And indeed, he has some arguments. First, he starts by reviewing the process that led to the invention of the theory. Let's remember it briefly: by the end of the 19th c. the Swiss linguist Ferdinand de Saussure proposed the idea that PIE had only one vowel (/e/) and a series of 'coefficients sonantiques' that could influence this vowel. Later on, the Danish linguist Hermann Möller introduced the concept of IE laryngeals, which he saw as arising from the supposed kinship between IE and Afro-Asiatic. The final impulse for the Laryngeal Theory came with the discovery oh Hittite, an IE language which showed a 'laryngeal' sound. In 1927, the Polish linguist Jerzy Kurylowicz combined Saussure's 'coefficients sonantiques' with the evidence from Hittite to produce the first version of the Laryngeal Theory.
According to Mańczak, the first problem is that Saussure's proposal is untenable, for various important reasons, among them the absurdity of proposing a language with just one vowel (/e/), when in fact the most common pattern in the world's languages is, at least /i a u/. (p. 26): "l'explication de ces alternances à l'aide des coefficients sonantiques est fausse. (...) les coefficients sonantiques n'ont existé que dans l'imagination de Saussure". This would be enough to invalidate the whole edifice of PIE laryngeals, which was based on Saussure's assumptions. But then Manczak goes on to analyse some further details of the theory, for example the fact that there are so many different versions of the set of laryngeals: (p. 29) "les laryngales n'existent que dans l'imagination de certains linguistes, la théorie des laryngales est un domaine où tout est permis, où rien ne freine la fantasie des chercheurs, où la notion de rigueur scientifique est inconnue". He quotes several authors who have expressed similar opinions, e.g . Hiersche (1964: 11): "die Laryngale sind in der Lage, bainahe jede Lautveränderung hervorzurufen oder selbst zu erfahren, was in der allgemeine Phonetik nicht seinesgleichen hat". The laryngeals were, and still are, the perfect solution to solve any possible PIE reconstruction mystery that could not be solved by other means, even if it's necessary to propose quite abnormal things, like consonants turning into vowels and other unlikely events.
In the final part of the article, Manczak asks himself why it is that the Laryngeal Theory has been so successful among linguists. According to him, there is a general lack of validity criteria in historical linguistics. (p. 31): "le terme "critères de verité" n'est jamais employé par les linguistes, bien que les linguistes soient unanimes pour dire que la linguistique est une science". The important thing is the 'authority' behind the theory, not the validity of the theory itself. (p. 32): "Comme les linguistes croient en l'infaillibilité des autorités, ils détestent ceux qui osent critiquer les autorités et adorent ceux qui approuvent ou développent les idées des autorités".
I find Manczak's proposals quite interesting, and I think anyone doing research in the field of IE studies should take them into account, instead of assuming the Laryngeal Theory as indisputable truth.
References:
- Hiersche, R. (1964). Untersuchungen zur Frage der Tenues aspiratea im Indogermanishen. Wiesbaden.
- Mallory, J. P. and D.Q. Adams (2006). The Oxford Introduction to PIE and the PIE World. Oxford University Press.
- Manczak, Witold (2006). «Invraisemblance de la théorie des laryngales». In Historische Sprachforschung, 119: 25-34.
33 comments:
"'[...]que la linguistique este une science.'"
That should be est not *este.
As for denying that laryngeals exist, the evidence is quite clear in Anatolian. What more is there to dwell on? However, it's also absurd to reconstruct more than three 'laryngeals'.
Thanks for your comments, Glen.
The evidence for Anatolian is that there is a consonant, generally transcribed as "ḥ", whose phonetic realization is thought to be /h/, or in some cases closer to /k/. The rest of the story is, in my humble opinion, highly speculative, and probably quite wrong.
Were there laryngeals in PIE? First of all: what is a 'PIE laryngeal' exactly? A vowel? A consonant? A consonant that can be a vowel? A vowel that can be a consonant? A 'colouring' device? Laryngeals are always described vaguely. In fact, the term 'laryngeal' does not even exist in modern phonology, where we have other terms to refer to places of articulation (dental, velar, labial, glottal, etc.).
Yet Anatolian clearly shows these 'laryngeals' in all the places where we would expect them in modern reconstructions.
