As I was passing by our pasta shelf, my eye fell on a box of macaroni that carried, below the legend ELBOWS, the Italian equivalent CHIFFERI. Wondering what this meant other than ‘elbow macaroni’ and wanting to make sure my tentative pronunciation /’kifferi/, with initial stress, was correct, I looked it up; I confirmed the pronunciation but could find nothing more about the word. Having dug around in Google Books, I’ve come up with the answer, and in the best traditions of Languagehat I am sharing it here so future generations will not have to hack their way through the uncharted jungle.

In the first place, unlike other pasta names (e.g., farfalle ‘butterflies’), it does not in fact mean anything other than ‘elbow macaroni.’ According to somebody writing in the Universidad de Chile’s Boletín de filología Vol. 32-34, p. 429, chiffero is the standard Italian equivalent of the Lombard dialect form chifel ‘croissant,’ and according to Giovan Battista Pellegrini, “Noterelle linguistiche bisiacche” in Günter Holtus, Z̆arko Muljac̆ic̆, and Johannes Kramer (eds.), Romania et Slavia Adriatica (Buske Verlag, 1987), p. 229, that is borrowed from the German (Austrian, according to my large German dictionary) Kipfel, also meaning ‘croissant,’ which in turn is from Latin cippus, which according to Robert Sedlaczek, Das österreichische Deutsch: wie wir uns von unserem grossen Nachbarn unterscheiden, p. 197, meant ‘stake, post.’ I’m pleased that Google has allowed me to assemble these obscure sources and present a coherent story, but once again I shake my head at the lack of scholarly attention paid to food and cooking terminology.


  1. Darn, I was hoping it had something to do with the chifferobe

  2. Here in Vienna it is usually “Kipferl” with an “r”. “-erl” is a common dimunitive suffix in Austrian German – “sackerl”, “schmankerl”, “jauserl”, etc.

  3. You probably didn’t need to expend quite so much effort. Garzanti gives the etymology “Prob. adattamento it. della voce lomb. chìfer ‘chifel'”, and at “chifel”, which it also defines as some sort of croissant, we find the derivation “Dal ted. Kipfel ‘cornetto’, dal lat. ci°ppu(m) ‘cippo, colonnina'”.

  4. “Food”? You mean that stuff you eat while reading the dictionary?

  5. I went from chifferobe to sideboard, which reminded me of credenza. Probably you all know this, but I’ll post it anyway:

    in Italian the name meant belief. In the 16th century the act of credenza was the tasting of food and drinks by a servant for a lord or other important person (such as the pope or a cardinal) in order to test for poison. The name passed then to the room where the act took place, then to the furniture.

  6. J. W. Brewer says

    Although a credenza is both linguistically and functionally quite similar to a credence (side table near the altar in certain sorts of churches, see I do not believe it was ever customary for the acolyte to sample the bread and wine before giving them to the priest for consecration to make sure the elements weren’t poisoned. This makes me at least mildly suspicious about the striking etymology for credenza AJPC found.

  7. J. W. Brewer says

    I know the word chifferobe, btw, solely from reading To Kill a Mockingbird as a boy. I don’t think I’ve ever heard anyone actually say it.

  8. J.W.B.: It’s one of the Southernisms that expats tend to repress, along with initial stress on umbrella and cement. Y’all, on the other hand, they tend to be proud of.

  9. Minor correction: CHIFFERO is the Italian form, and CHIFEL/CHIFER the Lombard dialect form, not vice-versa.

  10. It’s one of the Southernisms
    My grandmother used the term. She was born in Galicia (then Austro-Hungary, now western Ukraine), immigrated to Canada in 1905 and to her last days counted in Yiddish. I don’t think I’ve ever heard anyone else say chifforobe. (That’s the only spelling acknowledged by the AHD.)

  11. Woops, thanks! I’ll make the correction.

  12. The German wikipedia says the shape of the baked goods imitates the horns of a goat, which could explain a little better the name “stake, post”… If I have understood correctly, the Romans built a wall of sharpened wooden stakes around their outposts and that’s where the Germans took the name to the Yule buck horn pastries… I can imagine a basket filled with rugelach looks like such a wall 😀

  13. initial stress on cement
    How about final stress on concrete, do you know the origin of that? I used to work with someone from New England who did it, but when I asked he just acted bemused that there might be any other pronunciation.

  14. Trond Engen says

    initial stress on cement
    How about final stress on concrete
    Ah! Structural linguistics.

  15. Well, I’m from New England, too, and I’m the same way: conCRETE is what I’m used to.
    According to the paper OED:
    – “The stress has always been variable”
    – conCRETE is older than CONcrete
    – the latter now appears to be more common in the adjective and universal in the (construction material) noun
    I wonder if there a chance that in the eastern US people started saying conCRETE in order not to sound like the kind of people who say SEEment.

