Irmin, Aryaman, et al.

JC was kind enough to send me a link to John D. Bengtson’s Iarl and Iormun-; Arya- and Aryaman-: A Study in Indo-European Comparative Mythology. The abstract:

In 1854 Martin Haug of Heidelberg suggested a root connection between the obscure German god Irmin and the minor Indic god Aryaman. Almost a century later (1952) Jan de Vries of Leiden agreed, with some reservations, and since then this theory has remained in dispute. In my study of this subject several arguments support the Haug – Vries hypothesis […]. The argument that Irmin simply means ‘great, immense, elevated’ and is the sole remnant of the Indo-European middle participle in Germanic is implausible; the form Irmines- is clearly the genitive form of a name. The oldest sources and comparative mythology point to Irmin / Iǫrmun as some kind of divine or heroic entity closely connected with sovereignty, ancestry, and the collective life of the people (irmin-diot). In the post-Christian literary traditions of the Germanic and Celtic peoples the original patterns were transformed and distorted.

JC adds:

This paper is by John D. Bengtson, and the title “Iarl and Iormun- : Arya- and Aryaman-: A Study in Indo-European Comparative Mythology” says it all: it goes right back to the roots of IE and historical linguistics generally, and is a direct descendant of Grimm’s philological program: to turn prehistory into history using comparative linguistics.

Well, almost all: there is also the Irish hero Éremón, who is said to have arranged many marriage bonds between Irish noble families (of which there were A LOT). Hindu ceremonies today still appeal to Aryaman to bless the marriage, even though he gets only a few mentions even in the Vedas, much less later.

We don’t know much about the Germanic Irmin, except that there are a lot of old names in Irmin-this-that-and-the-other, especially in Upper Saxony, and a few words of mythological import: OE éormengrund ‘orbis terrarum’, ON iormungand ‘the Midgard Serpent’ (the same gand ‘staff, wand’ as in Gandalf) , OHG Irmunsúl ‘axis mundi’ (destroyed by Charlemagne in the 8C). The usual, but not the oldest, interpretation is ‘grand’, as if cognate with ὄρµενος ‘inciter’; but that is a middle participle, and if the etymology is sound it would be the only middle participle in all Germanic.

Bengtson points out that Arminius, was probably named with a close variant of Irmin-, and that the h in personal and tribal names mentioned by later Roman writers, like the tribe of the Herminones, was silent, suggesting that they too are Irmin- based (and that Luther’s Germanization of Arminius as Hermann is a spelling pronunciation!)

Thanks, John!


  1. Trond Engen says

    I haven’t had time to read the paper, but it sounds very interesting. Strange to think that Bengtson actually can do work on this level of detail.

    FWIW, the Germanic comparanda with clear referents all denote something to do with the concept of the world rather than “honour” or “family”. I once had the thought that they could be derived from the root of the “earth” word.

  2. David Marjanović says

    So, they’re all root cognates, and the root used to be overlooked because it’s rather short. And at the same time all the arya- stuff is explained, and I finally learn the (rather trivial) explanation for Karl vs. Kerl.

    The one sentence that needs a correction is this:

    At the present time Irmin is still a common German name, for both males and females.

    …followed by footnote 25, which cites one born in 1937 and one born in 1940. That explains why I haven’t ever encountered it: it was part of the brief revival of rare Germanic names that, um, abruptly ended in 1945.

    I’m related to an Irmgard who was born a bit earlier than that, and I’ve heard of people named Irmtraut, but that’s it.

    Interesting that Erich and Erwin may be root cognates, too.

  3. In untils iarmun kruntar, what does kruntar mean? Sea?

    And what is sul/seul in Irminsul/Armenseul? Pole?

  4. David Marjanović says

    Grund = ground, floor, bottom.

    Säule = column.

    Anyway, here’s the conference presentation about the paper. It adds this gem about Ursa Major:

    Cf. Scandinavian Karlavagnen, Karlsvogna, Karlsvognen (orig. < Tor-karl = Þórr, later interpreted as Karl den store; cf. Middle English Cherlemaynes-wayne); Old English Carlswæn > Northeastern English Charlie’s Wagon

    And so has el gran Carlemany ended up in the heavens.

    …while HEREMON O’NEILL has ended up in a not-quite-superhero comic.

  5. Jen in Edinburgh says

    I’m related to an Irmgard who was born a bit earlier than that, and I’ve heard of people named Irmtraut, but that’s it.

    Presumably the same names as Ermintrude of the the Magic Roundabout and Ermengarde from Frances Hodgson Burnett’s Little Princess.

    There seems to have been a bit of an Edwardian fashion for Germanic names in English – a Victorian fashion when they were named, I suppose, but I’m not so aware of it earlier in the Victorian period.

  6. David Eddyshaw says

    I actually used to know an Irmtraut. She bore it bravely.

  7. David Marjanović says

    …and I completely forgot to mention the Italian name Ermenegildo!

    Also, this is the paper where I finally noticed that Held “hero” and the name (-)hild(e)- (OHG hiltia “battle”) are related!

    She bore it bravely.

    Oh, there’s worse, much worse. Also, she has a choice of going by Irmi or Traudi!

