Odd natlang features thread

[Nortaneous]

Didn't see one of these, so.

Here are a few from "The Languages of New Guinea" by William Foley:

• Proto-Lakes Plain is reconstructed as having a consonant inventory of /p t k b d/.
• Iau has eight tones, but verbs are not lexically marked for tone; instead, they change tones to mark aspect/mood. There are also six aspect-mood particles, but those also change tones. Examples from the paper:
doe 'see' doe3 'look at, watch' (durative), doe8 'have seen' (resultative)
tai 'contact' tai2 'be falling' (telic, durative), tai4 'still being pulled toward' (incomplete), tai5 'has fallen' (telic, punctual), tai6 'alight at/on' (punctual, resultative)
y (declarative speech marker) y3 (simple response providing information), y4 (implied directive), y9 (strongly contrastive)
• Many languages in New Guinea have highly reduced pronoun systems. Kuman has four, with a number distinction only in the first person: na '1SG', no '1PL', ene '2', ye '3'. However, the verbal affixes distinguish three numbers (singular, dual, and plural) in all three persons. Some have no number contrast at all: Manem ga '1', sa '2', aŋk '3', and Golin lacks a third-person pronoun: na '1', i '2'. Some conflate 2SG and 1PL: Suki ne '1SG', e '2SG/1PL', de '2PL', u '3SG', i '3PL'. Nimboran apparently distinguishes between inclusive and exclusive first-person pronouns but doesn't have a number contrast in pronouns, but no examples are given.
• I'm quoting this next one because I don't get it at all:

Dani (Bromley 1981) exhibits a fascinating cline in this area. There are six independent pronouns, distinguishing three persons and two numbers. But the subject suffixes in hortative imperative mood neutralize the number contrast in third person. The hypothetical mood further conflates second singular with the third person form, leaving the second plural as the only distinct second person form. Finally, the future potential loses all person distinctions, with only a binary contrast for number remaining.

• Kiwai has verbal prefixes that mark for whether or not the subject is first-person and suffixes that mark the subject's number (singular, dual, trial, and plural). The number of the object is marked by vowel alternations in the root for nonsingular objects, but there are also optional suffixes marking the same four numbers as are marked for the subject.
• Yimas has, in addition to singular, dual, and plural pronominal affixes, a paucal suffix -ŋkt, which can combine with plural pronominal affixes for subject and object.

And some others that I've picked up from various other places:

• Kensiu puts Germanic languages to shame: it distinguishes six levels of height (seven if the diphthong /ie/ is included), and also has a retroflexed vowel /ɚ/.
• Reduplicative infixation in Aslian languages.
• Kusunda has vowel harmony between /i ə u/ and /e a o/, but the two sets are in free variation in many words. Words with uvulars always have the lower set. Some high-frequency verbs mark the irrealis by shifting the vowels from the upper to the lower set, backing apicals to laminals (consonants in Kusunda are only distinguished by their active articulators; the passive articulator appears to vary almost freely) and velars to uvulars.
• Seri derived its definite articles from nominalized verbs of motion and location.

[chris_notts]

I'm sure we did have one of these. Did you look in the L&L Museum? Maybe the "Tidbits from beyond IE" thread?

http://zbb.spinnwebe.com/viewtopic.php?f=10&t=53

Also, did you mean "The Papuan Languages of New Guinea", or has Foley written a new book? A quick google search didn't reveal anything.

But anyway, I'm sure I can add some things to your list. I'll probably come back later...

[Ser]

The "Tidbits from beyond IE" thread was just Ran bragging about his native language (Mandarin), and nothing but his native language. You're thinking of Cor's "Quirks in conlangs/natlangs", which in spite of making it to the L&L Museum, it never received the love it deserved, as I pointed out last time somebody (dhokarena56) opened a thread like this.

[chris_notts]

I wrote something a while ago about role insensitive agreement that might fit here:

http://chrisdb.dyndns-at-home.com/blog/ ... -agreement

The best examples come from a Papuan language called Manambu, which has verbal agreement with topical NPs with very little restriction on their role. When the topical NP isn't the subject, then there is also subject agreement. For example:

Topical subject
wun a yarək bə laku-na-wun
I DEM.DIST+fem.sg news already know-ACT.FOC-1fem.sgBAS.VT
'I already know that news'

Topical object
də wun-a:m laku-da-wun
he I-LK+OBJ know-3masc.sgSUBJ.VT-1fem.sgBAS.VT
'He knows me'

Topical location
wun a-də yaba:r yi-tua-d
I DEM.DIST-masc.sg road+LK+ALL go-1stSUBJ.VT-3masc.sgBAS.VT
'I went towards that road'

Also, Elamite marked animate nouns for the person of their referent, which isn't a common feature in natlangs.