*h1 is seen as a glottal stop. (If you follow Kloekhorst's reconstruction of Hittite phonology) or as the absence of a consonant.
*h2 is seen as the /h/.
*h3 is seen as the /hw/. That these laryngeals correspond perfectly with the places where we were reconstructing these three laryngeals before we found Anatolian is too much of a perfect fit to be called a result of 'chance'.
Now in Post-Anatolian Indo-European, I still think there were Laryngeals, but I think there's definitely a case to be made that maybe Laryngeals were not there anymore, at least not in their purely consonantal value as found in Anatolian.
Anyone who writes an article in 2006 about the implausibility of Laryngeals is obliged to give an alternative plausible explanation for Anatolian. If he doesn't he is simply leaving out facts to suit his purpose.
Thanks for your comments, PhoeniX. I think you're right: perhaps Manczak's criticism focuses only on the weak points of the theory, without offering an alternative explanation for Anatolian, but maybe this explanation is just impossible in the framework of traditional PIE reconstruction.
Personally, I think the Laryngeal Theory is based on the optimistic view that PIE laryngeals fit the expectations. In his article, Manczak shows us that this correlation is not so perfect. He points out that the supposed theory only 'works' in approximately half of the cases, and offers quotes from other authors, e.g. Tischler, who wrote things like this: (quoted on p. 28 of Manczak's article): "Die Zahl der Gegenbeispiele ist demnach in den verschiedenen Bereichen verschieden hoch (...) Eine Theorie, die allenfalls für die eine Hälfte aller beobachteten Fälle zutrifft (...) kann bei bestem Willen nicht als zutreffend angeschen werden. (...) Kurylowicz ging dagegen deduktiv von den Theorien de Saussures und Cunys aus und wollte im Hethitischen nur die Bestätigung für diese Theorie finden". On the other hand, Manczak adds another interesting fact: (p. 27): Kurylowicz's three-laryngeal theory is based on the analysis of no more than 24 Hittite words.
As I said, I think there is excessive optimism about the validity of the laryngeal theory. One of the aims of my blog is to offer alternative views about some aspects of historical linguistics, including PIE laryngeals.
Finally, PhoeniX, I don't understand the phrase "Post-Anatolian Indo-European": What do you mean exactly? I guess it implies an adherence to Renfrew's Anatolian Hypothesis.
It's important to keep in mind that just because there are 9 symbols doesn't mean that there are 9 laryngeals. Usually only 3 or 4 are reconstructed. A symbol such as h_2/3 means that it's unknown if the reconstructed laryngeal is h_2 or h_3, for example.
One of the problems with Szemerenyi's critique (and apparently also Manczak) of laryngeals is that he attacks the theory that also assumes that PIE had only one vowel, /e/. This theory has few, if any, promoters anymore. Just about everybody thinks PIE had at least two vowels /e o/. Some further assume that there was also a length distinction.
Personally I share some of your concerns about the Laryngeal Theory. Laryngeals are invoked to explain every little oddity in PIE that you can think of. This is unnecessary. I think a number of issues with laryngeals can be dealt with just by assuming that PIE did indeed have *a and that words can begin with vowels.
Mordrigar, I agree with you on some the things you say. Manczak's criticism of the Laryngeal Theory starts with criticising its origin, i.e. Saussure's porposals, that's why he comments on the model with just one vowel (/e/). There have been a wide variety of proposals since Saussure's times. You mention the fact that there's some general agreement on a vowel system with at least /e/ and /o/. I find this problematic. As I said, the most usual set in the world's languages is /a i u/. The problem is that the Laryngeal Theory, especially in its most extreme variants, usually requires exceptional facts.
You say the following: "A symbol such as h_2/3 means that it's unknown if the reconstructed laryngeal is h_2 or h_3", and you're right. Mallory and Adams use four laryngeals (from 1 to 4) and five more for cases which are not clear. This is linguistic reconstruction gone too far.
But there is really no possible way to reconstruct anything approaching PIE without suggesting the presence of some consonants that have been lost. The fact that they have been lost means that it is hard to say what they sounded like, to be sure, but obviously they aren't actually attested, ergo any reconstruction of their sounds is inherently speculative.
