The Index Diachronica

[Darren]

Page has very kindly sent me a copy of the Fayu grammar, so I'll shortly be posting an updated comparative wordlist with a lot more Fayu terms. (She also hints that a Fayu-based PhD thesis may be appearing in the future) This means that I now have at least one pretty good resource for all branches other than Duvle and Far West, and one more language with consistent tone marking on lemmas.

• For conlangers such as ourselves, it provides a way to quickly gain confidence as to which sound changes are attested or not.
• For historical linguists, it could act as reliable database with which to test phonological generalisations (as opposed to the unsupported claims I regularly see in the literature).
• For linguists more generally, it could be a place to make their diachronic work more widely known.

But I’m happy to hear other suggestions about what we should aim for!

I think your distinction of target audiences is important. I was assuming that this was mostly intended for conlangers rather than as a published work for academics, but I appreciate your more ambitious aims. I think it's fine for amateur work and curated sound changes like my LP work to stay on the board for fellow conlangers.

I’ve finished my mockup of some Polynesian changes: https://bradrn.com/files/polynesian-biggs-mockup.html. This is compiled mostly from Biggs 1978, plus the changes for Takuu, Nukumanu and Nukeria from Davletshin 2015. Please let me know your thoughts! (Comments on the sound changes themselves can go in the Polynesian thread, general comments can go here.)

I think this is a good format. The links for shared changes are a nice touch. One minor nitpick is that for *w, ‘was probably bilabial and possibly a voiced fricative rather than a semi-vowel’ implies to me [w~β̞~β] rather than just [w~β].

[bradrn]

Page has very kindly sent me a copy of the Fayu grammar, so I'll shortly be posting an updated comparative wordlist with a lot more Fayu terms. (She also hints that a Fayu-based PhD thesis may be appearing in the future)

Great to hear! Could you send a copy to me too? (You should have my email already, I think.)

• For conlangers such as ourselves, it provides a way to quickly gain confidence as to which sound changes are attested or not.
• For historical linguists, it could act as reliable database with which to test phonological generalisations (as opposed to the unsupported claims I regularly see in the literature).
• For linguists more generally, it could be a place to make their diachronic work more widely known.

But I’m happy to hear other suggestions about what we should aim for!

I think your distinction of target audiences is important. I was assuming that this was mostly intended for conlangers rather than as a published work for academics, but I appreciate your more ambitious aims.

Yes, I’m very excited about the possibilities. If you poke around the literature, there’s already linguists who have tried doing something like this, but it appears their efforts petered out at some point. But we’ve already proven once that we can make big databases like this, so what’s stopping us from doing it for a larger audience?

Also, to get this really big, we’ll need manpower. We can do quite a lot ourselves, but if we can appeal to the linguistics community at large, that would be extremely helpful.

(Incidentally, this stuff drove a lot of the choices I made in my Polynesian mockup. For one thing, having lots of duplicate changes is fine for conlangers perusing a website for ideas — but it would quickly mess up any kind of statistical analysis. Similarly, it’s nice to be able to restrict yourself to, say, ‘papers with ordered sound changes’ if you’re researching something where that’s relevant. And the quotes from the original sources make it clear when something is unclear or could be interpreted differently.)

I think it's fine for amateur work and curated sound changes like my LP work to stay on the board for fellow conlangers.

I reckon you can think bigger too — you’re doing high-quality work there, in my estimation. If there’s one thing I’ve learnt from being in my physics research group, opportunities for papers occur with surprising frequency.

I’ve finished my mockup of some Polynesian changes: https://bradrn.com/files/polynesian-biggs-mockup.html. This is compiled mostly from Biggs 1978, plus the changes for Takuu, Nukumanu and Nukeria from Davletshin 2015. Please let me know your thoughts! (Comments on the sound changes themselves can go in the Polynesian thread, general comments can go here.)

I think this is a good format. The links for shared changes are a nice touch.

Thanks!

Note that hyperlinks are rather tedious to do by hand, so I couldn’t add all the ones that I want. (That’s why the table of contents is unlinked, for one thing.) I’m thinking of computer-generating the final version with a lot more links and cross-referencing: e.g. it would be nice to show the reconstructed realisations when hovering over the asterisked symbols.

One minor nitpick is that for *w, ‘was probably bilabial and possibly a voiced fricative rather than a semi-vowel’ implies to me [w~β̞~β] rather than just [w~β].

Fair enough; should be fixed now.

