Whence aspirated stops and nasal vowels?

[Ketumak]

I have looked for these on the wider internet, but not had much luck. I turned up an old paper arguing that aspirated descended from glottalised stops. That may well be one source, but there seem to be quite a few languages around with both.

I suppose a language could develop glottalised stops which turn into aspirates and then later develop glottals again. Or maybe two types of glottalised stop are involved, one turning into aspirates, one not?

Are there any other theories as to where aspirated stops come from?

I'm also looking into the origin of nasal vowels. I always thought there was just one source:

VN > Ṽ

I read recently that the transfer of nasalisation has to take place before the N disappears. so the sequence would be:

VN > ṼN > Ṽ

This makes sense to me, but Wikipedia writes:

"In Min Chinese, nasal vowels carry persistent air flow through both the mouth and the nose, producing an invariant and sustainable vowel quality. That is, this type of nasalization is synchronic and suprasegmental to the voicing. In contrast, nasal vowels in French or Portuguese are transitional, where the velum ends up constricting the mouth airway."

I can see how VN can give rises to the transitional kind of nasal vowel, the constriction being a relic of the old nasal stop, but where do the suprasegmental kind come from? I can only imagine them being a later development from the transitional ones. Is this assumption correct?

Thanks for any light people can shed on all this.

[Salmoneus]

Regarding aspirated stops: I believe they develop from unaspirated stops through a process of aspiration.

As you know, when we say a language contrasts voiced and voiceless stops, what we really mean is that there are two series of stops distinguished by the VOT. If it's early, it's called a 'voiced stop', if it's later it's called a 'voiceless stop', and if it's very late (overlapping the following vowel) it's called an 'aspirated' stop. [And if it overlaps the entire following vowel, it's called a voiceless stop followed by a voiceless vowel]

So to make voiceless stops aspirated, you just shift the VOT marginally. Which might happen because, eg, it increases the distinction between them and the voiced stops.

[linguoboy]

I'm also looking into the origin of nasal vowels. I always thought there was just one source:

VN > Ṽ

Often, it's enough for the nasal to be adjacent. Examples of progressive nasalisation can be found even in dialectal French (e.g. Normand (Val de Saire) amin "friend"; Franco-provençal (Bagnes) [nẽĩ] "nest") and there are numerous languages where it continues throughout a word unless blocked by an oral segment.

I can't really speak to the "two types of nasalisation" except to point out that Min nasal vowels pretty clearly derive from VN sequences, e.g. OC *sɯːn(s)-sreŋs > Taiwanese Hokkien sian-siⁿ (SC xiānshēng).

[Sumelic]

"In Min Chinese, nasal vowels carry persistent air flow through both the mouth and the nose, producing an invariant and sustainable vowel quality. That is, this type of nasalization is synchronic and suprasegmental to the voicing. In contrast, nasal vowels in French or Portuguese are transitional, where the velum ends up constricting the mouth airway."

That's an odd quote, since my impression was that French nasal vowels are invariant. What I remember learning was that in languages like French, nasal vowels are monophthongal, and as far as I know have persistent nasal air flow, while in languages like Portuguese, Polish or Japanese, nasalized vowels are realized by starting with an oral vowel, and ending with a nasalized offglide. It may also be relevant that in both Polish and Japanese, nasalized vowels also alternate allophonically with oral vowel + nasal stop sequences before stops and affricates, while this is not the case in French.

Sorry to only provide more quotes from Wikipedia, which is not a reliable source, but I'll try to find a better source later:

Unlike French but rather like Portuguese ão, nasal vowels in Polish are asynchronous, meaning that they are pronounced as an oral vowel + a nasal semivowel [ɔw̃], or a nasal vowel + a nasal semivowel. For the sake of simplicity, it is sometimes represented as /ɔ̃/.

(This is from [https://en.wikipedia.org/wiki/Ą#Polish] but for some reason the bbcode isn't processing this url correctly.)

