electro-oculogram channels (VEOG) set above and below the right eye. Ocular
artefacts were mathematically corrected when the variance of the eye blink
model was below 0.005. Subsequently, EEG files were visually inspected for
remaining artefacts and contaminated periods were manually excluded.
Continuous recordings were cut into epochs starting 100 ms before the onset
of critical (adjective or noun) stimuli and ending 1000 ms after. Baseline correc-
tion was performed in reference to pre-stimulus activity and individual averages
were digitally re-referenced to the global average reference. Finally, individual
ERPs were averaged together in each condition to obtain grand-averages.
Peak detection was carried out automatically, time-locked to the latency
of the peak at the electrode of maximal amplitude on the grand-average ERP.
To detect modulations in the N2 range, mean ERP amplitude was measured
in the 260–360 ms time window, determined on the basis of global field
power inspection. Based on the topography reported previously for the N2
component the set of electrodes selected for amplitude analysis was: FC1,
FC2, FCZ, FZ. Mean ERP amplitudes were subjected to repeated measures
analyses of variance (ANOVAs) with congruence (match/mismatch) and
speech-part (adjective/noun) as the primary factors of interest.
Results and Discussion
Results
The P1/N1 complex was not affected by the experimental factors. The
first significant differences were found in the N2 range in response to the
first critical word.
The N2 elicited by the first critical word (adjective or noun in first posi-
tion) in all conditions was significantly modulated by factors speech-part
(adjective –noun) (F(1,32) = 5.909, p = 0.021) and congruence (match/^
mismatch) (F(1,32) = 38.841, p < 0.001). There was also a significant interac-
tion between speech-part, congruence and group (F(1,32) = 8.117, p = 0.008),
whereas there was no interaction between speech-part and group
(F(1,32) = 0.943, p = 0.339) and no interaction between congruence and
group (F(1,32) = 0.694, p = 0.411). Follow-up analyses, performed to break
down this interaction, showed a main effect of congruence (F(1,32) = 26.505,
p < 0.001), but no congruence by group interaction (F(1,32) = 1.104, p = 0.301)
when adjectives were in first position. The N2 component elicited by mis-
matching adjectives was more negative-going than that elicited by matching
adjectives. There was a main effect of congruence in the same conditions for
monolinguals (F(1,17 ) = 30.542, p < 0.001) and for bilinguals (F(1,15) = 5.763,
p = 0.03), indicating that there was greater negativity in the mismatch condi-
tion than in the match condition in both groups.
In short, in adjective-first conditions, we found a difference in the N2
range between the Go and No-Go conditions. Specifically, the N2 component
224 Part 5: The Bilingual Brain