Darren's Mitsiefa Thread

[bradrn]

While it's a bizarre analysis, it's the biggest discrepancy between consonant phones and phonemes I've ever seen – on average more than 6 distinct realisations of each phoneme. And yes, obviously it's complete bunk.

Moloko (Chadic) has a single vowel phoneme with 5 realisations, plus another 5 realisations of an epenthetic vowel, thanks to word-level ‘prosodic’ features. Quite a lot of the Chadic languages are similar. So I wouldn’t be so confident as to call this one ‘obviously complete bunk’.

(That being said, the details of this analysis do look a great deal more strained than the Chadic ones. I don’t see the point of positing prosodies if they only ever apply to one CV syllable at a time!)

[Darren]

While it's a bizarre analysis, it's the biggest discrepancy between consonant phones and phonemes I've ever seen – on average more than 6 distinct realisations of each phoneme. And yes, obviously it's complete bunk.

Moloko (Chadic) has a single vowel phoneme with 5 realisations, plus another 5 realisations of an epenthetic vowel, thanks to word-level ‘prosodic’ features. Quite a lot of the Chadic languages are similar. So I wouldn’t be so confident as to call this one ‘obviously complete bunk’.

(That being said, the details of this analysis do look a great deal more strained than the Chadic ones. I don’t see the point of positing prosodies if they only ever apply to one CV syllable at a time!)

Yeah I draw the line at having not one but three different syllable-level nasalisations. His argument mostly rests on the fact that /e ɛ/ and /o ɔ/ are non-contrastive before prenasalised stops and after nasals, which apparently is a much bigger problem than the fact that half of his "prosodies" only occur with two (2) consonant onsets.

[Darren]

How's this for obscure? A five-consonant inventory used in a (partly incorrect) transliteration of Persepolitan late Elamite texts by Edward Hincks in 1846:

Code: Select all

 p   t   k
     s
     n

Modern analysis suggests more like eleven consonants – still fairly small. I like Hincks's thinking though (he added the vowels /i u ə˞ a/ which is delicious).

[Raholeun]

How's this for obscure? A five-consonant inventory used in a (partly incorrect) transliteration of Persepolitan late Elamite texts by Edward Hincks in 1846:

Code: Select all

 p   t   k
     s
     n

Modern analysis suggests more like eleven consonants – still fairly small. I like Hincks's thinking though (he added the vowels /i u ə˞ a/ which is delicious).

So, do you think they corresponded with the Kings of ə˞?

Jokes aside, I had a lot of fun reading this thread. Good job in describing the typological issues surrounding minimalism.

[Travis B.]

How's this for obscure? A five-consonant inventory used in a (partly incorrect) transliteration of Persepolitan late Elamite texts by Edward Hincks in 1846:

Code: Select all

 p   t   k
     s
     n

Modern analysis suggests more like eleven consonants – still fairly small. I like Hincks's thinking though (he added the vowels /i u ə˞ a/ which is delicious).

An inventory of /p t k s n/ combined with /a i u/ (/ə˞ / is silly frankly) seems to be one of the smallest "sane" inventories to me.

[foxcatdog]

nah where is your *m

More: show

btw smaller inventories are much saner then big ones

[Travis B.]

nah where is your *m

/n/ can have labial and velar allophones based on adjacent consonants and/or vowels.

[Darren]

Something that rivals Orokolo for the least TNG-looking TNG language – Ai'aowe Kamoro. Even the name is obscenely Polynesian.

Code: Select all

 p   t       ʔ
 m   n
 w   r   j
 
 
 i iː        u uː
 e eː        o oː
       a aː

A nonesense sample (the words are real, not the sequence):

Epare pea ewe opo'o owena watepa amo yamapu piniwiri patayaro uuru namae apaoe

It also has one of the highest sonorant-to-obstruent ratios I know of (62.5%). The Ok language Muyu does beat it (66.7%) with a slightly larger inventory:

Code: Select all

 b   t       k
 m   n       ŋ
 w   l   j

As does the Oceanic language Mamusi (63.6%):

Code: Select all

 p   t   k
     s
 m   n   ŋ
     r
 w   l   j

(And of course not to forget Bandjalang; /b d ɟ g m n ɲ ŋ w l r j/)

[Darren]

Here's an interesting case: the language Purari, or Namau, or Koriki, or Baimuru; or Evorra, I'ai, or Maipua. It's spoken very close to the Eleman languages like Orokolo, and has a very similar consonant inventory... I think.

