15, p=0.299. For the semantic task, there were tone×antpost, F(2, 32)=8.55, p=0.003, and tone×lat, F(2, 32)=4.67, p=0.027, interactions. High tone was more positive than low tone in frontal, F(1, 16)=12.16, p=0.003, and central, F(1, 16)=12.84, p=0.002, RoIs ( Fig. 1C). The effect size was larger over mid, F(1, 16)=15.55, p=0.001, η2=0.493, and right, F(1, 16)=11.84, p=0.003, η2=0.425, than over left electrodes, F(1, 16)=5.66, p=0.030, η2=0.261. For the lexical word boundary task, a tone×antpost

interaction was seen, F(2, 32)=5.98, p=0.010. High tones produced more positivity at frontal, F(1, 16)=6.34, p=0.023, and central, F(1, 16)=22.59, p<0.001, leads ( Fig. 1D). There was no significant effect for delexicalized speech, F(1, 16)=1.55, p=.231. In the semantic task ERPs, there was a tone×suffix interaction between 400 and 550 ms following suffix onset, F(1, 16)=4.63, SB431542 ic50 p=0.047. High tone-inducing suffixes produced increased positivity as compared to low tone-inducing suffixes following low stem tones ( Fig. 1E), F(1, 16)=5.84, p=0.028, but not following high stems, F(1, 16)=0.01, p=0.921. There were no significant effects for the lexical or delexicalized word boundary check details tasks. The negativities

for high tone-inducing suffixes preceding and following the positivity were not significant. It has been hypothesized that the early stages of prosodic processing are reflected in variations in the N1 and P2 components. N1 increase is thought to show detection of salient auditory features that might be relevant for speech processing, whereas a P2 increase would index allocation of anticipatory attention to upcoming grammatical information cued by the prosodic features. The present study tested the ERP effects of high and low word-stem tones in Central Swedish. As previously found, high stem tones increased the P2 amplitude as compared to low

tones. Crucially, however, this was not the case for the delexicalized EGFR antibody inhibitor versions of the same stimuli. This finding supports the hypothesis that the P2 effect indexes allocation of attention to upcoming grammatical information – in this case, high stem tone-associated suffixes – which was not available in the delexicalized stimuli. The fact that the P2 effect was also present in the lexical boundary task blocks, where the stem tone was irrelevant for the task, might suggest that the P2 in fact indexes “passive” anticipatory attention. It should be noted that native speakers are often unconscious of the existence of high and low stem tones in Swedish. This is similar to the case of left-edge boundary tones, where the P2 has been argued to show passive anticipatory attention to upcoming main clause structures. The P2 onset was further found to be rather early, around 160 ms rather than the 200 ms onset previously reported. This is most likely due to the more exact tone onset and thereby earlier tone processing in the present study.