Talk:Grimm's law

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
WikiProject iconVital articles: Level 5 / Society C‑class
WikiProject iconGrimm's law has been listed as a level-5 vital article in Society. If you can improve it, please do.
CThis article has been rated as C-class on Wikipedia's content assessment scale.
WikiProject iconLinguistics C‑class High‑importance
WikiProject iconThis article is within the scope of WikiProject Linguistics, a collaborative effort to improve the coverage of linguistics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
CThis article has been rated as C-class on Wikipedia's content assessment scale.
 High This article has been rated as High-importance on the project's importance scale.

A Grammar of Modern Indo-European[edit]

The text is practically the same as the one you can find in A Grammar of Modern Indo-European (v.3.30), pages 43-45. You can check the book here. They do mention the Wikipedia as one of their resources in the bibliography at the end of the book. But I don't know what came first. -- (talk) 13:33, 16 December 2007 (UTC)Reply[reply]

Pull or Push?[edit]

The article presents the change of proto-Indo-European voiceless stops into voiceless fricatives as the event which "pushed" the whole system in that direction. Isn't there another, more popular theory according to which the loss of aspiration in voiced aspirated stops "pulled" the system? —Preceding unsigned comment added by Staseman (talkcontribs) 04:02, 20 April 2006

Can you cite a source and explain how that might have worked? Oh, and please get yourself a username if you want to take part in discussion. --Doric Loon 14:54, 20 April 2006 (UTC)Reply[reply]

An username is not required or obliged to take part in any discussion Doric. Sander 17:46, 20 April 2006 (UTC)Reply[reply]

No, but it helps. It's a courtesy. --Doric Loon 23:33, 20 April 2006 (UTC)Reply[reply]

To return to the beginning question, it is rarely possible, after the fact, to tell whether a shift was pushed or pulled. As the shift is actually going on, it is (theoretically) possible to clock variation and frequencies, and/or note that (say) voiceless stops are getting more and more aspirated. Under most circumstances the issue has little interest (as is true of most "functional" so-called explanations of phonological change: they sound nifty in formulation, but in actual practice contribute little or nothing). Indeed, even the concept of phases adds little to the discussion. At most one might say, for instance, that the change of voiced stops to voiceless ones can't have gone to completion before the (original) voicless stops started to move. But Pre-Germanic could well have gone through a phase when some (shifted) voiceless stops still alternated with voiced ones (say, varying with register or age or whatever) while another set (the original voiceless ones) didn't alternate, with the fricativization coming along only later.
Put differently, we have some idea of what cannot have happened, but very little idea of what probably happened, beyond the fact of the three-way shift itself. Declaring that voiceless stops became fricatives first of all is to claim to know more than we do. Alsihler 22:10, 3 February 2007 (UTC)Reply[reply]

How would a specific user name make his/her question more courteous? And how would it help understanding the question? Admonishing someone for transgressing a non-existent rule seems like needless antagonism and discourteous in itself.


Does *gʷʰ always become /w/, rather than /gw/? If so it should be removed from the table, as it doesn't exactly show Grimm's Law in action. --Ptcamn 00:04, 8 July 2006 (UTC)Reply[reply]

Yes and no. It only needs some clarification. The *w < *gʷh in all probability went through a *gʷ (or *γʷ) phase. An interplay between [w] and [gw] is commonplace in diachronic linguistics: PIE *gʷ > Latin /w/ as in Lat. veniō "come" < *gʷṃ-yō; on the other hand, PIE *w > Welsh /gw/, as in *wiros "man" > W gwr (not to mention all those Germanic borrowings in early Romance where *w is replaced by gw: *gwerra < war, etc.). And no, it doesn't always become *w: in *gʷhen- "strike dead" it becomes *b (Old English bana "murderer", etc.). In at least one pretty-good etymology it becomes *gw after a nasal: PIE *sengʷh- "phonate" (Greek omphḗ "voice" < *songʷhā) > PGmc. *singwanaⁿ "to sing", Gothic siggwan /singwan/.
—Frankly, the "wife" example is really, really weak. I've never seen a remotely convincing etymology for the Germanic forms: Kluge opts for *weyp- "hide, cover" (with elaborate discussion about why she might be veiled ... and neuter besides), which is also Pokorny's and Partidge's preference, though the latter mixes it up with *weyb- "vibrate"; the OED laconically says (correctly) "of obscure origin". As for the example in the article, I can find no mention of a root *gʷheybh-. And more to the point, there is an excellent etymology for the taking, viz., *gʷhor-mo- "hot, warm", PGmc *warmaz > OE wearm etc., etc. = Latin formus (not attested apart from Paulus-Festus, alas), Sanskrit gharma- "heat", etc. Alsihler 22:11, 3 February 2007 (UTC)Reply[reply]
The University of Texas, Austin seems to believe it generally turned to *g in Germanic, They don't show any examples from *gwher in common English, but the Germanic words given have a modern derivation in German "gerben" (prepare leather). If *gwhen would cause Germanic *gunð, English has the modern derivation "gun", originally borrowed from Old Norse, for a type of large weapon. 惑乱 分からん * \)/ (\ (< \) (2 /) /)/ * 22:24, 28 April 2007 (UTC)Reply[reply]
The "wife" etymology isn't all that weak. It's clearly unusual that this word for "woman" is neuter, unlike other Proto-Germanic words for women, so the idea of a metonymy "vulva" > "woman" is not unmotivated. The example *warmaz is hardly better, because there is also the verb *werH- "to be hot", so there is no guarantee that it contains an original labiovelar stop. --Florian Blaschke (talk) 06:20, 30 October 2014 (UTC)Reply[reply]

or ?[edit]

Which is the correct transcription?

