Linguistic Miscellany Thread

[Tropylium]

I'm not sure why Wiktionary stopped at 37.

They might follow the principle of seeing what is accepted in both of the two published Proto-Afrasian reconstructions; like bradrn's linked paper discusses, there's a different one from Orel & Stolbova, etymology-wise it agrees about almost nothing with Ehret. Both of them also still often pick relatively random words from the dozens of Chadic, Cushitic and Omotic languages without an attempt to show they would be native-and-reconstructible even to some smaller subgroup. Occasionally the same even with Semitic, though it's most often just Arabic that gets preferential treatment.

Wikipedia has some discussion of the numerals for one demonstration of the problems. Sometimes they could be just subgrouping issues (e.g. Egypto-Semitic-Berber #čan versus Omotic-Cushitic #lam for '2'), but if e.g. for 'one' we have from Omotic just Ometo ista versus Kafa tok proposed, we really would like to know if one of these can be actually reconstructed even for Proto-North Omotic. If one or both cannot, then it's more likely a chance resemblance — or for the latter, a loan: note that Kafa is spoken in direct contact with Oromo which has the almost identical tokko (does not strictly reconstruct for Proto-Cushitic, but does maybe have the best odds for this out of a handful of competitors).

[Moose-tache]

The Wiktionary entry is pretty sparse, so I went to the source: Christopher Ehret's book from 1995, which as far as I can tell is still the most confident reconstruction of PAA vocabulary out there. I'm not sure why Wiktionary stopped at 37. Ehret has over a hundred entries just for the labial consonants!

But it’s very likely that they’re almost all chance resemblances.

Nice link.

I think it's complicated by the fact that we're dealing with multple layers. There are (hypothetically) PAA words, then early borrowings between branches, then recent borrowings between branches. For example, dam is "blood" in Semitic and Berber, and unattested elsewhere. Is that a cognate, a random coincidence, or a borrowing?

As I mentioned, any very small roots (i.e. less than three phonemes) can be rejected out of hand because, as the paper shows, coincidences are real. Ehret's most eggregious example of this is *b, realized as bi- in Semitic, and words relating to place in two other branches, but as bw and ba, so we really are measuring a single phoneme. No doubt you could play this game and make countless "cognates."

[Man in Space]

The talk about Akana remound me of the fact that I once got an actual academic (well, for credit for a college class) paper out of Teyetáti. Language & Culture with Prof. O'Neill. I still have it, impossibly. It was about the speakers of T1 languages from a Whorfian perspective.

That class was notable for four reasons:

1. It was where I met Claire, who in a very literal sense saved my life a few years later. She said she thought I was cool because I spontaneously busted out a dead-on California surfer accent in class once to illustrate a point.
2. The aforesaid paper. It was about the speakers of T1 languages from a Whorfian perspective. It was one half of the final for that class and I got full marks. I still have it, impossibly.
3. I wrote the other half of the final in poetic meter à la The Rime of the Ancient Mariner on a dare. Specifically, Prof. O'Neill dared me to. He mentioned someone doing something similar in a previous class and he thought I was the kind of guy who could actually pull it off. (Again, I got full marks, and he requested, and was granted, permission to retain a copy.)
4. I rolled the bones on the very first exam and answered one of the questions from my religious perspective, and he not only accepted it, he wrote "Strong answer" in the margin. As I went to a public university, that rocked my world.

He was cool. He had a patent on some sort of circuit and I think he designed some sort of musical instrument (it had a resonating chamber, I recall as much because the geometry of the holes was of importance).

Also I guess he's the ultimate reason I was Facebook friends with Dan Everett, back when he had one.

[Estav]

It's not just acc-. Here are some words in French with /ks/ or /gʒ/. (Sorry, the wordlist I grabbed had no accents.)

acceder
accelerer
accent
accepter
acces
accident
occident
succeder
succes
succinct
vaccin

suggerer

These are likely all reborrowings, but that doesn't really explain anything... why is this rule followed for Latin words?

The precedent for not assimilating the first <c> would be Old French acciun, 'action', if it is truly inherited. A more typical pattern would be for the first <c> to vocalise to <i>, but that seem undependable - compare façon, inherited from factionem - and too deviant to have fed back into Latin.

Synchronically the rule makes sense; <c g> are /s ʒ/ before <i e y> and /k g/ anywhere else. The first <c> or <g> is before an anything else, so it's /k/ or /g/.

There's only one good inherited example I can think of, which is VL *auccidere → Old French ocire, which only later got its first <c> back. So it seems pretty likely that all <cc> /ks/ are learnèd or semi-learnèd. The problem is like 90% of French words have gotten learnèd Latin interference at some time in the past millenium so it's all very confusing.

