It has been argued that children with Specific Language Impairment (SLI) use language in a conservative manner. For example, they are reluctant to produce word-plus-frame combinations that they have not heard in the input. In addition, there is evidence to suggest that their utterances replicate lexical and syntactic material from the immediate language environment. Aims: This study investigated further the linguistic conservatism of children with SLI. It addressed the hypothesis that these children have difficulties using syntactic schemas in a verb-general manner. It employed a technique pioneered by Childers and Tomasello in 2001 that aims to develop verb-general representations through the presentation of appropriate exemplars. In addition, it tested the hypothesis that children with SLI present with input-dependent language use. Methods & Procedures: Twenty-four children with SLI (mean age 5 and 6 years) were matched to 23 typically developing children (mean age 3 and 5 years) on overall language abilities. The children experienced play sessions designed to simulate exemplar-based learning. Facilitative exemplars consisted of sentences that alternated between nouns and pronouns in the argument slots, e.g. Pooh 's churning the plate . See, he's churning it . The children were divided evenly into two groups. One group experienced facilitative exemplars, while the other group heard exemplars with no variation in the argument positions. The training effect was assessed with a procedure in which the children heard novel verbs in a non-transitive frame and were required to produce them in a transitive frame. In addition, the argument types of the children's responses (nouns versus pronouns) were analysed to investigate the issue of input dependence. Outcomes & Results: Few of the children produced novel verbs in the transitive. This result may indicate difficulties using the transitive, although a number of methodological difficulties may be responsible for finding. As hypothesized, the children with SLI demonstrated greater input dependence in terms of the kind of arguments they used in responses during the training sessions. Conclusions: Potential reasons for the low response rate of both groups during the testing stage are explored and a number of methodological adjustments proposed. It is argued that the greater input dependence of the children with SLI may reflect a tendency to reduce processing load within a limited capacity system. It is proposed that input dependence could be a fruitful new area for SLI research and an important issue in the design and delivery of therapy programmes. (Contains 3 tables, 1 figure, and 1 appendix.)