There's really only one secondary source on the language, which is Holmes 1913, calling it "Namau". He lists the following inventory:

Code: Select all

 p     k
 m  n
 v  l
    r

i.e. seven consonants. Petterson N.D. agrees, calling it "Baimuru" for some reason, that it has seven consonants, but he says it has /ʔ/ without mentioning what other consonants it has. Wikipedia opts to include /ʔ/ but otherwise uses Holmes's inventory. The SIL OPD (Anonymous 2004), calling it "Koriki", adds /h/ onto this again.

Going back to Holmes, he claims that:

"t" is found in words having totemic significance but is never used in every-day conversation.

[...]

"l" and "r" seem to be interchangeable; "v" and "m" seem to be used erratically.

[...]

I duly recognize the possibility to confuse letter sounds; I have been, at times, painfully aware of the inability of the native to discern letter-sound values; at all times I have avoided getting information from the native when he is under the influence of betel-chewing, hence I regard the foregoing note on the consonants, their sounds and the peculiarities of usage in this language as being, in the main, correct.

That would absolutely make sense for a heavily Eleman-influenced language; generally speaking /r~l/ is a single phoneme (usually an allophone of /n~l~r/ in fact), and so are /m~β/ in the same manner. And Holmes, who learnt both Toaripi and Purari, reckoned the rate of cognates and similar word formation was too high to be due simply to contact (or, as he puts it, "intercourse"). Anyway, if he's right then it might have an inventory more like this:

Code: Select all

 p     k
 m  n
    l

But then he was compiling this over 110 years ago, and I'm not sure if they believed in glottal stops back then, so perhaps it's more like this:

Code: Select all

 p     k  ʔ
 m  n
    l

Taking everything into account, I can safely say that it has the following inventory, with somewhere between five consonants and double that number:

Code: Select all

 p  (t)  k  (ʔ)
 m   n
(β)  l      (h)
    (r)

And here's some more amusing excerpts from this rather quaint little grammar:

It has already been remarked that the Namau people duplicate a vowel wherever they can, but often when using the genitive the sense of euphony seems to embarrass them and there is an apparent effort to follow their usual custom of duplicating the final vowel of the word, whereas it will not always yield to it.

The ablative gives me some concern as it is difficult to be sure if I have really found it, or whether I am compromising a postposition to meet the need.

Gender is unknown in this language; an object is either a man or a woman. Trees, birds, fishes, everything is thus spoken of and the native seems to be very accurate in determining the sex of a tree. All trees are male or female, and he is shrewd enough to speak of the best trees as males.

I reluctantly forego the pleasure of giving more illustrations to avoid making these notes too long, but I hope to return to the subject of this paper and shall then note in detail the suggestiveness of the expletives used in this language; I am inclined to think that in them the student will find a strong clue to the genius of the language.

[Talking of number-marking on nouns:] It must not be thought that the native is very particular in this matter, he seems very indifferent to numbers.

There are two terms, au, ua, undoubtedly expressing "it," but I am unable to place them because they seem to be used erratically; to be used when not expected, omitted when they seem most necessary.

The natives of Namau seem to have well-defined ideas on the propriety of using the imperative mood. A boy speaking to his chief, or any native speaking to God in the form of prayer, always studiously avoids using the imperative mood, whereas in speaking to another boy he invariably uses it. When a native, for reasons he will not state, wishes to avoid the use of this mood he falls back on the infinitive mood.

[Darren]

It also has one of the highest sonorant-to-obstruent ratios I know of (62.5%).

