IntroductionLinguistics research is filled with observations such as the following: 'There are three green books on the table' is an acceptable sentence, but 'There are green three books on the table' is not. Such judgments-as well as judgments about co-reference, ambiguity, pronounceability, and more-form a significant part of the evidence base for linguistics. This is in large measure due to Chomsky, whose work has exemplified the fruitfulness of such evidence and whose Aspects of the Theory of Syntax (Chomsky 1965, chapter 1) is a locus classicus for theorizing about their status. The prominence of judgment data in contemporary linguistics is crucially tied to Chomsky's mentalist reconception of the field.Judgment data were not completely absent prior to Chomsky's work. For example, field linguists did not always prescind from asking informants whether such-and-such was something they would say, and Chomsky's teacher Zelig Harris emphasized the importance for phonology of speakers' judgments concerning sound differences (Harris 1951). But the positivist, behaviorist, and structuralist positions that dominated American linguistics in the first half of the 20 th century tended to view the use of judgment data with suspicion and focused rather on produced sentences.The methodological strictures in part arose in reaction to problems encountered in earlier introspectionist psychology (discussed in section 3). But the focus on produced utterances reflected as well a particular conception of what languages are and thus what linguistics is about. Though linguists of this period differed in many ways, they shared a tendency to view languages as consisting in the totality of utterances speakers of that language can produce (an E-language in Chomsky's (1986) terminology); and much work focused on describing, analyzing, and taxonomizing languages so conceived-for example, the many Native American languages so apparently different from the Indo-European languages which were then more familiar to linguists. While such a conception does not of itself preclude the use of speakers' judgments (cf. the remarks on Devitt's (2006) views in section 2), it is more naturally combined with an emphasis on corpus data, especially given the methodological scruples already mentioned.Conversely, judgment data find a natural home in Chomsky's mentalist reconception of linguistics-a reconception, according to Chomsky (1966), that is in fact a recovery and development of earlier ideas about language. On this approach, linguistics aims, not just to describe linguistic products, but to provide a cognitive explanation of various of their distinctive features. One of Chomsky's core hypotheses is that there is an innately constrained computational procedure realized in the mind-brain-so-called I-language-that is implicated specifically in linguistic phenomena and whose character explains some of their distinctive features. As with aspects of our cognition more generally, we cannot directly observe I-language but must infer it from the effects to w...