A recent wave of linguistic research has attempted to demonstrate empirically that in learning the syntax of a second language, adults re-access their universal grammar (UG). However, the conspicuous lack of success of second language learners has caused researchers to seek evidence in experimental data. The most expedient source of this data is grammaticality judgments, and they present problems of reliability and validity. More trustworthy elicitation techniques have been developed, but grammaticality judgments are still being used with few methodological safeguards. The most serious barrier to the development of a "principled theory" of second language learning appears to be reliance on unprincipled data. (MSE)