[Soap]

Some Khoisan languages have a mandatory verbal suffix that indicates whether the subject of the sentence is standing, sitting, or lying down. e.g. "I ate (while sitting down)" is one single verb. A very similar feature appears in a language of Mississippi named Tunica. Tunica also has gender, and apparently treats the so-called postural suffix for "while standing" differently depending on the gender of the speaker. For a male, the suffix for standing up is the same as the suffix for sitting; for a female, it's the same as for lying down.


source: google books search

http://books.google.com/books?id=0CsI6v ... 22&f=false

[Earthling]

• Kensiu puts Germanic languages to shame: it distinguishes six levels of height (seven if the diphthong /ie/ is included), and also has a retroflexed vowel /ɚ/.

It doesn't appear to actually distinguish six levels: front vowels are /i ɪ e̝ e ɛ/ (five), central /ɯ ɚ ə ʌ a/ (five), back /u o̝ o ɔ/ (four).

[chris_notts]

• Kensiu puts Germanic languages to shame: it distinguishes six levels of height (seven if the diphthong /ie/ is included), and also has a retroflexed vowel /ɚ/.

It doesn't appear to actually distinguish six levels: front vowels are /i ɪ e̝ e ɛ/ (five), central /ɯ ɚ ə ʌ a/ (five), back /u o̝ o ɔ/ (four).

If you argue in that way then Spanish and other typical 5 vowel systems only contrast two levels, since /a/ gets to have the centre to itself. But normally such 5 vowel systems are counted as distinguishing three levels, even if /a/ is in many cases less front as well as lower than the higher front vowels. Personally, I would count the Kensiu system as having six vowel heights.

[MisterBernie]

I'm quoting this next one because I don't get it at all:

Dani (Bromley 1981) exhibits a fascinating cline in this area. There are six independent pronouns, distinguishing three persons and two numbers. But the subject suffixes in hortative imperative mood neutralize the number contrast in third person. The hypothetical mood further conflates second singular with the third person form, leaving the second plural as the only distinct second person form. Finally, the future potential loses all person distinctions, with only a binary contrast for number remaining.

I assume it just means there's some merging going on in verbal person suffixes, like this:

[Astraios]

Noun modification is probably the most odd thing in L/Dakota, because it's so complicated/restrictive in comparison to English.

1a) If a head noun is definite, it can be modified with the structure Head + Mod. + Art.: wičháša čík'ala kiŋ the little man

1b) If the head noun is plural, the modifier is reduplicated (even if the head noun is animate, which is odd because reduplication usually only marks inanimate plural): wičháša čigčík'ala kiŋ the little men

1c) Phrases of this kind are treated phonetically as compounds; the primary pitch accent is heard only on the head noun: /wi.'ʧʰa.ʃa.ˌʧi.k’a.la.kɪ̃/

1d) If the modifier is a common or inherent property of the head noun, it's made into a full compound: šuŋgwášte kiŋ the good horse instead of šúŋkawakȟáŋ wašté kiŋ

2a) If a head noun is indefinite, there is also the structure Head + Mod. + Art., which is only used in the case of common or inherent properties: šúŋkawakȟáŋ wašté waŋ a good horse

2b) The more common structure is Head + Art. + Mod. + Art., which can be used for both inherent and unexpected properties: wičháša waŋ čík'ala čha a little man ~ a man who is little

3a) With complex relative clauses (i.e. those with more than just a stative verb in them), the structure can only be Head + Art. + Mod. + Art.: wičháša waŋ amápȟa čha / wičháša waŋ amápȟe kiŋ a man who hit me / the man who hit me

3b) Complex relative clauses must be the first nominal element in a sentence, no matter the role of the noun they modify, so that A and B are ambiguous and C is impossible:

A - Wičháša waŋ theyáȟila kiŋ wíŋyaŋ waŋ waŋyáŋke. The man you love saw a woman.
B - Wičháša waŋ theyáȟila kiŋ wíŋyaŋ waŋ waŋyáŋke. A woman saw the man you love.
C - *Wíŋyaŋ waŋ wičháša waŋ theyáȟila kiŋ waŋyáŋke. A woman saw the man you love.

3c) There can only be one layer of subordination per head, so "The wolf [that killed the woman [who saw the man [you love]]] is dead" would have to be translated as something like: A woman saw the man [you love] and now the wolf [that killed that woman] is dead.

[Risla]

Aymara morphophonology. Full stop.

[Drydic]

Aymara morphophonology. Full stop.

THANK YOU

I wanted to ask you if you had anything like that when you posted about being done with it in the {happy|venting} thread, but decided against it for some reason.

