On Language

In the process of working on my random language generator, I have discovered something: human languages are very complex things, but the do follow some rules. What those rules are, however, is often pretty damn difficult to pin down.

See, even something as "simple" as Morphosyntactic alignment can turn out to be intensely complicated. Originally I assumed it'd be as simple as an ergativity parameter, but this quickly mutated into a whole mess of parameters, one for each type of alignment. This still failed to be either elegant or easy to implement.

I have now developed a new method, which has the advantage of (if anything) erring on the side of more permissive of random and arbitrary variation. See, each verb will exist as a data structure that, when initialized, queries the language's grammar, finds out how it's supposed to be aligned, builds a list from which cases are assigned to its arguments, and then waits until it is called with its arguments.

The verb can be told to build different alignments based on the voice it's called in, the inflections it's taking, or the category it belongs to, allowing phillipine, fluid-S, and split-S alignment systems to be generated without resorting to a "phillipine" or "split ergativity" parameter.

Just though you might want to know.

Put some pants on,
-Stinja