And it turns out that Awtuw, a Ram language, far exceeds this:

Code: Select all

 p  t     k
 m  n     ŋ
    r  ɻ
 w  l  j

It also has a vowel system /i u e ə o æ a/ but /u o/ are only "marginally phonemic".

[Darren]

Ntǎ^ntɔfẽ

Lērisama has produced a language attempting to break my universals wherever possible. Obviously this must be out-done. Hence Ntǎ^ntɔfẽ /¹²¹-³-³²⁵.N/ [ntã˩˨˩ntɔ̃˧ɸɪ̃˧˨˥]

​ ​ ​ ​ ​ 1a. All spoken languages have phonemes
​ ​ ​ ​ ​ 1b. There are never fewer phones than phonemes in a phonological system
​ ​ ​ ​ ​ 2a. All spoken languages have multiple consonant phonemes
​ ​ ​ ​ ​ 2b. All consonant inventories have multiple degrees of sonority including multiple obstruents
​ ​ ​ ​ ​ 2c. There are always multiple consonant phonemes which are more sonorous than the least sonorous series of phonemes
​ ​ ​ ​ ​ 2d. There are always multiple contrastive places of articulation in a consonant inventory
​ ​ ​ ​ ​ 2e. Consonant phonemes will always occur at at least two out of labial, coronal and velar POAs.
​ ​ ​ ​ ​ 3a. All languages have at least one vowel phoneme.
​ ​ ​ ​ ​ 3b. All languages have multiple vowel phonemes unless consonants or words have markedness for F2.
​ ​ ​ ​ ​ 3c. All languages have at least one [+front] vowel, or a [+front] consonant such as /j/.
​ ​ ​ ​ ​ 3d. No language distinguishes frontness (F2) without also distinguishing height (F1).
​ ​ ​ ​ ​ 3e. No language has fewer than three vowel phonemes which does not also have more than ten consonant phonemes.

Can I break all of these? Well, no, 3d and 1a/3a are incompatible, and 1b with 1a, so I'll have to make do with 10 out of 12. So here's the phoneme inventory:



I hope you enjoyed that. Moving on to suprasegmentals, we have three of these – word-level nasalisation, word-level ATR harmony and tone. The first two of these are binary – words are [+/–nasal] and [+/–ATR]. Tone is more complex; there are sixteen tonemes (/5, 4, 3, 2, 1, 53, 51, 21, 34, 14, 454, 352, 121, 231, 325, 213/), and a morpheme can have between one and three of these. This provides for about 16,000 morphemes.

1. Words are assigned one vowel |V| per tone.
2. Tone influences F1 heavily; the highest pitch of a tone determines what |V| is – 1 2 3 4 5 → [a a ə ə ɨ]
3. Stress is placed on the first tone with the highest peak (the first tone with a 5 pitch gets stress, otherwise first with a 4, and so on). Syllables two away from the tonic get secondary stress.
4. Vowels dissimilate in frontness, with the tonic fronting and the adjacent backing, [a ə ɨ] → [æ,ɒ e,o i,u]. A vowel which is not next to the tonic remains central. If there is only one syllable it remains central.
5. Hiatus is repaired with |C|, which is realised as [tɕ] adjacent to [i], [t] following a central vowel, [k] following a front vowel and [ɸ] following a back vowel
6. Before stressed or secondarily stressed word-initial [i e,æ ɨ,ə,a] is inserted [tɕ k t]
7. Intervocalically, [ɸ t tɕ k] lenide to [h θ s ʃ] except before a primarily stressed vowel
8. In +ATR words, consonants voice [ɸ t tɕ k θ s ʃ h] → [b d ɟ ʔ l ɾ j β], vowels split [i ɨ u e ə o æ a ɒ] → [i,ɪ ɨ,ɨ̙ u,ʊ e,ɛ ə,ɐ o,ɔ ɛ,æ ɐ,a ɔ,ɒ].
9. In nasal words, vowels nasalise and [ɸ t tɕ k b d ɟ ʔ l ɾ j β] become [mp nt ɲc ŋk mb nd ɲɟ ŋg n ɲ ŋ m]

