The Great Proto-Indo-European Thread

[KathTheDragon]

Other than the fact that all three were relatively weak consonants by late PIE, I don't see much reason to assume that the laryngeals go back to some kind of a more general fricative series. I might have noted this already, but phonotactically there seems to be a clear divide between *s and *h₂ on one hand (often occurring in two-obstruent clusters such as *ḱs and *th₂), *h₁ and *h₃ on the other (which seem to be more common in clusters of the type *HR, *RH).

Are you saying that you think that *h₂ is the most likely to have been a fricative, with *h₁ and *h₃ more likely having been something else? *h₁ could certainly have been a glottal stop, but what could *h₃ have been?

Edit: It just occurred to me to comment that you singling out *th₂ implies that you're sourcing these distributions from the LIV - which is notorious for assuming *h₂ whenever Sanskrit displays aspirated voiceless stops.

[kanejam]

Just found this interesting paper on the d ~ h₁ alternation, which it dubs the 'Kortlandt effect' (there's an English abstract at the end) which happens to touch on a few points we have just discussed.

In passing it mentions the water root, which it reconstructs as both a singular *wódr, *wéd-n-s 'water' and a collective *wéd-ōr ,*ud-én 'waters'; whether or not the forms are correct, mixing to paradigms would be one way to account for the apparent o/zero ablaut. It doesn't mention *wóh₁r.

In addition to the evidence in numerals, *déḱm̥t vs *h₁ḱm̥tóm, the paper also equates Latin ad 'to, at' with Vedic ā 'near' (< *h₂éd=V ~ *h₂éh₁=C), they reconstruct udder as *úd=dh₁r > *úh1=dh₁r with the root *deh₁- to connect it to a couple of verbs meaning milk or suckle, and links several doublets such as *terd- and *terh₁ 'pierce', med- and meh₁ 'measure'.

It also explains several anomalous long vowels (mostly in Greek or Vedic) through reduplication: Gr. δηρις, Ved. dāri- are explained as *déh₁ri- < *dé-dr-i- from the root *derH- and Ved. dāśati 'they offer' from *déh₁knto < *dé-dk-nto among others.

The evidence is fairly convincing but the change seems very sporadic. There are plenty of little conflicts such as Ved. ta-tṛd-āná- (< *te-trd-mh₁nó-) and Gr. τετρημενος (< *te-trh₁-mh₁nó-).

[KathTheDragon]

That is indeed interesting! It can't be quite that simple, though, due to the likely doublet *dwi- "two" and *h₁wi- "apart" (a preverb). This paper by Kloekhorst does touch on the matter of *d > *h₁ (but obviously not in detail) and explains it at least in part as a dissimilation from a following dental, though he does also posit an unconditional *d > *h₁ / _ḱC. A dissimilation process would also help explain the sporadicness, which is a known property of dissimilations. Analogy can help to explain it as well, with an originally perfect distribution being obscured over time.

[Tropylium]

Well, the triple reflex does directly imply that the three laryngeals remained distinct beyond the breakup of PIE, at least within pre-Proto-Greek, so it is at least thinkable that colouring persisted as well. Also, it does seem that colouring was a synchronic rule, so if the pre-Proto-Greek speakers identified *ə with *e, then colouring (directly to the synchronic phones) is only natural.

Well, either the laryngeals themselves, or some later reflex of theirs.

Nobody would claim that *eh₂ >> Germanic *ō means that syllable-final *h₂ actually o-colored *e in Germanic; this is universally considered to have instead involved a later vowel shift from an intermediate non-PIE vowel *ā. The same approach is possible with "syllabic laryngeals", only with the complication that this time we do not have other languages that would retain clear records of distinct *ə₁ *ə₂ *ə₃.

(Within this approach I would suggest that also correspondences along the lines of *r̥ > Greek ar ~ or should be assumed to have gone thru an intermediate stage *ə₂r ~ *ə₃r; there does not seem to be any reason to introduce a fourth, "uncolored" schwa.)

Latin displays a "triple reflex" in front of syllabic nasals: *(h₁)n̥ > in, *h₂n̥ > an, *h₃n̥ > un, if memory serves.

Oh, that's interesting. Looking up Schrijver's "PIE Laryngeals in Latin" gives e.g. unguis and umbilīcus as clear examples of *h₃N̥- >> *oN- > uN-. This apparently extends to Sabellic as well: unctus ~ Umbrian umtu.

