Isolate Languages

An "Isolate Language" is a human language with no detectable relationship (metaphorically "genetic") with any other language. French, Spanish and Italian are all very similar, Finnish, Estonian and Hungarian are less obviously related to anything. As with many things in linguistics, absolutes are hard to come by.

Further debate follows ...

(Finnish is closely related to Estonian, and both more distantly to Hungarian; all three are members of the Finno-Ugric language group. While not related to the Indo-European languages that largely surround them, they are related amongst themselves.

Do you have references for this? I am aware that a great deal of speculation has been offered on isolates, based on very little evidence, but I am not aware of any actual agreement in the linguistics community on speculations such as you have voiced. Are you voicing a minority opinion as fact?

[As a Finn, I'd like to point out that no finnish people in their right mind would doubt the relatedness of Estonian and Finnish. The two languages are so close that speakers of one understand short passages of the other without studying it. Better examples of IsolateLanguages are Basque, Albanian, Greek.]

That wasn't the intended question. What about the claimed relationship to Hungarian, is that firmly established? I had previously thought that it definitely was not.

My understanding is that Finno-Ugric is quite firmly established. Some random links:

Ok, that's what I thought. If you look at all of that closely, what you will find is that isolates are sometimes grouped on the basis of synchronic resemblance, such as "large number of noun cases", but IN NO SENSE do any of those, or any other references, claim that such resemblances are a result of GENETIC relatedness.

So it goes back to what I said originally; isolates are isolates, and there is no established theory that confirms genetic relationship for any traditional isolate, especially including Japanese, Hungarian, Finnish, and Basque (although there are others as well). These languages have NO KNOWN genetic relation to any other firmly known language, living or dead, despite sometime structural resemblance. Period, end of story, nothing new here in recent years.

While the above sites don't explicitly say Uralic languages have a common origin, it is somewhat implicit in talking about the Uralic family tree or the forefathers of the Uralic speakers. I have not been able to find a source which says they aren't related, and lists of language isolates do not include them. Also, there is the hypothetical Uralic-Altaic. That is very speculative, but objections to the idea don't seem to include that the Uralic languages are not a genetic family, which one would expect. Maybe you have some better reference you can point out?

This is a far more complex, controversial, and difficult subject than you seem to realize. It's not even a certainty that Proto-Uralic ever existed...I wouldn't argue against it, but it's not that clear cut. A quick web search reveals this reasonable-seeming page:

http://www.geocities.com/Athens/Acropolis/3093/Uralists_Against_History.htm

But think about it: if you go back far enough, presumably all languages are related, and therefore there cannot be an "isolate" by definition. But this is a counterproductive definition.

I'm curious why you assume that - why couldn't language have been invented at a number of different places and times once the brain circuitry was in place... Certainly, the enormous variety of syntaxes and different concepts being stressed by speakers of different languages makes it pretty unlikely that all languages sprang from the same root stock. "Spoken Here" by Mark Abley (ISBN 061823649X ) gives a good idea of this variety, and it probably just scratches the surface. -- PaulMorrison

In retrospect I think I did not answer this at all clearly. When you say "pretty unlikely that all languages sprang from the same root stock" -- I strongly disagree, I believe most linguists disagree, and further I am unaware of any evidence for that position.

I am aware of the variety. But it is quite clear that 30,000 years (if you accept Greenberg's work in the Americas, or less than that if you don't) is sufficient for a single language to splinter into fragments which do not appear related whatsoever, and can have literally any known language feature whatsoever. There is every reason to think that modern languages are older than that. So part of the point is that you absolutely can't distinguish between 1 primordial Ur language that gave rise to all current languages, versus the case of multiple independent language origins. The case today would be the same.
I am indeed assuming that there was only one original Mother Tongue ("Proto-World"), because I'm a believer in Greenberg's work, and it seems to point in that direction. But I freely admit that we don't know that for a fact. We don't know that it's false, either.

On the other hand, it's hard to justify any particular choice of cut-off for a time horizon. Surely it depends on what evidence is available in individual cases. Nonetheless, beyond 8,000 years B.C.E. evidence becomes extremely sparse indeed, which is why Greenberg's reconstruction of the families of the languages of the new world are so very controversial.

