An Interface Model of Phonologically Determined Agreement

It is an assumption of most models of syntax that phonological features are not present during syntactic derivations, thus cannot influence syntactic structure (cf. Pullum & Zwicky 1986, 1988 ). A process that might challenge such an assumption is phonologically determined agreement, where phonological and not syntactic or semantic features agree across elements in a clause or discourse. Such a system exists in Guébie and other Kru languages, where nominal concord markers are determined not by semantic class but by the phonological form of the agreement-controlling noun. Here I describe the typologically remarkable phonologically determined agreement system of Guébie, an undescribed Kru language (Niger-Congo) spoken in Côte d’Ivoire, with the goal of determining an ideal model to account for this and other cases of phonologically determined agreement. I demonstrate here that phonologically determined agreement systems can be modeled without requiring syntactic processes to be sensitive to phonological features. That is, we can maintain syntax as phonology free. Guébie is spoken by approximately 7000 people in seven villages in southwest Côte d’Ivoire. Note that this is not the information provided in Ethnologue (Lewis et al., 2013), where Guébie is listed twice, once as a dialect of Bété-Gagnoa, a nearby cluster of Kru languages, and once as an alternative name for Dida-Lakota. Based on my own findings in the field, Guébie is not mutually intelligible with Bété-Gagnoa, and though it is closely related to Dida-Lakota, there are many phonological and syntactic differences which lead me (and speakers) to consider Guébie a distinct language. The data here comes from original work with Guébie speakers in Berkeley, California and Gnagbodougnoa, Côte d’Ivoire from September 2013 through June 2015. The majority of the data comes from two speakers, a 28-year-old male and a 76-year-old male. Three other male speakers ages 35-52 and two female speakers ages 19 and 30 were also consulted. Data was collected in the form of text and elicitation. In section 2 of this paper I present the facts of phonologically determined agreement in Guébie. In section 3 I argue that neither a purely phonological nor a purely syntactic approach to modeling phonological agreement is satisfying, and in section 4 I propose an interface analysis of phonological agreement that relies on morphological agreement mechanisms and the details of the interactions of morphology with syntax and phonology. In section 5 I extend the model to account for all agreement in Guébie, and in section 6 I show that the predictions of the proposed model account for the other few but existing instances of phonologically determined agreement in the world’s languages.