Page 41 - Fall 2011

Fig. 6. Effect of total stop duration on glottal timing strategy. Long stop data (left), Short stop data (right). (P = preaspiration, S = simultaneous timing, V = voicing during closure). Figure from DiCanio (in press). Reprinted with permission. preaspirated stops, as in Icelandic (Helgason, 2002). Finally, voicing may cease during the stop closure duration. This is called partial voicing or voice ter- mination time (V.term). Figures 5a-5c show these three oral-glottal alignment configurations, respectively. In DiCanio (2012), acoustic and EGG recordings were made from four speakers producing short and long stops in Itunyoso Trique. These words were presented in contexts so that the duration of stop closure could be examined from the acoustic signal along with the alignment of devoicing from the EGG signal. The results found significant differences between the long and short stops in relation to the timing of devoicing. Long stops are produced with either simultaneous glottal timing (49.4%), as shown in Fig. 5a, or with devoicing prior to closure (43.8%), as shown in Fig. 5b. Short stops are typically produced with voicing which extends into the stop closure (84%), as shown in Figure 5c, and rarely with simulta- neous oral-glottal alignment (14.1%). The amount of voicing during closure for the short (lenis) stops varied in relation to the over- all duration of the stop, shown in Fig. 6. No such variation was observed for long (fortis) stops. Research on related languages like Zapotec is inconclusive as to whether short stops are voiced or voiceless (Avelino, 2001; Nellis and Hollenbach, 1980). These results sug- gest that such variability is conditioned by within-category changes in conso- nant duration. The effect of such vari- ability is that, for some short stops with particularly short duration, voicing may extend through the entire stop. This pattern is called passive voicing (Jansen, 2004; Westbury and Keating, 1986). Long stops show a different pat- tern. They are actively devoiced by an abrupt glottal spreading gesture, the timing of which does not vary with consonant duration. This work addresses a general descriptive question within Itunyoso Trique: what is the phonetic realization of short and long stops in the language? However, the answer to this question also informs a more general theory of speech production. Voicing is generally considered to be a discrete category within the phonology of a language. Sounds are typically classified as either “voiced,” with vocal fold vibration, or “voiceless,” without vocal fold vibra- tion. The Trique data suggest that for certain sound types, there is a continu- um of voicing that varies due to dura- tional differences. Findings like this illustrate one of the ways in which descriptive phonetic work on endan- gered languages has a broader impact on the linguistic sciences. Summary We have presented just two of the many phonetic investigations currently under way that examine the world’s extensive, but shrinking, variety of lan- guages. Documenting differences between the world’s most disparate lan- guages is of central importance to the field of linguistics and to the language community’s heritage. Such efforts are funded by such governmental agencies as the National Science Foundation and the Administration for Native Americans, and non-profits like the Endangered Language Fund. While the work presented here focuses mainly on the production of consonants, many more aspects of speech acoustics need to be investigated. For instance, prosody (speech timing, intonation) varies substantially across languages; yet, this topic is rarely addressed in 40 Acoustics Today, October 2011