  8. >>In untils iarmun kruntar, what does kruntar mean? Sea?
    >grund=ground, floor, bottom

    Now I get Bengtsen’s sentence — “ie sea” defining the 3-word phrase, not a meaning of iarmun nor of iarmun kruntar. Sorry. Not a lot of experience with kennings here.

    I’ve known a couple women named Irma, and if wiki is correct that “the anglicized form is Emma,” then a name derived from this root is the most popular girl’s name in the Anglosphere.

  9. AJP Crown says

    Not a million miles away there’s also the German surname Ehrmanntraut (and the character Mike Ehrmantraut in Breaking Bad, played by Jonathan Banks).

  10. PlasticPaddy says

    I have the perhaps unfair impression that the Celtic etymon is used inappropriately as collateral support. Since Ariomanus can be analysed in Gaulish as “the good lord”, it would seem equally simple to class the Germanic “cognates” as due to borrowing, with the various meanings due to a ripple effect, precisely because the root does not really mean anything in Germanic.

  11. David Marjanović says


    That’s mentioned in the paper.

    with the various meanings due to a ripple effect, precisely because the root does not really mean anything in Germanic.

    *lightbulb moment*

    Maybe that actually explains the seemingly random variation between *er- (including ir- and iVr- by umlaut) and *ar-: the former native, the latter Gaulish? The paper mentions ablaut, which works, of course, but does not attempt any morphological analysis to predict which ablaut grade we’d expect in which forms (before wave upon wave of analogy at least).

  12. >Mike Ehrmantraut in Breaking Bad,

    And I think the penultimate episode, “Ozymandias”, references the destruction of the Irminsul.

    Or maybe that was another ancient monument. 😉

  13. Modern wands would seem to be stiff. Sticks would be an appropriate metaphor. But the discussion in this paper reveals to me that the word is related to winding and wending, though perhaps more in the manner of a fresh bough bending stiffly as you swing it than a snake sinuously bending. Which in turn has me thinking of Moses and Aaron with pharaoh. The snake/rod/wand metaphor always seemed off to me, likely because wands and rods are solid in my mind. The phallic aspect is there too, I guess, which may unite these alternately flexible and stiff objects.

    Just a way of thinking about wands that was outside my kenning.

    The idea of a great snake monster as a “world-wand” still seems strange to me. But maybe hellbender (a salamander) is a useful clue on how to think about it.

  14. Also, it’s impossible to google any of this without ending up on a page whose author either identifies as “a Yellow Solar Seed in the Mayan system and an intuitive/feeling type” or has adopted a Norse god as avatar to express with great subtlety his devotion to things northern and snow-colored.

  15. David Marjanović says

    and an intuitive/feeling type

    I mean, I knew Myers/Briggs belonged to psychoceramics rather than psychology, but I did not expect such a stark illustration. 😀

    to express with great subtlety his devotion to things northern and snow-colored

    Like this.

  16. One tentative word in defense of Myers/Briggs:

    I was introduced to it at a faculty “workshop” in 1988. After we’d found our types and discussed them for a while in regard to teaching style and the like, the speaker asked, “Is there anyone here with type [whatever]?” One hand went up. “Well, that’s very interesting, because people with this type seldom enter teaching and, if they do, usually get pushed out soon.” The fellow who’d put his hand up grinned and turned very red, and I knew why: He was a member of my department who was always in trouble because no one could understand why he did the things he did, and he bristled when asked to explain. He was one of these people who bounce from one small institution to another, and he left a year later, I think for some other kind of job.

  17. Lars Mathiesen says

    Myers/Briggs does have some predictive power, though — in my experience, most of the people who really like the idea of a four bit personality fingerprint get classified INTJ.

  18. Lars: Haha! Nailed me.

  19. But some of us who think it’s modern astrology get put into that bucket too.

  20. Trond Engen says

    Does anyone know more about Martin Haug? His surname looks like it couldn’t be anything but Norwegian, but his Wikipedia biography says he was a farmer’s son from Ostdorf in Würtemberg.

  21. Lars Mathiesen says

    @Keith, I’m INTJ as well (or anything else with starting with an I, depending on the constellations and prevailing winds) — the causation only works one way, as far as I can tell.

  22. PlasticPaddy says

    This explains that there is a different etymology for German Haug. Martins father came from Glatten in Baden-Württemberg and there are still Haugs listed in the phone book there.

  23. This does not strike me as a very satisfactory or useful treatment of very interesting material. The appeal to stem variation was a way to get rid of the medial -ya- in Aryaman is not paralleled by any of the examples he cites in support, and would be rather odd. It certainly would deserve a lengthy discussion on its own, and is an absolutely essential preliminary to carrying on the discussion any further.

    I’d also note that the supposed Tune reading *arjostez that he cites in support of his Germanic *er- root doesn’t exist. The runes very clearly read sijostez, and it’s really wishful thinking (often of the nauseatingly nationalistic sort) that conjured up a supposed reversed to give Germanic some Aryans.

    I also can’t see what’s so objectionable about there being an occasional middle participle fossilized in a particular lexeme. The ‘ablaut’ is no objection, since this kind of suffix alternation was very common in Germanic, often in ways that bore little relationship to the IE ablaut patterns from which the various alternants formally derived. It wouldn’t have been variation in *-min-, but just in *-in-, in Germanic terms. Of course, this doesn’t prove that *erm-n- was from a middle formation, but it’s hardly a sound objection to that proposal.

