Of the Germanic consonant shift

[alice]

This past week I've read in two different books that the changes known as Grimm's Law were the most wide-ranging ever known to occur in any human language. To what extent is this known to be true?

[Nortaneous]

that's not known to be true - it's baseless pop-sci bullshit. how would one even establish a total ordering over "wide-ranging"

[dɮ the phoneme]

Where is this claim from? What is "wide-ranging" even meant to mean?

[alice]

The complexity of this shift appears to be without close parallel in known histories of consonant systems in the languages of the world...

and:

It is probably the most comprehesive group of sound changes that has been observed in the history of any language.

So "most wide-ranging" is a faulty remembrance, and the two quotes probably don't mean the same thing, but the first is what I was asking about: is the First Germanic Sound Shift really the "most complex" one we know about?

[zompist]

This seems like it'd be hard to document in a valid way. I don't think there's an official catalog of sound changes, nor an official way of counting them. E.g. is Grimm's Law one rule or three?

Off the top of my head, the change in Chinese where voiced consonants unvoiced, accompanying a split of the four tones into eight, seems about as complex a change... basically affecting every syllable in the language.

[Nortaneous]

- the Tocharian plosive collapse (incl. nine proto-phonemes becoming /k/... although this isn't quite accurate, Tocharian preserves some labiovelars)
- Tangut 'compression', which includes the complete transphonologization of consonant clusters into vocalic features - e.g. *ptə *stə *ltə *rtə *ʁtə *trə > twə tə̰ tə̰ tə˞ tə̠ tɑ̯ə - but which still isn't well-understood
- the innumerable Grimm's Law analogues in Armenian, where in certain cases you get ejection of [+front] onto following vowels
- regular initial consonant loss in Arandic
- regular initial syllable loss in Paman(?)

[bradrn]

- Vowel collapses and associated consonant splits in Arrernte, NW Caucasian, Marshallese
- Great Vowel Shift
- Nishnaabemwin syncope
- Practically any instance of tonogenesis

[WeepingElf]

The Germanic consonant shift is actually a rather simple affair, at least if my idea holds true that the PIE voiceless stops originally were aspirated (an idea I developed not to account for the Germanic sound shift, but for the root structure constraints). Perhaps, Germanic and Armenian were two dialects which were not reached by the deaspiration of the voiceless stops. Then, the Germanic sound shift is just a voicing loss in the voiced stops, followed by a change [+breath] > [+spirant]. And the Armenian shift is the same without the second step. (There even is evidence that the "voiced" stops of Old Armenian still were breathy-voiced.)

I also have a hunch that the PIE dialect that evolved into Germanic was isolated from the rest for some time. If Late PIE was the language of the Corded Ware culture, and the Bell Beaker people spoke something else (perhaps a sister of Late PIE), the Corded Ware people of southern Scandinavia were cut off from the rest of Late PIE by the Bell Beaker people in Germany, and their dialect evolved in isolation until Italo-Celtic spread into western Europe in the Late Bronze Age. This perhaps explains why people like Warnow and Ringe have so much trouble placing Germanuc in their IE family trees.

[kodé]

I’ll have to find this paper, but apparently the Armenian consonant shift then went seven different ways in daughter dialects. This is where the infamous Western vs. Eastern Armenian “voicing flip” came from. And you’re right, weeping elf, some dialects did have breathy-voiced stops, as documented by some linguists (though I don’t know of any extant varieties that retained this after the dialect leveling of the last hundred years).

I’m not sure how they’re purporting to measure complexity of sound changes, but I’m not convinced. Polar changes, like tone polarity in Athabaskan languages are way crazier to me, especially since there was tonogenesis first.

[vegfarandi]

OK, only marginally relevant, but I recently learned about this theory of extensive Phoenician contact with Germanic that may account for its unique vocabulary compared to other IE languages, Runes, word order and strong verb ablaut, that I find absolutely fascinating. https://benjamins.com/catalog/nss.32

[KathTheDragon]

It's an interesting theory but deeply questionable. Almost nothing it claims to explain is actually a significant problem. For example, Germanic's verbal ablaut is trivially derivable from PIE patterns, and the asymmetry of ablauting verbs and non-ablauting nouns/adjectives is paralleled in most other branches.

[WeepingElf]

I’ll have to find this paper, but apparently the Armenian consonant shift then went seven different ways in daughter dialects. This is where the infamous Western vs. Eastern Armenian “voicing flip” came from. And you’re right, weeping elf, some dialects did have breathy-voiced stops, as documented by some linguists (though I don’t know of any extant varieties that retained this after the dialect leveling of the last hundred years).

I think I know which paper you mean, though I have no idea ATM who wrote it and what it is titled. I am not going to argue with you about this; you are a native speaker and I can't claim expertise on Armenian dialectology.

[WeepingElf]

It's an interesting theory but deeply questionable. Almost nothing it claims to explain is actually a significant problem. For example, Germanic's verbal ablaut is trivially derivable from PIE patterns, and the asymmetry of ablauting verbs and non-ablauting nouns/adjectives is paralleled in most other branches.

Indeed, it's bullfrogs. There is nothing in Germanic which really points at a substantial Semitic influence. Vennemann is notorious for such nonsense (for instance, he claimed "Vasconic" etymologies for such transparent names as Ebersberg or Bischofsheim), and Mailhammer just parrots what his master is telling him. They are not that much better than Octaviano The Germanic ablaut is entirely Indo-European, derivable from PIE ablaut by a few changes to the vowel system of which none is out of the ordinary; there is nothing Semitic about it.