Greek transliteration[edit]

What transliteration scheme is it that renders πούς as pūs? I would think it would be pous. 00:49, 25 October 2006 (UTC)Reply[reply]

No, ού is in fact transliterated as ū. Greeks already pronounced the diphthong that way in the Classical period of Greek, even though they always spelled it OU. That's the accepted transliteration, and it holds in modern linguistics. RokasT 11:36, 25 November 2006 (UTC)Reply[reply]
I BEG to differ! There seems to be a confusion here between transcription (phonemic or otherwise) and transliteration. A translteration is a system for rendering each character of the original in an unambiguous way in a different signary. It doesn't have to be one-to-one (θ = th etc. is kosher), but the more straightforward and consistent the transliteration the better. Now, regarding the case in hand, we know that the sequence ou in Greek eventually became /ū/, but it is historically the outcome both of the diphthong *ow and secondary lengthening of *o. Besides, if you're going to take it on yourself to interpret Greek ou as ū, then presumably you would also render ei sometimes as ē and sometimes as ey depending on which it stood for in the original. And what about "iota subscripts"? Pretend they aren't there (which in a way is what the standard Greek orthography means)? In any case, I'm unfamiliar with any school of historical linguistics that would treat Greek transliteration in such a way, so far from it being "accepted" as a standard transliteration. Alsihler 22:11, 3 February 2007 (UTC)Reply[reply]
Your prayers have been heard. --RokasT (talk) 11:37, 24 January 2008 (UTC)Reply[reply]


Anyone think there should be a note somewhere that /k/ became /x/ before advancing to /h/? I'm inclined to think so, since it has implications for Verner's law. 01:05, 25 October 2006 (UTC)Reply[reply]

Of course it must have. In fact, we don't really know when it did become /h/: it might well have been [x] in Gothic, at least in some positions. And note that e.g. *hr in word-initial position becomes kr in Scandinavian languages, which implies (but does not prove) that in a word like *hringaz "ring" it was something more than an ordinary [h]. Alsihler 22:09, 3 February 2007 (UTC)Reply[reply]


I'm pretty sure that Icelandic "hvað" is not an example for the *kʷ→hw shift, since it's still pronounced with a /k/ (even though it's not written that way). I'll remove this. -- Schnee (cheeks clone)

A little learning is a dangerous thing. hvað is indeed the direct descendent of PGmc *hʷat-aⁿ (< *kʷod-ōm). It's probably not the best idea in any case to cite Icelandic, since Old Norse hvat does just as well (and isn't going to stir up irrelevant arguments about pronunciations that are 2,000 years too late).
In general, I find both that the etyma are poorly chosen (why trot out a contested and implausible set to demonstrate the reflexes of *gʷh when there are far clearer ones lying around?) and that the presentation the cognate forms as cited is questionable in detail. As hinted above, the Greek transliteration is unacceptable; forms of different meaning are sometimes glossed and sometimes not (Skt pāda is presented as meaning "foot", by implication; it does get that meaning, late, but earlier it meant "quarter", that is, "leg", of an animal), there are many small fluffs of detail (Latin "vivus" should have ī, and so on and on). I suppose someone (sigh, meaning me) should take the stuff in hand. I don't know how hard to be on the creator of the table, though the business about a root *gʷhíbh- (presumably a typo for gʷhībh-; and his or her discussion of it) has gone far to alienate my sympathies; whether or not the connection between the Germanic and Tocharian forms is valid, the other supposed cognates are both wrong (they reflect PIE *gʷenH₂- "woman") and are cited in a botched transliteration. Two of a trade never agree, and so on, and that goes double for historical linguists, but some things are out of place in what is supposed to be a rehearsal of basic principles. I'm working on a substantial revision. Alsihler 22:25, 3 February 2007 (UTC)Reply[reply]
hvað with a /k/ in front is a recent development, it was /x/ before that, with maybe some speakers still pronouncing it that way. (talk) 23:28, 3 October 2008 (UTC)Reply[reply]

Russian for 'foot'[edit]

Yesterday I commented in the Grimm's Law article that 'pod' is not the Russian for 'foot'. My comment had disappeared by this afternoon, without explanation. If Russian 'pod' is indeed considered cognate with English 'foot', Latin 'pes' etc., because of its meanings 'hearth', 'floor' and 'under', should not one of these meanings be stated in the article (perhaps in brackets immediately after the word) as justification? Levka 16:44, 19 February 2007 (UTC)Reply[reply]

Please read the summary of the revert: "Undid revision 108885647 by Levka (talk) This is talk page material, not article material" You do not make comments directly to article pages (unless they're commented out with: <!-- comment here -->. This is because articles are not the place for a discussion such as this. This talk page is the proper place for such discussion.
Addressing your real question, the words "Knight" and "Knecht" are both cognates of each other, yet mean fairly different things in both languages. We're not attempting to establish the similarity of meanings, but rather simply presenting a list of words that are cognates to those words. If you feel that the Russian entries are confusing, and misleading, and you know that the current meaning of the word, is different from the cognate that you're comparing it to, you can put it in parentheses: "Russian: pod (hearth, floor, under), ..." in order to remove that misleading nature. It is however not considered acceptable to attempt to discuss an issue directly on the article page. --Puellanivis 00:24, 20 February 2007 (UTC)Reply[reply]

Thank you, Puellanivis, for your patience with a newcomer to Wikipedia. Please accept my apologies for breaching the established etiquette – I should have studied it more carefully before leaping in!

I do of course realise that cognates will often have acquired quite divergent meanings over the centuries, which is precisely why I think it would be helpful occasionally to give meanings in parentheses, particularly when the majority of the words cited still share a basic meaning (e.g. "foot").

Finally, I'm not quite sure what you mean when you say in your reply above " can put it in parentheses...". Are you actually inviting me now to make this insertion in your article? If so, I will gladly do it.