So what's the bottom line? Well, on the one hand you could say there are no isolates. Or you could say that a few languages are isolates relative to everything else. We have excellent evidence of the precursors to and influences on English. We have no direct evidence at all about the precursors to Japanese; the languages most closely related to it 2,500 years ago apparently died long ago. The entire Uralic family is much more complicated, but although I'm not too strong on the diachronic aspects of Finnish, I was under the impression that the evidence was scanty. And similarly for Basque and some others. -- DougMerritt

I agree entirely that the matter is complicated, but that's no reason to ignore what's been done. Japanese and Basques have no known, or at least no relatives among languages, or at least none we can put much faith in. Finnish is different. However poor our understanding of Proto-Uralic, the commonalities and particular relationships among the Uralic languages, it doesn't seem many people have been questioning its relationship with Estonian at the least. And heck, there are still people who question the validity of Indo-European (curiously enough, the one other time I've seen appeal to real experts). Lists of language isolates, in direct contradiction to someone's claim up above, don't have it. Maybe it's a little premature to give it an exact placement with respect to all other tongues, but saying it has no known relatives is unjustified.

My "appeal to real experts" some months ago was merely a very badly phrased thought, which I've deleted. As for the existence of Proto-IndoEuropean?, I would call that very firmly established after 300 years of work, although details of reconstructions are always open to critique. As far as I know, the only people who challenge the very existence of PIE are those who challenge the notion of indirect evidence in the first place, and demand e.g. written records of a language. That doesn't constitute an intellectual position, it's more of a phobia. Neurotic. The world is full of uncertainty.

I'll accept that. I don't know that much about Uralic, and what I do know is out of date. My heat had purely to do with Japanese, but my phrasings weren't thus contained. -- dm

Meanwhile, Basque and Japanese are respectively derived from ancient Basque and ancient Japanese; we just don't have separate names for the older forms like we do for "Latin" and "Spanish".

It is irrelevant that Basque and Japanese have evolved from ancient forms to newer forms, because, unlike Latin, the ancient version has exactly a one to one relationship with the modern version; there are no "cousin" languages that diverged from the ancient versions.

Further, while neither is known to be related to another distinct contemporary language, both modern Japanese and modern Basque do consist of multiple dialects with greater or lesser degrees of mutual intelligibility.

Although the word "dialect" is infamously abused and often ambiguous, nonetheless, it is unusually clear in the case of modern Japanese and modern Basque, in that all variants of those two languages are mutually intelligible, the number one criteria for defining a "dialect".

Declaring two spoken tongues to be "dialects" or "related languages" isn't always a clear call.)

Very true, but that doesn't seem to be an issue here. "Dialects" of Japanese and of Basque are not so diverged as are "dialects" of e.g. Chinese.

Actually, speakers of different dialects of Japanese sometimes find it difficult to understand each other, which has traditionally been the border between different "languages".

Unless something has shown up since I went to university, Sumerian is an isolate. I have always found Sumerian interesting, because we have had to deduce its sounds through an Akkadian filter, so for all we know it may have been full of sounds that the Akkadians couldn't represent - but that is unlikely to change its status as an isolate. -- PaulMorrison

Then there's my own favorite: Etruscan. Here is a quote from http://www.netaxs.com/~salvucci/VTLfacts.html : "With the steady process of discovery, there is increasing confirmation that the language is related to no living language. In particular, Etruscan does not belong to the Indo-European family of languages, and thus cannot be connected to Greek, Latin, Sanskrit, or any other descendant of that branch. Only two languages show a resemblance to Etruscan: Rhaetic ... and an even closer language recorded on inscriptions and pottery from the Greek island of Lemnos. But though the connections are interesting, they regrettably do not bring us any closer to placing Etruscan into the scheme of world languages." -- PaulMorrison

I had to put a space between that URL and the following ":" to make it work, otherwise the ":" ended up part of the URL.
Interesting, I don't think I've looked much at Etruscan since before I was in college, so I didn't remember anything about it. I think I've had it mixed up with Phoenician before -- unforgivable considering the historical/archaeological distinction. My first thought was whether there was any evidence for a relationship with Basque, but I see that (of course) this has been checked rather thoroughly. On the other hand apparently it is related to Lemnian -- of which we know little -- and possibly Camunic. We probably need to lift up Rome and move it somewhere else, so we can look underneath for remaining Etruscan artifacts. More text! (Was there known to be some Etruscan texts at Alexandria?)

