EEG data was averaged across all trials, split by participant, electrode and condition. We then computed the mean amplitude of brain activity every 50ms into the epoch (from -200 to 1000ms from the onset of the appearance of the target on-screen). Preliminary analysis examined the onset of differences between conditions using these 50ms slices. Based on this preliminary analysis of significant differences between conditions in both L1 and L2 speakers, we chose to focus on the time-window from 400 to 600ms from the onset of the appearance of the target word on-screen, i.e., the typical N400 time window. Further analysis examined participants’ brain activity across frontal [F3, F4], fronto-central [FC1, FC2], central [C3,C4], centro-parietal [CP1, CP2], parietal [P3, P4] and parieto-occipital[PO3, PO4] electrode sites in both hemispheres. Data are reported using the Greenhouse Geisser correction where required.
We first conducted a 2×2×2×2×6 repeated measures omnibus ANOVA for the data averaged across 400-600ms with the within-subjects factors repetition (1st vs. 2ndoccurrence of the target within a text), word (real, pseudo), context (biasing, neutral), hemisphere (left, right), region (frontal, fronto-central, central, centro-parietal, parietal, parieto-occipital), and group as a between-subjects factor. The analysis revealed a near-significant main effect of context, F(1, 32)=2.95, p=.095, ƞp2
=.08, a significant interaction between repetition*word, F(1, 32)=6.98, p=.013, ƞp2
=.18, repetition*context, F(1, 33)=5.05, p=.032, ƞp2
=.14, word*context, F(1, 31)=15.06, p=.001, ƞp2
=.32, context*hemisphere, F(1, 32)=13.32, p=.001, ƞp2
=.29, word*context*hemisphere*group, F(1, 32)=14.06, p=.001, ƞp2
=.31, and near-significant interactions between repetition*word *context *hemisphere*group, F(1, 32)=3.62, p=.06, ƞp2
=.1, and repetition*word*context*hemisphere*region*group, F(5, 160)=2.24, p=.053, ƞp2
=.073. Based on the interactions reported above, especially the interactions repetition*word, repetition*context and word*context*hemisphere*group, we decided to split the analysis into eight smaller repeated measures ANOVAs based on the repetition of the target in the text (1st or 2nd), the type of word (real or pseudo) and hemisphere (left, right). The smaller ANOVAs were, therefore, computed separately for each repetition of the target in the text based on whether the target was a real word or pseudo word, separately for left and right
3 We also found a significant main effect of region, and significant interactions between repetition*hemisphere*
group*word*hemisphere*group, word*context*hemisphere*group, repetition*region, word*region, repetition*word*region, repetition*context*region, word*context*region and hemisphere*region*group, (ps<.05). Given the interactions between the critical factors reported in the main text, we do not report these interactions in detail.
hemisphere electrodes, and retained the within-subject factors context (biasing, neutral), region (frontal, fronto-central, central, centro-parietal, parietal, parieto-occipital) and the between-subjects factor group. A preliminary repeated measures ANOVA for the L2 learners’
group with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor native language revealed no main effect of native language or any interaction with native language within the L2 group of participants, p>.1, suggesting no influence of the L2 learners’ native language on their processing of real and pseudo words in texts.
First occurrence of a real word in biasing vs. neutral context
Across left hemisphere electrodes, a repeated measures ANOVA with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor group revealed a significant main effect of context, F(1, 32)=22.78, p<.001, ƞp2
=.42. There were no other significant main effects or interactions. Especially, there were no significant interactions with group. Across right hemisphere electrodes, a repeated measures ANOVA with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor group revealed a significant main effect of context, F(1, 32)=26.54, p<.001, ƞp2
=.45. There were no other significant main effects or interactions. Especially, there were no significant interactions with group. Nevertheless, to ensure that the effects were similar across native speakers and L2 speakers, we performed planned post-hoc tests within each group of participants, comparing brain activity across condition. Native speakers and L2 speakers alike showed a more negative deflection in brain activity 400ms after the onset of the appearance of the target in the neutral context condition relative to the biasing context condition across both left, native: t(16)=3.84, p<.001, d=.84, L2: t(16)=2.96, p=.009, d=.52, and right hemisphere electrodes, native: t(16)=4.38, p<.001, d=1.24, L2: t(16)=2.79, p=.013, d=.52. These differences suggest that both native speakers and L2 learners find it easier to recognise the target word, therefore, showing reduced brain activity upon encountering the target word, when it is embedded in a biasing context relative to a neutral context.
Figure 1 plots the mean difference in the amplitude of brain activity to real words in biasing and neutral contexts, separated according to whether the data represents the first or second repetition of the target word and according to language group4.
Figure 1. Mean difference in the amplitude of brain activity between biasing and neutral contexts in the real word condition.
Second occurrence of a real word in biasing vs. neutral context
Across left hemisphere electrodes, a repeated measures ANOVA with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor group revealed a near-significant interaction between region and context, F(5, 160)=2.65, p=.058, ƞp2
=.08. There were no other significant main effects or interactions. Especially, there were no significant interactions with group. Across right hemisphere electrodes, a repeated measures ANOVA with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor group revealed a near-significant main effect of context, F(1, 32)=3.39, p=.075, ƞp2
=.1, and a significant interaction between region and context, F(5, 160)=4.58, p=.003, ƞp2
=.13. There were no other significant main effects or interactions.
