{"pk":28556,"title":"Modeling Children’s Early Linguistic Productivity Through the Automatic\nDiscovery and Use of Lexically-based Frames","subtitle":null,"abstract":"A central question for cognitive science is whether children’s\nlinguistic productivity can be captured by item-based\nlearning, or whether the learner must be guided by abstract,\nsystem-wide principles governed by innate constraints. Here,\nwe present a computational model of early language\nacquisition which learns to discover and use lexically-based\nframes in a fully incremental, on-line fashion. The model is\nrooted in simple prediction- and recognition-based processes,\nsubject to the same memory limitations as language learners.\nWhen exposed to English corpora of child-directed speech,\nthe model is able learn developmentally plausible frames and\nuse them to capture over 70% of the utterances produced by\ntarget children aged 2 to 5. Across a typologically diverse\nrange of 29 languages, the model is able to capture over 68%\nof child utterances. Together, these findings suggest that\nmuch of children’s early linguistic productivity can be\ncaptured by item-based learning through computationally\nsimple mechanisms.","language":"eng","license":{"name":"","short_name":"","text":null,"url":""},"keywords":[{"word":"language learning; language acquisition; usage-\nbased approaches; computational modeling; chunking"}],"section":"Papers with Oral Presentations","is_remote":true,"remote_url":"https://escholarship.org/uc/item/644529zb","frozenauthors":[{"first_name":"Stewart","middle_name":"M.","last_name":"McCauley","name_suffix":"","institution":"University of Iowa","department":""},{"first_name":"Morten","middle_name":"H.","last_name":"Christiansen","name_suffix":"","institution":"Cornell University","department":""}],"date_submitted":null,"date_accepted":null,"date_published":"2019-01-01T18:00:00Z","render_galley":null,"galleys":[{"label":"PDF","type":"pdf","path":"https://journalpub.escholarship.org/cognitivesciencesociety/article/28556/galley/18427/download/"}]}