I'm curious, though, how you would try to explain things like the varying vowels of various languages (which without laryngeals have no conditioning phonological condition)? Or how you would explain roots like *h1es without a consonant at the beginning (**es? **esmi, **essi, **esti, **smos, **stis? where is the /e/ of Latin estis, or the /e/ of Greek esmen/εσμεν? or do we have a different conjugation here, and in that case, what did it develop from without laryngeals?) which would lead to a completely fraught and inconsistent verbal and nominal system?
It doesn't matter who thought them up originally, or what they're called - we can call them sons inconnus if you'd like. The fact is that PIE just doesn't make any sense without them.
Ethan, what makes you think that PIE can be reconstructed in such detail? Maybe many of the differences between the various IE languages are the consequence of hybridization processes that we know very little of.
You ask me to provide an alternative explanation for some IE phenomena, but you imply that this answer can only exist in the framework of a theoretical construct called a 'proto-language'. You seem to take for granted that a proto-language, in this case PIE, is necessarily an acceptable theoretical construct. I am spectical, I see it differently.
In the absence of any compelling evidence that there are hybridization processes which render the comparative method useless, there is no reason to assume that they exist. That's like saying evolution is all an unjustified theory, because there may be creationary processes that we know nothing of. Possible? Definitely. But not useful.
I don't know that it makes any sense to reinterpret the IE data in terms of a hybridization process that nobody knows anything about. Which hybridization processes do we know exist? There's pidginization, creolization, and language mixing. If there are any others I then I've never heard of them, but would love to know more. If you want to reinterpret the IE data as a hybridization process you should look at the processes that we already know about and how they are known to work.
When I talked about "hybridization processes that we know very little of" I was thinking about the specific processes that may have affected IE languages thousands of years ago, not about hybridization in general. I think we must assume that they existed and that they may have had a significant role in the formation of IE languages, including phonology.
I'm not against reconstruction. As I've said more than once in this blog, it is a necessary tool in linguistics. But again, I have the impression that PIE reconstructions with laryngeals are too artificial, or too 'optimistic'. And my impression is reinforced by what some experts have written about the subject, e.g. Ballester, Marcantonio and Manczak.
What specific processes are these? We can't very well assume they existed if we have no idea what they are.
What is it about the laryngeals that's so irksome to you? I saw you railed earlier against the construction of *h₁widʰweh₂h₁e; granted, wrapped in the terminology, that does look weird. But that's because it's wrapped in the specific terminology. If we take a given reconstruction of the sounds, say giving it instead as *h̥widʰwexʰe, it looks a lot more normal (especially considering this is an edge case, since you really wouldn't refer to two widows in the nominative very often, realistically). Or take *h₁neh₃mn; instead, *h̥neɣʷmn̥; those aren't even especially hard to pronounce.
Finally, PhoeniX, I don't understand the phrase "Post-Anatolian Indo-European": What do you mean exactly? I guess it implies an adherence to Renfrew's Anatolian Hypothesis.
Eek! No, what a terrible misunderstanding (and, I must say, completely my fault).
I believe that the Anatolian languages split off earlier than the other PIE languages. As there is very clear evidence of laryngeals in Anatolian, it is, to me, without a doubt that before the split of Anatolian and PIE laryngeals must have existed.
The PIE-stage form which the later PIE languages split has far less concrete evidence of Laryngeals. I still think that the evidence is strong enough to supposed these consonants to have existed, but it can definitely be contested.
Ethan: I have already written about hybridization, diffusion and other processes in this blog, and it would be difficult to summarize the whole thing in just a post comment. For a better understanding of the role of hybridization in prehistoric languages I would suggest reading Mario Alinei's writings.
PhoeniX, thanks for the clarifications. You talk about languages 'splitting off', and as I understand it, this 'splitting off' is merely a metaphor, not an explanation, or a fact.
Great blog!
Thank you!
Hello and thank you for an interesting blog. I agree that there are some problems with the way Indo-Europeanists treat "laryngeals" and with the way they treat novel ideas and approaches. Over the past 10 years, I've been involved in developing new approaches to etymologizing IE kinship terms by combining anthropological and comparativist methodologies. This project naturally took me into PIE phonology. What I realized is that comparativist method in linguistics is constrained by ad hoc definitions of cognate sets. Words get grouped on the basis of "similarity" in sound and meaning. But what if overtime words change both formally and semantically? Comparativist method has no way to handle this complex situation.
