There are two major themes to my research: the phonetics of endangered languages and the production and perception of tone and phonation type. My work on endangered languages has focused on explaining language-specific phonetic patterns and examining their broader implications for phonetic theory and phonological structure. This research includes, for instance, the examination of gemination contrasts in Trique, the study of laryngeal timing in Ixcatec glottalized consonants, and work on the coarticulation between laryngeal segments and tone in Trique. At Haskins Laboratories, I am part of a project investigating how corpora from language documentation projects can be used for phonetic corpus work. We are working with corpora from three languages: Yoloxóchitl Mixtec, Arapaho, and Tarahumara (Rarámuri). Using forced alignment (automatic speech-to-text recognition), we hope to be able to largely automate the process of segmentation of language recordings. This will allow us to more easily extract speech sounds for acoustic analyses and expand the utility of documentation materials.
The second area of my research concerns the phonetics of tone and phonation type contrasts. In terms of speech production, I am interested in questions related to phonetic alignment and coarticulatory overlap. In particular, I am investigating how external prosodic constraints, such as syllable structure, word size, and speech rate, interact with pitch targets in tone production. I am currently investigating this topic both in my work on Yoloxóchitl Mixtec tone production and in my work on rate-induced tonal coarticulation in Itunyoso Trique. Yoloxóchitl Mixtec is notable for its large tonal inventory and large number of tonal combinations surfacing on disyllabic couplets. Approximately 21 patterns surface on monosyllabic words and 28 on disyllabic words. These combinations create a complex set of dynamic tonal patterns, each which has a unique surface realization conditioned by word size and the presence of glottalization. We are currently investigating the effect of these conditioning contexts on the alignment of pitch targets using a controlled data set from 10 speakers. The preliminary results show a strong sensitivity of pitch inflection points to moraic structure and a dominant pattern of pitch-raising which expands the tonal space when pitch targets overlap with glottalization. The results from this investigation will be presented at the 2012 SSILA meeting in Portland. My work on Trique tonal oarticulation examines findings which argue that speakers often approach a ceiling in the maximum rate of F0-change during tone production (Xu and Sun, 2002).
In terms of speech perception, I am interested in the categorization and discrimination of word-prosodic contrasts as well as the weighting of perceptual cues in auditory discrimination. I have done cross-linguistic work examining the categorical perception of Trique tone by Trique and French listeners as well as work on the weighting of perceptual cues in Trique laryngeal contrasts. The methods used in this first study closely followed those used in a similar investigation on Taiwanese Mandarin and French done by Hallé, Chang, and Best (2004). The goals of this study were to generally understand the effect of language experience on pitch perception and to examine cross-linguistic differences in the categorical perception of tone. Trique listeners showed evidence for categorical perception of 5/8 tonal pairs, but behaved similarly to French listeners discriminating others. Tonal discrimination depended on both psychoacoustic distance and the number of phonetic cues which distinguished the tonal pairs. Contra Burham et al. (1996) and Hallé et al. (2004), French listeners were better at tonal discrimination than Trique listeners. Speaking a tone language may not confer an advantage at pitch discrimination in a non-linguistic task.
For the study on the perceptual cues of laryngeal contrasts, I investigated the role of duration, H1-H2, and laryngeally-induced pitch changes in the perception of laryngeal rime contrasts in Trique. Previous work on speech production showed these three cues to be relevant to distinguishing the contrast. A labelling task was carried out which manipulating three independent phonetic cues (vowel duration, pitch, and spectral tilt) used to distinguish between these rime types. The results show duration to be weighted more heavily than the other cues for both the /Vː/-/Vʔ/ and the /Vː/-/Vɦ/ contrasts, followed by H1-H2. Pitch is also perceptually relevant to listeners, though only for the /Vː/-/Vɦ/ contrast. Relatively few cues were used by listeners to identify /Vʔ/ stimuli. Where duration is neutralized, listeners showed greater sensitivity to subordinate cues. The accuracy of discrimination from a baseline lexical discrimination task corresponds with the number of acoustic cues used to distinguish the laryngeal contrasts. Where fewer cues are available, lexical discrimination is poorer. The results of this work are currently in preparation.
Fieldwork on Otomanguean Languages
I began doing fieldwork on Oto-Manguean languages in 2001, when I spent a semester abroad in Oaxaca, Mexico studying Yojóvi Zapotec. Since then, I've devoted most of my energies to work on the complex phonological and prosodic systems of this language family.
Since 2004, I have done fieldwork on San Martín Itunyoso Trique both in Oaxaca, Mexico, and in the central valley of California. I am primarily concerned with the phonetics and phonology of the language, but have investigated other aspects of the language too. In particular, I have done some work on the complex enclitic morphophonology in the language, and am working on a dictionary. I co-authored a short Trique dictionary with my collaborator/consultant Benigno Cruz Martínez.
In 2009, I began to collaborate with Michael Swanton and Denis Costaouec on Ixcatec, a moribund Oto-Manguean language spoken in Santa María Ixcatlán, Oaxaca, Mexico. With 8 remaining speakers, most of whom are elderly, we are working as a team to develop a comprehensive grammar, set of texts, and documentation for the language. My principle interest in this language right now is understanding the phonetics and phonology of the segmental and tonal contrasts. There are a number of rules of tonal spreading, tonal neutralization, and an extensive process of tonal downstep. This project is funded by the Hans Rausing Endangered Language Program.
In Fall 2010, I began collaborating with Jonathan Amith and Rey Castillo on Yoloxóchitl Mixtec, an little studied Oto-Manguean language spoken in San Luis Acatlán, Guerrero, Mexico. We are working together to write a very comprehensive grammar exploring all aspects of the language's phonetics, phonology, morphology, and syntax. This project is funded by the National Science Foundation's Documenting Endangered Languages (DEL) Program.