Funny you should mention that: Claudius was supposed to have compiled an Etruscan-Latin dictionary, but it has never been found. If there was a copy in the Alexandria library, it probably burned when they torched the library. A friend and I have fantasized that there may be a copy in someone's attic - after all, if they can find what quite likely is a genuine portrait of Shakespeare in an Ontario attic, why not Claudius' dictionary?!

Since wikizens don't shy away from controversy, I will state now that I still find Mayani's claim very plausible that Etruscan and Albanian are closely related. I know this is not currently accepted by most linguists, but I have yet to see a convincing proof why this is unlikely. -- PaulMorrison

"I'm curious why you assume that - why couldn't language have been invented at a number of different places and times once the brain circuitry was in place... Certainly, the enormous variety of syntaxes and different concepts being stressed by speakers of different languages makes it pretty unlikely that all languages sprang from the same root stock."

I thought it was widely believed that our first common ancestors to leave Africa already had language. Is anyone seriously suggesting that they didn't, or that some group lost language and re-invented it after that? -- EricHodges

Everything related to the topic is very controversial, and about every other month "strong evidence" comes out that turns the previous thinking around 180 degrees. Individual researchers have strong opinions, but it's hard to find a universal consensus.

It is unclear whether our first common ancestors left Africa, or whether they started from Asia, or perhaps evolved independently in multiple areas but still interbred.

It is unclear exactly when "language" first evolved, or in what form -- for instance, it could have been sign language. Or it could have been single words. The simplest forms could be millions of years old. Or not.

The intermediate forms between that unknown simplest language, and the current modern form of language, are equally unknown. One hypothesis championed by the linguist Derek Bickerton is that universal grammar, in the form evidenced by creoles, was such a late-stage intermediate form. Others disagree.

I had written "the late Derek Bickerton" originally because last year someone told me he had died late last year. Since he just published something (and presumably does not have ElRon?'s after-life publishing skills), happily I seem to have been mistaken.

See the discussion in the middle of CreoleLanguage - Claire Lefebvre seems to have proved pretty conclusively that Haitian Creole ties (Africanized) French words together with Ewe syntax. -- PaulMorrison

We know that modern language, in all its complexity, existed 30k years ago. It is widely thought that it existed 100k years ago, but this is disputed; some think that 30k years ago is just about when the fully modern form finished evolving. Others think that the full modern form has been around for several 100k years.

The descent of the larynx obviously was caused by evolutionary pressure to improve language, and that sort of thing takes a bit of time and has some prerequisites, but that's one of the few things that is known as opposed to being slippery and debatable.

So, back to your question: I thought it was widely believed that our first common ancestors to leave Africa already had language. Is anyone seriously suggesting that they didn't, or that some group lost language and re-invented it after that?

At the time of the OutOfAfrica? event(s) (whether it happened or not), most think that language already existed, but not in its modern form. No one is suggesting that some group lost language, to my knowledge.

-- DougMerritt

I thought all of the alternatives to "OutOfAfrica?" had lost their backing at this point. I used to be fascinated by the idea of multiple germ lines contributing to modern man, but the genetic evidence just hasn't supported that idea. Anyway, if humans had language when they left Africa and the population was small, then they probably only spoke one language. If no group since then has lost their language for more than a generation then every language on the earth must be a memetic descendant of that common ancestor. -- EH

Mostly I agree, but there's an awful lot of this sort of thing that is merely reasonable, as opposed to proven beyond reasonable doubt.

Side note: There may always be competing theories to "OutOfAfrica?", due to national pride in China and the rest of Asia if nothing else -- however, a steady stream of interesting new discoveries comes out of Asia that does help keep alternatives alive, so it's not completely crazy.
Sigh -- another side note, about genetic evidence: we don't know anywhere near enough. Very recent research about mitochondrial inheritance seems to completely invalidate the assumptions that previous research (such as the famous MitochondrialEve) were based on. We know less today than we thought we did 5 years ago.

One slippery point, though. If the "language" that existed when emigration from Africa occurred was very simple compared with modern language, then it wouldn't mean much to call it one language versus many, because the modern distinctions between languages might not have existed at that point.

There are two primary algorithms at work: for a single interbreeding population's gene pool to drift under pressure from natural selection (and sexual and kinship selection, which were more important in some ways), and for a geographically separated bud population's gene pool to innovate -- at which point either speciation occurs, or the new traits spread back to the parent stock even though the isolation slows the diffusion rate.

