zompist bboard

Nessari wrote:I never said v>b / #_, _#, _C, C_.

In Baltic, Slavic, and German what jmcd calls "v" was pronounced in those positions. In Baltic and Slavic furthermore "v" was pronounced even in the intervocalic position.

All these point to a reconstruction of jmcd's "v" as /b/.

You're being incredibly obtuse so I'm done with this.

Hi pot, I'm kettle, pleasure to meet you.
PIE *bh is in most of those positions in Germanic as well. It's perfectly plausible to reconstruct a phoneme *b~ƀ there, which was a stop initially, next to an obstruent, and finally (if it occurred there, I can't remember and it's not worth looking up), and was a fricative intervocalically. In the hypothesis we're discussing, this is supported by fricatives showing up in a lot of those same positions in Latin (though admittedly not all, the dative-ablative plurals in -Vbus spring to mind, but then again Germanic doesn't retain those, so moot point i guess), and the phoneme subsumed solely into the stop in Balto-Slavic (indeed as it did in most of Germanic, eventually.) I'm not saying this definitively was what the situation was, but your objections to it are entirely without merit.

In Latin, the fricatives /f/ and /h/ only appear word-initially though. Elsewhere it's /b/, /d/, /g/ and /w/ (also /gw/ after nasals). In Osco-Umbrian the fricatives appear in all positions, even intervocally (and after nasals? too lazy to look it up). So Latin -Vbus corresponds to Oscan (or was it Umbrian? again, too lazy to look it up) -Vfs. Evidence enough that all "voiced aspirates" turned into fricatives (probably voiced) in proto-Italic, and (more importantly) were distinct from the voiced stop series.

(This doesn't make Rusanov's objections any more valid though.)

Side-note: I think voiced aspirates could only occur word-finally in vocatives of certain root nouns, e.g. *sneyg^wh "O snow". Very rare indeed.

Sleinad Flar wrote:In Latin, the fricatives /f/ and /h/ only appear word-initially though. Elsewhere it's /b/, /d/, /g/ and /w/ (also /gw/ after nasals). In Osco-Umbrian the fricatives appear in all positions, even intervocally (and after nasals? too lazy to look it up). So Latin -Vbus corresponds to Oscan (or was it Umbrian? again, too lazy to look it up) -Vfs. Evidence enough that all "voiced aspirates" turned into fricatives (probably voiced) in proto-Italic, and (more importantly) were distinct from the voiced stop series.
(This doesn't make Rusanov's objections any more valid though.)

Thank you for the clarification; for the record, the Germanic voiced stop~fricative series is descended from the PIE voiced aspirate series, while the PIE plain voiced series (excepting cases where Verner's Law applied) devoiced to p t k kʷ.

Side-note: I think voiced aspirates could only occur word-finally in vocatives of certain root nouns, e.g. *sneyg^wh "O snow". Very rare indeed.

Sleinad Flar wrote:In Latin, the fricatives /f/ and /h/ only appear word-initially though. Elsewhere it's /b/, /d/, /g/ and /w/ (also /gw/ after nasals). In Osco-Umbrian the fricatives appear in all positions, even intervocally (and after nasals? too lazy to look it up). So Latin -Vbus corresponds to Oscan (or was it Umbrian? again, too lazy to look it up) -Vfs. Evidence enough that all "voiced aspirates" turned into fricatives (probably voiced) in proto-Italic, and (more importantly) were distinct from the voiced stop series.

Yes. I am of the opinion that the fricatives indeed were voiced in Proto-Italic, which would also explain how /f/, which would actually have been something like /β/ at that time, would be spelled with a letter that represented /w/ in the West Greek alphabet. In Latin, those fricatives devoiced in initial position and merged with the voiced stops in medial position; in Sabellic, they remained distinct from the voiced stops in medial position, perhaps devoicing there as well, perhaps not. At any rate, they were still voiced when the Greek brought in their alphabet.

WeepingElf wrote:

Sleinad Flar wrote:In Latin, the fricatives /f/ and /h/ only appear word-initially though. Elsewhere it's /b/, /d/, /g/ and /w/ (also /gw/ after nasals). In Osco-Umbrian the fricatives appear in all positions, even intervocally (and after nasals? too lazy to look it up). So Latin -Vbus corresponds to Oscan (or was it Umbrian? again, too lazy to look it up) -Vfs. Evidence enough that all "voiced aspirates" turned into fricatives (probably voiced) in proto-Italic, and (more importantly) were distinct from the voiced stop series.

