Research Projects

Here are descriptions of some of the research projects that I'm currently working on.

Cross-linguistic studies on speech perception

One key observation at the intersection of speech perception theory and phonology is that language experience influences speech perception. This project seeks to understand (1) what aspects of speech perception adapt to experience and (2) how pervasive this adaptation may be. Answers to these questions bear on speech perception theory, so that models need to be able to account for both cross-linguistic regularities in speech perception and for language-specific aspects of perception. Similarly, the linguistic theory of phonology and language sound change may also be impacted by our understanding of the maleability of speech perception.

This project has grown out of a very productive and influential (for me) collaboration with Beth Hume (OSU Linguistics). In that collaboration we edited a couple of books, and wrote several papers:

The Role of Speech Perception in Phonology (Hume & Johnson)

The Impact of Partial Phonological Contrast on Speech Perception (Hume & Johnson)

A model of the interplay of speech perception and phonology (Hume & Johnson)

A cross-linguistic study of place perception (Hume et al.)

Adaptive dispersion in vowel perception (Johnson)

Huang Tsan wrote her dissertation on the perception of tone and tone sandhi by speakers of Mandarin Chinese and American English. One thing that I think is very interesting about her studies is that the influence of language on perception seems to be present in very low-level psychoacoustic processing.

Also, be sure to check out Steve Winters' research on place perception and place assimilation processes. Including his PhD thesis Empirical investigations into the perceptual and articulatory origins of cross-Linguistic asymmetries in place assimilation

ViC - Variation in Conversation.

Mark Pitt (OSU Psychology) and I, together with Beth Hume (OSU Linguistics) and Eric Fosler-Lussier (OSU Computer Science), have been working on a large corpus of conversational speech (the speakers are all natives of Columbus Ohio so we sometimes call it the "Buckeye Corpus"). This work is funded by the NIH and resulted in a number of research papers, as well as its own web-page.


No, it isn't a cross between the X-Files and the Mod-squad, it's just a computational model of speech perception that draws on techniques in automatic speech recognition and ideas from mathematical psychology to model human auditory word recognition.  X-MOD uses exemplar-based categorical representations of words, where each exemplar is like a LAFS spectral sequence network (Klatt, 1980).  One of the more interesting things that happens in X-MOD word recognition is that smaller units of linguistic structure - such as syllables and segments emerge in the recognition process - and like word categories, these units are defined simply as sets of exemplars.  Of course, given my interest in processes of speaker normalization in speech perception, X-MOD handles talker differences beautifully.  An early non-DSP implementation was presented in the Johnson & Mullennix volume.  The more recent model, complete with auditory neural spectrograms, is described in a paper in Volume 50 of the OSU Working Papers in Linguistics.

Phonetic studies of the languages of Oklahoma

The goal of this project is to produce acoustic/phonetic descriptions of some of the languages of Oklahoma. Phonetic descriptions of endangered languages are a valuable contribution to the cross-linguistic study phonetics. We also are producing descriptions that can be of value in the design of language teaching materials - returning something to the speakers of the languages.

I also like this project because it gives me opportunities to return 'home' (I am a native Oklahoman) and visit with my family and friends there.

Creek. I am working with several colaborators on this project. Jack Martin (College of William and Mary) is an expert on the Creek language. Together with Margaret Mauldin he is producing a dictionary of Creek. Jack and I have written a paper on the vowels of Creek (download "Vowel reduction in Creek" PDF 128k) and are working on papers on the consonants and tone patterns. Here's a photo of us at Margaret's house.

Cherokee. Marcia Haag (University of Oklahoma) is an expert on syntax and morphology in native American languages.  She has published a dictionary of Choctaw. Durbin Feeling (Northeastern Oklahoma State) is an expert on Cherokee and together with William Pulte produced a famous dictionary of Cherokee. We are working together on a phonetic study of Cherokee - especially the tonal system.

Back to Keith Johnson