The Index Diachronica

[bradrn]

Thanks to Vilike, we now have some Bantu sound changes! (Specifically, for Herero.) I don’t have access to the source, so if anyone else can access it, I’d appreciate if they could double-check that everything’s correct. (Post in the Bantu thread, please.)

But also: this is the first time someone’s tried the submission process, and it proved a lot more complicated than I’d anticipated. I hadn’t realised just how much room it left for little mistakes and errors. For now I’d like a few more people to try it out, to see if these difficulties can be ironed out over time. But we may need to rethink this use of GitHub if it proves too difficult.

[bradrn]

I came across a nice paper from Blust on Palauan, so I’ve transcribed the sound changes there and added them to the Austronesian page. There’s some more in Flora 1974, but I actually can’t understand the notation there — it would be great if anyone is able to transcribe that! (For that matter, it would be great if other people could review the Palauan changes I just uploaded, too.)

[Darren]

Yo I'm trynna make a request thingy, am I supposed to add new stuff to the thingy or replace the stuff that's there? I got no idea how Github works (or what it is for that matter). I feel like I'm probably supposed to add new stuff.

[bradrn]

Yo I'm trynna make a request thingy, am I supposed to add new stuff to the thingy or replace the stuff that's there? I got no idea how Github works (or what it is for that matter). I feel like I'm probably supposed to add new stuff.

I’m not quite sure what the ‘stuff’ or the ‘thingy’ is meant to be, but the basic workflow is as follows:

1. Make a ‘fork’ of the repository on GitHub, to get your own copy of the files which you can edit.
2. Make whichever changes you need to make to the files. For this purpose you might find GitHub’s web editor a bit easier to use, though I prefer to make the changes on my computer and commit them to the repository once I’m done.
3. Once you’ve made your changes, there should be a big button saying ‘Compare & Pull Request’ (or something like that). Click the button, then fill in the form to submit your changes to the main repository.
4. I will review your changes, and either accept them to the repository as-is, or comment to ask that you make further edits.

[Darren]

I’m not quite sure what the ‘stuff’ or the ‘thingy’ is meant to be, but the basic workflow is as follows:

1. Make a ‘fork’ of the repository on GitHub, to get your own copy of the files which you can edit.
2. Make whichever changes you need to make to the files. For this purpose you might find GitHub’s web editor a bit easier to use, though I prefer to make the changes on my computer and commit them to the repository once I’m done.
3. Once you’ve made your changes, there should be a big button saying ‘Compare & Pull Request’ (or something like that). Click the button, then fill in the form to submit your changes to the main repository.
4. I will review your changes, and either accept them to the repository as-is, or comment to ask that you make further edits.

I reckon I've forked the thingy successfully at least and I can probably do the submitting of pull stuff, it's the bit in between (with all the adding of the changes and such) that's confusing. Actually, I'll just do whatever and submit it and if it's wrong you'll know what I'm talking about, and if it isn't then we're all good.

[bradrn]

I’m not quite sure what the ‘stuff’ or the ‘thingy’ is meant to be, but the basic workflow is as follows:

1. Make a ‘fork’ of the repository on GitHub, to get your own copy of the files which you can edit.
2. Make whichever changes you need to make to the files. For this purpose you might find GitHub’s web editor a bit easier to use, though I prefer to make the changes on my computer and commit them to the repository once I’m done.
3. Once you’ve made your changes, there should be a big button saying ‘Compare & Pull Request’ (or something like that). Click the button, then fill in the form to submit your changes to the main repository.
4. I will review your changes, and either accept them to the repository as-is, or comment to ask that you make further edits.

I reckon I've forked the thingy successfully at least and I can probably do the submitting of pull stuff, it's the bit in between (with all the adding of the changes and such) that's confusing. Actually, I'll just do whatever and submit it and if it's wrong you'll know what I'm talking about, and if it isn't then we're all good.

Yep, just do whatever you feel is right! I’ll let you know if there’s any problems. (Indeed, that’s the whole point of the pull request process.)

[Tropylium]

Does anyone know how to contact Tropylium? I want to ask them what the capital I in the Finnic sound changes means but their last activity here was 2 years ago.