As a result of re-comparing IE cognate sets I came across quite a few possible cases in which "laryngeals" correspond to... palatalized velars in satem languages (specifically, H2) and labiovelars in centum languages (specifically, H3). Exactly like "laryngeals" palatalized velars and labiovelars are not attested in IE languages.
here're a few examples.
IE *peh3ur 'fire' (Hitt. paHHur-, Greek puur, etc.) next to *pek(w)o- 'cook, bake'.
IE *paH2s- 'to protect' (Hitt. paHs, Lat, paasco-, paastor 'shepherd') next to IE *peku- 'cattle' (Skrt. pasu-, Lat pecu 'cattle', Slav *pasti- 'graze', etc.)
IE *swekuros 'husband's father' (Skrt. svasura, Gk hekuros, Arm skesur, Alban vjeher, etc.) next to
IE *wiH2ros 'man, husband' (Lith vyras, Lat viir, etc.). Arm skesur (< *s-gesur) suggests that s- in *swekuros is a secondary addition).
See more on my website at www.kinshipstudies.org
So, my overall impression is that "laryngeals" is an artifact of comparativist methodology that errs by assuming that "similarity" will ensure the correct placement of cognates into cognate sets from which sound laws are eventually derived.
Thank you very much for your comments, German. I see we have similar views on the status of PIE laryngeals.
I have browsed your web-page and your blog, and they look really interesting. I'll go back to them when I have some time.
One more interesting example. Comp. IE *bhreHter 'brother' and IE *mer- 'affine' (Latin mariitus 'husband', Germ *bruudi 'bride', Latv marsa 'brother's wife' - here is the clear morphosemantic link between the two cognate sets as marsa is identical to Lat fratria 'brother's wife', with suffixal -r- regularly dropping in Baltic). As we can see, the front aspirated labial bh can be derived from b and further from m in a br/mr cluster environment provided that there was an "aspiration throwback" caused by the medial, suffixal "laryngeal." For the connection between laryngeals and voiced aspirated stops see IE *dhugHter 'daughter' where H2 not only vocalizes in Sanskrit (dhuhIta) and Greek (thugAter) but also converts a voiced stop into a voiced aspirated stop in dhuHita. It's quite strange that a single laryngeal produces two effects: turns into a vowel and transforms a consonant at the same time. But this is exactly what we see in my other examples above: in some cases some ancient consonant disappears but colors an adjacent vowel, in others it turns stops into spirants (e.g., satem "s") and plain voiced stops into voiced aspirated stops.
There are all these puzzles in IE primary data and instead of addressing them, linguists tend to hide behind the forest of "laryngeals."
And another interesting one illustrating the identity of H2 and palatalized velar. Comp. IE *H2rkto- 'bear' (Hitt hartagga, Gk arktos, Lat ursus (< *urksus, *urkstus?), Skrt rksa, Arm arj) and IE *k'er(es)-/*k'rst- 'fur, animal hair' (OHG hursti 'crest', OCS srusti- 'fur', Lith serys 'bristle, animal hair') with the resulting *H2(=k')rst-/*H2(=k')er-. Borrowings into Uralic (Finn karhu 'bear') attest for a velar in place of a laryngeal. The etymology of bear as 'hairy one' can't get any better in view of frequent tabooing of this animal species across the IE area. Finally, the "fur" isogloss covers precisely the northwestern IE area in which the bear isogloss is not attested. This makes the two isoglosses not only historically related but geographically complementary.
When a phoneme is predicted by theory and subsequently discovered in newly discovered languages, that's about as close to 'proof' as you're going to get in a historical science.
There's certainly a danger of the things being used as a universal panacea, but discarding the whole theory on that account would be like discarding Evolutionary Theory because of a few over-zealous explanations of data.
Thanks for your comments, Naso. One of the problems about the Laryngeal Theory is that it is generally accepted with little or no criticism, and used, as you suggest, as a panacea for the reconstruction of IE proto-languages. I would also call it 'the mirage of structure'.
A model of PIE phonemic elements without an /a/ is, in my opinion, proof that there could be something wrong in the model.