[Darren]

Page has very kindly sent me a copy of the Fayu grammar, so I'll shortly be posting an updated comparative wordlist with a lot more Fayu terms. (She also hints that a Fayu-based PhD thesis may be appearing in the future)

Great to hear! Could you send a copy to me too? (You should have my email already, I think.)

If I ever had your email, I've lost it; could you dm me it?

(Incidentally, this stuff drove a lot of the choices I made in my Polynesian mockup. For one thing, having lots of duplicate changes is fine for conlangers perusing a website for ideas — but it would quickly mess up any kind of statistical analysis. Similarly, it’s nice to be able to restrict yourself to, say, ‘papers with ordered sound changes’ if you’re researching something where that’s relevant. And the quotes from the original sources make it clear when something is unclear or could be interpreted differently.)

As a non-coding person, I wonder whether it would be (easily) achievable to have an option to toggle between showing e.g. known ordering vs. uncertain/unordered changes on the site? If it's relatively easy, then maybe there's some other features we could toggle, like idk "hide incomplete change lists" or such.

I reckon you can think bigger too — you’re doing high-quality work there, in my estimation. If there’s one thing I’ve learnt from being in my physics research group, opportunities for papers occur with surprising frequency.

We'll see how it goes. I have no real idea how to write a paper and I'm not planning to go to uni, so I don't know how I would go about any of that.

[bradrn]

Page has very kindly sent me a copy of the Fayu grammar, so I'll shortly be posting an updated comparative wordlist with a lot more Fayu terms. (She also hints that a Fayu-based PhD thesis may be appearing in the future)

Great to hear! Could you send a copy to me too? (You should have my email already, I think.)

If I ever had your email, I've lost it; could you dm me it?

I’ll send you an email now.

(Incidentally, this stuff drove a lot of the choices I made in my Polynesian mockup. For one thing, having lots of duplicate changes is fine for conlangers perusing a website for ideas — but it would quickly mess up any kind of statistical analysis. Similarly, it’s nice to be able to restrict yourself to, say, ‘papers with ordered sound changes’ if you’re researching something where that’s relevant. And the quotes from the original sources make it clear when something is unclear or could be interpreted differently.)

As a non-coding person, I wonder whether it would be (easily) achievable to have an option to toggle between showing e.g. known ordering vs. uncertain/unordered changes on the site? If it's relatively easy, then maybe there's some other features we could toggle, like idk "hide incomplete change lists" or such.

I was thinking of doing that as part of the general search functionality. Possibly there may be a smoother way to do that along the lines you mention, but I suspect it’s one of those things which will only become really clear once you start building it.

…actually, now that I think about it, I should probably explain the meaning of the star rankings I’ve used for the sources:

• Ordering is how much information the source has on the relative ordering of the sound changes.
  (e.g. for Polynesian, Biggs talks about it very little. The book by Martin which linguistcat’s been using would probably be 2/3 or even 3/3, thanks to its full chronological chart.)
• Exhaustivity is how much effort it puts in to fully describe every single change from the protolanguage to the descendant.
  (It’s hard to assess for Biggs, since he covers so many languages: 1/3 seems reasonable to me. Davletshin gets 0/3: due to his very narrow focus on /l/ and /r/, he doesn’t even attempt to be exhaustive. A lot of other specific sources will probably be similar.)
• Detail is how detailed each sound change is, especially in terms of its conditioning factors.
  (Biggs gets 2/3 for this — although most of his rules are unconditioned, he’s pretty good at describing conditioning factors where they exist. I gave Davletshin 3/3… which actually may be a bit too high; there seem to be many words which don’t follow his changes, indicating there may be some randomness he’s not clear about. Perhaps I should downgrade it to 2/3.)
• Consensus is the well-acceptedness of the phylogenetic group in question by the majority of linguists. Most things should get 3/3. (For other examples, glottalic-theory IE might get 2/3; the various mutually-contradictory reconstructions of Afroasiatic should probably get 1/3; Altaic would be 0/3.)
• Transcription is how confident I am about the transcription of the source. To some extent this overlaps with the above conditions, insofar as poor transcription means that details about ordering, exhaustivity or detail could possibly be missed.
  (I gave Biggs 3/3, since it was pretty easy to follow; Davletshin I gave 2/3 since I’m not sure I found and transcribed all the conditioning factors.)

These criteria are of course somewhat subjective, but that can’t really be avoided. They’re also per-source (rather than per-language or per-family or something else… except ‘consensus’ which is a bit iffy in that regard). I’ve divided them up in a rather fine-grained way, which leaves me unsure as to whether I’ve made the best choices of criteria: I’m fairly happy with the first two, but more uncertain about the others. What does everyone else think?