    The whole thing reminds me a bit of Voltaire’s famous dictum…

  24. I’d also note that the supposed Tune reading *arjostez that he cites in support of his Germanic *er- root doesn’t exist. The runes very clearly read sijostez

    Yikes. That’s a pretty bad sign.

  25. David Eddyshaw says

    I took the online Myers-Briggs thingy, and I turn out to be Elon Musk. Evidently there is a problem with the methodology.

  26. @Ryan: Hellbenders (or at least the one at the zoo here) are really broad—not rod or wand like at all, in my mind.

    I have a certain fondness for hellbenders, since they were my patrol mascot during one of the years I was Boy Scout. Every year, the patrols in my troop were supposed to decide on new names, although for my first couple of years, patrol names were largely neglected. However, a new scoutmaster decided we needed more patrol cohesion and spirit, so at one meeting, each patrol was tasked with picking a new name and then eventually developing a patrol yell and other such things. There was some lack of enthusiasm in my patrol, and not many good ideas for names; so the senior patrol leader told us that if we couldn’t think of anything, we could use any animal listed in the nature part of the Boy Scout Manual. That included the hellbender, so for the next year, we were the “Hellbender (Salamander) Patrol.”

    The next year, we changed the name again, and had much more positive results as “Santa’s Sleighers.”

  27. J.W. Brewer says

    There was actually an important breakthrough in public recognition and respect for hellbenders just last year:

  28. John Cowan says

    Help! I’m an INTP trapped in a (local) world of INTJs!

  29. Trond Engen says

    Now I got curious and took an online test too. Turns out I’m an E trapped in a world of Is!

    (Apparently fairly clear E and P, also more N than S, and maybe a little more F than T. But given how haphazardly I assigned answers to the more introspective questions, that’s probably a random fluke.)

  30. David Marjanović says

    Yay, hellbenders!

    conjured up a supposed reversed to give Germanic some Aryans

    A reversed what?

    I took the online Myers-Briggs thingy, and I turn out to be Elon Musk.

    Ну и как тебе это, Илон Маск?” 😀

  31. I got curious too; it labeled me as “Adventurer ISFP-a.” But half the questions I would doubtless have answered differently at another moment.

  32. David Marjanović, that should be ‘reversed R’ — I made the mistake of putting it in angled brackets to indicate an orthographic form, forgetting how the computer would construe this…

    Anyone who wants to take a look, there’s a fairly legible drawing of the Tune Stone here:

    The relevant bit is on Side B (to the right, in the drawing), third line (leftmost), which is read bottom-to-top. Normal instances of R appear in the second rune in and the seventh from the end. Both show the R-rune facing the direction of writing (like a Latin R would), and give an idea of the expected angles and strokes. The SI in question is the seventh and eight runes from the start of that line, and is clearly carved differently (probably more than would be accounted for by the messiness of inscribing, though there’s some give on this point), and if read as R would be mirrored relative to the direction of the line, unlike every other R in the inscription. On the other hand, these particular squiggles are unremarkable as S and I.

    There’s an excellent discussion of this inscription in general, including quite a lot of this problem in particular, by Þórhallur Eyþórsson:

  33. Trond Engen says

    Thanks. I thought I had something recent on the Tune inscription, but I haven’t been able to find it.

    Bengtson’s reading is footnoted “Antonsen (1975)”. I haven’t access to that, but it doesn’t matter much. If it says what Bengtson claims it does, the reading was never accepted, and either way it’s obviously superceded by later scholarship. But if I understand it correctly, this is a resurrection of work Bengtson did in his university days in the seventies. That might also go some way to explain why it’s so different from his later work.

  34. Rodger C says

    Otto von Friesen, in his article “Runes” in the old Britannica, read the word as “si(bi)josteR.” I don’t know enough Germanic to see what he was getting at there. On the other hand, the article has a number of misprints in the transcriptions. (If Þórhallur’s [Þórhall’s?] article explains it, sorry, I don’t have access.)

  35. @Brett: Hellbenders (or at least the one at the zoo here) are really broad

    I’ve surely never seen one in the wild. Maybe at a state park visitor center once or twice. I’d pretty much forgotten exactly what they look like. From Hellbenders to Santa’s Sleighers. The most goth scouts ever, struggling for self-expression under the leadership of Gallant from Highlights magazine. “Well, they did keep Santa in there, so I guess they’re good boys.”

    I wasn’t thinking so much in terms of direct connection or analogy. Simply that hellbender makes me realize that a medieval name for a creature that means something like “earth-god wender” or “awesome winder” (with awe evoking more of the awe of god than it does today) shouldn’t surprise me as much as it does. Let alone if you consider iarmungandr a kenning rather than a name.

  36. I’m so unqualified to interpret the Tune Stone that I initially thought spellcheck had changed your post from runestone. But I am curious – the wiki page has

    >the b-side reads:

    and then

    >The transcription is
    >arbija arjostez(?) arbijano

    Which is footnoted to a site that is in a Scandinavian language, so that’s where the trail ends for me.

    Did Bengtson or someone close to him change the transcription line in the wiki page, but not the “rune reads” line or is there wider belief that arjostez is reasonable, contra your post? “Just another “nauseatingly nationalistic” person who can read runes?

