Lexical word formation in children with grammatical SLI: a grammar-specific versus an input-processing deficit?

An ongoing controversy is whether an input-processing deficit or a grammar- specific deficit causes specific language impairment (SLI) in children. Pre vious studies have focussed on SLI childrens' omission of inflectional morp hemes or impaired performance on language tasks, but such data can be accou nted for by either theory. To distinguish between these theories we study c ompound formation in a subgroup of SLI children with 'grammatical (G)-SLI'. An input-processing account (e.g. Leonard, L. (1998). Children with specif ic language impairment.. Cambridge, MA: MIT Press), in which perception and production of inflections requires extra processing resources, would predi ct that G-SLI children will omit the regular plural -s in compounds (e.g. r at-eater). A grammar-specific deficit account (e.g. Ullman, M. & Gopnik, M. (1994) The production of inflectional morphology in hereditary specific la nguage impairment. The McGill Working Papers in Linguistics, 10, 81-118; va n der Lely, H. K. J. & Ullman, M. (1996). The computation and representatio n of past-tense morphology in normally developing and specifically language impaired children. In A. Stringfellow, D. Cahana-Amitay, E. Hughes & A. Zu kowski, Proceedings of the 20th Annual Boston University Conference on Lang uage Development (pp. 816-827). Somerville, MA: Cascadilla Press, in which G-SLI children are impaired in regular inflectional morphology, would predi ct that G-SLI children will produce regular plural -s forms inside compound s (e.g. *rats-eater). We compared the responses of 16 G-SLI subjects (aged 10 years 4 months to 18 years) with those of 36 normally developing control children (24 matched on language ability and 12 matched on age and cogniti ve ability). All the groups produced irregular plural nouns in compounds (m ice-eater). The normally developing children and teenagers rarely, if ever, produced regular plural nouns inside compounds (*rats-eater), whereas the G-SLI subjects did so often. This pattern of results conflicts with the pre dictions of the input-processing deficit account. The: findings support the grammar-specific deficit hypothesis. The data provide further evidence tha t specialized grammatical abilities may be differentially impaired within t he language system. (C) 2000 Elsevier Science B.V. All rights reserved.