Wednesday, July 13, 2022

Some thoughts on conlang families and typology

This post details some thoughts about typology, and then comes up with some ideas for Dairwueh-Bryatesle-Sargaĺk. Or, to be long-winded about it:

Some thoughts on typology, relative (and other) pronouns and Dairwueh, Bryatesle and Sargaĺk in light of Indo-European and Uralic 

1. Typological preamble

I have sometimes come across the claim that a very conservative and thoroughly Indo-European feature that has survived in all branches since day one is the interrogative pronouns are also relative pronouns thing. It turns out this is wrong, but it's still interesting enough to spin an idea off it.
 
If anyone is not familiar with the gist of the idea, English has several obvious examples of this:
the man who bought the car
the areas which Caesar conquered
at the time when she arrived
 
Not all languages with relative pronouns have these correlations with interrogative pronouns, c.f. Hebrew, where relative pronouns and interrogative pronouns do not overlap at all.

Even then, of course, the overlap in IE languages is not complete - both English and Swedish have interrogative/relative conflation to some extent, but counterexamples exist: c.f. English 'the house that I bought', Swedish "bandet som hon spelar i" ("the band that she plays in", som being a cognate to English 'some', actually). Some "non-Q-root interrogatives" sometimes also work as relatives, sometimes not. In Swedish, some speakers dislike using "när" as a relative adverb ('when') which is an exceptional interrogative due to not having a historical "qw"-root, and prefer using 'då' ('then', quite clear a cognate and not interrogative as all) as a relative adverb for times, and even more strongly, people prefer "där" over "var" (there, where).

It turns out that this structure might go back to fairly early Indo-European, but must have been lost in several branches and later re-emerged in Germanic and Slavic, for instance, through Latin influence. Many languages in the "near-IE" sphere have also been influenced by the Latinate construction, possibly with Germanic or Slavic vectors of influence: Finnish has began to use mi-interrogatives (which are mainly for non-human referents) as a relativizer, in addition (in some dialects instead of) the joka-relativizers.
 
Joka is, in the singular nominative identical to 'each', but in the other cases they are distinct. First of all, as a determiner it is not (necessarily) inflected for case: joka mies, joka miehen, joka miehellä, (each man, each man's, by each man) ..., although there is also an inflected form jokainen/jokaisen/jokaisella which can be used independently or as a determiner. 

As a relative pronoun, 'joka' is inflected for case, but the root is jo-, and -ka is a suffix that goes after the case suffix or vanishes: mies, joka ... (the man who ...), mies, jo-n-ka ... (the man whose ...), mies jo-lla ... (the man with whom ...), miehet jo-t-ka ... (the men who ...). Apparently, some eastern dialects maybe retain the -ka, but my sources on this are a bit unclear. I am, alas, rather unaware of how relativization is handled in other Uralic languages.
 
Here we actually get a slightly disconcerting thing: Sihler's New Comparative Grammar of Greek and Latin suggests PIE had a relative pronoun yoh-. Why do tantalizing hints at Indo-Uralic jump out at every corner? Why!?!
 

3. Specifics of relativization

I like the overlap with four different types of pronouns that we find here: "each"-quantifiers, some-quantifiers, demonstratives and interrogatives. I also can see a reasonable grammaticalization path for each of these into being used as a relativizer, and it's also possible to find a slightly less demanding grammaticalization path if they start out as something more general and turn into relativizers AND this-or-that on the other hand.

Now, we can take the basic idea "formal conflation of so-and-so with relativizers as a stable feature in a language family".  So-and-so doesn't even have to be pronouns - it could be some auxiliary, it could be some type of pronoun, it could be some conjunction. Relativization happens in many different ways in different languages, but I have decided to go for relative pronouns as one of the strategies present in all BDS languages.

4. Stable features in families

It's clear Indo-European does have some rather stable features over times: the three gender system (say I, writing in one of the languages that has lost it and speaking another that has reassembled it natively), verbal prefixes that are largely overlapping with prepositions, -a as a feminine marker (somewhat less stable), ... so, having a few stable features in Dairwueh-Bryatesle-Sargalk (as well as others in Cwarmin-Ŋʒädär) might very well lend these groups a more "family"-like type of grammatical style.

It is conceivable, however, that this is a type of survivor bias! Proto-Indo-European probably had quite a large amount of features. Just by random chance, we'd expect a handful of features to survive in multiple branches: us thinking of these features as resilient or somehow "characteristic" features of a family might be a mistake. They may very well be features that just have survived the elimination lottery.

4. Implications for Bryatesle-Dairwueh-Sargaĺk

So, as a kind of tribute to Indo-european, I will have a 'similar' correspondence as the one we just ... well, partially debunked or threw on the trashheap or whatever. After a way too long post, here's the nugget:

In BDS languages, relative and reflexive pronouns greatly overlap, and this goes back to proto-BDS.

I think this might require some reworking of Bryatesle, Dairwueh and Sargaĺk.