Decomposition of morphologically complex words in English: evidence from event-related brain potentials

To explain processing differences between regular (e.g., start/started) and irregular (e.g., think/thought) word formation linguistic models posit either a single mechanism handling both morphological clusters or separate mechanisms for regular and irregular words. The purpose of the present study is to investigate how these processing differences map onto brain processes by assessing electrophysiological effects of English past tense forms, using the repetition priming paradigm. Event-related brain potentials (ERPs) were recorded from 59 scalp sites as 19 subjects read stem forms of regular and irregular verbs from a list of 1152 words; the stem forms were either preceded (5-9 intervening items) by their past tense forms (=primed condition) or by past tense forms of unrelated verbs (=unprimed condition). The difference between the ERPs to the primed and unprimed stems was taken as a measure of morphological priming. We found that the ERPs to regular verbs were clearly different from those to irregular verbs: the former were associated with an N400 reduction in the primed condition; primed irregular verb stems, however, showed no such effect. Control conditions demonstrated that the N400 modulation for regular verbs cannot be attributed to formal (i.e., phonological or orthographical) priming. These ERP effects indicate that regular verbs serve as more powerful primes for their corresponding stem forms than irregular past tense forms, suggesting that regular (but not irregular) past tense forms may be decomposed into stem plus affix.