Best wishes, Levka 15:41, 20 February 2007 (UTC)Reply[reply]

Well, it's not "my" article, it's our article. I am by no means an expert on Grimm's Law, so if you know something that I don't, it would be great to see you add it to the article.  :) --Puellanivis 16:22, 20 February 2007 (UTC)Reply[reply]

Peg/ Baculum?[edit]

According to Chambers' Dictionary of Etymology "peg" is borrowed from Middle Dutch "pegge" and is of uncertain origin after that. Deman7001 17:48, 26 May 2007 (UTC)Reply[reply]

If no one else disapproves, I would like to change the words in the table. Perhaps this could be changed to English warp, virbas(it's either Sanskrit or Lithuanian I can't remember). I'm still looking for a most ideal example. I checked the German page and it has Pegel listed as cognate with baculum, but I don't think peg is cognate with Pegel or that Pegel is cognate with baculum either. I checked an German online Etymological dictionary and Pegel doesn't appear. Also, one would expect that if peg and Pegel were cognates that Pegel would have shifted to Fegel or Pfegel.Deman7001 02:32, 29 May 2007 (UTC)Reply[reply]

No problem with your change if you like it better: warp is most closely related to German werfen (throw), and is cognate with Latin verber, whence English reverberate. You have to be careful with a /p/ in final position, because it can be a Germanic /b/ which has been subject to terminal devoicing; so initial or medial position are safer, but in this case it does seem to be a Germanic /p/ resulting from Grimm's law. But peg was also fine. Yes, it is borrowed from Dutch, but that is also a Germanic language and therefore does not affect its relevance. According to Kluge's Etymologisches Wörterbuch, German Pegel IS cognate with peg. As you know, the High German consonant shift moved p > pf in initial position, so any High German word beginning with a non-affricated /p/ must be a loan word. In fact, Pegel is taken from Low German, which is very close to Dutch, so that tallies nicely. The trouble is, good examples of a Germanic /p/ are hard to find. The symmetry of Grimm's law demands that there must have been an original IE unaspirated /b/ which became /p/ in Germanic, but it seems to have been a very rare phoneme in PIE, and Calvert Watkin's dictionary, which I happen to have beside me, only knows seven IE roots beginning with it - all of which have some question mark or peculiarity attached to them. Some scholars have questioned whether IE had the phoneme at all. On balance, I think it probably did. But we will always be struggling to demonstrate it with examples as clear as the father/pater or thou/tu pairs. --Doric Loon 08:13, 29 May 2007 (UTC)Reply[reply]

Simplifying the table[edit]

Apparently people don't know what "The most illustrative examples are used here." means, adding things that are shifted by separate sound changes and so don't illustrate Grimm's law very well. I suggest we include only English on the Germanic side (English is conveniently conservative with respect to Grimm's law), and maybe Greek and/or Sanskrit on the unshifted side. Thoughts? --Ptcamn 08:25, 29 May 2007 (UTC)Reply[reply]

And why on earth is φρατήρ commented out? --Ptcamn 08:28, 29 May 2007 (UTC)Reply[reply]
The point is to only include "illustrative examples", "φ" would be pronounced as a voiceless labiodental fricative or an aspirated voiceless bilabial plosive. To include cognates with other sound shifts than Germanic muddles things, and complicates the point. 惑乱 分からん * \)/ (\ (< \) (2 /) /)/ * 11:28, 8 June 2007 (UTC)Reply[reply]

When did this occur?[edit]

Hi, when in pre-history did this sound-change occur? --Kjoonlee 23:53, 24 September 2007 (UTC)Reply[reply]

I am currently studying the 'Grimm's Law' at the university and in my textbook 'Klaus Weimann, Einführung ins Altenglische (Introduction into Old English), Heidelberg Wiesbaden, 1995' there is also /bh, dh, gh/ to /v, ?, j/. ? = the voiced form of th (Sorry, can't find the correct symbol, but I hope everybody understands what I mean). My point is therefore that the author was totally ignored this part of the Grimm's Law —Preceding unsigned comment added by (talk) 11:35, 28 October 2007 (UTC)Reply[reply]

Dating Grimm's Law[edit]

I am currently working on several Germanic-related articles. My research has led me to compose a short passage relating to the dating of Grimm's law. It appears below with notes:

There is growing consensus among researchers that the First Sound Shift took place around 500 BCE.[1] One argument used to substantiate this date observes that the Greek word for hemp (kánnabis) appears to have been borrowed into that and other Indo-European languages sometime around the 5th century BCE from a hitherto unidentified but purportedly non-Indo-European source in the form *kanab-. Based on the cognates of the word in attested Germanic daughter languages (Old English hænep, Old High German hanaf, Old Icelandic hampr) it is assumed that the Proto-Germanic form of the word was *hanapaz. According to this reconstruction, it would appear that the word entered the Proto-Germanic vocabulary prior to the First Sound Shift. Assuming that hemp became known in both regions at or around the same time, the First Sound Shift can thus be considered to have taken place no earlier than the 5th century BCE.[2]


  1. ^ "The Gmc [Germanic] consonant shift will indeed have taken place around 500 B.C., that is, in the Iron Age or the overlapping time between the Bronze Age and the Iron Age, i.e., in the transition between EPGmc [Early Proto-Germanic] and LPGmc [Late Proto-Germanic]. This timing of the consonant shift seems to be more and more the consensus." (van Coetsem 1994:140-141)
  2. ^ For a detailed presentation of this argument, see Streitberg (1943:135-6). For discussion, see Voyles (1994:78).


If someone would like to include this information here, I can post the full citations for the references. There is also an original argument offered by van Coetsem (1994) which involves extra-linguistic evidence (Celtic in particular), but I have not been able to locate an occurence of its being cited in other work. With that being said, I view van Coetsem's argument to be the stronger of the two. If interested, please contact me on my userpage. Aryaman (☼) 17:30, 19 November 2007 (UTC)Reply[reply]

Comprehensability of the table - adjusting the column order.[edit]

Two suggestions for the tables:

1. Generally, we are trained to think in terms of time moving forward and represented as moving from left to right in diagrams or other representations on a page. The older becomes the newer, the earlier pronunciation changes into the more recent pronunciation.

With this in mind, I think the table would be more understandable, more intuitive, if the column order was changed. The columns would then also reflect the order in which the symbolic examples are given and the order in which the process actually took place.

