Zompist Bboard Again

conlangernoob wrote: ↑Tue Mar 14, 2023 2:53 pm In general, how many sound changes does a language go through every, say, 1,000 years? Also, how much variation is there?

It varies hugely. I seem to recall reading about one Australian language where the speaker community was so small it could change radically in a single generation. On the other hand you have languages like Icelandic (as Travis mentioned), which have changed to some extent but have retained many features. Practically the only constant is that there will be some kind of change.

Also, do some types of sound change occur more than others?

Yes! For a simple example, lenition (‘softening’ of phonemes) is much more common than fortition (‘hardening’). For more details, I will refer you to zompist’s Language Construction Kit.

Finally, how many years does it take between a sound change to begin and a sound change to finish?

This is an interesting topic. For one thing, sound changes themselves often can’t be gradual: to take an extreme case, how does one gradually turn a [t] into a [k], as happened in Samoan? Considered like that, such a sound change can’t really have a ‘beginning’ or an ‘end’ — just a time when it took place.

However, sound changes often do take place gradually. Usually they only affect a few words, then spread out to more and more of the lexicon until they affect many or all words. There is also spread in population: the sound change might start in a particular demographic (often young women, IIRC) and then spread out until it’s present in the speech of a whole generation. I don’t recall any numbers, but again, there are lots of details in one of zompist’s books. (Advanced Language Construction, I think, but I’m not sure.)

conlangernoob wrote: ↑Tue Mar 14, 2023 2:53 pm Four dumb questions: In general, how many sound changes does a language go through every, say, 1,000 years? Also, how much variation is there?

As others have said, it's very variable. However over time it's generally the case that languages generally take 1,000 years to change to a point where they're mutually incomprehensible with the original. How many sound changes this involves depends greatly on the phonology of the language, the speaker base, and how you define one sound change.

Bearing that in mind, I'll give you Spanish as an example. Ralph Penny's A History of the Spanish Language lists 33 changes over about 2,000 years for Latin to Spanish. However, some of these involve series rather than single phonemes; for example, one of the changes is:

20. Lenition [...]
(a) geminate > simple (and sonorant are additionally modified: LL > /ʎ/, NN > /ɲ/, RR > /r̄/);
(b) voiceless > voiced (e.g. -T- > /d/, -S- > /z/, /ts/ > /dz/);
(c) voiced plosives > fricatives (and are often eliminated)
(d) /j/ > /zero/.

If you wanted to break this up into individual steps, you could think of it as at least half a dozen changes. Penny's list of 33 could easily be expanded to 50 or 100 depending on what you think of as a single change.

Contrast this to Hawaiian for example, where you might only be able to list 10 sound changes for the same depth of time, yet with a similar degree of mutual unintelligibility with its ancestor. Basically, over 1000 years you probably want enough developments to significantly alter the language, but how many changes that actually involves is up to you. Although there are some exceptions like Icelandic where remarkably little has happened. You're pretty safe if you go anywhere between "not very many" and "quite a lot" of sound changes per millenium.

Also, do some types of sound change occur more than others?

Yes. There are some very common sound changes, like intervocalic lenition, palatalisation, cluster simplification, anticlockwise movement of back vowels etc. Other changes like large chain shifts or fortition are pretty rare, and wacky changes like /r/ → /ɡ͡ʟ/ or /j/ → /p/ are extremely rare but still attested. Sound change likelihood depends on the wider phonology of the language, e.g. if a language has four stops MOAs, it's more likely for one of them to do something than if there's only one. And then there's Sprachbundts where language contact makes sound changes areally more common even when they might be pretty rare everywhere else.

Finally, how many years does it take between a sound change to begin and a sound change to finish?

This is a very difficult question to answer - again it depends a lot on prestige levels and tons of other socioeconomic factors. Sound changes can take a while to spread throughout a community; generally starting with women, whose children learn it from them. It can also take effect at different times in different words (although hardcore neogrammarians would hate to admit this). For a single phoneme change (like /ʎ/ > /j/ or something) it might take only a generation, for more complex changes or changes involving intermediate stages it could be several generations before it fully catches on. I'll go back to Spanish for another example - the Old Spanish sibilant shift (basically /dz z ʒ/ > /ts s ʃ/, /ts/ > /tθ/ > /θ/ and /ʃ/ > /x/) took place within a period of 30 years for some speakers, but it wasn't carried out uniformly even in the standard language for a couple centuries and never completely applied in some dialects. Sound changes are always ongoing and there's about three or four generations of people using a language at any given time, so it's basically impossible to qualify when exactly a sound change has "finished".