Basically every variation on this theme that can be imagined did in fact happen, but the details are very unclear. We're not even sure of the speciation points -- e.g. was Neanderthal a separate species? This is far from settled, despite frequent triumphant press releases of research claiming one side or the other.

So whenever we say "language" or "human" or "migration" or any other key words on the topic, we're being too vague. What kind of "language" are we talking about? What kind of "human"? Even "homo sapiens" isn't specific enough, but if it were, we're not entirely sure when that species arose, exactly.

Two fundamental mechanisms of human language are morphology and grammar. Languages with simpler morphology, like Chinese (and English, to a lesser extent) have more complex grammar to implement the same mechanisms that other languages (e.g. Inuktitut, Finnish, and to a lesser extent, Latin) accomplish with morphology.

So in certain ways, morphology and grammar are functionally redundant -- just one of the two would be sufficient to get the job done. This suggests to me that the capability for each arose genetically in separate subpopulations, but then diffused backward into other populations, so that eventually all subpopulations were capable of both mechanisms (as today), with the specifics of use of the two mechanisms becoming a cultural trait.

If speculations of this sort are at all accurate, it means that questions like "did all languages descend from a single mother tongue?" are too simple and assume too much to have good answers.

And actually, even if my specific speculation is completely wrong, nonetheless similarly complex events are practically certain to have happened, so again, simple questions about our origins may not have simple answers even once we understand the whole process.

-- DougMerritt

I'm not a fan of "species", so I don't really worry about that kind of classification for organisms or languages. It seems to me that if all humans spoke one language at some time (no matter how different it was from modern languages) and each generation copied the previous generation's language with modifications, then all of our languages have a common memetic ancestor, even if the variations move between populations. I love the regional continuity theories, but I've come to doubt them. They seem to be based on a fundamental misunderstanding about the influence of environment on morphology. But that's an entirely different rant. -- EH

What could it possibly mean to not be a fan of species? But no matter what you mean, I only mentioned it in passing; the point is that it is an example of genetic change at a dramatic rather than tiny level.

If we began with a common language of a single word, and the evolution to multiple words occurred in separated regions, and you wish to call that "having a common ancestral language", well, that seems a confusing definition to me, but so long as you're clear about what you mean...

I didn't argue in favor of RegionalContinuity?, so perhaps that means I should have stuck to a single paragraph response. -- dm

Regarding what it means to not be a fan of "species", Darwin said:

"In short, we shall have to treat species in the same manner as those naturalists treat genera, who admit that genera are merely artificial combinations made for convenience. This may not be a cheering prospect; but we shall at least be freed from the vain search for the undiscovered and undiscoverable essence of the term species."

I'm not convinced that dramatic genetic changes are of interest, only accumulations of tiny changes. Tiny changes created the diversity of biological organisms we see on the planet today. I have no trouble believing they created the diversity of languages we hear, regardless of the difficulty we have in establishing ancestry.

I thought the most popular alternatives to the OutOfAfrica? theories were the regional continuity theories. I know you aren't arguing in favor of them, I'm just defending my belief in the OutOfAfrica? theory.

I didn't say the language spoken by our African ancestors consisted of a single word. I'm not sure where you got that impression.

It wasn't intended to paraphrase you per se. It is reasonable to guess that language began with single words, rather than springing full blown from the brow of Homo Zeus with sequences of verbs, subjects, and objects. Occham's razor.
But that misses the point. If all of the humans alive today are descended from humans who lived in Africa 50,000-100,000 years ago and those humans who lived in Africa had language (not just one word) then it follows that all of our languages are memetic descendants of the language(s) spoken by those ancestors. If the migratory population was as small as some believe, the chances are that a single language migrated out of Africa with others (if any) migrating later over that single language.
Well, yes, if you assume all those ifs. But that, in turn, misses the laborious points I tried to make, cautioning that such ifs are far from proven. If you examine your phrasing very closely, it's actually worse than that, you've basically set it up as a circular argument, immediately above. This might be because you regard the ifs as so reasonable, but that's not the same as true.

