NatLang Diachronic Challenge

A forum for discussing linguistics or just languages in general.
Post Reply
User avatar
spanick
roman
roman
Posts: 1297
Joined: 11 May 2017 01:47
Location: California

NatLang Diachronic Challenge

Post by spanick »

Taking inspiration from the “Quick Diachronic Challenge” thread in the Conlangs forum, I thought I’d throw together a similar challenge based on natlangs rather than conlangs.

So, below I have ten words from five closely related languages, which are labeled A-E. While the languages all descend from one proto-language (technically, it’s a sub-branch), there is more than one clade represented by these languages. The labels and their order do not necessarily indicate any particular grouping.

For this challenge I’ll propose the following goals:
1. Reconstruct each word as closely as possible to the actual proto-word (as judged by me).
2. Correctly group these languages into their clades.
3. Guess the language family (bonus goal)

I’ll also propose the following limits:
1. Each participant is limited to three guesses for each goal.
2. The challenge will end in two weeks during the evening of Sunday 12/12 (PST).

The challenge will end once goals (1) and (2) have been met or when the time limit is met, whichever comes first. Once the challenge is over, I will reveal the language family, the individual languages, and of course the proto-words.

I think this will prove to be a relatively easy challenge, especially compared to the conlang game this is based on, but I hope the fact that these are natlangs might add to the entertainment.

Data
Spoiler:
1.
A: /nũːp/
B: /nũːpa/
C: /nõba/
D: /nũːwe/
E: /nõːba/

2.
A: /nãːt͡ʃawa/
B: /nakpa/
C: /nãtːa/
D: /nãːdwa/
E: /nitːa/

3.
A: /ɾaːʃ/
B: /t͡ʃaʒe/
C: /ʒaːʒe/
D: /ɾaːja/
E: /ʒaːʒe/

4.
A: /waːs/
B: /aze/
C: /baze/
D: /baːðe/
E: /maze/

5.
A: /sĩt͡ʃ/
B: /sĩte/
C: /sĩd͡ʒe/
D: /θĩːd͡ʒe/
E: /sĩːde/

6.
A: /hosĩnĩ/
B: /sni/
C: /snĩ/
D: /θɾĩː/
E: /osnĩ/

7.
A: /seɾet͡ʃ/
B: ???
C: /st͡ʃed͡ʒe/
D: /θɾeːd͡ʒe/
E: /snede/

8.
A: /nãː/
B: /t͡ʃaː/
C: /ʒã/
D: /jãː/
E: /ʒã/

9.
A: /ɡuː/
B: /ku/
C: /gy/
D: /ɡuː/
E: /gi/

10.
A: /peːt͡ʃ/
B: /pʰeta/
C: /pːeːde/
D: /peːd͡ʒe/
E: /pːeːde/
dva_arla
cuneiform
cuneiform
Posts: 182
Joined: 25 Oct 2019 21:03
Location: Realm of Ideas

Re: NatLang Diachronic Challenge

Post by dva_arla »

Is the challenge still open to guesses? Should they be posted in this very same thread or PMed to your inbox?
Conlangs in progress:
Modern Khotanese
Modern Gandhari
?? - Japonic language in the Mekong Delta
Locna - Indo-European language in N. Syria
Wexford Norse
A British romlang, &c.
User avatar
spanick
roman
roman
Posts: 1297
Joined: 11 May 2017 01:47
Location: California

Re: NatLang Diachronic Challenge

Post by spanick »

Sure thing. No one has guessed so far. Just put them in the threads in a spoiler.
dva_arla
cuneiform
cuneiform
Posts: 182
Joined: 25 Oct 2019 21:03
Location: Realm of Ideas

Re: NatLang Diachronic Challenge

Post by dva_arla »

spanick wrote: 05 Dec 2021 15:20 Sure thing. No one has guessed so far. Just put them in the threads in a spoiler.
Let's go!
Spoiler:

1. Most probable protoform: /nõːpa/ ~ /nṹ:pa/
p > b in C while further > w in D
õː > ũː in A, B, D
*ũ (rather than *õ) is chosen for reasons of symmetry. #6 points to the presence of a high-close nasalised vowel in the protolang. *ũ would provide the latter with a back counterpart.