That's a pleasure, thanks for replying.
Well, there has been lots of discussion and criticism of Laryngeal Theory in the past, though there is currently - for the moment - general consensus.
Although the algebraic notation may seem 'artificial', and the lack of phonetic determinacy unsettling, there really is nothing about the reconstruction of laryngeals that's any less solid than, say, that of PIE *bh (and they're arguably more solidly based than, say, *b). They are predicted by the comparative method and directly attested in the Anatolian branch of the family. And the fact that they were predicted before they were discovered makes them, if anything, less vulnerable to accusations of methodological circularity than other reconstructed phonemes.
As to the uncertainty over phonetic realisation, the recent history of the Glottalic Theory, for instance, shows this is the case with the reconstructed language generally. With the laryngeals we're just more honest about it. Re the terminology, it's just a historical oddity that's stuck.
Re the difficulties with pronunciation of laryngeal-heavy words that you talk about in your essays, I'd say two things. Firstly, the kind of sounds you're talking about are not in fact particularly difficult to pronounce. Leaving aside the consonant clusters of languages like Slavic, listen to Klallam some time! http://www.lingtechcomm.unt.edu/~montler/Klallam/kttd.htm (Those are tongue-twisters, admittedly, but amazing ones!)
But secondly, all of that is essentially irrelevant anyway. Your objections suggest you've been misinformed about the comparative method. Laryngeals are not sounds; PIE is not a language. The former are phonemes, and the latter is a model, a theoretical construct. This is also why your objections about artificiality are not so much false as beside the point: the whole system is artificial and algebraic - it so happens that the laryngeals are written with little subscript numbers. That's not something that linguists are or should be embarrassed about: that's simply what the comparative method produces.
An excellent image for this is offered by James Clackson, my old supervisor, in his 'Indo-European Linguistics'... (see next post)
(Having trouble getting posts to work... I'm hoping my previous one got through - apologies if it's repeated seven times! Here's the rest...)
"Reconstructed PIE is a construct which does not have an existence at a particular time and place (other than in books such as this one), and is unlike a real language in that it contains data which may belong to different stages of its linguistic history. The most helpful metaphor to explain this is the ‘constellation’ analogy. Constellations of stars in the night sky, such as The Plough or Orion, make sense to the observer as points on a sphere of a fixed radius around the earth. We see the constellations as two-dimensional, dot-to-dot pictures, on a curved plane. But in fact, the stars are not all equidistant from the earth: some lie much further away than others. Constellations are an illusion and have no existence in reality. In the same way, the asterisk-heavy ‘star-spangled grammar’ of reconstructed PIE may unite reconstructions which go back to different stages of the language. Some reconstructed forms may be much older than others, and the reconstruction of a datable lexical item for PIE does not mean that the spoken IE parent language must be as old (or as young) as the lexical form."
See, the whole thing is a 'mirage of structure'! That's what models in historical sciences are. In the absence of time machines, that's the best we can do.
None of the above is to suggest that there aren't still lots of problems, controversies and mysteries in connection with laryngeals. But I do think that you do a disservice to linguists when you imply that they're just following authority in a sheeplike (h3ewine?) way. I think most of them have wrestled with the data and theories and found the LT well enough supported (by facts, not just scholars!) to accept. If you're sceptical, that's right and proper, but the onus is on you to come up with something as persuasive or more so: vague intuitions are not enough, because most of the time our intuitions about the structure of languages - even our own native language - are completely wrong.
"But secondly, all of that is essentially irrelevant anyway. Your objections suggest you've been misinformed about the comparative method. Laryngeals are not sounds; PIE is not a language. The former are phonemes, and the latter is a model, a theoretical construct."
Allow me to interject: models are fine, but the inability of the standard comparative method to meaningfully engage with other model-generating (sub)disciplines, such as linguistic typology or my kinship studies makes IE historical linguistic models rather "thin," the method underlying them based on circular logic and many etymological problems involving "laryngeals" unsolved.
It's really not about models vs. reality but about the degree of density of models and their ability to respond to data and reasoning embedded in adjacent models.
I agree with you, German.
No matter how PIE reconstructions are understood in theoretical terms, the truth is that they're usually taken at face value in order to prove all kinds of theories about PIE languages and societies. There is indeed a very high degree of circularity in all this.
