Phonological representation: Beyond abstract versus episodic

Phonology deals with systematic patterns in the sound structure of languages. Language scientists propose phonological representations with the goal of developing general and predictive understanding of these patterns. We want our understanding of phonology to be general, capturing the implicit knowledge that different speakers of the same language share, as well as similarities and differences across languages. We want it to be predictive, accurately characterizing what previously unseen words and phrases might occur in the future. The goal of having a predictive theory already means that the theory must be synoptic. Only by advancing generalizations is it possible to make predictions about examples that have not yet been seen. Many linguists impute phonological representations to individuals. Of course, the brain does not carry out derivations and calculations in the exact form that scientists write them. The assumption that people have phonological representations amounts to the assumption that when people acquire highly virtuosic capabilities – such as our rapid and adaptable ability to process speech and learn the forms of new words and expressions – they implicitly acquire generalizations that are effective in predicting new data. Scientific theories of phonology in turn succeed insofar as they categorize and generalize along the same lines that people do. The representations used in phonological theory should therefore be well-aligned with the way phonological information is encoded in the mind. Phonological representations are abstract. The representational apparatus of any successful scientific theory – from the theory of evolution to