I reckon you can think bigger too — you’re doing high-quality work there, in my estimation. If there’s one thing I’ve learnt from being in my physics research group, opportunities for papers occur with surprising frequency.

We'll see how it goes. I have no real idea how to write a paper and I'm not planning to go to uni, so I don't know how I would go about any of that.

Ah, for some reason I thought you were already doing a postgraduate degree. (Though if research is something you might be at all interested in, it may be worth contacting linguists whose work you appreciate: ‘Hi so-and-so, I really enjoy in your work on whatever-it-is, in my spare time I’ve been working on a reconstruction of Proto–Lakes Plain, would you be at all interested in talking about it?’ Unless academic people in linguistics are very different to the ones I know in physics, they’ll generally like the opportunity to talk to enthusiatic students.)

[Darren]

These criteria are of course somewhat subjective, but that can’t really be avoided. They’re also per-source (rather than per-language or per-family or something else… except ‘consensus’ which is a bit iffy in that regard). I’ve divided them up in a rather fine-grained way, which leaves me unsure as to whether I’ve made the best choices of criteria: I’m fairly happy with the first two, but more uncertain about the others. What does everyone else think?

I reckon they're all pretty good. Ordering and exhaustivity are for sure the most important, although the rest are definitely useful.

Ah, for some reason I thought you were already doing a postgraduate degree.

I'll take that as a complement

(Though if research is something you might be at all interested in, it may be worth contacting linguists whose work you appreciate: ‘Hi so-and-so, I really enjoy in your work on whatever-it-is, in my spare time I’ve been working on a reconstruction of Proto–Lakes Plain, would you be at all interested in talking about it?’ Unless academic people in linguistics are very different to the ones I know in physics, they’ll generally like the opportunity to talk to enthusiatic students.)

That's an idea for sure. I'll wait until I've worked through everything to a level I'm happy with, then I'll consider contacting someone. I'm not personally concerned with getting it published unless I come up with a good amount of new research and changes.

On a related note: I'll look over the LP changes on the old ID, and see if there's anything I can add with the resources I have, then reformat it in brad format (which I think is a good standard). There might be a few other languages I could add too. I don't think Mekeo's in the old ID, and Jones's grammar has some changes in it. I think I've also got a better set of Russian changes than what's on there currently. And anything else I can get my hands on.

[bradrn]

These criteria are of course somewhat subjective, but that can’t really be avoided. They’re also per-source (rather than per-language or per-family or something else… except ‘consensus’ which is a bit iffy in that regard). I’ve divided them up in a rather fine-grained way, which leaves me unsure as to whether I’ve made the best choices of criteria: I’m fairly happy with the first two, but more uncertain about the others. What does everyone else think?

I reckon they're all pretty good. Ordering and exhaustivity are for sure the most important, although the rest are definitely useful.

Thanks!

On a related note: I'll look over the LP changes on the old ID, and see if there's anything I can add with the resources I have, then reformat it in brad format (which I think is a good standard). There might be a few other languages I could add too. I don't think Mekeo's in the old ID, and Jones's grammar has some changes in it. I think I've also got a better set of Russian changes than what's on there currently. And anything else I can get my hands on.

And thanks for this too! Maybe also take the opportunity to double-check the already-existing changes… given the, ahem, questionable quality of the current ID, I wouldn’t be surprised to see errors.

(Actually, just checking them now, they’re surprisingly comprehensive. I guess that’s in large part thanks to Man in Space!)

---

Now that I think of it… while we’re having this conversation, I’ll take the opportunity to repost something I asked in the Japonic thread:

I’d be interested to get your thoughts on how to treat multiple sources. So far I’ve been keeping sound changes from different sources entirely separate, on the basis that they might not be entirely compatible with each other (e.g. they might assume different reconstructions, or it might be impossible to order them sensibly). But on the other hand, we should expect different sources to agree on many sound changes, so keeping them entirely separate would result in lots of annoying duplicate changes. Do you have any suggestions as to how this should be handled?

Any thoughts?

[Darren]

Now that I think of it… while we’re having this conversation, I’ll take the opportunity to repost something I asked in the Japonic thread:

I’d be interested to get your thoughts on how to treat multiple sources. So far I’ve been keeping sound changes from different sources entirely separate, on the basis that they might not be entirely compatible with each other (e.g. they might assume different reconstructions, or it might be impossible to order them sensibly). But on the other hand, we should expect different sources to agree on many sound changes, so keeping them entirely separate would result in lots of annoying duplicate changes. Do you have any suggestions as to how this should be handled?