4 Note, that in keeping with the previous literature, this effect appears earlier in native speakers relative to L2 readers (e.g. Ardal et al.; Weber-Fox & Neville, 1996; Hahne, 2001; Moreno & Kutas, 2005) indicating that L2 readers may need more time for target processing compared to native speakers. However, to ensure that we were analysing a similar time-period across both groups of participants, we retained the analysis comparing the 400 to 600ms window across both groups. Analysis using the earlier time-window for the native speakers reveals very similar results.
Especially, there were no significant interactions with group. Planned post-hoc tests within each group of participants found no significant difference in brain activity following the target word in the neutral context relative to the biasing context in native speakers across both left, t(16)=-.016, p=.9, and right hemisphere electrodes, t(16)=.57, p=.5. In contrast, L2 speakers showed a more negative deflection in brain activity in the neutral context relative to the biasing context across right hemisphere electrodes, t(16)=2.22, p=.041, d=.36, but not across left hemisphere electrodes, t(16)=1.51, p=.15. Combining data from both hemispheres, this effect was restricted to frontal, t(16)=3.12, p=.007, d=.04, fronto-central, t(16)=2.34, p=.03, d=.35, and central, t(16)=2.15, p=.047, d=.37, electrode sites. These differences suggest that only L2 readers differ in their recognition of the target in neutral and biasing contexts when presented with the target for the second time in the same text. Native speakers do not show any difference in their ease of recognition of the target word across biasing and neutral contexts when they have encountered this target word before in the same text.
First occurrence of a pseudo-word in biasing vs. neutral context
Across left hemisphere electrodes, a repeated measures ANOVA with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor group revealed a significant interaction between region and context, F(5, 160)=3.03, p=.04, ƞp2
=.09.
There were no other significant main effects or interactions. Especially, there were no significant interactions with group. Across right hemisphere electrodes, a repeated measures ANOVA with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor group revealed near-significant interactions between region and context, F(5, 160)=2.54, p=.06, ƞp2
=.07 and between region, context and group, F(5, 160)=2.71, p=.053, ƞp2
=.07. There were no other significant main effects or interactions.
Especially, there were no significant interactions with group. Planned post-hoc tests confirmed there were no significant differences in brain activity to the first presentation of the pseudo-word target in biasing and neutral context in native speakers or L2 speakers (ps >.14).
Analysis of the data grouped by hemisphere, similarly, revealed no significant differences in brain activity to the pseudo-word target in biasing and neutral context in native speakers or L2 speakers (ps >.2).
Figure 2 plots the mean difference in the amplitude of brain activity to pseudo-word targets in biasing and neutral contexts, separated according to whether the data represents the first or second repetition of the target word and according to language group.
Figure 2. Mean difference in the amplitude of brain activity between biasing and neutral contexts in the pseudo-word condition.
Second occurrence of a pseudo-word in biasing vs. neutral context
Across left hemisphere electrodes, a repeated measures ANOVA with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor group revealed a significant main effect of context, F(1, 32)=4.86, p=.035, ƞp2
=.13. There were no other significant interactions with region or group. Across right hemisphere electrodes, a repeated measures ANOVA with the within-subjects factors context (biasing, neutral) and region (see above) and the between-subjects factor group revealed no significant main effects of context or significant interactions with context and group. Nevertheless, planned post-hoc tests revealed a near-significant difference in the brain activity to the second presentation of the pseudo-word target in biasing and neutral contexts across left hemisphere electrodes, t(16)=-1.98, p=.065, d=.46, in native speakers but not in L2 speakers, t(16)=-1.23, p=.23.
There was no difference between conditions across right hemisphere electrodes for both native speakers and L2 speakers (ps >.15). Further tests per region revealed that the difference between brain activity to biasing and neutral contexts in native speakers was restricted to left frontal, t(16)=-1.88, p=.078, d=1.22, left fronto-central, t(16)=-2,12, p=.050, d=2.13, and left
central electrode sites, t(16)=-2,20, p=.043, d=.88. Across all three regions, there was a more negative deflection in brain activity to the second presentation of the pseudo-word target in biasing contexts relative to neutral contexts, i.e., the opposite direction of effects relative to the real word condition. Planned paired samples t-tests for the L2 learners group revealed no significant results for either hemisphere and any of the electrode sites, p>.2.
Figures 3 and 4 plot brain activity from -200ms before the appearance of the target on-screen to 650ms after the appearance of the target on-on-screen separated by language group, hemisphere and repetition of the target word in the text aggregated across frontal, fronto-central and fronto-central electrode sites. Figure 3 plots the data for real word targets while Figure 4 plots the data for pseudo-word targets.
Figure 3. Mean amplitude of electrical activity from -200ms before the appearance of the target on-screen to 650ms after the appearance of the target on-screen separated by language group, hemisphere and repetition of the target word in the text aggregated across frontal [F3, F4], fronto-central [FC1, FC2] and central [C3, C4] electrode sites for real word targets.
Figure 4. Mean amplitude of electrical activity from -200ms before the appearance of the target on-screen to 650ms after the appearance of the target on-screen separated by language group, hemisphere and repetition of the target word in the text aggregated across frontal, fronto-central and central electrode sites for pseudo-word targets.