Wikipedia on Polish nasal vowel allophony (I don't know if it's synchronic, or merely historical):

The nasal vowels do not occur except before a fricative and in word-final position. When the letters ą and ę appear before stops and affricates, they indicate an oral /ɔ/ or /ɛ/ followed by a nasal consonant homorganic with the following consonant. For example, kąt is [kɔnt] ('angle'), gęba ('mouth') is [ˈɡɛmba], and pięć ('five') is [pjɛɲt͡ɕ].[5] Before /l/ or /w/, the nasality is lost altogether and the vowels are pronounced as oral [ɔ] or [ɛ]. Many speakers also frequently denasalize /ɛ̃/ to [ɛ] in word-final position.

[----]

Several languages of New Caledonia have interesting sources of both aspirated stops and nasal vowels. Aspirated stops often derive from geminates, e.g. Pwapwâ tʰai < *ssa-i < *sasa-i. Vowels seem to nasalize out of nowhere sometimes, while other times they come from oral vowels adjacent to nasals, and in a couple languages they come from oral vowels after /h/.

[vokzhen]

Aspiration comes pretty commonly from voiced stops, they become breathy as a means of ease of pronunciation, and then end up devoicing to aspirated stops. Though off the top of my head the examples of such a shift already have aspirates, the voiced>aspiration change just increases the frequency. Well-attested in Chinese languages and Armenian at the very least, and I'm sure quite a few others, as well as hypothosized (without already-existing aspirates) at least for Late PIE > Greek.

Korean there's a set that's aspirated word-initially, voiced between sonorants, and plain elsewhere. Originally probably from a voiced set that devoiced everywhere except between sonorants (they still carry the low tone characteristic of voiced stops), and colloquially started being aspirated word-initially.

The Sotho-Tswana languages have aspirates from old prenasalized voiceless stops, e.g. nt > tʰ.

Languages without voicing or aspiration distinctions - either only plain or plain/ejective - fairly often seem to have allophonic aspiration in the coda before other stops or word-finally, though that may be an areal feature of Mesoamerica I've misinterpreted as a common feature. Mayan and Mixe languages, for example, aspirate final stops and stop-stop clusters either aspirate or don't release the first (free variation). It's not entirely just Mesoamerican, though, as stops can undergo "final devoicing" to something that's more aspirated than a normal plain stop in a number of languages. Iirc that includes Armenian and some of the Northwest Coast languages; I believe this is my source on that but I can't remember my old university login to get past the paywall.

Some Spanish dialects are in the process of gaining aspiration via s > h in the coda, which can turn a following stop into an aspirate and delete, so instead of standard /estaðo/ and widespread [ehtaðo], it's instead [etʰaðo]. A similar change is postulated for Sino-Tibetan where a preinitial (derivational prefix) s- aspirates a following stop in some daughters, including Chinese.

There's a few languages where aspirates are likely an actual cluster of C + h. Khmer is probably the most well-known one, where "aspirated" consonants can be split from Cʰ to C<infix>h. Tariana has an incredibly odd rule where an /h/ in a root or suffix after a prefix or root shifts to the onset of the first syllable. This creates new instances of aspirated stops, and as far as I can tell some of the other "aspirated" sounds like /dʰ/ and /wʰ/ seem like they came entirely from that rule before undergoing marginal phonemicization.

Also speaking of Khmer, it allophonically aspirates stops not only before other stops but also nasals (as in the name) and /l j w/. Notably implosives don't cause aspiration, tɓ- not tʰɓ-.

[Buran]

Some Spanish dialects are in the process of gaining aspiration via s > h in the coda, which can turn a following stop into an aspirate and delete, so instead of standard /estaðo/ and widespread [ehtaðo], it's instead [etʰaðo].

I've never heard of that before, but that's extremely interesting. Could you provide a link or more information?

[KathTheDragon]

Aspiration comes pretty commonly from voiced stops

I'd intuit that it's easier to derive from voiceless stops than voiced. After all, it's voiceless stops that allophonically aspirate in English, not voiced stops allophonically becoming breathy.

[Richard W]

Stop + /r/ can give aspirate stops, especially if there is tendency for /r/ > /h/. The change is late enough to manifest itself in the spelling in Northern Thai, and an earlier round of this change is now hypothesised to be the origin of what Li had as Proto-Tai voiceless aspirates.

