17 Jan 2008

Winter's Law in Balto-Slavic, "Hybrid Theory" and phonation - Part 1

From here on in, I'm going to coin the term "Hybrid Theory" which is my cute catchphrase to describe a phonological stance somewhere between the Traditional Theory and the Glottalic Theory concerning the interpretation of Proto-Indo-European (PIE) phonology. I still maintain, no doubt because I have a stubborn streak, that what are called "plain voiced stops" (Traditional Theory) or "ejectives" (Glottalic Theory) are better conceived of as phonemes that are in some way merely derivatives of ejectives, thereby showing the inherited traits of ejectives while explaining why ejectives are not featured whatsoever in the languages that sprang from PIE.

A recent comment by Phoenix concerning Proto-Balto-Slavic has left me pondering incessantly about something for the past few days now: Does Winter's Law prove conclusively that PIE had ejective stops and even disprove my Hybrid Theory stance?

The standard description of Winter's Law by Mr. Winter himself was that in Balto-Slavic, vowels had somehow lengthened before PIE plain voiced *d but did not lengthen before voiced aspirated *dʰ. Frederik Kortlandt expanded further on this by infusing it with the Glottalic Theory according to which PIE's ejectives became "implosive" in all dialects save Proto-Anatolian and Proto-Tocharian. His reinterpretation of Winter's Law was that the implosives somehow developed a glottal stop segment preceding a plain voiced stop. It was the glottal stop element that supposedly lengthened the preceding vowel and caused acute (i.e. rising) tonal accent[1]. It is afterall to be expected for voiceless codas to produce such tones as we see in, say, Middle Chinese and other budding tonal languages. In light of this, I was about to admit my inability in find an alternative to this account. If *d were "half-voiced" as I claim, it cannot produce the required rising tone since the presence of voice produces low acoustic pitch. In a coda, we then would expect either low level or falling tones to result, not rising ones. It seemed as though I had my pants down around my ankles. I was even prepared to give myself a beating for being so daft all this time.

Enter Miguel Carrasquer Vidal, a very informed individual who has been participating on forums for a decade. He apparently commented two years ago on sci.lang debunking the Glottalic Theory interpretation of Winter's Law in a single paragraph[2]. If I understand his point correctly, he's saying that a phonetic trigger wasn't the cause of acute tones at all. Miguel claims that Toshihiro Shintani discovered that it was merely the pretonic nature of Winter's Law itself that provoked Latvian's Brechton (the German name for "broken tone"). I presume he's indirectly citing Shintani, On Winter’s Law in Balto-Slavic (1985). Add to this, Oswald Szemerenyi's negative comments about Kortlandt's theory[3]. Recently in the turn of the century, Rick Derksen pulled no punches and told it like it was: "The sound law that is generally referred to as Winter’s law can by no means be regarded as well-established. Its reception has ranged from almost unreserved acceptance to categorical rejection."[4]

Oh dear lord, what do we do now? What's the answer and who do we have to kill to get it?

(Continued in Winter's Law in Balto-Slavic, "Hybrid Theory" and phonation - Part 2 ...)

[1] Remarks on Winter's law. Studies in Slavic and General Linguistics 11: Dutch Contributions to the 10th International Congress of Slavists: Linguistics (1988), p.387-396 (see pdf).
[2] Miguel Carrasquer, sci.lang - Re: Indo-European Typology and Sanskrit Phonology (21 Feb 2006) (see link).
[3] Szemerenyi, Introduction to Indo-European Linguistics (1999), p.153 (see link).
[4] Derksen, On the reception of Winter’s law, Baltistica 37 (1) (2002), p.5-13 (see pdf).

(Jan 19/08) I corrected the year of Shintani's first publication of On Winter’s Law in Balto-Slavic. I stated "1988" but it should be "1985" apparently, as cited in one of Kortlandt's online pdfs. I cite the full reference in Part 2.


  1. Fun to see Derksen's name pop up! That's the guy who made the Slavic Etymological Dictionary that I'm currently digitalising ;)

    There's one more rather dangerous theory that seems to give some plausibility to the glottalic theory, which I, due to its controversiality didn't want to pull out to soon.

    As you may know there's a very noticeable lack of *b in PIE. Apparently, when looking at more languages that have pre-glottalisation, they all tend to lack the pre-glottalised b.

    This is just what I've been told, I've never checked the facts, and in fact, I don't know any languages with pre-glottalisation.

    Besides that I thin linguistic universals are dangerous and can divert us from the point. But once a theory is well established it is often found to nicely come together with universals.

    So even though, I don't find the absence of *b a convincing point, I just thought I'd mention it for completeness ;)

  2. My immediate thought is that a glottal stop segment preceding a plain voiced stop seems phonetically weird. You'd think that the voiced stop would devoice.

    At least one IE language, Sindhi, has implosives fwiw.

