The Great Proto-Indo-European Thread

[jal]

The main point is most languages do not have one single ancestor

Exactly on-one here is claiming that, so you are again strawmanning.

So Dutch is not the only ancestor of Afrikaans

Again, no-one is claiming that, here or elsewhere. It is well known Afrikaans is partially creolised and has native African and Dutch Indian influences.

west-germanic is not the only ancestor of Dutch and proto-germanic is not the only ancestor of west-germanic.

Those are rather bold statements, that require some proof. I personally have never heard of non-germanic substrates in Dutch or west germanic.

But even then, who says that Portuguese was one monolithic language at the time?

Again, strawmanning.

Look, apart from the fact that you obviously lack any deeper knowledge on the things you discuss, and apart from the fact you are apparently completely blind to this (Dunning-Kruger anyone?), you keep on strawmanning time after time. Stop doing that! It is extremely annoying.


JAL

[jmcd]

@sirdanilot: You must distinguish between the language(s) of the founding population and later influences. In that way, one could say that Afrikaans has multiple origins but not Dutch.

[sirdanilot]

Yes, but my pointis that a situation such as Afrikaans is not very a typicaal. A population moving away into territory where unrelated languages are spoken and not having contact with the original population, that is quite normal. The original population here is not a monolithic language entity but a concoction of dialects and languages. Why would the construct we call 'proto Germanic ' not actually have been a collection of dialects and perhaps languages, with different mixtutes of people moving in different directions losing most contact with each other ?

[jal]

Why would the construct we call 'proto Germanic ' not actually have been a collection of dialects and perhaps languages

That has been explained both by me (because what we call Proto-Germanic is, by definition, a single language) and KathAveara. You seem to have problems with reading, as well as reasoning.


JAL

[Plaas]

Those are rather bold statements, that require some proof. I personally have never heard of non-germanic substrates in Dutch or west germanic.

There is a claim that the Germanic languages were influenced by a non-IE language, see Germanic substrate hypothesis; for Dutch, there many Old-French influences, and a disputed possible Celtic substrate and (Germanic) Frisian substrate, depending on the region.

[jal]

There is a claim that the Germanic languages were influenced by a non-IE language

I am aware of this, but then we're talking Proto-Germanic, not Dutch or West Germanic.

for Dutch, there are many Old-French influences

That would be strange, as Old French is influenced quite a bit by Old Frankish, itself an ancestor of Dutch. You are claiming the influence extended the other way as well?

and a disputed possible Celtic substrate

Dutch itself? I can imagine West Germanic has, but Dutch?


JAL

[Plaas]

for Dutch, there are many Old-French influences

That would be strange, as Old French is influenced quite a bit by Old Frankish, itself an ancestor of Dutch. You are claiming the influence extended the other way as well?

It is possible that they developed some new features together, yes. Language contact is of all times. A well known example is the shift /er/ to /jè/ in Brabantic dialects, with peerd -> pjèd in the Dutch, perdre -> pjed-re in the French dialects (Leuven region). A possible older example is the methatesis of Latin verbal prefix re- ("again") to /er-/ in Picardic dialects, borrowed as /her/ in Southern Dutch, that became part of the written standard in the North as well, altough it is not as normal as in the South (think of S "hervulbaar" vs. N "navulbaar" for refillable). There are many French morphemes that became productive in Dutch, like -aard (< -ard), -ier, -ij (<ie).

and a disputed possible Celtic substrate

Dutch itself? I can imagine West Germanic has, but Dutch?[/quote]

Some linguists considered shifts like ft -> cht and ol -> ou- to be Celtic (I never found it very convincing, though).

[Pabappa]

The Proto-Celtic shift of ft > xt is at least 1000 years older than that of Dutch, and was part of a wider shift of all other plosives to /x/ before /t/, and /f/ was a plostive because it came from /p/. Unless a similar shift happened within Celtic later on I cant see any way this could have influenced Dutch.

[jal]

It is possible that they developed some new features together, yes. Language contact is of all times.

