Don Ringe ties up some loose ends

« previous post | next post »

Don Ringe's guest post back on January 6, "The Linguistic Diversity of Aboriginal Europe", sparked a lively discussion, and Don responded with a series of other posts responding to questions and comments: "Horse and wheel in the early history of Indo-European", "More on IE wheels and horses", "Inheritance versus lexical borrowing: a case with decisive sound-change evidence", "The linguistic history of horses, gods, and wheeled vehicles", "Some Wanderwörter in Indo-European languages".

Now, despite his claim that "after this I'm going to have to stop posting for a while", he's sent another essay:

Here is an interim post clearing up some outstanding questions; the last part turned into an in-depth discussion of the "thorn cluster" problem that will probably be published somewhere eventually.  I do mean to get back to the "IE homeland" problem, but that won't be possible till the summer.

Here's hoping that he gets drawn back in again earlier than that!

Don is a user of the Athenian font, a polytonic Greek font produced by the American Philological Association. Now that the problems with Unicocode representations of polytonic Greek have been fixed, more or less, this font is deprecated; and for good reason, because setting up browser preferences, font installation, etc., so that you could see it correctly on a web page is difficult. In Don's earlier guest posts, I've gone through them and (along with re-doing the MS Word formatting) substituted Unicode characters for the Athenian code points. This time, I've decided to put a .pdf up on the web and link to it. So click here to read it!

(And Don has decided to switch to Unicode Greek in the future, though he echoes Rolf Noyer in complaining that "it clearly wasn't put together by a classicist".)



71 Comments

  1. John Cowan said,

    February 20, 2009 @ 5:24 pm

    Nick Nicholas has a great explication of the issues around Greek Unicode.

  2. dw said,

    February 20, 2009 @ 6:48 pm

    Thanks again for the wonderful, stimulating analysis!

    I have a question about the Indo-Iranian thorn clusters. In the case of a form like Skt. tákṣā = Av. tašā, your analysis gets us as far as *téḱtsō. What happened during the rest of the development of the thorn cluster? By the Satem shift we would expect *ḱ to become the unaspirated member of the first palatal series, which I'll notate here as *ć. The normal development of *ć would be Sanskrit ś and Avestan s. (see e.g. PIE *ḱm̥tóm > PII *ćatám > Skr. śatám, Av. satəm)

    I think I remember Burrow, who is probably seriously out of date by now, suggesting a series of developments something like ḱts > ćts > šš, which would give a final Proto-Iranian form *táššā. From here Avestan could have simplified the *šš to a single *š, while for Sanskrit Burrows postulated a further šš > ṣṣ > ṭṣ > kṣ, finally arriving at the desired form tákṣā.

    This seems an awfully long journey to get from *ḱts to kṣ! If *ć were phonetically something like [tʃ] then the shift ćts > šš could be phonetically plausible, I suppose, with the s being assimilated to the post-alevolar sibilant. On the other hand a direct leap from *ćts to *tš might also be possible.

    I would be really interested to know your current thinking on the post PII developments. Would you say that in the environment *ḱts the normal Satem shift was inhibited? This would simplify the derivation of the Sanskrit, but would leave us requiring a further set of developments for the Avestan forms.

  3. dr pepper said,

    February 20, 2009 @ 11:29 pm

    On a seperate note, is there anywhere i can find out how to pronounce all these consonant blends? I don't mean just hearing them, i mean instructions on how to make my mouth do that.

  4. Philip Spaelti said,

    February 21, 2009 @ 4:50 am

    And since we're talking about fonts (and Unicode), I have a question (or a request?). The css for this blog and website specifies to use Verdana. Why? The Verdana that I have on my mac is not a good font for such discussions as it doesn't include many of the phonetic characters used in such posts. Is Verdana on Windows (or whatever platform the bloggers use) better at this? Couldn't the website either (a) specify a different font, or (b) allow users to override this choice?

    [(myl) This choice was made by whoever created the WordPress theme that we picked when I set up the new version of this weblog site last spring. I'm a bit reluctant to mess with it, since it seems to work (and look) OK on the various machines that I use (including two Macs as well as various Linux and Windows computers), and I'm concerned that another choice might have other unintended consequences that I'd then have to deal with again later. But I'd be happy to consider suggestions.]

  5. Jesús Sanchis said,

    February 21, 2009 @ 7:16 am

    I wonder what could lead a speech community to use something like *h2ŕ̥tḱos to call a bear. Is it the onomatopaeia of a bear's claws scratching the bark of a tree? Or the sound of a bear chewing a salmon?… I also wonder what kind of value Mr Ringe assigns to these reconstructed 'PIE' words. On the other hand, I'd like to know if Mr Ringe can find a plausible explanation for the fact that, in many IE languages, this common word for "bear" was substituted by other, possibly euphemistic, words. Are these easy to explain in a post-Neolithic environment, like the one usually proposed in traditional theories?

  6. language hat said,

    February 21, 2009 @ 6:33 pm

    Nick Nicholas has a great explication of the issues around Greek Unicode.

    I hate to use the hackneyed "tl;dr," but that Nicholas piece is not only incredibly long, it includes lots of stuff that doesn't directly bear on Greek and is inconveniently broken up so that you have to click a link to find out what's in a given chunk. Is there some shorter explication available? I'm curious about the issues but truly don't have time to delve into Nicholas.

  7. marie-lucie said,

    February 21, 2009 @ 9:32 pm

    JS: quoting from Wikipedia:

    — the Proto-Indo-European word for bear, *hr̥ktos (ancestral to the Greek arktos, Latin ursus, Welsh arth (cf. Arthur), Sanskrit *ṛkṣa, Hittite hartagga) seems to have been subject to taboo deformation or replacement (as was the word for wolf, wlkwos), resulting in the use of numerous unrelated words with meanings like "brown one" (English bruin) and "honey-eater" (Slavic medved). Thus four separate Indo-European language groups do not share the same PIE root. The theory of the bear taboo is taught to almost all beginning students of Indo-European and historical linguistics; the putative original PIE word for bear is itself descriptive, because a cognate word in Sanskrit is rakshas, meaning "harm, injury".

    Greek arktos, Latin ursus, look like past participles, therefore descriptive words derived from verbs. Onomatopeia (imitation of actual sounds) is a rather marginal form of word formation, and it is found mostly in names of birds with distinctive calls (eg chickadee, whippoorwill). The PIE reconstructed word is not so strange as it seems: *h2 means that it is some type of laryngeal (could be h, a glottal stop, or similar sound), the r is syllabic (as in Czech and some other languages), and the -kt- cluster is not a pronunciation problem. The renaming (through nicknaming) of an animal because of a taboo (eg using the actual name of the bear could warn the bear of the hunter's intentions) is well-attested in a variety of hunting cultures.

  8. Jesús Sanchis said,

    February 22, 2009 @ 4:25 am

    Thank you for your comments, Marie-Lucy. In any case, if you take a look at this post: The Linguistic Diversity of Aboriginal Europe: http://languagelog.ldc.upenn.edu/nll/?p=980, which is the first in this interesting series by Donald Ringe, you will see what I think about reconstructed PIE and laryngeals, and you will also realize that my comment about onomatopaeias, etc., was ironic.

    One of the things I said then was that the term "laryngeal" is completely irrelevant in modern phonology. It does not refer to any point of articulation, like "bilabial", "palatal", etc. It has no real meaning, which is perfect for something as absurd and artificial as the laryngeal theory.

    Now, I'd like to focus on one detail in your comment. You say the following: "The renaming (through nicknaming) of an animal because of a taboo (eg using the actual name of the bear could warn the bear of the hunter's intentions) is well-attested in a variety of hunting cultures". According to the orthodox view, PIE was still a more or less unified language by 4000 BC, or even later. i.e. a post-Neolithic period, and the speakers of this 'protolanguage' are not supposed to be 'hunter-gatherers'. Why is it that, after this imaginary post-Neolithic expansion, new euphemistic words for 'bear' were coined in the various IE languages? Would a post-Neolithic society see a bear as some kind of totemic animal or as a animal with a tabooed name?

  9. marie-lucie said,

    February 22, 2009 @ 10:34 am

    marie-LUCIE to Jesus S:

    Even in the modern period, hunting can still play a significant part in the lives of rural speakers, and a bear in the vicinity of flocks of sheep is a danger (consider the opposition to the reintroduction of bears in the Pyrenees where a century ago farmers and sheepherders were still also bear-hunters).

    PIE did not arise from nothing. The lack of a single word for 'bear' in the various IE families does not mean that speakers abandoned the original word reconstructed as above: since that word appears to be a verbal derivative, it too must be a nickname or euphemism. The reason it is reconstructed for PIE in general is that it is more widely attested than the other nicknames, which are limited to Germanic and Slavic, but it may never have been used in those families.

    I disagree that "laryngeal" does not refer to any type of articulation: h and the glottal stop have definite phonetic descriptions and are used as phonemes in many languages. The glottal stop (although not written in normal spelling) is a phoneme for instance in German where it precedes initial vowels (including in derived and compound words). The alliteration schemes in Old English poetry show that the so-called initial vowels must have been preceded by the glottal stop, which was the alliterating consonant.

