Form and Substance in Phonological Development

If we adopt the standard assumption that a phonological grammar is appropriate for expressing regularities in child speech (cf. Hale and Reiss 1998), we are faced with the question of what type of grammar is best suited to this task, as well as the more basic issue of how the relationship between theory and child speech is to be conceived. In generative phonology, connections between theory and child phonology have been made at a formal level and a substantive level (cf. Jakobson 1941/1968 as a precursor). In rule-based theory, formal connections were made by arguing that the phonological generalizations in child speech are best captured in terms of a mapping from an Underlying to a Surface Representation, rather than simply in terms of statements of surface regularities, and that this mapping is accomplished by ordered rules (e.g. Smith 1973, Stampe 1973). Substantive connections were motivated by parallels between the content of child language rules and those found in the languages of the world. For example, learners of English devoice syllable-final voiced obstruents, in a stage that resembles languages like Russian and German. To capture such parallels, Stampe (1973) proposes that there is a set of innate processes that constrains early phonology. Final devoicing would apply in the early phonologies of all learners, would continue to remain in force in the fully developed phonologies of Russian and German speakers, but would be overcome by learners of English. It was quickly recognized that using a purely rule-based framework to deal with child phonology meets the same conspiracy problem that