Computational models of eye-movement control during reading: Theories of the 'eye-mind' link

摘要

During the last few decades, technology has made it possible to precisely measure when and where a reader is looking when he or she is reading. This eye-tracking methodology has made it possible for psychologists to ask - and to answer - questions about the cognitive processes that allow readers to build whatever mental representations are necessary to decipher printed text. This approach has been extremely profitable; one only has to consult a recent review (e.g., see Rayner, 1998) to gain some appreciation for how much our understanding of the psychology of reading has advanced through studying the eye movements of readers. It is also striking that so much of what has been learned about reading is in fact about the psychology of reading; by studying eye movements, we have learned a considerable amount about how attention is allocated, how words are identified, and how various linguistic representations are generated from the text (just to mention a few of the topics that have been studied). It should come as no surprise, therefore, that the question of what actually guides the eye movements of readers has largely remained a methodological one, with most researchers simply being content in knowing that -at least to some approximation - the reader is thinking about whatever bit of the text that he or she is looking at, and that the eye-tracking methodology is a valid one.