2. Most probable protoform: /nãk:jwá/
long vowel in A may perhaps be due to compensatory lengthening.
palatal glide neccessary to account for affricate outcome in A (and perhaps also dental ones in C, D, E)

3. Most probable protoforms: /ɾáːʒe/ ~ /ɾaːʝe/ ~ /ɾaːje/ /ɾáːʃe/ or /ɾáːɾe/
initial ɾ retained in A, D, while > ʒ in C, E and further devoiced to t͡ʃ in B, both outcomes through an intermediary d͡ʒ
also note elision of word-final -e in A (also seen in nos 4,5,7,10)

4. Most probable protoform /bá:ze/ ~ /pá:ze/ /pá:se/
note: voiced and voiceless initials as likely as the other voicing probably not contrastive in the protolang (also for #9).

5. Most probable protoform: /sĩ́ːte/
notes:
palatalisation in A must happen after elision of word-final -e
palatalisation trigerred by /e/ but not /i/ (see #6; perhaps at one point realised as a central or back vowel [something like /ɨ/])
isoglosses:
intravocalic voicing in C, D, E
palatalisation in A, C, D

6. Most probable protoform: /hosnĩ́́(:)/ (phonetically [həsnĩ́́(:)]?)
Long vowel in D might be secondary.

7. Most probable protoform: /sɾé:te/ (or /sé(:)ɾte/?)
notes on nos 6 & 7:
reflex in C points heavily to clusters /sn/ and /sɾ/ being differentiated in the protolang.
sɾ > θɾ in D

8. Most probable protoforn: /ɾãː/

9. Most probable protoform: /gu:/ ~ /ku:/

10. Most probable protoform: /pːéːte/
/d/ in C (against /d͡ʒ/ in nos 5 and 7) may be accounted for by intradialectal borrowing.

Notes on final (declensional?) vowels:
Spoiler:
nos 2,4,5,7 show a more or less homogenous distribution of final vowel (save for the regular elision of final -e in A). The varying outcomes of the finals in no 1,3,10, on the other hand, poses a problem, to which an explanation may be sought either in a:

- difference in vowel qualities/quantities in the protolanguage,
- conditioning of vowel quality by preceding consonant, or
- shifts in declension

To resolve this problem a comparison of the endings might be in order:

1: -p -pa -ba -we -ba
3: -ʃ -ʒe -ʒe -ja -ʒe
10: -tʃ -ta -de -dʒe -de
(cf. 5,7: -tʃ -te -dʒe -dʒe -de)

In three cases, only one of the reflexes differ from the rest. 10 differs from 5,7 by the -a found in B; the change is most probably irregular. On the other hand, note that the irregularities in cognate set 1&3, found in D, only happens when the vowel is preceded by a semivowel; this might have had a bearing on the quality of the final vowel, though the direction of the change is rather uncanny (/w/ triggering a change a > e and /j/ the change e > a??)

The final -a is in A elided in #1 but not #2. This may be accounted for by difference in word-stress in the protolanguage (however the stress may be realised, or not, in the daughter languages): /'nõːpa/ vs. /nãk:jwa/.
Notes on vowel length and nasalisation:
Spoiler:
D seems to be the most conservative of all the languages when it comes to vowel length; having words in long vowels corresponding to short vowels in the other languages -- one might therefore safely base the vowel quantities of the protolanguage based on D. The other languages, nevertheless, reflect the vowel quantities of the protolanguage to a varying degree, to which a brief note might be of merit:

/ũː/ length retained in A, B, D, E
/uː/ length retained in A, D
/ã:/ length retained in A, D; in B only when stressed in the protolang
/aː/ length retained in A, C, D, E
/eː/
in #7 vowel length only retained in D, but in #10 in A, C, D, E. But note the ɾ following (or preceding?) the e, reconstructible for the protoform. We might be seeing something like Slavic metathesis here.
/ĩ́́:/ length retained in D, E. In #6 only in D, but see notes below #6

Nasal quality of /ĩ/ and /õ/ (or /ũ/?) retained in all five languages. /ã/ denasalised in B, when umlauted/unstressed in E.
Reconstructed phonemic inventory:
Spoiler:

Consonants:
p t k kʷ (kʲ) kʲʷ
m n
s [ʃ]*
h
ɾ (j) (w)

Code: Select all

Vowels:
a o~ə e
aː eː uː
ãː ĩː ũː~õː
The vowel system is phonemically analysable as consisted of four, underlying, vowel qualities:

Code: Select all

 /a i u ə/
Notes:
*ʃ may be omitted from our inventory if one assumes -ɾ- for the reconstruction in #3. A more decisive appraisal wound no doubt necessitate further evidence.
Forms in parentheses (edit: save for /ʃ/) cannot be inferred from the ten cognate sets alone, but their presence would present the phonemic inventory with more 'balance'.
Consonant voicing not distinctive, but length is.
Word-stress is lexically distinctive.
Clades:
Spoiler:
Most indicative of internal classification is the treatment of intervocalic consonants.
-t- is voiced in C, D, E
t / _e is affricativised/iotated in A, C, D
-s- is voiced in B, C, D, E
-ʃ- or -ɾ- > ʃ in A,
> ʒ in B, C, E,
and > j ( < ʒ ?) in D.
Reflexes of k:ʲʷ --
affricativised in A (k:ʲʷ > t͡ʃw)
becomes dental in C, D, E
velar quality retained in B

The relationship between the languages is best described with the wave model, wherein changes emanating from certain areas (one which we are yet to determine) propagate throughout the family (and perhaps beyond), influencing certain languages and leaving unscathed others. Nevertheless, certain shared phonetic innovations points out to a close relationship between C and D, and then between C, D, and E.... I'll just let the chart below do the talking more succinctly:
Spoiler:

Code: Select all

A  (B  ((C  D)  E))
D has developed some peculiar changes independent of the other languages, pointing to isolation after long periods of contact -- something like English and the rest of the Germanic family. Perhaps D is the language of some sea-fared colony? Or a people that has wandered too deeply into the deserts or mountains? The Andes? Nevada?


Language family?
Spoiler:
I am yet to fathom any definite guess... but the change ɾ > ʒ is rather reminiscent of an isogloss separating a language family in Central America, while the shift kw > t: occurred in some Caucasian languages.
Last edited by dva_arla on 06 Dec 2021 01:56, edited 2 times in total.
Conlangs in progress:
Modern Khotanese
Modern Gandhari
?? - Japonic language in the Mekong Delta
Locna - Indo-European language in N. Syria
Wexford Norse
A British romlang, &c.
dva_arla
cuneiform
cuneiform
Posts: 182
Joined: 25 Oct 2019 21:03
Location: Realm of Ideas

Re: NatLang Diachronic Challenge

Post by dva_arla »

Update:
Spoiler:
I am beginning to suspect kʲʷ -- the palatalisation and dentalisation in #2 might as well be brought upon by gemination. The protoform might as well be reconstructed as /nãk:ʷá/.
Conlangs in progress:
Modern Khotanese
Modern Gandhari
?? - Japonic language in the Mekong Delta
Locna - Indo-European language in N. Syria
Wexford Norse
A British romlang, &c.
User avatar
spanick
roman
roman
Posts: 1297
Joined: 11 May 2017 01:47
Location: California

Re: NatLang Diachronic Challenge

Post by spanick »