[Niedokonany]

http://www.uni-leipzig.de/~bickel/resea ... 06.ppt.pdf (see p. 24)
includes:

- antipassive inflection which is also used to mark a 1st person patient in Puma language (Tibeto-Burman), it's said to have evolved as a politeness device
- free prefix ordering (from what I gather, it doesn't influence the meaning) in Chintang language:

http://www.eva.mpg.de/psycho/pdf/Public ... _in_07.pdf

(1) a. a-ma-im-yokt-e.
2-NEG-sleep-NEG-PST
b. ma-a-im-yokt-e.
NEG-2-sleep-NEG-PST
Both: ‘You didn’t sleep.’
(2) a. u-kha-ma-cop-yokt-e.
3NS.A-1NS.P-NEG-see-NEG-PST
b. u-ma-kha-cop-yokt-e.
3NS.A-NEG-1NS.P-see-NEG-PST
c. kha-u-ma-cop-yokt-e.
1NS.P-3NS.A-NEG-see-NEG-PST
d. ma-u-kha-cop-yokt-e.
NEG-3NS.A-1NS.P-see-NEG-PST
e. kha-ma-u-cop-yokt-e.
1NS.P-NEG-3NS.A-see-NEG-PST
f. ma-kha-u-cop-yokt-e.
NEG-1NS.P-3NS.A-see-NEG-PST
All: ‘They didn’t see us.’

also from the paper: "In order to demonstrate free permutation, we need to show (i) that the prefixes under consideration form a single grammatical word with their host in all other respects save ordering constraints, and (ii) that permutation does not reflect semantic scope, syntactic
constituency, or language variety choice. We take these points up in §§4 and 5, but by way of background information, we begin by describing the inflectional system of the Chintang verb."

- 'recursive inflection' in the very same language: there are some verbs with a stem composed of two parts, which must be separated by a suffix (a dummy if nothing better is available) so that the [stem1 + affix] segment would be bisyllabic; only then the second part of the stem can be appended to it, taking further suffixes, which may include the aforementioned suffix

[[STEM1-SUFF1]-STEM2]-SUFF1-SUFF2-SUFF3...

• Proto-Lakes Plain is reconstructed as having a consonant inventory of /p t k b d/.

So this is simpler than Rotokas. Well, there could have been a merger (a la Tsimshian, hehe) along the way in all members of the family.

(telic, durative)

These seem a little contradictory to me, though I don't know what phenomenon has been labelled as telicity by the author.

Reduplicative infixation in Aslian languages.

At this stage my diachronic imagination says "Hey, God, stop the world, I'm getting off".

[stuff about Lakota]

Does the article waŋ have any specific function on its own or is it really just a linker or relative marker?

Aymara morphophonology. Full stop.

The dominant/recessive affix thing resembles stable and labile morphemes that I have in my Tseltsolian; in an earlier stage, the stable ones used to attract word stress and subsequently effects connected to it like voiceless stop aspiration, while the labile ones could only receive the stress in the initial position, if there were no stable ones in the word. Later, the stress system was rebuilt and the currently visible stability-related phenomena include aspiration or lack thereof, vowel mutation due to reduction, vowel elision and the debuccalization of /k/ to a glottal stop.

A morpheme can take on a strong, weak or intermediate form; they depend on its 'stability', position in the word and properties of other morphemes in it.

[----]

I'm not sure how weird exactly this is, but I've never seen it in any other language. !Xóõ, for many verbs, has entirely different roots depending on whether a group of people or a single person is doing the action or having the state. Ex: tshûu vs. !ʻáã - both mean 'to sit' but the first is used with singular subjects, the second with plural subjects. They appear to not be phonologically related in any way.

[Mecislau]

I'm not sure how weird exactly this is, but I've never seen it in any other language. !Xóõ, for many verbs, has entirely different roots depending on whether a group of people or a single person is doing the action or having the state. Ex: tshûu vs. !ʻáã - both mean 'to sit' but the first is used with singular subjects, the second with plural subjects. They appear to not be phonologically related in any way.

It's called Pluractionality.

[Miekko]

• Proto-Lakes Plain is reconstructed as having a consonant inventory of /p t k b d/.

Not particularly weird - of the usual stops, /p/ and /g/ are the most likely to be missing.

[chris_notts]

• Proto-Lakes Plain is reconstructed as having a consonant inventory of /p t k b d/.

Not particularly weird - of the usual stops, /p/ and /g/ are the most likely to be missing.

Does he mean and nothing else? I.e. no nasals, fricatives, rhotics, laterals, semi-vowels...? That would be a bit weird, to have a stop-only consonant inventory.

[Solarius]

• Proto-Lakes Plain is reconstructed as having a consonant inventory of /p t k b d/.

Not particularly weird - of the usual stops, /p/ and /g/ are the most likely to be missing.

Does he mean and nothing else? I.e. no nasals, fricatives, rhotics, laterals, semi-vowels...? That would be a bit weird, to have a stop-only consonant inventory.

Yes. Those are the only phonemes Proto-Lakes Plain is reconstructed as having.