    Is the word sijostez not otherwise found? Or why is there a strain of scholarship that believes it should read arjostez?

    eilu v’ eilu divrei elohim chayim

    (I’ll leave this post, even though I suspect it’s mostly answered above in ways I’ll need time to understand and digest.)

  37. Ah, cripe. Boustrophedon is one thing, but when the top/bottom orientation changes from line to line…

    We need new terms. What’s the Greek for “periodic somersault” and “as the mower circles inward on the lawn”?

  38. The reading (a)rjostez I think goes back to Krause, who was a foundational and deeply influential runic scholar, but also at least Nazi adjacent (I’ve heard different accounts of just how enthusiastically he embraced the ideology). Not nearly everyone who’s repeated this idea is themselves a Nazi sympathizer, obviously, but the word could hardly have gained much initial traction without the significant aid of a lot of nationalist wishful thinking on Krause’s part (his stature and the wide use of his edition was enough to propagate it further).

    Antonsen is probably the most egregious offender here. As far as I know, he was not at all a Nazi sympathizer, but he repeats Krause’s view without caveat or qualification — contrary to his usual practice of noting irregularities in readings. I wonder he was just being a bit lazy at that point, working from memory rather than carefully rechecking the inscription as he usually would have done. Anyway, as far as I know that was the last major runological work to promote this reading, and basically everyone since has agreed that this is a pretty clear phantom.

    I see it’s time to go make supper, so I don’t have time to summarize Eyþórsson’s argument right now.

  39. Lars Mathiesen says

    Krause is covered in Eyþórsson’s (Eyþórssonar?) article, in a 1934 article he basically wrote “Look, the old tribes agreed that being Aryan was the shit!” — and mysteriously that phrasing disappeared in later recensions though the cognacy to the Indian word was maintained.

  40. >There’s an excellent discussion of this inscription in general, including quite a lot of this problem in particular, by Þórhallur Eyþórsson

    Yes. Also a lot resting on the author’s use of “ipsissimus” as a superlative reflexive to bolster sijostez as such a formation, “his ownest heirs”. “Suus heres”, which he says somehow supports this view, seems completely expected. For actual support, I’d want to see “suissimus heres” somewhere.

    It’s interesting to me that the author doesn’t mentioned what I pointed out above, that in addition to boustrophedon writing, the top/bottom orientation changes from B1 to B2, but not from B2 to B3. That seems like another sign, stronger than the ones adduced, that these are not part of the same 3-line passage.

    The reversed-R theory does seem strained. Two things not mentioned that it has going for it are the clumsy R in thrijoz, suggesting the inscriber had trouble with R’s, and the fact that the other two S’s (the second S in sijostez and the S in steina) start close to the preceding letter at the top (ie, L/R orientation does not change with the direction of the line, as it does in other letters.). To my eye, that’s an easy mistake to make. But then, to my eye, having the Rs and Ps (as Ws) “face forward” into the line of writing rather than remaining in the same orientation is unnatural. It seems to be natural for boustrophedontists. This scribe clearly had trouble with it, either in the S or the R.

    But the reversed-R theory still seems strained. You have to believe he got it backward and failed to link the staves.

    It needed better scribes.

  41. Gah! As you start to reckon with the change in top-bottom orientation, a lot of things become strange. Why would the scribe write B1 with the right as “top” and then switch to left/top for B2? That seems pretty unusual, as does the spacing between B1 and B2. This argues that they are separate passages.

    However, if you accept B2/B3 as being independent of B1, then B3 is “above” B2, seemingly giving a more natural reading of

    arbija si- (or ar-) jostez arbijano
    thrijos dohtris da(i)lidun

    I don’t know what new problems of word order that might create. The article does mention that verb-final is a possibility in the language of the early runic inscriptions. It’s certainly not a reading mentioned. But it seems like someone should at least address it, since it is a pretty obvious potential interpretation of the orientation of these two lines.

    Lest anyone think I just don’t get boustrophedon, and that I’m confusing top/bottom with right/left, understand the effect of the B-side of this rune is like when you’ve got Spanish and English books next to each other on the shelf, and you have to flip your head to read the title of one then the other (or read the letter-upside down). To read boustrophedon, you hold your head the same way, and you just have to move your eyes right to left, then left to right.

    But to read B1 and then B2, you have to lean your head right for B1, and then lean your head left for B2; and then for line B3, with your head now leaning left, rather than switching again, you go up a line. That seems pretty strange to me for a single passage. But if two passages, then why assume B3 is after B2?

    I hesitate to ask, but what cases can thrijos dohtris be interpreted as? Not knowing declensions, reading B3 as the first line seems to open the possibility that the three daughters were not heirs, but rather they were the inheritance. And while I defer to others on the possibility that daughters could be heirs in ancient Germanic cultures, inheriting them seems plausible, arguably more so.

  42. David Marjanović says

    If Þórhallur’s [Þórhall’s?] article explains it, sorry, I don’t have access.

    If the paper just doesn’t load, try again; is in open access, and I’m not sitting behind a VPN or anything here either.

    Haven’t finished reading it yet, but I’m convinced it’s si rather than a suddenly mirrored and broken r, and that dalidun is *dailidun with the L serving as a probably accidental bindrune for I and L.

