Tag
This paper introduces a data-free universal prior over syntactic structures from a cognitive model of language production, demonstrating it assigns higher probabilities to attested trees and correlates with language-specific data across 138 languages.