Electrophysiological Insights into Timing Aspects of Discourse Processing in Aphasic Patients

Abstract

In the present thesis, three ERP experiments were used to investigate timing aspects of discourse processing in two groups of healthy subjects and two groups of aphasic patients. In two experiments, subjects were auditorily presented with sentence pairs. The two factors Context (Discourse, Sentence) and Coherence (Coherent, Incoherent) were crossed: in the Discourse condition, the second sentence continued the context introduced by the first in a meaningful way; in the Sentence condition, the two sentences were unrelated in meaning. For both the Sentence and the Discourse condition, a coherent and an incoherent version were created by means of the respective fit of the final word of the second sentence to which ERPs were acquired. In Experiment 1, semantic relationships between the critical word (CW) and at least one content word in the respective preceding context determined the respective fit/misfit of the CW (CW in CAPS, semantic match underlined; Discourse: Bob covered his pancakes with maple syrup. He likes them very SWEET / SPICY. Sentence: We went to see the famous performer. The gardener has mowed the LAWN / HAIR). Note that this introduced a semantic match / violation for the Coherent and Incoherent conditions, respectively.For the control groups, the hypothesis of interactive processing of discourse information was confirmed by the absence of onset differences of the N400 effects in both conditions which shows that both local (sentence) and global (discourse) coherence are processed concurrently and in parallel rather than serially.For the groups of aphasic patients, the nature of their processing deficit should be further determined, i.e. whether their comprehension deficits were due to impaired lexical activation or lexical integration. Furthermore, the hypothesis of heuristic use of semantic information was investigated. Patients were classified as either High or Low Comprehenders based on their auditory comprehension scores on the Boston Diagnostic Aphasia Examination (BDAE). ERP evidence was obtained which confirmed the hypothesis of an underlying integration rather than an activation deficit. ERP evidence reveals an integration deficit which varies as a function of both the severity of the comprehension deficit and the amount of information that has to be integrated: High Comprehenders show a stronger delay of the N400 component in the Discourse than the Sentence condition, and Low Comprehenders do not show any effects in either context condition. The hypothesis of heuristic use of semantic information could not be answered with the present design, i.e. it could not be determined whether the presence of a semantic relationship facilitated processing of the CW or whether the semantic violation impaired processing.A second ERP experiment with a similar design was used to further investigate this aspect of aphasic comprehension. In this experiment, the fit of the CW was to be derived from the respective context in the absence of lexical relationships (Discourse: Joel s office is forty miles away. He does not like the long COMMUTE / CURTAINS. Sentence: Helen reached up to dust the chandelier. While skiing, Randy broke his LEG / NOSE.). This experiment confirmed the results for the control groups from Experiment 1, and for the patients, it showed that rather from profiting from the semantic relations, they suffered from the semantic violations. Yet, did the integration impairment vary as a function of the severity of the deficit. High Comprehenders did show integration with a normal time course in the Sentence and delayed integration in the Discourse condition. Low Comprehenders showed a tendency toward a delay in the Sentence and no signs of successful integration in the Discourse condition.Finally, in a three-stimulus-oddball experiment, the language specificity of the N400 effects was confirmed: control subjects show a frontally distributed P3a component and a centro-posteriorly distributed P3b component. High Comprehenders do not show a P3a component, but they show a P3b component. Low Comprehenders do show both a P3a and a P3b component which rules out that the absence of N400 effects can be attributed to brain damage per se but that they are language specific.