The surface inventory is thus

[mp b mb t nt d nd tɕ ɲc ɟ ɲɟ k ŋk ŋg ʔ] <mp b mb t nt d nd c nc j nj k nk g ʔ>
[ɸ β θ s ʃ h] <f v θ s ʃ h>
[m n ɲ ŋ] <m n ñ ŋ>
[l ɾ j] <l r y>

[i ɨ u] <i i u>
[ɪ ɨ̙ ʊ] <e e o>
[e ə o] <e e o>
[ɛ ɐ ɔ] <ɛ ɛ ɔ>
[æ a ɒ] <æ a ɒ>
[Ṽ] <V>

[a˥ a˦ a˧ a˨ a˩ ; a˥˧ a˥˩ a˨˩ ; a˧˦ a˩˦ ; a˦˥˦ a˧˥˨ a˩˨˩ a˨˧˩ ; a˧˨˥ a˨˩˧] <ā á a à ǎ ; ā' a' ǎ' ; a- ǎ- ; ā^ â ǎ^ à^ ; ã ǎ~>

Ntǎ^ntɔfẽ does follow two universals - it has more phones (58) than phonemes (0) and it does not distinguish F2 without also distinguishing F1.

Here's some example words:

/231.14.325/ → [tɐ˨˧˩θɛ˩˦kɔ˧˨˥] <tà^θɛ̌-kɔ̃>
/21.5.1/ → [ɒ˨˩tɕɪ˥sɒ˩] <ɒ̌'cēsɒ̌>
/14/ → [tɐ˩˦] <tɛ̌->
/14.2 N A/ → [ŋgẽ˩˦ɲɔ̃˨] <ngě'ñɔ̀>
/352.3 N A/ → [ɟĩ˧˥˨ɲõ˧] <jîño>
/5.5/ → [tɕɪ˥sʊ˥] <cēsō>
/34.454.4 N/ → [ɔ̃˧˦ɲcɪ̃˦˥˦sɔ̃˦] <ɔ-ncē^sɔ́>
/325.3 A/ → [ɟi˧˨˥ɾo˧] <jĩro>

Morphology can produce some interesting results, say adding a suffix like /-454 A N/:

/14.121.4/ → [kɛ˩˦ʃɒ˩˨˩hɐ˦] <kɛ̌'ʃɒ̌^hɛ́>
/14.121.4-454 A N/ → [ə̃˩˦ndɐ̃˩˨˩nõ˦ɲɟĩ˦˥˦] <ě'ndɛ̌^nónjī^>

Although I think this ought to be isolating, so that hapless field linguists don't realise that only the tones are real. I might if I feel keen try and do a more sane ground-up analysis which reduces the phonemic tones to probably six (?), bestows us a vertical vowel system and at least a few consonants.

[bradrn]

This is getting ridiculous now. But it’s brilliant. Well done.

(It reminds me of an idea I once posted in the Conlang Random Thread, for a language where consonants and vowels weren’t separate. But this is more cursed than that. Sadly, I can’t seem to find the post again.)

[Lērisama]

Congratulations, you have indeed outdone me, in both general cursedness and number of universals broken. I'm not sure what else there is to say, except that I'm slightly in awe. I'll still work on mine though; it's grown on me

---

(It reminds me of an idea I once posted in the Conlang Random Thread, for a language where consonants and vowels weren’t separate. But this is more cursed than that. Sadly, I can’t seem to find the post again.)

Do you mean this post: viewtopic.php?p=32999#p32999?

[bradrn]

(It reminds me of an idea I once posted in the Conlang Random Thread, for a language where consonants and vowels weren’t separate. But this is more cursed than that. Sadly, I can’t seem to find the post again.)

Do you mean this post: viewtopic.php?p=32999#p32999?

Well done! How did you find it?

