The influence of harmony on structural priming in artificial languages, extended

This is a combined replication of the experiment in Culbertson, Legendre and Smolensky (2012), Learning biases predict a word order universal. Cognition, 122, 306–329 and Experiment 1 from Feher, Wonnacott and Smith (2016), Structural priming in artificial languages and the regularisation of unpredictable variation, Journal of Memory and Language, 91, 158-180. We adapt the paradigm from the Feher et al. study to run online (on Mechanical Turk), and test all 4 language types investigated by Culbertson et al. In brief, participants learn a miniature language which consists of four sentence types: Numeral-Noun, Noun-Numeral, Adjective-Noun and Noun-Adjective. All four orders occur in their input, but depending on input condition, two of the orders (e.g. Numeral-Noun and Adjective-Noun for participants receiving a type 1 language) are the dominant orders. Experiment 1 constitutes an extended training exposure to one of these languages. Completion of Experiment 1 is a pre-requisite to participating in Experiment 2, which involves further training, followed by interaction with Glermi (their alien language tutor) in a director-matcher task which allows us to test for the present of structural priming, i.e. copying Glermi's word-order choices on a trial-by-trial basis.