A Cyclic and Multiple Agree account

Abstract

In this paper we propose that probe-goal relations are subject to greater variation than expected, such that both Cyclic Agree and Multiple Agree are possible not only typologically, but also within a single language and a single probe. Cheyenne (Algonquian) has two ways of marking 1st and 2nd person plurality and their conditioning differs between transitives and ditransitives. We propose a hybrid account of agree in order to account for this person/number marking which includes elements of both Cyclic Agree (potentially two probing cycles) and Multiple Agree (multiple simultaneous goals in the first cycle). Evidence against a pure Cyclic Agree account comes from the absence of bleeding effects in transitive forms, which indicates that probing is not always successive cyclic. Evidence against a pure Multiple Agree account comes from the presence of bleeding effects in ditransitive forms, which indicates that all arguments are not always simultaneously probed. Support comes from the properties of other probes in Algonquian and similarities with agreement patterns in other languages, such as Hungarian and Southern Tiwa.