No words spelled with -cc- and pronounced with /ks/ in French are regular inheritances. French lost syllable-final obstruents by sound change at several points in its history (both before/during and after Old French). However, these sound changes were followed by the development of new syllable-final consonants as a result of vowel syncope, after which speakers found it easy to reintroduce previously lost syllable-final consonants based on old spellings. As Darren said, the pronunciation /ks/ for -cc- is a spelling pronunciation that follows the rule that c = /k/ when not before <i e y>. This matches the pronunciation of syllable-final c in monosyllabic words such as sac, bec (/k/ in this context was originally lost when followed by another consonant by regular sound change between Old French and modern French, but has been restored by analogy from the pronunciation before a vowel, and maybe also from the spelling). Compare correct /kt/, direction /ksj/, concept /pt/, conception /psj/, none of which are inherited pronunciations. Old French acciun is not inherited. The regular development is as in façon. Latin -cti-, -ci-, -cci- regularly became Old French [ts] (voiceless with no preceding offglide).

[Darren]

1. I wrote the other half of the final in poetic meter à la The Rime of the Ancient Mariner on a dare. Specifically, Prof. O'Neill dared me to. He mentioned someone doing something similar in a previous class and he thought I was the kind of guy who could actually pull it off. (Again, I got full marks, and he requested, and was granted, permission to retain a copy.)

That's amazing

[Darren]

Reading over Hyman's universals paper, he says that he "can see no principled way to rule out" theoretical phoneme inventories like

Code: Select all

     #1               #2
     
 pʰ  tʰ  kʰ       p   t   k
 p   t   k        f   s   x

Because Rotokas has /p t k b d g/ (or /p t k β ɾ g/, same thing). But that got me thinking, is it true that all languages have some voiced consonant phonemes? In other words, are there any languages without nasals, approximants or voiced obstruents?

[Nortaneous]

is it true that all languages have some voiced consonant phonemes? In other words, are there any languages without nasals, approximants or voiced obstruents?

I think so, although my databases don't have perfect coverage. All languages with no nasals have either a /p b/ or a /p β/ contrast, except Crow and Hidatsa, where nasals are allophones of /w r/, and Pawnee, with /p t ts k ʔ s h w r/.

Stop voicing contrasts are too common in Oceania, with the exception of Polynesian, which has plenty of nasals and liquids. It'd be unlikely for such a language to exist there - probably not even in New Guinea.

It's possible to contrive a plausible-sounding path from languages with a minimal voiced consonant inventory to languages with none - the simplest is if Pawnee had r > x and reasons to analyze [w] as a -syllabic allophone of /u/.

[zompist]

Your periodic reminder that Sign languages are languages; also that the languages we have today are not a random sample and not a guide to the linguistic diversity of, say, 10,000 BCE.

If we did want to declare that no languages with voiced consonants exist— would that be like declaring that no human can be 9 feet tall (the record is 8' 11"), or like declaring that no human can be 90 feet tall? When we have close edge cases, it's pretty bold to declare a universal.

[Estav]

Reading over Hyman's universals paper, he says that he "can see no principled way to rule out" theoretical phoneme inventories like

Code: Select all

     #1               #2
     
 pʰ  tʰ  kʰ       p   t   k
 p   t   k        f   s   x

Because Rotokas has /p t k b d g/ (or /p t k β ɾ g/, same thing). But that got me thinking, is it true that all languages have some voiced consonant phonemes? In other words, are there any languages without nasals, approximants or voiced obstruents?

In terms of phonology, it seems questionable to describe an inventory with no voiceless/voiced contrast as containing specifically "voiceless" consonant phonemes: I think many theorists would interpret the phonemes as being unspecified for voicing in that context, not as all being redundantly marked for voicelessness.

In terms of phonetics, some presence of allophonically voiced consonants seems likely, although it's hard to imagine that we could prove that this would necessarily exist in a language with such a consonant inventory.

[Darren]

I think so, although my databases don't have perfect coverage. All languages with no nasals have either a /p b/ or a /p β/ contrast, except Crow and Hidatsa, where nasals are allophones of /w r/, and Pawnee, with /p t ts k ʔ s h w r/.

Stop voicing contrasts are too common in Oceania, with the exception of Polynesian, which has plenty of nasals and liquids. It'd be unlikely for such a language to exist there - probably not even in New Guinea.

Hey, anything can happen in New Guinea.

It's possible to contrive a plausible-sounding path from languages with a minimal voiced consonant inventory to languages with none - the simplest is if Pawnee had r > x and reasons to analyze [w] as a -syllabic allophone of /u/.

I can kind of buy that. It's interesting that /p t ts k ʔ s h/ are all apparently consistently voiceless, so they're not even voice-unspecified.

---

Your periodic reminder that Sign languages are languages; also that the languages we have today are not a random sample and not a guide to the linguistic diversity of, say, 10,000 BCE.