Guesses
Spoiler:
dva_arla wrote: 06 Dec 2021 01:30 1. Most probable protoform: /nõːpa/ ~ /nṹ:pa/
This is very close. The first consonant is wrong, but I'm inclined to give it to you, since I don't know if you'll be able to reconstruct it based on the evidence.
2. Most probable protoform: /nãk:jwá/
You're on the right track, but you've got the middle sequence wrong. Specifically, I think you've got the directionality of the sound changes wrong.
3. Most probable protoforms: /ɾáːʒe/ ~ /ɾaːʝe/ ~ /ɾaːje/ /ɾáːʃe/ or /ɾáːɾe/
/ɾáːʒe/ is the closest guess here. Like #1, you're off with the initial consonant but I'm inclined to give it to you, because it might be too difficult to determine without more data.
4. Most probable protoform /bá:ze/ ~ /pá:ze/ /pá:se/
/bá:ze/ is the closest. Again, the initial consonant is wrong and and the length of the vowel is off. I think it's due to the directionality of the relationship between /w p b/.
5. Most probable protoform: /sĩ́ːte/
[tick]
6. Most probable protoform: /hosnĩ́́(:)/ (phonetically [həsnĩ́́(:)]?)
/hosnĩ́́(:)/ is correct except for the vowel length
7. Most probable protoform: /sɾé:te/ (or /sé(:)ɾte/?)
/sɾé:te/ is correct except for the vowel length.
8. Most probable protoforn: /ɾãː/
See comment for #3.
9. Most probable protoform: /gu:/ ~ /ku:/
/ku:/ [tick]
I hope this helps establish the directionality of voicing related song changes.
10. Most probable protoform: /pːéːte/
This is essentially, it's a littel off with the initial consonant. There's a better answer, but this basically works and I've heard arguments that this is the actual reconstruction. So, [tick]
Clades:
Spoiler:

Code: Select all

A  (B  ((C  D)  E))
This isn't a bad model, but it's not quite right. I would suggest that you consider the vowel length of each root as a guide to how these languages are grouped. It's a little deceiving.
Language family?
Spoiler:
I am yet to fathom any definite guess... but the change ɾ > ʒ is rather reminiscent of an isogloss separating a language family in Central America, while the shift kw > t: occurred in some Caucasian languages.
Central America is closer than the Caucasus.
dva_arla
cuneiform
cuneiform
Posts: 182
Joined: 25 Oct 2019 21:03
Location: Realm of Ideas

Re: NatLang Diachronic Challenge

Post by dva_arla »

Second attempt:
Spoiler:

I have revised (2) to /nãk:wá/, with no palatal element in the glide.

4. /wáze/
6. /hosnĩ́́/
7. /sɾéte/

I have fathomed the long vowel in D to be secondary in #6's case, but didn't expect the same to hold for Nos 4 and 7.

8. /ɲãː/? There was a ɲ > (d)ʑ shift in the evolution of Mandarin and Hokkien.

I would suggest that you consider the vowel length of each root as a guide to how these languages are grouped.
In more than half of the cases A and D share the same vowel length. Perhaps a closer relationship between A and D?

To be frank I can't see how vowel length can be a better indication of linguistic distance or relation than phonological processes such as voicing or palatalisation.












Conlangs in progress:
Modern Khotanese
Modern Gandhari
?? - Japonic language in the Mekong Delta
Locna - Indo-European language in N. Syria
Wexford Norse
A British romlang, &c.
shimobaatar
korean
korean
Posts: 9281
Joined: 12 Jul 2013 23:09
Location: PA

Re: NatLang Diachronic Challenge

Post by shimobaatar »

spanick wrote: 29 Nov 2021 03:19 Taking inspiration from the “Quick Diachronic Challenge” thread in the Conlangs forum, I thought I’d throw together a similar challenge based on natlangs rather than conlangs.
spanick wrote: 29 Nov 2021 03:19 2. The challenge will end in two weeks during the evening of Sunday 12/12 (PST).
Interesting idea! I'd like to give this a shot at some point, but due to the time limit, Lexember, and offline responsibilities, I'm afraid I'm going to have to sit this first round out.
User avatar
spanick
roman
roman
Posts: 1297
Joined: 11 May 2017 01:47
Location: California