    I also find the language intriguing: like the Horn of Gallehus inscription, which is always said to be the same age, it’s a bit before Proto-Northwest-Germanic – short /o/ is fully phonologized, but word-final long /oː/ still hadn’t become /u/. Also, the paper features an obsolete German word I didn’t know (Gesippen “relatives”) and hyggelig.

    Ausrichten, BTW, means “organize and announce” in this context.

  43. And indeed, Vigfusson in the 19th century read it this way starting with B3 then B2.

    Then things get muddled. The citation to Vigfusson in a Liverpool Royal Proceedings has him reading the words completely differently, arbijas ijos tez-arbijan(d) (That book ascribes the D to a poor copy of the inscription.)

    However, tracing it back to Vigfusson, one edition has him using this order, but reading -ng for the -j rune. In his later Corpus Poeticum Boreale, he sticks with arbijan(d), realizes the error of -ng, but reverses himself on line order.


    He argues tez- is Latin dis-, and that you should read tez-arbijand that they are “dis-owning,” ie, pulling apart the inheritance into portions. I wonder whether realizing that the rune is pretty clearly o, not d, made him (or later readers) abandon that reading completely.

  44. John Cowan says

    Eyþórsson’s (Eyþórssonar?)

    Nope. As the Hat said in defense of the hoi polloi, to speak English you only need to know the English language.

    suus heres

    That’s a technical term of Roman law meaning an heir who was in the power of the testator at the time of his death (a wife by a strict-form marriage, a child, a child’s descendants, in principle a slave). The antonym is alienus heres.

  45. David Marjanović says

    He argues tez- is Latin dis-

    And German zer-, but I’d interpret *zererben in analogy with zerreißen “rip apart”, zerstören “destroy”, zertrümmern “smash”… also, I’d be really surprised if the e is that old.

    However, concerning sijostez, German has a superlative of “own”: ureigenst- “one’s very own (interest, usually)”.

  46. Trond Engen says

    @ Ryan: Thanks for the discussion and the observations on the boustrophedon.

    Eyþórsson does a very good job summing up and sorting out the evidence, and he makes a strong case for his reading. But I really, really think the need for a full word on the end of line B1 is a problem, even if the reading needs a verb there. One possibility I don’t see mentioned is that B1 is the continuation of A1-2, and that the missing runes at the start of B1 belong to the word beginning with the broken rune at the end of A2. That would have the double benefit of minimizing the lost shard from side A and not messing with the straight edge on top of side B. I just wish I had a suggestion for piecing that together.

  47. Trond Engen says

    @Ryan: I meant to add that I think the meaning of lines B2-3 would be independent of their internal order, The language here is poetic in word choice and rhythm. If anything, I think the masculine agreement is somewhat more plausible if the inheritants are mentioned before the daughters have been introduced,

    Inheritance, closest of relatives,
    three daughters shared.

    .. but that would mean that the first line of the inscription started from the lower end of the stone and stopped before reaching the top, while the last line filled the whole height. All in all I’d prefer starting with the three daughters.

  48. John Cowan says

    the three daughters were not heirs, but rather they were the inheritance

    Indeed they would be by Roman law. The heres stepped completely into the position of the testator, and whoever was in the testator’s power when he died, including all his children, would now be in the power of the heres, including his unmarried daughters.

    I’m sticking with the Latin word because its meaning is subtly but profoundly different from what common-law countries know as an heir. In particular, the liabilities of the testator were now the liabilities of the heres, even if they exceeded the assets he received. In the Empire it was common for men of property to appoint the Emperor as heres in their wills, as he was expected to pay whatever legacies the testator left to particular persons.

    (My reference to “strict-form marriage” in a previous comment was incorrect: all married women were in their husband’s power unless (a) they had not been married in strict form and (b) they absented themselves from the marital home at least three nights in a year. Many perfectly respectable Roman matrons did this in the late Republic and the Empire in order to keep their own property rights; if they failed to do so, they would become married by usus, the Roman-law version of prescription.

  49. Trond Engen says

    Also forgot: Þrijos dohtris can be accusative as well as nominative.

    Note that these words have -s, while others have -z. It’s interesting, but I don’t know if anything deep can be said about the distribution.

  50. David Marjanović says

    No, they have z.

  51. Trond Engen says

    Oh, I noticed the s-es in the quoted words now and didnt go back to check. Sorry for the confusion

  52. Trond Engen says

    The danger of commenting while bored in a Teams meeting.

  53. And of typing from memory when you only know Latin, no old Germanic whatsoever. Sorry. The confusion was all mine.

  54. David Marjanović says

    I keep forgetting to ask: how did Torkarl happen? Did Christians start referring to him as “y’know, the thunder guy”?

  55. The declension of haubith/haubidis (head) in Gothic has me thinking about thrijoz dohtriz da(i)lidun. If “th” was perceived as an intrinsic part of a -d paradigm, then that line arguably had an alliterative triplet, another minor point in favor or arjostez. Arbija sijostez arbijano hardly seems to count as poetic alliteration since the only alliteration comes from lightly differentiated forms of the same root.

  56. PlasticPaddy says

    Maybe not Christians…

  57. David Marjanović says

    If “th” was perceived as an intrinsic part of a -d paradigm

    Gothic had word-final fortition of fricatives as a special feature that does not go back to Proto-Germanic.