Anyway, I had some good things for this thread, but Nortaneous took mine from this thread's CBB sister!

(goes to go find some nutinesss)

EDIT: Found some. Tariana has aspirated nasals and an aspirated /w/.

[chris_notts]

EDIT: Found some. Tariana has aspirated nasals and an aspirated /w/.

Tariana has everything. Well, not literally, but it has an awful lot of grammatical complexity crammed into it. I own a copy of Aikhenvald's grammar of Tariana (is that the only complete grammar of the language in English?)...

[Yng]

Where can I get one of these?

[Radius Solis]

• Proto-Lakes Plain is reconstructed as having a consonant inventory of /p t k b d/.

Not particularly weird - of the usual stops, /p/ and /g/ are the most likely to be missing.

Does he mean and nothing else? I.e. no nasals, fricatives, rhotics, laterals, semi-vowels...? That would be a bit weird, to have a stop-only consonant inventory.

That's what I understood him to mean, and it may well be the best reconstruction, but that's no reason to assume the original language was limited to those consonants. Information loss can happen over time - even if the language had ten consonants, there is no guarantee its daughters will collectively retain enough traces of the original contrasts that they'd all be reconstructible millennia later. For example PIE can be reconstructed with only *e *o for vowels, even though all known languages have a height contrast in their vowel systems and therefore the balance of probability is that PIE did too.

Or, Proto-Lakes Plain may have been like Central Rotokas, which has only /p t k b d g/ for its entire consonant system. We will probably never know.

[chris_notts]

That's what I understood him to mean, and it may well be the best reconstruction, but that's no reason to assume the original language was limited to those consonants. Information loss can happen over time - even if the language had ten consonants, there is no guarantee its daughters will collectively retain enough traces of the original contrasts that they'd all be reconstructible millennia later. For example PIE can be reconstructed with only *e *o for vowels, even though all known languages have a height contrast in their vowel systems and therefore the balance of probability is that PIE did too.

But the typical reconstruction of PIE does have the phones [ i] and , IIRC, it's just that they're vocalic versions of /j/ and /w/. It's purely an analytical choice, based on the symmetry between the behaviour of the glides and the other consonants that can be syllabic (e.g. the nasals), to simplify the description. And while it may do that, it also has the result of making comparisons of PIE phonemic inventories with other languages misleading, because in other languages which have alternations between glide and vocalic allophones, it's normal to take the vowel realisations as basic, not the glide realisations.

The main exception to this is a few other languages in places like the Caucasus and Papua New Guinea which are also analysed as having extremely small vowel inventories, with the analysis typically relying on positing underlying glides that turn into vowels and/or colour neighbouring vowels and merge with them. For example, I think this analysis has been applied to Kabardian, some of the Ndu languages, Kalam, etc.

[Nortaneous]

EDIT: Found some. Tariana has aspirated nasals and an aspirated /w/.

Are they actually aspirated, or just partially devoiced, as most voiceless nasals/liquids are?

Angami has nasals with a positive VOT. (i.e. actually aspirated)

[Solarius]

EDIT: Found some. Tariana has aspirated nasals and an aspirated /w/.

Are they actually aspirated, or just partially devoiced, as most voiceless nasals/liquids are?

Angami has nasals with a positive VOT. (i.e. actually aspirated)

Alas, my Tariana grammar doesn't say, but since it doesn't really elaborate much on them, I would suspect that they are actually aspirated.

[Jashan]

Some Khoisan languages have a mandatory verbal suffix that indicates whether the subject of the sentence is standing, sitting, or lying down. e.g. "I ate (while sitting down)" is one single verb.

In college I had a professor of linguistics who specialized in Native American languages, specifically Yuchi / Euchee. In Yuchi, noun "genders" are instead based on position: standing, sitting, or lying down. This is often somewhat symbolic (trees are 'standing', rivers are 'lying', etc.) and is more like "standing" = "taller than it is wide", "lying" = "wider than it is tall", and "sitting" = "about as wide as it is tall."

Every noun has a 'default', but the classes can be changed for humorous or poetic effect. For instance, my instructor once asked some elderly woen whether a penis was considered to be standing, sitting, or lying. After much giggling, the women replied that the default was "sitting", but you could definitely make a joke about a "standing" or "lying" penis as well.

Also, for weird feature, Yuchi also has mandatory pronoun distinctions for not only number and person, but also relative age (younger or older than the speaker) and ethnicity (Yuchi vs. non-Yuchi).

[Niedokonany]

EDIT: Found some. Tariana has aspirated nasals and an aspirated /w/.

Tariana has everything.

Except native speakers, these days.

Some Khoisan languages have a mandatory verbal suffix that indicates whether the subject of the sentence is standing, sitting, or lying down

How do they classify kneeling or squatting?