Individual Differences in Language Processing
The goal of the studies in this project is to examine the time course of the influence of individual word meanings (e.g., lexical association), syntactic factors, and the larger discourse context as a function of main effects or interactions on measures of working memory, cognitive control, and language experience. In this approach we use ERPs and EEG oscillatory measures during reading and listening, and use multiple regression modeling to identify which individual characteristics modulate the amplitude of the ERPs. Thus far the results of these studies suggest that there are different routes to comprehension supported by different neural networks in the brain:
The good enough processing network which relies on meaning relations between words, and stored discourse scenario’s. Processing via this network leads to representations of the gist of the meaning of sentences and discourse. We hypothesize that this network is supported via a ventral pathway that connects regions in the temporal lobes to the Inferior Frontal Gyrus. Processing in this network will lead to good-enough representations, but can also lead to mistakes such as semantic illusions or errors in thematic role assignment when precise combinatorial rules dictate an interpretation that is different from the sum of the meanings of individual words.
A second network, the combinatorial network unifies phonological, syntactic and semantic information and is necessary for precise understanding of sentence meaning. The representations produced by this network are monitored and a feedback loop allows for retrieval of alternative representations when combinatorial processing runs into errors. This is the perysilvian language network where a dorsal pathway connects temporal lobe areas with the LIFG via the angular gyrus. We further suggest that the division of labor between the ventral and the dorsal networks could potentially account for variability in findings with respect to the timecourse and priority by which syntactic and semantic information influence language processing, since the representations that are computed by each of these networks may become available at different moments in time.
The third network contributes to controlled maintenance of context; and in language comprehension we suggest that it is recruited when unification demands are particularly high. In this network, the LIFG is connected to the PFC, which is connected via a dorsal pathway to the inferior parietal lobe and the temporal lobes.
Syntactic Priming in Comprehension.
To understand a sentence, comprehenders identify the meanings of the words in the sentence, and they also determine how the words in the sentence relate to one another. Comprehenders build syntactic structure representations to determine how words relate to one another. Different theories provide different explanations of how comprehenders build syntactic structure representations; and different theories make different claims about how comprehenders mentally represent syntactic structures. According to Chomskian approaches to parsing, syntactic structure representations are generated anew each time a new sentence is encountered. Lexicalist approaches to syntax contend that some aspects of syntactic structure information are pre-built, stored in long-term memory, and associated with specific words. We use eye-tracking and electrophysiological (ERP) methods to assess how comprehenders respond to sentences. In syntactic priming experiments, comprehenders read a prime sentence immediately before they read a target sentence. We manipulate the relationship between the prime and the target sentence and assesse comprehenders' response to the target. Our results show that processing a prime sentence speeds processing of a target sentence if the two have similar syntactic structures and if a critical verb is repeated across the prime and the target. These results indicate that syntactic structure information is pre-compiled and stored in long term memory.
Cognitive Control and Language Impairments in Schizophrenia
Schizophrenia is a prevalent mental health disorder that creates enormous social, economic, and interpersonal hardships for patients and their families. Although hallucinations and delusions are the most salient symptoms of this disease, language abnormalities are among the most prominent cognitive deficits in schizophrenia. In our lab we explore the processes and circuits that underlie impaired discourse comprehension in schizophrenia. We examine whether deficits in controlled integration and maintenance of discourse context in schizophrenia leads to impairments in discourse comprehension, but relatively spares processing of meanings of words and sentence structures. To do so we combine electrophysiological (EEG/ERP) measures of language comprehension with measures of cognitive, social and occupational functioning in schizophrenia. Our approach allows us to examine whether discourse comprehension deficits in schizophrenia relate to impaired cognitive, social and occupational functioning, and the outcome of this research can be used in the development and assessment of new treatments for this disease.