Rate of sound change

[Clearsand]

I don't know if this has been asked before, but I thought I'd give it a shot: Does anyone have any idea how often, on average, a language in a given, say, city undergoes an individual sound change, whether through outside changes or individual innovation? I realize these things tend to be fluid, especially with vowels, but is there a normal rate of sound changes that are recorded seperately? If so, are there differences between rural and urban areas or other factors that would alter the sound change rate?

[KathTheDragon]

Short answer: There is no normal rate.

[jal]

is there a normal rate of sound changes

I don't think there is, other than a very broad average. Sound changes may be triggered by multiple sources, both internal (e.g. chain shifts) and external (influence by other languages). Some triggers may cause more rapid changes than others.


JAL

[Soap]

There's definitely a wide variation. "Stable" phonologies seem to include groups like proto-Uralic and proto-Austronesian, which have some languages that have changed so little over the past 5000 years that sentences in the proto-language can still be understood today. However these same languages also have some descendants that have changed radically. THen you have families like PIE where all of the descendants have changed radically. Another example I remember is in this thread:
viewtopic.php?f=7&t=41813
peponwi --> aa in about 2000 years, but a sister language to the above kept the proto-form much better preserved as peponwe.

[Tropylium]

Short answer: There is no normal rate.

Slightly longer answer: there is still a normal range.

"Zero sound changes over 20000 years" scores you 300 milliNylands on the scale of linguistic crackpottery.
"Zero sound changes over 2000 years" would be exceptional, though I suppose it's happened occasionally in history under particularly suitable conditions. (Sanskrit has, as a liturgical language, allegedly remained unchanged for longer than that.)
"Zero sound changes over 200 years" would be a bit unexpected but not particularly worthy of concern.
…
"50 major sound changes over 100 years" sounds perhaps a bit high but again not terribly worthy of concern (except, perhaps, for the great-grandmothers who fail to understand what the kids are saying to them).
"50 major sound changes over 10 years" should make the "or wait, is this a creole" warning lights go off.
"50 major sound changes over 1 year" requires the default hypothesis "cryptolect artificially created by a socially prestigious cabal of conlangers".

[jal]

"50 major sound changes over 1 year" requires the default hypothesis "cryptolect artificially created by a socially prestigious cabal of conlangers".

ɐ dɔ̃ nø wɔ yo ɐ tokĩ bɔ!


JAL

[jmcd]

"Zero sound changes over 2000 years" should surely be a cabal of prescriptivists.

[zompist]

[(Sanskrit has, as a liturgical language, allegedly remained unchanged for longer than that.)".

I have a hard time buying that. For the same religious reasons, people are going to be loth to admit that they speak it differently.

Plus, I know the Sanskrit grammarians were exceptionally good phoneticians for ancient times, far better than the Greeks... but surely not better than modern phonology, especially now that we have access to actual recordings and frequency analysis?

[Nortaneous]

THen you have families like PIE where all of the descendants have changed radically.

Probably a good way to do this is to create a protolang and then kick it out of its sprachbund and have major influence from lots of different substrates, which is what happened with PIE.

(Has anyone researched the pre-IE substrates in Western Europe? Fortis resonants everywhere. Fortis resonants and weird shit going on with the labial nasal, except in Old English.)

[Salmoneus]

Population sizes are probably important in this regard. Uralic and Polynesian have both historically been spoken by small, non-dense populations. [and indeed, at a glance it looks like the places with the most rapid change in Austronesian are the most populated areas]

[jmcd]

I think the level of interaction with other societies is important too. The more interaction there is with other societies (including immigration), the more change there is in the society and in the language. It reminds of a document I read which compared the development of Old Norse and Classical Arabic into their respective modern forms.

[R.Rusanov]

you're gonna find substrates anywhere you see rapid linguistic expansion into a new region

unless there are no inhabitants there, but then it's probably a different kind of environment and the new inhabitants will change at least the vocabulary as they adapt, or if there are inhabitants but they're so radically different that mixing and substrate influence won't happen, like bantu-pygmy interactions.

[Shrdlu]

When person X forgets/doesn't notice to comment that person Y pronounces word A slightly different, and then begins to do it her self.

[linguoboy]

unless there are no inhabitants there, but then it's probably a different kind of environment and the new inhabitants will change at least the vocabulary as they adapt, or if there are inhabitants but they're so radically different that mixing and substrate influence won't happen, like bantu-pygmy interactions.

Huh? Most Pygmy groups speak Bantu languages, with considerable substratum effects. (About 30% of the lexicon of Aka, for instance, is of non-Bantu origin.) Neighbouring languages have at the very least borrowed vocabulary relating to Pygmy trades (such as honey-gathering). In this respect, it's something like the relationship between indigenous and settler languages in the New World.

[R.Rusanov]

it took a few hundred years of killing or "hunting" as the bantu thought of it before the situation had settled down enough for this kind of exchange to take place.

[CatDoom]

you're gonna find substrates anywhere you see rapid linguistic expansion into a new region

unless there are no inhabitants there, but then it's probably a different kind of environment and the new inhabitants will change at least the vocabulary as they adapt, or if there are inhabitants but they're so radically different that mixing and substrate influence won't happen, like bantu-pygmy interactions.

Considering that the southern Bantu languages have picked up *clicks* from their hunter-gatherer neighbors and predecessors, I'm not sure there's such a thing as "too radically different" in terms of linguistic mixing.

[R.Rusanov]

this happened very very recently, given we can still observe it and accurately identify it, and also given that these people were engaged in a frontier, fight flight or die mentality for the past millenium.

it's like if wolves starting picking up human speech and vice versa

[Clearsand]

Tropylium's answer was closest to what I was looking for. I just wanted a general average or ballpark range.

