Zompist Bboard Again

Moose-tache wrote: ↑Sat Aug 05, 2023 6:47 pm So a reference grammar only needs to do these things in emergency situations in which a reference grammar is required to do more things other than be a reference grammar. You might as well ask how to design a car so that it can also double as an underwater shelter. Sure, that could be very useful in certain edge cases, but it tells you very little about good car design overall.

But people named the edge cases, and they're much more applicable to conlangs than the underwater shelter scenario is to cars - most conlangs will be like underresearched natlangs, the reference grammar plus maybe a lexicon will be the only documentation that will ever be available. I mean, even zomp didn't write more than a couple of lessons for Verdurian. So there is an argument for the reference grammar doubling also as an introduction.
(Now, if this is just about finding in the tin what's written on the tin, maybe just name it "An Introduction to Conlang X", and make Moose happy?)

For conlangs, it's an exercise in pure optimism to write a learning textbook.

Linguists do have another model for us, though— what I've called a grammatical sketch. I have a couple of books full of them— Comrie and Lyovin for the world, Woodard for ancient languages. These can be pretty thorough— e.g. the chapter on Akkadian is 70 pages of small type.

zompist wrote: ↑Sun Aug 06, 2023 4:10 pm Linguists do have another model for us, though— what I've called a grammatical sketch. I have a couple of books full of them— Comrie and Lyovin for the world, Woodard for ancient languages. These can be pretty thorough— e.g. the chapter on Akkadian is 70 pages of small type.

I think most so-called ‘grammars’ for conlangs should really be called grammatical sketches, in terms of their size and detail. It’s a rare conlang which can stand up to 200 pages of grammatical analysis — and that’s on the small side for natlang reference grammars.