One criticism of both singleand dual-route models of regular/irregular inflection is that the proposal that irregulars are stored fails to account for regularities in the structure of irregular forms (see Embick & Marantz, 2005). Far from showing the sort of arbitrary variation allowed by storage, most irregular forms are phonologically similar to their stems and, in many cases, to regularly inflected forms (e.g., many English irregular past forms end in the affixes –d or –t: made, left, slept). This has led many to conclude that nearly all irregular forms are generated via affixation and stem Readjustment rather than storage and retrieval (Chomsky & Halle, 1968; Halle & Marantz, 1993). On such views, English past-tense forms vary along two dimensions: (i) whether their affixes are phonologically realized or phonologically null (‘‘zero affixes’’), and (ii) whether or not the stem undergoes Readjustment following affixation. Overtly affixed and Readjusted past forms are irregulars such as slept, made; overtly affixed and non-Readjusted past forms are simply regulars. Zero-affixed and Readjusted past forms are irregulars like gave, sang; zero-affixed and non-Readjusted past forms are irregulars identical to their stems (e.g., hit). In English, zero-affixed irregulars constitute a large percentage of irregular past-tense forms. Could the proposed agrammatic preference for English irregular forms (e.g., Ullman et al., 2005) reflect a preference for zero affixes over overt affixes? Most studies of English regular and irregular past tense have not distinguished overtly affixed and zero-affixed irregulars, but Ullman et al. (2005) report that an anterior aphasic patient produced 80% of zero-affixed forms correctly, while producing only 50% of overtly affixed forms correctly (cf. 20% of regulars correct). Given the general proposal that agrammatic aphasia reflects impaired phonological spellout of morphosyntactic features (Fix, 2005), we hypothesized that past-tense forms with overt affixation (including both regulars and overtly affixed irregulars) should be more difficult for agrammatic speakers than forms with zero affixation. Method