Well yeah, but when I'm talking about phonological universals, I'm talking about spoken languages. Obviously sign languages are exceptions to all phonological universals, but they're trivial cases. And there isn't any reason to assume that the languages of 10,000 BC were any different to those of today, or for that matter those of 20,000 BC. Language change seems by all accounts to be fairly cyclical, so I'd argue that we do have a random sample. Not a complete sample, but a fairly good one at least.

If we did want to declare that no languages with voiced consonants exist— would that be like declaring that no human can be 9 feet tall (the record is 8' 11"), or like declaring that no human can be 90 feet tall? When we have close edge cases, it's pretty bold to declare a universal.

True enough. Until Northwest Mekeo was discovered (well, until someone actually got far enough through Jones' 800-page grammar to see the phonology) we could wrongly declare that "all languages have coronal phonemes" even though there were known edge cases like Samoan. I'm not sure you can really call Pawnee an edge case, because it does have two voiced consonants, the phonemic status of which isn't in any doubt, and in an inventory which only has 6 or 7 other consonants anyway.

In terms of phonology, it seems questionable to describe an inventory with no voiceless/voiced contrast as containing specifically "voiceless" consonant phonemes: I think many theorists would interpret the phonemes as being unspecified for voicing in that context, not as all being redundantly marked for voicelessness.

I think we can safely say that nasals and approximants can be considered "voiced" phonemes even without a voicing contrast, although I can't really explain why other than that they are voiced most of the time.

In terms of phonetics, some presence of allophonically voiced consonants seems likely, although it's hard to imagine that we could prove that this would necessarily exist in a language with such a consonant inventory.

Well, we can't prove anything in science, and linguistics is barely even a science, so I'd be satisfied with just a lack of counterexamples plus some good reasoning. I've got the former, but unfortunately not really the latter.

---

Yous've raised some good points. I'm pretty certain that it's a true universal in the sense that there aren't any exceptions – the only place I could plausibly see going against it would be North America where there aren't any new languages to be found unfortunately – but I don't reckon it's a proper structural universal.

[zompist]

Your periodic reminder that Sign languages are languages; also that the languages we have today are not a random sample and not a guide to the linguistic diversity of, say, 10,000 BCE.

And there isn't any reason to assume that the languages of 10,000 BC were any different to those of today, or for that matter those of 20,000 BC.

Sure there is: because agriculture was invented, and thus empires, and thus continent-spanning language families. We have lost a lot of linguistic diversity, and weird exceptions to 'universals' unfortunately don't fossilize.

[Darren]

Your periodic reminder that Sign languages are languages; also that the languages we have today are not a random sample and not a guide to the linguistic diversity of, say, 10,000 BCE.

And there isn't any reason to assume that the languages of 10,000 BC were any different to those of today, or for that matter those of 20,000 BC.

Sure there is: because agriculture was invented, and thus empires, and thus continent-spanning language families. We have lost a lot of linguistic diversity, and weird exceptions to 'universals' unfortunately don't fossilize.

Bizzare languages can appear even within large families. Mekeo is from Austronesian, an enormous family; Ontena Gadsup is Trans-New Guinea, and those both refute otherwise very secure universals. There were also only about a million people in 10,000 BC, so there's no way there were more languages then than now. That's not to say we have a picture of more than 10% of all languages in history (at a very rough guess), but I reckon it's still a fairly generic sample.

[jal]

Sure there is: because agriculture was invented, and thus empires, and thus continent-spanning language families. We have lost a lot of linguistic diversity, and weird exceptions to 'universals' unfortunately don't fossilize.

Not sure if I agree. First, it's an untestable hypothesis: we just don't know about the linguistic diversity of yesteryear, only that at this moment in time, we've lost a lot of it, not because of agriculture, but because of globalization. We can assume linguistic diversity was larger, but that leads me to my second counterargument: it seems, from the languages and language families we do know, that an increase in population led to an increase in diversity. When you have a small number of highly mobile people, linguistic diversity seems to be low(er/ish): there's enough intermingling to prevent wild divergence, as well as too little people overall. It's not until the population size gets bigger, and people travel further, that diversification seems to kick in, but even then it takes quite a lot of time, and people to stay more put, to turn mutual intelligible dialects into different languages. tl;dr aggriculture may have led to more linguistic diversity, through populations getting bigger en staying put more.


JAL

[zompist]

Sure there is: because agriculture was invented, and thus empires, and thus continent-spanning language families. We have lost a lot of linguistic diversity, and weird exceptions to 'universals' unfortunately don't fossilize.