    Linguists who use h1, h2, h3 for PIE are being extremely cautious about exactly what the articulation was, but more accurate descriptions might be possible in the future, in which case the proper symbols will be used.

  10. David Marjanović said,

    February 22, 2009 @ 12:08 pm

    Ardagastus, if you read this, please check out the bottom of this thread!

    On a seperate note, is there anywhere i can find out how to pronounce all these consonant blends? I don't mean just hearing them, i mean instructions on how to make my mouth do that.

    Wikipedia. No, seriously.

    the r is syllabic (as in Czech and some other languages)

    Most relevantly Sanskṛt. That's what the underdot or -circle in ṛkṣa means (…not the dot under the s, though, which indicates the retroflex consonant [ʂ]).

    Euphemisms for "bear" are well attested from 19th-century Europe, from Wales to Serbia at least. Often kinship terms like "aunt" or "mother-in-law" were used.

    Why have a euphemism for "bear"? Because if you call the bear, he comes. Just like the God-be-with-us, the In-the-flesh One (German: der Gottseibeiuns, der Leibhaftige).

    One of the things I said then was that the term "laryngeal" is completely irrelevant in modern phonology. It does not refer to any point of articulation, like "bilabial", "palatal", etc.

    It was once used for "glottal". Russian Caucasianists still talk about "laryngeals" (glottals) and "emphatic laryngeals" (epiglottals).

    But several people on the very thread you link to have already told you that IEists only still use this term for historical reasons! *h2 and *h3 were almost certainly velar, uvular, or pharyngeal, and *h1 glottal.

    The glottal stop (although not written in normal spelling) is a phoneme for instance in German where it precedes initial vowels (including in derived and compound words).

    I disagree. In northern German, a glottal stop is predictably inserted in front of every stressed syllable that would otherwise begin with a vowel (example I heard on TV: Asteroiden und Kometen [ˌʔastɐʀoˈʔiːdn̩ʊndˌkoˈmeːtn̩]). In southern (even Standard) German, the glottal stop is only used in front of otherwise vowel-initial utterances, so when a northern German says Naomi, I hear the sentence Na — Omi! "Now — granny!", because I'd never put a glottal stop inside a word.

    The alliteration schemes in Old English poetry show that the so-called initial vowels must have been preceded by the glottal stop, which was the alliterating consonant.

    Now that is interesting!

  11. David Marjanović said,

    February 22, 2009 @ 12:12 pm

    Oh. The long IPA string in my comment contains [n̩] twice, which is not displayed correctly on my computer (after I submit and in the preview, that is — in the text field it works just fine!) It's an n with a vertical line under it and means a syllabic [n].

    [(myl) It looks OK on my machine now (Firefox, Ubuntu):

    Maybe it depends on what fonts exist in the viewing environment, or what characters are in the font?]

  12. David Marjanović said,

    February 22, 2009 @ 12:31 pm

    Certainly. Internet Explorer 7 for Windows XP speaking here, and the font I see is distinctively Verdana.

    Sometimes fonts with the same name have different character ranges. For example, Arial and Times New Roman for Windows Vista contain lots of Unicode glyphs, such as IPA symbols, which the earlier versions all lack — I have to use fonts like Arial Unicode MS, which looks slightly different from Arial.

  13. Jesús Sanchis said,

    February 22, 2009 @ 2:00 pm

    Mr Marjanovic, I am aware that the 'laryngeal' is used by IEists as a traditional term, that's why I said it is irrelevant in MODERN phonology, if by "modern phonology" we understand things like IPA. If h1, h2 and h3 had clear places of articulation, they would not be called "laryngeals". They would be called "glottals", etc. On the other hand, I wrote my comment as a question to Mr Ringe: I wanted to know the value that he assigns to these reconstructed forms. That is an important question: we all think that reconstructions are more or less necessary, or practical, but can they actually be a description of a human language or are they primarily a a scientific tool?

    The use of euphemistic words to refer to animals is an interesting area in historical linguistics. The motivation behind this phenomenon may find its roots in the Paleolithic, with various linguistic realizations through time. I'm not saying that the various words for "bear" are the definite evidence to prove the earlier dating of PIE, but it they are a remarkable detail, among others.

  14. marie-lucie said,

    February 22, 2009 @ 3:14 pm

    David, my German is not too great and I am not aware of all the subtleties of German dialects, but how would you pronounce, for instance, Beamte or bearbeiten or beenden and similar words? Are you saying that you do not separate the adjacent vowels by a glottal stop? or is your quarrel with my saying that the glottal stop is phonemic in German? I can see the argument from predictability (in some dialects only), but on the other hand it seems to me that in initial or intervocalic position the glottal stop is in contrast with h just as with other consonants.

  15. A Reader said,

    February 22, 2009 @ 3:23 pm

    I am very glad that prof. Ringe has returned with another fascinating piece. I had one question, which I suppose it mainly methodological, concerning the statement:

    Since the Hittite reflex is the only one that's distinctive, it makes sense to base our reconstruction on that

    I can see perhaps basing off of Hittite on the assumption that that form is more archaic, but wouldn't it be more economical to suppose that the Hittite form had metathesized, rather that the other forms? I realize we then don't have a good mechanism for getting from the PIE form to the variant stops/sibilants- but that doesn't exactly seem to be the rationale here either. But why is the distinctiveness of the Hittite form what makes it our base form rather than an anomaly?

    If you get a chance to respond to this, thank you very much! I hope your semester is going well.

    @JS, I couldn't say what Professor Ringe's opinion is, but the Indo-Europeanist Calvert Watkins has this to say on reconstructed languages:

    But the stronger claim, in spite of all the cautionary hedges we may put up, is that these reconstructions are a real model, constructed to the best of our ability, of how we think certain people talked at a remote period before recorded history…
    {How to Kill a Dragon, Oxford University Press, 1995}

    Of course, there are all the 'hedges' which can't be ignored in this sort of discussion- considerations of levelling irregularities, dialectal/register variation, and incompleteness of data all matter. But Watkin's point still stands.

    David's evidence, if nothing else, should have made it clear that a simplistic equation of 'euphemistic terms for bear [or another animal] = paleolithic cultural trait' is so far from the truth that it is essentially irrelevant as evidence in dating PIE. Anyway, David Anthony's argument posits PIE culture as being what could be called a chiefdom society- even if we had no ethnographic evidence at all and had to reason it out, why would euphemistication of animal names be out of place is such a society? As one archaeologist put it, some animals (especially physically powerful ones) are not just good to eat, but are good to think. If the Assyrians were troubled by ravening lions who were worthy of royal hunts, why wouldn't the PIE tribes be able to develop superstitions about the name of a powerful predator?

  16. Jesús Sanchis said,

    February 22, 2009 @ 5:18 pm

    Dear "Reader", you're right when you say "why wouldn't the PIE tribes be able to develop superstitions about the name of a powerful predator?" I think what we have here is a question of probabilities: as far as I know, the religious thoughts of pastorialist or farming societies are quite different from those of hunter-gatherers, even though some elements of ancient forms of religion can be traced in later times. The words for "bear" are just an example among many that point to a Paleolithic context for 'PIE'. Another one would be kinship terminology. You say "why would euphemistication of animal names be out of place is such a society?" That's a question that should be answered by anthropologists.

    As for Watkins' definition, I would say that the phrase "a real model, constructed to the best of our ability" is curious. In what sense is this model "real"? Why should "the best of our ability" be enough to build a "real" model?

  17. A Reader said,

    February 22, 2009 @ 6:14 pm

    @JS, I am studying anthropology, and my comments weren't without basis. The relationship between social structures, subsistence strategies, and religion is very complicated, and arguments based on them need careful consideration of many kinds of evidence (David Anthony's book is just such a consideration for PIE, a very well done and carefully considered piece of research)- not simplistic claims about social typologies (these claims are reminiscent of the Victorian 'armchair anthropologists' who made many broad claims about the way societies, especially 'primitive societies' worked- their claims weren't actually all totally wrong, it just turns out that reality is much, much more complicated than Tylor or Morgan's social categories imply). Most features of the posited PIE religion fit well with the kinds of religious features we have come to expect, through ethnographic research, from societies with similar subsistence patterns. But the range of social configurations is very great, and it would be hard to posit a religion so radically unusual for what we expect that it would actually be useful evidence (not that what we have here is even remotely approaching the radical).

    And as marie-lucie pointed out, we actually have a very solid ethnographic analogy for the animal taboos in what is certainly not a gatherer-hunter society. This reduces the claim that 'The words for "bear" are just an example among many that point to a Paleolithic context for 'PIE" from baseless (in that there was no good reason to think that animal taboos are somehow unique to 'paleolithic' societies) to false (in that we have real evidence saying the exact opposite).

    .

    You probably use kilograms a great deal in your everyday life, but it is practically certain that you have never dealt with a perfect, or 'really real', kilogram. That is an approximation (a very good one, much of the time), but one we find very useful. PIE is less good than the kilogram, but it too is an approximation of reality, and a very useful one at that. The fact that it isn't perfect doesn't destroy its utility, or even its reality, any more than the fact that a 1 kilo weight is perhaps really 1.00001 'kilograms' means that we should throw away your weight as imperfect and unreal (or perhaps more true to the analogy, it doesn't mean that we should abandon the metric system just because the model measurements don't precisely reflect the actual measurements).