  16. final stress…Structural linguistics
    That reminds me. Since there are at least three structural engineers here, not to mention natural scientists and mathematicians, what’s the etymology of the algebraic symbols in structural engineering? For instance, if I remember right, a load is always “p” and a stress is “f”. Why these particular letters? Are they the initial letters of words from French or German?

  17. I’m a conCRETE sort of fellow too in the adjective, but not in the noun (except, like, the deadjectival noun, like talking of “abstracts and concretes”).
    And by “expats” I mostly mean my North Carolinian wife, though I had other people in mind too.

  18. Learning from OnlEtyDict that the word “chifforobe” dates from around 1917, and from WiPe that Sears advertised one in 1908, I began searching for more info. Among the first few hits for “chifforobe sears 1908” are documents beginning variously “The foundation from the chifforobe extends back to 1908.” “The cause with the chifforobe goes back to 1908” “The origin with the chifforobe goes back to 1908.” Can anyone explain this? They go on even more strangely, some of them. Are they all machine-translated form some other language?

  19. This version of the “chifforobe sears 1908” text seems less garbled, so may be the original text, though not necessarily the original source.

  20. You might like to have a look at this chart to see different chifferi sizes:
    Smaller ones are called chifferini or chifferetti, bigger ones chifferotti and chifferoni.

  21. David Marjanović says

    Um… vanya? Are you still in Vienna?

    Here in Vienna it is usually “Kipferl” with an “r”. “-erl” is a common dimunitive suffix in Austrian German – “[S]ackerl”, “[S]chmankerl”, “[J]auserl”, etc.

    Of course there’s no etymological /r/ in there. The r just signals that the whole thing is pronounced [ɐl].
    And yes, it’s not just Vienna.

  22. David Marjanović says

    Also Swiss Gipfeli. The /g/ probably means it’s a loan from the north or east (the Swiss have shifted /k/ to [qχ] or [χ] depending on the dialect… I hope this displays correctly).

  23. marie-lucie says

    I had never seen or heard such as word in English or French, but I guessed that it must be a blend of chiffonnier and garde-robe, as Wikipedia confirmed (thanks Ben Zimmer). The word must be an English (rather American) coinage, not a French one.
    This obviously hybrid piece of furniture has drawers on one side and a place for hanging clothes on the other. Un chiffonnier (from un chiffon ‘rag’, here referring to miscellaneous pieces of underclothing, ties, handkerchiefs, etc) is a tall, narrow piece of furniture with many drawers. A garde-robe (lit. “keeps-clothes”), usually called in European French une penderie, is mostly a large vertical space for hanging full-length clothes from a bar. It could be a piece of furniture or a walk-in closet. Une armoire could also conceivably have two separate spaces, or just shelves, but any shelves or drawers would be hidden behind the main door or doors, unlike with the chiffonnier.
    I have seen similar hybrids in Canada, usually in secondhand furniture stores or in old hotels. Often the side with the drawers is shorter than the wardrobe side, making the piece look unbalanced. They seem to have been popular in the period between the two world wars.

  24. The OED attributes chifforobe‘s second syllable directly to wardrobe, which is far more common in English than garderobe; the g/w distinction is Parisian/Norman. Ward itself is Germanic, and adopted directly as such into Old Norman, but surfaced as gard in the other Romance languages. Nowadays, garderobe is used only in historical or historical-fictional contexts, and mostly in the sense of ‘privy’.
    Wardrobe has a lot of derived senses in English, including not only ‘privy’ but ‘the excrement of the badger’ (!)

  25. JC, I guess I wrote a little too fast about the robe part, which is common to the French and English words. Like English wardrobe, French garde-robe now refers mostly to a person’s full set of clothes, as well as remaining a word for the closet or similar space where such clothes are stored.
    Ward itself is Germanic, and adopted directly as such into Old Norman, but surfaced as gard in the other Romance languages
    I guess you are summarizing a more complex evolution. I don’t know about Old Norman, but Old Germanic words starting with [w] were normally borrowed into very late Latin with [gw], written gu- in Romance languages, witness Spanish and Italian guardia ‘guard’, or guerra ‘war’. I once met a Catalan whose last name was Guasp, which I suspect was from a Germanic wasp- word (cf German Wespe). Those languages have preserved [gw] while French has changed it to [g] while preserving the gu spelling before i and e, as in guerre ‘war’, guêpe (from older guespe) ‘wasp’.
    English has some doublets going back to Germanic both directly and through French, eg ward(en)/guard(ian), warranty/guarantee, where the initial gua- reflects the older French pronunciation, not the modern one in either language. Most French words now starting with ga used to have gua, not only garde, gardien and garantie but for instance gagner ‘to win, gain’ from earlier guaagnier (Italian guadagnare).

  26. Trond Engen says

    I’ve thought of warranty as a loan from Norman, albeit on a Germanic base, since the ending sounds un-Germanic.