Yes. I am of the opinion that the fricatives indeed were voiced in Proto-Italic, which would also explain how /f/, which would actually have been something like /β/ at that time, would be spelled with a letter that represented /w/ in the West Greek alphabet. In Latin, those fricatives devoiced in initial position and merged with the voiced stops in medial position; in Sabellic, they remained distinct from the voiced stops in medial position, perhaps devoicing there as well, perhaps not. At any rate, they were still voiced when the Greek brought in their alphabet.

While I don't disagree that this is the most likely path*, that's not really evidence for it. What other letter would they have used? As we said before, Hurrians spelled /f/ as interchangeably either <p>, <b> or <w> (in the same words), yet it's not thought to be /v/. People just grab whatever symbol or symbols they think are closest, or even just whichever symbols aren't being used for something else (eg Bantu languages spelling clicks with <C>), without too much worry whether the exact sounds match perfectly in all regards. You can only draw conclusions if they chose between two equally unused symbols, one for /f/ and one for /w/, and picked the /w/ one - and even then it may be because of the labiovelarisation rather than because of the voicing.

*not the only way, though. An alternative would be devoicing of initial voiced aspirates, then turning voiceless aspirates into fricatives - both phases of that are reasonable enough. The remaining voiced aspirates would then eventually merge with unaspirate voiced stops in Latin - the other Italic evidence showing them as fricatives could be either a secondary transformation into fricatives not shared by Latin, or simply an orthographic decision, with fricative symbols representing a class of voiced aspirates for which there were no dedicated letters available.

Nessari wrote:<...> the PIE plain voiced series (excepting cases where Verner's Law applied) devoiced to p t k kʷ.

Verner's law didn't apply to this series. I am sorry.

Basilius wrote:

Nessari wrote:<...> the PIE plain voiced series (excepting cases where Verner's Law applied) devoiced to p t k kʷ.

Verner's law didn't apply to this series. I am sorry.

I guess there were no such cases then

Which "such"?

Verner's law is about the f T x -> B D G change (word-internally).

More specifically, it's about a subset of exceptions to that change.

Also, the change (with the exceptions) arguably wasn't Proto-Germanic, although it was/is common Germanic.

PGerm f T x have very little to do with the traditional plain voiced series of PIE.

I am sorry again if I missed your point.

WeepingElf wrote:I am of the opinion that the fricatives indeed were voiced in Proto-Italic, which would also explain how /f/, which would actually have been something like /β/ at that time, would be spelled with a letter that represented /w/ in the West Greek alphabet.

The Italic alphabets went thru Etruscan too though, and its ‹FH› for /f/ could have been a model also for a voiceless value of ‹F›.

The Latin development *-sr- → /-br-/ seems to demonstrate that it has fortified Proto-Italic fricatives, not retained PIE stop values. It doesn't follow though if this was thru *-θr- → *-fr- or *-zr- → *-ðr- → *-vr-.

Osco-Umbrian has also some evidence for voiceless values specifically, e.g. *ts → *θ → /f/ (in one of the langs, don't recall right now which one). This also could still have been secondary, I suppose.

Basilius wrote:Which "such"?

Verner's law is about the f T x -> B D G change (word-internally).

More specifically, it's about a subset of exceptions to that change.

Also, the change (with the exceptions) arguably wasn't Proto-Germanic, although it was/is common Germanic.

PGerm f T x have very little to do with the traditional plain voiced series of PIE.

I am sorry again if I missed your point.

Nessari wrote:

Basilius wrote:

Nessari wrote:<...> the PIE plain voiced series (excepting cases where Verner's Law applied) devoiced to p t k kʷ.

Verner's law didn't apply to this series. I am sorry.

I guess there were no such cases then

I've bolded and italicized the relevant related phrases.

Drydic: ah, OK then.

Tropylium: ... → *-vr- → -br- does not seem to imply that all other b's were *v before the last change. Strictly speaking.