Additionally, internal consistency within the table is easier to understand.

Note that column 1 states: *p→f But then the language examples reverse that order, citing the shifted examples before (to the left of) the non-shifted original state. So the example order conflicts with the rule's order.

To resolve both these issues, I think the column of older non-shifted forms should appear to the left of the column of the newer shifted forms.

2. Not as critical, but perhaps more logical in terms of process, would be to put the symbolic rule in the middle column. So, one would read a row like this:

Col 1 | Col 2 | Col 3

"the old state" | "underwent this change" | "to the new state."

Hlaufman (talk) 16:58, 22 November 2007 (UTC)Reply[reply]

Norwegian "hva" *kʷ→hw[edit]

Some dialectal Norwegian still uses *kʷ "kva" in speech instead of hw "hva". Don't know if it's relevant. /per —Preceding unsigned comment added by (talk) 10:45, 3 December 2007 (UTC)Reply[reply]


What exactly is meant by "gaelic" in the Non-Germanic (unshifted) cognates column? It is not Irish\Gaelic, but the examples given are extremely close to it. Is it the common root of all the Goidelic languages? Bogger (talk) 15:55, 28 December 2007 (UTC)Reply[reply]

These are the Scottish Gaelic forms. --Doric Loon (talk) 18:45, 28 December 2007 (UTC)Reply[reply]

Source(s) for Tocharian Cognates[edit]

Might I ask for sources for the Tocharian cognates of wife? I'll happily add this information to the entry once I come upon a good hard source. Dr. C.S. Lewis-Barrie, Ph.D. (talk) 18:23, 28 January 2008 (UTC)Reply[reply]

I think it's from Calvert Watkins' appendix of Indo-European Roots to the American Heritage Dictionary. Apparently, there are no other known cognates, and both the phonetic and semantic shifts seem to be unusual, so it seems wildly disputed. 惑乱 分からん * \)/ (\ (< \) (2 /) /)/ * (talk) 14:28, 10 March 2008 (UTC)Reply[reply]

Hidden cognates[edit]

I re-hid the less illustrative cognates once again. I know they are correct, but it would be confusing for a layman to see a bunch of examples shifted in (for a layman) haphazard ways. This shouldn't be a list to show all known cognates to words, but the most illustrative examples for explaining Grimm's law, in particular. 惑乱 分からん * \)/ (\ (< \) (2 /) /)/ * (talk) 15:30, 7 March 2008 (UTC)Reply[reply]


Why does it say that *b→p and then show English: warp; Swedish: värpa; Dutch: werpen; Icelandic, Faroese: varpa and shows the Latin version as verber. Should it not be v→w or something? (talk) 01:14, 10 March 2008 (UTC)Reply[reply]

Aaahh, the v/w distinction isn't primarily phonetic, as much as ortographic. Apparently both Latin and Proto-Germanic is believed to have had one single v-sound /w/ (as in wait). The distinction here is between warp and verber, but the problem is that *b was such a rare phoneme in PIE, that it's difficult to find good examples. 惑乱 分からん * \)/ (\ (< \) (2 /) /)/ * (talk) 14:25, 10 March 2008 (UTC)Reply[reply]

CHaracter set ?[edit]

Many of the characters in the table are appearing as filler symbols rather like a very fat lower-case "L". Is there some "proper" character set we are supposed to have loaded for these pages to display correctly? And how do we make certain our browser uses it in this instance?

Install Unicode, I think that should fix it. 惑乱 分からん * \)/ (\ (< \) (2 /) /)/ * (talk) 09:15, 1 May 2008 (UTC)Reply[reply]


"Viv" was dialectal/poetic only in Swedish since at least 1922 ( I have never heard it, "Fru" or "Dam" are always used instead. Please remove if it should be removed. —Preceding unsigned comment added by (talk) 06:51, 10 May 2009 (UTC)Reply[reply]

This is a list of cognates, not necessarily common words. 惑乱 Wakuran (talk) 15:25, 9 November 2009 (UTC)Reply[reply]

I propose redirecting the article Great consonant shift here. Please see Talk:Great consonant shift for discussion. +Angr 20:47, 16 June 2009 (UTC)Reply[reply]

Oppose - Grimm's law is certainly related to the Great consonant shift article, but does make a clear distinction between the First and Second shifts. The new article Great consonant shift is intended to be a short summary for the beginner. Although it is a stub and needs more material for both shifts, the massive detail in the Grimm's law article is for people who have more than a beginner's knowledge of the subject. I added a "Main article" link to the Grimm's law article. Greensburger (talk) 04:14, 17 June 2009 (UTC)Reply[reply]

Please keep the discussion centralized at Talk:Great consonant shift. +Angr 13:23, 17 June 2009 (UTC)Reply[reply]

Alleged first non-trivial[edit]

Some of the consonant laws in the Semitic group are non-trivial. —Preceding unsigned comment added by (talk) 14:36, 2 July 2010 (UTC)Reply[reply]

It was the first to be explicitly discovered, not to occur. --Trɔpʏliʊmblah 18:46, 6 July 2010 (UTC)Reply[reply]

Spanish example words[edit]

Hello, I have given Spanish examples for practically every Germanic word in the 'non Germanic' section to the right of the page; they have disappeared today with no explanation. I can also notice that no Romance language examples are being accepted as examples, being those the retainers of more than one IE feature not already present in Germanic. I have observed in latter years a biasing towards Germanic when it comes to IE research, and wouldn't like to think that Wikipedia follows that trail.