  27. Trond Engen says

    gagner ‘to win, gain’ from earlier guaagnier (Italian guadagnare)
    Nice. Is there a further etymology? *wad(ja)- + *aigVna- “prize-own”?

  28. marie-lucie says

    TG, there is probably something else going in in warranty/guarantee, eg Latin admixture towards the end of the word, but I don’t know.
    Regrettably, I don’t know either about the Germanic source for guadagnare and guaagnier.

  29. Marie-Lucie: John Cowan is quite correct, Norman French (+ some other Northern varieties) preserved Germanic /w/ in loanwords unchanged: only further South, in Central (Parisian) French was it adapted as /gw/, which subsequently shifted to /g/. Thus, the /w/ to /gw/ change cannot be dated back to Vulgar Latin.
    Most early Germanic loans in more Southern Romance languages (Provencal, Spanish, Portuguese, Italian…) came from Central (Parisian) French. Hence, their having /gw/ for Germanic /w/ may simply be due to their Parisian origin. Of course, the /w/ to /gw/ change may have taken place more than once. Spanish GUADALQUIVIR, from Arabic WAADI AL-KABIIR, is a good example of this.
    As for the English doublets you refer to, I am almost certain that WARRANTY/GUARANTEE is a Norman/Parisian doublet (of Germanic origin of course). Naturally, the /w/-/g/ correspondance will also be found between native (Germanic) English words and (Parisian) French words of Germanic origin.

  30. marie-lucie says

    Etienne, you undoubtedly know best, but I thought it was a given that the change /w/ to /gw/ had occurred in very late Latin, when original Latin /w/ had already become /v/ and therefore no longer existed in Latin (eg uenire, It venire, Sp and Fr venir, with /v/ in Old Spanish). Perhaps I need to read some new sources.
    I am also surprised that Germanic loans even in Portuguese would have come from “Parisian” French! The Wisigoths are supposed to have influenced Spanish rather than French, isn’t that right?
    The change /w/ to /gw/ initially is apparently still going on in some Spanish varieties, eg in (possibly nonstandard Argentinian?) gwevo instead of huevo (where hue = standard [we]).

  31. gagner < guaagnier … Is there a further etymology?
    < Old Low Franconian *waidanjan < waida ‘pasture’ (Ger. Weide, Span. guadañar ‘mow’, maybe Eng. weed). So, profit from land.

  32. Marie-Lucie: My guess is that Northern French preserved /w/ in Germanic loanwords not because they still had original Latin /w/ (as you say, all original /w/ > /v/ long ago), but simply because enough were bilingual in a Germanic language (Norse, English, Flemish, or what not) with /w/ that they could pronounce the phoneme correctly. There must have been at one time a sort of w-Sprachbund covering Welsh, English, and Normand which maintained the phoneme in all three languages long after their nearest relatives had lost it.
    Etienne: There is clear documentary evidence that warranty/guarantee, like wardrobe/garderobe, is indeed a Norman/Parisian borrowing pair. (The -ee ending of guarantee replaced the original -y within English: formerly a guarantee was one who made a guaranty.) Other such pairs are cattle/chattel, catch/chase, leal/loyal, real/royal, glamour/grammar (with a detour through Scots). Ward itself, however, is a native word as against Parisian guard.

  33. Marie-Lucie: probably I should have written “Central” rather than “Parisian” French above, to which should be added Old Provencal. The /w/ to /gw/ change is widespread, but the preservation of /w/ in some Northern French varieties means we cannot really call it a Late Latin treatment: “Late Latin” would normally refer to a pan-Romance change.
    Wisigothic did indeed have some influence upon Spanish and Portuguese, but this is dwarfed by the huge number of Frankish loanwords which entered both languages indirectly, as part of the heavy lexical contribution French and Provencal both made to Ibero-Romance (For example, ESPANA is a native word, but ESPANOL owes its suffix to Provencal). Thus, most words of Germanic origin in Spanish and Portuguese today are ultimately Frankish in origin (leaving aside, perhaps, the recent flood of anglicisms). The same is true of Italian, incidentally: despite the Ostrogoths and the Langobards having ruled parts of Italy, Germanic loanwords of Frankish origin (via French and Provencal) far outnumber words of Ostrogothic or Langobardic origin.
    For more details on the topic, I recommend William Elcock’s THE ROMANCE LANGUAGES: the section on the Germanic influence on Early Romance is excellent.
    John Cowan: Thanks for confirming my hunch above. Now, some nitpicking:
    Old Norse only appeared in Normandy at a much later date. Tellingly, its loans entered (Norman and Parisian) French with /v/ for Germanic /w/: cf. French VAGUE “wave”, from Old Norse VAGR (The English and Old Norse words are cognates, of course). I suspect the /w/ to /v/ change had already taken place in Old Norse before it was transplanted to Normandy.
    Nor can I agree with your claim that Welsh must have been a member of a “w-Sprachbund”. British Celtic *w had become *gw before the fifth/sixth century AD, as is proved by the change having affected Welsh, Breton and Cornish.
    (Because Latin loans also undergo this change (Welsh GWIN from VINUM, “Wine”) it is sometimes claimed British Latin was exceptionally conservative and kept Classical Latin /w/ long after it had become /v/ on the Continent. This is quite silly of course. First, there is no evidence that British Latin was in any way conservative, and second, a British Latin /v/ would likely have been borrowed in British Celtic as /w/, which would subsequently undergo the same change to /gw/ as all /w/ phonemes in the language, whatever their origin).
    Meaning that when, later, Frankish *w was borrowed as such in Northern dialects of Early Old French, *w had already disappeared in Welsh and its close relatives.