I do check back here every … few years … currently, though this would probably have been more often if I knew about this sub-forum — a great project, I should some day submit various thoughts I have about how it might be turned into a serious academic resource and/or other individual improvements (I've been working at the tech / science interface for academic database building projects for a couple years now and it's been a great learning experience). For future reference, if anyone's got questions or just wants to chat about Uralic diachrony or other stuff, I'm reachable on Discord nowadays, same username.

anyway, I means the same as all capital vowel letters in Finnic reconstructions: front/back vowels doublets. Looking back where this appears I have no idea why I've used it though, since the sound changes in question come later than the putative *i / *ï merger and in positions (or at least etyma) where only *i is known.

[bradrn]

Welcome back, Tropylium! Please feel free to start a Uralic thread in this subforum to discuss sound changes for that family.

I should some day submit various thoughts I have about how it might be turned into a serious academic resource

It’s definitely something I’ve been thinking about seriously. That’s one reason why I’m placing such an emphasis on rating the reliability of sources.

For future reference, if anyone's got questions or just wants to chat about Uralic diachrony or other stuff, I'm reachable on Discord nowadays, same username.

I’m on Discord too, I’ll message you now!

[Tropylium]

various thoughts I have about how it might be turned into a serious academic resource

Straight off it's easy to say though that the most important point for more-than-conlanging usability would be improved sourcing. And I don't mean settling just for giving or requiring "a" source for a sound change, but the ability to have an explicit bibliography and at least some amount of comments on it in basically prose. There's two main reasons:

1. Debate over the sound change. To take for an example something like Brugmann's Law…
1.1. first, what exactly is the input and the output? Is it merely *o > *ō and then followed by the usual Indo-Iranian *ō > *ā? Or is it *o > *ā immediately? Is it perhaps *ɑ > *ɑː or *oˑ > *oː or some other phonetically elaborate interpretation? Could it even be just lack of *ā > *o (if following the idea that *e *o < pre-PIE *a *ā)?
1.2. what's the conditioning? Do all open syllables work or are there any blocking consonants? What counts as an "open syllable" precisely? Does ablaut block this (also does this maybe mean ablauting "*o₁" and non-ablauting "*o₂" were not the same vowel)?
1.3. what's the evidence anyway? what are examples, what are sources for them, or perhaps sources for refutations / debate of if something is an example?

2. Research-historical interest. Who first proposed this? Who named it (if applicable)? is there one or more "standard" locus classicus sources that has all the generally accepted examples and maybe lists of more dubious cases? which overviews or introductions mention the sound change (and maybe, which conspicuously do not)?

A lot of "big-name" sound laws can be easily discussed for a few pages and can end up with bibliographies a page or two long; for Indo-European big-name sound laws in particular, there's already a monograph The Laws of Indo-European of just this (N. E. Collinge, 1985 = Current Issues in Linguistic History 35; should be findable on Libgen I think).

Initially anything like these issues is perhaps a reason to avoid more debated reconstruction stages and stick to more recent ones where it's fairly clear what's going on; in the long run though, it may also mean that even something like "this was proposed by Ehret as being from Proto-Afro-Asiatic to X" may not require throwing the data away entirely, and it could be just corralled under some status of "dubious" or "unsupported in the rest of literature", which makes poorly enough supported changes by default not come up on general searches like "show me sound changes from *ŋ".

[bradrn]

various thoughts I have about how it might be turned into a serious academic resource

Straight off it's easy to say though that the most important point for more-than-conlanging usability would be improved sourcing.

Indeed, this is a really core part of how we’re structuring the new ID. Specifically, we’ve abandoned the idea of reviewing all the literature to get one canonical set of sound changes for each language, because it runs into all the problems you’ve mentioned.

Instead, our current approach is to treat this as more of a transcription project: an attempt to take all the sound changes which have been suggested in the diachronic literature, and put them in one consistent and searchable database. This comes with its own problems, most notably duplicate sound changes between sources. But I think the advantages strongly outweigh the disadvantages.

(In case you missed the link, you can see what this looks like so far at https://bradrn.com/index-diachronica/. The website is still pretty basic, but should give some idea of our approach.)