Spanish example words[edit]

Hello, I have given Spanish examples for practically every Germanic word in the 'non Germanic' section to the right of the page; they have disappeared today with no explanation. I can also notice that no Romance language examples are being accepted as examples, being those the retainers of more than one IE feature not already present in Germanic. I have observed in latter years a biasing towards Germanic when it comes to IE research, and wouldn't like to think that Wikipedia follows that trail. Druiweplukker (talk) 00:59, 19 October 2010 (UTC)Reply[reply]

I was the one to remove them (with an explanation, but also as a minor edit, which may have hidden it from you?). I would think having Latin examples covers the Italic branch adequately. You may also note we don't have any examples from Hindi, Bengali, Modern Greek, Albanian, Welsh, etc. The idea is not to list every known cognate, just sufficiently many to demonstrate non-occurrence elsewhere.
Also, a Germanic focus in an article that's about a Germanic sound change is hardly a bias.--Trɔpʏliʊmblah 18:28, 19 October 2010 (UTC)Reply[reply]

All right, your point sounds logical; but I have seen Spanish hardly ignored when exemplifying about IE, not to mention Portuguese or Italian, much less covered. There is also a good deal of Gothic-origin words here, such as escanciar (skagkjan, to pour) and Rodrigo (Roderick) or Alfonso (Alafuns), so we could deepen into what's the scope of Germanic, but there would not be any point on it I guess. I didn't happen to see the explanation, but it's OK as you have given plenty of it. Also, interesting profile that of yours. —Preceding unsigned comment added by Druiweplukker (talkcontribs) 02:11, 20 October 2010 (UTC)Reply[reply]

This sound change has also affected the English language, as many latin loan words arrived both directly from latin and indirectly (shifted) via German. English often indicates the loan source in the spelling. I would think that it would be useful for many readers to see the removed spanish cognates if they illucidate similar phenomena.OrenBochman (talk) 13:31, 31 March 2012 (UTC)Reply[reply]

Two questions[edit]

1) Did the shift cause any clashes at the end of the chain that we know of? Like p... shifting to f... clashing with a pre-existing f...? 2) Are there any known exceptional words? Words that didn't shift, where not imported from other languages, and are not covered by the exception rules? — Preceding unsigned comment added by (talk) 03:23, 2 April 2012 (UTC)Reply[reply]

There was no pre-existing "ɸ" or "f" before the changes. There were various contextual partial exceptions (such as that the shifts did not apply after "s") and later complications (such as Verner's law), but Grimm's law itself does not seem to have neutralized contrasts... AnonMoos (talk) 07:56, 2 April 2012 (UTC)Reply[reply]

Angr proposed to merge these two articles on June 16, 2009, but did not initiate a discussion nor state the reasons for his merge. I'm not really an expert on this subject, so I am not certain how to go about merging the two articles. But this should be dealt with soon, as the merge request is outstanding for over three years now. Does anyone have any thoughts on this? WTF? (talk) 16:08, 12 July 2012 (UTC)Reply[reply]

Not much to merge, I'd just blank it and redirect it to Grimm's law.·ʍaunus·snunɐw· 16:10, 12 July 2012 (UTC)Reply[reply]
There's discussion at "Redirect from Great consonant shift" directly above. Maybe make it a disambiguation page... AnonMoos (talk) 18:26, 12 July 2012 (UTC)Reply[reply]
Per the previous discussion even User:Greensburger who started the former appears to consider it a "layman introduction" into the same topic. I think easing on the technicalities in this article would be a better way to reach the same goal. So yeah, not much to actually merge from there. --Trɔpʏliʊmblah 19:06, 12 July 2012 (UTC)Reply[reply]
A disambiguation page would be better because Grimm's law and the High German consonant shift have very little to do with each other, other than that they both affected German and involved a chain shift of consonants. It would be misleading to have a single page treat them both as though they were somehow related. CodeCat (talk) 21:05, 12 July 2012 (UTC)Reply[reply]
WTF (That's a vocative, not an interjection), I did initiate a discussion and state the reasons. It's all at Talk:Great consonant shift. I've reopened the discussion there in the form of an RFC at Talk:Great consonant shift#Redirect to Grimm's Law, 2012 remix. Please discuss this there so we don't have different people discussing it in different places. Angr (talk) 10:15, 13 July 2012 (UTC)Reply[reply]

Effect on geminates?[edit]

The Indo-European word *átta and its Germanic descendants show that Grimm's law preserves geminate plosives (at least -tt-) and does not turn them into fricatives. I believe geminates were rather rare in PIE, but a few did exist so what effect did Grimm's law have on them? It seems likely to me, from a common sense point of view at least, that geminates were devoiced, but did not become fricatives; after all, why would a language that devoices voiced plosives keep them when geminated even though this would have been (at the time of the chance) a non-phonemic distinction? Are there any sources that consider this question and give examples? CodeCat (talk) 19:54, 21 November 2012 (UTC)Reply[reply]

"Atta" was originally a kind of baby-talk word (meaning "daddy", not actually "father"), and such don't always follow normal sound changes, so I'm not sure that I'd deduce too much from it. There were few geminated stops within roots or stems in Indo-European, but when t + t came together across a morpheme boundary, the first "t" affricated or assibilated, as seen in a number of forms in Greek (including dental consonant final middle-voice perfects), in Latin ēst "he eats", etc. AnonMoos (talk) 01:02, 22 November 2012 (UTC)Reply[reply]
*atta is the only PIE word I can think of that has a geminate, but there may be more that I'm not aware of. I don't think it still was a baby talk word in Germanic, as it became the normal word for father in Gothic. Sound laws are normally considered exceptionless (Grimm's law certainly is!) so if this word was not affected, it must be part of the law. CodeCat (talk) 04:05, 22 November 2012 (UTC)Reply[reply]
But in non-Gothic Germanic, pater-cognates prevailed. Here's something which comes from a rather unreliable source (all I can find right now by Googling), but which I think is probably true nevertheless: [Geminates] "appear mostly in words of expressive origin, children vocabulary, onomatopoeia, etc., which makes it more likely that PIE inherited gemination mainly as an expressive resource, different from its central phonological system"... -- AnonMoos (talk) 05:12, 22 November 2012 (UTC)Reply[reply]
Re "I don't think it still was a baby talk word in Germanic, as it became the normal word for father in Gothic": just because it became the normal word for father, that doesn't mean it wasn't a baby talk word before. The common Brythonic words for father and mother (Welsh tad and mam and something similar in Cornish and Breton), both started out as baby talk words, *tata and *mamma, and I'm sure there are plenty of examples from other languages. Angr (talk) 10:45, 22 November 2012 (UTC)Reply[reply]