Now there's however the complication that Sabellic instead vocalizes *N̥ as aN (e.g. inter ~ Oscan anter < Italo-Celtic *ənter < *(h₁)n̥ter)! So, contrary to what Schriver suggests (the known Latin development *N̥ > *eN taking place before laryngeal coloring), this seems like strong evidence to rather reconstruct here *h₃R̥- > *h₃əR- > *ə₃R-, followed by *ə₃N > *oN (and elsewhere *ə₃ > *ə₂ > a). Laryngeal coloring taking place (or remaining active) until something like early Proto-Greek seems possible at least, but dating it as post-Proto-Italic does not.

Are you saying that you think that *h₂ is the most likely to have been a fricative, with *h₁ and *h₃ more likely having been something else?

No, just that the laryngeals are unlikely to have been a set of voiceless back fricatives, fully paralleling the stop series. They very well could have been fricatives, but without forming any kind of a clear triplet. If there is a voiceless (post-, labio-)velar in there, that's most likely *h₂. If we wanted fricative values for the others, something along the lines of h₁ = [h], *h₃ = [ɣ~ʁ] seems fairly reasonable.

[KathTheDragon]

I agree that *h₁ was more likely [h], though I would defend the claim that *h₃ was voiceless uvular.

[kanejam]

That is indeed interesting! It can't be quite that simple, though, due to the likely doublet *dwi- "two" and *h₁wi- "apart" (a preverb).

I think the word for twenty is also pretty strong evidence of the alternation within the prefix for two itself. The question is then what was the conditioning factor other than dissimilation in reduplicated forms.

I agree that *h₁ was more likely [h], though I would defend the claim that *h₃ was voiceless uvular.

I can't find the paper but I remember reading that borrowings from the Anatolian languages show that ḫ was velar or uvular, which then strongly suggests that both PIE *h₂ and *h₃ were velar or uvular, and a change into pharyngeal would be likely for the non-Anatolian languages.

Kath, you mentioned a *ḫw in Anatolian as the reason *h₃ isn't just a labialised version of *h₂. I hadn't heard of it before, what's the evidence for it?

[Zju]

That is indeed interesting! It can't be quite that simple, though, due to the likely doublet *dwi- "two" and *h₁wi- "apart" (a preverb).

I think the word for twenty is also pretty strong evidence of the alternation within the prefix for two itself. The question is then what was the conditioning factor other than dissimilation in reduplicated forms.

According to this paper (mentioned earlier somewhere in the thread) dissimilation was the only factor and later dissimilated forms spread by analogy as their origin was forgotten, especially Indo-Iranian *wi.

I agree that *h₁ was more likely [h], though I would defend the claim that *h₃ was voiceless uvular.

I found this (from the same paper) pretty convincing that *h₁ was, in fact, a glottal stop:

According to De Vaan, the preform *h₁ḱuo- may explain the presence of i- in ippos: just as in initial clusters of the type *CCC- an epenthetic -i- developed in Greek (*ptnéh₂mi > pitnemi ‘to spread out’, *skdnéh₂mi > skidnemi ‘to scatter’), so did such a vowel develop in an initial cluster of the type *h₁CC-. So, *h₁CC- yielded *h₁iCC- > (iCC-).

Note that the fact that in *h₂CC- > aCC- (e.g. aster ‘star’ < *h₂stēr) and *h₃CC- > oCC- (e.g. ossomai ‘to look’ < *h₃ḱie/o-) no epenthetic -i- emerged may be used as evidence in favor of the view that *h₁ was a (glottal) stop, whereas *h₂ and *h₃ were (pharyngeal) fricatives (cf. the fact that the fricative *s in initial position does not count as a stop when it comes to the placement of the epenthetic vowel -i-, e.g. splen, ‘spleen’ < *splV). This pattern can also be found when looking at the distribution of the laryngeals among the roots in LIV². In root-initial, preconsonantal position, *h₁ only occurs in roots of the shape *h₁ReC- and *h₁TeR-. Since these latter roots can be regarded as derivatives of roots of the structure *h₁eT-(cf. footnote 43), we see that in original roots *h₁- never occurs before stops, probably because it was a stop itself. This is different for *h₂ and *h₃, however, which not only occur in roots of the structure *h₂/₃ReC- and *h₂/₃TeR-, but also of the structure *h₂/₃Te(R)C- (*h₂teuǵ-, *h₃peus-). Therewith they pattern as *s (*sReC-, *sTeR-and *sTe(R)C-), which would fit their identification as fricatives.

