Tone Shift is Phonological, and Phonology is not Emergent: Linguistic knowledge and the acquisition of tone in Kinande

Abstract: Recent work by Archangeli and Pulleyblank (2014) has argued that phonological systems are “emergent” in the sense that they can be acquired without the benefit of a rich innate human language faculty. In support of this view, A&P argue that a complex system of tonal alternations in Kinande (Bantu), standardly analyzed as resulting from a phonological rule (namely leftward H tone shift) made available to the learner via some form of phonology-specific knowledge, actually results from allomorph listing and selection which draws only on domain-general statistical learning abilities. In this talk, I argue that when the tonal (and morphological) system of Kinande is analyzed properly, it actually constitutes strong evidence against the emergent position. Learners do not simply memorize surface allomorphs, but project instead a rich system of morphology-phonology interactions that are only hinted at in the basic data they are exposed to. This suggests an acquisition pathway that is deeply guided by innate language-specific knowledge.