Natasha Warner's current and recent research

• Production and processing of reduced speech. The main focus of the Douglass Phonetics Lab at this point is reduced speech, the kind of "sloppy" speech one hears all the time in normal conversation and many other speech styles.  We are examining both acoustics of reduced speech and perception and psycholinguistic processing of it. We have been working on several sub-projects in this area. Daily-life casual speech is filled with rampant reduction and variability (sounds and syllables disappearing, voiceless stops appearing as voiced approximants, vowels appearing as voiceless fricatives, etc.), and yet the vast majority of speech research examines careful, non-reduced speech, often words produced in isolation. This is not the type of speech most of us process during most of every day. We are studying reduced speech through a large acoustic study of American English flaps, analyzing for example how often flaps surface as approximants rather than true flaps, how much variability within and among speakers there is in reduction, how much reduction is explained by speech style vs. speech rate vs. word frequency, etc. We are also studying processing of reduced speech through phonetic identification, lexical decision, and cross-modal identity priming experiments.  We are beginning to explore how second language learners or listeners from another dialect perceive reduced speech.  Recent and current students of the lab have also examined how listeners process Japanese vowel reduction/devoicing (Naomi Ogasawara) and whether Mandarin tones show reduction in the tone space (Jeff Berry).  Collaborators on this overall project include Ben Tucker (U. Alberta, Edmonton), Amy Fountain (U. Arizona), and Mirjam Ernestus (Radboud U., Nijmegen).
• Revitalization of the Mutsun language. Mutsun was spoken near San Juan Bautista, California, until approximately 1930, when the last fluent speaker died. The Mutsun community has been working on revitalization since 1996. Natasha Warner and a group at the University of Arizona participate by compiling a dictionary from published sources and field notes, and by assisting in the development of language teaching materials. NEH Preservation and Access funding (2005-2007) and a Woodrow Wilson Foundation Public Scholarship Grant (2003) have allowed us to enter all of the existing documentation about Mutsun into databases and analyze it, leading to a much clearer understanding of the language for revitalization purposes. Collaborators: Quirina Luna (Amah Mutsun tribe), Lynnika Butler (University of Arizona), Heather van Volkinburg (Columbia University), and numerous student volunteers.
• The relationship of intonation and speech segmentation. This project investigates whether the pitch rise at the beginning of an accentual phrase in Japanese can serve as a word boundary cue.  A corpus study suggests that it should make a good boundary cue, and a perception experiment indicates that listeners can use this cue for segmentation.  We have also investigated the pitch rise cue in several speech styles (e.g. conversation vs. newscasting). Collaborators: Takayuki Arai (Sophia University, Tokyo), Takashi Otake (Dokkyo University, Tokyo), Lynnika Butler (University of Arizona).
• Navajo spoken word recognition and phonetics.  The spoken word recognition aspect of this project investigates how speakers of Navajo process verbs and nouns with either few or many morphemes, focussing especially on the role of the verbal morphology in recognition. The phonetics aspect of this project examines influences of the first and second language phonologies on phonological categories, where Navajo and English are the first and second languages, in fluent speakers, semi-fluent speakers, and novice learners of Navajo. Collaborators: Mary Willie (University of Arizona), Alina Twist (University of Arizona), James McQueen (Max Planck Institute, Nijmegen).  This project is currently on hold.
• Incomplete neutralization in Dutch. (Completed project) This project investigates the production and perception of neutralization of voicing in final position in Dutch, which appears to involve the maintenance of very small differences in duration (incomplete neutralization). This study also investigates the influence of orthography in creating such sub-phonemic differences. Collaborators: Allard Jongman (University of Kansas), Joan Sereno (University of Kansas), Rachel Kemps (University of Nijmegen).
• Phonology and speech segmentation in Korean. (Completed project) This project investigates listeners’ use of language-specific phonology in segmenting speech in Korean. In this project, we found that segmentation is affected by phonotactic constraints, even though these phonotactic constraints are variable and are created by several interacting aspects of Korean phonology. Collaborators: Jeesun Kim (University of Melbourne), Chris Davis (University of Melbourne), Anne Cutler (Max Planck Institute, Nijmegen).
• Stop epenthesis. (Completed project) This project investigates the production and perception of epenthetic stops between nasals and following obstruents, such as the optional [k] in “youngster.” The project focuses primarily on Dutch. This research has shown that listeners quite often interpret phonetically variable epenthetic stops as real tokens of the stop phoneme, and that listeners are influenced by phonotactic constraints of their language in doing so: they are less likely to report epenthetic stops when the epenthetic stops would violate phonotactic constraints of the language. Collaborator: Andrea Weber (Max Planck Institute).
• Vowel epenthesis. (Completed project) This project uses articulatory data on the production of /l/ before epenthetic schwa in Dutch, and argues that the light/dark /l/ alternation in Dutch contradicts the assumption of Articulatory Phonology that gestures cannot be inserted in the process of speech production. Collaborators: Allard Jongman (University of Kansas), Anne Cutler (Max Planck Institute, Nijmegen), Doris Mücke (University of Cologne).
• Perception of gated Dutch diphones. (Completed project) This project provides a large database on how listeners’ perception of sounds changes over time as the acoustic signal unfolds. This information is being used as input to a new Bayesian model of spoken word recognition, Shortlist-B. Collaborators: Roel Smits (formerly Max Planck Institute, Nijmegen), James McQueen (Max Planck Institute, Nijmegen), Anne Cutler (Max Planck Institute, Nijmegen).