The contributions of phonology and orthography to skilled reading: An electrophysiological investigation of phonological priming effects

Prior research has shown that pronounceable nonwords, or pseudowords, that are homophones of real words (pseudohomophones) are processed differently than other pseudowords. These results have consistently been interpreted as evidence for phonological recoding during reading, but the role of this phonological information has been debated. In a series of three separate experiments, skilled readers categorized words, pseudohomophones, and control pseudowords while behavioral and electrophysiological responses were recorded. In Experiment 1, the behavioral "Pseudohomophone Effect" was replicated, with slower and less accurate responses to pseudohomophones than other pseudowords in a lexical decision task (LDT). Concurrent recording of event-related potentials (ERPs) revealed that pseudohomophones generated N400 components that were more intermediate to those for real words and non-homophonic pseudowords. When pseudohomophones served as primes for the real words they were based on (i.e., their "basewords"), behavioral priming was greater from identity primes than pseudohomophone primes, but N400 priming effects were similar for targets following both of these types of primes. These results as well as those from a subsequent recognition memory task suggested that pseudohomophones accessed the representation of their basewords. In the following two experiments, primes were masked in order to determine the extent to which priming effects observed in Experiment 1 were due to automatic extraction of phonological information. In Experiment 2 participants performed an LDT, and during Experiment 3 participants performed a semantic monitoring task. Effects of masked priming were small but were generally consistent across the two experiments. Masked identity primes facilitated behavioral and ERP responses to their unmasked targets. In contrast, both kinds of masked pseudoword primes resulted in inhibitory effects on their unmasked target basewords. These inhibitory effects varied depending on the type of information shared by the primes and targets (phonological vs. orthographic) and the task. Results are interpreted in a framework of lexical processing in which interactions between orthography, phonology, and semantics serve to activate representations of words. Furthermore, the N400, as an index of the ease with which a consistent lexical-semantic representation is achieved during reading, may be sensitive to these interactions.