The weird natlang phoneme thread

[vec]

Palatal lateral ejective affricate

[Grunnen]

I'd like to hear that one, cause the sound coming out of my mouth when I try it doesn't sound very convincing at all.

[Nortaneous]

Teribe has a prelabialized retroflex labial flap ʷɺ̢ .

Yi has four fricative vowels out of a total set of ten vowels: /ɿ ʮ/ <y u> pattern with the high/'tense' vowels /i ɯ u/ <i e o>, and then there are retracted variants <yr ur> (which can only take the mid tone or its falling variant) that pattern with the low/'lax' vowels /e a o/ <ie a uo>. /ʮ/ can be realized as a syllabic bilabial trill.

Big Nambas has linguolabials.

Nias has a bilabial trill that can occur before any vowel and that comes from *mb; there's also a unit trilled-release stop /dr/ that comes from *nd.

Kusunda phonetically has a pharyngealized uvular nasal, but it's probably /ŋʕ/. Greenlandic might have a phonemic uvular nasal.

Kensiu distinguishes all of /i ɪ e̝ e ɛ/.

Central !Kung has subapical retroflex clicks that contrast with alveolar clicks.

Wikipedia: "The XiNkuna dialect of Tsonga has this affricate, as in [tiɱp̪͡fuβu] "hippopotami" and aspirated [ɱp̪͡fʰuka] "distance" (compare [ɱfutsu] "tortoise", which shows that the stop is not epenthetic), as well as a voiced labiodental affricate, [b̪͡v], as in [ʃileb̪͡vu] "chin". There is no voiceless labiodental fricative [f] in this dialect of Tsonga, only a voiceless bilabial fricative, as in [ɸu] "finished". (Among voiced fricatives, both [β] and [v] occur, however.)"

[CatDoom]

Hmu has a total of six aspirated fricatives, including two laterals: /fʰ sʰ ɬʰ ɕʰ ɬʲʰ xʰ/, though the aspirated velar fricative doesn't contrast with a corresonding plain voiceless version.

As of the 1930's, Badaga apparently had two degrees of "r-coloring" or retraction on all five of it's vowels, though some of these distinctions are no longer productive.

[vec]

Big Nambas has linguolabials.

Apparently, so do a few other languages. New mission. Make conlang with linguolabials. Do we have any data on how they arise?

[Nortaneous]

Labials before unrounded vowels.

This makes it sound like Kajoko unconditionally shifted /b/ to a linguolabial.

[----]

Teribe has a prelabialized retroflex labial flap ʷɺ̢ .

O'odham has a retroflex lateral flap as well. What's more surprising is that it's the only liquid consonant the language has.

Yeli Dnye has an array of unusual coarticulated consonants. These can even be palatalized and labialized (or both): /mʷʲa:/ and /t̠͡pⁿʲa/ are examples of attested words with some really strange consonants.

[Nortaneous]

The operative word there is 'prelabialized', but as for languages that have an odd liquid as their only liquid, Hiw's only liquid is gʟ.

Aspirated fricatives are an areal feature of East Asia. Burmese and Korean are the only major languages to have them, and they only have /sʰ/, but a lot of minor languages have larger sets of them. dGudzong Tibetan even has preaspirated aspirated fricatives /ʰsʰ ʰɕʰ/, contrasting with /ʰs sʰ ʰɕ ɕʰ/, but the preaspiration may be best analyzed as a consonant cluster -- the paper I get this from also lists consonants like /ᵐg/ and /ᵖʈʰ/.

[gach]

Most Iwaidjan languages contrast both alveolar and retroflex lateral approximants with lateral flaps in addition to already contrasting the non-lateral retroflex approximant and flap. In Iwaidja this contrast might also be phonemic with palatal laterals.

dGudzong Tibetan even has preaspirated aspirated fricatives /ʰsʰ ʰɕʰ/, contrasting with /ʰs sʰ ʰɕ ɕʰ/, but the preaspiration may be best analyzed as a consonant cluster -- the paper I get this from also lists consonants like /ᵐg/ and /ᵖʈʰ/.