Thanks! This is really helpful.

Is it a general trend the simpler a languages phonology is the less changes it goes through as it evolves. I know Japonic has a similarly simple phonology yet it's languages seem to have gone through much more changes.

foxcatdog wrote: ↑Thu Mar 16, 2023 12:14 am Is it a general trend the simpler a languages phonology is the less changes it goes through as it evolves. I know Japonic has a similarly simple phonology yet it's languages seem to have gone through much more changes.

One thing I’d mention is that if a language has a simple phonology, its phonemes often have quite a bit of allophonic variation, so sound changes are less likely to cause changes in the phoneme distribution — consider e.g. how Samoan famously underwent /t/→/k/ yet that made no difference at the phonemic level.

I have [ɲ c cʰ cʼ] as allophones of /ŋ k kʰ kʼ/ before front vowels /e i/. I also have the phonemes /q qʰ qʼ/, which causes a following vowel to allophonically retract (in the case of /e i/, to [ɛ ɪ]). Is it plausible, therefore, to have [cɛ cɪ] as an allophone of /qe qi/, for example (converting the velar/uvular contrast in the consonant to a non-retracted/retracted contrast in the vowel)?

Apologies if I asked something like this before.

StrangerCoug wrote: ↑Thu Jun 22, 2023 9:56 pmI have [ɲ c cʰ cʼ] as allophones of /ŋ k kʰ kʼ/ before front vowels /e i/. I also have the phonemes /q qʰ qʼ/, which causes a following vowel to allophonically retract (in the case of /e i/, to [ɛ ɪ]). Is it plausible, therefore, to have [cɛ cɪ] as an allophone of /qe qi/, for example (converting the velar/uvular contrast in the consonant to a non-retracted/retracted contrast in the vowel)?

That doesn’t seem unreasonable to me, though I might tweak the particulars a bit:

- Retraction
- Fronting of uvulars to velars before front vowels
- Palatalization

So I have a phonology that starts off with a syllable structure of CCVC and only five vowel qualities: /a e i o u/. Germanic-style umlaut ends up adding /ɛ œ ø y ɔ/, the umlaut also creates a new consonant phoneme /ɥ/ from a conditional merger of /j w/. All ten vowels have a length distinction, though vowel laxing causes a conditional merger of short /e ø o/ with short /ɛ œ ɔ/ (short monophthongs that end a syllable don't lax if the syllable is stressed or final). A later change triggers palatalization adjacent to front unrounded vowels and /j/, labio-palatalization adjacent to front rounded vowels and /ɥ/, labio-velarization adjacent to back rounded vowels and /w/, and velarization elsewhere; syllable-medial /j ɥ w/ then elides. (The first consonant in remaining syllable-initial consonant clusters gets the same secondary articulation as the second consonant in the cluster.) The exception to the rule is that consonants that are already labial don't labialize; what would be the labio-palatalization of a labial consonant becomes just palatalization, and likewise the labio-velarizartion of a labial consonant becomes just velarization (e.g. what would be /pᶣ pʷ/ instead becomes /pʲ pˠ/, and a syllable could theoretically start with, say, /pʲrᶣ/).

I have an /l/ that I'd somehow like to split into /lˠ lʲ ʎ/. How do I get that? (I don't want stable labialized lateral consonants; feel free to propose what happens to /lʷ lᶣ ʎʷ/.)

Also (I may or may not decide to have this in the same language family as above), how common are languages like Ancient Greek and Mandarin Chinese where a coda consonant can be a sonorant, but not a stop?

