Sound changes are selected by a bias against morphotactic ambiguity

Abstract This paper hypothesises that there is a universal cognitively and semiotically grounded preference for word-form shapes that signal the morphotactic structure of the word forms they represent. By exploring three historical changes in English, we demonstrate how such a bias might plausibly constrain the actuation and implementation of sound changes. The changes we discuss are the emergence of /z/ as the underlying form in English plural, genitive, and 3rd person present suffixes; Middle English Open Syllable Lengthening; and the emergence and stabilization of irregular past tense and participle forms ending on sonorant+/t/ clusters instead of the expected ones resulting from regular /d/-suffixation. The present study shows how a preference for morphotactically indicative (and unambiguous) word-form shapes can be detected and measured and suggests that the existence of such a preference appears to be highly plausible. We conclude by pointing to more ways in which this hypothesis can be tested and potentially corroborated.