during the process of speech planning in the L2 in a context where no
instruction or stimulus was presented in the L1 (Hoshino & Thierry, 2011).
In this study, Spanish-English bilinguals were asked to name pictures in
their L2 English while ignoring visually presented L2 English distractor
words. Similar to the seminal picture-word interference study with bilin-
guals by Hermans et al. (1998), a ‘phono-translation’ condition in which
distractor words were phonologically related to the translation of the pic-
ture name (e.g. celery for the picture ‘brush’ whose Spanish translation is
‘cepillo’) were included to evaluate the time-course of cross-language acti-
vation. The results replicated Hermans et al. (1998) behaviorally and pro-
vided ERP insights into phono-translation interference. Specifically, the
Spanish-English bilinguals named pictures presented with phono-transla-
tion distractor words more slowly than those with unrelated distractor
words. The phono-translation effect was also observed in the mean ampli-
tude of the N2 peak (200–260 ms) and the N3 range (350–400 ms), but not
between 400–450 ms. These results suggest that alternative candidates
from the non-target language are active and compete for selection until 400 ms
and that language selection does not occur until after 400 ms. The time-
course of cross-language activation, which was indexed by the phono-
translation effect, and the locus of language selection were compatible with
the findings of Christoffels et al. (2007).
Another source of converging evidence for cross-language activation
from ERP research with immediate naming paradigms is a study by
Strijkers et al. (2010). Strijkers et al. tracked the time-course of lexical access
in bilingual word production millisecond-by-millisecond by manipulating
frequency (high vs. low) and cognate status (cognate vs. noncognate). In
this study, Spanish-Catalan and Catalan-Spanish bilinguals named pic-
tures in their L1 Spanish or in their L2 Spanish, respectively. The logic of
the study was to identify the locus of frequency and cognate effects in
simple picture naming. Both bilingual groups named pictures with high
frequency names and cognate pictures faster than pictures with low fre-
quency names and non-cognate pictures. The frequency and cognate
effects were also detectable in the mean amplitude of the P2 (second posi-
tive peak, 160–240 ms in L1; 180–260 ms in L2), N3 (third negative peak,
240–320 ms in L1; 260–350 ms in L2), and P3 (third positive peak, 320-
420 ms in L1; 350–450 ms in L2). Furthermore, the ERP elicited by pictures
with high frequency names and that elicited by pictures with low frequency
names started to diverge 172 ms after picture onset in the L1 group (Spanish-
Catalan) and 184 ms in the L2 group (Catalan-Spanish). Similarly, the ERP
elicited by cognate pictures and that elicited by noncognate pictures started
to diverge at 200 ms in the L1 group and at 184 ms in the L2 group. More
critically, P2 mean amplitude differences were significantly correlated with
differences in naming latencies between high frequency or cognate items
and low frequency or noncognate items (between 160-240 ms in L1 and
Language Selection During Speech Production in Bilingual Speakers 211