[PS-1.4] Discrimination learning, suffixing and prefixing in an artificial language learning experiment

Vujovic, M. 1 , Ramscar, M. 2 & Wonnacott, E. 1

1 Department of Language and Cognition, University College London, UK
2 Department of Linguistics, University of Tubingen

There is evidence for a cross-linguistic preference for suffixes over prefixes in marking grammatical distinctions (Greenberg, 1963). According to discrimination learning models, prefixes facilitate learning of probabilities of subsequent elements in a sequence, and thus vocabulary learning, whereas suffixes facilitate category learning (Ramscar, 2013). In the current study, adult participants were exposed to an artificial language describing alien characters, with two noun categories marked by both phonological and semantic cues and each accompanied by one of two affixes (Category1_affix: ge; Category2_affix: ma). There were two versions of the language: Suffixing_Language: nouns followed by the affix; Prefixing_Language: nouns preceded by the affix. Training was followed by a vocabulary test and a generalization test. From the performance of computational models trained on each version of the language (using the ndl package in R implementing Resorla-Wagner (1972)), we predicted that Prefixing_Language participants would show better vocabulary learning, whereas Suffixing_Language participants would be better at generalizing the correct affix to new category members. Data collection is ongoing. Preliminary analyses suggest that, as predicted, the Prefixing_Langauge participants were better at vocabulary learning than the Suffixing_Langauge participants, who performed at chance. There was no difference between conditions in generalization, although both conditions performed above chance.