    Watkin's operative word is model- the model approximates a real language, without actually being that language. Does the German grammar I like to reference precisely reflect the German language in every way? No, it too is a model, an approximation of the actual language (should I throw away my German grammar because it isn't perfect?). The only difference the too grammars is the methodology for building each model, and completeness of each model- but reality is still there, and, at least for Watkins, we couldn't do comparative linguistics without it.

  18. AJD said,

    February 22, 2009 @ 8:27 pm

    "Reader":

    I think what Don meant by "distinctive" is that the Hittite "tk" is not found in words outside the set of those that show the "thorn cluster" correspondence. That is, for instance, if you followed Sanskrit and reconstructed *ks, you'd have to explain why there are other instances of reconstructed PIE *ks that don't correspond to Hittite "tk" and Greek "kt" and so on; if you reconstruct *tk you don't have this problem.

  19. dr pepper said,

    February 22, 2009 @ 10:27 pm

    @David Marjanović

    You recommended i check Wikipedia for pronunciation instructions. I went
    and looked up "phonics" and "pronunciation", but all i could find was the technical specs for the creation of individual sounds, which doesn't help.

    To be more specific, i can say "k" and "th", but i can't say "cth" as in "cthomic" i can make the vowel sound between the two consonants really short, but i can't make a true consonant blend. Same thing with sounds like "pt", "ps", "pn", "kn" (though strangely i have no trouble with "gn"), "ks" and so forth.

    Reading about stops and such doesn't give me any clues about that. Is there some other keyword i could use?

  20. Jesús Sanchis said,

    February 23, 2009 @ 5:09 am

    Dear "Reader", when I say "The words for "bear" are just an example among many that point to a Paleolithic context for 'PIE" I mean exactly that. I am convinced that 'PIE' must be traced back to Paleolithic times, because there's a lot of evidence suggesting it, from archaeology, population genetics, anthropology, dialectology, etc.). I agree that analysing the religious elements of ancient societies, or even trying to reconstruct them, is obviously a very complex task. You talk about 'armchair anthropologists', and I'm afraid there have been, and possible there are still, a lot of 'armchair linguists', busy with their laryngelas and their laws of language change. And we also have some 'armchair linguists' talking about recursion and Universal Grammar.

    About your German grammar book, you say: "it too is a model, an approximation of the actual language". What is the "actual language"? Your grammar of German describes a Standardized form of language, based on a dialect that has gained a high social status. It is not an approximation to the 'German language'. It is the opposite process: speakers of 'German' have adapted their dialects to the standard form in the last centuries, therefore creating this image of 'the language', which is typical of modern, highly uniform societies. Most of the reconstructed PIE is based on written standards of 'languages' belonging to the 'IE family'. The reconstructed models, especially the ones that include laryngeals as a last resort towards theoretical 'perfection', are very unlikely to be a "real" language. At most, they might be considered a scientific tool.

  21. marie-lucie said,

    February 23, 2009 @ 8:13 am

    dr. pepper,

    It is hard for English speakers to pronounce clusters at the beginning of words that they have no trouble pronouncing in the middle and sometimes at the end. So you can train yourself by finding cases where you have no problems (as in caps</b or capsule for the /ps/ cluster, or apnea for the /pn/ cluster), silently mouthing the part before the cluster and only sounding out the cluster itself and the rest of the word. Sometimes you will not find the relevant cluster in a single word but the cluster will occur in words that are adjacent to each other (which you may have to invent), ex. rock theatre for the cth cluster you mention. Try it!

  22. A Reader said,

    February 23, 2009 @ 10:56 am

    @JS, you are wrong about the German grammar not being a model. The dialect of High German now spoken fairly standardly in much of Germany is indeed the product of various social forces (going back at least to Luther's bible and the Reformation)- but the abstraction of the rules of that language contained in my grammar are not the language. No one is speaking my grammar. As you yourself said, it 'describes' High German, it itself isn't 'High German'.

    PIE is the same way. Of course it 'isn't' the language (no more than my German grammar 'is' that language), but it is an abstraction, approximation, or representation of something once actually spoken.

    As for 'armchair linguists', that terms doesn't quite apply so well. 'Armchair anthropology' refers to the practice of making theories without looking at the evidence (in their case, relying entirely on secondhand accounts of events they could have observed first hand)- Don Ringe in his series has presented a very careful account based on very detailed evidence which was as direct as possible. This is not 'armchair' research, as Ringe has clearly done the legwork and is basing his linguistics off of primary texts. Other linguists are doing very solid field or laboratory research- perhaps there are some 'armchair linguists' out there, but I doubt there are many.

    This is not the place to debate Paleo-Continuity Theory (which I have actually looked at, and feel more than justified in rejecting as sloppily argued from weak evidence)- we've strayed far enough as is.

  23. marie-lucie said,

    February 23, 2009 @ 12:07 pm

    sorry about my mistake above:

    caps</b : I forgot to cancel the Bold after caps.

  24. A Reader said,

    February 23, 2009 @ 2:44 pm

    I'd like to soften my critique of PCT just a bit: there are some critiques contained in it which may well be valid, but I am not convinced by the argument as it is currently presented. It is at odds with solid linguistic evidence, and doesn't seem to a responsible interpretation of archaeological evidence.

  25. Jesús Sanchis said,

    February 23, 2009 @ 4:02 pm

    To "Reader":

    Please notice that I talked about German as a 'language' from a diachronic perspective. I agree with you when you when you say "the abstraction of the rules of that language contained in my grammar are not the language".

    I could discuss some of the things that you mention in your comments, but it's true that this is not the right place for yet another discussion about the PCT, and it's not my intention to start one now. Many arguments in favour of and against it were raised in a previuos post:The Linguistic Diversity of Aboriginal Europe (http://languagelog.ldc.upenn.edu/nll/?p=980), and it's not necessary to repeat them here.

  26. David Marjanović said,

    February 23, 2009 @ 5:59 pm

    If h1, h2 and h3 had clear places of articulation, they would not be called "laryngeals".

    Tradition can go all the way to laziness, you see. h1 almost certainly was a glottal, but people don't feel it's necessary to state that — they prefer, it seems, talking about its behavior.

    As I mentioned, it has not been possible to tell if the place of articulation of h2 and h3 was velar, uvular, or pharyngeal. However, both of them most likely had the same place of articulation, whichever of the three it was.

    I am not aware of all the subtleties of German dialects

    This difference extends into the standard language. TV/radio newsreaders from northern and central Germany insert glottal stops in front of all stressed syllables, TV/radio newsreaders from Austria, Switzerland, and (to a lesser extent) southern Germany only use utterance-initial ones.

    how would you pronounce, for instance, Beamte or bearbeiten or beenden and similar words? Are you saying that you do not separate the adjacent vowels by a glottal stop?

    Exactly.

    or is your quarrel with my saying that the glottal stop is phonemic in German?

    Yes, because I think it's predictable in all varieties, so that its presence vs absence is never used to distinguish meanings.

    on the other hand it seems to me that in initial or intervocalic position the glottal stop is in contrast with h just as with other consonants.

    Sure, but one could just as easily say that the absence of a (phonemic) consonant is in contrast with the presence of one — just like in English or French.

    I think what we have here is a question of probabilities: as far as I know, the religious thoughts of pastorialist or farming societies are quite different from those of hunter-gatherers, even though some elements of ancient forms of religion can be traced in later times. The words for "bear" are just an example among many that point to a Paleolithic context for 'PIE'. Another one would be kinship terminology.

    Didn't I just tell you that euphemisms for "bear" are recorded from all over 19th-century Europe? Maybe I didn't make clear enough that these euphemisms were in actual use at that time.

    To be more specific, i can say "k" and "th", but i can't say "cth" as in "cthomic" i can make the vowel sound between the two consonants really short, but i can't make a true consonant blend. Same thing with sounds like "pt", "ps", "pn", "kn" (though strangely i have no trouble with "gn"), "ks" and so forth.

    As usual in English, you aspirate your /p t k/, and to make clusters that begin with an aspirated consonant is difficult or impossible, because the aspiration is automatically released into a vowel.* /g/ is not aspirated, and that's why you have no problem with /gn/.

    However, /p t k/ are not aspirated at the ends of words in English, and this is why Marie-Lucie's examples will most likely work for you.

    * And this does make me wonder how the ancient Greeks pronounced their clusters — there are lots of words like chthon, ophthalmos and so on. Probably it was just an orthographic convention, though; I'm told graffiti tend not to obey it.

    grammar of German describes a Standardized form of language, based on a dialect that has gained a high social status.

    Not even. From the start (a bit before Luther), what later became Standard German was an eclectic and (grammatically) especially conservative mixture of High and Middle German dialects. It is more artificial than most other standard languages.

    Most of the reconstructed PIE is based on written standards of 'languages' belonging to the 'IE family'.

