This study was based on a modification and extension of Greene's (1969) test wherein children choose structurally-transformed sentences which are synonymous with key sentences, thereby indicating recognition of the transformational relationships involved and internalization of those transformations. A similar test was devised to investigate the relation between dialect and reading ability. It examined in particular the following hypotheses: (1) Speakers of non-standard dialect will do more poorly on standard English structures than speakers of a standard dialect; and (2) Performance will be affected by the age of the non-standard-speaking children. Comparison of overall test performance among white, middle-class children with that of inner-city children does not substantiate the former hypothesis; yet the differences between control and experimental groups of comparable age are significant, confirming the latter hypothesis and provisionally confirming the former. Specific transformations and their treatment by test groups are discussed, as are pedagogical implications and possible explanations for group-specific difficulties. (DB)