A little reconstruction game

[Salmoneus]

I've recently been toying with some ideas for sister languages within the same family, and it occured to me they might be appropriate for a little challenge.

Unlike most of these challenges, rather than posting a plain word list, I'm posting a series of sentences. Rather than start off by overwhelming people with a wall of text, I'm going to start off with just two languages for now, and add a few more in the coming days (I have three more ready to post, and hopefully might get to add a few more after that, RL allowing). The sentences in each case mean the same thing, although there are some grammatical differences in expression (they're meant to be 'the most natural' way of saying it in each language, so it's possible that parallel constructions may be found, but be marked, in the other languages). For your sake and mind, the vocabulary is mostly the same, with only a little borrowing and lexical replacement - although there is some.

The game: from the given languages, reconstruct how you think the common ancestor may have looked.

Other questions: while you're thinking about that, and perhaps helping in doing that, I'd also be interested to hear...
- when you think the last common ancestor may have been
- what subfamilies can be seen (obviously only an issue once I post a third language...)
- whether you can guess anything about the grammar from comparing the languages, even without knowing what the sentences mean. Can you spot possible parts of speech or word order rules?
- whether you think the various languages really are languages, or only mutually intelligible dialects
- what your general aesthetic impression is of these languages (what they remind you of, etc)

Couple of caveats:
- I'm giving the sentence in something approaching IPA rather than using native orthographies, as this may bias you unduly
- I'm giving apparent phonetic form, not presupposing anything about deep phonemic analysis
- similarly, you can take spaces as indicating phonological units for purposes of stress and the like, but I'm not presupposing anything about what is and isn't a morpheme or which morphemes are clitics vs inflection markers, etc
- similarly I've not tried to add any sort of punctuation indicating either prosodic or grammatical divisions of each sentence
- 'X' is an uvular fricative, since this board makes chi look like x
- I actually worked out the vowel diachronics in particular on a vowel space diagram, and it shouldn't necessarily be assumed that these symbols represent exact cardinal values, but just best approximations given the vowel systems as a whole
- I've tried to avoid any 'mistakes'; but if you do see something that looks 'wrong', feel free to ask me to double-check that I meant it

So let's begin!

We're starting with two languages: bɑ (A) and fɔɪnɛsæne (B).

EDIT: now including a third language/dialect [/b]k’lakæ (C). And yes, those are IPA ejective apostrophes, not SAMPA retroflex primes.

Our sentences are:

1.
A: ʔɑ haɾ ɣe dɾɐɪ aɾ ɸɛːʒæ nɐʊze þi aɾ ɸɛː ðɛːɾ æ vægjuːvɛ Xeːðɛ
B: æd ɣɪtɪz nɛʊsɪ dɾɔɪ pøʒäzo fɑːʊ ä βɪːdʒɨβɤ keðe nɛʊsɪ ti pøɣʏz
C: ət ɾɛ ɣes nəːse dɾəː ɾɛ pɛdʒæz nəse ti ɾɛ pɛgz ɸäʊn æ βæk’jʉβu kjəjɛ
D: dɹs ʊ getʃ tɹæɪ poːdʒəs in ti poːx i ʋikiːʋo kind


2.
A: dɛ ɸɔː þi ʔɐɪ m̊ɐɪ bæ dʒæ vaːðɛːɾæ
B: ti βoð-ʏz-o ɔɪd-ɪː mɔɪd βä wäː-z βɑʒɪteɾjɪː
C: ti ɛjɛ wud əː-s məːt wæ wæn βäːtsəɾn
D: ti äɾ bɔːtʃ æɪtʃ moɪ bɛ wɛːntʃ ʋɑːʃɛɾ

3.
A: me ɸɛː ɸɛː ʔaː zɐɪ vuːɾ ʔwoː dʒɑheː boɾɑː ðoː æ luː ɣɾɐ ɸaː ðɛ ʔugaːgʉ dɛː
B: æɪð sɔɪð ðɛh mʏh βøɣ pøɣ møɾʉ ʃæhɪː buɾɪ ðuɣ ä luː ɣɾɔh βæːɪ ðɛh ɪlgɑːgɨ de
C: æɪd səːd jɛ meʃ wɛg pɛg mɛɾʉβ tʃəʃeː sʉ boɾə jog æ luː gɾɑ wɑkɛβ jɛ k’ləgäːk’ʉ dzə
D: äːf sæɪ iɹ boːx poːk tɔːs tʃʊz.n äxto dv.pfɔɹɑ än eːʋ lɔːn gɹæz-əs ʊbæg iɹ kɾɑːmpf.kɾ tʃoːɾ