  34. ESPANOL owes its suffix to Provencal
    The things I learn around here!

  35. Trond Engen says

    Joseph B. Voyles has ON /w/ > /v/ at 1100 CE, as one of the very last changes on the way to the attested language, far too late to appear in Norman French. He doesn’t give an argument, though. Of course, sound changes aren’t instantaneous. I seem to remember some other author placing it in the late 900s on Skaldic evidence. I’ll try and find the book.

  36. David Marjanović says

    There must have been at one time a sort of w-Sprachbund covering Welsh, English, and Normand which maintained the phoneme in all three languages long after their nearest relatives had lost it.

    West Flemish retains it.
    (And while I am at it, Frisian kept /θ/, spelled th, into the 14th or 15th century.)

    The things I learn around here!

    I’ve read about this before. If it were native, it’d be *espano or *españuelo.

  37. David Marjanović says

    British Celtic *w had become *gw

    Shouldn’t that rather be worded as *w and *gw having been reanalyzed as allophones linked by consonant mutation? – Of course that would make it very easy for loans with /w/ to show up with gw when not preceded by whatever triggered the mutation, without the speakers even noticing that they had changed anything.

    enough were bilingual in a Germanic language (Norse, English, Flemish, or what not) with /w/ that they could pronounce the phoneme correctly

    When the value of “enough” is high enough, this only requires a pretty low value of “bilingual”. Around here, everyone aged 40 or less has Rock as a minimal pair: with [ʀ] or even [r] as the native word for “skirt” (historically also “coat”), with imported [ɹ~ɻ] as the music style. This includes lots of people (roughly those above 30) who can’t say [θ], and, well, almost nobody even tries to get [ð] right.

  38. David: your formulation may be true: I think even Celticists would disagree on the relative chronology here (did *w turn to *gw everywhere, with the resulting phoneme then lenited intervocalically? Or was this lenition already active, and *w only became *gw when it was unlenited?).
    Trond: the more I think about it the more I think that the /w/ to /v/ shift must have taken place in Old Norse before it was transplanted to Normandy. Norman French already had both /v/ (from Latin /w/) and /w/ (from Frankish loans) word-initially: why should native speakers of Norman French have borrowed (to repeat my example) VAGR with initial /v/ (VAGUE) unless the initial phoneme in the Old Norse word was /v/? At this time speakers of Norman French could have reproduced a /w/ or a /v/ with equal ease.
    (Of course, considering its geographical spread at the time, Old Norse may have had some registers/dialects with /v/ and others with /w/.)
    Hat: there is a Spanish Romance scholar who writes in Catalan (his name escapes me at the moment) who has argued that ESPANOL was probably originally an exonym with an ironic or diminutive/pejorative meaning among native Provencal speakers, which native Spanish speakers gradually adopted, taking away its demeaning connotations in the process (the history of the N-word among African Americans offers an interesting parallel). And yes, *espanuelo (with palatal n) is indeed what the native form should have been. Stressed mid-low /o/ turns into /ue/ in Spanish, but remains intact in Old Provencal, whereas Old Provencal eliminates final unstressed
    -o, which Spanish preserves.

  39. Penny says the expected Spanish form is españón. Welsh does have /w/ that’s not from mutated /gw/, though, as no other Celtic language does.

  40. Penny says the expected Spanish form is españón.
    How does that work?

  41. Ya got me, it’s just a throwaway line: the main point he’s making is that español is an occitanism.

  42. Hat, John Cowan: I think I get it. We’re dealing with two different suffixes here, one from (Late) Latin -ONEM, and the other from -OLUM.
    Penny’s *ESPANON is the “expected” Spanish form with the (attested, productively used) Spanish suffix -ON, from -ONEM. *ESPANUELO, on the other hand, is the “expected” Spanish form with what the Old Provencal suffix (-OL, from -OLUM) would have looked like if this same suffix had been a native Spanish one in productive use at the time.

