Linguistic aspects of voice quality with special reference to Athabaskan

1. Introduction Many languages of the world use differences in voice quality or phonation type to contain linguistic functions. Thus, in certain languages, lexical items are differentiated solely on the basis of differences in voice quality. For example, in Jalapa Mazatec, an Otomanguean language spoken in Mexico (Kirk et al. 1993), the three words /ja! / meaning 'tree', /jaª ! / meaning 'he wears' and /ja0 ! / meaning 'he carries' differ only in the voice quality associated with the vowel. In the first word, the vowel is characterized by modal voicing, the default type of voice quality found in all languages of the world and involving regular, periodic vibration of the vocal folds. The second word contains a breathy voiced vowel, which is marked by a greater aperture of the vocal folds relative to modal voice (Ladefoged 1971, Laver 1980). The increased opening of the vocal folds allows more air to pass through the glottis and creates turbulence not present with modal phonation. Finally, the third word has a creaky vowel, which involves a more tightly adducted glottis than either modal or breathy voice (Ladefoged 1971, Laver 1980). This laryngeal setting gives the auditory impression of a " rapid series of taps, like a stick being run along a railing " (Catford 1964:32). Within the Athabaskan language family, although voice quality is not known to contrast in vowels, many languages distinguish between modal voiced and creaky voiced sonorants. For example, in Hupa, the word /xoN) / 'fire' with a final creaky voiced velar nasal contrasts with the word /xoN/ 's/he' containing a final modal voiced velar nasal. In addition to this (somewhat limited) use of distinctive voice quality, in most, if not all, Athabaskan languages, non-modal voice qualities arise as predictable variants of modal phonation in certain contexts. For example, vowels adjacent to phonemic glottal stop are often realized with a creaky voice quality immediately adjacent to the glottal closure. Similarly, vowels in the vicinity of phonemic /h/ also typically display some breathiness. This paper examines some of the linguistic uses of different voice qualities, in particular, the two most common types of non-modal voice, breathy and creaky voice, and their phonetic realization in a number of languages, focusing on the Athabaskan languages, Hupa and Western Apache. The organization of the paper is as follows. Section 2 discusses some of the phonetic correlates of voice quality differences, and how …