    I don't think so. I've often found references to a wide variety of dialects in what little historical-linguistics literature I've read. It would also surprise me if all historical linguists had managed to completely overlook the literature that exists on the words of some Austrian dialects which can be found nowhere else in West Germanic, but are known from Gothic.

    ——————

    How serious, as a representation of really once spoken languages, should reconstructed proto-languages be taken? Right here in his second post, Prof. Ringe talked about that.

  27. David Marjanović said,

    February 23, 2009 @ 6:10 pm

    Because that post is so long, I copy the relevant paragraph:

    This raises a methodological point that we can no longer avoid. Is there any difference between a word which is reconstructable for a protolanguage and a word which spread from dialect to dialect of the protolanguage as it was breaking up? As usual, it depends on the individual case. If the real-world separation of the daughters was genuinely abrupt—that is, one group picked up and moved within a generation or so, and subsequent contacts were infrequent and brief—then there is a clear difference between the two scenarios. But most disintegrations of speech communities don't happen like that; dialects remain in contact as they diverge, continuing to trade linguistic material until some event finally makes them lose touch altogether. (The best discussion of these processes is Ross 1997.) In such cases the "protolanguage" which we reconstruct is most unlikely to correspond to a single, completely uniform dialect that existed in the real world before its speaking population became large enough to exhibit significant linguistic diversity; it almost inevitably corresponds to a dialectally diversified speech community, still unified but no longer uniform, simply because we can't tell the difference between words and grammatical forms which had been in the language for generations and those which had arrived very recently. It is also likely that our reconstruction will be temporally "out of focus", including some inherited words and forms which were no longer characteristic of all the dialects and some new words and forms which were still spread­ing from dialect to dialect. There are good reasons to suspect that our reconstruction of PIE is like that.

  28. marie-lucie said,

    February 23, 2009 @ 6:21 pm

    David M:

    – it seems to me that in initial or intervocalic position the glottal stop is in contrast with h just as with other consonants.

    – Sure, but one could just as easily say that the absence of a (phonemic) consonant is in contrast with the presence of one — just like in English or French.

    In fact, in standard phonemic theory (eg 50's-60's) that was one part of the definition of a phoneme: it can contrast with 0, as in cat contrasting not only with hat and pat but also with at.

  29. David Marjanović said,

    February 23, 2009 @ 6:58 pm

    But the glottal stop never contrasts with 0.

    You'll find such a contrast proposed in some — not all — of the relevant publications (always implicitly for northern German only, and only word-internally), but all of these are, as far as I can tell, stress contrasts. Even though the orthography hides it, stress is already phonemic in German; as an example where nobody has a glottal stop, I can offer [ˈ]umgehen "to haunt" (of a ghost, intransitive; ich gehe um) vs um[ˈ]gehen "to circumvent" (ich umgehe).

  30. Jesús Sanchis said,

    February 23, 2009 @ 7:07 pm

    Marjanovic, thank you for quoting from Ringe's previous post. I had forgotten that he had already dealt with this issue. He describes some of the logical limitations of reconstructed PIE, but it seems that, in his view, these are minor issues. He doesn't mention the fact that, in many aspects, reconstructed PIE looks very little like a natural language, or at least like any of its 'descendants'. Let's see another example from Ringe's last text:

    – " *ń̥-dhgwhi-tos 'imperishable'. "

    Don't you get the feeling that, at some point, the reconstruction of PIE becomes something a bit too artificial?

  31. marie-lucie said,

    February 23, 2009 @ 8:23 pm

    David M: But the glottal stop never contrasts with 0.

    I guess I did not put it very well: I did not mean that a phoneme has to contrast with 0, only that that is the case in many languages.

    Jesus S: from Ringe's last text: – " *ń̥-dhgwhi-tos 'imperishable'. "

    Don't you get the feeling that, at some point, the reconstruction of PIE becomes something a bit too artificial?

    The artificiality is largely due to the transcription, especially when the symbols are not properly written as they would be in handwriting or specialized typesetting. For instance, the sequence -dhgwh- is not a sequence of 5 sounds d-h-g-w-h but only two, the aspirate dh and the labialized aspirate gwh. In a proper transcription the h and w would not be written consecutively in the same size as the basic consonants but small and raised in order to indicate that they belong to the same sound as the consonant written just before. But other transcriptions, not just that of PIE, look equally strange: for instance the consonant sequences in words like ichthys or chthonian or diphthong or ophthalmology, derived from Greek, where the various Ch-type sequences are the Latin alphabet transcriptions of individual Greek letters.

    The sound dh is found for instance in the name Buddha where the d and the dh are two distinct sounds. That there is no vowel between the consonants dh and gwh is due to vowel loss, a frequent process between two root consonants when there is a vowel following (such as i in the example)(this process is not peculiar to PIE but widely attested in the patterns of word-formation in many languages). Writing the prefixes and suffixes separately makes the overall structure clear but also contributes to the strange appearance of the words.

  32. Jesús Sanchis said,

    February 24, 2009 @ 3:31 am

    Marie-Lucie, even if we write the reconstructed PIE phonetic symbols more accurately, the whole picture still seems too artificial in my opinion. Maybe something as speculative as PIE should be expressed in a different way.

    Marjanovic, I think there is a lot of anthropological evidence suggesting the possibility that PIE emerged in a Paleolithic framework. The name of the bear is just an example. You seem to reject this possibility straight away, without even considering that it would be a good idea to go further into this material and analyse it more carefully. The funny thing is that you very readily accept something as baroque and artificial as the laryngeal theory for PIE. I think that, in a scientific context, it is advisable to doubt things more.

  33. marie-lucie said,

    February 24, 2009 @ 9:16 am

    Jesus S,

    When written with the proper symbols and without the hyphens and stress marks, the reconstructed word you gave as an example is not more baroque than the word "imperishable" done in close phonetic transcription with the prefix and suffix separated with hyphens.

    As for the laryngeal theory being "baroque and artificial", perhaps you have a better suggestion: science progresses as new theories are proposed when the old ones start showing signs of wear, and the best new theory eventually wins (although not necessarily without a fight). Thus far the laryngeal theory appears to be the best one. If you have something better to propose, then do so. But any new theory must not only explain what the old one did, but do it at least as well AND also account for more phenomena (See Kuhn, The structure of scientific revolutions). It is dicy to propose a new theory without being thoroughly familiar with the arguments in favour of the old one.

    PIE and paleolithic: Languages do not spring up ready-forned, but evolve. Any language we know has evolved, and represents a stage in a long continuity. Surely there must have been a PPIE and even a PPPIE, etc, but as far as I know the reconstructed PIE vocabularly appears to correlate with a Neolithic lifestyle. But elements of previous lifestyles can subsist, eg as I wrote earlier, rural people often hunt even if their main livelihood comes from cultivation and animal raising. Elements of old beliefs also often subsist besides more organized religion: bear taboos documented in modern Europe (see some previous messages) are a case in point and do not mean that the cultures using them are still pursuing a paleolithic lifestyle in general, or that their languages can be traced further back than PIE itself.

  34. Jesús Sanchis said,

    February 24, 2009 @ 11:31 am

    Marie-Lucie, you have your view on PIE and I have mine. I could write a very long comment with criticism of traditional IE theories, etc., to illustrate my points, but I won't do it now, because, as I said earlier, I did it already in the comments to the first of Ringe's articles in Language Log (there were more than 160 comments for that post).

    In any case, I find these discussions quite enlightening, and I would like to thank Mr Liberman for this initiative, and for his openmindedness.

  35. David Marjanović said,

    February 24, 2009 @ 7:12 pm

    Marie-Lucie, even if we write the reconstructed PIE phonetic symbols more accurately, the whole picture still seems too artificial in my opinion.

    Why? Voiced aspirated plosives are common in Sanskrit and its modern descendants; syllabic nasals are common in English and German; word-initial syllabic nasals are common in various languages in central Africa (take N'djamena, the capital of Chad, or Nnamdi Azikiwe, the first president of Nigeria).

    Also, keep in mind that the transcription used for PIE is phonemic, not phonetic. It doesn't take a stand on whether *ṇ́dʰgʷʰitos was pronounced [ˈn̩dʱgʷʱitɔs], [ˈəndʱəgʷʱitɔs], or something in between, or something similar; it only tells us that there were just two phonemic vowels in this word. In fact, the presence of a (non-phonemic!) vowel at the beginning of the word is suggested by the fact that the prefix *n-̣ is the ancestor of the Greek a- and an-. But, for the record, I am capable of pronouncing [ˈn̩dʱgʷʱitɔs], if we kindly assume that I get those pesky voiced aspirates right.

    I also still don't understand what you consider "baroque and artificial" about the "laryngeal" "theory" (yes, two sets of scare quotes). Yes, IEists don't agree on the exact place of articulation of *h2, but they don't agree on the exact manner of voicing of *d either!

    Marjanovic, I think there is a lot of anthropological evidence suggesting the possibility that PIE emerged in a Paleolithic framework. The name of the bear is just an example. You seem to reject this possibility straight away

    In this thread I have only been rejecting the possibility that the name of the bear is evidence for a Paleolithic age* of PIE. Whether there is other evidence for this hypothesis is irrelevant here, and I neither will nor need to discuss it in this thread.

