We examined word superiority and sentence superiority effects in post-cued letter identification by embedding target letters in a letter string that was part of a sequence of letter strings separated by spaces. Experiment 1 compared letter identification in words versus random consonant strings (i.e., nonwords), thus involving three conditions: grammatical word (e.g., HE RUNS OVER THERE; the target being the letter V), ungrammatical word (e.g., THERE HE OVER RUNS), and nonwords (e.g., THPRN HJ GVTR LPDKS). Stimuli were displayed for 500 ms and post-masked. Letter-in-word identification was greater in the grammatical than in the ungrammatical word context (a sentence superiority effect, SSE). Moreover, letter-identification accuracy was greater in words than in nonwords (a word superiority effect, WSE). Experiment 2 used pronounceable pseudowords instead of nonwords and replicated the SSE and WSE seen in Experiment 1, with the size of the WSE being substantially reduced relative to Experiment 1. Experiment 3 tested letter identification in words, pseudowords, and nonwords, either in a grammatical or in an ungrammatical context. We again found a significant SSE on letter-identification accuracy combined with the standard pattern of the WSE (word > pseudoword > nonword). Finally, the classic WSE pattern was also found when stimuli were presented in isolation in Experiment 4.
