*k series) triggers the sibilant assimilation in the same way the palatal sibilants do. More recently, in a manuscript which is still being revised, Krauss (20) is more explicit about the phonetics of PAE *kw and its PA counterpart, and he suggests that the interpretations of *ts, *ts, *k, and *kw as tY, tYw, kY, and kYw respectively were once true for PAE, but were no longer true for PA. Viitso [as reported by Krauss (18)] and Tharp (32) observed the inadequacy of these interpretations primarily because it is *ts, but not *kW, that has developed into kw in Dogrib and p(f) or kw in some Slavey dialects, all of which can be explained in terms of flattening or rounding (labialization). A more striking case for this type of development is seen in Chilcotin, in which alveolars and velars oppose phonemically in terms of sharp and flat (pharyngealized) features, ts vs ts k vs q (see 4). Interestingly, flat alveolars develop from *ts, and its sharp counterparts develop from the merger of *kw and *t? series, e.g., This content downloaded from 157.55.39.112 on Wed, 07 Sep 2016 04:24:01 UTC All use subject to http://about.jstor.org/terms ATHAPASKAN LINGUISTICS 257 *ts > ts (flat): *ts'an > ts an 'bone', *sad? > zi 'mouth' *tsi9 > tsi 'head'. *8 > ts (sharp): *tsa > tsae 'beaver', *tsu > tsu 'grandmother' *kw > ts *xwun > zun 'good', *gwe-n > dzin 'day' *sWa > sae 'sun'. The above Chilcotin data provides some support for Viitso's interpretation of *ts, *ts, *k, and *kw as tYw, tY, kYw, and kY, respectively,3 where the w-feature is considered to be [+flat] in the Jakobsonian framework (13). Now the development of the flat alveolar sibilant in Chilcotin from *ts, that of the Tsetsaut labial consonant from *kW, and that of the retroflex obstruents in Alaskan languages are all attributable to the w-feature, i.e. [+flat]. On the other hand, the merger of *ts and *kw (tY and kY a la Viitso) is explainable in terms of the loss of opposition between two points of articulation. Krauss, however, still prefers his own abstract interpretation of the four sets of obstruents to those of Viitso's, taking a more concrete approach to the development of labials from the *ts series. He believes the labials have been developed from *ts through an intermediate stage, namely an interdental, i.e. ts > tO > pf. But a similar interpretation is unlikely for the development of kw (Dogrib, Slavey) from *ts, i.e. there is no reason to believe that kw has gone through an intermediate stage, ts or kW. Nor does this pushchain type of interpretation make sense for the development of Chilcotin ts (flat sibilant series) from *ts. While this problem remains unresolved, Krauss now transcribes PA reflex of PAE *kw as [tsw] or [ts'w ] which can be abstractly interpreted as tywr. This heavily marked segment is obviously intended to account for such a labialized alveolar palatal as Hupa t?w on the one hand and a retroflex alveopalatal as Kiowa-Apache tsr on the other. In terms of the Jakobsonian distinctive feature framework, both the w-feature and the r-feature are considered to be [+flat], hence these two phonetic features are mutually exclusive realizations of the same distinctive feature. While it is generally accepted that labialization (rounding) has to be independent of pharyngealization, since both may be phonemic in one language, i.e. labialization and pharyngealization are not mutually exclusive with respect to a given language, it has not been explicitly questioned, at least to my knowledge, whether or not these two features are mutually exclusive with respect to a given segment. Chilcotin velars seem to suggest that these features are not mutually exclusive with respect to a given segment either: q (as in -qan 'husband') 3This interpretation is reported by Krauss (18), but I have not been able to see Viitso's article. This content downloaded from 157.55.39.112 on Wed, 07 Sep 2016 04:24:01 UTC All use subject to http://about.jstor.org/terms