  43. What are other comparable adjectives in –ón?

  44. I’m no Spanish specialist, but I believe -ON in Spanish today is productively used as an augmentative only, with former -ON diminutives having lexicalized: an example of this is TERRON “lump, clod of earth” (cf. TERRON DE AZUCAR “lump of sugar”), which *historically* is the -ON diminutive of TIERRA “earth, ground”. I don’t have any example of adjectives with diminutive -ON, unfortunately (some other reader might have examples and/or insights on the topic. Julia perhaps?)

  45. Yes, I’m familiar with it as an augmentative, but it seemed strange to me in this usage.

  46. Penny says that the productive use of -ón is a combined augmentative and pejorative, as in novelón ‘long boring novel’, caserón ‘large rambling house’, huevón ‘jerk, asshole (literally, someone whose testicles are so large as to be animalistic)’, and ricachón ‘filthy/stinking rich person’ (which includes the neutral augmentative -acho as well).
    Some frozen forms are not pejorative, as hombrón, which can be neutral, ‘large man’, or even ameliorative, ‘a man distinguished for talents, knowledge, and valor’. But I don’t think there are any frozen diminutives in -ón within Spanish itself, as opposed to (perhaps) Iberian Vulgar Latin.

  47. marie-lucie says

    All those Spanish words are nouns, not adjectives. Others I am thinking on offhand are ladrón ‘thief’ and cincuentón/a ‘fifty-year old man/woman.
    None of the words quoted are comparable to the allegedly expected form *españón which would be based on the name of a country. Are there Spanish words in -ón meaning ‘inhabitant of’? there are a few in French, like berrichon ‘from the province of Berry’, bourguignon ‘from the province of Bourgogne (Burgundy)’.

  48. « España » came from Latin « Hispania » ( land of rabbits, experts say) because of a palatalization of “n” in “ñ” before the diphthong “ia”. “Español”, such as you have said, came from the Provencal. For example, “peseta”, our old (in the future I don’t know) currency, came from the Catalan, as “paella” (the famous dish of rice), and “jamón” (ham) from the French; our own emblematic words are import! “Guerra” is Germanic but also “tregua” (truce) came from Gothic. On this matter, the sound “gw” is in “ga”, “go” and “gu”; however we need to add an “u” to write this sound to “gue” and “gui”; without this “u” these sounds would be as your “h”, like my name.
    As regards augmentatives, the suffix are “-on”, “-ote” and “-azo”. Diminutives are “-ito”, “-ico”, “-illo” and also “-ino” in some places, like Extremadura. Besides, obviously, we change the “-o” by “-a” at the end of words to do feminine genre. Actually not all words with these ends are augmentatives or diminutives, of course. Some words are nouns as “novelón” but others are adjectives as “ricachón” and “huevón”.
    “Terrón”, obviously from “tierra”, is used to mean a little of earth piled up, like when you use a hoe. The resemblance between the sugar piled up with the clod explains this word. Even you use, besides “clod”, “lump” to speak about the two things.

  49. Etienne, I haven’t got anything more to add… In fact I’ve never realized that “terrón” has an augmentative ending but a diminutive meaning. I can only think of one other word similar to “terrón” which is “tapón” (stopper, cork, cap, etc.) that comes from “tapa” and has a “diminutive meaning” (perdón, no sé si estoy diciendo burradas lingüísticas, ya avisé que soy muy ignorante en ese campo).

  50. >Etienne
    It occurs to me some words like the Julia’s example: “ratón” (mouse), that comes from “rata” (rat) but it is not exactly a diminutive. We’d say “ratita, ratilla” for that, and “ratoncito” or “ratoncillo” as diminutives of “ratón”. Also “tablón/tabla” (plank/board), “cajón/caja” (drawer/box).

  51. David Marjanović says

    ladrón ‘thief’

    Straight from Classical Latin latro “robber”.


    *lightbulb moment*
    Truth! 🙂 Closer in meaning, and just as closely related, is German Treue “fidelity/loyalty”. Gothic, like the North Germanic languages, had a sound shift that changed /wː/ into /ɣʷː/; the West Germanic languages merged /wː/ in various ways with the preceding vowel.

  52. corazon? tiburon

  53. You bet. In Gothic it was triggws, and it’s another of the bits of Gothic in Lest Darkness Fall, along with aiw ‘yes’, nu ‘well’, atta ‘father’, and the famous Ho, frijond, alai skattjans sind waidedjans.
    Note the English triplet truce, truth and troth, the latter mostly in betroth nowadays, which originally meant ‘swear fidelity.

