Chauncey et al., 2009; Christoffels et al., 2007; Hoshino & Thierry, 2011;
Strijkers et al., 2010; Verhoef et al., 2009a, 2009b). In the section that follows,
we will review in detail three of these recent studies which integrated ERPs
into classical paradigms that have often been used in past behavioral research
on bilingual production: language switching/mixing, simple picture naming,
and picture-word interference.
Time-course of Language Selection in Bilingual
Speech Production
As stated above, ERPs reflect, in a fine-grained manner, the cognitive
processes involved from the time when a stimulus (e.g. a picture) is presented
to the time when a response (e.g. speaking a word) is made. The ERPs’ exqui-
site temporal resolution allows us to investigate the stages of production at
which language and lexical selection processes are influenced by the activa-
tion of the non-target language by combining ERPs with traditional behav-
ioral paradigms.
A first ERP study on bilingual production using an immediate naming
paradigm was a language switching/mixing experiment with cognates and
noncognates (Christoffels et al., 2007). In this study, German-Dutch bilin-
guals undergoing ERP recoding named cognate and noncognate pictures
overtly in their L1 only, in their L2 only, or in either of their languages based
on the color of the picture. The bilinguals named cognate pictures faster than
noncognate pictures in the L1 blocked (i.e. when naming pictures in L1 only),
the L2 blocked (i.e. when naming pictures in the L2) conditions and in the
mixed condition. Furthermore, the magnitude of cognate facilitation was
greater for L2 than L1 in the blocked conditions but vice versa in the mixed
condition. These results suggest that the L1 is more permeable to the activa-
tion of the non-target language when the activation of the two languages is
required, whereas the activation of the two languages has little impact on L2
naming. In other words, the L1 is normally active when speaking in the L2.
The cognate effect was also observed in the ERPs from 275–375 ms and from
375–475 ms,^3 which suggests that the non-target language is activated to the
level of phonology and that language selection does not occur until 400 ms
after the onset of the stimulus presentation or even later. According to the
meta-analysis on speech planning by Indefrey and Levelt (2004), phonologi-
cal encoding would occur between 275 and 400 ms when speakers produce
a word in their native language. The time windows in which the cognate
effect was observed by Christoffels et al. (2007) were therefore in line with
the time estimate by Indefrey and Levelt (2004), even though their partici-
pants were bilingual.
Another recent study combined immediate naming with a picture-
word interference paradigm to examine the locus of language selection
210 Part 5: The Bilingual Brain