Yes, but the movement was South, where the Franks influenced French, then started speaking French themselves (or the other way around). Northern Frankish developed independently.

There are many French morphemes that became productive in Dutch, like -aard (< -ard), -ier, -ij (<ie).

But are those Old French morphemes, or later ones? It is well known Dutch has been influenced by French, which makes sense as they share a language border, but I'm not too sure this is Old French / Old Dutch as opposed to Middle French / Middle Dutch.


JAL

[Salmoneus]

It seems reasonable to assume there was at one time a celtic substrate for dutch or its predecessor. After all, if not celtic then what? The celts appear to have been neighbours to the germans, and the Netherlands are outside the core germanic region, so it's likely that sometime around 1000BC say there was a germanic push (genetic and/or linguistic) into celtic-occupied territory.

Of course, whether such a substratum had any detectable long-term influence is another matter.

[Plaas]

There are many French morphemes that became productive in Dutch, like -aard (< -ard), -ier, -ij (<ie).

But are those Old French morphemes, or later ones? It is well known Dutch has been influenced by French, which makes sense as they share a language border, but I'm not too sure this is Old French / Old Dutch as opposed to Middle French / Middle Dutch.

These morphemes were borrowed in the early Middle Dutch period (written evidence for -aard or -ij dates back to the 13th century), maybe earlier.
The traditional dating of Old Dutch/Middle Dutch and Old French/Middle French causes overlaps (Middle Dutch: early 13th century), Middle French: mid 14th century), this can be confusing. But there is no doubt Dutch was influenced by French and is therefore not a pure stam tree language, there are non-Germanic influences and they are pretty old. It's a member of the SAE-Sprachbund as well, of course, it shares many developments with other Western-European languages, Germanic and non-Germanic.

[WeepingElf]

It seems reasonable to assume there was at one time a celtic substrate for dutch or its predecessor. After all, if not celtic then what? The celts appear to have been neighbours to the germans, and the Netherlands are outside the core germanic region, so it's likely that sometime around 1000BC say there was a germanic push (genetic and/or linguistic) into celtic-occupied territory.

Of course, whether such a substratum had any detectable long-term influence is another matter.

There is also the "Nordwestblock" hypothesis according to which an IE language that was neither Germanic nor Celtic was wedged between Germanic and Celtic, extending eastwards to the Aller river in Lower Saxony and southwards to the Somme river in northern France. This language is sometimes also called "Belgian".

[GreenBowTie]

It seems reasonable to assume there was at one time a celtic substrate for dutch or its predecessor. After all, if not celtic then what? The celts appear to have been neighbours to the germans, and the Netherlands are outside the core germanic region, so it's likely that sometime around 1000BC say there was a germanic push (genetic and/or linguistic) into celtic-occupied territory.

Of course, whether such a substratum had any detectable long-term influence is another matter.

There is also the "Nordwestblock" hypothesis according to which an IE language that was neither Germanic nor Celtic was wedged between Germanic and Celtic, extending eastwards to the Aller river in Lower Saxony and southwards to the Somme river in northern France. This language is sometimes also called "Belgian".

hoooooooooooooooly shit

[linguofreak]

I thought up an interesting potential reconstruction of the PIE obstruents:

To start with, I'm using a notation in which I use one letter to label a point of articulation and another to notate a manner of articulation.

I assume a fairly traditional reconstruction of the points of articulation: labial, alveolar, velar, uvular, and labiovelar/labiouvular, which I'll label P, T, K, Q, and W.

I propose four "phonetic" manners of articulation and four "phonemic".

The "phonetic" manners of articulation are unvoiced Stop, voiced Ztop, unvoiced Fricative, and voiced Vricative.

The "phonemic" manners of articulation are voiced sDops and unvoiced stOps, "voiced" Aspirates, and Laryngeals.

My reconstruction assumes that S = O and Z = D, and that these correspond directly to the voiceless and voiced stops in traditional reconstructions of PIE.