    * By PIE, I strictly mean the stage of the language at the split between Anatolian and the rest = the most recent common ancestor of the attested IE languages.

  36. Jesús Sanchis said,

    February 25, 2009 @ 3:46 am

    Reconstructed PIE is full of exceptional things: a very limited list of vowels, a set of phantom phonemes called "laryngeals", etc. I know that it is always possible to find some language or another with this vowel set or with this initial consonant, etc., etc., but why should PIE be so exceptional, or so unnatural? We know that there are NBA basketball players who are around 1.75m tall, or even shorter. One of them, Nate Robinson, even won the Slam Dunk contest this year. Now, would we expect an NBA team with four or five 1.75-metre-tall players? It's not impossible, but, is it likely?

    Many linguists who reconstruct PIE think that there wasn't an /a/ phoneme in this protolanguage. Maybe this idea fits their 'laboratory' work, but does it really make sense to propose something like this? Or: how should something like this be proposed? I'm not saying that all reconstructions are necessarily wrong, but I would like to strees that they're only relatively significant.

  37. marie-lucie said,

    February 25, 2009 @ 9:03 am

    Jesus S,

    PIE is not "perfect", and there is probably room for improvement as more detailed work is done, but thus far it is the best thing historical linguists have been able to come up with.

    The "exceptional things" you mention are not that exceptional. In addition to the ones David M mentioned:

    – a very limited set of vowels: this is typical also of languages of the Caucasus, which have lots of consonants but very few distinctive vowels (it is typical for vowels and consonants to influence each other).

    – laryngeals are not "phantom phonemes", it is just that the exact pronunciation is not definitely known at the moment. The laryngeal theory was derived from otherwise unexplained vowel alternations. It was long disputed but was vindicated overall when h-type sounds were found to exist in Hittite in just the right places.

    As I mentioned earlier, before criticizing a theory or proposing a new one, one should be familiar with the reasons the theory was proposed in the first place.

  38. David Marjanović said,

    February 25, 2009 @ 7:46 pm

    Reconstructed PIE is full of exceptional things: a very limited list of vowels,

    No, the only odd feature of the vowel system is the (near-?)absence of /a/ (the phoneme, mind you, not the sound). Granted, that's a very odd feature, though. But I can't see how one could avoid it.

    Bomhard has proposed that the /e/-/o/ ablaut was originally an /e/-/a/ or /ə/-/a/ ablaut. I don't know how testable this is.

    Incidentally, Arapaho has been described as having [a] only as an allophone of /o/, though probably it makes more sense to interpret it the other way around.

    a set of phantom phonemes called "laryngeals"

    Why do you keep insisting that the "laryngeals" are so odd just because the place of articulation of two of them can't be reconstructed precisely? Why?

    I'll go ahead and say that PIE is probably less unusual in this respect than German or Tatar with their distinction between /h/ and /x/, and in any case had a smaller "laryngeal" inventory than Arabic (which has six instead of three or maybe four).

    It was long disputed but was vindicated overall when h-type sounds were found to exist in Hittite in just the right places.

    And not just Hittite (and the other Anatolian languages, including Lycian, which used the Greek letter χ for its descendant of *h2). Armenian preserves word-initial *h2 and *h3 as /h/ (hoviw "shepherd", from older howi-pa, of which the first part comes from PIE *h3owi-), and Greek preserves the word-initial cluster *h1y- as /h/ (in spite of turning word-initial *y- into whatever ζ was really pronounced as, probably /d͡z/).

    a very limited set of vowels

    I wouldn't call */e eː o oː i u/ "very limited" — the West Caucasian languages all have two or three vowel phonemes… the globally most common number is five, and even without */a/, let alone */aː/, PIE had six.

    (There is evidence from Hittite that */i/ and */u/, distinct from */j/ ( = *y) and */w/, existed.)

  39. Jesús Sanchis said,

    February 26, 2009 @ 2:34 am

    Thank you for the information. When I said "a very limited set of vowels" I meant some models which have just two vowels (e, o). By the way, you include two diphtonhgs in your list (ei, oi). You must assume that this is not common practice in phonology. When I say "vowels", or when I say "a limited list of vowels", I'm not thinking of diphthongs, obviously. If we count diphtongs, then the count is different.

    Personally, I think the methodology and theoretical background that lead to laryngeals and the 'disappeance' of /a/ must be challenged, and in fact, as I mentioned in a previous discussion, it has already been challenged (you can take a look at my blog for some more information on this). Marie-Lucie wrote: "there is probably room for improvement (…) thus far it is the best thing historical linguists have been able to come up with". This is true, but I think it's not only a question of improving the current methodology. Maybe some major shift is needed.

  40. David Marjanović said,

    February 26, 2009 @ 2:30 pm

    I meant some models which have just two vowels (e, o).

    Such models forget to count the phonemic length and haven't incorporated the recent (possibly controversial, I don't know — it's Bomhard after all) research that is claimed to show the existence of /i/ and /u/.

    By the way, you include two diphtonhgs in your list (ei, oi).

    No, I didn't. Read my comment again. I counted six vowel phonemes: */e/, */eː/, */o/, */oː/, */i/, */u/.

    The reason I didn't include the diphthongs is that — as far as I know, and I'm not an IEist — they behave just like sequences of vowels + *y: in the zero-grade, the *e/*o disappears, and only *i — the syllabic allophone of *y — stays behind.

    You must assume that this is not common practice in phonology.

    That depends. For example, it doesn't make sense to consider the English /ɑɪ̯/ a sequence of /ɑ/ and /ɪ/, does it? Or take the German /aʊ̯/ a sequence of /a/ and /ʊ/, because /ʊ/ doesn't occur behind any other vowel than /a/ — there is no other diphthong that ends in [ʊ̯] in the whole language.

    Have you written a blog post specifically on PIE */a/? If so, I'd appreciate a link.

  41. David Marjanović said,

    February 26, 2009 @ 2:32 pm

    And you still haven't explained what's so scary about the so-called "laryngeals".

  42. Jesús Sanchis said,

    February 26, 2009 @ 3:23 pm

    Mr Marjanovic, thank you very much for your comments. This discussion about PIE laryngeals is very interesting but we must agree it's quite off-topic here. In my blog (Language Continuity: http://languagecontinuity.blogspot.com/) I have written a couple of posts about lE laryngeals, as you can see if you click on the "Phonology" Label: http://languagecontinuity.blogspot.com/search/label/Phonology.

    In one of those posts I mentioned an article about the missing /a/ phoneme in PIE, written by Xaverio Ballester. This article is written in Spanish. Unfortunately, it's not available on-line.

  43. marie-lucie said,

    February 26, 2009 @ 8:34 pm

    Jesus S.,

    Thank you for mentioning your blog and the continuity people. I have had a look and followed some links, and although some of the articles available on line on the PCT website look interesting (the ones with more localized topics) I am afraid that what I can gather of the work of Drs. Alinei and Ballester from the titles (and I can read both Italian and Spanish) does not inspire confidence in their work. For instance, for someone who is concerned to conflate the data from anthropology, archaeology and linguistics, Dr. Alinei's choice of putting together Etruscan and Hungarian is at the very least eyebrow-raising.

    You point to the Wikipedia charts of consonants in the PIE languages, contrasting the simpler, obvious list of Bopp with the more complex reconstructions of modern specialists, but let me assure you that historical linguists do not make suggestions for more complex reconstructions just for the fun of it or in order to confuse people: it is because the simpler forms are not enough to tie together all the languages of the family. Similarly, earlier Indo-Europeanists used a lot of *a's in their reconstructions, but that was because they assumed that Sanskrit must represent the oldest form of the original language, but later linguists realized that that was not the case: from other languages you could predict the Sanskrit vowel, but not the opposite. Finally, a lot of the work of PIE reconstruction (which is still undergoing some modifications) was done in the 19th century, long before "structuralism". It is true that Saussure, who proposed the laryngeal theory, went on to become one of the sources of structuralist theory, but that was decades after he had proposed the laryngeals in his twenties, so the laryngeals were not a product of structuralism.

    I note that you seem to be partial to the theories of R.M.Dixon (The rise and fall of languages) but Dixon is not a historical linguist, and his theories are incompatible with the attested facts of sociolinguistics and of what is known about historical conditions of change.

  44. Bryn LaFollette said,

    February 26, 2009 @ 8:37 pm

    As I've mentioned in earlier posts on this subject, there are plenty of real-world natural languages that resemble the reconstructed phonology of PIE. One only needs to familiarize themselves with languages of the Caucasus Mountains (which incidentally contain extensive use of Pharyngeals both voiced and voiceless and even labialized) or the Pacific North-West.

    Here are a few examples from Kwakiutl:
    (representations are orthographic)
    m'kwla – 'moon'
    yudxw – 'three'
    duqwla – 'see'
    ts'sqwana – 'bird'
    dxluł – 'owl'

    Here's an example of a sentence from Abkhaz:

    sara' a-píYâ's a-sap'â'n s-xarp a-la-l-sâ-r-dZYdZYa'-jt'

    "I got the woman to wash my shirt with (the) soap"

    (The hyphens represent morpheme boundaries, and should serve to demonstrate the polysynthetic nature of the language.)