True, but...[edit]

After all, you must admit that the Germans and other Nordish tribes implemented a great amount of latin languages into their vocabulary. Sure, they've chosen to say it differently due to their different tongues. But the worlds mostly - if not all - came from the greek and latin vocabulary. Does this proof the indo-european language thesis? Not at all, it just proves that languages were more dynamic in the past than we could ever imagine nowadays: They interchanged words very heavily, especially from the South to the North, from "high culture" to "low culture". -- (talk) 00:23, 3 August 2014 (UTC)Reply[reply]

No, the words did not mostly come from Greek and Latin. There are some Greek and Latin loanwords in the oldest Germanic languages, but really not very many. —Aɴɢʀ (talk) 09:10, 3 August 2014 (UTC)Reply[reply] -- There are no attested Greek or Latin loanwords during the proto-Germanic period, when Grimm's Law was in operation, and Germanic speakers were presumably mainly located in southern Scandinavia, and not in any kind of direct contact with Mediterranean civilizations. AnonMoos (talk) 15:39, 3 August 2014 (UTC)Reply[reply], stop using that shit. The Germanics in ancient times didn't borrow anything from the so-called "civilization" - they didn't even have a written language (runes) untill 2 AD, they didn't really needed it. The maximum of peaceful interaction is trade. Like the Celts, they were free and fearless people, not cowards behind stone walls. I would say that the further north the "higher" culture - it was a completely oral and living phenomenon for all tribesmens, unlike the "civilized" southerners, where whole culture was available only to a few percent of literate people who could read Livius or Homer poems.

MHG wülpe[edit]

Currently the article claims there was a Middle High German word wülbe "she-wolf" and cites Kuiper (1995) for it. I don't have access to the paper, but someone else does and says Kuiper claimed no such thing; instead Kuiper cited Middle High German wülpe and derived it from Proto-Germanic *wulƀjō- (in his transcription). By West Germanic gemination, this would have become Proto-West-Germanic *wulbbij-, nominative singular *wulbbī, which regularly yielded MHG wülpe (compare modern Standard German Rippe "rib", Sippe "extended family"). **Wülbe doesn't seem to be attested anywhere.

If this isn't fixed in a month, I'll just delete it – unless I can improve on it after reading this paper (German is my native language). David Marjanović (talk) 18:58, 19 October 2014 (UTC)Reply[reply]

The West Germanic gemination only applied after -j-, which only occurred after light syllables because of Sievers' law. After heavy syllables -ij- appeared, which did not cause gemination. So not only is -ulbj- not possible in Proto-Germanic, but -ulb(i)j- could never undergo gemination. If it did, it would be the only attestation of this in all of West Germanic, so that would need some very convincing sourcing to be any more than a badly researched fringe theory. CodeCat (talk) 19:43, 19 October 2014 (UTC)Reply[reply]
Oops. But anyway, Seebold (1967: 129) reconstructed PGmc. *wulgw-ī-/jō to explain ON ylgr, adding (my translation): "In the remaining Germanic dialects the root-final consonant has been assimilated to that of 'wolf'." PGmc. *b is, according to that paper, the regular outcome in initial position and only there. David Marjanović (talk) 21:06, 19 October 2014 (UTC)Reply[reply]
Also, are you aware that b > p is a regular part of the High German consonant shift in Upper German dialects? CodeCat (talk) 19:45, 19 October 2014 (UTC)Reply[reply]
I natively speak an Upper German dialect. Today at least, /b d g/ are voiceless lenes (except in the Carinthian dialects, because those have reinterpreted the entire sound system in Slovene terms) – all the way to Austrian Standard German, incidentally (Carinthians try to devoice their lenes when speaking it, often either failing or overshooting and landing at the fortes). When I started learning French, I had to consciously learn to articulate voiced obstruents ([z] included, but I digress)! I strongly suspect the voiceless lenes go straight back to the High German consonant shift, and the Old Bavarian spellings with p and k (also Lombardic crapworf "the crime of 'throwing' a corpse out of a grave") only indicate this voicelessness, perhaps because the Romance dialects spoken in the same areas had voiced [b d g], perhaps because Latin was taught with a Romance (or Old Irish...) pronunciation and everyone who was literate at all had learned to read & write in Latin, perhaps both.
Notker (Old Alemannic) did use b d g in some environments, but his famous spelling rule presupposes that b d g were voiceless nonetheless.
By Middle High German times, however, the spellings were much more regularized; I wouldn't expect p for /b/ anymore.
David Marjanović (talk) 21:06, 19 October 2014 (UTC)Reply[reply]
I wouldn't assume that. After all, the devoicing d > t was reflected in Middle High German spelling, so presumably the g > k and b > p changes of Upper German were reflected the same way. CodeCat (talk) 21:16, 19 October 2014 (UTC)Reply[reply]
I don't think that's a devoicing either. Modern northern German (including northern Standard accents) has syllable-final fortition; southwestern German wouldn't show traces either way due to various lenition phenomena across the board, except maybe in Walser dialects that are beyond me; southeastern versions like mine have no trace of syllable- or word-final phenomena; epic MHG, based on a Swabian (southwestern but not too southern) dialect, spelled out word-final fortition, but that doesn't mean d was voiced.
On the other hand, many place names and elements in personal names still begin with p, and the oldest attestation of potatoes in Upper Austria (from 1644: ertöpfl) has t for /d/. To return to your actual point, then, I don't have another explanation for the p in wülpe.
David Marjanović (talk) 21:43, 19 October 2014 (UTC)Reply[reply]
Oh, sorry, you mean the change from Proto-West-Germanic *d to t. That's a qualitatively different thing: this *d must have been the voiced plosive [d], while *b must have been [β] or [v] in most positions and *g certainly was [ɣ] in almost all positions. Maybe the High German consonant shift changed voiced plosives into fortes, and voiced fricatives into lenis plosives. That would explain why *d and *dd behaved identically (t, tt), while *b and *bb as well as *g and *gg show a split (b, pp, g, cc/ck). I have wondered if short *b and *g did at first split, too, becoming p and c/k in the positions where they were already plosives (behind nasals, at least *b also word-initially), and then the complementary distribution was averaged out; but in order to test that hypothesis I'd need to study basically all surviving Old Bavarian and Old Alemannic writings, and I have no way of doing that, LOL.
David Marjanović (talk) 21:53, 19 October 2014 (UTC)Reply[reply]