The fricatives are postulated to have had the following phonetic values:

PF = [ɸ]
TF = [s]
KF = [x]
QF = [χ]
WF = [χʷ]
PV = [β]
TV = [z]
KV = [ɣ]
QV = [ʁ]
WV = [ʁʷ]

The laryngeals correspond to the following phonemes in traditional reconstructions:
PL, KL = /h₁/
TL = /s/
QL = /h₂/
WL = /h₃/

PL and KL are sounds that I postulate were separate in PIE, but were both lost and did not color vowels, and so cannot be distinguished today, both being reconstructed as h₁. I'll explain a bit later why I postulate two separate sounds where one has generally been seen as sufficient to explain the data.

/s/ is not traditionally regarded as a laryngeal, but I propose that it shared a manner of articulation with the laryngeals.

The voiced aspirates correspond to the sounds reconstructed as such in traditional reconstructions. (PA = bʰ, etc)

My reconstruction then posits that one of the following is true:

L = F and A = V

or

L = V and A = F

That is, the laryngeals and the "voiced aspirates" are reconstructed as voiced and voiceless fricatives, leaving the question of which was voiced and which was voiceless open.

This is why I postulate two separate, but diachronically indistinguishable, phonemes corresponding to /h₁/: To make the reconstructed system of fricatives symmetric between voiced and voiceless fricatives.

The rarity/lack of /b/ in PIE can then be explained by PZ and PV merging in most or all environments in early PIE.

At some point, the traditional laryngeals were lost. TL (traditional /s/), however, generally was not lost. However, there are several PIE daughter languages in which /s/ became /h/ or Ø in certain environments. Thus I propose that a general rule of the form L -> h -> Ø applied (laryngeals first become h, then later disappear entirely), but some process (perhaps voicing interactions between [s] and [z], one of which would have been /s/ and the other /dʰ /) interfered with it being applied to /s/, except in a restricted set of environments in certain dialects. I'm not quite satisfied with this, and this is probably the weak point in this reconstruction.

Anybody want to poke holes in this?

[jmcd]

The Proto-Celtic shift of ft > xt is at least 1000 years older than that of Dutch, and was part of a wider shift of all other plosives to /x/ before /t/, and /f/ was a plostive because it came from /p/. Unless a similar shift happened within Celtic later on I cant see any way this could have influenced Dutch.

On the other hand, Dutch might have had ft>xt about the same time as Picard dialects had pt>kt. (I'm not sure on the dating of the latter.)

[finlay]

Making the system symmetrical shouldn't be a goal of linguistic analysis. Besides, you have no scientific method to prove it, as it seems to be unfalsifiable.

[linguofreak]

Making the system symmetrical shouldn't be a goal of linguistic analysis. Besides, you have no scientific method to prove it, as it seems to be unfalsifiable.

Are you talking about the reconstruction in general, or the presence/absence of PL specifically? As for the reconstruction in general, I have a feeling that it may well be falsifiable by data that is known to people on this board but not to me, which is why I posted it. As I said, the area around TL strikes me as a weak point.

As for the presence or absence of PL in the reconstruction, the data fits either the presence or absence of PL, and does not allow us to distinguish it from KL in PIE roots, but it's desirable to choose the simplest of the available options. We could view this as the case with the smallest phoneme set (in which case we just use KL), but I see it as the case that leaves us with the fewest questions (in which case we use both PL and KL so that we don't have to answer the question "Why is there a gap in the labial series when all the other places of articulation are filled, and why is there a gap in the [voiceless/voiced] fricative series when all the other manners of articulation are filled?").

[KathTheDragon]

I thought up an interesting potential reconstruction of the PIE obstruents:

<snip>

Funnily enough, I came up with almost exactly the same idea several months ago (you can probably find some of my posts about it). However, I reconstructed h₁ as /h/, which may have come from earlier /ɸ/ and /xʷ/ (which merged into the former phoneme first). I also reconstruct h₃ as /x/, as I do not believe /o/ was rounded (at least at the time of laryngeal colouring). This reconstruction has the added advantage of offering an explanation of why /s/ is a special fricative - it's the only non-dorsal/glottal fricative, both of which are known to be prone to loss.