Tropylium wrote:The Latin development *-sr- → /-br-/ seems to demonstrate that it has fortified Proto-Italic fricatives, not retained PIE stop values. It doesn't follow though if this was thru *-θr- → *-fr- or *-zr- → *-ðr- → *-vr-.

Don't some Celtic languages have sr > fr?

Basilius wrote:Tropylium: ... → *-vr- → -br- does not seem to imply that all other b's were *v before the last change. Strictly speaking.

Strictly speaking no, indeed. It does mean that there was a development *-v- → -b-. This in hand, *bʰ becoming /f-/ ~ /-b-/ could be "freely" rewinded to the less outstanding /f-/ ~ *-v-.

(Side note, I wonder if the widespread Romance lenition of medial voiced stops could have been in part a retention. Classical Latin merging *-v- *-b- as , Vulgar Latin as [β]?)

A slightly weaker version of this argument can be also reapplied. *s → *-z- → /-r-/ shows that a intervocalic fricative voicing rule existed. So if a single set of voiceless fricatives *f *θ *s *x *xʷ is set up for Proto-Italic, the Latin representation is derivable from it with minimal assumptions.

The general picture of Latin could be derived also by Sal's initial spirantization/medial deaspiration suggestion, with just as much ease — but this does not seem to work for Osco-Umbrian and Faliscan. Moreover, Latin also has another relevant conditional development, *-dʰr- → /-br-/ as in e.g. *h₁rudʰros → rubrus, ruber (and IIRC *dʰ → /b/ in a couple other similar environments too). This seems to call for specifically the interdental → labiodental shunt seen with *sr, since there is no change *-tr- → ˣ-pr- or *-dr- → ˣ-br-.

Nortaneous wrote:

Tropylium wrote:The Latin development *-sr- → /-br-/ seems to demonstrate that it has fortified Proto-Italic fricatives, not retained PIE stop values. It doesn't follow though if this was thru *-θr- → *-fr- or *-zr- → *-ðr- → *-vr-.

Don't some Celtic languages have sr > fr?

Yes, it seems, at least word initially.
PIE *sregh- -> Gaul *frogná, OBret fron, MW ffroen, OIr srón
PIE *srew- -> OCo frot, OBret frut, MW ffrwd, OIr sruth
PIE *tem-[esla?] -> Lat tenebrae, OCo. tiwoulgou, OBret. temoel, OW timuil, W tywyll, Olr. temel

Cf. also Gaul.
Phroudis [Hydronym], which is sometimes claimed to show that Gaulish
shared the Brittonic change of *sr- > *fr-.

(All from Leiden's Celtic Etym Dict)

Are interdentals preserved anywhere? Is there *sr- θr-?

Basilius wrote:Drydic: ah, OK then.

To be fair, I should have realized your point myself when I originally wrote that, but I was freshly woken up on not near enough sleep. At the least I should have checked, but I didn't, so my apologies.

Sidenote on this: It'd be awesome if someone updated Prokosch's A Comparative Germanic Grammar to not use two fricative series for its PIE reconstruction (in place of the voiced aspirate series and Szemerényi's voiceless aspirates.) Maybe Ringe is good enough for that, I really need to get back to going through that.

Tropylium wrote:

WeepingElf wrote:I am of the opinion that the fricatives indeed were voiced in Proto-Italic, which would also explain how /f/, which would actually have been something like /β/ at that time, would be spelled with a letter that represented /w/ in the West Greek alphabet.

The Italic alphabets went thru Etruscan too though, and its ‹FH› for /f/ could have been a model also for a voiceless value of ‹F›.

The Latin development *-sr- → /-br-/ seems to demonstrate that it has fortified Proto-Italic fricatives, not retained PIE stop values. It doesn't follow though if this was thru *-θr- → *-fr- or *-zr- → *-ðr- → *-vr-.

Osco-Umbrian has also some evidence for voiceless values specifically, e.g. *ts → *θ → /f/ (in one of the langs, don't recall right now which one). This also could still have been secondary, I suppose.

Fair. As Salmoneus said, the pathway via voiceless aspirated stops is perfectly plausible, after all, just that has happened one peninsula further east - in Greek. The earliest Italic inscriptions indeed spell /f/ as FH, which would be a natural way of spelling a "voiceless counterpart of /w/", so the argument from the script I gave is invalid. Only later, the H was dropped from the digraph as F was only used in this digraph and the H was therefore redundant.

