Abstract
<jats:p>A host of common and recurring phonological patterns, from both typological and diachronic perspectives, lend weight to the presence of a limited set of universals in phonology. At the same time, languages vary greatly in terms of the size of phonemic inventories, their phonetic content, and how segments and features can be ordered and manipulated. These observations run counter to each other in debates on the existence and potential usefulness of universals in modeling the mechanisms, and explaining the outcomes, of sound change. Still, all languages have phonologies that consist of a definable set of contrastive elements and that occur within a range of fairly predictable forms. There is a general characteristic of underdeterminacy between phonetics and phonology, which is captured in the relationship between ‘language‐internal’ universals, both formal and substantive, and how those universals interface with ‘language‐external’ aspects of speech. Specifically, formal universals in phonology are the principles that guide the structure of contrastive representations. The content of these representations, that is, substantive universals, comprise abstract features that are translated into speech in a modular fashion. The conversion from feature to surface (phonetic) form is learned and specific to a given language community, in a particular sociohistorical context, yet constrained by the architecture of the modular system. Modularity therefore unifies learned characteristics of speech and universal (formal and substantive) properties of the phonology, and models how the contrasts of sound systems are affected when those relationships change over time and space.</jats:p>