Bizarre Sound Changes

[kodé]

Some Levantine Arabic varieties: short *i and *u are neutralised to schwa in stressed syllables but remain separate in unstressed syllables (for examples see Wikipedia).

Whoa, super weird! Also some theories of vowel stress and quality (e.g., Paul de Lacy's) predict that this should be impossible.

[Vijay]

I read a paper years ago on spontaneous nasalization. From the little I can remember, the argument was that it's related to aspiration, and IIRC both aspiration and nasalization blur the lower formants of vowels. I should probably look that paper up...

Oh wow, that sounds very interesting, thanks! Don't worry too much about finding the paper or anything for me, though. I'm awful about actually reading stuff.

[jmcd]

Machvano Vlax Romani is a variety of (Vlax) Romani formerly spoken in Mačva, Serbia but now spoken almost exclusively in the Americas (and especially in Central and South America) IIRC. In Machvano Vlax, apparently, this sound change took place (and the pronunciation of these sounds is a shibboleth of Machvano Vlax; I'm not aware of any other variety of Romani in which this particular sound change took place):

/t͡ʃʰ d͡ʒ/ > [ʈr ɖr]

I wonder how often something like that happens.

That looks like it might a backwards sound change, like when acrolectal varieties of Réunion Creole use unetymological high rounded vowels for etymological high unrounded vowels e.g. [føv] for fèv (bean).

Having researched on Picard dialect at Edinburgh University (I'm seeing my family on holiday), I have discovered that my initial impression of fev>føv being a hyperacrolectalisation is probably wrong in fact. It is likely of Picard origin, as is the case with many other longstanding features of Réunion Creole. The exact same change occurred in Picardy, where it is due to the surrounding labials.

[TaylorS]

English ɡenerally has [ɻ] or [ɹ] while some dialects have [ʁ]

Which English dialects have [ʁ]???? [ɔzɛʁ zan fʁɛnʃ pipɔl]???

http://en.wikipedia.org/wiki/Northumbrian_Burr
It's apparently somewhat out-dated and pretty rare, though...

Someone posted on here a long time ago a link that showed that in some American dialects the /l/ is in fact shifting to uvular R in some poisitons. Its just an allophone, and /l/ is a very flexible consonant, so they still have [l], they just also have [R]. I do it myself on purpose most of the time, I have to admit, because it "feels right". e.g. /lORipOp/ for "lollipop". Basically all dark /l/ becomes [R]. One of the few downsides of having 600 conlangs to maintain without knowing a single real-world language other than my native.

This is just asking for a future chain shift of /ð/ > /d/ > /l/ > /ʁ/

[----]

Armenian /ʁ/ comes from a velarized l which was probably plain at some point in the past.

[----]

In Yaur, an Austronesian language, we have the change *m,n > g/_r. (cf. Kamholz 2014)

[opipik]

Trung (rGyalrongic? I don't think it's the same language as the Nungic Trung) apparently had z → k.
Wassu (rGyalrongic too) had ŋ → j.

(I was browsing the Index Diachronica)

[gufferdk]

Some Danish dialects do some weird things with high vowels with stød. In a lot of places these changes got, or are getting fully or partially reverted though. Source is http://jyskordbog.dk :

Midvestjysk + Sydfanø + Rømø (also Sydsamsø and Vesthimmerland, though with unlabialised k, and only in codas):
iˀ yˀ uˀ > ikʲ ykʲ ukʷ

In Vesthardsysselsk (subdialect of Midvestjysk) also usually: eˀ oˀ > ejkʲ owkʷ (øˀ > øjkʲ only sporadically, or not at all)

The vowels with stød are historically long, Rømø also has a somewhat similar development for long wowels without stød:
iː yː uː > iɣʲː yɣʲː uɣʷː (> Vŋː / _N, no length in codas)

Thy and Mors have iˀ yˀ uˀ > ix yx ux (possibly ç after i, y)

Vestsønderjysk (unconditionally) and Vendsyssel (/_{s,#}) have iˀ yˀ uˀ > itʲ ytʲ uk

All the unconditional changes affecting the vowels with stød have following schwa-insertion except in codas and possibly before /s/.