Not sure if I agree. First, it's an untestable hypothesis: we just don't know about the linguistic diversity of yesteryear,

But we do-- we can see the Indo-European, Sino-Tibetan, and other families taking over huge territories over recorded history, and what we know of those areas from say 5000 years ago is that there were more families there. Where Europe is now almost entirely IE, and the Middle East is almost entirely Semitic, there were multiple language families or isolates.

I don't think we know much about what preceded Niger-Congo, but what would make you think it was a single continent-wide family like that?

it seems, from the languages and language families we do know, that an increase in population led to an increase in diversity. When you have a small number of highly mobile people, linguistic diversity seems to be low(er/ish): there's enough intermingling to prevent wild divergence, as well as too little people overall. It's not until the population size gets bigger, and people travel further, that diversification seems to kick in, but even then it takes quite a lot of time, and people to stay more put, to turn mutual intelligible dialects into different languages. tl;dr aggriculture may have led to more linguistic diversity, through populations getting bigger en staying put more.

Again, I don't get this. With agriculture, you get single families extending over entire continents. How is that more diverse?

It seems far more likely to me that Eurasia once resembled the precolumbian Americas: dozens of families. The Americas did have agriculture, of course, but it's precisely the areas with long-term agriculture (i.e. Peru, Mesoamerica) where diversity is less. Though again, in Peru as in Europe, we can see the spread of Quechua knocking out local languages-- Moche, Jaqaru, Pukina, etc.

[zompist]

Bizzare languages can appear even within large families. Mekeo is from Austronesian, an enormous family; Ontena Gadsup is Trans-New Guinea, and those both refute otherwise very secure universals.

Good point; certainly new weirdness can appear within an existing family.

Still, on the average, do you really think you're more likely to find weirdness in a previously undocumented Niger-Congo language (1000 languages in the family), or in a previously undocumented isolate?

There were also only about a million people in 10,000 BC, so there's no way there were more languages then than now. That's not to say we have a picture of more than 10% of all languages in history (at a very rough guess), but I reckon it's still a fairly generic sample.

That's a low estimate... Wikipedia suggests 3 to 8 million people. One pitfall is that remaining hunter-gatherers live in marginal areas, like the Kalahari Desert; in the paleolithic they could range over what is now agricultural land. The record density of hunter-gatherers seems to be the Chumash, at an impressive 21 people per square mile, mostly due to seafood.

The number of languages depends on the number of speakers, which is anyone's guess. I found a paper that suggested a median size of 800, which gives us 3750 to 10,000 languages.

[hwhatting]

Not really arguing for or against anything said here, but also keep in mind that hunter-gatherer language families can spread over vast territories as well; see Australia, which was mostly Pama-Nyungan before the European colonists arrived.

[Travis B.]

It's that time of the year again... I have heard multiple people pronounce pączki as [ˈpʰũʔtʃci(ː)] as of late. How in the world do you get from [ɒ̃], which would be close to the original Polish, to [ũ] of all things is beyond me... This is probably one of the worst hyperforeignisms I have ever heard. Surprisingly enough I have not heard the anglicizations that one would expect, with [æ]~[ɛ] or [ɑ]~[a], from anyone.

[TomHChappell]

Punchkey. Stress on first syllable. (Detroit area).

[Travis B.]

Punchkey. Stress on first syllable. (Detroit area).

Well that is more reasonable than [ˈpʰũʔtʃci(ː)] to me.

[TomHChappell]

(I don’t know where to post this!).

I have found one or a few natural language(s) with terms for
a consanguine relative’s spouse’s consanguine relative’s spouse’s consanguine relative:
(for instance [sibling’s or parent’s or child’s] spouse’s [sibling’s or parent’s or child’s] spouse’s [sibling or parent or child]);
or even more specifically
Sister’s husband’s sister’s husband’s sister
Or
Brother’s wife’s brother’s wife’s brother.

So, a fifth-degree kinship relation, in which
the first and third and fifth links are to blood-relatives,
but the second and fourth links are conjugal or connubial or marital or spousal,
has a kinterm meaning that relationship in this or these language(s).

(That term is an affine relative’s affine relative.).

But I’ve never found a language with any triply affine kinterms;
such as wife’s brother’s wife’s brother’s wife; or
husband’s sister’s husband’s sister’s husband.

A little more generally, a spouse’s sibling’s spouse’s sibling’s spouse.

Or either or both of the two occurrences of “sibling” could be replaced by “parent” or “child”.

….

Does anyone know of any natlang having a kinterm for such a a triply affine kinsperson?
A spouse’s blood-relative’s spouse’s blood-relative’s spouse?

(Note such a kinterm would be at least 5th-degree!)

What natlang, if you can say, and what is the kinterm and its English gloss?
Can you cite a reference? Does that reference have a URL ?

If you don’t know the answers, you might nevertheless still have a good idea how to go about finding them!
If that’s the case, can and will you please tell me/us how to find these answers?

….

Thank you!