    Basically, limiting one's conception of what looks "natural" for language to what typifies modern European languages is not very realistic. There is an incredible diversity in the phonology and morphology of the languages of the world today, and there's no evidence to support the claim that the same diversity wasn't possible in the past. As for the degree to which the characteristc of a language can change, one needs only to look to the reconstructions of the Sino-Tibetan family for similar examples of an amazing degree of change occurring between the ancestral and (some branches of) the descendant languages (take a look at A Handbook of Old Chinese Phonology by William H. Baxter)

  45. Bryn LaFollette said,

    February 26, 2009 @ 8:56 pm

    Oops, I didn't mean to sound contrary to the prior mention of Caucasus Languages. I just meant to give examples that demonstrated the shape of them as an example. I ought to refresh my browser more regularly.

  46. marie-lucie said,

    February 26, 2009 @ 9:43 pm

    Bryn,

    And to the uninitiated like myself, the transliteration of Tibetan words into the Latin alphabet looks exceedingly weird (although since the language is alive and well, it must be just because I have not taken the trouble of finding out how to pronounce it), for instance these ancient place-names:

    bDe-legs-gling
    gNya'-lam
    mTsho-sngon
    lHun-grub-rdzong
    rGyal-bzang

    and many similar ones.

    Among the languages of the Pacific Northwest, the Nuxalk language (also known as Bella Coola), a member of the Salishan family (which is very consonant-heavy) can string so many consonants together that many words have no vowels at all, for instance:

    t'xt "stone, rock" (x is like German (a)ch but longer)
    tl'lh "dry" (lh= L, a voiceless lateral fricative)
    ts'xlh "true"
    nmnmk' "animal" (the first m is syllabic)
    plhtknknlhp "bitter cherry tree" (the first n is syllabic)

    Many other words have only a single vowel next to a large number of consonants:

    kstcwa "to do, make, fix" (c = x as in German (i)ch)
    lhts'tsut "to duck, dodge"
    q'slctu "to tighten'
    q'suslmclhp "to scold somebody"

    Some of the Wakashan languages related to Kwakiutl have even more unusual consonants, as they have pharyngeals (which can be labialized and/or glottalized also) in addition to glottals, uvulars and labio-uvulars.

    Seen in a more global perspective, the PIE reconstructions no longer look so strange.

  47. Jesús Sanchis said,

    February 27, 2009 @ 3:15 am

    Marie-Lucie, I know Alinei's theory about the Etruscan language is controversial but as far as I know it hasn't been refuted yet. In my blog I try to talk about theories and proposals that challenge the linguistic establishment (especially the American, Chomskyan one), because I'm convinced it needs to be challenged in many ways, as some linguists have already shown. Otherwise, I wouldn't publish a blog: I'm not interested in the perpetuation of traditional ideas. You talk about Saussure and structuralism. You can call it "structuralism", "classical grammar", "generative grammar" or "Chomskyan linguistics"; they all share a common characteristic: seeing language primarily as a set of rules, a 'system', a collection of parameters, etc.

    You also wrote the following, referring to me: "you seem to be partial to the theories of R.M.Dixon" and then "his theories are incompatible with the attested facts of sociolinguistics and of what is known about historical conditions of change". What is the next step in your line of thought? Are you going to say that languages evolve at a given speed, which can even be calculated? What I like about Dixon is his idea that language change can happen at different speeds depending on external factors. I also like his critique of geneaological trees. The things he says are just common sense. You say Dixon is not a historical linguist. Do you think his ample experience with Australian languages does not qualify him to say something about language change? What do you prefer, linguists who spend their entire academic lives dealing with mere abstractions they call 'language'?

  48. marie-lucie said,

    February 27, 2009 @ 8:33 am

    Jesus S,

    From what you have written thus far (here and earlier) I understood that you were interested in historical linguistics. I am a historical linguist (though not an Indo-Europeanist), and the Chomskyan paradigm (especially in its present incarnation) has almost nothing to offer to the historical linguist. Historical linguistics is interesting to me precisely because of its link to real life: the present vocabulary and functioning of a language reflects the history and life of its speakers. But there is also a strong formal, systematic component to language (its skeleton, if you like, which is the primary object of linguistics), and therefore to its evolution, and in tracing the history of the sounds of its words that component is particularly important. It is the systematic aspect of sound change that allows competent historical linguists to weed out borrowings from among inherited words (those transmitted in a continuous line), as well as to trace the evolution of sounds and sound systems (yes, systems) over time and to reconstruct both stages of evolution and the probable remote ancestor.

    Dixon's idea that "language change can happen at different speeds depending on external factors" is hardly original. I thought that his book was interesting too the first time I read it, but the problem is that that he follows a geological model, and geology is not the best model for something strongly linked with society. As for his dissatisfaction with genealogical trees, again he is not the only one to have voiced this opinion. Genealogical trees as models useful for certain types of relationships, but yet other models have been proposed, even for the Indo-European languages.

    As for the fact that Alinei's work on Etruscan and Hungarian "has not been refuted yet": as an analogy, let me quote the response of a specialist in Algonquian languages who was asked why Native American specialists had not given a point-by-point reply to the new classification proposed by Greenberg: he said something like "Geographers do not waste their time debating with members of the Flat Earth Society." Perhaps Alinei's work has not been refuted, but more to the point, has it been taken up by reputable historical linguists (not just his colleagues) and (since he emphasizes a multidisciplinary approach) confirmed (or even favorably reviewed) by reputable archeologists and other specialists?

    I would not be replying to you if I did not think you were sincere and dedicated in your attempts, but historical linguistics is not something that can be improvised, or criticized with as little acquaintance with the subject as you seem to have.

  49. Jesús Sanchis said,

    February 27, 2009 @ 9:52 am

    Marie-Lucie, you seem to have blind faith in all the established dogmas of (historical) linguistics, and I'm not even trying to convince you of anything really. The only thing I can say is that I strongly disagree with you. You talk about systems, systematic change, "competent historical linguists"… What can I say? On the other hand, maybe I what I said about Chomskyan linguistics was a bit confusing. I criticise both traditional historical linguistics and Chomskyan linguistics, but I know quite well the clear differences between them and also their subtle similarities.

    You are right in one thing: Alinei's ideas receive little attention by the members of the linguistic establishment, and the lack of complete confirmation or refutation of his theories is partially due to that fact. I don't think his theories are sacred or necessarily true, but I think they're interesting. They offer new views on the study of languages. They are useful at least to prove some wrong assumptions in historical linguisitcs, especially one: the idea that a group of people in the 4th millennium BC started an intercontinental expansion that took them thousands of kilometres away and caused a general and quite dramatic process of language substitution everywhere they went, always in favour of IE languages. What I like about Alinei is that he is providing the right tools to prove that this account of IE expansion is a fairy-tale, and that some assumptions about the history of languages must be changed because they're simply untenable, no matter how "competent" their advocates may seem to be. That's why I like Alinei's ideas and why I've written about them in my blog, because I'm tired of streamline linguistics and linguists who would just ignore Alinei or any other linguists who dear challenge the core of their discipline.

    Finally, you say "historical linguistics is not something that can be improvised, or criticized with as little acquaintance with the subject as you seem to have". I don't see why I should repeat once and again the same arguments. You seem to know very well the doctrine of historical linguistics. Congratulations! Even if many of the things that are researched in linguistics (historical, Chomskyan or any other type) are completely useless, they can at least provide a nice university career!

  50. marie-lucie said,

    February 27, 2009 @ 11:53 am

    Jesus S,

    some wrong assumptions in historical linguisitcs, especially one: the idea that a group of people in the 4th millennium BC started an intercontinental expansion that took them thousands of kilometres away and caused a general and quite dramatic process of language substitution everywhere they went, always in favour of IE languages.

    This idea has nothing to do with historical linguistics itself, it is a proposed scenario for why the languages in question are distributed the way they are and how that situation might have arisen. Different scenarios can be proposed (and have been), without invalidating the relationships between the various languages and the extremely high probability that their resemblances mean that they have a common ancestor.

    As I wrote just earlier, I am not an Indo-Europeanist, and the languages I am considering in a historical perspective are in America. But the methods used in IE linguistics have been proven time and again for a variety of other language groups, without coming with the extra baggage of the IE invasion scenario. A case in point is Austronesian, which includes the Polynesian family, which expanded into hitherto uninhabited islands within the historical period. Another is the recent (just about a year ago) demonstration that the Dene languages in North America are related to the Ieniseian languages (most of them extinct) of Siberia. Most of the Dene populations (in Alaska and Northern Canada) are still hunter-gatherers. If you consider these matters totaly useless, you are entitled to your opinion, but then don't pretend you are interested in historical linguistics. (And as for 'a nice university career', historical linguistics is not the way to go if you really want one in North America).

  51. marie-lucie said,

    February 27, 2009 @ 2:42 pm

    (a nice university career…) The thread "jobs in linguistics" on Feb. 21 did not even list historical linguistics as a specialty in its count of job offers and applicants. Contrast this with the number in Syntax (Chomsky's specialty).

