The Dene (Athabaskan) languages are a group of closely related North American indigenous languages. They are the most geographically widespread language family in the Americas, spoken in Alaska, across the Northern Canadian provinces, and in the Southwestern USA (Navajo). Other than Navajo, the Dene live in small, isolated and inaccessible communities, even today.
Navajo is one of the best documented indigenous languages in the Americas in terms of the grammar (Young and Morgan, 1980, 1987, 1992), the phonology and the documentation of phonetic patterns (McDonough, 1990, 1999, 2001, 2003). Moreover, recordings of Dene speakers from several communities in the Mackenzie Basin made at the University of Alberta, Edmonton, over the past 6 years are available. Phonetic illustrations of phonemic inventories are available for the three of the four Mackenzie Basin Dene language groups, North Slavey, Dogrib and Dene Sųłiné, online in the Dene Speech Altas (McDonough 2012). Finally, comparative studies of Navajo and these languages are being undertaken, including a comparative study of lexical tone.
Dene people have maintained a distinct culture and remarkably stable languages with striking similarities, seemingly resistant to outside influences and borrowings. They share a typologically unusual and complex morphological structure and phonemic inventory, and appear to maintain a consistency at the phonetic level in the articulatory features of their phonemes (McDonough and Wood 2008).
The tonal split in Dene/Athabaskan languages
The family has one significant division which is geographically determined. The languages in the Mackenzie River Basin and the American southwest (Navajo and Apache) have lexical tone. Those in Alaska, the Yukon and the Pacific coast are toneless. The toneless languages have retained glottal sonorants in their stem codas; the tone languages lack glottal sonorants in this position. The Athabaskan tonogenesis hypothesis (Krauss 2005) argues for the genesis of tone from the incorporation of the laryngeal gesture into the nucleus of the. Several aspects remain unexplained: First, both high and low tone developed from this incorporation (Kingston, 2005). However, a tonal ‘reversal’ process appears to have occurred more than once, well after the initial loss. Neighbouring communities have opposite tones on similar lexical items (Figure 4-6).
Verbs are likely to play an important role in understanding tone. The minimal Dene verb is a disyllabic unit of two separate domains, a verb stem and a pre-stem conjugational morpheme (Young and Morgan 1987; McDonough 1999, 2003). All current Dene tone languages inflect verb stems for aspect; tonal alternations are part of that inflection (Krauss, 2005; Leer, 1999). Furthermore, the pre- stem complex is very rich; over a dozen possible morpheme positions have been proposed. How tone developed in non-stem morphemes is not well understood. Tone in the pre-stem domain appears to be structurally distinct from the tone in stems. For instance, is it not uncommon for morphemes in the pre- stem domain to be termed ‘underlyingly toneless’ (Young and Morgan 1987). Finally, crucial to the present project, until quite recently, there has been a widespread lack of phonetic documentation for these languages.
One hypothesis suggests that the original contrast was stem laryngealisation, rather than tone. In this view, tonal contrasts arose due to the interaction of tonal alignment and tonal asymmetries in the pre-stem complex responding to the distinct laryngeal specifications of the stem. Evidence for this is the existence of a major boundary in the verb word between the stem and the pre-stem domains in both verb and noun forms (McDonough 1999). Despite the diachronic consistency in their grammars and lexicons, a great deal of prosodic variation has been observed across the group. The documentation and modelling of tonal specification within a closely related group with near identical tonogenesis patterns is important to theories of tone, tonal alignment and language change.
Tone-intonation interaction in the Dene/Athabaskan languages
Distinctly different intonational systems have been observed among the Dene languages. For Navajo, McDonough (2002) found no discourse related pitch events associated with either declarative vs. yes/no questions or in focus constructions, though intonational effects are clearly audible in the northern Dene languages. Holton (2005), writing on intonation in Tanacross, observed that these differences across the Dene languages may be due to the interaction of tone and intonation. Tanacross has intonational right edge tones, which however do not completely override, or neutralise, the lexical tone on the final syllable. Interestingly, Holton points out that in the Dene family, Tanacross is somewhat unique in this regard. On the basis of impressionistic observations of the final syllable in declaratives, he argues that in Gwich’in, the intonational contour dominates the lexical tone. In Han, the converse is true in this position: lexical tone takes precedence over intonation. “In looking at the Athabaskan languages as a whole, it seems very possible that observed differences in pitch patterns may be due not so much to differences between tonal or intonational patterns, but may rather be attributable to differences in the way tone and intonation interact in the individual languages.” (Holton 2005:275). These differences (cf. Hyman’s 2011 ‘avoidance’ and ‘submission’) call for a model of the interplay between lexical tone and intonation which incorporates gradience in the relative impact of tones from each source on the pitch contour