Note to self: Seebold's (1967) summary:
1) PIE *gʷʰ- becomes PGmc *b-, except before *-u-, where it becomes *g- (boukólos rule, not explicitly mentioned by Seebold).
2) PIE *-gʷʰ-, and *-kʷ- affected by Verner's law, become:
a) PGmc *-h- before *-t-.
b) pre-PGmc *-gw- behind nasals, possibly later simplified to *-g- before *-u-, *-ō-, *-j-, *-r-, *-l-, *-n-; *-gw- may also have been the outcome behind liquids, but the only example of that position is Old West Norse ylgr, which – Seebold doesn't say – could have come from *-gw- or from *-g- as far as I can tell.
c) pre-PGmc *-w- ~ *-u- before semivowels, liquids and nasals.
d) Following the developments b) and c), *-w- is dropped before *-u-, *-ū- and *-ō-; *-g- is dropped behind *-i- and *-e-.
3) Lots of analogical leveling within paradigms and within word families.
4) There are few examples of PIE *gʰw and *gʲʰw, but they are "evidently" treated like *gʷʰ. (No idea how Seebold even wanted to tell *gʰw from *gʰʷ.)
David Marjanović (talk) 22:13, 19 October 2014 (UTC)Reply[reply]

Ah yes, that was a tiny bit of OR by me, trying to make sense of the *gʷ problem that mystified me myself. It seems wülpe (which I converted from Upper to Standard MHG as wülbe, probably incorrectly) shows the same artikulatorischer Sprung as *wulfaz, and *gʷ when product of Verner voicing (and thus word-internal) never really became *b at all. Sorry for that. How should we proceed now? What do you advise?
As for the geminates, it seems Upper German retained them even after "heavy" syllables, so West Germanic *wulbbj° would really become *wülpp°, but MHG then simplified that outcome to wülp° eventually. In any case, the West Germanic geminate *bb always hardened, even in Central German, such as in Sippe and Rippe, as you mentioned.
It would seem in any case that the paradigm was something like *wulbī- vs. *wulgjō-z originally, so the introduction of b into the weak cases in West Germanic is clearly secondary. That would mean Seebold is right and there is no problem. However, we would have to assume that the delabialisation prior to *j preceded the artikulatorischer Sprung, as in the weak cases it evidently did not happen.
I believe it is sometimes assumed that in Proto-Germanic, Kw clusters were still distinct from labiovelars, but I'm not sure why, either. --Florian Blaschke (talk) 06:02, 30 October 2014 (UTC)Reply[reply]
Another remark: As far as I'm aware, Middle High German spelling wasn't all that standardised when you look at the actual manuscripts. Indeed, I think that p for Proto-Germanic (and Central/Standard German) b is found word-initially all the way from Bavarian and Alemannic Old High German to the Upper German written language of the early modern period, whence it is still frequently found in place names and surnames, as well as in some Standard German words (such as Pilz) which originally came from Upper German. --Florian Blaschke (talk) 05:31, 1 July 2015 (UTC)Reply[reply]

Can someone shift this to layman's terms?[edit]


I love linguistics, but I find this article hard to follow as it seems to be written for linguistics majors. Can someone add more explanations of stops, fricatives, sibilants, shibboleths, geminis, etc., for those who don't quite understand them? ;) DBlomgren (talk) 20:07, 24 March 2017 (UTC)Reply[reply]

We have Wikipedia articles for all of those terms. CodeCat (talk) 20:34, 24 March 2017 (UTC)Reply[reply]


I feel like the "History" section of this article is written for linguistics specialists who happen not to be familiar with this topic and want to learn about the history of research on the topic rather than what researchers actually say about the origins of this phenomenon, how it happened, why it happened, when it happened, and where it happened. John McWhorter (in several of his "Great Courses" lecture series) seems to think it originates with native speakers of a Semitic language, possibly Phoenicians, who learned Proto-Germanic en masse, and "badly", and passed their version of the language onto future generations. I this a common view among Germanicists? What do others say? Hijiri 88 (やや) 01:33, 18 October 2021 (UTC)Reply[reply]

