Authors:

Cutler, A., Mehler, J., Norris, D. & Segui, J.

Reference:

Year of publication:

1987

CBU number:

2053

Abstract:

In seven experiments reaction time to detect the initial phoneme of words and nonwords was measured. Reaction time advantages for words over nonwords come and go according to the particular characteristics of the experimental situation. One relevant characteristic is degree of task monotony, an effect which is most parsimoniously explained by attention shifting between levels of processing. General classes of models of the relationship between levels of processing in comprehension are discussed in light of the results. Serial models incorporate an attention shift explanation of the monotony effect more elegantly than do interactive models. Alternative serial models are available in the literature in this area. One recent model, which allows only a single outlet point for phoneme detection responses, and hence requires that apparent reaction time advantages for words are artefactual, can be unambiguously rejected on the basis of the present data. It is argued that a serial model involving competition between target detection based on a prelexical representation and detection based on a lexical representation most satisfactorily accounts for the overall pattern of results.