4.
A: dɛ poɾɑː ðoː ðɾeːgeː vuːɾ sʉ ʔɐʊgɑː ɑːɾ ɾɐkɐʒɛː βɐpɐ wan ðɛ glɔ ɸɛː zɐɪ vuːɾ
B: ðɛh buɾɪ ðuɣɪð dəð ðɾɪːgɪː ɛʊgɑʊ ɑʊɾ ɾɔkɔʃo-z βɔpɔg wɑn ðɛh lod pøɣ sɔɪ ɛd
C: jɛ boɾəd jogd wɛn dɾeːk’eː əːgäʊ ɸɾʉ sʉ ɾɑk’ɑtʃuz wɑpɑk ulju wän jɛ k’lut pɛgd səːd huɾ
D: iɹ pfɔɹɑ än ʃæɪt boːn dʒiːŋ.gŋ jʊgəɾ äɹ dʒ.ɹæktʃəs bæpək wɑːn iɹ lɔ poːk ʃæɪs sæɪ ɔːɹ

5.
A: dɛ poɾɑː ðoː tɐ ʔɔʎɔ vuːɾ fɘ tɛ naːzaː ʔwoː zlɐɪ.eː jʉ ʔɔʎɛ eːmɛ ʔɐ tʃɐɪ plɐdɛ duːɾ tɛ ðɛ hæʔ
B: ðɛh buɾɪ ðuɣɪð tɔgä fəd uː ʃømɔgɨ ojo jɨz slɔɪbɪː ðɔd emo ojo uː ʃɔɪb blɔtɛg ðɨɾ uː ðɛh äd
C: jɛ boɾəd jogd tɑkt jɛ fʉt un sʉ sləːpeː tʃəmɑgʉ ulju jʉd jəmud un jɛ tʃəːp blɑtɛk jʉɾ dɛ jɛ hæt ulju
D: ɪɹ pfɔɹɑ än ʃæɪt boːn täk ɪɹ fetf dʒ.ʃlæɪb.m boɪ iːt boːn imod.n iɹ tʃæɪp pläxtək dʊːɹf ɪɹ hɛt

6.
A: doː maːnɑː þaː vɛːɾɑː vaːlæ kɾɐɪʒɛ vɛːvovræ leː
B: mɑnɪ tæːʏz ðuɣɪzo mäʃ βæjɛläd gɾɔɪʃoz mäʃulɪg lez
C: mänə tæɪz jogz mætʃ βæɪlæt gɾəːtʃuz no mætʃləg ne ljəːz
D: ʋɔv.n äntso sibo ʋäɾ kɹæɪdʒəs zopəxt limfs

7.
A: leː ðɛːɾ næː æ vægjuːvɛ Xeːðɛ ze maː βɘ sɑ
B: mɪ βəg sɑz nɪ lezoː fɑːʊ ä βɪːdʒɨβɤ keðe
C: ne ljəːz no ɸäʊn æ βæk’jʉβu kjəjɛ zwo mæ wʉk säz
D: mä bek sʊːs dæ dɹs ne limfs i ʋikiːʋo kind


8.
A: maː βɘ ʔaheː zɑ dɾi
B: ɑdɪː mɪ βəg sɑzɪz dɾi
C: ädeː mæ wʉk säz dɾi
D: ɑːɾ dæ mä bex sɑːts dɹi

9.
A: ʔeːse ɾɛɸɐɪ wæ ʔaː ʒæ sɑ vuːɾ
B: ɾɛpɔɪd nɑʊ wäd æɪð sɑð
C: ɾɛpəːt dʒi nə: wæt æɪd säd ɜt
D: ɹpæɪ ji dɑːn wɛ æɪf dʒɛs sɑːt

