ABSTRACT
Considerable empirical evidence indicates that perceivers construct a grammatical representation of sentences during comprehension. The principles underlying constituent structure analysis have been investigated in some detail, and their predictions are beginning to be tested in languages other than English. If the results of these investigations continue to be positive, this will encourage the view that psycholinguistics is indeed making progress in the endeavour to develop a theory of human language processing, not just a theory of processing English. However, even within English, the interaction of structural principles with item-specific lexical preferences and with discourse constraints continues to be debated.
Several relatively neglected areas of research are now receiving attention, including theories of recovery from misanalyses, the role of thematic relations in comprehension, and the processing of various types of long-distance dependencies. Conclusions in these areas remain tenative. However, the hypotheses being explored clearly indicate psycholinguistics has changed considerably. The question is no longer whether the human language comprehension system is structured or whether it uses various broad classes of information (e.g. the grammar of the language): Instead, the focus of attention is on fairly detailed and articulated hypotheses about the nature of that structure and the principles underlying the co-ordination of the myriad information sources implicated in language comprehension.
