PREDICTING LETTER SEARCH TIME THROUGH WORDS AND NONWORDS: THE ROLES OF STATISTICAL FREQUENCY AND LEXICAL STATUS IN THE WORD-SUPERIORITY EFFEC

The present study sought to answer two questions: (1) does target letter statistical frequency determine search time and (2) can the word-superiority effect be explained in terms of the summed statistical frequencies of the component letters of the stimulus string? To answer the first question, subjects searched for predetermined target letters placed in either their high or their low statistically frequent positions. The target letters were embedded in nonwords of differing context (i.e., high, medium and low summed statistical frequency), and in words of medium summed statistical frequency. The results indicated that although search time was not predicted by the statistical frequency of the single target letter, the summed statistical frequency of the entire stimulus string could predict letter search time through nonwords.;To address the second question, subjects searched for predetermined target letters through words and nonwords which were equated for summed statistical frequency. The results yielded a clear word-superiority effect. Thus letter search time could not be accounted for solely by summed statistical frequency. Lexicality, pronounceability, trigram frequency and number of syllables/vocalic center groups were examined as possible explanations. The results ruled out all factors but lexicality and were interpreted as providing evidence against a strict hierarchical model and in favor of an interactive/heterarchical model of word processing.