Title: Sound structure in speech production: A comparative study of Fijian and Berber languages

Funding: Social Science and Humanities Research Council

Research Team: John Alderete (PI, Simon Fraser University), Alexei Kochetov (University of Toronto), Stefan A. Frisch (University of South Florida), Abdelkrim Jebbour (formerly at Ibn Tofail University)

Description: Speaking is a complex behavioral task. It involves selecting words from an immense mental lexicon, ordering them in well-formed sentences, and articulating their sounds in fluid speech. Dominant models of speech production assume speaking is accomplished with a richly structured mental lexicon; a web of interconnected words and the sounds inside them. Word retrieval in these models involves searching this “word web” and selecting the correct sounds in the right order. While this approach has shed light on the nature of the human mind, the research on which universal claims are being made has relied overly on data from Indo-European languages. Important world languages such as Mandarin Chinese and Arabic, and many lesser-known languages, have received scant attention, despite findings that suggest these languages reveal important structural differences in speaking patterns. This team-based project will carry out a detailed study of three little-studied but structurally significant languages, Fijian, Berber, and Cantonese, in order to explore what contributions these languages can make to our understanding of the cognitive capacities for speech.

Speech production studies analyze the structure of speech sounds—including syllables, sound frequencies, and the internal relationships between sounds—in order to understand the structural foundations of language. Fijian, Cantonese, and Berber are ideal languages in which to study speech production because they differ strikingly from Indo-European languages in sound structure, and represent opposite poles in the two key areas of syllables and consonants. Fijian and Cantonese both have a highly reduced syllable inventory and simple consonant systems. Berber, by contrast, has very complex syllable inventory and consonant systems. Our team will explore whether these differences mean that these languages have different word webs, and what impact these unique word webs may have on how native speakers produce speech. These cross-linguistic investigations will shed light on the possible existence of different forms of word webs, and will further develop our understanding of word webs and their place in human cognition.

An international team of researchers and students will combine linguistic, computational, and experimental methods in a comparative study. First, we will build a database of the words that make up Fijian and Berber languages. Second, we will create a computational word web, building a precise formal model of the internal dynamics of sound retrieval in speaking. Computational modeling is the best way to answer our questions about sound structure because it provides detailed, quantitative predictions about how speaking unfolds in real time. The speech patterns predicted by the computational model will then be tested for validity by conducting fieldwork with native speakers of Fijian, Cantonese, and Berber, collecting slips of tongue from natural speech, and running experiments that probe speech planning in word webs.

Results from our comparative study will be used to enhance knowledge about the complexities of speech production. Dissemination will occur in national and international academic venues. Our team will also produce two extensive web-based lexical databases, and a suite of computer programs for use in mapping word webs and speech production in diverse languages. We will mobilize the accumulated research findings by translating their insights into aligned areas outside of academia, including language technology, pedagogy, child language development, and language deficits. This knowledge mobilization plan extends deeply into Fijian, Cantonese, and Berber communities, contributing linguistic studies to their larger community project of language retention and revitalization.