Re: NatLang Diachronic Challenge

Post by spanick »

So, I'm sorry that I didn't get back to you sooner. You have gotten most of the guesses pretty close. Since no one else has played, I'm going to end the game and post the answers in a separate post. But if you want to make another guess, please feel free to.
dva_arla wrote: 07 Dec 2021 02:01 Second attempt:
Spoiler:
I have revised (2) to /nãk:wá/, with no palatal element in the glide.
Closer, but no cigar.
4. /wáze/
6. /hosnĩ́́/
7. /sɾéte/
[tick]

8. /ɲãː/? There was a ɲ > (d)ʑ shift in the evolution of Mandarin and Hokkien.
Nope. This one is surprisingly simple and not one that requires a phoneme which has every feature of the daughters.
In more than half of the cases A and D share the same vowel length. Perhaps a closer relationship between A and D?

To be frank I can't see how vowel length can be a better indication of linguistic distance or relation than phonological processes such as voicing or palatalisation.
[tick]

AFAIK, this actually is a common feature between these two languages which despite being closely related look incredibly different.
User avatar
spanick
roman
roman
Posts: 1297
Joined: 11 May 2017 01:47
Location: California

Re: NatLang Diachronic Challenge

Post by spanick »

shimobaatar wrote: 11 Dec 2021 15:44 Interesting idea! I'd like to give this a shot at some point, but due to the time limit, Lexember, and offline responsibilities, I'm afraid I'm going to have to sit this first round out.
Yeah, I chose the time poorly. I also think I may have poorly curated my dataset.
User avatar
spanick
roman
roman
Posts: 1297
Joined: 11 May 2017 01:47
Location: California

Re: NatLang Diachronic Challenge

Post by spanick »

Answers

I'll start with the language family, then do the groupings, and then finally show the reconstructions.
Spoiler:
These are five languages from the Mississippi Valley branch of the Siouan langauge family.
Language groups
Spoiler:
Group 1
A = Hoocąk
D = Chiwere

Group 2
C = Kanza
E = Omaha-Ponca

Group 3
B = Lakota
Reconstructions
Spoiler:
1. *rũːpa 'two'
2. *nãtpa 'ear'
3. *jaːʒe 'name'
4. (Wa-)aze'breast'
5. *sĩːte 'tail'
6. *snĩ(-hi) 'cold'
7. *srete 'long'
8. *jãː 'sleep'
9. ku 'come'
10. *saːtʰã 'five'
Last edited by spanick on 13 Dec 2021 10:45, edited 1 time in total.
dva_arla
cuneiform
cuneiform
Posts: 182
Joined: 25 Oct 2019 21:03
Location: Realm of Ideas

Re: NatLang Diachronic Challenge

Post by dva_arla »

The initials in the reconstructed forms #1 and #9 were quite unexpected. What could have led to their assumption?
Conlangs in progress:
Modern Khotanese
Modern Gandhari
?? - Japonic language in the Mekong Delta
Locna - Indo-European language in N. Syria
Wexford Norse
A British romlang, &c.
User avatar
spanick
roman
roman
Posts: 1297
Joined: 11 May 2017 01:47
Location: California

Re: NatLang Diachronic Challenge

Post by spanick »

dva_arla wrote: 13 Dec 2021 05:12 The initials in the reconstructed forms #1 and #9 were quite unexpected. What could have led to their assumption?
#9 is just a typo. The proto form was *ku.

#1 is pretty simple, actually: r > n / __ [+nasal]. It was my mistake to put that word, since obviously no one could reconstruct that form without more data. The Missouri Valley Siouan languages (Hidatsa, Crow, Mandan) preserve the /r/ there.
Post Reply