Abstract
Introduction:
In language comprehension, the analysis of the grammatical category of words occurs during the very early phase of processing (Friederici, 2011). When the category of incoming words violates syntactic rules, increased negativities are observed compared to grammatical constructions (Early Left Anterior Negativity – ELAN, Friederici et al., 1993); Early Syntactic Negativity – ESN, Hasting & Kotz, 2008). The early latencies of these components have been attributed to the presence of structural predictions (Lau et al., 2006). For example, after hearing “The”, a noun, alone or with an adjective, is expected to build a phrase. Accordingly, the ELAN and ESN effects would reflect a mismatch between the predicted category and the observed ungrammatical one.
Functional studies implicated Broca’s area in syntactic categorical prediction (Bonhage et al., 2015), but causal evidence for this claim is still missing. We tested this hypothesis by simultaneously combining Transcranial Magnetic Stimulation (TMS) and Electroencephalography (EEG) in a two-word ESN paradigm. With TMS, we transiently perturbed activity in Broca’s area at the stage of prediction (e.g., “the”) and measured the negativity elicited by predicted (e.g., noun) and unpredicted (e.g., verb) categories. We hypothesized that disruption of Broca’s area during the processing of the first word would prevent an ESN effect (ungrammatical vs. grammatical) on the second word, as no prediction is initiated. Therefore, categorical effects on the second word would be similar for both grammatical and ungrammatical items.
Methods:
We employed a German version of the auditory two-word ESN paradigm. The first word could either be the determiner “Ein” (a, predicting a noun) or the pronoun “Er” (he, predicting a verb). The second word could either be a noun or a verb. Two types of two-word constructions were therefore included: grammatical (“Ein” and a noun, “Er” and a verb) and ungrammatical (“Ein” and a verb, “Er” and a noun). To selectively interfere with the predictive stage, we transiently perturbed activity in Broca’s area by means of a 10 Hz train of five TMS pulses starting at the onset of the first word. Three stimulation conditions were included: Broca’s area (target site) and two control conditions (superior parietal lobe and sham stimulation), with the order counterbalanced across subjects.
Data from twenty-nine German speakers entered the final analysis (fifteen female; mean age: 27.1 years, standard deviation: 4.1 years). The statistical analysis focused on the second word. We analyzed the main effect of grammaticality (ESN effect) and its modulation by TMS (grammaticality*TMS interaction) with cluster-based permutation tests (Maris & Oostenveld, 2007). By employing state-of-the-art electrical field modelling (Weise et al., 2020), we further correlated the TMS-induced electrical field in Broca’s area with changes in the ESN effect relative to the sham condition.
Results:
The cluster-based permutation test revealed a main effect of grammaticality, with the presence of a significant early negative cluster (approximately from 190 to 430ms, P < 0.0005, cluster-corrected) and a late positive one (approximately from 440 to 800ms, P < 0.0005, cluster-corrected). The grammaticality*TMS interaction was not significant (P > .5, cluster-corrected). No significant correlation was found between changes in the ESN effect and the TMS-induced electrical field in Broca’s area (r = 0.142, p > 0.1, BF01 = 3.302).
Conclusions:
The main effect of grammaticality is consistent with an early analysis of word category, followed by late repairing processes (Friederici, 2011) in case of ungrammaticality. The lack of a significant grammaticality*TMS interaction suggests that Broca’s area may not be causally involved in categorical prediction. Our findings are compatible with functional studies proposing that this region might be involved in the bottom-up integration of words into syntactic structures (Bhattasali et al., 2019).