10.
A: dʒɔːɾ sɐɪnjɛː dʒi ɾɛba nɐʊze ðɛːɾ dɐʒɛ βɘ Xoː ɾɛba Xo n̊æheː
B: ɾɛbɑg ʒoɾ sɔːɪjoz nɛʊsɪ ðeɾ βəgɪz kuz ɾɛbɑg kuh nädɪː
C: ɾɛbäk dʒi dʒuɾ səːnjuz nəːse jəɾ dɑʃɛ wʉks koz ɾɛbäkeː kox nædeː
D: ɹpfɑk ji jʊːɹ sæɪndəs in dʊːɹ bex kɔːs ɹpfɑk kɔs nɛd.n

11.
A: dʒi ɣo bɘ ʔɛ Xæːvɛ zɛː nɐʊze ʔɛː keː ʔɛzɛʔ
B: ði ʒi ɣud βəg ɔs käβɤð ä seg nɛʊsɪ eg gɪː
C: se dʒi ɣot sʉ wʉk wɛd kæβuweː æ ʃək nəːse jək gen ɛzɛ
D: di ji gɔ dv.bek dætʃ kɛʋot wot sjʊk in jʊk xoːn

12.
A: maː βɘ dʒi ɛː mɐɪ aː fɾʉ ʔæglɘɪ eː nɘʊze ʔaː ʔeː ʔɐɪvʉ zʉː
B: eb mɔɪ ä æɪðɪː mɪ βəg fɾʉ ɪːglɔːɪ eːz nɛʊsɪ æɪz dɛʊsɪ ɔɪβɨ sʉz
C: jəp dʒi məː jɑt æɪ hæɪdz mæ wʉk ɸɾʉ k’nək’ləːn hjəz nəːse æɪz nəːse əːβʉ sʉz
D: jʊp ji mæɪ wot iɹ äːd.ntf mä bek fɹm ŋkɾæɪn hintʃ äːs in æɪʋɾ sʊːs


...any thoughts?

[KathTheDragon]

Not even a translation?

[Salmoneus]

Not even a translation?

Well, most of what you can get out of it is phonological anyway, so the meaning doesn't matter. And I'm curious to see what if anything people might be able to guess about the grammar without having the crutch of a translation.

Think of it as, if you must, "we've found these prehistoric inscriptions - we know the script so we're confident what they sound like, but we don't know what they mean. Are they related, and if so how?"

EDIT: I suppose I might give translations later, if people do have a go but it turns out no-one can work out anything other than phonology. But I think it's fun to try without the translations first.

[dhok]

A few observations...

a) Language B appears to preserve original word-final voiced obstruents, while language A drops them at least sometimes. Examples include ʔɑ~æd, m̊ɐɪ~mɔɪd, ðo:~ðuɣ, le:~lez...in fact there may have been a general process of voiced-obstruent loss in language A: witness apparent cognate pairs as ʔɐɪ~ɔɪdɪ: and sɑ:~sɑzɪz, possibly even ɸɔ:~βɔðʏzo (very tentative)...but there are some apparent exceptions like ðɾe:ge:~ðrɪ:gɪ:--a borrowing, perhaps?

b) Do þa:~tæ:ʏz form a cognate pair? If so, we might add þi~ti in sentence 1, though there has been some clear syntactic moving-about if so. No such process appears to have occurred for *ð, except that language B's ðɛh corresponds to A dɛ at the beginning of sentences 4 and 5 and ðɛ apparently elsewhere; clarification that this is or isn't a typo would be quite helpful (perhaps, for example, language A has a consonant mutation process that creates alternation between stops and fricatives...we see both porɑ: and boɾɑ: for B buɾɪ, so I think it probably does. But why ðɛh buɾɪ in sentence 4 and ðɛh purɪ in 5?)

c) Language B seems to have some morphology that language A does not, though it is too early to say whether this morphology is innovative or a retention. ðuɣ~ðuɣɪð~ðuɣɪzo looks like some sort of inflection.

d) At least one *VjV sequence, seemingly preserved in language B, has yielded a long vowel in A: væ:læ~væjɛläd. However, I suspect B ojo is cognate not to A ʔwo: but to ʔɔʎɔ~ʔɔʎɛ, since ojo appears twice in B while ʔwo: appears once.

