Writing a grammar

[náʼoolkiłí]

After many years of developing a conlang, I figure it's time I reify it by writing a grammatical sketch. I'm curious to hear your advice for going about this.

One problem I'm facing is that everything is currently too abstract. The phonology, morphology, and syntax are all very well fleshed out in my head, but I neither have a large collection of texts nor a proper lexicon to draw examples from that would illustrate the abstract rules. The obvious answers are "translate texts" and "compile a lexicon", but I'm wondering what a smart an efficient way to do that is.

So any tips? Pitfalls to avoid, strategies for organization?

[elemtilas]

After many years of developing a conlang, I figure it's time I reify it by writing a grammatical sketch. I'm curious to hear your advice for going about this.

One problem I'm facing is that everything is currently too abstract. The phonology, morphology, and syntax are all very well fleshed out in my head, but I neither have a large collection of texts nor a proper lexicon to draw examples from that would illustrate the abstract rules. The obvious answers are "translate texts" and "compile a lexicon", but I'm wondering what a smart an efficient way to do that is.

So any tips? Pitfalls to avoid, strategies for organization?

The only thing I'd (strongly) suggest is what you already know: write some original or translate some texts; work on increasing your lexicon a bit. But not having texts or lexicon oughtn't stop you from at least planning and blocking out your grammar. Even if it's just rules without native words or forms to showcase.

As for the nuts and bolts, someone over on CBB asked basically the same question and Salmoneus gave an awesome answer:

Part 1
Part 2

Both are well worth the read!

[Qwynegold]

How do you even keep all that in your head??! The way I do it is just that I think about how the grammar can be divided up into chapters and sections. The most basic things are the most important and ought to be written down first. I think it's easier to first just list your morphology, because there is a finite amount of that, and then write about how these different grammatical categories are actually used. You can write short descriptions at first, and then you just keep fleshing things out.

I don't know if that helps any. ¯\_(ツ)_/¯

[zompist]

You may have read it already, but the LCK includes a guide to writing a grammar.

You don't need a lot of lexicon to write the grammar, but you do need some. Personally I feel you can't write a syntax section without writing full sample sentences. And if you have a complicated morphology, you'll need enough words in each part of speech to illustrate the complications. But 100-200 words should be fine. (Grammatical words like pronouns, numbers, and ad positions will probably add another 100.)

[Moose-tache]

I find myself struggling to find a balance between describing form and describing usage. For example, you probably want to organize your nominal morphology by case endings, talking at some length about the -ka dative, and the -sli oblique, and so on. But then in the syntax you'll want to describe how to form ditransitive clauses, and how prepositional phrases are organized. If you go into sufficient detail you'll end up repeating yourself.

If you organize your grammar based on form, it doesn't give your reader a good sense of how the language actually operates ("English Grammar Chapter 3: The Suffix -ing"). But if you organize it according to the things you can do and how to do them, your information will feel scattered and incoherent ("Here's another thing that uses '-ing.' Hope you're keeping track, 'cuz we aren't!"). And if you do both, as I tend to do, your grammar gets bloated with redundant information pretty quickly ("Chapter 3: Locatives ... Chapter 4: Locating Things").

[So Haleza Grise]

If you go into sufficient detail you'll end up repeating yourself.

I've decided this is OK, and may even be a good thing. In a sense, you can't really trust anyone to read your grammar from cover to cover (not even you) so if you mention things more than once it may be useful. Plus I think lots of cross-referencing of sections is good.

[náʼoolkiłí]

Many thanks for the replies!