Phonological Gain

[linguoboy]

/_g/ > /M\/ > /h\/ > /+h\/ seems perfectly possible. Still, there are more straightforward ways to acquire breathy voiced stops, I would think. E.g.: lenition of aspirated stops in specific positions followed by the loss of those conditioning factors; simplification of C+/h/ or C+/h\/ cluster; V+/h/ or V+/h\/ leading to breathy voicing on vowels which is then transferred to consonants.

For Middle Chinese you simply have an unconditioned shift of the voiced series to voiced aspirated/breathy voiced. (This eventually become a voiceless aspirate series after tonogenesis in the following vowel nucleus.) Some sublanguages (e.g. Minnan) then acquire a new voiced series from spontaneous denasalisation of nasal stops.

[Grath]

and as a last question, does anyone know a program that is capable from a root word generate soundchanges onto each other ultimately ending with several different sounding words for various languages?

Like this?

[merijn]

One thing I really wonder is, how does a language build up new stop series? In particular, other than that stop~top example, is there another way to build, say, /t_h,t,d/ contrast from /t,d/? How zulu end up with /ph,p,b,b_</ from an original /p,b/?

Let me try to answer the Zulu question. Keep in mind that I am not a historical linguist and not a phonologist, but I do know an awful lot about Zulu. Most of what I say here comes from an article I read years ago which I believe was an MA dissertation but I can't seem to find it here. It is also most certainly not the whole story. I'll use the Zulu graphemes ph for the aspirated plosive, p for the ejective plosive, bh for the voiced aspirated plosive, b for the implosive, mp for the prenasal ejective plosive and mb for the prenasal voiced plosive.

Proto-Bantu had a voiceless plosive series, and a "voiced" series where it is not clear whether it was a plosive series or a series of liquids. It had also prenasal variants of both. Somewhere along the line the normal (that is not-prenasal) voiced series changed; it disappeared in the palatal and velar series; it became (remained?) an l in the alveolar series, and it became an implosive b in the bilabial series. The next step was that voiceless plosives became aspirated, so PB *p became ph. Now we have the following series: aspirated ph voiceless prenasal mp, voiced prenasal <mb> and the implosive b. What happans next is that a consonant-agreement rule has come into effect in Zulu. All plosives have to have the same value for voiced-ness as the initial consonant of the stem. This consonant agreement does not affect prenasal plosives, but prenasal plosives can trigger it. So the hyopothetical pre-Zulu *ndotha "man" became ndoda "man". I couldn't find any examples with a voiceless prenasal plosive but that may have gone something like *ntapha =>ntapa. These agreement rules effectively reintroduced the voiced plosive and the ejective plosive. Also, many nouns starting with a prenasal consonant belong to noun class 9 and lose the prenasalization if they are transferreed to other noun classes or if the are verbed, so the plural of ndoda is madoda and there is a verb -doda " to act like a man". I doubt however that that explains all of the cases of words starting with voiced plosives, there are just too many of them. It could be that most cases of a voiced plosive are loans but it is also possible that there is some additional way of creating words with initial voiced plosives.

[boustrophedon]

Informative answer, Merijn. You don't happen to know the answer to the question about clicks, do you? Like some folks here it seems like there's just too many words with clicks for them all to be borrowings - and I seem to recall that some words with clicks in them may have cognate words in non-click Bantu languages but I could be totally off the mark.

[sangi39]

One thing I read, can't remember what it was though, said that clicks, purely as phonemes, were "borrowed". That is to say that certain native sounds, under the influence of neighbouring languages with clicks and strong contact with the speakers of such languages, became clicks through some sound change or another. That is to say, the phonology of one language led to sound changes causing the non-click language to develop clicks.

Another idea is that clicks arise as the result of complex consonant clusters while another suggests they're the result of either direct borrowing of a word from a click language or a phonological feature used for taboo avoidance which eventualy spreads into other words.

Clicks are one of the worst sound groups in the world, though, for working out how they develop in non-click languages. I don't think there's a single idea right now and maybe only a handful that might actually be right.

[Etherman]

the gain it from another language doesnt quite answer it in my eyes as it becomes circular. How did they originally gain the sound? and so on. How would any language alone gain any sounds that it originally didnt have?

It may have to do with the intensity and duration of contact. Borrowed words tend to adapt themselves to the phonology of the borrowing language, especially when contact between languages is not strong. However, as the hearer has more and more contact with someone who pronounces the word correctly they'll start pronouncing it more correctly.

