Effects of Type of Agreement Violation and Utterance Position on the Auditory Processing of Subject-Verb Agreement: An ERP Study

Abstract

Previous ERP studies have often reported two ERP components-LAN and P600-in response to subject-verb (S-V) agreement violations (e.g., the boys (*) runs). However, the latency, amplitude and scalp distribution of these components have been shown to vary depending on various experiment-related factors. One factor that has not received attention is the extent to which the relative perceptual salience related to either the utterance position (verbal inflection in utterance-medial vs. utterance-final contexts) or the type of agreement violation (errors of omission vs. errors of commission) may influence the auditory processing of S-V agreement. The lack of reports on these effects in ERP studies may be due to the fact that most studies have used the visual modality, which does not reveal acoustic information. To address this gap, we used ERPs to measure the brain activity of Australian English-speaking adults while they listened to sentences in which the S-V agreement differed by type of agreement violation and utterance position. We observed early negative and positive clusters (AN/P600 effects) for the overall grammaticality effect. Further analysis revealed that the mean amplitude and distribution of the P600 effect was only significant in contexts where the S-V agreement violation occurred utterance-finally, regardless of type of agreement violation. The mean amplitude and distribution of the negativity did not differ significantly across types of agreement violation and utterance position. These findings suggest that the increased perceptual salience of the violation in utterance final position (due to phrase-final lengthening) influenced how S-V agreement violations were processed during sentence comprehension. Implications for the functional interpretation of language-related ERPs and experimental design are discussed.

Keywords: ERPs; auditory modality; subject-verb agreement; type of agreement violation; utterance position.

Figure 1

Example of images used for the verb cook/cooks.

Figure 2

Representative waveforms and spectrograms illustrating the time-locking point used for ERP analysis; (A) illustrates the inflected verb (cooks) and (B) the uninflected verb (cook). The dotted arrow indicates the stop closure of the oral-stop coda /k/ and the solid arrow indicates the end of stop closure that was used as the time-locking point in grammatical and ungrammatical experimental conditions.

Figure 3

Approximate placement for the electrodes included in the regions of interests (ROI) analysis for MANOVA. The rectangles indicate the levels used to demacate the nine ROI [anterior midline (Fz, FCz), central midline (Cz, CPz), posterior midline (Pz, POz), anterior left (F7, F5, F3, FT7, FC5, FC3), central left (C3, C5, T7, CP3, CP5, TP7), posterior left (P7, P5, P3, PO7, PO5, PO3), anterior right (F4, F6, F8, FC4, FC6, FT8), central right (C4, C6, T8, CP4, CP6, TP8), posterior right (P8, P4, P6, PO4, PO6, PO8)].

Figure 4

Grand average ERP waveforms for grammatical and ungrammatical conditions across positions and type of agreement violation at the F3, Fz, F4, C3, Cz, C4, P3, Pz, and P4 electrodes and the topographic maps of the significant ERP effects. The first row of the figure shows the anterior electrodes while the second row shows central electrodes and the third row shows the posterior electrodes. The ERPs are time-locked to the offset of the verb-stem (end of stop closure) and positivity is plotted upwards. The topographic maps show brain voltage distributions for the negative and positive clusters. These maps were obtained by interpolation from 64 electrodes and were computed by subtracting the grand averages of grammatical from the ungrammatical conditions. Electrodes in the significant clusters are highlighted with a black circle and the F3, Fz, F4, C3, Cz, C4, P3, Pz, and P4 electrodes in the significant clusters are highlighted with a white circle. Time-windows for significant clusters is highlighted in gray over the waveforms.

Figure 5

Difference in mean amplitude between grammatical and ungrammatical conditions in the utterance-medial and utterance-final position across the 9 ROIs, showing error bars representing +1/−1 standard error.

Figure 6

Grand average event-related potentials elicited by errors of omission (red) and correct verb (blue) in medial position. Gray bar highlights the significant time-window for the P600 effect.

Figure 7

Grand average event-related potentials elicited by errors of commission (red) and correct verb (blue) in medial position. Gray bar highlights the significant time-window for the P600 effect.

Figure 8

Grand average event-related potentials elicited by errors of omission (red) and correct verb (blue) in final position. Gray bar highlights the significant time-window for the P600 effect.

Figure 9

Grand average event-related potentials elicited by errors of omission (red) and correct verb (blue) in final position. Gray bar highlights the significant time-window for the P600 effect.