Do morphophonological rules impact both regular and irregular verb inflection? Evidence from acquired morphological impairment

• ¹ New York University, United States

Introduction The role of morphophonological rules in production is a frequent point of contention in competing theories of morphological processing. Dual-mechanism theories have posited that a single default rule (stem+ed) is used to produce the regular past tense, while all other past-tense forms are memorized and retrieved whole. However, research has suggested that a series of stochastic morphophonological rules plays a role in morphological productivity for both regular and irregular novel verbs (e.g. Albright & Hayes, 2003). Under this view, the likelihood of a given rule applying to a verb is partially dependent on its lexical support, measured as the number of phonologically similar verbs in the lexicon which take the same inflectional change (e.g. weep→wept, sweep→swept, etc.). To date, most evidence supporting this view has come from the morphological productivity of novel forms (e.g. Albright & Hayes, 2003) and from visual word recognition (Fruchter, Stockall, & Marantz, 2013). The present study extends this work by demonstrating that morphophonological rules impact production more generally, including the production of both regular and irregular past tense in RMI, an aphasic individual with an established morphological deficit (Rimikis & Buchwald, 2014). While we previously reported differences in production for irregular verbs with either high or low levels of lexical support, the current study further examined this phenomenon, and we found that the production of the past-tense form for both regular and irregular verbs was affected by the support for that verb’s inflectional rule . Subject RMI, 39, right-handed male with aphasia secondary to L-MCA CVA. His production includes frequent morphological errors across tasks, with semantic and phonological errors also occurring. Procedure RMI was administered a past-tense elicitation task. Sentence frames including regular and irregular verbs were presented verbally and visually (e.g. “Every day I run a mile. Just like every day, yesterday I ____ a mile”; N=800). Analyses A confidence score was calculated for each verb, representing the amount of lexical support for its inflectional pattern based on morphophonological rules (Albright & Hayes, 2003). Logistic-regression mixed-effects models were used to assess the impact of this confidence score on RMI’s production above that of other lexical variables. Accuracy and error patterns were analyzed separately for regular and irregular verbs. Results Confidence score was a significant predictor of performance for both regular and irregular verbs. For regular verbs, words with higher confidence scores were more likely to be produced accurately (see Figure 1). For irregular verbs, words with lower confidence scores were more likely to be regularized (i.e. affixation of –ed, e.g. "digged"). Discussion For irregular verbs, lexical support for the particular morphophonological rule affected error type, with words with less support being more likely to regularize. Interestingly, our findings indicate that even regular verbs are sensitive to lexical support as words with less support for the regular pattern were more vulnerable to error. These findings provide additional evidence that morphophonological rules impact spoken production, and may be problematic for accounts in which a single, default rule is used in production.