Irregular past tense forms in English: How data from children with specific language impairment contribute to models of morphology

Abstract

Two cognitive models of inflectional morphology are widely debated in the literature-the Words and Rules model, whereby irregular forms are stored in the lexicon but regular forms are created by rule, and Single Mechanism models, whereby both regulars and irregulars form an associative network, with no rules. A newer model, the Computational Grammatical Complexity (CGC) model, recognises the contribution of hierarchical complexity in three components of the grammar, syntax, morphology and phonology, to the construction of morphologically complex forms. This model has previously been tested for regular past tense inflection in English, and in this study we test its predictions for the English irregular past tense, in four groups of children: a group with Grammatical Specific Language Impairment (G-SLI; aged 9;8-17;8), and three groups of typically developing children (aged 5;4-8;5). Children with G-SLI provide an important test case for the CGC model because they have deficits in syntax, morphology and phonology. As predicted, children with G-SLI produced fewer tense-marked irregulars than expected for their age, and fewer over-regularisations than their language-matched controls. The effect of verb-end phonology on over-regularisation and null-marking errors was the same for all groups: both G-SLI and typically developing children were more likely to over-regularise verbs ending in a vowel, and more likely to null-mark verbs ending in an alveolar consonant. We interpret these results as providing further support for the CGC model. © 2011 The Author(s).