[Soap]

Just a few things I thought where interesting:
proto-Germanic was apparerntly pretty conservative of the basic structure of words inherited from Proto-Indo-European, despite at least 2500 years having elapsed since PIE (3000 BC is the latest commonly proposed date for PIE, some people go even further back), and many major sound changes. Basically every word had the same number of syllables as it had had in PIE, apart from laryngeals if those were even syllabic in the first place. e.g. hamiþiją "shirt" (likely the source of Latin camisia), freusaną "to freeze" (with a 2-syllable infinitve suffix), samədaz "sand", fedwōr "four". Then the accent changed to the first syllable and everything fell apart. All of those words are monosyllabic today in modern English and in German if you ignore the retention of the infinitive -en (German has Hemd for shirt), even though less time has gone by between proto-Germanic and today than between PIE and proto-Germanic. However, word-initial stress doesn't always predict sound change chaos, right nearby we have the counter example of Finnish, which has preserved Germanic loanwords such as kuningas "king" in an almost unchanged state for 2000 years, along with other Finnic languages.

[Ketumak]

I'd often wondered about this point myself. Thanks, Tropylium, it's good to have a ball-park figure to work from. From what other people are saying, I guess the exact rate of change depends on your people's history. How much change have they been through and what kind?

Their history would influence the nature of change, too. A trading and empire-building nation like Britain would bring back lots of words from around the world. The sounds of colonial and metropolitan speech patterns would diverge due to their isolation from each other. A remote mountain tongue like Romansch might have fewer sources for loan words but gain a still end up with a number of dialects, as the terrain created communications problems.

[Curlyjimsam]

To measure this you'd need a solid definition of what constitutes "one sound change". If you have five voiceless sounds /p t k f s/ and they all become voiced between vowels, is that one sound change or five? If /e/ gradually raises to /i/ in a very large number of very small steps, is that one change or lots?

[Clearsand]

To measure this you'd need a solid definition of what constitutes "one sound change". If you have five voiceless sounds /p t k f s/ and they all become voiced between vowels, is that one sound change or five? If /e/ gradually raises to /i/ in a very large number of very small steps, is that one change or lots?

Yeah... That's the problem. you can really ony approximate the number of sound chnges even if you know exactly what they are. For the first example, I would probably count thas one sound change ([p,t,k,f,s] [+vcd]/V_V) because all the changes are based on the same principle. However, if the changes were spread out over several hundred years with other changes in between, I would probably count each seperately. With vowels it's always trickier because they tend to gradually drift more than consanants. One sound change would probably be counted as when a vowel drifts far enough to use a different ipa (or X-sampa) letter without diacritics. So that means a little more than halfway between two ipa symbols.

[Tropylium]

Population sizes are probably important in this regard. Uralic and Polynesian have both historically been spoken by small, non-dense populations. [and indeed, at a glance it looks like the places with the most rapid change in Austronesian are the most populated areas]

FWIW most Uralic languages are not very phonologically conservative at all, including some of the small isolated groups. It's pretty much just a couple of the Finnic languages and even that is mainly with respect to the vowels.

To measure this you'd need a solid definition of what constitutes "one sound change". If you have five voiceless sounds /p t k f s/ and they all become voiced between vowels, is that one sound change or five? If /e/ gradually raises to /i/ in a very large number of very small steps, is that one change or lots?

Good point.

You can get some discreteness to this by looking at the phonology instead of the phonetics, at least. If /e/ becomes slightly closer but remains distinct from /i/, then "nothing has changed" yet. (At least, unless /i/ itself is also becoming something else, say [ɪj].) Or, suppose that [c] goes to [ts] — at some point speakers are going to start thinking of this as a sibilant and as non-palatal. It might be difficult to tell when exactly if there are no mergers or anything though, and it's possible for a segment to have a disconnect between its phonological representation and its phonetic value for a while. (Like how [ʁ] still gets perceived as "rhotic" across large parts of Europe, although most languages that have uvulars see nothing "r-like" about them.)

A "re-phonologization" would need to have happened at a point before some other change could intervene though, e.g. to create a new [e] from /ai/, or make a rule that word-final non-alveolar consonants disappear.

Some group shifts might operate in parts (there's evidence that e.g. the Greek spirantization of φ θ χ β δ γ did), others could occur all at once, I don't think there's any hard and fast rule for these?

[CatDoom]

Population sizes are probably important in this regard. Uralic and Polynesian have both historically been spoken by small, non-dense populations. [and indeed, at a glance it looks like the places with the most rapid change in Austronesian are the most populated areas]

This is interesting, because I've actually read a paper that argued (based on some manner of statistical modeling... I don't remember the details, unfortunately) that language change is likely to be more rapid in very small populations. As I recall, the effect leveled out fairly quickly when the group in question got much larger than an isolated village, so it's possible some other factor kicks in at higher population sizes.

That said, if very small, relatively isolated communities are prone to rapid language change, it might help explain the linguistic diversity of regions like pre-colonial California, the Caucasus, and New Guinea, where the topography and/or population density makes those conditions more likely.

[jal]

language change is likely to be more rapid in very small populations

I think what comes in play there is that within a generation or two, there's no memory of how words used to be pronounced. In larger communicities, people must communicate with others that are removed more than a few miles away, and the "original" pronunciation is reinforced. That said, in large communities, sound changes can also progress rapidly, think e.g. of the Great Vowel Shift. A factor there can also be social differences, where e.g. royal speech is imitated by the population, or when the upper class speaks a different language altogether (e.g. Norman French).


JAL