  52. Jesús Sanchis said,

    February 27, 2009 @ 2:43 pm

    The 'Kurganic' theory of PIE exists because there is a linguistic background for it, provided by traditional historical linguistics.

    Please notice that when I said "a nice university career" I was not talking about historical linguistics in particular but about the linguistic 'establishment' in general.

    Many of the conclusions reached by Indo-Europeanists derive from the study of codified, written standards. I don't see what this has to do with Austronesian or Dene languages, apart from some obvious coincidences in the general idea of 'comparing' languages.

    Once more I'll have to repeat this: I agree that the results obtained in traditional historical linguistics may be scientifically useful. Why not? But there is an important difference in the degree of relevance that one adscribes to those results, and also in some of the basic theoretical considerations underlying the methodology.

    But anyway, who am I to talk about historical linguistics if I'm no more than an 'amateur' according to you, Marie-Lucie (or whoever it is that's hiding behind that name)? How dare I doubt the dogma? Well, it's been a pleasure talking to you.

  53. Jesús Sanchis said,

    February 27, 2009 @ 2:52 pm

    Answering your last comment (I hadn't seen it before), I'd like to say that I'm not surprised that historical linguistics, and IE studies in particular, are not seen as interesting fields of study. They need some renovation! (I hope you agree with me on that).

  54. marie-lucie said,

    February 27, 2009 @ 5:31 pm

    Jesus S,

    It does not seem useful to continue this conversation. If you have access to a good university library you might be able to find the following volume: Historical Linguistics 1999, published by Benjamins in 2001 (this is a collection of conference papers) and look for the following article: "On the eve of a new paradigm: the current challenges to comparative linguistics in a Kuhnian perspective."

  55. Jesús Sanchis said,

    February 27, 2009 @ 5:51 pm

    Thank you very much for the information. I'll try to read the article.

  56. David Marjanović said,

    February 27, 2009 @ 6:48 pm

    I have written a couple of posts about lE laryngeals, as you can see if you click on the "Phonology" Label:

    Thank you, I'll check them out as time permits.

    In one of those posts I mentioned an article about the missing /a/ phoneme in PIE, written by Xaverio Ballester.

    I can read scientific articles in Spanish (and Italian), but unfortunately I currently have no way to access anything that isn't online.

    Some of the Wakashan languages related to Kwakiutl have even more unusual consonants, as they have pharyngeals (which can be labialized and/or glottalized also)

    The "glottalized pharyngeal approximant which functions as a stop" of Haida turned out to be an "ordinary" (ha!) epiglottal stop; probably the same holds for what has been described the same way in the geographically close Wakashan languages, but AFAIK nobody has tried to measure the place of articulation or anything.

    Anyway, here are sound files of various pharyngeal and epiglottal consonants from the same East Caucasian language.

    I know Alinei's theory about the Etruscan language is controversial but as far as I know it hasn't been refuted yet.

    It was already refuted before Alinei proposed it. Lack of systematic sound correspondences, deliberate ignorance of the fact that the two known languages that are most closely related to Hungarian are spoken in western Siberia, ignorance of the fact that the history of the Magyar conquest of the Pannonian plain is well attested (as… history), ignorance of the loanword layers in Hungarian, and so on… I'm sorry, it's pseudoscience. It's like the creationist argument that evolution is impossible because random chance as the sole cause is too improbable, which completely ignores the role of selection.

    What do you prefer, linguists who spend their entire academic lives dealing with mere abstractions they call 'language'?

    Do such people even exist?

    you seem to have blind faith in all the established dogmas of (historical) linguistics

    You seem to be blissfully unaware that this is the gravest imaginable insult one can make to a scientist. Well, OK, it's about on par with a charge of having invented data.

    Yes, of course, scientists are only human, too. Scientists regularly overlook data or make other sloppiness mistakes, and regularly interpret data according to the hypotheses they are familiar with even when those hypotheses are not the best possible ones; but "blind faith", "dogma" and "doctrine" go way beyond that.

    The "nice university career" argument, on the other hand, is nothing but an amusing display of ignorance. Do keep in mind that the argument from ignorance is a logical fallacy…

    The 'Kurganic' theory of PIE exists because there is a linguistic background for it, provided by traditional historical linguistics.

    Well, do you remember the last few of Prof. Ringe's posts, during the discussion of which it came out that it's entirely possible that *eḱwos may originally have meant "donkey" rather than "horse", in which case an eastern Anatolian Urheimat (more or less as proposed by Renfrew and Gamq'relidze & Ivanov) suddenly becomes a lot more plausible? I wonder if the Kurgan culture might be associated with something else than PIE, such as Proto-Satem-sensu-stricto or even Proto-Indo-Iranian (though it's probably too early for at least the latter).

    Please notice that when I said "a nice university career" I was not talking about historical linguistics in particular but about the linguistic 'establishment' in general.

    Ha. Getting a job in Romanistics won't become any easier if you proclaim a fervent belief in the Kurgan hypothesis.

    Many of the conclusions reached by Indo-Europeanists derive from the study of codified, written standards.

    You keep repeating this assertion. Have you got any evidence for it?

    Marie-Lucie (or whoever it is that's hiding behind that name)

    That's actually fairly easy to find out.

  57. Jesús Sanchis said,

    February 27, 2009 @ 7:09 pm

    Marjanovic, you raise a lot of issues that I could reply to but I agree with Marie-Lucie that it's no longer useful to continue this conversation. I realize that my sentence about the 'blind faith in the scientific dogma' is very serious when addressed to a scientist, but I didn't know Marie-Lucie was a 'scientist' as you seem to suggest. In any case, I started my sentence with "You seem to", which softens the message in the sense that it's possibly just an impression. And obviously, it was an impression, what else could it be? Now, you 'seem to' know who "Marie-Lucie" really is. What can I say? I don't think the Language Log is the best place for this kind of identity games, or whatever they are. Anyway, I must say I have learned a lot from this discussion and I hope to keep learning. Best Wishes from the Mediterranean!

  58. Mark Liberman said,

    February 27, 2009 @ 7:44 pm

    Jesús Sanchis: "I'm not surprised that historical linguistics, and IE studies in particular, are not seen as interesting fields of study. They need some renovation!"

    In my opinion, historical linguistics is one of the more dynamic areas of linguistics these days.

    New kinds of work abound — the applications of computational phylogeny by many groups, including for example Ringe and Warnow's work as well as Russell Gray's terrific project on Austronesian; the work on dynamic models by Charles Yang, by Martin Nowak's group, or by Patha Niyogi; the work on historical syntax by Tony Kroch and others; the connections between sociolinguistic variation and language change, explored by Bill Labov and his colleagues; the many sorts of typological work that have implications for historical change; and many other lines of work less well known to me, including collaborations with archaeologists and geneticists.

    Meanwhile, excellent work of a more traditional kind continues to challenge old ideas or add to them — Don's recent posts here give you a snapshot of that kind of research in action.

    If there are fewer jobs in historical linguistics than there should be, it's not because of a lack of excellent people or excellent research. Instead, what seems to have happened (at least in North America) is that language departments (where many historical linguists used to be employed) have moved away from linguistic studies of all kinds, from phonetics to metrics; something similar has happened in anthropology; and the field of classics as a whole has shrunk. There never have been all that many jobs in linguistics departments proper — there were essentially no such departments in the U.S. before WW II — and I doubt that the number of historical linguists employed in those departments is significantly lower, overall, than in previous time periods.

    Despite the rather poor job prospects in historical linguistics, we get many heartbreakingly stellar applications every year from students who want to go into that field. We accept as many as we can, consistent with the small size of our graduate program and the competition from other areas of specialization; and we urge them (along with all of our students) to prepare broadly for possible jobs in several different areas.

    There never were very many jobs for historical linguists in the academy. Now there may be a few less, and a larger fraction of the people studying language change are in non-traditional disciplines like computer science or biology. But I believe that the apparent motivation for your animus against historical linguists — the lack of success of PCT — has nothing at all to do with this trend.

  59. marie-lucie said,

    February 27, 2009 @ 8:06 pm

    David M,

    Some of the Wakashan languages related to Kwakiutl have even more unusual consonants, as they have pharyngeals (which can be labialized and/or glottalized also)

    – The "glottalized pharyngeal approximant which functions as a stop" of Haida turned out to be an "ordinary" (ha!) epiglottal stop; probably the same holds for what has been described the same way in the geographically close Wakashan languages, but AFAIK nobody has tried to measure the place of articulation or anything.

    I listened to a presentation a few years ago by two phoneticians (one of them Dr. John Esling of the University of Victoria, B.C.) who used a miniature video camera to see what was going on in the throat of an elderly speaker of one of the Wakashan languages while she pronounced glottalized and other pharyngeals. It was fascinating to see the vocal cords snapping during the process, and there were definitely some strange coarticulating mechanisms going on. As I am not a phonetician I don't remember the technical details, but you should be able to access the paper from John Esling's web page (it was at an LSA conference, within the last 5 years at most). Haida, as far as I know, does not have such complex sounds.

    Jesus S,

    Any linguist worth their salt approaches linguistic problems with a "scientific attitude", hence David's use of the word "scientist". Not all science is done in labs. But one important component is never taking anything for granted and always being aware that one could be wrong.