    Maybe not Christians…

    Oh, I had overlooked the “old man” meaning of karl. Everything makes sense now.

  58. Haven’t finished reading it yet, but I’m convinced it’s si rather than a suddenly mirrored and broken r, and that dalidun is *dailidun

    Those are two things I also take as certain. The interpretation beyond that could maybe be improved, but those are really extremely likely to be right.

    I also find the language intriguing: like the Horn of Gallehus inscription, which is always said to be the same age, it’s a bit before Proto-Northwest-Germanic

    A bit after, rather. West Germanic had not only separated by the date of this inscription, but Ingvaeonic had likely emerged as a distinct dialect within that group. I’d mention that word-final *-ō had indeed shortened to -u in the language of these inscriptions (no examples in this particular text though). The examples of -o you see here are all from *-ôⁿ (as in the genitive plural) or *-ōⁿ (as in the verbal ending of worahto). The merger of *-ô(ⁿ) and *-ōⁿ as, initially, *-ō is actually one of the relatively few distinctively North Germanic innovations visible in Early Runic.

  59. David Marjanović says

    Oh! I didn’t expect nasality in a vocalic 1sg ending.

    So, is the Horn of Gallehus North Germanic, too, then?

  60. It sort of depends on your terminology. Linguistically, Gallehus is completely normal Early Runic, which shows extremely few innovations from Proto-Northwest-Germanic — but the corpus as a whole does show a very small number, and by the time the Gallehus horn was made (if the archaeologists are right in their dating), West Germanic would have lost several features seen on the horn.

    Antonsen, rather bizarrely in my view, decides to call this language ‘Northwest Germanic’, since it’s very little changed from PNWGmc. He’s perfectly explicit that it’s not literally a proto-language, and that this ‘Northwest Germanic’ coexists with West Germanic. This needlessly confusing terminology is a result of his irritation with people regarding Early Runic as ‘Proto-Norse’, and trying to interpret (often anachronistically) inscriptions through later Norse eyes. A fair enough objection, but his replacement term is just a bit too weird for me. Hans Frede Nielsen has really promoted the term Early Runic as a language name, which works for me.

    It’s interesting that Gallehus is found in an area that we’d otherwise think of as probably Ingvaeonic-speaking. But I’m not sure how much significance to place on this. For one thing, runic objects move around. Early Runic may also have been used artificially beyond places it was actually spoken (this might also explain the relative lack of dialect variation we seem to find, though there are possible linguistic explanations for this too). And that area would eventually end up as North Germanic — it’s possible that process began earlier than the conventional histories would have it (on different grounds, people have periodically suggested that the movement of ‘Angli’ to Britain was a result of people being pushed out by expanded Danish power). It’s all a bit murky, but the main point is that we don’t really have any good geographical grounds for taking it as West Germanic.

  61. or *-ōⁿ (as in the verbal ending of worahto)
    Just to refresh my memory – the ending goes back to PIE *-dhoH1-m?

  62. The opera character Norma is a druid priestess of the god “Irminsul”. I always wondered if the set design for the scenes in Irminsul’s temple ever features a column or tree trunk or something, or if the opera and all of its interpreters have always just treated the whole word as a god’s proper name, e basta?

  63. or if the opera and all of its interpreters have always just treated the whole word as a god’s proper name, e basta?

    Why wouldn’t they? Opera lovers and performers are not etymologists, nor were meant to be.

  64. Sure, I was just wondering if the “pillar” idea ever bled into the set design … from a quick look it seems like earlier sets were just temples, but it seems a likeness of the Saxon Irminsul has shown up at least once in a recent production:

    Wagner certainly fancied himself quite the philologist, although in my opinion his efforts in that area are more something that audiences endure rather than enjoy. For example, I think at one point he decided that “Nibelungen” = “Wibelungen” and thus that the Nibelungs were the Guelphs and wrote a treatise on it, an idea that I doubt anybody but Wagner was all that interested in.

  65. Rodger C says

    @AG: I strongly suspect that this recent design feature is a result of the recent ability to do a Google image search of “Irminsul.”

  66. Just to refresh my memory – the ending goes back to PIE *-dhoH1-m?

    That depends on your theory of the weak preterite. I’m not sure any of the recent treatments of the problem would suppose a form like that for PIE proper, but Hill figures that that, or the similar *dʰō-m, was created analogically in the history of Germanic (from an aorist subjunctive reinterpreted as a present indicative — this happens in Germanic — which then acquired a new imperfect, the 1sg. of which was *dʰṓm, in his reconstruction).

    Ringe, on the other hand, figures its a form of the 1sg. imperfect as it was in PIE, *dʰé-dʰeh₁m, and assumes a sound change of *-ēm > *-āⁿ > *-ōⁿ.

    Basically, we don’t have a really convenient PIE *dʰoh₁m ready-made in the right kind of paradigm to just become the 1sg. of the weak preterite, so some kind of analogy or sound change is needed to get PGmc *-dōⁿ. This is one of the many wrinkles that makes the weak preterite so debated, and not the most problematic one by a long shot.

  67. David Marjanović says

    from an aorist subjunctive reinterpreted as a present indicative — this happens in Germanic —

    It happens particularly to replace reduplicated forms like *dʰé-dʰeh₁m.


