Angelo of Sauvage Noble has translated Hamlet’s soliloquy into Proto-Indo-European, as “H₃regs suhnus gʷʰn̥ntosyo” (The Slain King’s Son). It begins:
eg̑oh₂ h₁esoh₂? way! ne h₁esoh₂? h₁r̥h₁yoh₂er:
upo de melyos teh₂ smereses bʰeroh₂
mn̥teyi Hih₁tleh₂ dusmeneses smr̥tos,
kʷoynoybʰos wē toybʰos tl̥neh₂oh₂ h₁r̥meh₂,
h₂enti yeh₂ stisth₂ents peh₂woyh₁m̥?
Or, in what he aptly calls Old High Translationese:
Should I be? Alas! Should I not? I ask myself:
shall I, having been allotted, better suffer in (my) mind
those missiles of ill-disposed fate?
or should I raise arms to those troubles
which, standing against them, I might stop?
Very enjoyable for this Indo-Europeanist manqué!
Tip for easier reading: just ignore the various hs, which represent the laryngeals (nobody knows how to pronounce them anyway): “Eg̑o eso? way! ne eso? r̥yoer…”
Obercool, thank you very much!
I sometimes pronounce the laryngals as Arabic ع. Freaks the hell out of other Indo-Europeanists :o)
I usually pronounce them like ﺡ, myself.
According to Craig Melchert at a conference last weekend, the best guess is that h1 was /h/, and h2 and h3 were phayngeal fricatives.
N.B.: That’s an unattested soliloquy. 😉
I like the simplicity of Rasmussen’s view, where the consonant allophones are [h], [x], [ɣʷ] and the corresponding vowel allophones are [ə], [ɐ], [ɵ]. These are the least unusual sounds that satisfy the conditions: (1) that /h₁/ is as neutral as possible; (2) that /h₂/ is a-coloring, voiceless and unrounded; (3) that /h₃/ is o-coloring, voiced, and rounded; and (4) that the vowels are simply the consonants after deaffrication. The disadvantage of this scheme is that you need a pronunciation of /H/, the ambiguous laryngeal: my inclination is to pronounce it as /h₂/, because /h₁/ is too hard to hear and /h₃/ is too specialized.
Velars are quite unlikely because they wouldn’t color enough. I bet on uvulars. There are several other arguments for this as well (foreign transcriptions of Anatolian words, developments within Anatolian).
Rounding in *h₃ is only postulated because of the o-coloring. It doesn’t fit the Anatolian evidence (see link above) and fails to explain why the entire labialized velar series, */w kʷ gʷ gʱʷ/, caused no coloring whatsoever. More likely, I would say, is that o-coloring dates from a time before *o was rounded: perhaps *h₂ originally colored toward [ɐ] or [ʌ] and *h₃ toward [ɑ]. (…and I just noticed that in a comment to that post I had a good, though vague, idea how such a split might theoretically happen.) The paper linked above also points out there’s no reason to think *o, once it was phonetically rounded, was phonologically specified as rounded: rather, it was rounded because it was back (and there wasn’t a specifically unrounded back vowel to contrast with).
A further argument against *[ɣʷ] is that there’s pretty good evidence that the inordinately common *h₂w was actually (at least in most cases) a unit phoneme, *h₂ʷ, *[χʷ]. (This is uncontroversial for Anatolian, where the reflex of *h₃ was, again, not rounded.) That makes it likely that the same distinction existed between *h₃ *[ʁ] and *h₃ʷ *[ʁʷ]. Due to the rarity of *h₃ in current reconstructions, though, more research is definitely needed. It is also easily possible that */ʁʷ/ merged into */w/ very early – but, again, *h₃ never did.
Anyway, *h₂ is so common that most cases of *H were probably *h₂.
[h] for *h₁ gets a big boost from Bozzone’s law; the evidence that the difference between the Hieroglyphic Luwian signs a and á (before they became hopelessly confused in the mid-9th century BC) was that á began with a consonant that is a direct reflex of *h₁ makes a lot more sense for [h] than for [ʔ].
Velars are quite unlikely because they wouldn’t color enough. I bet on uvulars.
It’s a matter of degree. I’m definitely okay with /χ/ instead of /x/ (and most of the time I tend to articulate /x/ as /χ/ anyway unless I am careful).
Kümmel’s claim for an origin in uvular stops sounds unnecessary to me, though: bad enough we have to postulate three post-coronal stop series, and he wants to make it four? Naah. His only real argument is that stop > fricative is more natural than fricative > stop, but that’s a statistical argument only. The process changing interdental fricatives into dental stops is so probable that except for some weird people in little islands off the edge of civilization (and even they are starting to lose them) and a bunch of crazy mountaineers, it affects everybody.
it was rounded because it was back
Or vice versa. It’s impossible to say which. Most five-vowel systems lock backness to roundness irrevocably.
more research is definitely needed
Agreed; h₃ is a mess.
Kümmel cites and comments on it, but doesn’t make it except for allowing it as a (likely) possibility for some pre-PIE stage.
I continue to have a suspicion that if there is some part of various versions of laryngeal theory that is really just a misunderstanding of how ablaut operates, it’s most likely *h₃ — I wouldn’t even flinch too much to see a reconstruction that has still three laryngeals, but they’re *h₁ *h₂ *h₄, and classical *h₃ has been replaced with something like *h₄ + *o-grade.
For those not acquaintanced, *h₄ supposedly differs from *h₂ in not surviving as a consonant in Albanian or Armenian or some others of those languages with more marginal lines of direct evidence. If *h₂ was [q], then that would probably open up also a good phonetic identity, *h₄ = [χ]: similar coloring expected, further along on lenition already. (I don’t think I’ve seen any all-fricatives proposal that would look good; some attempts of [χ] versus [ħ] in either order.)
Conversely, I wonder if the frequency of *h₃ has been dramatically underestimated because in prevocalic position it doesn’t leave any trace other than *o, which could usually just be an o-grade, or a voiced plosive, which were in the inventory anyway.
…except *b, of course, but that’s too rare to do statistics with, I’m afraid. *h₃ is very useful in explaining *pib- as the reduplicated version of *poh₃-, i.e. *pi-ph₃-; the only other example I’m aware of that’s accepted by more than one person is Celtic *abonā “river” as containing the “Hoffmann suffix” (which may or may not be the same as Latin onus “burden”): *h₂ap-h₃on- “with water”.
However… why does everyone seem to be assuming that all laryngeals disappeared immediately when Germanic branched off? If they were still there when Grimm’s and Verner’s laws struck, suddenly it makes sense to interpret Holtzmann’s and Lühr’s/Müller’s laws as simply Grimm’s and Cowgill’s as Verner’s. It may be telling that all the examples given here (pp. 52–54) for Holtzmann’s & Lühr’s/Müller’s have the voiceless *h₁ or *h₂. (Cowgill’s follows on pp. 56–57; *h₂ is represented, the first example is usually reconstructed with *h₃ but in this paper with *h₁ for reasons perhaps given elsewhere, and a possible example with *h₃ is such a mess that it’s discussed at the end with no conclusion.) Similarly, the exceptions to Verner’s law all seem to have *h₁ or *h₂.
The other way around: it does not survive in Anatolian (in writing anyway), but does supposedly survive as h in Albanian or Armenian… but not both in any single one of the different proposals, AFAIK.