[Lērisama]

I remembered it involved consonant/vowel alternations, amd was pretty sure it involved the word altern(wasn't sure about the ending), so I searched "altern*" in the conlang random thread, then got bored, so I searched for "altern*" in Conlangery with the author being bradrn*, only looked at the topic names until I found something in the Conlang random thread, and I think it was the 4th or 5th one of those


* How do you pronounce Bradrn, by the way? In my head it is /b̥ɹad̥ɜːn/, or possibly /b̥ɹad̥ɹən/

[bradrn]

I remembered it involved consonant/vowel alternations, amd was pretty sure it involved the word altern(wasn't sure about the ending), so I searched "altern*" in the conlang random thread, then got bored, so I searched for "altern*" in Conlangery with the author being bradrn*, only looked at the topic names until I found something in the Conlang random thread, and I think it was the 4th or 5th one of those

Wait, phpBB supports wildcards in searching…‽ Wow, that’s convenient.

(Also, I hadn’t realised you’d been around here back then.)

* How do you pronounce Bradrn, by the way? In my head it is /b̥ɹad̥ɜːn/, or possibly /b̥ɹad̥ɹən/

It’s just my initials, so I say it as ‘brad-r-n’. Though feel free to say it however you want.

[Lērisama]

Wait, phpBB supports wildcards in searching…‽ Wow, that’s convenient.

It does. There are a few other things you can do in searching, but I have to look at the advanced search every time

(Also, I hadn’t realised you’d been around here back then.)

I didn't make my account until just before my first post, but I did read the forum for quite a while before

* How do you pronounce Bradrn, by the way? In my head it is /b̥ɹad̥ɜːn/, or possibly /b̥ɹad̥ɹən/

It’s just my initials, so I say it as ‘brad-r-n’. Though feel free to say it however you want.

Good to know, thanks

[bradrn]

Wait, phpBB supports wildcards in searching…‽ Wow, that’s convenient.

It does. There are a few other things you can donin searching, but I have to look at the advanced search every time

Oh yeah… I use the advanced search reasonably often but I must have skipped over that part. I guess I’ve gotten so used to the search being rubbish that I never thought it could have actually useful features.

(Also, sorry Darren for derailing your thread completely…)

[Darren]

Attempting to analyse Ntǎ^ntɔfẽ as a hapless field linguist

I'll start by listing monosyllables. There are 64 of these; under my first analysis, just 16 tones with +/-ATR and +/-Nasal marking. The table below suggests an alternative analysis by which the sixteen tones are viewed as six tones with three realisations depending on vowel height (while missing the two rising tones on low vowels).

Code: Select all

      1      2     3       4       5      6

/tɐ/  ta˩    ta˨   ta˩˨˩   ta˨˩
      ntã˩   ntã˨  ntã˩˨˩  ntã˨˩

/dɐ/  dɐ˩    dɐ˨   dɐ˩˨˩   dɐ˨˩
      ndɐ̃˩   ndɐ̃˨  ndɐ̃˩˨˩  ndɐ̃˨˩

/tə/  tɐ˧    tɐ˦   tɐ˨˧˩   tɐ˨˩˧   tɐ˧˦   tɐ˩˦    
      ntɐ̃˧   ntɐ̃˦  ntɐ̃˨˧˩  ntɐ̃˨˩˧  ntɐ̃˧˦  ntɐ̃˩˦   

/də/  də˧    də˦   də˨˧˩   də˨˩˧   də˧˦   də˩˦    
      ndə̃˧   ndə̃˦  ndə̃˨˧˩  ndə̃˨˩˧  ndə̃˧˦  ndə̃˩˦   

/tɨ/  tɨ̙˥    tɨ̙˥˧  tɨ̙˧˥˨   tɨ̙˥˩    tɨ̙˦˥˦  tɨ̙˧˨˥
      ntɨ̙̃˥   ntɨ̙̃˥˧ ntɨ̙̃˧˥˨  ntɨ̙̃˥˩   ntɨ̙̃˦˥˦ ntɨ̙̃˧˨˥

