How local is allomorphy?

A long-standing issue in morphological theory is allomorphy (e.g. good vs. bett-er, go vs. went, book-s vs. sheep-∅) and the potential locality conditions that hold between the trigger and target in an allomorphic conditioning relation (e.g. Embick 2010, Bobaljik 2012, Moskal & Smith 2016, Choi & Harley 2019). While a prominent proposal is that allomorphic relations must be very local, i.e. within the same complex head (Bobaljik 2012, Bobaljik & Harley 2017), there are well-known counter-examples to this (e.g. Toosarvandani 2016, Oseki 2016).

In this talk, I report on a recent project with Bob Offer-Westort (UChicago) in which we revisit this question from the perspective of the Cushitic language Bidhaawyeet (Beja). Object clitics in Bidhaawyeet relative clauses appear to show non-local allomorphy conditioned by the head of the relative clause. While this appears to be non-local allomorphy, we will argue that this relation is actually local due to general concord processes in the language. Furthermore, it will be shown that the wider distribution of object clitic forms outside of relative clauses supports the idea that this is local allomorphy constrained to the morphological word. The analysis we develop gives us a way of understanding apparent exceptions to the otherwise stringent locality conditions on allomorphic relations.

Location: Royce Hall 362