e) Both languages seem to have a two-way voicing distinction in the obstruents. But I wonder if there was a three-way distinction or some other sort of trickery originally. A voiced obstruent in A corresponds to a voiceless one in B in nɐʊze~nɛʊsɪ, dʒɑhe:~ʃæhɪ:, zɐɪ~sɔɪð and zlɐɪ.e:~slɔɪbɪ:; it is probably not coincidental that none of these are pure stops, but the implications of that factoid are not yet clear to me. The reverse is true in ʔuka:kʊ~ɪlgɑ:gɨ--there's a labiodental approximant at the end of that first word; was it an upsilon? a /u/? plɜdɛ~blɔtɛg and kɾɐɪʒɛ~grɔɪʃɔz seem to have correspondences in both directions. Sometimes language B's /p/ corresponds to A /ɸ/ (pøʒäzo~ɸɛ:ʒæ), sometimes to /b/ (plɜde~blɔtɛg, though I think there was a general process of voicing before liquids in B), sometimes there is no difference (βɔpɔ).

[KathTheDragon]

there's a labiodental approximant at the end of that first word

There're a couple cases I've seen:
- ʔukaːkʋ in sentence 3A
- møɾʋ in sentence 3B
- sʋ in sentence 4A
- jʋ in sentence 5A

[bradrn]

there's a labiodental approximant at the end of that first word

There're a couple cases I've seen:
- ʔukaːkʋ in sentence 3A
- møɾʋ in sentence 3B
- sʋ in sentence 4A
- jʋ in sentence 5A

How does those last two words even make sense? As far as I'm aware, it's not too common to have words without vowels - how would something like /jʋ/ even be pronounced?

[evmdbm]

My sense is that A and B didn't separate very long ago - perhaps the same sort of relation as German and Dutch? Particularly in 4A and 4B there are some very close cognate words and I think I detect the loss of some final consonants in A, which are not lost in B. Compare ɾɔkɔdʒɛ with ɾɔkɔdʒoz. Not quite sure whether these are inflectional suffixes though. Compare 4A poɾɑː with 3A boɾɑ. Inflectional prefixes maybe?

[KathTheDragon]

B seems to have had a process of labialising certain vowels near labial consonants. These are all the front rounded vowels:

• pøʒäzo
  pøɣʏz
  βoðʏzo
  mʏh
  βøɣ
  pøɣ
  mørʋ
  ʃømɔgɨ

[Hallow XIII]

A tabulation of initial correspondences:
(Language B first, for reasons I will explain later.)
[strike][/strike]

No.	B	A	Ex.
1	∅	∅	ä ~ æ
2	∅	ʔ	ɛʊgɑʊ ~ ʔɐʊgɑː
3	p	p	puɾɪ ~ poɾɑː
4	p	ɸ	pøɣ ~ ɸɛː
5	b	p	buɾɪ ~ poɾɑː
6	b	b	buɾɪ ~ boɾɑː
7	β	ɸ	βøɣ ~ ɸɛː
8	β	β	βɔpɔ ~ βɔpɔ
9	β	v	βɪːdʒɨβɤ ~ vægjuːvɛ
10	t	t	tɔga ~ tɜ
11	t	θ	ti ~ θi
12	d	d	dɨɾ ~ duːɾ
13*	ð	d	ðɛh ~ dɛ
14	ð	ð	ðuɣ ~ ðoː
15	k	X	keðe ~ Xeːðɛ
17	g	k	gɾɔɪʃoz ~ kɾɐɪʒɛ
18	ɣ	ɣ	ɣɾɔh ~ ɣɾɐ
19	f	f	fəd ~ fə
20	v	v	væjɛläd ~ vaːlæ
21	s	s	sɑz ~ sɑː
22	s	z	sɔɪ ~ zɐɪ
23	ʃ	tʃ	ʃɔɪb ~ tʃɐɪ
24	ʃ	dʒ	ʃæhɪː ~ dʒɑheː
25	m	m̊	mɔɪd ~ m̊ɐɪ
26	m	m	mɑnɪ ~ maːnɑː
27	n	n	nɪ ~ næː
28	l	gl	lɔ ~ glɔ
29	l	l	luː ~ luː
30	ɾ	ɾ	ɾɔkɔdʒoz ~ ɾɔkɔdʒɛː
31	w	w	wɑn ~ wan

Words ending in -h in B appear to condition devoicing in A:
ðɛh buɾɪ ~ dɛ poɾɑː (5)
ɣɾɔh βæːɪ ~ ɣɾɐ ɸaː (7)
mʏh βøɣ pøɣ ~ me ɸɛː ɸɛː (7)
ðɛh lɔ ~ ðɛ glɔ (possibly from *łɔ, or non-coalescence of *hlɔ) (28)

