A three-minute test of implicit language processing for large sample sizes: Proof-of-concept application in a public science event
Research into human language processing is challenged by its implicit nature of processes, the short time-window in which they take place, and heterogeneity within the population. We used a word monitoring task to measure implicit sensitivity to grammar, semantics, and the strength of word collocations, when listening to natural spoken sentences. We report substantial and novel adjustments to the paradigm, which allowed a participant to be tested in about three minutes, and data from a proof-of-concept study for which we recruited 125 visitors of different backgrounds to a science event. Linear mixed-effects models suggest that participants’ response time to target words was 45ms slower when words were less predictable in their grammatical, semantic, or collocational context (p = .034). There was no significant difference between language conditions. However, in non-native speakers, grammatical manipulation was less effective than manipulations of semantics and word collocations (p < .08), which is consistent with theories of shallow language processing in non-native speakers. There was no significant effect of age on language sensitivity. Effects generally replicated findings based on standard word-monitoring methods, suggesting that the novel adjustments can be applied in situations where participants’ time is limited, e.g. at events, in schools or online.