Parallel Mapping Circuitry in a Phonological Model

Abstract M3P is a computational model of phonology based on parallel mapping matrices. Rules govern the mappings between successive levels of representation of an utterance, from morphphonemic, to phonemic, to phonetic. The mapping matrix architecture performs several types of sequence manipulation. M3P shows that human phonological processing can take place in a highly constrained architecture, using purely feed-forward circuitry and tight limits on depth of derivations.