/dɨ/  dɨ˥    dɨ˥˧  dɨ˧˥˨   dɨ˥˩    dɨ˦˥˦  dɨ˧˨˥
      ndɨ̃˥   ndɨ̃˥˧ ndɨ̃˧˥˨  ndɨ̃˥˩   ndɨ̃˦˥˦ ndɨ̃˧˨˥


1: Low  2: Mid  3: rise-fall  4: fall  5: High rise  6: Low rise

This is excellent for vowels, but sadly reduces the consonant inventory to only /t d/ – I'm analysing vowel ATR as secondary with assimilation of stop voicing within a word, but nasality is still considered at the word level, which gives us three vowels /ɨ ə ɐ/. Polysyllables allow us to introduce new consonants; I call them /p t c k b d ɟ g/ realised as [ɸ~h~ᵐp t~θ~ⁿt tɕ~s~ᶮc k~ʃ~ᵑk b~ᵐb~β~m d~ⁿd~ɾ~n ɟ~ᶮɟ~ɾ~ɲ ʔ~ᵑɡ~j~ŋ]. These then influence vowel frontness as you'd expect;

• /c ɟ/ front neighbouring /ɨ/ to [i], but only the first if they're intervocalic (/ɨcɨp/ → [icu] etc.)
• /k g/ front preceding /ɨ ə ɐ/ to [i e ɛ]
• /p b/ back preceding /ɨ ə ɐ/ to [u o ɔ]

For final vowels we can assume lenided final */p k/ where necessary. Lenition occurs as before, although stress is now placed on the first highest vowel with the highest tone. Lenition and nasalisation are the same. Hence we get a very nice little system of:

/p t c k/ <f~h~mp t~th~nt c~s~nc k~sh~nk>
/b d ɟ g/ <b~mb~v~m d~nd~r~n j~nj~r~ny '~ngg~y~ng>
/ɨ ə ɐ/ <i~ɨ~u e~ə~o ɛ~a~ɔ>
/L M LHL HL MH LH/ <i í î ì ĩ ǐ> etc.

Going back to our example words:

tə̂thěkǒ [tɐ˨˧˩θɛ˩˦kɔ˧˨˥] <- /təLHLtəLHkəpLH/
ɔ̀chisɔ [ɒ˨˩tɕɪ˥sɒ˩] <- /ɐHLcɨLcɐpL/
tǎ [tɐ˩˦] <- /tɐLH/
nggěnyó [ŋgẽ˩˦ɲɔ̃˨] <- /gəLHjəpM.N/
njînyo [ɲɟĩ˧˥˨ɲõ˧] <- /ɟɨLHLɟəL.N/
cisu [tɕɪ˥sʊ˥] <- /cɨLsɨpL/
õncĩsó [ɔ̃˧˦ɲcɪ̃˦˥˦sɔ̃˦] <- /əpMH.cɨMH.səM.N/
jǐro [ɟi˧˨˥ɾo˧] <- /ɟɨLHɟoL/

And the language would be Ntântofǐ [ntã˩˨˩ntɔ̃˧ɸɪ̃˧˨˥] <- /tɐLHLtəLpiLH/. Of course the major problem with this analysis is that there are major distributional restrictions on stops depending on their POA; in fact following stressed syllables only /p b/ occur and following unstressed syllables only /c ɟ k g/. You could definitely analyse away /c ɟ/ and probably /p b k g/ too (I know you could because that was the first analysis. You might even be able to reduce the tone system still further and come up with something like /p t k b d g ɨ ə ɐ ???some tones/ but you'd have to be a hell of a lot smarter than me to work it out.

[Lērisama]

I feel sorry for this linguist. I would definitely follow their system though (except maybe I'd merge /p k/ and /b ɡ/, since they seem to be in complimentary distribution), because the real one is clearly a hoax. Like that zero vowel analysis of Kabardian¹.

¹ Yes, I am aware that that one was supposed to be serious, but it makes so much more sense as a joke.