Two experiments examined how lexical status affects the targeting of saccades during reading by using the boundary technique to vary independently the content of a letter string when seen in parafoveal preview and when directly fixated. Experiment 1 measured the skipping rate for a target word embedded in a sentence under three parafoveal preview conditions: full preview (e.g. brain-brain), pseudohomophone preview (e.g. brane-brain), and orthographic nonword control preview (e.g. brant-brain); in the first condition the preview string was always an English word while in the second and third conditions it was always a nonword. Experiment 2 investigated three conditions where the preview string was always a word: full preview (e.g. beach-beach), homophone preview (e.g. beech-beach), and orthographic control preview (e.g. bench-beach). None of the letter string manipulations used to create the preview conditions in the experiments disrupted sub-lexical orthographic or phonological patterns. In Experiment 1 higher skipping rates were observed for the full (lexical) preview condition, which consisted of a word, compared to the nonword preview conditions (pseudohomophone and orthographic-control). In contrast Experiment 2 showed no difference in skipping rates across the three types of lexical preview conditions (full, homophone and orthographic control), though preview type did influence reading times. This pattern indicates that skipping depends not only on the presence of disrupted sub-lexical patterns of orthography or phonology but is also critically dependent on processes that are sensitive to the lexical status of letter strings in the parafovea.
展开▼