Further evidence for laryngeals can be seen in apparent contact words shared with Finno-Ugric, Semitic, and Kartvelian. I believe some of the initial negative reaction to the laryngeal theory is partly rooted in the only partial digestion at that point of Anatolian data.
By the way, laryngeals in Hittite do not work exactly as they do in other Anatolian languages, which is the sources of some of M&A's more complicated coefficients. They only believe in 3-4 laryngeals, but wished to present reconstructions in such a way that individual readers can colour them however they wish, according to their own beliefs. This is because their works are concerned with semantic fields and the resulting picture of Indo-European material culture, not with abstract phonological reconstruction.
As regards vowels, the most popular reconstructions today have four vowels, not one (i, e, o, u). *a is controversial, and obviously *e and *o and the most important in morphophonological patterns (e.g., Aktivum vs. Perfektum).
As for the affinity between some laryngeals and velars...that is not an argument against laryngeals. It could, however, influence how we reconstruct the laryngeals themselves (e.g., leans towards uvular/velar [x] or [q], not pharyngeals for *H2/3). See the first part of Sarah Rose's dissertation on the Hittite -hi conjugation.
Nothing is 100% clear in reconstruction, as not all language change takes place as part of a scientific process!!! But the laryngeal theory is not one of the odder elements of the modern-day conception of PIE (the reconstruction of nominal cases is much cloudier!).
I do not think it is wrong of some commenters here to sense a high degree of anti-scientific mentality here. I will, however, have to read more of this blog to see if that be the case (and to see what the blogger thinks about other aspects of reconstruction).
http://www.utexas.edu/cola/centers/lrc/books/piep03.html
Interesting statement:
"Even though Hittite has supplied the clinching evidence for the laryngeal theory, the Hittite evidence is not without difficulty, and almost disappointing as a support for the theory."
:)
H-dropping and h-adding:
http://en.wikipedia.org/wiki/Phonological_history_of_English_fricatives_and_affricates#H-dropping_and_h-adding
Lastly, how can one be sure that the script reflects the language as spoken correctly?
It is assumptions piled on assumptions.
Anyway, Hittite cuneiform on Wiki
You're right, Arun. There's a lot of wishful thinking in the idea that Hittite can support something as artificial as the set of supposed PIE laryngeals.
We know about Hittite from a series of inscriptions in Old Assyrian cuneiform. The Hittites adapted the script to their language but it is obvious that, on the whole, these signs are more suitable for a Semitic language than for an IE one. This is an additional difficulty in our understanding of Hittite and the other languages of ancient Anatolia.
I haven't commented before but I really enjoy reading you blogs.
I generally do not dispute that all the so-called sub-branches of IE sprang from a common language, but find many problems with the PIE reconstructions. Many are jumping to defend laryngeals, but so far no one has attempted to offer an explanation for the /a/ less model. Currently there is only one language in the world that lacks an unrounded open vowel.
Languages are presumed to have been more complicated as one goes back further in history. PIE reconstruction has followed suit with the consonants, surely enough, but has reduced the number of vowels to two (generally). I don't see any reason for this. It's not like any documented "descendants" of PIE have an /a/ less vowel system, or is there one?
Exactly, Ken, the absence of /a/ is one of the most strinking and absurd features of the traditional PIE model, together with the laryngeal set. There's an interesting article by Xaverio Ballester (in Spanish) about this missing /a/. I don't know if you've read it, but anyway, this is the link to the post where I duscssed this issue, with a llink to Ballester's article:
http://languagecontinuity.blogspot.com.es/2009/01/laryngeal-theory-revisited.html
To me, if you do some complex calculations and you get an unlikely answer, you treat it as a miscalculation and redo the calculations. You don't just invent a theory that justified the seemingly erroneous calculations.
Thanks for the link to the article. I have read parts where you have quoted Ballester in your blog. Although I have some elementary knowledge of Spanish that could probably get me by if I was lost in the streets of Madrid, that article is slightly beyond me... I will fight it with a dictionary sometime :)
There is now a brand-new comparative theory with only one laryngeal:
https://helda.helsinki.fi/handle/10138/41760
Post a Comment