Where is the locus of the lowercase advantage during sentence reading?

While most models of visual word identification and reading posit that a word's visual codes are rapidly transformed onto case-invariant representations (i.e., table and TABLE would equally activate the word unit corresponding to "table"), a number of experiments have shown a lowercase advantage in various word identification and reading tasks. In the present experiment, we examined the locus of this lowercase advantage by comparing the pattern of eye movements when reading sentences in lowercase vs. uppercase. Each sentence contained a target word that was high or low in word-frequency. Overall, results showed faster reading times for lowercase than for uppercase sentences. More important, while the word-frequency effect was sizeable in the first-fixation durations on the target word, the lowercase advantage only arose in the gaze durations (i.e., the sum of durations of first-pass fixations on the target word, including refixations). Furthermore, we found an effect of word-frequency, but not of letter case, in the first-fixation duration on target words with multiple first-pass fixations. Taken together, these findings suggest that the lowercase advantage reflects operations that do not occur in the initial contact with the lexical entries.