03 July: Carmen Saldana and Jenny Culbertson

In this study we use Artificial Language Learning techniques (e.g., Culbertson & Adger, 2014) to test a hypothesized link between cognitive biases active during language learning and morpheme order cross-linguistic regularities; in particular, we investigate the tendency for number morphemes (e.g., singular or plural markers) to be ordered closer to the noun stem than case morphemes (e.g., nominative or accusative markers) (Greenberg, 1963, Universal 39). We hypothesize that this universal tendency might be driven by learners’ biases towards orders that match semantic scope relationships (Bybee, 1985; Rice, 2000). Across a series of experiments, we taught English-native participants an artificial language with noun stems, and case and number morphemes. Crucially, the input language indicated only that each morphemes preceded or followed noun stems; examples in which the two morphemes co-occurred were held out—i.e., no instances of plural accusatives. At test, participants were asked to produce utterances, including the held-out examples. We find that learners consistently produce number morphology closer to the noun stem than case. We replicate this basic effect with free and bound morphemes, and when case markers are higher frequency. However, we are able to reverse the preference to some degree by strengthening the relationship between nouns and particular case markers. Taken together, our results provide evidence of a cognitive bias towards scope-isomorphic patterns which may play a causal role in this strong typological generalization.