Doctoral Dissertation Research: The interaction between tone, prosody and accent

University Of Chicago, Chicago IL

Investigators

Abstract

The world's languages have traditionally been divided into one of three categories (stress-accent, pitch-accent, or tone) based on how pitch is utilized to distinguish words from one another. However, this division of languages is based largely on the impressionistic judgments rather than on detailed quantitative measurements of linguistic properties. Recent work has demonstrated that this typology is deeply inadequate for characterizing both the precise role(s) of pitch in language, and the relevant sources of diversity among languages. For example, although a language may utilize pitch as the primary acoustic cue to differentiate between words, it may also use stress, suggesting that these two phenomena are not mutually exclusive. The prevailing typology, though arguably flawed, has formed the basis for many important claims about the nature of language and linguistic complexity, which in turn have shaped how linguistic pedagogies are developed for use in foreign language teaching, and the way that prosodic systems are modeled for developing tools such as speech recognition software. Given the importance of such resources in today's global society, it is important for linguists to form a clear picture of the ways in which the acoustic properties of speech are used by speakers both in production and in perception. The current project aims to take a closer look at pitch, accent and prosodic structure in an effort to better understand how these structural elements interact, with the end goal of developing a more accurate linguistic typology, and a more adequate set of tools for language teaching and natural language processing. Research for this project will take place in Southwestern Cameroon, focusing on the Bamileke languages, a subgroup of the Mbam-Nkam branch of the Benue-Congo group (Niger-Congo family). These languages have some of the most complex tone systems in the world, and much remains to be understood about their structure. Recent work suggests that word- and phrase-level accentual properties reminiscent of those found in prototypical stress languages are responsible for previously unexplained tonal phenomena in these languages, and that accent may play a much more important role in the organization of sounds than was formerly thought. This research will endeavor to explore in detail the way that acoustic measures such as pitch, vowel quality, amplitude, syllable and vowel duration interact in language production, and the way that these cues are weighted by listeners when perceiving speech. Data will be collected in the form of controlled production and perception experiments, as well as in the form of natural conversation between native speakers of the language. Research findings will contribute to our understanding of the ways in which pitch can be used alongside other acoustic cues at different levels of grammatical structure in a language with lexically-based tonal contrasts. Since many of the languages deemed 'critical' for learning by the US government also fit this description (Mandarin Chinese, Punjabi, Hausa, etc.), this work will have important implications for a wide variety of national language training initiatives.