8 Apr 2010

A revisal of the PIE sound system

As Jésus Sanchis in his blog Language Continuity takes up a noble battle against academic insanity and laryngeal overindulgence, I decided to finally upload a small pdf to my Lingua Files section regarding my summarized views on PIE phonology which have remained somewhat static for the past couple of years. Just as I've blogged before, I still advocate uvular and creaky-voiced phonemes to help rebalance the system and bring it in line with what we currently understand about phonology and its tendencies. I've included a paragraph or two about the ill-named laryngeals as well.

Of course, the staunchest traditionalists will be outraged by my vision as they are with most change but the rest of is, most notably qualified phonologists as well as rational Indoeuropeanists, should at least be intrigued if not in agreement.


  1. I very much agree with most of what you say here. A few little questions/comments:

    1) I've also seen reconstructions that use a tense-lax distinction rather than voicing. What are your thoughts on that, esp. if voiceless = tense * voiced = lax ?

    2) In regards to *k = [q], how do you account for stems with *keC?

    3) An idea I've been throwing around about *h1 is that it functioned as /h/ initially, and as a voiced /h/ (/ɦ/) medio-finally. My only worry is that the medio-final form could be too easily lost. Does it still sound plausible?

  2. 1) I simply don't see how voicing can be sensibly, nor should be, done away with in PIE because there's no doubt that we're dealing with a fundamental 3-way contrast in the plosives and voicing is a clear feature in the majority of its branches.

    2) I think stems of the form *keC- are not correctly reconstructed. Via Eichner's Law, a long vowel is unaffected by colouring so this may be just a matter of lengthening the root vowel. Ergo *kēC-. Indeed, as per Jasanoff, presents were originally normally characterized by a lengthened grade (ie. Narten presents).

    3) It's not impossible I suppose but it's just that this initial /h-/ of yours is in want of evidence. A glottal stop is more parsimonious to the data.

  3. I've been reading Derksen's Etymological Dictionary of the Slavic Inherited Lexicon, and now I'm curious - how would you account for Winter's Law in Balto-Slavic, where, according to the pro-glottalic Derksen, "the PIE glottalic stops dissolved into a laryngeal and a buccal part. The former merged with the reflex of the PIE laryngeals [i.e. /ʔ/ in Balto-Slavic] and the latter with the reflex of the lenes [i.e. traditional "voiced"] stops, e.g. Latvian pȩ̂ds [ê has cedilla] 'footstep' < *pedóm, nuôgs 'naked' < nogʷós, duômu 'I give' < *dodH₃mí."

    In addition, the root *bel- has reflexes in European and Sanskrit; for example, Gk. βελτίων "better", Proto-Slavic *boļjь [ļ = palatalized /l/] "bigger, better", Skt. bálīyas- "better", all from extensions of the o-grade of *bel. I'd imagine it to be a post-Anatolian, possible post-Tocharian borrowing into the PIE dilect continuum. You agree?

    Finally, there's a handful of roots that look suspiciously substratum-y, among these are some that seem worth further investigation for Aegean connections:

    *čemerica; *čemerika fem. 'hellebore'
    East: Russian čemeríca
    West: Czech čemeřice; Slovak čemerica; Polish ciemierzyca; Old Polish czemierzyca
    South: Serbo-Croatian čemèrica; Slovene čemeríca 'hellebore, morose person'; Bulgarian čemeríca

    See -> *čemerь where also the non-Slavic plant-names are mentioned. The Slavic derivatives of *čemer- denote both the highly toxic white hellebore (Veratrum album) and various species of Helleborus, which are also toxic.

    *čemerь; *čemerъ masc. 'hellebore, poison'
    Church Slavonic: Russian Church Slavonic čemerь 'hellebore'
    East: Russian čémer (dialectical), čémer' (dialectical) 'crown (of head), forelock, headache, belly-ache, horse's disease'; čémer (dialectical) 'poison, West: Czech čemer 'name of an illness, aversion'; Slovak čemer 'illness caused by coagulation of the blood, weakness'; Polish czemier (dial.) 'hellebore, stomachache (of a horse)'
    South: Serbo-Croatian čȅmēr 'venom, andger'; CHakavian čȅmer 'venom, anger'; Slovene čemę́r 'venom, anger, gall'; čmę́r 'venom, anger, gall'

    Balto-Slavic. kemero-
    Baltic: Lithuanian kė́meras 'hemp agrimony (Eupatorium cannabium), burr marigold (Bidens tripartita)'; Latvian cemeriņš 'hellebore'

    Cognates: Greek κάμαρος masc. 'larkspur (Delphinium)'; Greek κάμ(μ)αρον neut. 'aconite'; Old High German hemera fem. 'hellebore'

    Undoubtedly a non-Indo-European plant-name.

