ABSTRACT
Theoretical developments in the study of visual word recognition are surveyed from three viewpoints; that of Huey, Neisser, and the present day. Early work tended to treat words as linguistically empty spatial patterns. A major focus of interest was on the facilitatory effect of stimulus structure on recognition. Theoretical argument largely rested on the quality of perceptual experience. Attentional hypotheses played a dominant role in accounts of the limits of performance and of strategic flexibility. In the 1960s, discovery of the perceptual advantage of unfamiliar but word-like strings heralded more detailed manipulation of word properties, requiring, in turn, more precision and differentiation in theoretical accounts. With development of the concept of a mental lexicon as the interface linking the perceptual to the linguistic domain, the lexical decision task began its vertiginous climb to prominence. Experimental variables attracted interest as means for diagnosing the nature of lexical representations and the procedures used for gaining access to them. To this end, it became increasingly crucial to distinguish between universal and task-specific effects. As an approach to modelling the traditional boxes and arrows representation of lexical architecture came to be rivalled by parallel distributed processing networks.
