Statistical preemption is a process through which speakers learn to observe arbitrary restriction in language, allowing them to avoid meaningful utterances judged as questionable due to the existence of an established alternative formulation used to express the same meaning-in-context. The process relies on indirect negative evidence ¬- that is, evidence based on absence, in this case the absence of an expected formulation when the alternative is produced instead. Negative evidence is sometimes considered irrelevant in linguistics; statistical preemption is able to utilise it due to the core assumptions of its specific theoretical framework, construction grammar. Statistical preemption is contrasted with conservatism via entrenchment to highlight how different negative evidence explanations approach questionable formulations; both consider frequency, but statistical preemption additionally focuses on the idea of competing constructions. Non-native speakers do not benefit from statistical preemption to the same extent as native speakers. The thesis addresses the suggestion that learners might find evidence for it in corpora, arguing that while it may be possible to consider negative evidence in corpora, this requires extensive knowledge of linguistics not available to the average language learner.
|