Any thoughts?

You could cite each change individually, like

p → v /V_V (Xi 2003, Zi 2004)
s → Ø /_# (Xi 2003)
m → n /_# (Zi 2004)

Which is still clunky but considerably less so. Or you could just number the sources with a key at the beginning; p → v /V_V (1,2) or whatever. That might only be necessary when you're dealing with like three or four sources though.

[bradrn]

Now that I think of it… while we’re having this conversation, I’ll take the opportunity to repost something I asked in the Japonic thread:

I’d be interested to get your thoughts on how to treat multiple sources. So far I’ve been keeping sound changes from different sources entirely separate, on the basis that they might not be entirely compatible with each other (e.g. they might assume different reconstructions, or it might be impossible to order them sensibly). But on the other hand, we should expect different sources to agree on many sound changes, so keeping them entirely separate would result in lots of annoying duplicate changes. Do you have any suggestions as to how this should be handled?

Any thoughts?

You could cite each change individually, like

p → v /V_V (Xi 2003, Zi 2004)
s → Ø /_# (Xi 2003)
m → n /_# (Zi 2004)

Which is still clunky but considerably less so. Or you could just number the sources with a key at the beginning; p → v /V_V (1,2) or whatever. That might only be necessary when you're dealing with like three or four sources though.

I’ve considered this, but it has a few big problems. Such as: how should the sound changes be ordered? As a worst-case scenario, for instance, you could imagine a situation where two sources agree on two sound changes but put them in opposite orders. More common are situations where it’s simply unclear how to order the changes.

There’s also problems when one source subsumes another. For instance, in the changes for Takuu (Polynesian), Biggs presents a change *l→{l,r}, leaving the condition unclear. Then Davletshin, after considering the data more carefully, formulated a set of six more detailed rules as a replacement for that one change. Putting both Biggs’s and Davletshin’s sound changes in one place would be confusing, because the latter replaces part of the former. (If I were taking the approach of creating a single authoritative set of changes, I’d just merge those changes in the obvious way, but like we discussed, I’d rather not do that for now.)

[Darren]

I’ve considered this, but it has a few big problems. Such as: how should the sound changes be ordered? As a worst-case scenario, for instance, you could imagine a situation where two sources agree on two sound changes but put them in opposite orders. More common are situations where it’s simply unclear how to order the changes.

There’s also problems when one source subsumes another. For instance, in the changes for Takuu (Polynesian), Biggs presents a change *l→{l,r}, leaving the condition unclear. Then Davletshin, after considering the data more carefully, formulated a set of six more detailed rules as a replacement for that one change. Putting both Biggs’s and Davletshin’s sound changes in one place would be confusing, because the latter replaces part of the former. (If I were taking the approach of creating a single authoritative set of changes, I’d just merge those changes in the obvious way, but like we discussed, I’d rather not do that for now.)

Those are some good points. I think clunkiness is inevitable.

[bradrn]

I’ve considered this, but it has a few big problems. Such as: how should the sound changes be ordered? As a worst-case scenario, for instance, you could imagine a situation where two sources agree on two sound changes but put them in opposite orders. More common are situations where it’s simply unclear how to order the changes.

There’s also problems when one source subsumes another. For instance, in the changes for Takuu (Polynesian), Biggs presents a change *l→{l,r}, leaving the condition unclear. Then Davletshin, after considering the data more carefully, formulated a set of six more detailed rules as a replacement for that one change. Putting both Biggs’s and Davletshin’s sound changes in one place would be confusing, because the latter replaces part of the former. (If I were taking the approach of creating a single authoritative set of changes, I’d just merge those changes in the obvious way, but like we discussed, I’d rather not do that for now.)

Those are some good points. I think clunkiness is inevitable.

I suppose that’s true. Maybe we can just include these relations in the underlying database, and not attempt to represent them in the user interface unless the user asks for it.

[bradrn]

While going through Blust’s *t to k: An Austronesian Sound Change Revisited (https://doi.org/10.1353/ol.2005.0001), I’m beginning to rethink my current approach of cross-referencing all shared changes.

