Simulating syllable frequency effects within an interactive activation framework

Abstract

Simulation data are presented for a novel computational model of visual word recognition containing syllabic representation units. The model is based on the multiple readout model MROM (Grainger & Jacobs, 1996) and it was used to simulate data from a syllable frequency experiment. The model successfully simulates the inhibitory syllable frequency effect in lexical decision obtained so far in Spanish, German, and French. In addition, this model, fully representing the principles of interactive activation between the three representation layers of letter, syllable, and word units, proved successful in correctly parsing all presented stimuli contained in its lexicon with different types of syllabic structures. Multiple regression analyses on the model's output and on human data confirmed the model's ability to still account for effects of word frequency and orthographic neighbourhood like the original MROM (Grainger & Jacobs, 1996) it was derived from.