  54. About -ON: it looks like Penny made a mistake: neither Julia nor Jesus have given any example of an adjective in -ON.
    It should be added that the “duality” of the ending (historically both augmentative and diminutive, with Modern Spanish now productively using the former meaning only: thank you, Julia, for confirming that native Spanish speakers do not perceive the link between TERRON and TIERRA) is not a Spanish peculiarity.
    Thus, in French the suffix is no longer productively used in either meaning, but pairs with and without -ON show the same double meaning of the suffix (thus, a CHAÎNON is smaller/more delicate/daintier than a plain CHAÎNE, but a BALLON is bigger than a plain BALLE)
    David, Marie-Lucie: Spanish TREGUA is identifiable as an unambiguous Gothic loan because of its form. French has TRÈVE as its word for “truce”: it comes from Frankish, which, being West Germanic, kept -ww- intact (hence German TREUE). Had Spanish borrowed the word from Old French we’d have expected a form *TREVA.
    Marie-Lucie: you had mentioned that a shift from initial /w/ to /gw/ seems to be “ongoing” in some Spanish varieties, quoting /gwevo/ as the realization of HUEVO. Well, an unpublished dissertation on the Spanish Philippine Creole (Chavacano) of Ternate which I’ve been reading these days indicates that this very change is found in this creole…but *only* in words of Spanish etymology: the author adds that it is likewise found in many American varieties and Ladino (Judeo-Spanish). Since all of these varieties were born in the sixteenth century or thereabouts, I think it is fair to assume that the change in question had taken/was taking place in (colonial) Spanish at that time.

  55. >Etienne
    “…neither Julia nor Jesus have given any example of an adjective in -ON.” I understood that your question was about an adjective diminutive in –on. There are some adjectives in –on , like the aforementioned “ricachón” and “huevón”.

  56. Etienne, please let me explain myself in Spanish:
    En cuanto a “terrón”, sí percibimos la relación entre “terrón” y “tierra” o al menos yo lo percibo. Lo que te decía que nunca había notado era que terrón podía ser un DIMINUTIVO de tierra, con la particularidad de terminar en -ON, dado que los adjetivos terminados en -ON tienen siempre un valor aumentativo.
    En cuanto a los adjetivos y sustantivos, desde el principio en tu pregunta estaba esa confusión que te señaló M-L y por eso no insistí yo en mi respuesta. “Terrón” es un sustantivo, de eso no hay duda, un sustantivo que designa un ente de la realidad más pequeño (o una porción pequeña) de otro ente mayor.
    Veo que el Diccionario de uso del español de María Moliner, designa a palabras como “cabezón” o “panzón” como “nombres calificativos”. Ustedes son los lingüistas y entenderán mejor estas cosas!
    En definitiva como hablante de español, la única palabra que puedo pensar semejante a la mencionada “terrón” es “tapón” (debe haber otras). Ésa también proviene de un ente de mayor tamaño, “tapa”, y designa algo más pequeño, por más que tenga esa terminación -ON, que en la lengua actual se utiliza para dar la idea de algo más grande.
    ¿Pude aclarar algo?

  57. David Marjanović says

    The (mag)pie thread is closed, so I’ll try to hijack this one. I have no idea about Lest Darkness Fall; I suppose I’ll have to remedy that…

    Not that dw is doing so well in Low Germanic either.

    Oh. I suppose that could explain its disappearance as a category in German.

    dwine ‘pine, waste away’ (of which dwindle was originally a sort of diminutive)

    The German verbs in -eln are fossilized frequentatives. A few have taken on a somewhat diminutive meaning due to reanalysis as the diminutive suffix -(e)l((e)(i)((n))).

    In a rather smaller Dutch dictionary, I find […] dwang ‘coercion’ (cf. English dwang above)

    German Zwang as expected.

    dwarrelen ‘swirl’

    I suppose German zwirbeln could be related, but “twirl” is a much better match in meaning and fits just as well in sound.

    dwars ‘transverse’

    Extinct in German except for Zwerchfell “diaphragm between chest and belly”.

    dweil ‘cloth’

    Hm. There is a cloth called Zwil(li)ch, but the Duden derives it from “2” because it’s doubly sewn or something. (Would make sense; “twin” is Zwilling.)

    dwingen ‘compel’

    Again zwingen.

  58. The (mag)pie thread is closed, so I’ll try to hijack this one.
    Yeah, sorry about that, but when the spam is coming so thick and fast that immediately after deleting a batch of comments I find another one already there, I don’t have much choice. Needless to say, I am delighted to have the discussion continue here.

  59. I don’t have much choice.
    Dammit, I missed a perfect opportunity to say I was gezwungen.

  60. Is Zwilch the same as English twill?

  61. marie-lucie says

    dwars ‘transverse’
    This must be related to thwart, as in a crosspiece in a canoe.

  62. Spanish TREGUA is identifiable as an unambiguous Gothic loan because of its form.
    Or East Germanic, at least. The change /ww/ > /ggw/, called “sharpening”, is one of the shared states between East and North Germanic. Unfortunately, there are also shared states between West and East Germanic. It doesn’t help that almost all the Gothic we have is in an archaic state compared to North and West Germanic.
    David: I was going to write “frequentative” for dwindle, but the OED says “diminutive”. However, Etymonline says both.
    Empty, Marie-Lucie: Right on both counts: dwile = twill (originally Scots), dwars = thwart (not only as in a canoe, but also the verb meaning ‘hinder, prevent’).