Of course, in many ways doing a proper review like you mention is preferable. But to get to that point, in any case you’d need to do all this transcription work beforehand. So we might as well focus on collecting that raw data, and leave it to others to make a coherent story out of it.

Initially anything like these issues is perhaps a reason to avoid more debated reconstruction stages and stick to more recent ones where it's fairly clear what's going on; in the long run though, it may also mean that even something like "this was proposed by Ehret as being from Proto-Afro-Asiatic to X" may not require throwing the data away entirely, and it could be just corralled under some status of "dubious" or "unsupported in the rest of literature", which makes poorly enough supported changes by default not come up on general searches like "show me sound changes from *ŋ".

We’re doing something like this already. One of the criteria on which we’re ranking sources is ‘consensus’, which is a subjective measure of pretty much exactly this. (All the ranking criteria are subjective, really, but I think it‘s important to allow for this sort of filtering.)

[bradrn]

Clearly, my clever strategy of ‘just force everyone to use GitHub’ has not worked. I have a few other things going on at the moment, but perhaps I’ll just go back to transcribing the collected sound changes in this subforum and putting them into the repository myself. We can work out an easier method to contribute sound changes at the same time, if it’s necessary.

[Zju]

Clearly, my clever strategy of ‘just force everyone to use GitHub’ has not worked. I have a few other things going on at the moment, but perhaps I’ll just go back to transcribing the collected sound changes in this subforum and putting them into the repository myself. We can work out an easier method to contribute sound changes at the same time, if it’s necessary.

I don't know about others, but for me this is still just a hobby, which takes a back seat to more important stuff (such as jobhunting... bleugh) so getting around to doing it takes a while.

But I digress. What I'm getting at is that if when I get to it, I anticipate I'll find pull requests more straightforward.

[Tropylium]

By the way, how many of you have read / browsed Martin Kümmel's Konsonantenwandel (2007, Wiesbaden: Reichert)? At its core it's basically a 300-page Index Diachronica of consonant sound changes in IE + Semitic + Uralic, primarily in order by type of sound change, but already has also a by-language index that looks something like this for another 80 double-column pages:

osk.
[0 > t /n_s Buck 1904: 72]
n > l /_l Buck 1904: 70
[x > s /_s$ = /_s#, /_sC Meiser 1986: 169-172]
n > 0 /_s>s:# Buck 1904: 72 {präsynkopal}
d > 0 /#_j Buck 1904: 84 {früh}
w > 0 /d_ Meiser 1986: 185f. {fraglich}
n > 0 /V>Ṽ_t Buck 1904: 70f.
t > 0 /s_m,l Buck 1904: 87
p > m /_m Buck 1904: 79
[x > 0 /l,n_t Buck 1904: 91f.]
C > CC /_j Buck 1904: 66, 99
C > CC /_r,w Buck 1904: 99
s > ss /_t Buck 1904: 99
s > r /r_ Buck 1904: 76 {nach Synkope}
z > 0 /V>V̄r_ Buck 1904: 76; Meiser 1986: 172f.

[Zju]

n > 0 /_s>s:# Buck 1904: 72 {präsynkopal}
n > 0 /V>Ṽ_t Buck 1904: 70f.
z > 0 /V>V̄r_ Buck 1904: 76; Meiser 1986: 172f.

Are those supposed to mean this?:
ns → sː / _#
Vn → Ṽ / _t
Vrz → V̄r

[Tropylium]

Suspect yes.

[Man in Space]

Speaking of the Konsonantenwandel, the ID got cited and the authors specifically chose it over the Konsonantenwandel due to a) ease of use and b) it has vowels.

Come to think of it…let me e-mail the authors and ask them what they would like to see out of it.

EDIT: Just did.

[Man in Space]

Since I have the domain for the ID parked, should I maybe look into setting up access for everyone? That way we’d have a clean testbed or sandbox.

[bradrn]

But I digress. What I'm getting at is that if when I get to it, I anticipate I'll find pull requests more straightforward.

This was exactly my point. They’re an extra barrier which is slowing down work.

Speaking of the Konsonantenwandel, the ID got cited and the authors specifically chose it over the Konsonantenwandel due to a) ease of use and b) it has vowels.