The main problem I’ve encountered is that there seem to be a lot of almost-shared changes, so to speak. For instance, Biggs (my original source) has a straightforwardly shared *t→k in Hawaiian, Luangiua and Colloquial Samoan. But Blust lists Niʻihau Hawaiian as having *t → k / anywhere before [k] in a word, and Tahitian as having *t → k / _Vt. Do those count as shared or not? I don’t know.

There’s also the fact that the delineation of shared changes is, as far as I can tell, completely arbitrary. Biggs is happy to take one change and say, ‘this is shared by languages X, Y and Z’. But Blust is less definite about it. And of course, if some sources don’t mark changes as being shared, that completely negates the deduplicating benefits of marking shared changes!

So, for these reasons, I suspect it may not be worth our time to mark shared changes. The benefits are marginal, and largely conjectural at this point; meanwhile, it increases the effort required on our part, and it doesn’t seem possible to mark shared changes in a consistent way anyway. (And, if someone wants to do statistical analysis of our data and runs into Galton’s problem, they can damn well decide how to deduplicate the data themself!)

[Darren]

While going through Blust’s *t to k: An Austronesian Sound Change Revisited (https://doi.org/10.1353/ol.2005.0001), I’m beginning to rethink my current approach of cross-referencing all shared changes.

The main problem I’ve encountered is that there seem to be a lot of almost-shared changes, so to speak. For instance, Biggs (my original source) has a straightforwardly shared *t→k in Hawaiian, Luangiua and Colloquial Samoan. But Blust lists Niʻihau Hawaiian as having *t → k / anywhere before [k] in a word, and Tahitian as having *t → k / _Vt. Do those count as shared or not? I don’t know.

There’s also the fact that the delineation of shared changes is, as far as I can tell, completely arbitrary. Biggs is happy to take one change and say, ‘this is shared by languages X, Y and Z’. But Blust is less definite about it. And of course, if some sources don’t mark changes as being shared, that completely negates the deduplicating benefits of marking shared changes!

So, for these reasons, I suspect it may not be worth our time to mark shared changes. The benefits are marginal, and largely conjectural at this point; meanwhile, it increases the effort required on our part, and it doesn’t seem possible to mark shared changes in a consistent way anyway. (And, if someone wants to do statistical analysis of our data and runs into Galton’s problem, they can damn well decide how to deduplicate the data themself!)

I also noticed this when looking at the Papuan Tip (including mekeo) changes. Lots of languages have *l → Ø /_i, while lots of other ones have *l → Ø /#_i. They're clearly related changes; but they're not just one shared change. I could say that *l → Ø /#_i is shared by all of this group, while further *l → Ø /V_i is shared by the rest of them, but that's implying a level of internal differentiation which isn't stated in the source. So I agree it's not worth doing shared changes (although it's a nice idea). It should be pretty easy to tell when changes are all connected; for statistical analysis you could probably lump all repeated instances of a change in one family together as one data point and be reasonably safe.


Some unrelated questions (for everyone):

* The Polish changes on the old ID are really good, but need reformatting. The references are both in Polish though; does anyone a) have these books and b) read polish? It would be nice to flesh out the Slavic section.

* I have a copy of Joseph Wright's English Dialect Grammar which pretty thoroughly (albeit confusingly) goes through sound changes for a number of British dialects. Is this worth formatting up for the new ID? Or are we mostly ignoring dialectal variation?

* I think there's potential for some reasonable-quality Skou changes which I could try writing up too. Maybe a couple other random families which didn't make it into the old ID. Do yous mind a piecemeal approach rather than filling out entire major families at a time?

[bradrn]

I also noticed this when looking at the Papuan Tip (including mekeo) changes. Lots of languages have *l → Ø /_i, while lots of other ones have *l → Ø /#_i. They're clearly related changes; but they're not just one shared change. I could say that *l → Ø /#_i is shared by all of this group, while further *l → Ø /V_i is shared by the rest of them, but that's implying a level of internal differentiation which isn't stated in the source. So I agree it's not worth doing shared changes (although it's a nice idea). It should be pretty easy to tell when changes are all connected; for statistical analysis you could probably lump all repeated instances of a change in one family together as one data point and be reasonably safe.

OK; I’ll get rid of the shared changes then!

* The Polish changes on the old ID are really good, but need reformatting. The references are both in Polish though; does anyone a) have these books and b) read polish? It would be nice to flesh out the Slavic section.

IIRC, Pedant is Polish, though I’m not sure if they’re active here any more. Maybe worth sending a PM?

