| dc.description.abstract | This study examines the effect of regularity and inflectional entropy on the production of Greek verbs. A previous study on the production of Greek verbs found no difference in producing Greek regular and irregular verbs (Terzi et al., 2005). It was suggested that Greek irregular verbs, like regular verbs in Greek and unlike, for instance, irregular verbs in English, involve decomposition and consequently the two types of verbs have no difference regarding the memory systems involved in their past tense formation. However, in this study, I suggest that an effect of regularity on past tense formation is nonetheless expected due to a morphological priming effect. It is proposed that error rates, the measurement used in the Terzi et al. (2005) study, are not sensitive enough to detect the effect of regularity, at least for healthy people. Therefore, a sentence completion task was administered to 32 healthy Greek speakers to test for the effect of regularity on verb production, using 48 verbs, half of which were regular and half irregular, measuring both the response times and the error rates. The results revealed a significant effect of regularity on the response times of the correct answers, with the regular verbs being produced faster than irregular verbs. Although a relatively high error rate was observed (4.69%), regularity was not attested to be the source of it. The high error rate was, however, explained by the inflectional entropy of the verbal paradigm, an information-theoretic measure of complexity that has been previously shown to correlate with processing latencies (e.g. Van Ewijk, 2013; Tabak, 2010; Baayen & Moscoso del Prado Martin, 2005). Specifically, the results showed that the higher the inflectional entropy, the higher the probability of error. No significant effect of inflectional entropy on response times was detected. In an attempt to see if inflectional entropy has the same effect on the error rates of other populations, the data of the Terzi et al. (2005) study on the production of verbs by Greek Parkinson’s disease patients were reanalyzed. This analysis revealed no significant effect of inflectional entropy on error rates possibly because of the small amount of data and the small range of inflectional entropies of the verbs in the Terzi et al. study. Therefore, it cannot be concluded that inflectional entropy has a different effect on the performance of Parkinson’s disease patients than that of healthy people. This is the first study, at least to my knowledge, suggesting an effect of regularity on the production of verbs in a relatively understudied language, Greek, and it also provides further data for a (relatively) new measure of complexity, inflectional entropy. Based on all the data collected for Greek, a model of past tense formation is proposed, which could be the baseline for future research for a more holistic theory of lexical access. | |