I am incredibly surprised to discover that. Emailing the authors is certainly a great idea, and I’d be fascinated to hear what they say!

(Perhaps it’s a stretch, but could you even invite them to join the conversation on this subforum…?)

EDIT: On further reading, I note that they cite Schweikhard & List 2020 against the reliability of the old ID. It seems their arguments are essentially the same as mine, that the main issue is the use of unreliable sources. Perhaps it’s worth emailing them too and getting their ideas on what would be most useful?

EDIT2: I did a quick search on Google Scholar to see if anyone else cited it. This turned up Rubehn 2022, which again uses it as input to a neural network while warning about its reliability. (And seems to hold against it that we’re conlangers…) There’s also Anderson et al. 2018, which mentions its existence but doesn’t actually use it in any way.

Since I have the domain for the ID parked, should I maybe look into setting up access for everyone? That way we’d have a clean testbed or sandbox.

That would be great, thanks!

[Man in Space]

I am incredibly surprised to discover [the citation in the ACL paper].

It also got used as a comparand and basis for the CLTS (your Anderson et al.), and Kevin Alberto López Porcheron used it in his thesis (he's a nice guy, I've messaged with him). It also got turned into a ChatGPT plugin, which is extremely cool.

Emailing the authors is certainly a great idea, and I’d be fascinated to hear what they say!

I did. I also reached out to Jeffrey Punske—he taught me for a few classes under him at OU and said he'd put a good word in with potential contributors. (I should see if Drs. Haag and Herrick might be interested…)

(Perhaps it’s a stretch, but could you even invite them to join the conversation on this subforum…?)

I did. I swung for the fences. We'll see what happens.

On further reading, I note that they cite Schweikhard & List 2020 against the reliability of the old ID. It seems their arguments are essentially the same as mine, that the main issue is the use of unreliable sources. Perhaps it’s worth emailing them too and getting their ideas on what would be most useful?

I will do so!

I did a quick search on Google Scholar to see if anyone else cited it. This turned up Rubehn 2022, which again uses it as input to a neural network while warning about its reliability. (And seems to hold against it that we’re conlangers…)

I didn't quite read it as hostile as that. (One wonders if some of them are put off that it took a hobbyist bulletin board to build this instead of some formal academic program…)

Since I have the domain for the ID parked, should I maybe look into setting up access for everyone? That way we’d have a clean testbed or sandbox.

That would be great, thanks!

I use DreamHost if it's relevant—tell me what you need me to do and I'll try to figure it out.

[bradrn]

Emailing the authors is certainly a great idea, and I’d be fascinated to hear what they say!

I did. I also reached out to Jeffrey Punske—he taught me for a few classes under him at OU and said he'd put a good word in with potential contributors. (I should see if Drs. Haag and Herrick might be interested…)

(Perhaps it’s a stretch, but could you even invite them to join the conversation on this subforum…?)

I did. I swung for the fences. We'll see what happens.

On further reading, I note that they cite Schweikhard & List 2020 against the reliability of the old ID. It seems their arguments are essentially the same as mine, that the main issue is the use of unreliable sources. Perhaps it’s worth emailing them too and getting their ideas on what would be most useful?

I will do so!

Thanks very much!

(As it happens, I was in Canberra for a conference last year, and I did seriously consider contacting some linguists at ANU to get their thoughts on this project. But I decided it was too immature at that stage to do so. Perhaps I should have…)

I did a quick search on Google Scholar to see if anyone else cited it. This turned up Rubehn 2022, which again uses it as input to a neural network while warning about its reliability. (And seems to hold against it that we’re conlangers…)

I didn't quite read it as hostile as that. (One wonders if some of them are put off that it took a hobbyist bulletin board to build this instead of some formal academic program…)

Oh, I’m quite sure that’s what it is. Linguists can get dreadfully snobby.

Since I have the domain for the ID parked, should I maybe look into setting up access for everyone? That way we’d have a clean testbed or sandbox.

That would be great, thanks!

I use DreamHost if it's relevant—tell me what you need me to do and I'll try to figure it out.

Not sure, actually — I’ve never used DreamHost. Perhaps it would be easiest if you just DM’d me the login details.