WORD FREQUENCY EFFECTS ON LEXICAL SELECTION: EVIDENCE FROM A PICTURE–WORD INTERFERENCE (PWI) TASK

Given that a high frequency (HF) word advantage exists in lexical processing, a question arises about the locus of that frequency effect. This locus may be important for AAC research and clinical practice to provide an empirical rationale for graphic symbol-based AAC interfaces. \ud \ud This study’s first specific aim was to identify whether word frequency affects lexical selection using a picture-word interference (PWI) task. Fifty healthy, monolingual, Native American English speakers, between 40 and 64 years of age, participated in the study. Response times (RT) for semantic, phonological, and mixed distractor conditions served as the dependent variable during a PWI task. Data were analyzed using generalized linear mixed models (GLMMs).\ud \ud In all distractor conditions, participants named HF pictures significantly faster than low frequency (LF) pictures. This result revealed that the word frequency effect occurs not only with phonological encoding, but also with lexical selection and interactive processing between these two steps. This finding is at odds with the Discrete Two Stage (DTS) model that states that word frequency selectively affects the phonological encoding. \ud \ud The secondary aim was to determine whether the target item’s frequency interacts with the distractor item’s frequency. HF distractor words had a stronger interference effect on the retrieval of target words than LF distractor words, resulting in a more delayed RT when naming the target pictures in semantic and mixed distractor conditions. However, an interaction effect was observed between target word frequency and distractor frequency only in the mixed distractor condition. \ud \ud A third aim was to determine whether RT analyses provide a more sensitive measurement than error type analysis for healthy adults. No significant word-frequency effect or interaction was found for error type. A high rate of correct responses and the characteristics of the errors as the end product of inefficient word retrieval are considered as main reasons for this negative result. This finding supports the hypothesis that RT is a more sensitive measure than error type for indexing the inefficiencies that affect naming behavior in the PWI task for healthy adults.\u