The CHAT transcription standard of the Child Language Data Exchange System (CHILDES), the world's largest computer archive of child language data, is critiqued. CHAT is being considered as a possible standard for the field of child language as a whole. It is argued that CHAT is biased toward three views that are no longer current in the field: (1) toward written rather than spoken language; (2) toward a reductionist, compositional perspective on language and discourse; and (3) toward the view of child language as defective adult language rather than as systematic in its own right. A minimalist standard is proposed that is consistent with mainstream tenets, less theory-committed, and more readable than CHAT. It also can be accommodated by CHILDES' existing computer programs, and it enables exhaustive searches necessary for unbiased results. The minimalist proposal advocated is summarized in the appendix, which also contains tables on the following: CHAT's utterance delimiters; an example of a prosodically-based alternative to CHAT's unit-final delimiters; a comparison of Bloom's original format and CHILDES' version of the data in CHAT format; Bloom's data modified as necessary for line-oriented searching; CHAT's "special form" tags; and a sample from Clark's data on CHILDES in a non-CHAT minimalist format. (LB)