Long-distance reflexivization in Chechen and Ingush
Johanna Nichols
University of California, Berkeley
Ingush (Northeast Caucasian) and its close sister Chechen display
long-distance reflexivization so regular and so extensive that they
can be taken as definitive of a canonical type of
long-distance-reflexivizing language. Any subject (whether in main or
non-main clause) can (and generally does) reflexivize any coreferent
in the same or any lower clause. These are clause-chaining languages,
and narrative texts abound in examples where the first clause in a
chain contains a reflexive whose antecedent occurs many clauses
later, commonly with other, non-coreferential subjects in the
intervening clauses. There is no fade-out effect, no matter how long
the chain. Distant reflexivization is comparably regular and strong
in complementation stacks, which are rare in natural texts but can be
elicited.
Two distinct reflexivization chains in different person categories can co-occur and intertwine in the same sentence, the only constraint being that two overt reflexive pronouns with different antecedents cannot co-occur in the same clause. Chechen-Ingush reflexive pronouns are morphologically simplex but distinguish person lexically, and this indicates that it is not literal morphological complexity but the ability to distinguish person that makes multiple reflexivization possible. The impossibility of multiple reflexivization into the same clause shows that long-distance reflexivization is very different from ordinary long-distance anaphora.
Both ordinary and logophoric reflexivization occur in Chechen, and the different treatment of person reference in the two types, as well as the ability of logophoric but not ordinary long-distance reflexivization to cross tensed sentence boundaries, show that, in principle, long-distance reflexivization cannot simply be equated with logophoricity.
Blocking factors include certain combinations of first and secondperson; and purpose clauses, further evidence that long-distance reflexive is not the same thing as logophoricity.
Preliminary results of a survey now underway indicate that canonical long-distance reflexivization can be expected in languages with (1) clause chaining, (2) no category of person, (3) no one-to-one morphological signaling of subject (e.g. ergative languages), and (4) reflexive pronouns which (5) can distinguish person. Languages having most or all of these properties except (2) typically have switch reference.
I hope with this typologically oriented paper to show that Chechen
and Ingush provide an ideal natural laboratory for more theoretical
work on long-distance reflexivization.