The paper is at http://ir.minpaku.ac.jp/dspace/handle/10502/4629. Besides preaspiration there are complex initials with prenasalisation (including the nasal POA contrasts /ŋg/ vs. /mg/ and /n̪̥tsʰ/ vs. /m̥tsʰ/, Sect. 2.2.2.2), labial prestopping (Sect. 2.2.2.3) and continuant prelabialisation (realised as /ɸ ~ w/, Sect. 2.2.2.4). I can't see any reason why it would be preferential to analyse these as unitary phonemes instead of consonant clusters and this is also the position the paper takes. Expect for an alveolar ~ retroflex allophony in the nasal preinitials, all of the preinitials also appear as fully independent simple initials. The preinitial consonants are, however, always articulated as weaker than the main initial following them which seems to be the reason why the author chose to write them with the potentially misleading superscript notation.

[Buran]

Greenlandic might have a phonemic uvular nasal.

Apparently it's from the cluster /ʁŋ/.

[Porphyrogenitos]

And we can't forget the voiceless bidental fricative /h̪͆/ in Shapsug Adyghe, which apparently developed from /x/.

[vec]

Oro Win, Wari’ and of course Pirahã, all from the Amazon, reportedly have a voiceless bilabially post-trilled dental stop /t͡ʙ̥/. Dafuq?

[TaylorS]

Nobody has mentioned English /ɻʷˤ/, yet? I am disappoint!!!

[Nortaneous]

Khoisan languages:
* Uvularized stops /tᵡ dᵡ tʃᵡ/ (Ekoka !Kung), /c̟χ/ (Nǁng), /dzᵡ dʒᵡ/ (Juǀʼhoan), /tqχʼ tsqχʼ/ (Gǀui), /cqχʼ/ (ǂ’Amkoe).
* Epiglottalized stops /t^H k^H g^ʢ/ contrasting, in the case of the coronal, with uvularized stops. (Juǀʼhoan)
* Voice contours on stops: /b͡pʰ d͡tʰ d͡tsʰ d͡tʃʰ ɡ͡kʰ/. (Juǀʼhoan)
* Voice-contour ejectives: /d͡tsʼ d͡tʃʼ/. (Juǀʼhoan)
* Strident vowels. Nasal strident vowels. (Nǁng)

Also clicks.

[TaylorS]

Khoisan languages:
* Uvularized stops /tᵡ dᵡ tʃᵡ/ (Ekoka !Kung), /c̟χ/ (Nǁng), /dzᵡ dʒᵡ/ (Juǀʼhoan), /tqχʼ tsqχʼ/ (Gǀui), /cqχʼ/ (ǂ’Amkoe).
* Epiglottalized stops /t^H k^H g^ʢ/ contrasting, in the case of the coronal, with uvularized stops. (Juǀʼhoan)
* Voice contours on stops: /b͡pʰ d͡tʰ d͡tsʰ d͡tʃʰ ɡ͡kʰ/. (Juǀʼhoan)
* Voice-contour ejectives: /d͡tsʼ d͡tʃʼ/. (Juǀʼhoan)
* Strident vowels. Nasal strident vowels. (Nǁng)

Also clicks.

People can actually PRONOUNCE this stuff???

[Dē Graut Bʉr]

Khoisan languages:
* Uvularized stops /tᵡ dᵡ tʃᵡ/ (Ekoka !Kung), /c̟χ/ (Nǁng), /dzᵡ dʒᵡ/ (Juǀʼhoan), /tqχʼ tsqχʼ/ (Gǀui), /cqχʼ/ (ǂ’Amkoe).
* Epiglottalized stops /t^H k^H g^ʢ/ contrasting, in the case of the coronal, with uvularized stops. (Juǀʼhoan)
* Voice contours on stops: /b͡pʰ d͡tʰ d͡tsʰ d͡tʃʰ ɡ͡kʰ/. (Juǀʼhoan)
* Voice-contour ejectives: /d͡tsʼ d͡tʃʼ/. (Juǀʼhoan)
* Strident vowels. Nasal strident vowels. (Nǁng)

Also clicks.

People can actually PRONOUNCE this stuff???

Miracles still exist.

[gach]

People can actually PRONOUNCE this stuff???

The keyword here is contour. I get the feeling that accurate pronunciation of a segment like /tᵡ/ is [tχ].