[Pabappa]

One of the flaws, I think, with that theory is that you need to come up with a good explanation for why most of the voiceless fricatives disappeared but the voiced fricatives survived and even hardened into stops in most branches of the family.

[linguofreak]

One of the flaws, I think, with that theory is that you need to come up with a good explanation for why most of the voiceless fricatives disappeared but the voiced fricatives survived and even hardened into stops in most branches of the family.

Recall that I leave open whether L = F and A=V, or L=V and A=F.

I posit that one or the other of aspirates and laryngeals was voiced fricatives, and the other was unvoiced fricatives, but I don't specify which.

[Pabappa]

That theory is a lot more interesting, I think, although still not without problems. I think if you identify the laryngeals with original voiced fricatives, which then mostly disappear and in some cases leave behind a vowel, that leaves only the voiceless fricatives, which then change to voiced ones (but maybe a few laryngeals remain, and merge with these) and then into voiced aspirates. It's still a pretty steep final step but more plausible I think than losing /ɸ x χ χʷ/ unconditionally but (near-)unconditionally keeping /β z ɣ ʁ ʁʷ/ and /s/.

[KathTheDragon]

Actually, there's another example of debuccalising only the voiceless dorsal fricatives - Germanic. Recall that /x/ is [h] word-initially, but /ɸ θ s ɣ/ are still [ɸ θ s ɣ] in the same position (the other voiced fricatives are stops in initial position). So, I doubt that xʷ>ɸ; ɸ>h; x,χ>h without similar changes in the voiced fricatives is at all unplausible (but note how I've ordered the changes)

[linguofreak]

and then into voiced aspirates.

I was thinking more along the lines of "the voiced aspirates never existed as such, but were really fricatives". So instead of a final step where the non-laryngeal fricatives become voiced aspirates, in cases where traditional reconstructions have "the voiced aspirates changed into sound type X in daughter language Y", I have "the non-laryngeal fricatives changed into sound type X in daughter language Y".

The only reason I use the term "voiced aspirate" in my reconstruction is that the sounds I reconstruct as non-laryngeal fricatives are *traditionally* reconstructed as voiced aspirates.

[Valdeut]

Anybody want to poke holes in this?

Correct me if I'm misunderstanding what you're proposing. But the core idea seems to be that PIE had a nice symmetrical system of four series based on two contrasts: plosive–fricative and voiceless–voiced. In particular, the proposal seems to be that the voiced aspirates and ”laryngeals” were two symmetrical series distinguished only in that one of them was voiced and the other voiceless.

Now, if we look at the phonotactics of PIE words and roots, do voiced aspirates and ”laryngeals” actually appear to share any properties? Do they pattern with each other in terms of distribution?

In roots that have two stops, there seem to be a strong preference for these stops to have different places of articulation. Can the same preference be observed if you add the ”laryngeals” into the mix?

If there is no other patterning between ”laryngeals” and voiced aspirates, why would you think that there should be any sort of symmetry just in terms of what segments exists? Why both PL and KL?

And if laryngeals and voiced aspirates were distinguished only in voicing, wouldn’t you expect a lot of alternations between the two due to voicing assimilation? For example, wouldn't you expect the zero grade of the roots *bʰes- and *ǵʰes- to at least in some daughter languages appear to reflect *h₁s, *bʰdʰ or *ǵʰdʰ? Can you find any such alternations?

Are there any restrictions in terms of what other segments you voiced and voiceless fricatives can coöccur with in a root? For example, do your voiced fricatives only occur next to voiced stops in roots, and vice versa?

And are there any examples where voiced aspirates appear to vocalize like laryngeals between two voiced or voiceless segments (depending on what you think they were)?

How is Siebs’ law explained within this reconstruction?
http://en.wikipedia.org/wiki/Siebs%27_law

And traditional *s and *dʰ, did they really differ only in that one was [z] and one was [s]? So the word *misdʰós would have had either a cluster [sz] or [zs]? And what about the cluster *sd and *st? They have very different outcomes in Sanskrit compared to *dʰt.

[masako]

Just gonna put this here.