Page 36 - Fall 2011

PHONETICS OF ENDANGERED LANGUAGES D. H. Whalen Speech-Language-Hearing Program Graduate Center of the City University of New York 365 Fifth Avenue New York, New York 10016 and Haskins Laboratories 300 George Street, Suite 900 New Haven, Connecticut 06511 and Endangered Language Fund Christian T. DiCanio Haskins Laboratories 300 George Street, Suite 900 New Haven, Connecticut 06511 Patricia A. Shaw First Nations Languages Program University of British Columbia 1866 Main Mall Vancouver, British Columbia V6T 1Z1, Canada The world is filled with an astounding array of languages, 6,909, by the count of the Ethnologue (Lewis, 2009). Most of these use an acoustic signal as the main ele- ment in signal transmission, though vision affects speech even for typically hearing individuals (e.g., McGurk and MacDonald, 1976); sign languages (126 are listed in Lewis, 2009) use the visual channel almost exclusively. The acoustic signal for speech is powered mostly by the larynx and shaped by the vocal tract. Because human popula- tions have essentially the same anatomy, there is a great deal of similarity in the sounds that languages use. However, there is an impressive range of variability as well. The largest survey of sound systems (Maddieson, 1984), for example, lists no sound that occurs in all languages, even though broad patterns are seen. The number of significant sounds, or phonemes, ranges from about 12 (Pirahã, Rotokas) to over a hundred (!Xóõ), and the mechanisms used vary greatly as well. The acoustics of speech have proven to be extraordinar- ily complex. Early estimates that simple acoustic pattern matching would make automatic speech recognition practi- cal (e.g., Juang and Furui, 2000) proved to be wrong. Current high levels of recognition are founded on acoustic analysis of huge amounts of data combined with statistical inference about common co-occurrences among words and sounds (e.g., Jelinek, 2009). Understanding what it is that listeners are sensitive to in this complex acoustic signal has been fruit- fully guided by examining how those sounds are generated in the vocal tract (e.g., Iskarous et al., 2010). For sounds in less- er-studied languages, having articulatory data to help inter- “Documenting differences among the world’s most disparate languages is of central importance to the field of linguistics and to the language community's heritage.” pret the acoustic signal is even more valuable. A few of the world’s languages have been well-studied, but most of them have yet to be explored in any detail, if at all. Pioneering efforts by Peter Ladefoged and Ian Maddieson to record the sounds of the world's languages resulted in descriptions of many of the less typical sounds used (Ladefoged and Maddieson, 1996) and phonetic sketch- es of several languages (e.g., Ladefoged et al., 1998; McDonough et al., 1993; Silverman et al., 1995; Taff, et al., 2001). Funded for many years by the National Science Foundation, this work has provided an invaluable basis for further pho- netic work, given that it provides an initial understanding of the expected production and acoustic bases of virtually all the sounds that are used in language. These descriptions are far from complete, however, as can be seen from two trends. First, we are still learning new and important facts about even the best-known languages, including English, as evidenced by the continuing appear- ance of phonetic studies in the pages of scientific journals. Second, even though the “same” sound may be described in phonetic studies from different languages, we can not assume that it shares the same characteristics across languages. For instance, ejective fricatives occur in several languages (Maddieson et al., 2001), but they vary widely in how they are produced. Such wide variability calls into question the valid- ity of these phonetic categories across languages. There is much left to learn. Phoneticians have begun stepping up their efforts to study endangered languages while there are still fluent speak- ers left, especially those who acquired the language as their Phonetics of Endangered Languages 35