Why does copying from a pdf document have to be such PITA?

[Pole, the]

Why does copying from a pdf document have to be such PITA?

It's probably because pdf is CRAP.

[jmcd]

And which file format do you prefer in that case? (Assuming you aren't taking the piss)

[garysk]

PDF(off topic) is a very versatile file format. It can store text (as text or formatted text), it can store text as binary images, it can store images as themselves, it can be password protected, both from reading and from various modifications, and it can be induced to do any and all of these things at the author's control, provided s/he has suitable software. If a document is "scanned to PDF", it almost certainly will be stored as an image, and so there is no text to select or copy. Such an image may be fed to an OCR program, and produce the text from the images, with varying quality. Good, expensive, possibly not commercially available, OCR systems can do a terrific job, cheap or free OCR programs can do a middling to horrible job. In any case, the output of an OCR system must be proofed carefully by a human, as printed material submitted as input for digitizing can be anywhere from pristine to nearly illegible. Keep in mind that "text" on a computer represents an idealized version of "writing", translating that to actual writing is a complex graphical operation, which is intended to appear as though actual metal type has been inked and pressed to paper, with all the variability that those processes can impart; undoing that via software is an even more complex operation.

If a PDF is a PITA, don't blame the format, blame the author/duplicator/process.

[Zju]

I'm talking about copying from PDF per se, without any modification it looks like this:

More: show

*
ptnéh
2
mi
>

‘to spread out’, *
skdnéh
2
mi
>
$
‘to scatter’), so did such a vowel develop
in an initial cluster of the type *
h
1
CC-
. So, *
h
1
CC-
yielded *
h
1
iCC-
>
(
CC-
.
39

And I have to painstakingly remove all the new lines and replace the faultily coppied characters with the correct ones. The PDF displays one character, but copies another one to the buffer.

[zompist]

PDF is designed for exact display, which is why it's used for printing, or to make sure documents look exactly as you want on all computers. It's not designed for word processing.

[KathTheDragon]

Kath, you mentioned a *ḫw in Anatolian as the reason *h₃ isn't just a labialised version of *h₂. I hadn't heard of it before, what's the evidence for it?

Probably the best evidence that I can think of off the top of my head is the correspondence of the Hittite participle tarhunt- with Lycian Trqqñt-. On the basis of the base verb showing spellings like ta-ru-uḫ-zi alongside tar-ḫu-uz-zi, completely parallel to e-uk-zi alongside e-ku-uz-zi, Hittite must possess a labialised laryngeal. As shown by Kloekhorst, Lycian q is also labialised, and since these correspond exactly, it's very likely that it was already Proto-Anatolian.

[Zju]

According to Wiktionary proto Germanic *bedō (“request, plea”) comes directly from PIE *gʷʰedʰ-eh₂. Since when is *gʷʰ → *b established for PGm? Shouldn't that rather be considered a loan from proto Celtic?

[Sumelic]

According to Wiktionary proto Germanic *bedō (“request, plea”) comes directly from PIE *gʷʰedʰ-eh₂. Since when is *gʷʰ → *b established for PGm? Shouldn't that rather be considered a loan from proto Celtic?

Is there any reason to suppose it wasn't a PGm change? There also seems to be "https://en.wiktionary.org/wiki/Reconstr ... ean/gʷʰen-"

[KathTheDragon]

Because the change *gʷʰ > *b / #_ only has a very few examples, many of which can be explained otherwise, while *gʷʰ > *w (which is the default in every other position) does have examples word-initially, like *warmaz < *gʷʰormos

[Sumelic]

I see. There's some discussion by David Marjanović in the following Languagehat thread: http://languagehat.com/sounds-and-meanings-revisited

[KathTheDragon]

Interesting! He cites a paper (which I unfortunately could not access, and seems to have been in Russian anyhow) and summarises it as saying that *gʷʰ > *b / #_ is irregular, and for all positions in the word, *gʷʰ > *g before a stressed vowel, and *gʷʰ > *w everywhere else (although it should be mentioned that the latter change is blocked after *n). Without having the paper, I'll have to investigate this for myself at some point.

