Studies of second language acquisition often employ analytical concepts defined relative to the target language (such as obligatory context, error, etc.). It is demonstrated here that this practice can be a serious obstacle to the investigation of crucial questions about the nature of learners' languages. The difficulties caused by this practice—called here the “comparative fallacy”—are illustrated in an investigation of the framework proposed by Tarone, Frauenfelder, and Selinker (1976) to study the issue of interlanguage systematicity and variability. Their framework is shown to result in incorrect or unrevealing analyses of systematicity and to obscure the character of the interlanguage. These defects are shown to be the results of the comparative fallacy.