In Pali, Old Indic sibilant + stop and stop + sibilant gives an aspirated geminate stop, which simplifies to a simple aspirate word-initially and after nasals.

[vokzhen]

Some Spanish dialects are in the process of gaining aspiration via s > h in the coda, which can turn a following stop into an aspirate and delete, so instead of standard /estaðo/ and widespread [ehtaðo], it's instead [etʰaðo].

I've never heard of that before, but that's extremely interesting. Could you provide a link or more information?

I'm not finding my original source that says the coda h can be dropped as well, but this has measurements of VOT for -C- and -hC- and this has some more. This actually proposes metathesis of coda h into the onset.

Aspiration comes pretty commonly from voiced stops

I'd intuit that it's easier to derive from voiceless stops than voiced. After all, it's voiceless stops that allophonically aspirate in English, not voiced stops allophonically becoming breathy.

It's certainly possible. I was just taking a quick mental note of how many instances of p>pʰ versus b>pʰ I knew of once and I wasn't really coming up with more instances of origin from plain stops than voiced, but there's also a lot more available info about European (where they tend to come from plain) and Chinese-influenced (where they tend to come in part from voiced and in part from other sources) languages than other parts of the world. The latter especially is probably areal feature rather than underlying tendency. I also didn't intend to imply voiced stops are the the most likely source, but a well-known one.

[Sumelic]

As others have mentioned, nasal vowels don’t have to come from VN specifically, but nasality in vowels often originates from assimilation to other nasal segments (which can also include preceding or nearby nasal consonants, or other nasal vowels). This is the general pattern I see, but the set of languages I’m familiar with might be biased.

What’s weird is that as far as I know, nasality in consonants most often also originates from assimilation to other nasal segments, like other nasal consonants or nasal vowels. (By "originates" above of course, I'm only referring to cases when an origin can be determined. Some sounds may have been nasal for as far back as we can tell.)

What I mean by this is, most languages I can think of that have a development (synchronic or diachronic) of a non-nasal consonant to a nasal consonant do so only in the presence of another nasal segment.

Example: In the history of Korean and Latin, you get regressive assimilation from nasal consonants to preceding non-nasal plosives. In Korean this is fairly recent and still reflected in the orthography, so I won’t give any examples. In Latin, two examples are *swepnos > somnus, *dh2pnom > damnum.

There are plenty of examples of languages where nasal vowels cause synchronic assimilation of voiced plosive phonemes, or sometimes other voiced phonemes, to nasal allophones.

But I can’t think of many examples of nasal sounds arising from only non-nasal sounds. The opposite seems more common (the “lazy pronunciation” of /n/ as [l] in Cantonese, the currently in-progress denasalization of word-initial nasal stops in Korean). It makes me wonder why nasals don’t get eliminated entirely from a language... although as long as a language preserves some nasals, nasality can always spread from them to other sounds, so I guess the extinction of nasals is not theoretically inevitable after all.

I guess voiced plosives can sometimes evolve into nasals? Japanese voiced plosives are the first example I can think of, though they might not be the best, since I've read somewhere that they might historically have been pronounced as prenasalized stops. But anyway, in some modern varieties of Japanese, intervocalic /g/ is pronounced as [ŋ]. And IIRC there's a variety of Rotokas with a phoneme inventory where voiced plosives and nasals are in free variation allophonically.

The only tendency I can think of that causes nasals to arise completely from non-nasals has already been mentioned, rhinoglottophilia.

[gestaltist]

Sorry to only provide more quotes from Wikipedia, which is not a reliable source, but I'll try to find a better source later:

Unlike French but rather like Portuguese ão, nasal vowels in Polish are asynchronous, meaning that they are pronounced as an oral vowel + a nasal semivowel [ɔw̃], or a nasal vowel + a nasal semivowel. For the sake of simplicity, it is sometimes represented as /ɔ̃/.

