Involvement of the Intraparietal Sulcus in Sentence Comprehension - An rTMS investigation

Semantically reversible sentences, i.e., sentences wherein both characters can potentially perform the given action, have long been used to understand the various mechanisms involved in successful sentence comprehension. Over the decades, studies have established that sentences with non-canonical word-orders such as passive voice sentences are more difficult to process than canonical counterparts such as active voice sentences using psycholinguistic, neuroimaging, lesion-based, and more recently transcranial brain stimulation methods such as transcranial magnetic stimulation (TMS). In addition to understanding the cognitive processes involved, these studies have also attempted to uncover the underlying neural correlates. Various parts of the frontal, temporal, and parietal lobes have been thought to be critical for different functions. In the recent years, the parietal regions have garnered considerable attention. In particular, various studies have found the intraparietal sulcus (IPS) to be involved in the comprehension of semantically reversible sentences, particularly when they have non-canonical word-orders. In this thesis, I attempted to build on this literature and further understand the role of the IPS in sentence comprehension. More specifically I followed-up on two TMS studies by Finocchiaro and colleagues that looked at reversible Italian active and passive sentences. Online repetitive TMS (rTMS) to the posterior portion of the left IPS (henceforth, l-pIPS) affected only the processing of passive sentences in both experiments. In fact, one of the studies also found an effect on ‘passive’ pseudosentences, thus, prompting the authors to suggest that the l-pIPS affects passive sentences irrespective of semantic meaningfulness. The authors concluded that this region is likely to be involved in sentence comprehension, particularly at the stage where thematic reanalysis, i.e., a revision of the initially ascribed thematic roles, occurs. However, these studies were unable to discern if these effects were seen on reversible passives due to their reversibility, passive voice, or the non-canonical word-order seen in passive sentences. They also raised questions regarding whether the region is important in processing only thematic reversibility or is also involved in comprehending reversible sentences without thematic roles (e.g., comparative sentences). I ran three experiments to better understand these factors. In Chapter 1 I summarise the literature on the comprehension of reversible sentences. I discuss findings from behavioural, lesion-based, neuroimaging, and TMS studies that have formed the basis of the current thesis. I conclude by bringing to attention some of the questions raised by these studies that I have attempted to answer in the subsequent chapters. In Chapter 2 I attempted to understand the role of passive voice and reversibility in reanalysis and sentence comprehension. To do this I used a sentence comprehension task while administering online rTMS to the l-pIPS. Participants performed a forced-choice task where they were required to read reversible and irreversible Italian active and passive voice sentences and identify either the agent or the theme in alternate blocks. The experiment showed an effect of rTMS only on reversible passive sentences. While these results are in line with the previous studies, they also draw attention to a critical aspect of comprehension of passive voice sentences. They suggest that passive voice in itself may be insufficient to trigger reanalysis. Instead, it is likely that reanalysis is triggered by the co-occurrence of numerous factors such as voice and reversibility. However, as with the previous studies, this study still does not clarify the role of non-canonical word-order or passive voice per se. In an attempt to distinguish between the two, I ran an rTMS experiment with a sentence-picture verification task in Chapter 3. The experiment used reversible active and passive voice sentences in Hindi. The advantage that Hindi offers in this regard is that both actives and passives are typically presented in the agent-theme-action order. Consequently, such passives may not require the reassigning of originally established thematic roles. Stimulation to the l-pIPS showed no effects on these actives or passives. Interestingly, these null results serve as supporting evidence (albeit, weak) that non-canonical word-order may be essential for thematic reanalysis. If passive voice alone, or even a combination of passive voice and reversibility were sufficient to engender reanalysis, stimulation should have had effects on the passives even in the absence of a non-canonical word-order (as seen in the stimuli of this experiment). Finally, I attempted to understand if the region was involved only in comprehending reversible sentences with thematic role assignment, or also played a role in reversible sentences without thematic role assignment such as comparative sentences (where one of the two characters is the owner of a given property/feature). To do this, I ran the final rTMS experiment reported in Chapter 4. Participants received online stimulation to the l-pIPS while performing a sentence-picture verification task. The stimuli sentences consisted of reversible Italian declarative active and passive sentences, and comparative of majority or minority sentences. Comparatives were used because unlike Italian actives and passives, both types of comparatives are identical in word-order and other morphosyntactic features. In an attempt to maintain the visual complexity across stimuli pictures, same pictures were used to depict the relationships in both declarative and comparative sentences. The results of this study are rather puzzling. Unlike previous studies no effects were found on passive sentences. In terms of the comparative sentences, a selective effect was seen on the ‘easier’ comparatives of majority. These results call into question previous findings which have found an effect on the more ‘difficult’ sentence type, i.e., the passives. The effect of TMS on the comparatives of majority indicates the involvement of the l-pIPS in sentence comprehension yet again. However, these results to do not clarify what specific features of a sentence the l-pIPS helps comprehend. Moreover, given the consistent effect of rTMS on reversible Italian passives in the previous experiments, and the effect on comparatives of majority, it is possible that the current results may have been confounded by the use of complex stimuli. Coupled with the findings from the Finocchiaro studies, this thesis establishes the role of the l-pIPS in sentence comprehension. In particular, the results of the two Finocchiaro studies and Chapter 2 suggest that the co-occurrence of passive voice and reversibility is essential for reanalysis. While the results of Chapter 3 concur with these findings, they also strongly indicate that these features must co-occur with a third factor, namely a non-canonical word-order, to trigger reanalysis. Lastly, the results of Chapter 4 clarify the l-pIPS’ involvement in sentence comprehension. However, it leaves the exact role of the l-pIPS unclear in comprehending reversible sentences without thematic role assignment. The current thesis has advanced our understanding of some key factors responsible for reanalysis, and its neural correlates. Future studies can aim to understand these factors better by exploiting parallel versions of the same sentence type, and by studying different sentence types in isolation. For example, Hindi passives can be presented in both agent-theme-action and theme-agent-action word-orders. Contrasting such versions may help answer questions pertaining to word-order. On the other hand, studying sentences like declaratives and comparatives in separate experiments or even separate sessions may help simplify stimuli, thereby giving us clearer results.