The role of assembled phonology in English visual word recognition: Towards a nonlinear model of assembly

The role of assembled phonology in English word recognition has been the focus of a considerable research effort. Yet, the empirical findings regarding the role of assembly exhibit marked contradictions, which are further linked to the properties of the experimental method. This state of research may be partly due to a failure to consider the structure of the assembled representation as a linguistic entity. In an attempt to account for these contradictions, a new model of assembly is proposed. In accord with current autosegmental theories of phonology, the model postulates a multi-planar representation that segregates consonants and vowels in different planes. This structure further determines the on-line process of assembly: The consonant and vowel components are derived in two consecutive cycles that differ in their automaticity. The model is motivated by its observational and explanatory adequacy and its computational plausibility. The model's claims are supported by a series of English masking experiments demonstrating that the relative contribution of consonantal and vowel components changes as a function of the target's exposure duration. The observation of these results in the performance of the Stroop task and a concurrent secondary task of digit recall suggests that the assembly process is mandatory. However, the difference in the susceptibility of the consonant and vowel components to digit load indicates that their computations differ in their relative automaticity. An additional set of masking studies in Hebrew further examines the linguistic motivation for the structure of the assembled code. Finally, the methodological implications of the model, identifying the exposure duration allowed by the experimental method as a critical determinant of the facet of assembly it reveals, are supported by an experiment comparing the masking and the naming techniques.