EDIT: There still is the issue that Proto-Italic */f/ gives /b/ in medial position in Latin. So it may have been voiced in that position?

Nortaneous wrote:Are interdentals preserved anywhere? Is there *sr- :> θr-?

Not that I see. However, the introduction only covers sound changes from PIE to PCelt (not PCelt to whatever), so I don't know.

There is PIE/PCe *tisres > Gaulish tidres (Old Irish teoir) "three" (fem.) I guess that's close enough.

Sleinad Flar wrote:There is PIE/PCe *tisres > Gaulish tidres (Old Irish teoir) "three" (fem.) I guess that's close enough.

Not really: as far as we can determine the <ð> of <tiðres> probably represented /ts/, not /θ/.

WeepingElf wrote:What speaks against your proposed correspondence are questions of frequency and markedness. The PIE voiceless stops are the most frequent and least marked stops in the language. Uralic geminates, in contrast, are highly marked and much less common. Hence, it is very unlikely that these two classes of stops correspond to each other directly.

That's a good point. This could be taken into account by only some *p, *t and *k corresponding to *pp, *tt and *kk. The scenario could be somewhat similar to the exceptions to Grimm's Law: The voiceless stops remained voiceless stops in various clusters, especially next to *s. Then again, given what Tropylium says about geminates actually being clusters, it may well simply be that the Uralic branch of Indo-Uralic abandoned voicing distinction.

WeepingElf wrote:I don't know where the Celtic t-preterite came from; it may indeed be a parallel development to the Germanic weak preterite, which is AFAIK not well-understood either.

I know that the Germanic weak preterite is understood as being a replacement of the PIE reduplicated forms, one of the few of which that survived was "did". And from there, the -d- forms spread by analogy into other verbs. Thus there is some similarity between the development of the Gaulish and the Germanic coronal preterite but it could simply be parallel development. On the other hand, they could have arisen next to one another within the same time period at which point they could have encouraged one another.

What about the coronal past in Iranian?

Astraios wrote:What about the coronal past in Iranian?

That's just the passive participle suffix in *-t-os.

jmcd wrote:

WeepingElf wrote:What speaks against your proposed correspondence are questions of frequency and markedness. The PIE voiceless stops are the most frequent and least marked stops in the language. Uralic geminates, in contrast, are highly marked and much less common. Hence, it is very unlikely that these two classes of stops correspond to each other directly.

That's a good point. This could be taken into account by only some *p, *t and *k corresponding to *pp, *tt and *kk. The scenario could be somewhat similar to the exceptions to Grimm's Law: The voiceless stops remained voiceless stops in various clusters, especially next to *s. Then again, given what Tropylium says about geminates actually being clusters, it may well simply be that the Uralic branch of Indo-Uralic abandoned voicing distinction.

This is of course possible, and from my first attempts at Indo-Uralic comparison, it seems like all IE stops correspond to single voiceless stops in Uralic, with no easily discernible criteria as to what kind of MOA surfaces in IE. So far, it looks as if Uralic simply collapsed the three sets of stops into one (as happened in Tocharian).

jmcd wrote:

WeepingElf wrote:I don't know where the Celtic t-preterite came from; it may indeed be a parallel development to the Germanic weak preterite, which is AFAIK not well-understood either.

I know that the Germanic weak preterite is understood as being a replacement of the PIE reduplicated forms, one of the few of which that survived was "did". And from there, the -d- forms spread by analogy into other verbs. Thus there is some similarity between the development of the Gaulish and the Germanic coronal preterite but it could simply be parallel development. On the other hand, they could have arisen next to one another within the same time period at which point they could have encouraged one another.

Yes. The Germanic strong preterites continue Late PIE perfects which, however, have shed their reduplication for some reasons. For the many verbs which lacked a perfect in Late PIE, Germanic innovated the weak preterite, apparently by suffixing a perfect of 'to do' - i.e., what became the form did in English - to the verb stems. Something similar may have happened in Celtic, but I have read somewhere that the Celtic t-preterites came from a generalization of the 3sg. *-t throughout the paradigm. Alas, I know too little about these matters to say anything of value. Ask Dewrad.