Conspiracy and sabotage in the acquisition of phonology: dense data undermine existing theories, provide scaffolding for a new one

This paper presents a major extension of the exemplar-based approach of radical templatic theory [Vihman, M.M., Croft, W., 2007. Phonological development: toward a 'radical' templatic phonology. Linguistics 45, 683-725] and new data from early phonological acquisition, and supports the psycholinguistic appropriateness of frequency-sensitive representations. In order to account insightfully for the striking regularities as well as the irreducible irregularities of early phonology, our Linked Attractor model maintains both underlying and output representations for each lexical entry, and additionally posits both input-to-output rules and feedback (output-to-input) rules to account for patterns of fossilization and developmental progress. Our long-term goal is to make the Linked Attractor model explicit enough for statistical modeling and testing, and to offer it to phonological theory as the basis for a psycholinguistically viable model for adult phonology as well.