Perfecting research techniques in an imperfect world: response to MacWhinney & Snow

Edwards (1992c) concerns data and data use rather than manuals. It presents principles to facilitate reliable archive research in an imperfect world. MacWhinney & Snow (1992) endorse the principles but disagree on important details, discussed in this response. M&S underrate the impact of corpus inconsistencies on substantive conclusions. Since acquisition research concerns events which often trickle in a few at a time, main findings often involve a handful of examples, so overlooking a single early variant can seriously skew results. Until archives become perfect, if they do, type-token listings can help in identifying variants before computer search and avoiding this hazard.