The contribution of frequency and semantic similarity to morphological processing

This research investigated the involvement of semantic factors in the processing of morphologically complex words. Lexical decision latencies for unaffixed targets that were immediately preceded by morphologically related primes were examined. The primes were either inflected or derived words (e.g., seemed-seem vs. folder-fold, respectively), which differ in semantic similarity to their targets. ^ Experiment 1 found equivalent priming effects for inflected and derived words, suggesting that factors other than semantic similarity alone mediate the magnitude of morphological priming. Experiment 2 revealed that frequency and semantic transparency interacted in determining the magnitude of morphological priming. Whereas low-frequency inflected and derived primes yielded equivalent levels of priming, high-frequency inflected primes yielded more priming than high-frequency derived primes. Subsequent experiments examined whether this difference in priming can be attributed to differences in semantic similarity. This question was addressed by manipulating variables that influence semantic priming and examining their influence on morphological priming. Experiment 3 demonstrated that both semantic and morphological priming varied as a function of prime-target stimulus onset asynchrony (SOA). At a short SOA (50 ms), semantic priming was unreliable and inflected and derived words gave rise to comparable levels of priming. In contrast, at longer SOAs (150 and 250 ms), semantic priming was found and inflected words yielded more priming than derived words. Moreover, the difference in priming of inflected and derived words increased as SOA increased. Experiment 4 exploited the finding that the effect of semantic relatedness varies in magnitude depending on depth of prime processing. Participants either read the primes (deep processing) or performed a letter search on the primes (shallow processing). This manipulation, however, did not yield the expected effect on the magnitude of semantic priming and hence did not allow a meaningful examination of morphological priming. ^ Taken together, the results suggest that frequency and semantic similarity interactively modulate the magnitude of morphological priming, and by inference, affect the processing of morphologically complex words. The results were interpreted within the connectionist approach to morphological processing, according to which morphological effects reflect the influence of intercorrelations of form and meaning on the dynamics of word recognition.