  3. Phoenix: "As you may know there's a very noticeable lack of *b in PIE. Apparently, when looking at more languages that have pre-glottalisation, they all tend to lack the pre-glottalised b."

    I think this idea is getting off track. Glottalic Theory and my "Hybrid Theory" position already establish ejectives as the original phoneme for pretty much the same articulatory reason, putting aside which stage of PIE or Pre-IE had ejectives for the moment. Ejectives are surely more common than the "preglottalized" phonemes (which I can only assume is a bizarre term for an "implosive"). If Occam's Razor means anything at all, we should be seeking the most optimal solution that requires the least amount of assumption (including the least amount of exotic phonemes). There's nothing particularly exotic about ejectives in world languages, ergo it explains the absence or scarcity of *b (< *p') the best thus far.

    Phoenix: "Besides that I thin{k} linguistic universals are dangerous and can divert us from the point."

    I would counter-argue that the dismissal of seeking regular structure is an anti-scientific attitude and even more dangerous. Since you offer no better alternatives to using linguistic universals as a guide, your statement lacks constructivity and therefore without further elaboration is illogical. Theories "come together" precisely because of the adherence to structure, while others that lack logically structured methodology continue to fail (e.g. "Nostratic Theory").

  4. Goofy: "At least one IE language, Sindhi, has implosives fwiw."

    Yes, but on the other hand, may I refer you to Lehmann, Theoretical Bases of Indo-European Linguistics (1993), p.98 (see here): "Some efforts have been made to ascribe glottalic stops in Armenian dialects and in Sindhi to retention from the proto-language, but these have not been widely accepted; rather, in both languages the glottalics are assumed to be recent."

  5. Might I suggest further reading concerning the origin of implosives from voiced stops rather than ejectives?

    1. Miyake, Old Japanese: A Phonetic Reconstruction (2003), p.120 concerning the history of Vietnamese phonology: "Vietnamese initial unaspirated labial and dental stops became voiced implosives: [...]"

    2. Fallon, The Synchronic and Diachronic Phonology of Ejectives (2002), p.283: "The development of implosives from voiced stops to me more common than the reverse, [...].

    There are also implosives in the Wu dialect of Chinese and I don't recall seeing ejectives in Middle Chinese. So then, we have to ask ourselves: Is it really likely that Sindhi retains an archaicism that validates Glottalic Theory, or is it far likelier that Kortlandt has a dirty fetish for glottalized phonemes?

  6. is it far likelier that Kortlandt has a dirty fetish for glottalized phonemes?

    I could go round his office and ask him. :D

    I should really consider doing that. Or just sending him an e-mail.
    I mean, being in Leiden, they practically bomb me with the glottalic theory, which, the more you talk about it, becomes more implausible. At least, Kortlandt's non-hybridic form.

  7. It's also important to note that implosives are found in only 10% of world languages (see here and here). Implosives aren't the only phonemes that can derive from an ejective either.

    It seems to me that it's sort of ironic that ejectives (which are so common) are used to explain away PIE's typological patterns brilliantly and yet the theory is self-sabotaged by the relative rarity of implosive/preglottalized phonemes that the Glottalic camp and others feel are required to explain why traditional *d and *g surface far more often than not as voiced phonemes. The unnaturalness of such a solution is then further compounded by the implausibility of most PIE dialects supposedly doing away with ejectives independently.

    Plus, as an aside, apparently we're not the only ones having trouble grasping what linguists really mean by "preglottalized" (see Gussenhoven/Warner, Laboratory Phonology 7 (2002), p.313).

  8. Although the amount of languages that has implosives might be a bit small, but the amount of speakers of language with implosives is not that small at all. Swahili comes to mind, and there's several other largely spoken Bantu languages.

    But as can be clearly seen in the above paragraph, Bantu languages, so unless someone goes really crazy and wants to have a go at proposing relation or strong linguistic contact of Indo-European with Bantu speakers, I'd say that implosives are indeed difficult to believe to have existed.

    It seems odd to me to have an ejective develop into an implosive anyway, the direction of airflow is exactly opposite. But I guess it has happened before. nevertheless, I agree that it's wholly unnecessary to assume that it went ejective > implosive > voiced stop in all Indo-european language (except for Tocharian/Hittie).

    I think I'm starting to see the implications and merit of voiceless-semivoiced-voiced.

  9. Population is not a statistic to go by. If there are a million speakers of one language with a bilabial trill, we still have only one language with a bilabial trill and it's still rare.

    Phoenix: "I think I'm starting to see the implications and merit of voiceless-semivoiced-voiced."

    Well, hold on. I want to talk more about that. I don't think I've been as accurate about these contrasts as I would like to be. I believe there is a way to reinterpret Glottalic Theory with minimal fuss, optimal naturalness and a possible solution to Winter's Law without preglottalized phones.