The most recent typological work on this I know of is a paper by Krämer & Zec. They mostly focus on nasals but they seem to have a sizeable database for more general information in the supplementary materials.Here is a link: https://www.cambridge.org/core/journals ... rticle-tab

StrangerCoug wrote: ↑Tue Sep 19, 2023 12:25 pm Also (I may or may not decide to have this in the same language family as above), how common are languages like Ancient Greek and Mandarin Chinese where a coda consonant can be a sonorant, but not a stop?

Umm, doesn't Ancient Greek permit non-sonorant codas, e.g. in the word polis? (I assume you meant to say non-plosive here.)

Travis B. wrote: ↑Tue Sep 19, 2023 1:52 pm

StrangerCoug wrote: ↑Tue Sep 19, 2023 12:25 pm Also (I may or may not decide to have this in the same language family as above), how common are languages like Ancient Greek and Mandarin Chinese where a coda consonant can be a sonorant, but not a stop?

Umm, doesn't Ancient Greek permit non-sonorant codas, e.g. in the word polis? (I assume you meant to say non-plosive here.)

Good catch—non-plosive was indeed what I meant in that language’s case.

Creyeditor wrote: ↑Tue Sep 19, 2023 1:45 pm The most recent typological work on this I know of is a paper by Krämer & Zec. They mostly focus on nasals but they seem to have a sizeable database for more general information in the supplementary materials.Here is a link: https://www.cambridge.org/core/journals ... rticle-tab

Incidentally, this is a very interesting paper in its own right; thanks for linking it!

Nasalization before uvulars, which then front to velars, creating a nasality contrast before dorsals. Am I good or nah?

Man in Space wrote: ↑Wed Nov 08, 2023 10:50 am Nasalization before uvulars, which then front to velars, creating a nasality contrast before dorsals. Am I good or nah?

Why do uvulars trigger nasalization, and velars not?

WeepingElf wrote: ↑Wed Nov 08, 2023 2:41 pm

Man in Space wrote: ↑Wed Nov 08, 2023 10:50 am Nasalization before uvulars, which then front to velars, creating a nasality contrast before dorsals. Am I good or nah?

Why do uvulars trigger nasalization, and velars not?

Maybe by having at one time had glottalization - and uvulars are known to turn into glottals (look at Egyptian Arabic) - which then results in nasalization via rhinoglottophilia? Of course, then it'd be hard to explain the subsequent merger with velars, unless glottalization was not complete.

Is any of the following sound changes at all plausible?

ɾ > ∅ / V_V
ɾ > h / V_V
ɾ > ʔ / V_V

I.e. can I get away with deleting intervocal /r/ provided that there weren't many minimal pairs to begin with?

Zju wrote: ↑Sat Feb 03, 2024 3:36 pm Is any of the following sound changes at all plausible?

ɾ > ∅ / V_V
ɾ > h / V_V
ɾ > ʔ / V_V

I.e. can I get away with deleting intervocal /r/ provided that there weren't many minimal pairs to begin with?

The first one sounds fine, considering that /r/ in many languages is realized as a tap/flap, and NAE has flap elision.

The others also sound fine, if you have /r/ become [ʁ] or [ʀ] as an intermediate step, which is very plausible. (The second actually happened in Brazilian Portuguese.)

Zju wrote: ↑Sat Feb 03, 2024 3:36 pm I.e. can I get away with deleting intervocal /r/ provided that there weren't many minimal pairs to begin with?

Absolutely. Numerous Lakes Plain languages did it, and they only had five other consonants in intervocalic position. Hell, you can delete anything intervocalically really.

Cross-posting from the Linguistic Miscellany Thread:

Per A grammar and dictionary of Gayogo̱hó:nǫˀ (Cayuga):

Dyck, Froman, Keye & Keye (2024) wrote:SR – as in węhnihSRí:yo: ‘nice day’ – sounds like the SHR [ʃɹ] in shrink. Some speakers pronounce SR as FR [fɹ] instead, for example in words like ganǫ́hkwasraˀ (ganǫ́hkwaFRaˀ) ‘love’. SR syllabifies as two separate consonants, [ʃ.ɹ] or [f.ɹ].

A fronting of [ʃ] to [f]. I never would’ve figured on that as a direct step. ʃ to x and thence to f, sure, but this may be useful to some.