  63. David Marjanović says

    Is Zwilch the same as English twill?

    …Probably. I’ve actually never seen either.

    Or East Germanic, at least.

    If only we had any East Germanic other than Gothic and a few personal names. *wince* *wail*
    Oh, sure, we have a few legal terms from Lombardic. But, at least according to the Pffft, people can’t even agree on whether Lombardic is an East Germanic language or an Old High German dialect!

    It doesn’t help that almost all the Gothic we have is in an archaic state compared to North and West Germanic.

    In other words, we’re lacking comparably old samples of North and West Germanic, so we can’t tell if loss of the features which Gothic alone retains from PIE was a series of innovations of a North + West Germanic branch or a set of areal features that spread by borrowing after the Goths moved off to Romania (…in more than one sense).
    All we can do is count the contradictory innovations, stick with the most parsimonious hypothesis, and ascribe the evidence that contradicts it to borrowing or whatever. Has there ever been a comprehensive count? (It definitely wouldn’t be trivial to catalog all those features.)
    The thread on the smell of old Soviet books is now closed, too, so:

    There may also be a feminine form with root stress *láho:- with meanings circling around “shallow water”: ON f. “water close to the beach, tidal plain”, No. dial. “reddish water in a bog”, MHG la: “pond, bog, swampy grassland”. B&L see a possible etymology for the name of the shorebird lo “Charadriidae”

    Oh yeah. Thanks! The MHG one, which I’ve read had /o:/ instead of /a:/*, lives on in a few placenames, like the Laaerberg, the next hill over from the one I’m sitting on here in Vienna. Reportedly, dialectal forms of the Laa- part used to have /g/ and /x/ in free variation.
    * Doesn’t make sense if you look at the modern form, but da was /do:/ in MHG, too.

  64. David Marjanović says

    A spambot, no doubt, alerted me to the long-dead Starodub thread:

    It’s to avoid Ambiguities like “cold cicada” that the official Common Names of Birds and other Wildlife are nowadays generally Capitalized, as in Red-winged Blackbird

    No, as in Red-winged blackbird. This looks silly enough that pretty much only the ornithologists have adopted this convention. See here for examples by somebody who isn’t an ornithologist but reads and writes a lot about birds in addition to other limbed vertebrates.

  65. From that last link I found The noble tradition of military goats, which I hereby offer to AJP to do with as he wishes. (Sorry about the smell of old Soviet books thread; as you will have guessed, the spammers swamped that one as well—bezek unto their khothar!)

  66. I read this lovely (if rather lengthy and perhaps only peripherally relevant) passage in Rickard’s History of the French Language yesterday:

    In 1565 [humanist Henri] Estienne […] expressed his conviction that French is superior to Italian, on the grounds that it is ‘demonstrably’ nearer to Greek than any other language, and that Greek is ‘as all men acknowledge’ manifestly superior to all other languages. His attack on the Italianising affectations of courtiers was comparatively mild in this first work. In 1587, however, he inveighed against them with considerable truculence in his Deux dialogues du nouveau langage françois italianizé, et autrement desguizé, principalement entre les courtisans de ce temps.

    ‘Celtophile’, in whom we may see Estienne himself, takes to task his courtier acquaintance ‘Philausone’ (the name means ‘lover of Italy’) for his Italianised jargon, for saying things like j’ay l’usance de spaceger par la strade aprés le past (‘I’m in the habit of talking a walk in the street after dinner’); quelque volte ‘sometimes’; in case ‘at home’; Dieu soit ringratié ‘thank God’; m’incresce fort ‘I’m very sorry’; sa maison est fort discoste ‘his house is a long way off’; ragasch ‘boy’, ‘page’; cattif ‘bad’, and for using se fermer in the sense of ‘to stop’, ‘to come to a halt’.

    Celtophile clearly regards the pronunciation [ɛ] for [wɛ] as an Italianism too, for he attributes to Philausone forms which he spells dret, endret, voudret, alet, francés, while he is careful to attribute to himself the forms which he writes droit, endroit, voudroit, aloit, françois [ɛ] for [wɛ] was certainly a common substitution, but it was not necessarily an Italianism for all that.

    Estienne followed up this onslaught in the following year with his De la precellence du langage françois, in which he demonstrated to his own satisfaction that, judged by such subjective criteria as gravité, gentilesse, bonne grâce, brièveté and richesse, Italian is clearly inferior to French. It is of course not possible to prove the inferiority or superiority of one language to another in this way, or indeed in any other.