Similar rules appear to hold also for historically stop-final words:
ʃɔɪb blɔtɛg ~ tʃɐɪ plɜdɛ (5)
væjɛläd gɾɔɪʃoz ~ vaːlæ kɾɐɪʒɛ
ɔɪdɪː mɔɪd ~ ʔɐɪ m̊ɐɪ (25)

There is, however, an exception in blɔtɛg dɨɾ ~ plɜdɛ duːɾ -- one explanation is that, rather than voiced stop plus voiced stop resulting in a tenuis stop in A, the innovative language is B and tenuis stops are voiced immediately following voiced stops. However, there is the mysterious d~ð reflex; perhaps -G d- becomes d because -V d- becomes -V ð-?

More obviously, initial voiceless stops usually lenite following a vowel- or resonant-final word:
βɪːdʒɨβɤ keðe ~ vægjuːvɛ Xeːðɛ
mɑnɪ tæː(ʏz) ~ maːnɑː þaː
lɔ pøɣ ~ (g)lɔ ɸɛː
mʏh βøɣ pøɣ ~ me ɸɛː ɸɛː

Again, there is an exception in ðuɣ(ɪð) tɔga ~ ðoː tɜ, which makes me suspect that the process is not entirely phonological in nature.

[Salmoneus]

Damnit!

You think you've caught everything until you show it to someone else, and then suddenly...

So, I'm having to make some corrections. I'm updating the OP with the correct versions, but to avoid any confusion I'm listing here what I've changed...

2A: first sound is /d/, not /ð/
2B: it's hard to make out, but there was a rogue /r/ in the last word that is now the appropriate /ɾ/. And the /d/ is now /t/, so /βɑʒɪteɾjɪ/, not /βɑʒɪderjɪ/

3A: /ʔugaːgʋ/ not /ʔukaːkʋ/ (damnit, I prefer the second one!)

4A and 4B: a rogue /d/ crept in before /ʒ/... and in 4B, that /ʒ/ should actually be /ʃ/ anyway, damnit. So 4B now has /ɾɔkɔʃoz/, not original /ɾɔkɔdʒoz/.
Also, vowels wrong in 4A: now /ɾɐkɐʒɛː βɐpɐ/ (I did 4 at a different time from the others and clearly forgot a few differences that should have been there)

5A: /ʔɑʔ/ has been corrected to /ʔɐ/ and /plɜdɛ/ to /plɐdɛ/. The first is a phonemic error on my part, the second is just that I forgot I was calling it /ɐ/...
5B: it should indeed be /buɾɪ/, not as I originally wrote /puɾɪ/, well spotted. And /ðɨɾ/, not /dɨɾ/.

6B: shouldn't have /v/, but rather the labiodental. B doesn't have /v/ as a phoneme.

7A and 8A: both sentences should have /sɑ/, not /sɑ:/

General transcription issue:
I'm a colossal idiot. You're quite right, /ʋ/ is a labiodental approximant and is not what I intended in the slightest. Well, not exactly. Where you see /ʋ/, just imagine a /ʉ/ instead (or indeed a /ʉ̞/ would be a bit more accurate, but the exact position doesn't really matter).

Now, hopefully that's everything, although there's bound to be at least one more thing I've missed...


Oh, and I meant to include a note on stress: there isn't any. That is, there's no phonemic stress, and the vowel qualities here do not reflect (synchronic) reduction in unstressed environments. I'm leaning toward initial stress in A and penultimate stress in B, but I'm not sure, and it doesn't really matter.

And finally, I've added a 9, because it seemed like an easy way to clarify a couple of things at once.


So, sorry for the errors, hope I didn't waste your time too much.

I'll be back in a couple of minutes to see if anything needs clarifying or correcting in your responses [you responded! yay! thanks!]

[dhok]

The lack of devoicing in glɔ is easily explained by rule ordering, if *gl became /l/ before the devoicing applied. C.f. Osage, where Proto-Mississippi-Valley-Siouan *kr gives /l/.

[Salmoneus]

but there are some apparent exceptions like ðɾe:ge:~ðrɪ:gɪ:--a borrowing, perhaps?

