Skip to main content

Quantitative Analysis of Intonation in an Endangered Language

Intonation is the meaningful melody of spoken language. Consider, for example, the contrasting melodies in the following two sentences: "What do you want?" and, "Do you want tea?" In English, intonation marks phrase and clause boundaries; sentence types, e.g., yes-no questions vs. statements; topic focus; and new vs. old information. In spite of how important intonation is for conveying meaning and its apparent universality in language, there is no generally accepted approach for investigating the intonational structure of a language.

The recent development of computerized instruments now allows researchers to measure the acoustic properties of recorded speech, including the fundamental pitch (f0), which is the number of vocal cord vibrations per second. Many variables affect f0 that are neither syntactic nor semantic, including individual speaker variation and the physical limitations of the human vocal system. To avoid these effects, many researchers, relying on expert judgment, select individual f0 measurements on which to base their analyses. In contrast we use all the f0 measurements. Our goal is to explicitly model and adjust for these nuisance variables, teasing out the intonational signal.

We approach intonation as a relationship between f0 and a set of explanatory variables related to syntax and semantics. Our task is to measure the syntactic and semantic effects on f0 in the presence of the other variables affecting f0. Our data consist of recordings from 16 speakers of two dialects of Unungam Tunuu (Aleut), an endangered Alaskan language and member of the Eskimo/Aleut language family. It is predicted that within one generation, 50% of the world's 6000+ languages will be gone. Thus, at this time it is crucial to gain a deeper understanding of these languages--both for linguists, who study the scope and parameters of the human language faculty, and for small language communities, who want to preserve knowledge for future language learning.

Our work has appeared in the Journal of Phonetics (2001) and at conferences of the Linguistic Society of America and the Acoustical Society of America.