A visible cue for CVC-final consonant voicing

[Ted Moallem <ted.moallem@xxxxxxxxx>]

Dear List,

I'm wondering if anyone can point me to references concerning a
particular visual speechreading phenomenon that I've encountered in my
research, developing a speech aid the Deaf and hearing-impaired.

As part of the study, my subjects performed a pairwise consonant
discrimination task under two sensory conditions:  lipreading-alone
and lipreading+speech-aid.  As expected, in the lipreading-alone
condition, corresponding voiced and unvoiced consonants (e.g., /b/-/p/
or /d/-/t/) are virtually indistinguishable in CVC-initial context.

However, in CVC-final context, most subjects discriminate consonant
voicing well above chance, just through lipreading (unaided).
Performance is far from perfect, but d' > 1 is the norm, and a few
subjects (both hearing-impaired and normal-hearing) consistently score
d' â 2.

My working theory is that the duration of the vowel (or of the CVC
utterance as a whole) is providing a visible cue that is correlated
with CVC-final consonant voicing.  Vowel duration as an acoustic cue
for final consonant voicing is well-known (e.g., House and Fairbanks,
1953; Denes, 1955; House, 1961; Raphael, 1972; Mack, 1982; Luce and
Charles-Luce, 1985).  But I cannot seem to find a paper that describes
or even refers to the visual correlate.

I assume that visual cues for final consonant voicing are
well-characterized somewhere in the lipreading literature.  Perhaps
I'm using the wrong search terms.  In any case, I'd appreciate the
benefit of your collective expertise.

Thanks in advance!

Ted


-- 
________________________________

Theodore Moallem, Ph.D.
moallem@xxxxxxx
646-872-0283

http://www.BlindLead.org
--
Sensory Communication Group
Research Laboratory of Electronics @ MIT
Harvard-MIT Division of Health Sciences and Technology