As a native Polish speaker, I can confirm this is in fact what’s (mostly) happening. It’s especially noticeable in word-final /ɔ̃/

Wikipedia on Polish nasal vowel allophony (I don't know if it's synchronic, or merely historical):

The nasal vowels do not occur except before a fricative and in word-final position. When the letters ą and ę appear before stops and affricates, they indicate an oral /ɔ/ or /ɛ/ followed by a nasal consonant homorganic with the following consonant. For example, kąt is [kɔnt] ('angle'), gęba ('mouth') is [ˈɡɛmba], and pięć ('five') is [pjɛɲt͡ɕ].[5] Before /l/ or /w/, the nasality is lost altogether and the vowels are pronounced as oral [ɔ] or [ɛ]. Many speakers also frequently denasalize /ɛ̃/ to [ɛ] in word-final position.

It’s synchronic. In careful pronunciation, this denasalization doesn’t always happen. <kąt> is typically pronounced [kɔnt] but you could also hear [kɔw̃t], especially if the word is pronounced on its own and not as a part of a sentence. (I wonder if you could use a similar process in a conlang to have /n/ > /w/.)

Historically /ɑ̃/ > /ɔ̃/, hence the non-intuitive orthography for that sound: <ą> (Polish has completely lost the /ɑ/, apparently due to Eastern influences.

[Nortaneous]

Tsakonian:
sp st sθ sk sx > pʰ tʰ tʰ kʰ kʰ
mf nθ ŋx > pʰ tʰ kʰ
kt xθ > tʰ
rp rt rð rk > mb nd nd ŋɡ

Arapaho:
http://www.academia.edu/2107195/The_sou ... in_Arapaho

Nyole:
*p > *h > ŋ

Kinalug:
*r > n / _C

Some Aslian languages turned word-final (syllable-final?) plosives to prestopped nasals.

The Correspondence Library says Vietnamese had *ɓ *ɗ > m n at some point.

[Sumelic]

Tsakonian:
sp st sθ sk sx > pʰ tʰ tʰ kʰ kʰ
mf nθ ŋx > pʰ tʰ kʰ
kt xθ > tʰ
rp rt rð rk > mb nd nd ŋɡ

Do you know if all of these are diachronic changes, or are some or all just synchronic correspondences with other dialects of Greek?

[Ketumak]

Wow! Thank you all. This is very informative.

For me the headline lesson here is the variety of possible sources for both these types of sound. It shows how normal they are, I tend to think of aspirated stops as a bit esoteric and therefore only arising under certain specific conditions. This is interesting from a general linguistic point of view and from a conlanging point of view it's good to see I have a range of options for deriving them. I'll say more on the conlanging applications another day in the C&C forum.

I'll post more here tomorrow, just dropping by to read the thread again and say I'd not forgotten. Can't believe it's three days since I asked, but I've been busy.

[Ketumak]

Regarding aspirated stops: I believe they develop from unaspirated stops through a process of aspiration.

... So to make voiceless stops aspirated, you just shift the VOT marginally.

Surprisingly, I didn't think of this one. I got hung up on the idea that sound shifts have to have external causes. But, yes, they could just happen or have a motivation (rather than a cause), as you say. This sounds like a versatile and usable reason for conlanging purposes, too.[/quote]

Well, there's a reason here. You've got two series of stops distinguished primarily by VOT, this leads to people having trouble telling them apart sometimes, so the distinction is heightened by making the two VOTs more different - by shifting one later, making the stops aspirated.

And really, aspiration is not a rare thing! It's widespread in Indo-European, Sino-Tibetan, Dravidian, Tai-Kadai, and probably other families too.

[Nesescosac]

Sumelic's second post gives me this image of nasalisation spreading forwards and backwards along a clause, like a bushfire which sounds like good metaphor to work with.

If this interests you, check out the phenomenon of nasal harmony, as found most famously in Tupian languages like Guaraní.

[CatDoom]

A phenomenon similar to the one Theta describes in Pwapwâ was a major feature of the mainland Chumashan languages of Southern California. Synchronically, where morphology would be expected to produce voiceless geminate consonants, the result was always realized as an aspirated singleton. This produces topologically unusual aspirated fricatives, as well as aspirated stops and affricates.

