Page Piccinini (UCSD) will be giving the following talk at 10AM on Thursday, October 31, in Herter 301.
Accessing Cross Language Categories in Learning a Third Language
Current theories differ greatly in explaining how bilinguals organize
their two languages, including at the sound level. The heart of the
debate is whether bilinguals have constant access to all of their
sounds across their two languages, or only access to sounds from one
of their two languages at a time. The present study examines these
theories by testing the ability of early Spanish-English bilinguals to
access phonetic distinctions within the voice onset time (VOT)
continuum that exist across their two languages (negative, short-lag,
and long-lag VOT). To this end, bilinguals were tested on a third
language that has all three contrasts phonemically: Eastern Armenian.
The effects of both language mode and language dominance were
examined. One production and two perception tasks were carried out. In
the production task participants heard words and nonce words in
Eastern Armenian and were told to repeat the words back to the best of
their abilities. Participants produced the three types of stops
significantly differently both for bilabials and velar. However this
was modulated by language dominance, with those who were more English
dominant producing less of a contrast between the negative and
short-lag categories (the phonemic categories in Spanish). There was
no effect of language mode. The first perception task was an AX
discrimination task. Participants heard two words and said if the
words were the same or different. Participants were consistently good
at negative versus long-lag VOT, but did poorly at both negative
versus short-lag VOT and short-lag versus long-lag VOT. There was no
effect of language mode or language dominance. The second experiment
was an ABX discrimination task. Participants heard three words and
said if the third word was the same as the first or second word.
Participants performed best at negative versus long-lag VOT, also
preformed well at short-lag versus long-lag VOT, but performed poorly
at negative versus short- lag VOT. However, there was an interaction
with language dominance such that those who were more balanced
bilinguals preformed better at the negative versus short-lag VOT
contrast than those were more English dominant. Language mode was not
significant. These results support a theory whereby language dominance
and not language mode is a key determining factor both in speech
production and perception. Furthermore, the results of the AX and ABX
experiments together suggest that while more balanced bilinguals can
accurately perceive the three way contrast when forced to assign
category labels (the ABX task) they do poorly when categorization is
open (the AX task). This may show a preference by participants for a
language with a two-way contrast, as would match their native
languages, even if they can produce and perceive a three-way contrast.