Also you should factor in substratum effects. An invading language may replace an indigenous one. The indigenous speakers may have sounds not present (or at least not phonemic) in the invading language which are retained in native vocabulary. They may have a phoneme which is merely an allophone of a phoneme in the invading language, so they'll retain the distinction.

Just the other day a pdf was posted (somewhere) on how retroflex consonants develop. There are actually several ways. Indo-Aryan famously had them develop from alveolar consonants in contact with a rhotic. The rhotic draws the tip of the tongue back past the alveolar ridge causing the stop to be retroflex. Apparently the sequence /ut/ > /ut'/ because the back vowel tends to retract the tongue just enough to turn the alveolar stop into a retroflex stop (and similarly with /ud/). I'm not exactly sure what the process is but /tw/ > /t'w/ can also happen, sometimes with the loss of the labial element.

[Etherman]

One thing I really wonder is, how does a language build up new stop series?

Some Finnic languages developed voiced stops from voiceless stops near sonorants. These would have been simply allophones at first but after the languages borrowed words with these sounds in other environments they became distinctive.

In particular, other than that stop~top example, is there another way to build, say, /t_h,t,d/ contrast from /t,d/? How zulu end up with /ph,p,b,b_</ from an original /p,b/? How did Hindi end up with /b_h/, and Sindhi with /d_h/ and /d_</? It's usually not too hard to figure out a path for other sounds (basically successive palatalization for multiple places of articulation and some fricative, plus lenition and fortition for some other sounds) but forming stop series seems to be definitely harder (especially in all those families that have only CV syllable structures).

I seem to recall that there was a language (probably Native American) that developed aspirate stops before voiceless vowels. Between voiceless vowels the voiceless stop became a voiceless fricative. These were all allophonic differences. It wouldn't be too hard to imagine a change occuring where voiceless vowels become voiced (perhaps a change in position of stress which induces voicing) and then the aspirate/fricative feature becomes phonemic.

Subsidiary question: How are "inner" vowel series (typically /y 2/, /1 @/ or /M 7/) formed?

Front rounded and back unrounded vowels often arise by rounding harmony in vowels. Labial/palatal features of consonants can spread to contiguous vowels. Retroflex consonants can cause vowels to centralize (as can lack of stress).

[merijn]

Clicks can come from loans. They are not that common, I just did a count on the webpage of a Zulu newspaper and of the first 50 words 5 had clicks. It is not out of the ordinary to assume that 10% of the words of a language are loans.
In addition to loans clicks may come from ideophones. Ideophones are words that are either onomatopeic or rely on sound-symbolism. In Zulu they are commonnly used in combination with a verb meaning "say". Many ideophones have clicks; clicks are very expressive sounds. Quite a few nouns and verbs are derived from ideophones so that may be a source of clicks in the language.
I once saw a proposal that a third way in which clicks would have entered the system is via hlonipha. Hlonipha is a type of language Zulu women use when they are married to avoid words that sound like their in-laws names. I have seen it proposed that some women may have subsituted some consonants with clicks, and this way some words with clicks entered the lexicon. I don't know if this is anything more than speculation, or that it is a theory that actually has some support

[Trebor]

Front rounded and back unrounded vowels often arise by rounding harmony in vowels. Labial/palatal features of consonants can spread to contiguous vowels. Retroflex consonants can cause vowels to centralize (as can lack of stress).

I wonder how Thai and other Southeast Asian languages developed /M 7/, when the area is mostly isolating in typology...

[CatDoom]

Lake Miwok provides a rather extreme example of a language gaining stop series. Most Costanoan and Miwok languages have only one series of stops and affricates; the series reconstructed for proto-Western Miwok is fairly typical: /p t ʈ tʃ k ʔ/.

Lake Miwok, on the other hand, has a much more complex set of stops:

/p t ts ʈ k ʔ/
/pʰ tʰ ʈʰ kʰ/
/p' t' ts' ʈ' k'/
/b d/

For the most part, these additions are the result of massive borrowing from neighboring languages, including Patwin, Wappo, and multiple varieties of Pomo. Lake Miwok is, in fact, completely surrounded by these languages, probably having been isolated from the rest of the Miwok languages by the southward expansion of Patwin speakers. However, these additional stops also appear in words that clearly have Utian etymologies.

