Zompist Bboard Again

bradrn wrote: ↑Thu Feb 15, 2024 7:04 pm The point is that this is a database, not just a bibliography. It’s comparable to other databases, like WALS and PHOIBLE.

And the key thing about being a database is that it combines all the data from all these different sources, in a consistent and searchable form. That makes it possible to answer questions which cannot otherwise be answered: for instance, ‘what are the attested sources of /ŋ/?’. Or, ‘under what conditions has /t/→/s/ occurred?’. Or, ‘which languages in this family have undergone sound change X→Y?’. Or… anyway, you get the idea, hopefully.

To be blunt, this sounds like a programmer-led project more than a conlanger/linguist-directed one.

I'm kind of with Moose on this -- I'd be fine with a lightly annotated bibliography, and probably find it more useful in the long run. You mention WALS and PHOIBLE, but after playing with them excitedly for an hour or so, how often have I ever gone back to either one? Never.

Even if there’s no single monograph, surely there’s at least individual papers which set out the sound changes for individual languages or subgroups? That’s what we’ve mostly been relying on for the other families.

Yes... technically. But unless I'm missing things entirely (which is entirely possible), not in a way that really moves the needle significantly. There's new and better stuff coming out these days, though, so maybe checking back in a couple years would pay off. (Sounds weird to say, but, really!)

fusijui wrote: ↑Fri Feb 16, 2024 8:27 pm

bradrn wrote: ↑Thu Feb 15, 2024 7:04 pm The point is that this is a database, not just a bibliography. It’s comparable to other databases, like WALS and PHOIBLE.

And the key thing about being a database is that it combines all the data from all these different sources, in a consistent and searchable form. That makes it possible to answer questions which cannot otherwise be answered: for instance, ‘what are the attested sources of /ŋ/?’. Or, ‘under what conditions has /t/→/s/ occurred?’. Or, ‘which languages in this family have undergone sound change X→Y?’. Or… anyway, you get the idea, hopefully.

To be blunt, this sounds like a programmer-led project more than a conlanger/linguist-directed one.

Quite possibly… all I can say is that when I first discussed this idea, no-one seemed to have any significant objections.

You mention WALS and PHOIBLE, but after playing with them excitedly for an hour or so, how often have I ever gone back to either one? Never.

On the contrary, I use WALS all the time when I’m conlanging, to get an idea of what languages do. I use the current ID for much the same reason, but it’s hard when its sound changes are so unreliable and incomplete.

(PHOIBLE, not so much… though its client Pshrimp was very useful, when it was working.)

Even if there’s no single monograph, surely there’s at least individual papers which set out the sound changes for individual languages or subgroups? That’s what we’ve mostly been relying on for the other families.

Yes... technically. But unless I'm missing things entirely (which is entirely possible), not in a way that really moves the needle significantly. There's new and better stuff coming out these days, though, so maybe checking back in a couple years would pay off. (Sounds weird to say, but, really!)

Fair enough.

bradrn wrote: ↑Mon Feb 12, 2024 7:03 pmhow reliable is this source? It would be good if you could track down the original source where these sound correspondences were published

Yeah fair question, just wanted to put out something to start with (that we might be able to later cite in more detail). And I know the conlangers of the forum will appreciate any showcase of examples already. *p >> s for example is pretty cool in any case (and FWIW I believe I've recently discovered the same in a different language group I've been studying and would have been more hesitant about it if I didn't already know of precedents).

bradrn wrote: ↑Thu Feb 15, 2024 7:04 pmEven if there’s no single monograph, surely there’s at least individual papers which set out the sound changes for individual languages or subgroups? That’s what we’ve mostly been relying on for the other families.

This only works in very large and relatively "continuous" families like Austronesian or Bantu (edit: or also families like Romance or Indo-Aryan where, if not quite the protolanguage, then something very close to it is actually attested), where you can trust having gotten a good picture of the proto-language together independently of some particular language you're studying. I am not even ankle-deep into Tungusic myself but, if it's anything like the research history of e.g. Uralic, the primary literature is instead going to be very piecemeal things like "I propose the existence of *x, here's what happens to it in the following eight varieties" (the fate in six more varieties only settled in five other papers later on) or "here's several observations about what happens to consonant clusters in Udege".

In these smaller families, often the literature is still good enough that you could reliably assemble an overall outline with some work and discernment; and also, if you did that with proper sourcing, explicit commentary and explicit examples, the result would probably be a publishable paper all by itself already.

Tropylium wrote: ↑Sat Feb 17, 2024 9:02 am*p >> s for example is pretty cool in any case (and FWIW I believe I've recently discovered the same in a different language group I've been studying and would have been more hesitant about it if I didn't already know of precedents).

(Aside: *p > *ts is a conditioned change in some varieties of Ryukyuan according to Thorpe (1983).)