No, that is decidedly not a common view. I would go so far as to describe it as WP:FRINGE. No one knows why Grimms law happened, any more than we know why sounds change in general.—Ermenrich (talk) 01:56, 18 October 2021 (UTC)Reply[reply]
Do you have a source that says it's fringe? I'm an "interested layman" (or "hobbyist") when it comes to the historical linguistics of the Germanic languages, but judging by your user page you are as well. Don't get me wrong: I'm not under any impression that "classical Japanese literature" and "medieval German literature" are equidistant from historical linguistics as it relates to the Germanic languages. That's why I'm asking you for clarification rather than assuming you don't know what you're talking about and walking away.
McWhorter also does not appear to be a germanicist, but unlike me and apparently you he is a professional linguist, and apparently a reputable enough one that the The Teaching Company have called him back five times to record courses for them.[1] Great Courses have pretty poor coverage of the fields I do have a specialist's knowledge of, but I'm fairly certain that the guy they get to talk about the New Testament and early Christianity (Bart Ehrman) is one of the foremost scholars, if not the foremost scholar, in his field in North America (he also wrote the textbook that is apparently the most widely used in undergraduate courses on the topic in North America), so I'm inclined to believe they commission well-regarded specialists in the fields they are talking about, and said specialists are careful not to present idiosyncratic views as though they were widely held by other scholars. I would hope that, if McWhorter were uncritically presenting a fringe view that was widely rejected by specialists in the field to a mostly lay audience, some colleagues either in Columbia or in a linguistics department at another institution would politely explain to him that he is wrong and he should stop (it does seem that someone corrected him on a number of errors he made in discussion of Japanese that I won't go into, as he didn't repeat those errors in his later courses). The above view has, however, been presented in several of his courses (including the relatively recent "Myths, Lies, and Half-Truths of Language Usage", where several of the errors regarding Japanese were notably absent). One would think that, if the view were being promoted uncritically to a lay audience in such a prominent forum, sources explicitly addressing it as "fringe" would be in no short supply.
I should specify that I don't have a dog in this race, and would be very pleased if you could present me with a source that supports your above statement. So far, browsing around Wikipedia, I've found only a 39-page scholarly review of a book that seemingly predates several of the Great Courses series and that (without working through the entire PDF and rather taking Wikipedia's word for it...) appears to present the view that there was a subsumption of a "Vasconic" substrate ancestral to Basque by a Semitic superstrate to form Proto-Germanic, which seems to be saying that Proto-Germanic wasn't an Indo-European language, which I agree is fringe and bears no resemblance to the view seemingly promoted by McWhorter. (On a loosely related note, do you agree with my reading that McWhorter's view appears to be different from that of Theo Vennemann?)
Hijiri 88 (やや) 00:25, 19 October 2021 (UTC)Reply[reply]
It's fringe because basically no one else cites this idea, which originates with Vennemann, not McWhorter. To quote one review by Nelson Goering (which you can find by searching Vennemann Semitic Germanic at Google scholar):

The result of this, it should by now be clear, is that I must conclude that the central thesis of the book – that there was linguistic influence from Punic on pre-Germanic – is not supported by the evidence marshalled here. More than that, I would say that this Punic hypothesis may now be confidently rejected as very probably untrue. If such influence were real, then the combined labours of two talented linguists, one working on the problem for very many years, and with ample (if often critical) feedback to build on, would have uncovered convincing traces by now. That they have not strongly suggests that such do not exist. If this is the best argument that can be made for the Punic hypothesis, then the hypothesis would appear to be incorrect.

Or here:

Despite the great depth of the hypothesized contact influence, the argumentation is rather detailed, but also speculative and it competes with Nostratic attempts to explain Indo-European Semitic similarities (Bomhard & Kerns 1994).

Nostratic is yet another fairly marginal idea in scholarship.
If you want a standard account in relatively simple language look at Salmons A History of German or Young and Gloning's A History of German Through Texts. You'll find no mention of this thesis. I believe Austronesier and Pfold will back me up on this. We just don't know why sound changes happen, however unsatisfying as that may be, and this idea is not the only one accepted by the vast majority of linguistics.--Ermenrich (talk) 00:41, 19 October 2021 (UTC)Reply[reply]
Hmm... okay, someone should really fix Germanic substrate hypothesis (and perhaps Vasconic substrate hypothesis), which appears to date Vennemann's formulation of the hypothesis to the 2003 book, which (given that McWhorter seems to cite it before that date) led me to the assumption that the one McWhorter presented (which, to be clear, I never assumed to be McWhorter's original formulation) was a separate hypothesis. (That combined with the fact that wording of our article implies that Vennemann considers Proto-Germanic to have been "formed" from a subsumption of a Vasconic language by Semitic language rather than a direct descendant of Proto-Indo-European; something I'm still not sure if he did.) That said, thank you! That pretty handily solves my problem as far as this article goes. Hijiri 88 (やや) 00:56, 19 October 2021 (UTC)Reply[reply]
The idea that languages were spoken in Northern Europe before the arrival of PIE or rather (pre-)Proto-Germanic speakers, and that these languages might have played a role in the emergence of the characteristic Germanic sound shifts is quite old and dates back to the the first half of the 20th century. Vennemann's and McWhorter's approaches are very similar in this context, as they both draw on documented languages and language families as potential contact languages. This is highly unimaginative since we know from fragmentary evidence that pre-IE Europe must have displayed a linguistic diversity that was completely erased with the sole exception of Basque. An interesting recent piece of research about the topic is this paper[2].
But FWIW, sound change does not have to be triggered by contact. Sound change is mostly internally driven. Contact can be an additional factor, and often change is accelerated by social disruption and periods of population bottlenecks, but usually it just happens. A very interesting and readable source about this is Peter Trudgill's Millennia of Language Change (2020)[3].
A word about McWhorter who otherwise is known as anything but unimaginative. He is an established scholar, but that doesn't preclude that some of his bolder ideas are controversial (e.g. his proposal that all languages with an isolating structure have undergone a creolization process) and occasionally utter fringe (e.g. linking the radical typological restructuring of the Central Flores languages to potential contacts with Ebu gogo/Homo florensis![4], the paper is accessible via the search bar of Wikipedia Library). –Austronesier (talk) 09:57, 19 October 2021 (UTC)Reply[reply]
And yes, Germanic substrate hypothesis needs to be radically rewritten. –Austronesier (talk) 10:11, 19 October 2021 (UTC)Reply[reply]

Research history[edit]

Speaking of this heading: it might be worth noting that the similar sound law *p- > f- in Hungarian was identified a bit earlier (Sajnovics 1770, Gyarmathi 1799) and some histories of linguistics claim this as an inspiration for GL. Any opposition to mentioning this? --Trɔpʏliʊmblah 09:07, 22 October 2021 (UTC)Reply[reply]

The pioneering work of Sajnovics and Gyarmathi is mentioned in Comparative_method#Origin_and_development, but without concrete examples. If RS draw a line from their research to the formulation of Grimm's law, a brief mention here is ok, but the wider context could then also be expanded in Comparative_method#Origin_and_development. –Austronesier (talk) 10:34, 22 October 2021 (UTC)Reply[reply]