[Author Login]
Title:Augmentation as Affixation in Athabaskan Languages
Authors:Sharon Hargus, Siri G. Tuttle
Abstract:Augmentation as Affixation in Athabaskan Languages

Sharon Hargus and Siri G. Tuttle

University of Washington

The well-known disyllabicity requirement for Athabaskan verbs has

been variously analyzed as satisfaction of a disyllabic verb template

(Slave, Rice 1990), satisfaction of a monosyllabic prefixal portmanteau

consituent (Navajo, McDonough 1990, 1995), or the result of stray

consonant syllabification (Ahtna, Causley 1994). In this paper, we

present data from other, less-well known languages of the family

(Witsuwit'en, Tanana, Sekani, Deg Hit'an) which suggests a different,

family-wide analysis of augmentation phenomena. Some of the languages

we discuss contain disyllabic verb stems and/or verb stems with

syllabic prefixes. The fact that even these forms undergo augmentation

indicates that augmentation is independent of base syllable count. We

propose that the augment is a tense prefix whose phonological

properties follow largely from its vocalic shape (a reflex of Proto-

Athabaskan *@) and its particular location in the verbal position class

template. The theoretical interest of our simple morphological

analysis lies in what it does not include: no disyllabic template, no

special prefixal 'stem' morphemes, and no unusual phonological domains.

The systematic existence of monosyllabic verbs in some of the

languages presents the greatest challenge for any account of disyllabic

minimality, including the present one. In our account of monosyllables

in one of the languages, we are forced to tackle a longstanding problem

in Athabaskan phonologies, namely, a problematic set of onset/coda

alternations associated with certain verbal prefixes. We trace the

exceptional behavior of these prefixes to a small set of prosodic

subcategorization constraints which these morphemes obey. Our analysis

relieves Athabaskan verb prefix phonology of much prior complexity, so

that the truly phonological constraints required are just members of

the normal constraint arsenal. Our analysis thus supports Generalized

Alignment in particular and Optimality Theory in general: for the

first time in Athabaskan linguistics, these previously problematic

alternations can be analyzed entirely within the limits of the assumed

theoretical framework.

Our analysis affirms Anderson's (1996) and Potter's (1996) proposed

accounts of position class morphology as a ranked set of Alignment

constraints. Because the constraints which regulate affix order, as

well as the prosodic subcategorization constraints mentioned above, are

individually ranked, it is possible to encode within the bounds of the

theory some of the morpheme-specific phonological phenomena which have

bedeviled previous analyses. Consequently, our analysis has uncovered

evidence that prosodic constraints may dominate morphological

constraints even in a language family without reduplication or

prosodically-governed infixation.
Type:Paper/tech report
Article:Version 1