It appears that a large part of this is due to sound symbolism, whereby certain phonological features became associated with particular meanings and were subsequently incorporated into etymologically unrelated words with similar meanings. For instance, inherited plain stops became ejectives preceding the stressed vowel in certain verbs of position (e.g. /joˈk'onːe/, "to hang," /waˈʈ'anːe/, "to squat"), and a similar process affected stops in words related to quick, small, or accidental movement (e.g. /ˈts'itak/, "to wink," /ts'aˈkaːtataʂi/, "to drip," /ˈt'ipt'ipiʂi/, "to flip (of a fish)").

A number of regular sound changes contributed to spreading the borrowed phonemes as well; initial /p/ inherited from proto-Western Miwok became ejective /p'/ before stressed short /o/ and /u/ when they were followed by another consonant, and this change subsequently spread to derived forms. Likewise, the geminate consonant /ʈː/ became /dː/ after high vowels, while short /ʈ/ became /d/ following any unstressed vowel. Again, this spread to derived forms, and reduplicated forms shifted to agree with the new voiced stops. It appears that some instances of /n/ also shifted to /d/ under similar conditions, and there is at least one example of /m/ > /b/ in those conditions as well.

[Terra]

Aside from just borrowing words containing a given sound, speakers of languages in close contact can also have their language undergo sound changes which introduce a sound found in neighboring languages. I have no idea if this is what happened with Southern Bantu languages and the click languages of southern Africa, but it's happened in other places. That's the reason linguistic areas often contain distinctive sounds shared by many languages of various families (cf. the uvular rhotic in many parts of Europe, including France, Germany, Denmark, Belgium, etc.). Languages can also share or borrow sound changes from one another--the sound change /j/ > 0 after a consonant diffused from Cree into Ojibwe-Potawatomi, for instance, and the loss of nasal consonants diffused among the Wakashan languages Makah and Nitinaht (the latter is now pronounced Ditidaht, but it wasn't at the time of first contact).

In north-east Asia/Siberia, there was a sound change 'p->f->h' that affected Japanese, Mongolian, Evenki, and even Ket (a language that is most certainly not Altaic!). I assume that it happened by diffusion, because it happened in each language at about the same time, but it colud've happened independently in each language; 'p->f->h' is a common sound change, after all.

[zompist]

CatDoom, that's really interesting, because Quechua has a very similar situation. Most Quechuan languages have a simple series of stops /p t k q/ plus the affricate /ch/, but one, Cuzqueño, has three series-- plain, ejective, and aspirated. Notably, its neighbor Aymara has these three series too, and there is a rich set of borrowed words.

Some simply reconstruct all three series for the proto-language, but this is not very satisfying-- it just seems more likely that the feature was borrowed from Aymara. But if so, it spread somehow to other native words. So it's fascinating that there seems to be another instance of this with Lake Miwok.

[CatDoom]

CatDoom, that's really interesting, because Quechua has a very similar situation. Most Quechuan languages have a simple series of stops /p t k q/ plus the affricate /ch/, but one, Cuzqueño, has three series-- plain, ejective, and aspirated. Notably, its neighbor Aymara has these three series too, and there is a rich set of borrowed words.

Some simply reconstruct all three series for the proto-language, but this is not very satisfying-- it just seems more likely that the feature was borrowed from Aymara. But if so, it spread somehow to other native words. So it's fascinating that there seems to be another instance of this with Lake Miwok.

Glad you find it interesting! Something similar may have happened in the early history of the Yokuts languages as well, since there's some strong evidence that they're related to the Utian languages (although the relationship would be much more distant than that between the Miwok and Costanoan languages). However, in her Proto-Utian Grammar and Dictionary (the same source I used for the above information), Catherine Callaghan points out that the development of aspirated and ejective stops in Yokuts could be secondary developments internal to the Yokuts family, rather than the result of borrowing.

Utian obstruents correspond to Yokuts aspirated and glottalized stops more often than they do to Yokuts plain stops, and word-final stops in Yokuts are aspirated by default, suggesting that the plain stops may be a more marked category arising from sound changes or borrowing in (pre-)proto Yokuts. Perhaps more tellingly, proto-Yokuts ejective stops and glottalized sonorants never seem to have occurred word-initially or in syllable onsets following a coda consonant, and it's common for stem-final consonants in Yokuts to become glottalized before certain suffixes. It has therefore been suggested that glottalized consonants in the Yokuts languages are mostly the result of fusion in clusters of a buccal stop with a glottal stop, which makes sense since a number of consonants in the Utian languages (particularly glides) seem to regularly correspond with glottal stops in Yokuts.