I wonder if this is at all related to the changes in Faroese posted by Magb:

Faroese has a pair of sound changes that are very roughly:
Ø > ggj / V[+long][+front]_#
Ø > gv / V[+long][-front]_#

Examples: oy > oyggj, trú > trúgv

<ggj> is pronounced [dʒː].

[Cedh]

A similar change is also attested in Huishu (Tangkhul, Sino-Tibetan), where *i *u in word-final open syllables became /ik uk/ [icʰ ukʰ].

Source: David R. Mortensen (2004), The emergence of dorsal stops after high vowels in Huishu, Proceedings of the Annual Meeting of the Berkeley Linguistics Society 30/1, pp. 292-303.

(The article also mentions a couple of other languages with apparent word-final obstruent epenthesis after high vowels: Lom (Austronesian) has *i *u > /ic ek/, Singhi (Austronesian) has *i *u > /is ux/, Maru/Langsu (Lolo-Burmese, Sino-Tibetan) has *əj *əw > **i **u > /it uk/, and Moghamo (Grassfields Bantu, Niger-Congo) may have *i *ɨ *u > /ek ək ok/.)

[Soap]

Are these changes independent of tone? This sounds like the sort of thing I'd expect to see develop from syllable-final [ʔ], which is often present only on specific tones. In fact, that was one of the key sound changes I used in my conlangs to get rid of tone: high-tone open syllables all developed [ʔ] at the end, which later in some daughters changed to [k] or, when after a rounded vowel, [p] (/uk/ > /ukp/ is known from Vietnamese).

[Cedh]

According to Mortensen, [ʔ] cannot have been an intermediate in Huishu because there was an independent coda /ʔ/ which was not affected by the change. He argues that the sequence was > [i̥ u̥] > [iç ux(ʷ)] > [icʰ ukʰ]; i.e. via vowel devoicing and fricatives, with the last step helped by the fact that the language already had onset /kʰ/ and coda /p ʔ/, but no phonemic dorsal fricatives.

[dhok]

Everyone likes to trot out Plains Algonquian for bizarre North American sound changes, but Salishan is just as bad. Sources are Galloway's 'Some proto-central Salish sound correspondences' in In Honor of Mary Haas, and chapter 1 of Kroeber's The Salish Language Family.

*p *p' *m to tʃ tʃ' ŋ in most of Northern Straits, except before *u, where they are retained.

A chain shift *a > ɛ and *u > a in Halkomelem and some of Northern Straits.

*t' > tɬ', unconditionally, in Shuswap, Thompson and Lillooet.

Fortition of *w *y to kʷ tʃ (fixed!) before a vowel in half of Northern Straits, which isn't that weird, I suppose, but most such fortitions seem to give sonorants...

Additionally, Salishan provides numerous examples of 'inverted roots', where a root will appear as C₁VC₂ in some forms, and as C₂VC₁ elsewhere. It's unclear what's going on--it's not regular, but it's too common to be coincidence or spoonerism. Noonan (1997, 'Inverted Roots in Salish') gave a good hundred examples of such sets, and noted that reduplication and ablaut seem to have something to do with the process, but the problem is unsolved.

[jmcd]

Is that y>kʷ and w> tʃ? The weirdest thing in that would be that it isn't w>kʷ and y> tʃ.

[Pole, the]

I believe it was a typo.

[dhok]

Yeah, typo.

[Markski]

More a snychronic rule, but Hup the plain voiced stop series (as opposed to the plain voiceless and glottalized, which lacks phonemic voice) is pre-nasalized word initially and post-nasalized word-finally. Now, Hup also has a rule that it really likes syllables to be in the CVC format, so if we have a voiced stop in the coda of a CVC root and add a vowel-initial suffix, we get a a post-nasalized stop followed by a pre-nasalized stop, resulting in what Epps (2008) calls a "nasal contour"

/wób-óy/ [wób^m.^mbóy] [rest.on-DYNM] "be resting on (sth)"
/tǒɟót/ [tǒ^jdⁿ.ⁿd^jót̚] [nose-OBL] "in the nose"

In other words essentially D (> DD) > DND.