[Nortaneous]

A phenomenon similar to the one Theta describes in Pwapwâ was a major feature of the mainland Chumashan languages of Southern California. Synchronically, where morphology would be expected to produce voiceless geminate consonants, the result was always realized as an aspirated singleton. This produces topologically unusual aspirated fricatives, as well as aspirated stops and affricates.

Speaking of aspirated fricatives:

Cs- z- s- > s- s- sʰ-
Cz- Cs- z- s- > z s s sʰ
*xʔ > sʰ

[Ketumak]

nasal harmony, as found most famously in Tupian languages like Guaraní.

That's very interesting. An average sized phoneme inventory, but with all sorts of allophony going on, including nasal harmony, as you say.

CatDoom and Nortaneous mention aspirate fricatives - I was wondering if there was such a thing, so thank you, too.

[Nortaneous]

yep -- mostly an areal feature of East Asia. most extensive in Tibetan (esp. Cone) but also present in Burmese, Korean, some Hmongic, and dialects of Pumi. generally not used in conlangs, along with all the other weird shit in East Asia. also appears in some American langs. Mongolic and Armenian occasionally pattern fricatives (only unvoiced fricatives in Armenian I think) with aspirated stops.

http://www.academia.edu/2118759/Proto-Q ... nt_Changes
http://www.academia.edu/3318754/A_contr ... fricatives
http://www.academia.edu/3318814/The_ris ... ves_poster_

and apparently some speakers of Scottish English have (nonphonemic) *pre*aspirated fricatives
http://www.academia.edu/2940161/Non-nor ... cteristics

[Hallow XIII]

By the way, the aspiration of voiceless stops after nasals has the motivation of increasing the salience of the voiceless stop. In Bantu languages, presumably to increase differentiation from sequences of nasals and voiced stops.

[CatDoom]

CatDoom and Nortaneous mention aspirate fricatives - I was wondering if there was such a thing, so thank you, too.

I'm mildly embarrassed to admit it, but I actually have a conlang that features an aspirated retroflex lateral fricative. There's no reason that such a sound couldn't exist, of course; Toda has a voiceless retroflex lateral fricative and Northern Hmu has two different aspirated lateral fricatives, but it's obviously a very unusual combination of features.

[Thry]

Some Spanish dialects are in the process of gaining aspiration via s > h in the coda, which can turn a following stop into an aspirate and delete, so instead of standard /estaðo/ and widespread [ehtaðo], it's instead [etʰaðo].

I've never heard of that before, but that's extremely interesting. Could you provide a link or more information?

I have this, though it's more like [eʰ.ˈtʰao̯], and I've heard some people where I live go as far as [ˈeʰ.tʃe] for este (but it's not representative, maybe it was an anomally).

I wonder to what extent it happens with /p/ and /k/ - I've paid less attention to it.
/ka.ˈpi.ʝa/ [ka.ˈpi.ʝa] capilla 'chapel' (for comparison)
/res.ˈpwes.ta/ [reh.ˈpweh.tʰa] respuesta 'answer' but
/es.ˈpi.na/ [eh.ˈpʰi.na] espina 'thorn'
/es.ˈpa.ra.go/ [eh.ˈpʰa.ra.ɣo] espárrago 'asparagus'

Maybe it's not about -t- and being tonic or not but about what follows? The /we/ diphthong seems to block it. But I'm not sure, it's so subtle, and the -esta has a very notable aspirate t.

[Thry]

It's also worth noting it's not only coda /s/ that triggers this phenomenon because basically most codas are [h] and trigger aspiration in my dialect - so for instance actitud is also ahthitú(h), apto/acto áhtho, etc... practically anything that isn't a nasal or a liquid. Though by no means a common ocurrence, if I were to pronounce igteísmo in my D1 I would say ihtheísmo - basically any plosive is [h] (atlas = áhlah, this one is more common).

In one of my romlangs I have also turned /kt/ into aspirate t, so that noct or lect are read noth and leth (night, bed) - but not coda -pt.