zompist bboard

I think he honestly believes that, though.

Well, no one said it but it's probably true... I mean, I have no education beyond primary school and although I haven't had any proper delusions or anything like that for years and am not even depressed at the moment, maybe nothing I say makes sense and I'm typing practical gibberish half of the time that's on the level of insane as Time Cube, without even realising it. In that case it should be ignored and rejected. And I do believe I'm spouting bullshit most of the time even if I don't understand why it's bullshit and believe that what I say has value.

This... isn't the case? Like, I don't even know how to respond to this.

Turkic, Mongolic and Tungusic all show clear signs of being Mitian languages, if that is a thing. They also seem to show some innovations at the exclusion of the other Mitian families, such as the bi ~ min alternation in the first person pronoun, so Altaic may be a valid node within Mitian. Yet, we don't know yet whether these seeming relationships are real or not; we may well be dealing with a Sprachbund here. Perhaps Mitian as a whole is a Sprachbund (though it looks more like a family to me); perhaps Altaic is a Sprachbund within a Mitian family.



The term "related", in historical linguistics, usually means "related by descent from a common ancestor", and any other usage is misleading and thus should be avoided. Yet, a term like "mesh" or "cluster" is useful to designate a group of languages that show resemblances that are hard to tell whether they are due to descent from a common ancestor, or due to contact, and Indo-Uralic, Altaic and Mitian are, at the current state of knowledge, examples of just this.



There probably is no language that isn't influenced by other languages.



You are confusing crackpots and wound-down clocks But I know what you mean: a hypothesis insufficiently backed up by evidence, arrived at by faulty reasoning, and therefore methodically invalid can still be right - it may hit the truth by accident. Yet, to show that it has hit the truth requires proper scholarship.



No - I am an amateur in this field as well, and have a "mental" condition (a mild autism-spectrum disorder), but at least I try to adhere to proper method. That's what counts, not academic credentials (or at least, it should be like that; in practice, some good work is dismissed due to lack of academic credentials in the author, and some people get a large audience because of their academic credentials despite saying utter crap).



As I've said earlier, the majority of historical linguists don't consider Starostin's work valid because of its methodological flaws. Sure, it is better than most of the crap one finds on various crackpots' personal home pages on the Web, and I wouldn't reject something like the Etymological Dictionary of the Altaic Languages out of hand - some of the etymologies therein are probably right - but there are problems with it which need to be sorted out.

But there are also problems of similar kinds with more accepted works. For instance, there are two mutually incompatible reconstructions of Proto-Afroasiatic - one by Orel and Stolbova, and one by Ehret, both published in 1995 - on the market; they cannot both be right, which probably means that both are at least in part wrong.

As I've said earlier, too, that Afroasiatic is widely accepted while Mitian or Eurasiatic is not, is to a large part due to the different social dynamics of African and Eurasian historical linguistics - the latter seem to be more critical of macro-comparison than the former - rather than the different quality of evidence.



Surely, language relationship operates differently than genetic relationships between people (or other organisms), and this "language family" thing is a metaphor. A language usually descends from one parent language, though it is influenced by others. Modern English, for instance, does not descend from Norman French and Old English equally; it is a lineal descendant of Old English but has borrowed extensively from Norman French. The common claim that English was a Romance language as much as it is a Germanic one is wrong. People who say such things confuse language and lexicon (a common error also among macro-comparison crackpots, BTW!). Things like the strong verbs are purely Germanic, there is nothing Romance in them.

But the languages we are talking about here don't look much like creoles. It is hardly an overstatement, for instance, that in PIE every verb was irregular - there are about 20 ways of forming the present tense, and four ways of forming the aorist; show me a creole that is like that!

Also, the "hourglass model" according to which multiple strains of language converged in a single point from which daughter languages branched out is a gross simplification. Apart from linguistic converge usually not leading to a single mixed language but a Sprachbund - a set of languages sharing some traits, but still distinct languages - there is both branching and influence all the time.



I am definitely not a "hypermegalumper"! I have some lumperist leanings, but to me, things like Indo-Uralic or Mitian look intriguing, yet I am not convinced that they are real yet. And I don't seriously expect to make meaningful contributions to their exploration. This is not something an enthusiastic amateur could cobble together in a few years - otherwise, it would have already been done long ago - but something which could support the entire careers of several academic linguists. See Tropylium's roadmap, originally posted to the late, lamented Nostratic-L mailing list. Sure, amateurs and students can achieve something in this discipline (as the examples of Linear B - deciphered by a young architect - and PIE laryngeals - first reconstructed by a university student - show), but I don't seriously expect to be the next Michael Ventris - yet, I try hard not to become yet another Octaviano!



Fine. As for me, I suspect both Indo-Uralic and (Micro-)Altaic to be valid nodes within a still larger enitity, which I call Mitian; yet, all this may turn out to be illusionary. What I am sceptical of is Uralic-Yukaghir, though: while Yukaghir shows all the hallmarks of a Mitian language, it is IMHO not particularly close to Uralic. The morphological resemblances between Uralic and Yukaghir are somewhat less than those between Uralic and Indo-European, and the Uralic-like words in Yukaghir look like borrowings from a Samoyedic language; while Yukaghir and Samoyedic are now about 2,000 km apart, they probably were closer in earlier times, as both the Yakut and Evenki languages which now occupy the space between them are fairly recent intruders.



Fair.

_________________
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

If one is only interested in reconstructing linguistic family trees, that is a fair point. However, in so far as it is useful to ascribe a language to a family, Vlürch has a valid point. As a less extreme example, modern English seems to show enough North Germanic features that it may not be useful to classify it as a West Germanic language. And is it actually useful to describe Frisian as a member of the Anglo-Frisian group rather than as part of the West Germanic dialect continuum, which excludes English? Furthermore, English seems to have a rather strong dash of relexified Norman French.

