In this dissertation, I argue that phonology is ---at least partly--- grounded in phonetics, and that the phonetic information that is relevant to phonology must be given dedicated scalar representations. I lay out the foundations of a theory of such representations, which I call SUBFEATURES. A typology and case-study of SUBPHONEMIC TEAMWORK, a kind of multiple-trigger assimilation driven by partial coarticulatory effects, serves as the empirical basis of this proposal. I argue, on the basis of instrumental evidence, that such partial coarticulatory effects are relevant to the phonological grammar, and must accordingly be represented in it.

In doing so, I take a stand in several debates that have shaped phonology. First, the debate surrounding phonological substance: I argue against a substance-free approach (cf. Blaho 2008 and references therein) by showing that some phonological phenomena (here, subphonemic teamwork) require a phonetically grounded, or ``natural", approach. Furthermore, within the field of phonetically based phonology, I argue in favor of maintaining a separation between phonology and phonetics, which are conceived of as distinct components of the grammar that communicate through the mediation of abstract phonetic knowledge. Finally, the theory I propose is representational: phonetic knowledge, which is what allows phonology access to some phonetic information, is reified into subfeatural representations that can be manipulated by the phonological grammar.