The distribution of consonants in the languages of the Khoisan area is very onset heavy with only a very limited set of single consonants available in syllable final or intervocalic positions. If complex consonant phones only appear in word initial positions, you might end up in a situation where it's not clear whether you should analyse these onsets as consonant clusters or single complex phonemes. Drive for either a maximally simple phonotaxis or phonemic inventory can push you into different descriptions. For the Khoisan languages this is a question of whether you allow any consonant clusters of not.

Personally I like to take the position that there has to be explicit language internal basis for the phoneme status of every proposed phoneme. That might include phonotactic, distributional or allophonic reasoning. This is essentially saying that simply being able to chain /t/ and /ʃ/ doesn't imply that a language has a postalveolar affricate. For that the sound chain has to behave like single phoneme, most likely in parallel to simple stops.

[----]

At least in !Xóõ, the complex onsets seem to behave phonetically like single consonants. The velar/uvular fricative release surfaces as a sort of aspiration.

[Nortaneous]

Right, I should have mentioned Khoisan phonotactics -- if you analyze them as unit consonants, it's all CV with a small set of word-final consonants, but almost all consonants can only occur word-initially.

The Ju|'hoan mixed-voice aspirates can't sanely be analyzed as clusters, unless you also decompose the voiceless aspirates. If you analyze the mixed-voice aspirates and the ejectives as units, why not the mixed-voice ejectives? (Also, mixed-voice aspirates are attested elsewhere as unit consonants, so it's not positing consonants that only occur in Khoisan languages, like the uvularized consonants might be.)

The uvularized consonants are another matter. I think it's simpler to just call them units. But at some level it doesn't really matter; there's no way to tell the difference between units and clusters from a language-internal basis, so if you make that decision, it has to be done based on cross-linguistic sanity, which would probably include other languages where people had to make that decision based on cross-linguistic sanity, and so on. It might be better to just talk about onsets.

(There are languages with very strange phonotactics, of course -- that one Qiang dialect where all clusters are ʂ x χ + C, and the same exact set of clusters appears in both syllable-initial and syllable-final position...)

[----]

Hmu has a total of six aspirated fricatives, including two laterals: /fʰ sʰ ɬʰ ɕʰ ɬʲʰ xʰ/, though the aspirated velar fricative doesn't contrast with a corresonding plain voiceless version.

Yapese has two ejective fricatives: /fʼ θʼ/
Abkhaz has a set of labialized consonants at various points of articulation, but their surface realizations are not completely symmetric, so to speak. The labialized alveolars /dʷ tʷ tʷʼ/ actually surface as coarticulated [d͡b t͡p t͡pʼ].

[CatDoom]

Yapese has two ejective fricatives: /fʼ θʼ/

The Chumashan languages actually had both ejective and aspirated fricatives (Barbareño, for instance, has /s s' sʰ/ and /ʃ ʃ' ʃʰ/, as well as /x x'/, but no *xʰ), which matches the contrasts in the languages' stop series. Synchronically, the asperated series most commonly appeared as the surface realization of geminate stops and fricatives, and Proto-Chumashan probably didn't have any phonemic aspirated consonants at all.

[Zaarin]

Tlingit has labialized ejective uvular and velar fricatives /xʼʷ χʼʷ/.

[Nortaneous]

Yapese has C + glottal stop sequences, not ejective fricatives

[Hallow XIII]

there's no way to tell the difference between units and clusters from a language-internal basis, so if you make that decision, it has to be done based on cross-linguistic sanity

wat

There are plenty of languages where there are e.g. strong distributional arguments for whether a given segment is a cluster or not.

[Trebor]

I'd like to know if any natlangs out there:

- make a phonemic distinction, particularly in underived roots, between /h/ ~ /h:/, /X\/ ~ /X\:/, and/or /?\/ ~ /?\:/. I'm familiar with the Arabic words /s_?\aX\:at/ 'health' and /s_?\aX\:aX\a/, but the history of the triconsonantal-root system might have something to do with these.

- allow geminate consonants in clusters, e.g., /m:b/, /nz:/.

- have liberal phonotactics (perhaps somewhere between English and Georgian), yet disfavour or prohibit such basic clusters as consonant plus /l/.