What are some strategies of anaphor binding/coindexation in languages and other strategies to resolve or compensate referent ambiguity?

I should know a good answer to this, as part of my apprenticeship (being a research assistant) was tracking referents in Acehnese discourse for Mark Durie.

The obvious answers I think have already been given. Gender in all its manifold forms, extending to noun classes. Deixis. Politeness strategies and social deixis. Reflexives, including long-distance reflexives and logophors, where a special pronoun refers out from an embedded clause back out to the top clause subject. (Logophoricity). English really struggles with this—

After being acquitted of Veseth’s murder, Red Dog testified at Lilly’s retrial that he, Red Dog, was responsible for shooting Veseth, and that he, Red Dog, had previously lied under oath. Red Dog v. State

I was actually googling for an instance in Red Dog (film), where it was even more awkward.

One thing worth pointing out, which came out of the Acehnese work and the work by Thomas Givon that had inspired it: topicality helps too. Discourse establishes what referents are the main topics being talked about. They tend to be referred to by pronoun rather than full noun phrase, and in fact the use of pronouns confirms that they are the main topics. So pronouns pragmatically are their own disambiguation.

Lojban, as you might well expect, has some whackadoodle strategies, which probably shouldn’t count. It’s an artificial language with an artificial language’s obsession about ambiguity, but it’s taken that a lot further than many. Letters as anaphors is one, which it gets from algebra. I wrote up the perverse motivation there for defaulting to long-distance reflexivity in Folk Functionalism in Artificial Languages: The Long Distance Reflexive vo’a in Lojban.