  4. *màkъ masc. (with fixed stress on stem) 'poppy'
    CS: Church Slavonic makъ
    E: Russian mak), gen.sg. máka
    W: Cz. mák; Slk. mak; Pl. mak
    S: SCr. mȁk, gen.sg. mȁka; Čak. mȁk, gen.sg. makȁ; Sln. màk, gen.sg. máka; Bulg. mak

    Baltic: Lith. aguonà fem. {with variants maguona and magūna}; māguonė fem.; Latvian maguône fem. {The /m/ was lost in Lithuanian, but not in Latvian; Lith. forms with /m/ are restricted to the area around Klaipėda, and may be due to the (Latvian) language of the fishermen of the Curonian Isthmus}; Old Prussian moke (Elbing Vocabulary)

    Cognates: Gk. μήκων fem., Dor. μᾱ́κων fem.; OHG māho masc.; OS magosāmo masc. 'poppyseed'; OS mēcopin masc. (Königsberg); OSw. valmoghe {val- means 'sleep', c.f. Nw. vale (dial.) 'deep sleep'}; Est. magun; Livonian maggon

    The Germanic forms show grammatischer Wechsel as well as an alternation : *a. The vocalism, which could reflect PIE *eh₁ : *h₁, does not match the of the Greek and Slavic forms, which leads us to assume that the vowel alternation arose when at a comparatively late stage the root māk- was loaned into Germanic (cf. Kluge-Seebold: 565). The Lithuanian and Latvian forms are usually considered borrowings from Germanic, where as OPr. moke may have been borrowed from Polish. The Estonian and Livonian forms must be borrowings from Baltic, probably Latvian. It is generally agreed upon that ultimately we are dealing with a word of non-Indo-European (Mediterranean?) origin.

    Derksen misses a potential connection between *gvozdi, *gvozdь; *gvosdъ "nail" and *gvĕ̄zdà "star" (< BSl. *g/źwoizdeʔ (*g/źwoiźdeʔ): Lith. žvaigždė̃ fem.; žvaiždė̃ fem. (Old Lith., dial.); Latv. zvàigzne fem.)

    *orkỳta fem. 'brittle willow'
    E: Ru. rakíta; rokíta (dial.); Ukr. rokíta
    W: Cz. rokyta; Slk. rakyta; rokyta (popular); Pl. rokyta
    S: SCr. ràkita; Čak. Rakȉta an island; Sln. rakȋta; Bulg. rakíta

    B: Latv. ẽrcis masc. 'juniper'

    Cognates: Gkk. ἄρκευθος fem. 'juniper'

    It is very likely that we are dealing here with a substratum word, c.f. the Greek variant ἄργετος (Hesych.).

  5. Seadog Driftwood: "[...] how would you account for Winter's Law in Balto-Slavic, where, [...] 'the PIE glottalic stops dissolved into a laryngeal and a buccal part. [...]'"

    I've already written about this in Paleoglot: Winter's Law in Balto-Slavic, "Hybrid Theory" and phonation - Part 2.

    This lengthening can easily be created when the creaky phonation precedes the stop. Since creaky voiced vowels will acquire a low-frequency tone, not a high one, I assume that the resulting rising tone in Winter's Law was caused when the newly lengthened syllable (with a natural low tone due to creaky voice) had pulled the accent away from the neighbouring stressed syllable (naturally with high tone); the rising tone is what you get when you start low and go high. Another way of looking at it is that the original state of low first syllable and high second syllable became "squeezed" together into the rising tone on the first syllable.

    So basically:
    PIE *pedóm 'footstep' [pʰed̰ōm]
    > *[peə̰dōm] (so-called "preglottalization", as in "pre-creaking")
    > *[pe:dōm] (lengthening)
    > *[pé:dom] (tone retraction to rising tone)

  6. By the way, *màkъ is interesting. I've already suspected Minoan *maka 'poppy' to explain the Greek form but wasn't aware of the Slavic or Germanic forms. I would guess that the Germanic forms are late too.