Two experiments are reported on understanding complement clauses embedded into four Greek matrix verbs, Ask (ask information), Promise, Request and Tell (give information), equivalent to English in syntactic and semantic constraints. Experiment 1 was conducted with educably mentally retarded (EMR) and nonretarded (NR) children matched on verbal mental age (MA), and the second with fifth and sixth graders and young high school adults. The results suggest that:(1) EMR and N R children are similar in performance, despite other differences, according to Guttman coefficients of reproducibility and scalability, (2) verbal MA and Digit Span cannot best predict EMR and young NR children's linguistic behavior, (3) the minimal distance principle (MDP) advocated by Chomsky (1969Chomsky ( , 1972Chomsky ( , 1982 fails to account for performance by either population sample on constructions conforming to the same syntactic and semantic constraints as in English, (4) the semantic role principle (SRP) approach (Lederberg and Maratsos 1981; Maratsos 1974) falls short of making consistent predictions of performance by EMR and N R children on constructions requiring the same semantic role allocation, ( 5 ) evidence from Experiment 2, despite differences, supports the results of Experiment 1, (6) the results on understanding complement clauses of the type tested are discussed within the framework of the SRP approach in general, but the emphasis is placed on interaction of semantic and pragmatic presuppositions.Since Chomsky's (1969) seminal study, understanding complement clauses by preschool and school-aged children has been the focus