* I have a copy of Joseph Wright's English Dialect Grammar which pretty thoroughly (albeit confusingly) goes through sound changes for a number of British dialects. Is this worth formatting up for the new ID? Or are we mostly ignoring dialectal variation?

Yes please! I’ve included some dialectal variation in my changes.

* I think there's potential for some reasonable-quality Skou changes which I could try writing up too. Maybe a couple other random families which didn't make it into the old ID. Do yous mind a piecemeal approach rather than filling out entire major families at a time?

Piecemeal is fine with me! (I’ve been doing things by source rather than by language anyway.)

[Darren]

* The Polish changes on the old ID are really good, but need reformatting. The references are both in Polish though; does anyone a) have these books and b) read polish? It would be nice to flesh out the Slavic section.

IIRC, Pedant is Polish, though I’m not sure if they’re active here any more. Maybe worth sending a PM?

I'll give it a go. He was online only a couple months ago so you never know. The original changes are credited to "Xiądz Faust" but they're probably just from the Polish conlang forum which I can't search. If we're lucky Xiądz Faust is Pedant but that's a long shot.

* I have a copy of Joseph Wright's English Dialect Grammar which pretty thoroughly (albeit confusingly) goes through sound changes for a number of British dialects. Is this worth formatting up for the new ID? Or are we mostly ignoring dialectal variation?

Yes please! I’ve included some dialectal variation in my changes.

I'll get on it then. It's not very clearly formatted (being almost 120 years old) so it might take some time to sort through it.

* I think there's potential for some reasonable-quality Skou changes which I could try writing up too. Maybe a couple other random families which didn't make it into the old ID. Do yous mind a piecemeal approach rather than filling out entire major families at a time?

Piecemeal is fine with me! (I’ve been doing things by source rather than by language anyway.)

Great!

[Man in Space]

I'll give it a go. He was online only a couple months ago so you never know. The original changes are credited to "Xiądz Faust" but they're probably just from the Polish conlang forum which I can't search. If we're lucky Xiądz Faust is Pedant but that's a long shot.

Nope, those changes came from the old ZBB. Faust himself drafted the PDF (which I still have somewhere) specifically because I asked for it.

[bradrn]

Somewhat random question: should we asterisk protolanguage categories, or not? For instance, if a sound change has ‘all vowels’ in the target, should that be notated as V or *V? Personally I think the former (on the basis that the asterisk represents reconstructed phonemes), but I’m not sure.

[Darren]

Somewhat random question: should we asterisk protolanguage categories, or not? For instance, if a sound change has ‘all vowels’ in the target, should that be notated as V or *V? Personally I think the former (on the basis that the asterisk represents reconstructed phonemes), but I’m not sure.

I think if we have a key at the beginning of a section like

V = *i *e *o *a

Then there's no need for an asterisk, since it's already accounted for.

[bradrn]

Somewhat random question: should we asterisk protolanguage categories, or not? For instance, if a sound change has ‘all vowels’ in the target, should that be notated as V or *V? Personally I think the former (on the basis that the asterisk represents reconstructed phonemes), but I’m not sure.

I think if we have a key at the beginning of a section like

V = *i *e *o *a

Then there's no need for an asterisk, since it's already accounted for.

Interesting idea, but do we really want to redefine V for each and every family? (It feels like a surefire way to make mistakes, too…)

[Darren]

Somewhat random question: should we asterisk protolanguage categories, or not? For instance, if a sound change has ‘all vowels’ in the target, should that be notated as V or *V? Personally I think the former (on the basis that the asterisk represents reconstructed phonemes), but I’m not sure.

I think if we have a key at the beginning of a section like

V = *i *e *o *a

Then there's no need for an asterisk, since it's already accounted for.

Interesting idea, but do we really want to redefine V for each and every family? (It feels like a surefire way to make mistakes, too…)

I guess not. But if V represents "all vowels", and the vowels are listed at the beginning with asterisks, then it's still accounted for.

[bradrn]

So, at this point I’ve posted a few mockups, and we seem to have decided on what the final result should be like, more or less. Now we just need to build this thing.

At the moment, I’ve been handwriting an HTML page for each family. This is annoyingly laborious and boring. It also limits the capability of other people to contribute. And, of course, it’s not particularly searchable. So we need a better solution for this.

What I’ve proposed is a custom-built CMS, which generates the website from some kind of simple text format. I still think this is a good idea. Initially, we could just make it generate static HTML files, basically automating what I’ve been doing by hand; once we’re confident it works well enough, we can extend it to support searching and so on.

But what does everyone else think?