S1 (FriAM 1-65) - The Psychonomic Society

Thursday Evening Posters 1008–1013
changes in discrimination and perceived similarity within and across
category boundaries, even when category learning is unsupervised.
The online mixture estimation model (Vallabha, McClelland, Pons,
Werker, & Amano, in press) provides an account of unsupervised
vowel category learning by treating categories as Gaussian distributions
whose parameters are gradually estimated. We extend the model
to pairwise discrimination, defined as the extent to which two stimuli
are members of different estimated categories. We address three findings:
Infants’ discrimination of speech sounds is better after exposure
to a bimodal rather than a unimodal distribution (Maye, Werker, &
Gerken, 2002), infants’ discrimination of vowels is affected by
acoustic distance (Sabourin & Werker, 2004), and perceived similarity
is affected by order of typical versus atypical category exemplars
(Polk, Behensky, Gonzalez, & Smith, 2002). The model also makes
predictions about the development of discrimination during perceptual
learning.
(1008)
Nonsimultaneous Context Effects for Speech-Like Stimuli. JEREMY
R. GASTON, RICHARD E. PASTORE, JESSE D. FLINT, & ANJULI
BOSE, Binghamton University—A well known speech perception
finding is that the physical properties that cue consonant identity can
differ substantially as a function of vowel context. Might vowel properties
alter the accessibility of consonantal cues? Addressing this
question, the present work investigates the effects of vowel formantlike
stimuli on the ability to recognize aspects of speech-like targets.
For both frequency transitions and brief noise bursts, a subsequent
stimulus provided significant backward recognition masking relative
to those same stimuli in isolation, with the amount of interference dependent
on both the target–masker frequency relationship and the
complexity of the masker. The opposite effect, forward recognition
contrast, was found for the temporal reversal of the stimuli. If these
results for speech-like stimuli generalize to speech, they would, at a
minimum, provide a perceptual account for the physical changes in
consonantal cues across vowel contexts.
(1009)
Perception of Cross-Modal Speech Alignment. RACHEL M.
MILLER, KAUYUMARI SANCHEZ, & LAWRENCE D. ROSEN-
BLUM, University of California, Riverside (sponsored by Lawrence D.
Rosenblum)—Talkers are known to produce utterances that partly imitate,
or align to, the speech they hear. Alignment occurs both during
live conversation and when a talker is asked to shadow recorded
words. Recent evidence reveals that talkers also align toward the
speech they see. When asked to shadow lipread words from a speaker,
subjects will produce utterances that sound more like those of the
speaker. Previously, alignment has been determined by naive raters
asked to judge the auditory similarity between a speaker’s and subjects’
utterances. To determine whether alignment is also apparent
across modalities, raters were asked to judge the relative similarity between
a speaker’s visual (video) utterance and two audio utterances of
the subject. Raters judged the speaker’s visual utterance as more similar
to subjects’ shadowed utterances than to subjects’ baseline (read
text) utterances. This suggests that raters are indeed sensitive to the
cross-modal similarity that occurs during alignment.
(1010)
Speeded Choice Responses to Audiovisual Speech. KAUYUMARI
SANCHEZ, RACHEL M. MILLER, & LAWRENCE D. ROSENBLUM,
University of California, Riverside—Speakers are nearly as fast at
shadowing an unpredicted syllable (/ba/, /va/, or /da/; choice condition),
as they are at uttering a single assigned syllable (/ba/; simple
condition) following presentation by a model. One explanation for
these speeded choice reaction times is that listeners are extracting articulatory
information that can serve to prime their own gestures. To
test the articulatory versus auditory nature of this priming information,
an audiovisual test of choice reaction times was conducted.
Speeded choice reaction times were observed for an audiovisually
52
fused syllable (audio /ba/ + visual /ga/, perceived as “da”), suggesting
that priming information does exist at the gestural level. These
findings suggest that the common currency facilitating a speeded production
can take an informational form that exists cross-modally. The
results are also consistent with transcranial magnetic stimulation findings
that show responsiveness in specific articulatory muscles dependent
on audiovisually integrated speech information.
(1011)
Cognitive Consequences of Parallel Language Processing in Bilinguals.
HENRIKE K. BLUMENFELD & VIORICA MARIAN, Northwestern
University—Bilinguals’ activation and control of word networks
across languages was examined using eyetracking and negative
priming. Experiment 1 found that, when listening to words, bilinguals
activated cross-linguistic competitors from sparse, but not dense,
phonological neighborhoods, suggesting that cross-linguistic competitors
from large neighborhoods are inhibited by similar words.
Competition resolution from cross-linguistic networks may place additional
demands on inhibitory control in bilinguals. Experiment 2
tested the hypothesis that bilingual experience modulates cognitive inhibition
mechanisms during auditory comprehension. The results suggest
that monolinguals inhibit within-language competitors to resolve
ambiguity, whereas bilinguals may resolve competition equally efficiently
without the use of overt inhibition. To draw connections between
consequences of bilingualism in the linguistic and nonlinguistic domains,
Experiment 3 compared performance on the eyetracking/
negative priming paradigm with performance on measures of executive
control (linguistic Stroop, nonlinguistic Stroop, and Simon tasks).
Together, results suggest a link between parallel language activation
in bilinguals and executive control mechanisms.
(1012)
The Nature of Memory Representations for Surface Form of Spoken
Language. SABRINA K. SIDARAS & LYNNE C. NYGAARD, Emory
University (sponsored by Lynne C. Nygaard)—This study examined
the nature of memory for surface characteristics of spoken words by
evaluating the effects of indexical and allophonic variation in implicit
and explicit memory tasks. During the study phase for all tasks, participants
were presented with a list of words that varied by speaker and
pronunciation. In the test phase, words were repeated either in the
same or different voice and with the same or different pronunciation.
The results showed that in both explicit and implicit memory tasks,
changes from study to test in pronunciation and voice interacted to influence
memory performance. A memory benefit was found for items
repeated in the same rather than in a different voice but only when pronunciation
stayed the same from study to test. These findings suggest
that the relative specificity of memory representations for spoken language
may depend on the interaction of surface and phonological
form.
(1013)
Cognitive Predictors of Lipreading Ability in Hearing Adults.
JULIA E. FELD & MITCHELL S. SOMMERS, Washington University
(sponsored by Mitchell S. Sommers)—Previous research has
shown extensive individual variability in lipreading ability, even
among participants sampled from homogenous populations. Despite
considerable research on factors that predict individual differences in
lipreading, it remains unclear what traits or abilities underlie the observed
variability in visual-only speech perception. In part, the absence
of reliable predictors of lipreading can be attributed to the use of narrow
sets of predictor variables within individual studies. The present
study therefore examined the ability of a broad range of factors, including
working memory, processing speed, verbal learning, personality,
and perceptual closure, to account for individual differences in
lipreading ability. Multiple measures of each of these factors were obtained
along with lipreading measures for consonants, words, and sentences.
To date, only verbal processing speed has emerged as a significant
predictor of lipreading ability for all three types of stimuli.