At the very least, there may be some advantage to being able to describe a language as 'your average Eurasiatic language", regardless of the existence of a Proto-Eurasiatic language.

What else would you be interested in? Moreover, given this is a thread about reconstructing linguistic family trees, why shouldn't we only be interested in it?

Now I'm confused because it's impossible to reconstruct any of the proposed families in the title...

And yet, there are people claiming that Nostratic has been reconstructed already by the likes of Illich-Svitych, Starostin, Dogopolsky and Bomhard. I think that is nonsense, and those macro-reconstructions are interesting conlangs at best. Another question is whether we need a reconstruction to assume a language family. There are certain parts of a language that are more resistant to borrowing, like morphology, basic vocabulary and pronouns. If we see a lot of commonality in those categories, we can at least say that these languages are possibly genetically related. And then we can investigate this relationship using the comparative method. And I think the best way to do this is bottom-up, by first looking at the language pairs/groups that may be more closely related, like Koreo-Japanese and Indo-Uralic.

science is fake. nothing is true. everything is permitted. hth

_________________
Siöö jandeng raiglin zåbei tandiüłåd;
nää džunnfin kukuch vklaivei sivei tåd.
Chei. Chei. Chei. Chei. Chei. Chei. Chei.

Classificatory labels that predict many of the more conservative features of a language can also be of use. At least, I assume that it should be useful to know that a language is Afroasiatic, regardless of whether Afroasiatic is truly a genetic entity.

Pair by pair reconstruction is the worst way to do it. For example, consider reconstructing the Swadesh list. Every time at least one of the pair has replaced a word, one can no longer reconstruct the word in the parent language. There's a good deal to be said for comparing the languages or groups four at time.

The argument for reconstructing Indo-Uralic is not that it is a valid node, but that there may be enough material attested across the Indo-European and Uralic groups to enable construction of reasonably likely features of the common ancestor. Many perfectly valid reconstructed items will I think formally be doubtful on the basis solely of Uralic and Indo-European. That is in the nature of history. We may then be able to add other groups as either relatives or Indo-Uralic, or as independent members of Indo-Uralic. Not knowing these independent members is a misfortune, not a benefit.

But even then, you'd have to take language contact and chance (especially typological) similarities into account, too, since these can also help explain striking linguistic similarities.

You can't compare whole language families just based on proto-languages and be totally oblivious to the language branches down the tree. So when you are comparing IE and Uralic, you are actually comparing Finnic, Saamic, Volgaic, Permic, Ugric and Samoyedic against Hittite, Tocharian, Indo-Iranian, Hellenic, Italo-Celtic, Germanic, Balto-Slavic etc. Any approach that just takes a fixed Swadesh list from a proto-language and limits itself to that is bound to fail. So why would anyone want to do it that way? The big danger of macro-comparisons of more than 2-3 language families is that the set of language branches down the tree becomes to big. And then it does become just a comparison of the proto-languages themselves. And when a little later, a better reconstruction of one of the proto-language reshuffles some phonemes, your whole reconstruction comes crashing down. And you are left with arguing against progress, just like the Moscow school is doing with proto-Japonic *e and *o.

I've seen pointed out that language classification is mostly "useful" as a taxonomy; and the people who need taxonomies (in the intro chapters to synchronic, typological or micro-comparative linguistic works) do not actually care if the taxonomy is historically accurate or not. So in this fashion it would be totally workable to treat English as a "North Germanic" offshoot.

History enters the picture for a couple of different reasons, some of which just conflate concepts with one another, but one practical point is the possibility to choose between different classifications: if the North Germanic features of English can be shown to be younger than the West Germanic ones, then that's a generally applicable unique criterion to prefer the latter classification over the former. (The opposite doesn't work: connections get weaker as they get younger.)

A lot of the time this doesn't really work though. Are commonalities between groups like Germanic and Balto-Slavic shared archaisms, late areal connections, or signs of a real genetic subgroup?


Case in point: grouping Mordvin and Mari together has been obsolete since the 80s, but this has been more due to negative evidence (most commonalities can be shown to be superficial, trivial, or shared archaisms) than due to any newly proposed subgroup being now considered clearly true. The currently "active" alternative proposals are West Uralic (Samic + Finnic + Mordvin) and Central Uralic (Mari + Permic), but they are only barely better, far from the quality of the primary groups like Mordvin.

_________________
[ˌʔaɪsəˈpʰɻ̊ʷoʊpɪɫ ˈʔæɫkəɦɔɫ]

I used the Swadesh list as a simple example of the problems of iterated binary comparison. While the multiple branches are what make comparing Uraic and Indo-European appealing, you still face the issue that anything simple lost on the way to Proto-Uralic or on the way to Proto-Indo-European will be missing from reconstructed PIU.



And here we reach the issue of judicious reaching down.

Or signs of a substrate language?


JAL

I suspect the classification of Romance is an example worth considering.

Most of the BSL-Germanic commonalities, except for the "m" in the plural case endings, look like areal developments to me.*1) And those are hard to distinguish from substrate influence, but assuming substrates only adds value if we have language material that is structured in a certain way that makes it likely that a different language (or language grouping / family) was involved, like the Temematic material that shows a different development to BSL of the PIE stop system, or the "bird language" material with its non-PIE ablaut a-CRC vs. CaRC, or the un-etymologised Germanic nautical lexicon (these are example where I at least see a basic case for positing substrates; I'm not sufficiently up to speed on all these hypotheses to say how well they hold up).
*1) The m-endings may be areal as well, but at least that must go back quite a way to when the IE plural case system was still being formed. And if you look at the verbal systems, BSL on one hand and Germanic OTOH look to me as if they parted ways relatively early.