    As for my name, I am not the only one on Google, but you can find me there if you try.

  60. Jesús Sanchis said,

    February 28, 2009 @ 3:14 am

    Marie-Lucie, after a simple search in google I think I know now you who you are, and I'm glad to have shared this discussion with you. As I said, I'll try to read the article you told me about.

    Mr Liberman, I am sure there isn't a "lack of excellent people or excellent research" in historical linguistics. In fact, I've been trying to use the word 'traditional' in my criticism, to make it clear that I refer to a specific type of approach, the one that has dominated IE studies for a long time, and which has often been seen as a model for the study of other language 'families'. But outside this traditional 'sphere' thre are linguists trying to apply new methods, as can be seen in the examples you have provided. The problem is that the weight of tradition is still felt in many ways. I pointed out some of these problems in a post I wrote about Chritiansen and Kirby's "Language Evfolution" (2003). In this post (The Speed of Change: http://languagecontinuity.blogspot.com/2008/07/speed-of-change.html) I analyse some sentences from some of the contributors in which they use evidence from IE studies. The concept of 'speed of change' and the chronology that they use is the current one, based on what I would call the 'traditional approach'. I'm not saying that these ideas are necessarily wrong, but I think that it would be interesting for the non-historical linguists to realize that the data from historical linguistics might not be as clear as they seem. That's a problem for anyone engaged in multidisciplinary research, or who's trying to incorporate data from another discipline: with a limited capacity of criticism, one has to trust the available data. One also tends to simplify research questions that would otherwise be extremely difficult to deal with, not to say impossible.

    The decline in historical linguistics is not a recent phenomenon. For many years it was the most important area of linguistics, but it was gradually substituted, as the main focus of attention, by what we may call 'synchronic studies'. In the 20th century, notions like 'system' or 'universal grammar' became the main focus of atttention for linguists. Now, is it possible to separate both aspects of language, the 'historical' ones and the 'static' ones? Can we really study language as if it were a photograph, with no movement? Can we study language outside the natural habitat of language: conversation, oral exchange? I personally think that the first step to make historical linguistics 'interesting' is to set it in the appropriate place inside linguistics, not just as a secondary 'appendix'. It requires a 'higher' status. Or a different status.

  61. David Marjanović said,

    February 28, 2009 @ 8:21 am

    I didn't know Marie-Lucie was a 'scientist' as you seem to suggest.

    She did act like one, though.

    I may not have made clear enough that by "scientist" I didn't mean "professional scientist" but anyone who, at the moment, does science ( = applies the scientific method: falsification and parsimony). That M-L also happens to be employed at a university is not really relevant — nothing would change if she were an unusually well-read random bored teenager on the Internet.

    As long as you can answer the question "if I were wrong, how would I know?", you are doing science.

    In any case, I started my sentence with "You seem to", which softens the message in the sense that it's possibly just an impression. And obviously, it was an impression, what else could it be?

    Please. If I say "you seem to be a baby-eating monster", how are you going to react?

    but you should be able to access the paper from John Esling's web page (it was at an LSA conference, within the last 5 years at most).

    Thanks, I'll try to find it.

    Mr Sanchis, see you on your blog as soon as I find the time.

  62. marie-lucie said,

    February 28, 2009 @ 8:21 am

    Can we study language outside the natural habitat of language: conversation, oral exchange?

    The subfield of sociolinguistics, spearheaded by Bill Labov, has been particularly active in this field in recent years, and some of its conclusions have been invaluable for historical linguistics, especially as concerns the conditions and rates of change.

  63. David Marjanović said,

    February 28, 2009 @ 8:54 am

    What I forgot to mention:

    And to the uninitiated like myself, the transliteration of Tibetan words into the Latin alphabet looks exceedingly weird (although since the language is alive and well, it must be just because I have not taken the trouble of finding out how to pronounce it), for instance these ancient place-names:

    bDe-legs-gling
    gNya'-lam
    mTsho-sngon
    lHun-grub-rdzong
    rGyal-bzang

    and many similar ones.

    The modern Tibetan "dialects" have simplified the consonant clusters to varying degrees (some, like that of Lhasa, developing tones in the process; BTW, that lh is the voiceless lateral approximant), so that the spelling system — unchanged for over 1000 years — is now like that of French, only worse, because there can be silent letters on both ends of a word. Also, if I've understood that correctly, the script makes a distinction between normal consonants and the one-consonant prefixes; this is symbolized in the transcription by capital letters, and may indicate that there was an epenthetic [ə] in that place (as there is in many of today's Tibeto-Burman languages, some of which have been called "sesquisyllabic languages"). Finally, the transcription uses two or three letters for some single phonemes like /nʲ/, /gʲ/ or /t͡sʰ/. But still, this leaves consonant clusters which must have been pronounced as such when the Tibetan script was developed, and which most native speakers of, say, Italian would find seriously difficult.

  64. marie-lucie said,

    February 28, 2009 @ 9:28 am

    David,

    Thank you for your explanations. I had figured out that the lower-case consonants at the beginning of words must be some kind of prefix or clitic, and the digraphs seem obvious, but the initial clusters still would need getting used to. My point was that without some key to the transcription the words look at least as weird as the PIE reconstructions, if not more.

  65. marie-lucie said,

    February 28, 2009 @ 10:29 am

    Mark Liberman,

    You say that historical linguistics is one of the most dynamic areas of linguistics right now, but practically all the people you quote in your initial paragraph are specialists in non-linguistic areas, and there are many problems about how to adapt the methods of other fields to linguistic data, which are not so easy to evaluate and manipulate as one might think, even if they are free of errors (See a comment by Mary Kuhner in the "Scrabble tips" thread). And what those people are working on is classification and time-depth, which is related to history but not the same. As far as I can tell, most if not all of this work relies on comparisons of vocabulary. There is more to pay attention to than vocabulary in historical or classificatory linguistics: in my view as a historical linguist (in the American Indian area), morphology is especially important as it tends to change less than vocabulary, to preserve irregularities which point to earlier structures, and to involve fewer sources of potential error.

    It is indeed heart-breaking that so many students are interested in historical linguistics but do not find places where they could receive thorough training in the field. In my opinion the pressure of such students (and the often expressed interest of the public) will eventually lead to a reevaluation of the importance of the historical dimension in linguistics, but in the meantime there are many linguistic programs which do not even have a historical specialist on staff, and where it is possible to get an advanced degree in linguistics without having taken a single course in the subject. Hiring biologists or computational specialists to teach courses in historical methods will not fill the gap: such methods are at best adjuncts to the traditional ones, but cannot replace them.

  66. A Reader said,

    February 28, 2009 @ 9:02 pm

    It is indeed heart-breaking that so many students are interested in historical linguistics but do not find places where they could receive thorough training in the field.

    This is a problem I am grappling with myself at the moment. My college only offers Linguistics as a concentration (study formed by taking a selection of linguistic oriented courses in various departments), not even a major- so I have majored in Anthropology. We are lucky enough to offer a course on historical linguistics (although only one, taught by an English professor), which is more than we can boast for phonology or syntax (at least outside the minimal coverage in Intro Linguistics- taught by a professor of Russian). So I am finding myself in the position of starting to look at graduate programs of historical linguistics while fully aware that I have nowhere near the formal background I would like in the field.

    It will be interesting to see where things go in the future. The chair of our linguistics department is trying very hard to expand the department. If she's successful, we would get a professor to teach syntax- I don't know if there is even the slightest chance for further growth. We still wouldn't exactly have a well rounded department, but it would be very refreshing to have someone specifically trained in linguistics on campus.

  67. marie-lucie said,

    February 28, 2009 @ 9:31 pm

    Reader,

    I would encourage to apply even with your limited background, and with good recommendations and an application letter explaining your circumstances and your interests you have a chance to be admitted at least provisionally, but you probably will have to take extra courses to bring you up to scratch for a graduate program. Make sure the program of your choice has enough faculty with historical interests.

  68. David Marjanović said,

    March 1, 2009 @ 9:15 am

    Marie-Lucie, you might like this and this pdf, which are two examples of recent research in historical linguistics that centers mostly on phonology and morphology.

  69. marie-lucie said,

    March 1, 2009 @ 10:36 am

    DM, thank you for the references.

  70. Jesús Sanchis said,

    March 30, 2009 @ 8:16 pm

    It's a bit late but anyway, just a quick note: Ballester's article about the missing /a/ in Indo-European, which I mentioned in a previous comment, is now available on-line, here: http://www.continuitas.com/ballester_vocalismo.pdf

  71. David Marjanović said,

    March 16, 2011 @ 3:26 pm

    Just for the record, I've now read that paper, and the only argument it makes for PIE /a/ is that the lack of such a phoneme is exceedingly unusual. It does not even try to provide examples of roots with PIE /a/ or regular sound correspondences.

    Nor does it even try to provide any alternatives to "las fantasmagóricas laringales" (p. 19).

    It's just a rant with 272 footnotes. It brings up a few points that would be good in principle, but it doesn't go anywhere with them; it just throws them out there and leaves. That's practically useless.

RSS feed for comments on this post