Generative and item-specific knowledge in sentence processing

How do generative linguistic knowledge and knowledge of item-specific usage preferences jointly contribute to language processing? We investigated this using two test cases: binomial expression ordering preferences (e.g. "bread and butter" vs. "butter and bread") and the dative alternation (e.g. "I gave the man a book" vs. "I gave a book to the man"). In both cases, preferences are driven by both generative constraints (e.g. short-before-long) and by item-specific experience (e.g. experience with a specific expression or idiosyncratic verb biases). Using a combination of computational models and psycholinguistic data, we quantify how generative and item-specific knowledge jointly contribute to alternation preferences, and how their contributions vary based on an item's frequency. We demonstrate substantial reliance on item-specific knowledge at much lower frequencies than has been suggested previously. We also found that item-specific experience plays an increasingly large role with increased item frequency. Our findings bring a quantitative perspective to the debate over the joint roles of generative and item-specific knowledge in language processing.