In this book I have offered both a programmatic account of a substancefree approach to the analysis of phonological patterning and an extended discussion of the sound patterns of a Breton variety. It has thus been my intention not only to provide a framework of assumptions that is capable of providing new theoretical insights but also to show the viability of my chosen approach in facing phonological data. By way of conclusion, in this brief chapter I point out three main themes that a theoretical phonologist (or a phonetician or morphologist interested in how ‘their’ module interacts with phonology) might find useful.

First, I have insisted on a substance-free approach to phonological representation. Under this view, the mapping between phonological symbols and phonetic realisation is not at all fixed, certainly not by any innate mechanism, and is instead language-specific (hence learned), subject to a range of influences both from phonology and other cognitive components, and – in principle – arbitrary. This last point is perhaps the most contentious one. However, the mainstream generative position – that representations are trivially recoverable from pronunciation – appears to be untenable. It is incompatible with a number of empirical facts, such as:

All this, coupled with significant evidence of non-categorical behaviour in many processes formerly assumed to be featural, leads us towards an abandonment of a maximalist approach based on universal features. Once we abandon this tight coupling between representation and substance, we are left with the question of how this mapping can be constrained. Crucially, in a modular framework it appears the correct answer is that, in principle, it cannot. This is because any such restrictions – which are normally connected to factors such as perceptibility (or perceptual reliability), learnability, biases, or even sheer frequency – cannot be phonological. Their role cannot be denied, of course, but there is, I suggest, no good reason to account for these effects in the phonological grammar itself, which concentrates on ‘core’ phonological phenomena such as contrast and morphophonological alternations.