[Zju]

I've been thinking of starting my own (semi-)serious research on PIE and or PIU. The first step in any case would be compiling materials and I've been wondering if it's cool to redistribute what I've compiled. As in, if only one IE-ist has reconstructed a paradigm in a certain way and that reconstruction is his, is it okay to copy it on the basis that I've read and learnt it?

I'd be publishing my compilation and results in another thread or maybe even a blog or a quickie website.

[KathTheDragon]

It's probably ok if you give a citation.

[WeepingElf]

Is the idea plausible that the verbal thematic vowel originally was a transitivity marker, perhaps ultimately the same morpheme as 3sg. stative *-e, as a last trace of a lost bipersonal paradigm?

[hwhatting]

Is the idea plausible that the verbal thematic vowel originally was a transitivity marker, perhaps ultimately the same morpheme as 3sg. stative *-e, as a last trace of a lost bipersonal paradigm?

I know scholars who argue that the thematic paradigms are linked to stative / mediopassive, but that would IMO argue against it being a transitivity marker, because one would expect intransitive verbs to be more typical for the stative class.

[WeepingElf]

Is the idea plausible that the verbal thematic vowel originally was a transitivity marker, perhaps ultimately the same morpheme as 3sg. stative *-e, as a last trace of a lost bipersonal paradigm?

I know scholars who argue that the thematic paradigms are linked to stative / mediopassive, but that would IMO argue against it being a transitivity marker, because one would expect intransitive verbs to be more typical for the stative class.

According to this paper (page 18), most thematic verbs are transitive, and those that are not, are change-of-state verbs. Yet, this does not imply that the thematic vowel once was a transitivity marker (why then should change-of-state verbs have it?). Also, of course, the thematic verbs, just like the thematic nouns, are a late formation and entirely the wrong place to look for archaisms.

[hwhatting]

Is the idea plausible that the verbal thematic vowel originally was a transitivity marker, perhaps ultimately the same morpheme as 3sg. stative *-e, as a last trace of a lost bipersonal paradigm?

I know scholars who argue that the thematic paradigms are linked to stative / mediopassive, but that would IMO argue against it being a transitivity marker, because one would expect intransitive verbs to be more typical for the stative class.

According to this paper (page 18), most thematic verbs are transitive, and those that are not, are change-of-state verbs. Yet, this does not imply that the thematic vowel once was a transitivity marker (why then should change-of-state verbs have it?). Also, of course, the thematic verbs, just like the thematic nouns, are a late formation and entirely the wrong place to look for archaisms.

So you're arguing against the idea that the thematic vowel is "ultimately the same morpheme as 3sg. stative *-e"?

[WeepingElf]

Is the idea plausible that the verbal thematic vowel originally was a transitivity marker, perhaps ultimately the same morpheme as 3sg. stative *-e, as a last trace of a lost bipersonal paradigm?

I know scholars who argue that the thematic paradigms are linked to stative / mediopassive, but that would IMO argue against it being a transitivity marker, because one would expect intransitive verbs to be more typical for the stative class.

According to this paper (page 18), most thematic verbs are transitive, and those that are not, are change-of-state verbs. Yet, this does not imply that the thematic vowel once was a transitivity marker (why then should change-of-state verbs have it?). Also, of course, the thematic verbs, just like the thematic nouns, are a late formation and entirely the wrong place to look for archaisms.

So you're arguing against the idea that the thematic vowel is "ultimately the same morpheme as 3sg. stative *-e"?

I am not sure. This identity of morphemes is an old idea of mine which I have grown mildly skeptical about (given that the thematic conjugation apparently is not old), but I just don't know what I should think about it. It may be that in the Late PIE thematic conjugation, two different conjugations were confused, namely the stative conjugation which we find in Late PIE middles and perfects, and an originally bipersonal transitive conjugation. Both conjugations would have been used on the same verbs - the transitive conjugation in the active, and the stative conjugation in the middle/passive (and the athematic active conjugation in the antipassive). The stative conjugation would also have been used, of course, on verbs denoting states and verbs denoting changes of state, i.e. on verbs which do not express an action performed by the subject. Thus, this early stage of PIE would have had an active-stative verb conjugation.

Yet, we have no evidence other than perhaps the thematic vowel itself in any attested IE language, not even in Hittite.