    Ah, but *-eh₁m never acquired a long vowel. This isn’t Indo-Iranian, where first the syllabic nasals were resolved, then the laryngeals dropped out, and then the syllable boundaries between two nuclei were abandoned. (With the syllable boundaries spelled out, that would mean *-e.h₁m̩ > *-a.ham > * > *-ām.) In Germanic, illustrated most beautifully in Hill’s paper on the wind, the laryngeals dropped out first (*-e.h₁m̩ > *-e.m̩), and then the syllable boundaries between two nuclei were abandoned (*-e.m̩ > *-em) before the remainng syllabic resonants were resolved (and *u was inserted in a few other situations as well, e.g. *melk > *meluk).

    (I’ll look for links later if nobody beats me to it.)

    In any case, the *h₁ is syllable-initial, so it is not expected to cause compensatory lengthening when it is dropped. Ringe, Kroonen and many others seem to have assumed that every laryngeal which happened to follow a vowel caused lengthening; they’ve ended up with far too many long vowels which they need to get rid of by contrivances like Osthoff’s law, for which there doesn’t seem to be any Germanic evidence otherwise.

    Assuming a period when Pre-Germanic allowed vowel clusters also seems to get rid of the very oddly distributed overlong vowels (which Ringe assumes and Kroonen, as far as I understand, sweeps under the carpet).

  68. Stangs law explains the > *-ēm word-finally, and this is plainly correct for Germanic. No one* would reconstruct the ō-stem ending as short, this is clearly *-ōⁿ < *-eh₂m. I don't recall Hill ever challenging this, and certainly his views on 'wind' (which rather suggest Germanic behaved _more_ like Indo-Iranic, not less, unless there's something beyond his 2005 paper that I'm not familiar with) are unrelated. Hill also explicitly says he follows Schrijver’s (incorrect) views on the Germanic auslautsgetsetze, who also accepts Stang’s law.

    Kümmel has a rather good paper on the anaptyctic *u in *meluk, etc.

    Aorist forms replace quite a few varieties of characterized presents, but yes, reduplicated presents were particularly thoroughly eliminated in Germanic. On the whole, I do like Hill's attempt to look for morphological ways to get rid of the reduplication than Ringe's hapologies, and the subjunctive is a good way to bring in the very 'thematic'-looking o-e-e alternations of the singular. Though I suspect there may have been some in-between developments that are too complex and too old to actually reconstruct in all their detail, and I've tried to keep from getting overly attached to any one theory about the exact development of the weak preterite.

    *(I say this realizing that nearly every permutation, no matter how untenable, has probably been suggested by _someone_. But this is not a position that could find much reasonable support.)

  69. David Marjanović says

    Stangs law explains the > *-ēm word-finally

    Ah, I didn’t know it had been extended to *h₁. My skepticism about its original extension to *h₂ rests on this paper by Tijmen Pronk on “Stang’s law in Baltic, Greek and Indo-Iranian”. (Unfortunately it was written a bit too early to profit from Byrd’s work on PIE syllables.)

    That paper barely mentions Germanic. But I don’t think the question has ever been asked what we’d expect the outcome of word-final *-a.m̩ (itself the expected outcome of *-e.h₂m̩) to be. How about a long nasal vowel that preserves both moras and the nasality? In other words, *-a.h₂m̩ > *-a.m̩ > *-āⁿ > *-ōⁿ?

    If so, we’d also expect *-e.h₁m̩ >> *ēⁿ, which could have developed as Ringe suggested or been analogically replaced.

    I don’t recall Hill ever challenging this

    Not explicitly; the paper I had in mind is indeed this one from 2005 (p. 112 onwards). Word-final position isn’t mentioned anywhere in the whole paper (and Schrijver is cited for something else). However, when word-internal *-e.h₁n̩- ends up as *-en- with a short vowel, it isn’t unreasonable to assume that word-final *-e.h₁m̩ should become *-em (> *-eⁿ or whatever).

    That said, word-final position is the most likely place for a mora-preserving change like I suggest above.

    Kümmel has a rather good paper on the anaptyctic *u in *meluk, etc.

    Yes! Here it is.

  70. Hill says in his 2010 paper that he’s following Schrijver’s developments for word-final syllables. I’d have to read his paper again to see if he’s ever explicit about a word-final *-VHN > *V̄N, but it seems implicit, both in the Schrijver reference, and in his proposal in general.

    Proto-Germanic *-eⁿ is probably impossible. There was no unstressed *e in Germanic at all, so we’d expect this to either become *-iⁿ or *-aⁿ, both of which have well-understood outcomes in the various languages. A source like *-am̥ is imaginable, I guess, but I don’t think very well motivated.

    I haven’t seen anyone challenge Stang’s law for Germanic, and even if Hill is right, I wouldn’t particularly expect the word-internal developments (which likely took place much earlier, and are in rather distinctive prosodic contexts) to mirror word-internal ones (which I’d attribute to Osthoff’s law anyway, though this is a murky area, and Hill might be right after all).

  71. David Marjanović says

    There was no unstressed *e in Germanic at all

    Not that it really matters here, but what do you think of this?

