6.1 Introduction
In the previous chapters I have examined the comprehension of pronouns in Broca’s agrammatic aphasia. The aim of the first part of this chapter to provide a brief overview of the results emerging from the experiments presented here and to suggest issues for future research. The goal of the second part of this chapter is to place the findings on the comprehension of pronouns and the weak/delayed syntax model in the broader framework of a general language deficit in agrammatic Broca’s aphasia.
6.1.1 Summary of the experimental findings
The main objective of the experimental studies presented here was to unravel the funtional organisation of the linguistic system that governs pronominal reference assignment discussed in the introductory chapter. In the experiments with Broca’s aphasic patients and children I focused on comprehension of pronouns in a number of different constructions.
In Chapter 2, two experiments testing the comprehension of pronouns in simple transitive and ECM constructions in agrammatic patients were discussed. The results of the first experiment indicated that Broca’s patients experience more difficulties interpreting pronouns in ECM environments than in the simple transitive sentences. The results of the second experiment showed that agrammatic patients could both access and use the morphosyntactic information on the pronoun in both types of constructions when linking the pronoun to its antecedent. I argued in this chapter that the Primitives of Binding model best explains the results of these experiments. In combination with the assumption that agrammatic patients have a reduced capacity for syntactic processing, I proposed that the economy hierarchy in these patients is different from the economy hierarchy in healthy non-brain-damaged adults. The most economical dependency in healthy adults is created in narrow syntax. In Broca’s patients the syntactic route is more costly, which allows other levels such as discourse to be used to assign reference without violating the economy hierarchy in these patients.
In Chapter 3 comprehension of pronouns in environments where Rule-I does not apply was assessed. Two types of conjoined constructions with the pronoun in the subject and object positions of the second conjunct were tested where the only way to connect a pronoun to its antecedent is through coreference. Agrammatic patients performed at chance level on pronouns in these sentences, which indicates that the errors they make are not limited to instances where Rule-I is at work. The error pattern Broca’s patients exhibited were related to a competition between two different discourse constraints, topic preference and the constraint of parallelism of thematic roles.
In Chapter 4 interpretation of possessive pronouns in VP-ellipsis was examined. In this experiment I examined the availability of semantic and discourse dependencies in agrammatic patients. The patients I tested were able to use variable binding to assign reference to the possessive pronoun. However, they were incapable of establishing a discourse dependency between the antecedent DP and the pronoun. They also had a preference for the bound variable interpretation that is much stronger than what the controls showed. I proposed that their grammatical knowledge of these constructions is not impaired and that their errors are not a result of a lack of knowledge but a consequence of a slower- than-normal syntactic processing.
In Chapter 5 experimental data from pre-school children were presented and discussed. The four experiments conducted with agrammatic patients were also used to test children’s ability to comprehend pronouns. The results of these experiments showed similarities but also differences in the patterns of errors of the two populations. The two populations exhibited a similar pattern of errors in ECM, transitive and coordinate constructions. The results of the experiments testing the interpretation of possessive pronouns in VP- ellipsis in these two populations were very different. I argued that children’s syntax is weak and not fully operational at this stage in their development. Their syntactic processing is thus delayed, which is similar to the delayed syntactic processing in Broca’s aphasic patients. The syntactic route is not the most economical route for children. Like in agrammatic patients, the economy hierarchy in children is different from the economy hierarchy in adults. In ECM sentences, children allow the pronoun to receive its referential value through coreference without violating the economy hierarchy. Unlike agrammatic patients and healthy adults, in elided VPs children relay on the non-linguistic way of assigning reference to the pronoun though deixis.
The results of these experimental studies provide evidence for a hierarchical organisation of the healthy, impaired and developing linguistic systems. The way in which in my view the impaired system deviates from the healthy system will be summarised below.
6.1.2 General conclusion
On the basis of the experimental results presented in the previous chapters, I argued that brain damage in Broca’s aphasia reduces the capacity of such patients to process syntactic information on time. I assume that the syntactic machinery in these patients is weakened, resulting in a delayed syntactic structure building. In healthy non-brain- damaged adults syntactic operations are the fastest, most automatic operations used to establish pronoun-antecedent dependencies. As such, syntactic operations block other possible operations that can potentially be used to establish these kinds of dependencies. In agrammatic patients and pre-school children, syntactic information is not ready on time to be used in the process of connecting pronouns to their antecedents. As a consequence of this delay, other levels of information, such as discourse or the non-linguistic level, which would normally be blocked by syntax, come into play and provide information for pronoun resolution sometimes resulting in erroneous dependencies. The observed error patterns in these populations thus reveal a competition between narrow syntax and other systems.
The main issue left open for future research is related to the time course of pronoun processing in agrammatic aphasia. The data presented here have all been gathered using off-line methods, which provide no precise indication of how pronouns are processed in real time. The way to examine language processing in real time is by using online paradigms such as cross-modal lexical priming or lexical decision where this time-course is measured indirectly through a reaction time measured by a secondary task, mainly lexical decision. Another way would be to measure Event Related Potentials, which is a much more direct way of measuring brain response to a particular stimulus without the need for a secondary task.
For example, an important issue to be addressed in the case of elided VPs is whether all possible interpretations for the pronoun are activated at the point of reconstruction in the second elided conjunct. It would, of course, be necessary to establish first how healthy adults interpret pronouns in these constructions. The model I argued for predicts that in agrammatic patients only the semantic (bound variable) interpretation will be available at the point of reconstruction. For the neurologically
intact adults it could be the case that the cheaper interpretation (bound variable) is available sooner in time than the more costly option (coreference). This could possibly be examined using the cross-modal lexical priming method, where priming for different interpretation would be evaluated. The ERP technique, on the other hand, could be used to examine how Broca’s patients process gender and number violations on the pronoun in different structures. The effects of these violations have been studied in healthy adults (Berkum, van, Brown & Hagoort, 1999; Brown, van, Berkum & Hagoort, 2000, Osterhout & Mobley, 1995; Barber & Carreiras, 2005), so the agrammatic data could be compared to the findings from the unimpaired population.