When I said on Saturday that Germanic appears to have been isolated from the rest of IE for some time, I wasn't thinking of an early breakaway like Anatolian, only of a dialect that occupied a geographically marginal position within the post-PIE dialect continuum that had only little contact with its peers, and thus evolved by its own rules. Kind of like Tocharian (which evolved in isolation from other IE branches, but unlike Anatolian doesn't require an earlier stage of PIE), but still less so. Also, the question remains how old the Germanic consonant shift is; the names Cimbri and Teutones show no trace of it around 100 BC, but on the other hand, we can't be sure whether these names are actually Germanic or perhaps Celtic or whatever, and about 100 years later, names like Cherusci and Chatti clearly show that it had already happened by then.

[Pabappa]

the theory i like explains the /k/ in Cimbri and a few other names as simply being the closest sound Latin had to the Germanic /x/. They didnt use their /h/ because it was mostly found initially and was very weak, and already in the process of going silent. i got that idea from this board, i think dewrad.

[KathTheDragon]

Also, the question remains how old the Germanic consonant shift is; the names Cimbri and Teutones show no trace of it around 100 BC, but on the other hand, we can't be sure whether these names are actually Germanic or perhaps Celtic or whatever, and about 100 years later, names like Cherusci and Chatti clearly show that it had already happened by then.

Looking at my notes on Germanic sound changes (which are, imo, very out of date now), Grimm's law can actually have occurred very late if you allow it to have followed Verner's law (which is independently motivated).

[WeepingElf]

the theory i like explains the /k/ in Cimbri and a few other names as simply being the closest sound Latin had to the Germanic /x/. They didnt use their /h/ because it was mostly found initially and was very weak, and already in the process of going silent. i got that idea from this board, i think dewrad.

Yes, this sound was foreign to Latin. As was the dental spirant (the first t in Teutones); also, they may have not recognized the second one as voiced. It is well known that many Romance speakers do not recognize Germanic voiced stops as voiced, and that may have already been the case back then. Only later, when the Romans had more dealings with Germanic tribes, they began systematically writing the Germanic spirants in question as th and ch (perhaps helped by the shift of the Greek apsirated stops to spirants).

Also, the question remains how old the Germanic consonant shift is; the names Cimbri and Teutones show no trace of it around 100 BC, but on the other hand, we can't be sure whether these names are actually Germanic or perhaps Celtic or whatever, and about 100 years later, names like Cherusci and Chatti clearly show that it had already happened by then.

Looking at my notes on Germanic sound changes (which are, imo, very out of date now), Grimm's law can actually have occurred very late if you allow it to have followed Verner's law (which is independently motivated).

Yes. There are Celtic loanwords in Germanic that have undergone the shift, so the shift cannot have happened in my hypothetical and perhaps entirely mythical isolation period (which would have lasted from about 2400 to 1600 BC). It almost certainly was an Iron Age thing.

[KathTheDragon]

Apparently there's enough evidence across Germanic to reconstruct [t] as the default stop allophone of *d, with [tʰ] for *t, and respectively the other stops (though I don't have a source to hand). If so, it seems plausible that the pre-Grimm's stop inventory was *tʰ *t *d, much like Greek, whence *tʰ > *þ, *t > *tʰ, *d > *t~*ð.

[WeepingElf]

Fine - I always have opined that such an "Armenian-like" stage (of course, utterly unconnected to the Armenian development) was intermediate between the PIE state of affairs and the result of the consonant shift.

This brought me to the following idea. Read at your own peril, as I am adding speculation to speculation, which is not very advisable Perhaps the isolation of Pre-Germanic between 2400 and 1600 BC resulted in the deaspiration of the *T set never reaching it (alternatively, Pre-Germanic innovated the aspiration of the *T set in this time). In a second step, maybe still during the isolation period, maybe later, the *D set was devoiced, resulting in the "pre-Grimm" *Th-T-Dh system.

However, Celtic voiced stops are devoiced in the oldest loanwords from Celtic, e.g. Celtic *rig- > Germanic *rik-. With the *Th-T-Dh system already in place in the Bronze Age, one would expect Germanic **rig- instead. So this development appears to have been in the Iron Age. At least the devoicing of the *D set thus cannot have happened in the isolation period already.

[KathTheDragon]

I personally think that something like [t d] for *t *dʰ is the most likely set of values for these two series in pre-Germanic at the time of Verner's law, if it preceeded Grimm's law. Neither series aspirated. *rīks < PCelt *rīgs is admittedly a challenge to fit into this picture...

[WeepingElf]

I personally think that something like [t d] for *t *dʰ is the most likely set of values for these two series in pre-Germanic at the time of Verner's law, if it preceeded Grimm's law. Neither series aspirated. *rīks < PCelt *rīgs is admittedly a challenge to fit into this picture...

The two sets may both have been aspirated, in contrast against *D, which wasn't. What forms like *rīks show is that the devoicing of the *D set was late - but the aspiration of *T may have been much earlier (at any rate, it wasn't later, since that would have led to a merger of *D and *T). Note that with both *T and *Dh aspirated/breathy, voicing of *T via Verner's Law would have led to *Dh (as observed), not to *D. What is your idea of what the value of *D was when neither *T nor *Dh was aspirated?