    In the journal article which constitutes the fourth paper, the research situation concerning i-umlaut is scrutinised and, based on internal reconstruction, the defectiveness of previous attempts to explain the distribution of fronting in the vocabulary is illustrated. In the paper, ill-fitting data are reconfigured to facilitate a phonological explanation for why ‘front umlaut’ (term preferred over “palatal umlaut” or “front mutation”) occurs variably in light-stem paradigms, even when least expected, as in the feminine abstracts in *-iþu (cf. Old Swedish dygþ ‘virtue’). A genuinely novel solution is proposed, based on the assumption that the contrast between Pre-Germanic */i/ and */e/ was upheld, not only in main stressed syllables, but also in syllables of relative prominence. A chain shift affecting the descendants of the proto-vowels is postulated and verified by their alterability in main stressed syllables when targeted by rounding umlauts and breaking. The same distinction and chain shift applied to trigger vowels and only descendants of Pre-Germanic */e/ triggered a front umlaut unconditionally.

    (The idea is that, in North Germanic, *i – following the Pre-Germanic *e…i > *i…i umlaut – changed to [ɨ] or something except under certain conditions like a following *z, *e changed to [i], [i] triggered umlaut, and then [ɨ] turned back into i if it was still there. Other people have wondered if in West Germanic, *i triggered umlaut while *e did not, explaining such things as 3sg fährt vs. 2pl fahrt in Standard German.)

    A source like *-am̥ is imaginable, I guess, but I don’t think very well motivated.

    Well, I’d say the lack of phonetic sense the extension of Stang’s law makes is a pretty good motivation – and the assumption behind that extension is exactly the same assumption that led people to expect a long vowel in the history of wind, which Hill has shown is unnecessary at best.

    Full disclosure: I may underestimate how awkward awkward syllable boundaries are for most people. /em̩/ is a word in my Central Bavarian dialect (eben), and /aˈɛɪa/ – four syllables, no consonants other than the automatic postpausal glottal stop – is a sentence: “oh, so it is me after all, good”.

  72. Lars Mathiesen says

    What’s with all the *-eⁿ lately, why not *-ẽ? IPA calls ◌̃ “nasalized” and ◌ⁿ “nasal release,” and releases are things that stops have innit.

  73. David Marjanović says

    Half of Schalin’s thesis isn’t in his thesis, but in two papers that are only accessible on dead trees or his ResearchGate page. I’ve downloaded a presentation he gave this February and will read it now.

    What’s with all the *-eⁿ lately, why not *-ẽ?

    It’s not meant to be IPA. I’m just copying my interlocutor’s usage, who is following one of the two common Germanist traditions. Both of these traditions happily ignore; the other tradition (found e.g. in Ringe’s books) uses the ogonek for nasal vowels (ą).

    See also: IEists using the under-ring to mark syllabic consonants, while in the IPA that means “voiceless”. And never mind the macron, which means “middle tone” in the IPA and has no relation to length there.

    The macron may actually be the motivation for not marking nasality on top of vowel letters. Stacking diacritics used to be really hard.

  74. People don’t use the IPA symbol for nasalization, because there’s a tradition of using that to mark ‘circumflex’ vowels (though I use a carrot-top for those). Since we don’t know the phonetics of those, we can’t really just use IPA throughout. Ringe uses the alternate IPA notation of an ogonek, which is sensible — so *-ę, etc. — but I avoid that because it creates a little awkwardness with Norse, where ǫ = [ɔ], not [õ]. I’m generally in favour of IPA over traditional orthographies, but the phonetic indeterminacy in reconstruction does mean traditional notations have their place. (Also, some people hold that final *ⁿ was actually just *n — I don’t believe this, but the notation is more neutral in that regard.)

    I’ve seen Schalin’s thesis. His arguments are interesting, though as usual morphology and analogy complicate things severely. But even if he’s right about the non-merger of pre-PGmc *e and *i, the evidence still favours early raising of *e, so the change would perhaps not be of PGmc *e > *i, but of pre-PGmc *i > *ɪ/ɨ/whatever we want to call this, followed by pre-PGmc *e > PGmc *i.

    Regardless, a final *-eⁿ would be unique, and so I guess we could stipulate it goes however is most convenient. It would be at odds with all other short-vowel developments for this to end up giving the reflexes usually assigned to *-ōⁿ.

  75. Lars Mathiesen says

    Yeah, I just noticed the thing about supposed ring vs tick for syllabicity the other day (read that etymologically), but by the time my brain finally went “d̥ and m̥, one of these things is not like the other!” you guys had started using the tick and I didn’t know where to complain.

  76. David Marjanović says

    Having read the presentation and a few pages of the “Scandinavian front umlaut revisited and revised” paper (the less theoretical one), it turns out I’ve misrepresented two things: first, the North Germanic preservation of the *e-*i distinction was limited to the main foot – after it, the merger to umlaut-triggering *i proceeded as traditionally thought; second, the supposedly Pre-Germanic umlaut evidently didn’t happen in North Germanic, where *e in the main foot is preserved ( > umlaut-triggering *i) even if *i follows later.

    the evidence still favours early raising of *e, so the change would perhaps not be of PGmc *e > *i, but of pre-PGmc *i > *ɪ/ɨ/whatever we want to call this, followed by pre-PGmc *e > PGmc *i.

    Yes. Any *-eⁿ would probably have become something else or would have been analogically replaced.

Speak Your Mind