Advances in the Study of Bilingualism

intended language without using the co-activated non-target language. In
attempting to elucidate the nature of these mechanisms, much of the
research on bilingual speech production has been conducted with a variety
of behavioral paradigms (see Kroll et al., 2006; Kroll et al., 2008 for reviews)
and, more recently, with neuroscience methods such as functional magnetic
resonance imaging (fMRI) and ERPs. The number of cognitive neuroscience
studies on bilingual language processing has been increasing dramatically
for the past 10 years (see Kotz, 2009; Van Heuven & Dijkstra, 2010, for
reviews on cognitive neuroscience research on bilingual comprehension).
The application of these methods allows us to examine the time-course and
locus of language processing in bilinguals in a fine-grained manner.
Although fMRI provides insights into the functional anatomy underlying
language production in bilinguals, ERPs provide temporal information with
a resolution compatible with the time-scale of neural function, that is in the
millisecond range.
First we review models of lexical access in bilingual production in the
context of psycholinguistic research and then discuss how ERPs can con-
tribute to the debate over the locus of language selection by studying a few
demonstrative ERP studies in detail. ERPs reflect cognitive processing
from the time when a stimulus is presented to the time when a response
is made and beyond. Therefore, they are suitable to address the question
of when bilinguals select the language of production if both languages are
active.

Lexical Access in Bilingual Speech Production

Considering that both languages are active in bilinguals regardless of
their intention to speak in one language only, research on bilingual speech
production has been focusing on when and how bilinguals select their
target language (see Costa, 2005; Kroll et al., 2006, 2008 for reviews). Figure
9.1 shows an illustrative model of bilingual word production adapted from
Poulisse and Bongaerts (1994) and Hermans (2000). In this figure, the
intention to speak a word in one language (represented by the language
cue) is insufficient to selectively activate lexical alternatives in the intended
language alone; lexical alternatives from both languages are therefore
simultaneously active. For example, when a Spanish-English bilingual
intends to name the pictured object ‘cat’ in English, this will result in an
activation of not only semantically related English lexical representations
(, ), but also their Spanish translation equivalents (,