The influence of syntactic priming on syntactic prediction in SOV languages

Syntactic prediction and syntactic priming are known to be important determinants of processing in Subject-Object-Verb (SOV) languages. In this work we investigate how they interact during sentence processing of an SOV language Bengali. In a series of sentence completion studies, we find that both prediction and priming show consistent and robust effects. Interestingly, the strength of priming induced prediction is modulated by the structural similarity between the prime structure and the structure of the predicted construction in the target -- when this similarity is high, prediction strength increases, conversely, when the similarity is low, prediction strength decreases. We additionally find that priming induced prediction reaches a saturation point when prime structure is very complex (operationalized in terms of number of verbal arguments and thematic structure). These novel findings pose a challenge to the current theories of priming such as `residual activation theory' and the `implicit learning' model. The current work suggests that syntactic priming during comprehension could be modulated by language typology (SVO vs SOV). Broadly, we find that the parser in an SOV language predominantly prefers to posit syntactically simple structures and this tendency is not overridden by priming.