Tropylium wrote: ↑Sat Feb 17, 2024 9:02 am I am not even ankle-deep into Tungusic myself but, if it's anything like the research history of e.g. Uralic, the primary literature is instead going to be very piecemeal things like "I propose the existence of *x, here's what happens to it in the following eight varieties" (the fate in six more varieties only settled in five other papers later on) or "here's several observations about what happens to consonant clusters in Udege".

That’s fine, then we can transcribe that paper as *x→phoneme1 in Variety A, *x→phoneme2 in Variety B, and so on. It’s not ideal, but it’s the best we can do with what it is. We already have some Skouic changes which are basically this.

This is also really the reason I don’t want to delve too deeply into editorialising multiple sources a single ‘canonical’ set of sound changes. When you have multiple sources like this, how do you figure out which are correct? Well… you have to use the comparative method. That’s already a major research project. And I don’t think we have the time or the resources to do that.

bradrn wrote: ↑Sat Feb 17, 2024 5:29 pm

Tropylium wrote: ↑Sat Feb 17, 2024 9:02 am I am not even ankle-deep into Tungusic myself but, if it's anything like the research history of e.g. Uralic, the primary literature is instead going to be very piecemeal things like "I propose the existence of *x, here's what happens to it in the following eight varieties" (the fate in six more varieties only settled in five other papers later on) or "here's several observations about what happens to consonant clusters in Udege".

That’s fine, then we can transcribe that paper as *x→phoneme1 in Variety A, *x→phoneme2 in Variety B, and so on. It’s not ideal, but it’s the best we can do with what it is

Obviously we can, but I mean though… if you don't insist on citing things from the most primary literature there is, these various tertiary sources like course materials or dictionary prefaces can have a lot more things in one place, even if they might not themselves have all the proper citations to the actual primary literature (or might simply not have them broken down conveniently), might not have examples readily listed, might not discuss unclear exceptions, etc. Adding primary citations too is good for bibliographic purposes, but may have a very low ratio of effort : progress-in-exactness or progress-in-coverage.

And then there's a simple user-friendlyness issue: it would be nice to not have the sound changes for a given language split in too many different sections. I like your current choice of treating secondary & tertiary sources as headers — "the sound changes from Proto-Fooic to Swampy Foo, according to N.N.". Primary literature that contributes just little bits should probably still be just in-line citations though, especially if there is no secondary literature. Some language varieties don't really even have tertiary literature yet!

bradrn wrote: ↑Sat Feb 17, 2024 5:29 pmWhen you have multiple sources like this, how do you figure out which are correct?

That's another reason to prefer tertiary sources to primary sources: if it's assembled by some topic expert, they will have already done some of this editorializing, even if the process of it is not spelled out explicitly (as it might be in some in-depth secondary source "Historical Phonology of Foo", if it existed).

Tropylium wrote: ↑Sun Feb 18, 2024 7:19 am

bradrn wrote: ↑Sat Feb 17, 2024 5:29 pm

Tropylium wrote: ↑Sat Feb 17, 2024 9:02 am I am not even ankle-deep into Tungusic myself but, if it's anything like the research history of e.g. Uralic, the primary literature is instead going to be very piecemeal things like "I propose the existence of *x, here's what happens to it in the following eight varieties" (the fate in six more varieties only settled in five other papers later on) or "here's several observations about what happens to consonant clusters in Udege".

That’s fine, then we can transcribe that paper as *x→phoneme1 in Variety A, *x→phoneme2 in Variety B, and so on. It’s not ideal, but it’s the best we can do with what it is

Obviously we can, but I mean though… if you don't insist on citing things from the most primary literature there is, these various tertiary sources like course materials or dictionary prefaces can have a lot more things in one place, even if they might not themselves have all the proper citations to the actual primary literature (or might simply not have them broken down conveniently), might not have examples readily listed, might not discuss unclear exceptions, etc. Adding primary citations too is good for bibliographic purposes, but may have a very low ratio of effort : progress-in-exactness or progress-in-coverage.

[…]

That's another reason to prefer tertiary sources to primary sources: if it's assembled by some topic expert, they will have already done some of this editorializing, even if the process of it is not spelled out explicitly (as it might be in some in-depth secondary source "Historical Phonology of Foo", if it existed).

I don’t actually have any problem with citing tertiary sources, per se. I’d greatly prefer primary or secondary sources, of course… but if a tertiary source is the best we can get, so be it. If I implied otherwise, my apologies, I did not mean to do so.

(It needs a proper citation, of course, which is why I asked for the reference earlier. But then again, everything else needs a proper citation too.)

The other thing is, incidentally, that there’s nothing to stop us from including both the secondary and the tertiary sources. Indeed, if we want to build a complete database, we should really be doing that anyway. It will result in duplicate sound changes, but there’s no avoiding that — indeed, we have duplicates already.

In any case, Man in Space has emailed some linguists to see what they would like out of a new ID, which should hopefully resolve such issues. Until they respond, I’m going to hold off on making any big decisions about this stuff. (Though, by all means, please keep on collecting sound changes either way!)