Non-finite constructions in Khanty: their unity and diversity

Abstract This paper investigates non-finite forms in Kazym Khanty (Ob-Ugric, Uralic). In Khanty, almost all subordination makes use of one of the two non-finite forms: -ti (nfin.npst) or -əm (nfin.pst). We propose that their uses are best treated under the headings ‘bare non-finites, ‘head-agreement non-finites’, and ‘verb-agreement non-finites’. The three classes are defined based on the subject agreement pattern used in non-finite constructions. Different syntactic tests, such as interaction with argument alternations, island effects, and adverbial modification, are used to show the morphosyntactic nature of the classification. The three classes differ in the amount of functional structure they involve: bare non-finites can vary from vP to TP, head-agreement non-finites are of AspP–TP size, and verb-agreement non-finites are CPs dominated by a nominal shell. The findings support both Givón’s Binding Hierarchy and the Implicational Complementation Hierarchy proposed by Wurmbrand and Lohninger, showing that both are also true for languages where all the subordination is done by means of a single form. We also discuss the Khanty data in light of Cristofaro’s Subordination Deranking Hierarchy and argue that the amount of functional structure does not correlate with the degree of deranking/balancing of the embedded clause.