To confirm: not a mistake on my part.

language B's ðɛh corresponds to A dɛ at the beginning of sentences 4 and 5 and ðɛ apparently elsewhere; clarification that this is or isn't a typo would be quite helpful

In general, this isn't a typo; however, as noted, I did make a typo for the first word of 2, which should have the stop, not the fricative.

But why ðɛh buɾɪ in sentence 4 and ðɛh purɪ in 5?)

Because I fucked up. Now /b/ in both sentences.

A voiced obstruent in A corresponds to a voiceless one in B in nɐʊze~nɛʊsɪ, dʒɑhe:~ʃæhɪ:, zɐɪ~sɔɪð and zlɐɪ.e:~slɔɪbɪ:

To confirm, these aren't typos.

; it is probably not coincidental that none of these are pure stops, but the implications of that factoid are not yet clear to me. The reverse is true in ʔuka:kʊ~ɪlgɑ:gɨ

Whereas this was. Now /g/ in both. Although of course this doesn't quite match your pattern either.

--there's a labiodental approximant at the end of that first word; was it an upsilon? a /u/?

My failure to use IPA from memory... it's just a barred-u rounded (almost) high central (or slightly back) vowel.

plɜdɛ~blɔtɛg and kɾɐɪʒɛ~grɔɪʃɔz seem to have correspondences in both directions. Sometimes language B's /p/ corresponds to A /ɸ/ (pøʒäzo~ɸɛ:ʒæ), sometimes to /b/ (plɜde~blɔtɛg, though I think there was a general process of voicing before liquids in B), sometimes there is no difference (βɔpɔ).

Not typos. Although I messed up a couple of vowels - now /plɐdɛ/ and /βɐpɐ/


Some really good thoughts here btw.

[Salmoneus]

A tabulation of initial correspondences:

A smart idea, and very pretty! although not so much in the edit screen...

So, to make clear the corrections I made above, correspondances 3, 12 and 20 are due to my own typos. Sorry, transcribing errors by the field researchers... The pair in 3 is actually in 5, the pair in 12 is actually in 13, and the pair in 20 is actually in 9.

Similar rules appear to hold also for historically stop-final words:
ʃɔɪb blɔtɛg ~ tʃɐɪ plɜdɛ (5)
væjɛläd gɾɔɪʃoz ~ vaːlæ kɾɐɪʒɛ
ɔɪdɪː mɔɪd ~ ʔɐɪ m̊ɐɪ (25)

There is, however, an exception in blɔtɛg dɨɾ ~ plɜdɛ duːɾ -- one explanation is that, rather than voiced stop plus voiced stop resulting in a tenuis stop in A, the innovative language is B and tenuis stops are voiced immediately following voiced stops. However, there is the mysterious d~ð reflex; perhaps -G d- becomes d because -V d- becomes -V ð-?

A nice explanation but the simpler one is just that I fucked up...

More obviously, initial voiceless stops usually lenite following a vowel- or resonant-final word:
βɪːdʒɨβɤ keðe ~ vægjuːvɛ Xeːðɛ
mɑnɪ tæː(ʏz) ~ maːnɑː þaː
lɔ pøɣ ~ (g)lɔ ɸɛː
mʏh βøɣ pøɣ ~ me ɸɛː ɸɛː

Again, there is an exception in ðuɣ(ɪð) tɔga ~ ðoː tɜ, which makes me suspect that the process is not entirely phonological in nature.

But here I did NOT fuck up!


Since I messed things up with some typos, it's only fair I give you something in return: yes, you're all right, there is indeed a mutation process in A, and you're mostly right about the rules for it (though not perfectly).

[Salmoneus]

My sense is that A and B didn't separate very long ago - perhaps the same sort of relation as German and Dutch?

To be fair, German and Dutch separated not much less than 2,000 years ago, so 'very long ago' is clearly relative! However, it's likely that the close contact of German and Dutch kept them closer than they otherwise 'ought' to be...

[Ryan of Tinellb]

Sorry, transcribing errors by the field researchers...

Or typos (writeos? paintos?) by the original author. Which is probably a realistic obstacle to things like this anyway .

[KathTheDragon]

Where you see /ʋ/, just imagine a /ʉ/ instead

For which? Just A, or B as well?

[Hallow XIII]

Nice. I'll use this opportunity to draw up a table of vowel correspondences as well.

