Research

My research investigates systematic variation in the sound properties of speech and the perceptual and communicative consequences of such variation. I am particularly interested in ways in which that variation is constrained or conditioned by communicative or representational factors.

Some questions that drive my work:

• How do communicative factors condition phonetic variation in speech production?
  • What are the properties of confusability-conditioned (especially neighborhood-conditioned) phonetic variation?
  • How do the characteristics of a listener (e.g., a child, someone hard of hearing, a non-native speaker) affect the phonetic properties of speech?
  • What are the perceptual consequences of this variation?
• What constitutes confusability? What contributes to communicative load or listener difficulty?
• What constitutes hyperarticulation? What phonetic properties are manipulated in speech designed for clarity?
• How do patterns of coarticulation (especially nasal coarticulation) vary within and across languages?
  • How does coarticulation interact with contrastive phonetic features?
  • What are the acoustic cues to nasality?
  • What role does sub-phonemic variation (e.g., coarticulation) play in speech perception?

Ongoing projects

(1) Previous work (mine and others') investigates the production of neighborhood-conditioned phonetic variation: we know that English speakers produce words from dense phonological neighborhoods with increased hyperarticulation and coarticulation. Current projects investigate neighborhood effects in French, alongside other clear speech effects, looking at coarticulatory nasality as well as contrastive vowel nasality in this more-dimensional vowel space.

(2) We are also investigating the perception of this neighborhood-conditioned phonetic variation. I am interested in exploring whether the increased hyperarticulation and coarticulation in high neighborhood density words are perceptually beneficial (and potentially listener-directed), compensating for the inherent difficulty of high ND words. We are currently looking at the role of degree of nasal coarticulation and hyperarticulation in the perception of high and low neighborhood density words.

(3) To support work on the perception of linguistic nasality, I am interested in identifying and manipulating acoustic cues to nasality.

(4) In joint work with CU colleague Bhuvana Narasimhan, we are looking at the realization of information structure cues in child-directed speech. We know that people produce speech with hyperarticulated vowels and exaggerated pitch in speech to children. In this work, we are investigating whether they modulate these acoustic properties in ways that are sensitive to information structure (new vs. given, focus) and confusability to make child-directed messages as easy to understand as possible. We are further interested in whether these child-directed cues differ in speech to neurotypical vs. autistic children.

Past work

• In work with Kuniko Nielsen (Oakland U), I have been interested in phonetic imitation. We have described imitation effects in nasal coarticulation (along with Georgia Zellou, UC Davis). Our work also explores whether imitation targets specific acoustics or whether imitation is more linguistically interpreted. We have looked particularly at imitation of both speaker-specific f0 patterns and linguistically-conditioned f0 patterns.
• I have worked on description of contrastive and coarticulatory nasality in Lakota, a Siouan language spoken in the upper plains of the US. With Georgia Zellou (UC Davis), Armik Mirzayan (USD), and David Rood (CU), we looked at how these two roles for nasality interacted in shaping the patterns produced by native speakers.
• Work with Georgia Zellou looked at listener-directed speech adjustments for infants, taking into account a novel measure of lexical difficulty - age of acquisition. We explored the relationship between age of acquisition (or the age at which a child is reported to have learned a given word) and neighborhood density in conditioning adjustments in hyperarticulation and coarticulation in infant-directed speech.
• Work with Pat Keating (UCLA) and colleagues used facial point tracking and articulator tracking to identify visual correlates of prosody.