-- EH

I agree with Doug's cautions above. Also I don't know how you get from the first grunts (I assume) to languages with complex inflections (e.g. Latin pluperfect subjunctive). I know that "grammaticalization" has been suggested as the mechanism - or maybe that's a general term for a process that we don't understand, but presumably must have taken place. Also, and this may or may not relate to this last point, I have always found it interesting that both PIE and the Semitic languages seem to attach most of the "meaning" to the consonantal backbone of words, rather than their vowels. There are even some Caucasian languages with almost no vowels. Far-out idea: what if this was a result of the pre-human larynx?! -- PaulMorrison

I have no reference for this, but I remember reading that some have suggested that the lower primates can't speak because they can't form consonants. To test this, a language was devised that was entirely vowels. The researchers learned it, and then found that they could not understand each other. A language comprised of only vowels seems very hard to understand when spoken., whereas replacing all the vowels with just one doesn't seem to reduce the understandability by much. Consider that the New Zealand, South African and Australian accents tend to be largely vowel shifts and/or replacements, and yet they are still easily understood. Replace all vowels in an utterance by "i" and you get a Sith Ifricin (say it aloud) accent, for example.

[There must have been more to it than that. A consonant is approximately a partial or complete obstruction of air flow during voicing or sibilance. Other primates can surely achieve at least one such obstruction, e.g. by closing the mouth and lips, so a more reasonable test would be a language with just one consonant, not zero. -- DougMerritt]

I have often wondered if you could use Information Theory on sound complexity - in my experience languages with fewer consonants tend to have longer words, e.g. Polynesian languages. OTOH, the written form of many Semitic languages didn't bother to show vowels, without much loss of comprehensibility - and the words didn't get any longer. -- PaulMorrison

Yes, it's been applied quite a bit. The primary issue is not, however, the number of consonants, but rather the number of distinct phonemes in a language. As I recall, English is in the neighborhood of 45 phonemes, Hawiian 25, and some other languages 60. Any given utterance translated into those languages will, on average, require more phonemes per utterance in the languages with a smaller number of phonemes, and vice versa, exactly as you'd expect -- although the issue is much more complicated for any given utterance, and by questions of specialty vocabulary, idioms, etc.

I phrased the above to avoid the really nasty issues of "word length", "phrase length", "sentence length", because it's not in general to do an apples to apples comparison on those units, nor even to rigorously define those terms in a universal way.

Surely the point of the previous discussion was not to lump all phonemes together, but to see whether vowels and consonants contribute differently to the information content... -- pm

Yes, it's just that that is a harder question, since it is phonemic distinctions that are fundamental, whereas vowel versus consonant is fuzzier both to define, to discern, to figure out equivalence categories, isn't actually just a two-way division of all possibilities (it is appropriate to have further categories), and of course varies a lot per language family. It's true that Semitic languages have less reliance on writing vowels because vowels tend to carry less information and thus can be easily inferred, while Polynesian languages quite obviously carry far more information in vowels, since many words are distinguished only by vowel differences. The problem is in well-defining this situation for languages that are not at the extremes!

However, it has been frequently pointed out that, although the complexity M of the morphology system of languages varies sharply (e.g. Chinese has low morphological complexity, Latin medium, and Finnish high), and although the complexity G of sequence-based grammar of languages varies sharply, that nonetheless the total complexity seems to be more or less conserved across languages; M * G = constant. Simple morphology leads to complex word-order grammar and vice versa.

This should be unsurprising, since constituent role has to be signalled somehow, so if it is not done by morphology, then it needs to be done by grammar, and vice versa. (Then again, it's not quite that simple, since some roles are simply inferred by the listener, as is frequently the case e.g. in many languages of Asia with pronouns.)

Your point about Asian pronouns (I assume e.g. Japanese) opens up the issue of the total context of an utterance - beats me how you would tackle that mathematically! -- pm

I was actually thinking of e.g. Chinese, Thai, Vietnamese, but yes, Japanese too. (Since Japanese is not geneologically/grammatically related to those other languages, presumably there is a regional cultural influence at work on these languages that produced this similarity; inappropriate use of pronouns in all of these languages is a faux pas/politeness issue.) Beats everyone else how to tackle total context, too, but the situation isn't that dire here -- I mentioned the pronoun issue just to make it clear that M * G = k is just an approximation. It's not a bad approximation, though. So despite the fact that linguistics is, in some ways, still in the dark ages, there are nonetheless some approximate quantitative forces that can be identified. We're not totally in the dark.

ZipfsLaw does turn up as a cross-linguistic universal. -- DougMerritt

Do Finnish and Estonian programmers use HungarianNotation? The world's most famous Finnish programmer (LinusTorvalds) is known to despise it... :)

CategoryNaturalLanguage