[Porphyrogenitos]

If this is the right thread for this - am I correct in thinking that early French would have entirely lost /k/ to palatalization if not for the introduction of Frankish loanwords? (And for the shift from /kʷ/ to /k/, but I don't know when that took place.)

[Terra]

If this is the right thread for this - am I correct in thinking that early French would have entirely lost /k/ to palatalization if not for the introduction of Frankish loanwords? (And for the shift from /kʷ/ to /k/, but I don't know when that took place.)

[k] before front vowels, yes.

[Hallow XIII]

If this is the right thread for this - am I correct in thinking that early French would have entirely lost /k/ to palatalization if not for the introduction of Frankish loanwords? (And for the shift from /kʷ/ to /k/, but I don't know when that took place.)

[k] before front vowels, yes.

And /a/, for the record.

[Terra]

If this is the right thread for this - am I correct in thinking that early French would have entirely lost /k/ to palatalization if not for the introduction of Frankish loanwords? (And for the shift from /kʷ/ to /k/, but I don't know when that took place.)

[k] before front vowels, yes.

And /a/, for the record.

For Standard French, yes. There are/were dialects (Norman and Walloon, iirc) of French where /k/ was never palatalized before /a/.

[jal]

For Standard French, yes. There are/were dialects (Norman and Walloon, iirc) of French where /k/ was never palatalized before /a/.

Afaik, it hasn't palatalized in Occitan, but I don't think that's generally considered a dialect of French (but a separate language).


JAL

[Nortaneous]

Front rounded and back unrounded vowels often arise by rounding harmony in vowels. Labial/palatal features of consonants can spread to contiguous vowels. Retroflex consonants can cause vowels to centralize (as can lack of stress).

I wonder how Thai and other Southeast Asian languages developed /M 7/, when the area is mostly isolating in typology...

Khmer developed some of its vowels/diphthongs through breathy voice loss: breathy-voiced vowels tend to be higher than modal vowels (Kri has /a e e_t i_t but no a_t or i) (_t = breathy voice in X-SAMPA), and some SEA languages turned voiced stops into voiceless stops with breathy voice on the following vowel.

I don't know how Thai developed its vowels, but one theory for Tangut and Chinese is that they came from loss of unstressed vowels in presyllables, like umlaut in English but the other way around. I forget the details, but something like s1"ta > st1a > t1a

[jal]

breathy-voiced vowels tend to be higher than modal vowels

Really? I've always understood they are lower in pitch, because of the vocal folds being less constricted. Or do you refer to the POA?


JAL

[Richard W]

If this is the right thread for this - am I correct in thinking that early French would have entirely lost /k/ to palatalization if not for the introduction of Frankish loanwords? (And for the shift from /kʷ/ to /k/, but I don't know when that took place.)

No. There are numerous verbs preserving the Latin prefix com- in its various forms, eg, connaître 'to know'. coup from Greek kolaphos is another example.

[Richard W]

I don't know how Thai developed its vowels, but one theory for Tangut and Chinese is that they came from loss of unstressed vowels in presyllables, like umlaut in English but the other way around. I forget the details, but something like s1"ta > st1a > t1a

And a popular idea is that Tai-Kadai is a branch of Austronesian close to Malayo-Polynesian, so Thai mɯː 'hand' would come from Proto-Austronesian *lima 'hand', 'five'. There are traces of /l/ in the 'Kra' languages within Tai-Kadai in the apparently cognate word for 'hand'.

[Nortaneous]

breathy-voiced vowels tend to be higher than modal vowels

Really? I've always understood they are lower in pitch, because of the vocal folds being less constricted. Or do you refer to the POA?


JAL

POA, hence the Kri example

[Terra]

Btw, another areal change I came across the other day: Ket has few words that begin with [m] or [n]. The ones that do are either cradle words or borrowings. Some Turkic languages have the same aversion to initial nasals. Ket was historically in contact with Turkic languages to its south and north-east. Again, I don't know for sure, but it looks like a good candidate for an areal change.

[----]

For the record, initial nasals were present at one time but merged with the corresponding voiced plosives (m > b and n > d)