No.	B	A	Ex.
1	æ	ɑ	æd~ʔɑ
2a	æɪ	aː	æɪð~ʔaː
2b	æjɛ	aː	βæjɛläd~vaːlæ
2c	æːɪ	aː	βæːɪ~ɸaː
2d	æːʏ	aː	tæːʏz~þaː
3	ä	æ	wäd~wæ
4a	ɑ	aː	mɑnɪ~maːnɑː
4b*	ɑ	a	ɑdɪː~ʔaheː
4c*	ɑ	aː	βɑʒɪteɾjɪː~vaːðɛːɾæ
4d	ɑ	ɑ	sɑð~sɑ
5	ɑʊ	ɑː	ɑʊɾ~ɑːɾ
6	ɑː	aː	ɪlgɑːgɨ~ʔugaːgʉ
7	ɑːʊ	???	fɑːʊ~???
8a	ɛ	ɛ	ðɛh~dɛ
8b	ɛ	∅	βæjɛläd~vaːlæ
9	ɛʊ	ɐʊ	ɛʊgɑʊ~ʔɐʊgɑː
10a	ɔ	ɐ	βɔpɔ~βɐpɐ
10b	ɔ	ɔ	lɔ~glɔ
11	ɔɪ	ɐɪ	sɔɪ~zɐɪ
12a	e	eː	lez~leː
12b	e	ɛː	de~dɛː
13	ø	ɛː	pøɣ~ɸɛː
14	ə	ɘ	βəg~βɘ
15	ɤ	ɛ	βɪːdʒɨβɤ~vægjuːvɛ
16a	o	ɔ	ojo~ʔɔʎɛ
16b	o	ɛ	emo~eːmɛ
16c	o	ɛː	ɾɔkɔʃoz~ɾɐkɐʒɛː
16d	o	ɔː	βoð(ʏzo)~ɸɔː
17	oː	???	lezoː
18a	ɪ	ɑː	buɾɪ~poɾɑː
18b	ɪ	aː	mɪ~maː
18c	ɪ	æː	nɪ~næː
18d	ɪl	u	ɪlgɑːgɨ~ʔugaːgʉ
19a	ɪː	æ	βɪːdʒɨβɤ~vægjuːvɛ
19b	ɪː	eː	slɔɪbɪː~zlɐɪ.eː
20	ʏ	e	mʏh~me
21	i	i	ti~þi
22a	ɨ	ʉ	jɨz~jʉ
22b	ɨ	uː	βɪːdʒɨβɤ~vægjuːvɛ
22a	ʉ	???	møɾʉ~???
24a	u	o	buɾɪ~poɾɑː
24b	u	oː	ðuɣ~ðoː
25	uː	???	uː~???

[Salmoneus]

Where you see /ʋ/, just imagine a /ʉ/ instead

For which? Just A, or B as well?

Both.

Hallow: wow.

But yes, there was at least one other thing I got wrong, thank you - I can take your 10b of your hands by saying it should have been /lod/, not /lɔ/ as I originally wrote.

[KathTheDragon]

I've gone and counted up the vowel correspondences (A~B) in my data, and sorted them by frequency of occurence:
(Note that I've rewritten ʋ as ʉ)

