The topic of 'negative evidence', i.e., of linguistic corrections, is focused upon. Its denial in the recent literature is briefly documented and various dimensions of this denial are specified. After a brief historical survey of research on adult corrections of children's speech, verbatim evidence of maternal corrective feedback is presented for the following categories of filial mistakes: Nouns and Labels, Verbs, Determiners, Prepositions, Bound Morphemes, and Syntax. It is also argued and demonstrated that most 'expansions' involve grammatical corrections and that expanding corrections predominate by far during the early developmental period. Finally, diverse aspects of corrections, their generality, and their effect upon language acquisition are discussed from epistemological and methodological perspectives.In any type of skill learning, feedback as to incorrect or sub-optimal performance has a very important function in leading to improvements (cf., e.g., Newell 1981). It would, therefore, be expected that the same principle might apply for the acquisition of language skills, whether in a first language or a foreign one. Gold (1967) formulated a model of language learning in which corrections are seen as highly conducive to learning. Corrective feedback is, of course, widely used in educational settings where foreign languages are taught. For first language acquisition, much of the recent literature asserts, however, that this general principle does not apply.Four forms and degrees of this denial are found. (a) Most general is a comprehensive denial of the existence of corrections (Atkinson