Case/agreement matching: evidence for a cognitive bias

In an artificial language experiment, participants were taught two different artificial languages consisting of English content words and novel morphological marking. The first of the languages had matching alignment in both case and agreement, as attested in natural languages such as Basque, Belhare and Tsez. The other language combined accusative case alignment with ergative agreement alignment, a combination which is apparently unattested amongst natural languages. There was no significant difference between the languages in terms of the proportion of participants that showed awareness of the agreement pattern, nor in the ability of aware participants to recall case markers and inflections during training, or select the correct verb inflection in the generation post-test. However, amongst participants who remained unaware of the agreement pattern there was a significant difference in recall of verb inflections and case markers during the exposure phase task – recall was more accurate in the (attested) language with matching case and agreement alignment than the (nonattested) language in which case and agreement alignment were unmatched. We take this as evidence that there is a cognitive bias against the unattested non-matching alignment, reflected in implicit learning