• ɐɪ ~ ɔɪ (zɐɪ ~ sɔɪ, sɔɪð; zlɐɪeː ~ slɔɪbɪː; etc.)
• ɐ ~ ɔ (βɐpɐ ~ βɔpɔ; ɾɐkɐʃɛː ~ rɔkɔʒoz; etc.)
• æ ~ ä (ɸɛːʒæ ~ pøʒäzo; æ ~ ä; etc.)
• ɛ ~ ɛ (ðɛ/dɛ ~ ðɛh; plɐdɛ ~ blɔtɛg; ɾɛbɐɪ ~ ɾɛpɔɪd)
• ɛː ~ ø (ɸɛː ~ pøɣ, βøɣ, pøɣʏz; ɸɛːʒæ ~ pøʒäzo)
• ɑ ~ ɑ (sɑ ~ sɑð, sɑz, sɑzɪz)
• ɑː ~ ɪ (poɾɑː/boɾɑː ~ buɾɪ; maːnɑː ~ mɑnɪ)
• eː ~ e (leː ~ lez, lezoː; Xeːðɛ ~ kede)
• eː ~ ɪː (dʒɑheː ~ ʃæhɪː; ðɾeːgeː ~ ðɾɪːgɪː)
• oː ~ u (doː ~ ðuɣɪzo; ðoː ~ ðuɣ, ðuɣɪð)
• ɐʊ ~ ɛʊ (nɐʊze ~ nɛʊsɪ; ʔɐʊgɑː ~ ɛʊgɑʊ)
• ɑː ~ ɑʊ (ɑːɾ ~ ɑʊɾ; ʔɐʊgɑː ~ ɛʊgɑʊ)
• e ~ ɪ (ɣe ~ ɣɪtɪz; nɐʊze ~ nɛʊsɪ)
• ɘ ~ ə (fɘ ~ fəd; βɘ ~ βəg)
• i ~ i (dɾi ~ dɾi; ti ~ ti)
• o ~ u (poɾɑː/boɾɑː ~ buɾɪ)
• uː ~ ɨ (duːɾ ~ ðɨɾ; vægjuːvɛ ~ βɪːdʒɨβɤ)
• ʉ ~ ɨ (jʉ ~ jɨz; ʔugaːgʉ ~ ɪlgɑːgɨ)
• a ~ ɑ (wan ~ wɑn)
• aː ~ æːɪ (ɸaː ~ βæːɪ)
• aː ~ æːʏ (þaː ~ tæːʏz)
• aː ~ æɪ (ʔaː ~ æɪd)
• aː ~ æjɛ (vaːlæ ~ væjɛläd)
• aː ~ ɑ (maːnɑː ~ mɑnɪ)
• aː ~ ɑː (ʔugaːgʉ ~ ɪlgɑːgɨ)
• aː ~ ɪ (maː ~ mɪ)
• æ ~ ɪː (vægjuːvɛ ~ βɪːdʒɨβɤ)
• æː ~ ɪ (næː ~ nɪ)
• ɑ ~ æ (dʒɑheː ~ ʃæhɪː)
• e ~ ʏ (me ~ mʏh)
• ɛ ~ e (Xeːðɛ ~ kede)
• ɛ ~ ɤ (vægjuːvɛ ~ βɪːdʒɨβɤ)
• ɛ ~ o (kɾɐɪʒɛ ~ gɾɔɪʃoz)
• ɛː ~ o (ɾɐkɐʃɛː ~ rɔkɔʒoz)
• ɔ ~ ɔ (glɔ ~ lɔ)
• ɔː ~ o (ɸɔː ~ βoðʏzo)
• u ~ ɪl (ʔugaːgʉ ~ ɪlgɑːgɨ)
• uː ~ uː (luː ~ luː)

I've also noticed that a in a number of words, there's a string of sounds in B corresponding to nothing at all in A:

• doː ~ ðuɣɪzo
• ðoː ~ ðuɣɪð
• ɣe ~ ɣɪtɪz
• leː ~ lezoː
• ɸɛː ~ pøɣʏz
• ɸɛːʒæ ~ pøʒäzo
• ɸɔː ~ βoðʏzo
• sɑ ~ sɑzɪz
• ʔɐɪ ~ ɔɪdɪː

. I think it's quite likely that these are inflectional suffixes that never arose in A, giving us the morphemes -o, -oː, -ɪː, -ɪz, -ɪð, and -ʏz if this isn't an allomorph of -ɪz.

[Hallow XIII]

Okay, let's see what I can make out of this. Kath is clearly right; the rounded vowels /ʏ ʉ ø/ have exactly the same correspondences as /ɪ ɨ e/ and only occur either adjacent to labial consonants or in words with other non-back rounded vowels, apparently a rightward-spreading form of rounding harmony.

The low vowels are an absolute disaster. It looks like the ancestral language had all three of /æ a ɑ/, with /a/ merging into /ɑ/ in B and /æ/ becoming /ä/, /æ/ being afterwards recovered from /ɑ/ or maybe a diphthong involving it. Vowel length in general is a clusterfuck as well; it looks like there was some compensatory lengthening in A (βɑʒɪt- ~ vaːð-), but fuck knows why ɪlgɑːgɨ has a long /ɑ/ but mɑnɪ has a short one.

It looks like there was a chain shift in B where u merged into ʉ, dragging u <- o <- ɔ.

Kath: I would wager that the ɪ is epenthetic, given also the forms sɑð and sɑz.