    Moreover, Estienne was at fault in making no distinction between affectations of the kind he attributes to Philausone, and quite un-affected loan-words of real practical value — technical terms associated with a host of new concepts, above all relating to military, nautical, financial, architectural and artistic matters. In fact it has been calculated that some 460 Italian words were borrowed in the course of the 16th century, the peak period being 1540-60.

    Of these, a very high proportion have proved their usefulness and are still current, having long since lost all association with Italy or with Italian: they are to all intents and purposes an integral part of the French vocabulary today. Even such common verbs as attaquer, briller, manquer and réussir, which one takes completely for granted today, were borrowed from Italian in the sixteenth century. Architectural terms borrowed from Italian during the same period are antichambre, appartement, arcade, architrave, balcon, balustrade, corniche, façade, frise, médaillon, piédestal, pilastre, site, volute, together with the words architecte and architecture themselves.

    Financial: bilan, escompte, faillite (banqueroute had been borrowed in the fifteenth century). Military: bastion, bataillon, campagne, caporal, cartouche, casemate, cavalerie, cavalier, escorte, fantassin, infanterie, parapet, sentinelle, vedette. Nautical terms include: accoster, bourrasque, boussole, escale, fanal, frégate, gondole, mousse, remorquer. Fine arts: arabesque, artisan, cadre, damasquin, estampe, figurine, galbe, relief. Textiles and clothing: brocart, brocatelle, capuchon, peluche, soutane. Music: cantilène, concert, contrebasse, fugue, madrigal, sérénade, sourdine, théorbe, trio, trombone, villanelle. Food: artichaut (transmitted from Arabic), cervelas, marron, saucisson, semoule, vermicelle. Literary: sonnet, tercet.

    Useful adjectives borrowed at this time include altier, bizarre, brave, brusque, burlesque, fruste, grotesque, jovial, and leste. French also acquired a new suffix from Italian — -esque, at first as an integral part of Italian loan-words, but soon available for use in new French words uninfluenced by Italian in other respects; while the superlative or hyperbolical ending -issime, already present in French as a Latinism, underwent a further slight extension under Italian influence.

    One minor later development is curious enough to be worth mentioning here. The Italian suffix -one was augmentative in force, but on becoming -on in French it came under the attraction of the French diminutive suffix -on.. This explains why vallon, which originally, when borrowed, meant a wide valley, soon came to mean a small valley, a dell; it also explains why carafon, borrowed somewhat later, meant for a time ‘large carafe’ and ‘small carafe’ (it now means only the latter), and why, to this day, medaillon may mean either a large or a small medallion!

    I debated whether to put this comment here or in the more recent malapropisms/mispronunciations thread, but I decided it belongs here: I’ll make a cross-reference (O frabjous day!)

  67. principalemtn

    Arrgh! Pkease fix, O Hat.

  68. Done, and while I was at it I fixed celarly, cocnepts, and aove.

  69. Thanks. Proofreading is so difficult, especially of one’s own work, especially without a preview mode (though I did copy and paste the text into another window to make sure the HTML was right). I put most of my energy into checking the French words that were the evidence rather than the French titles or the English matrix text.

    Pkease fix refers to a common failure mode of the Lear-Siegler ADM-2, an 1970s-era computer terminal, that showed up by causing both the L and K keys to generate k, leading to emails like “My terminak is broken. Pkease fix.” Terminak became a slang term for this type of terminal as a result.

  70. Thanks, I love that sort of thing!

  71. (The explanation, I mean, not the problem.)

  72. David Marjanović says

    [ɛ] for [wɛ] was certainly a common substitution

    Is that where the modern forms français, voudrait etc. come from?

  73. marie-lucie says

    David, yes.

    I have not found an explanation of why some forms went to [ɛ] (now written ai) for [wɛ] while others went to [wa] (still written oi). The verbal forms all have suffixes in -ai- (for the imperfect and conditional). With demonyms and ethnonyms, both variants of the suffix are found (though never with the same stem), something which suggests local variation, but apparently not consistently. Examples are anglais ‘English’, portugais ‘Portuguese’, dijonnais ‘from Dijon’, but albigeois ‘Albigensian’, chinois ‘Chinese’, siamois ‘Siamese’. I only know one stem which is found with both suffixes, but it occurs in forms which have become completely differentiated in meaning and usage: français ‘French’ and François ‘Francis’.

  74. David Marjanović says

    Oh, anglais reminds me: the law that ordered the English colonists in Dublin not to speak Irish begins:

    ITEM ORDINE EST que chescun engleys use la langue engleyse

    When does it date from again? I don’t know how to search for this quickly.

  75. It seems clear that the [wɛ] > [ɛ] change was in progress at the time of standardization in the 17C. “Advanced” speakers had only [ɛ], whereas more conservative speakers had either just [wɛ] or a mixture of [wɛ] and [ɛ], but both groups wrote oi (or oyas in soyez, soyons) in any case. Standardization froze an irregular result into place: Rickard says that [ɛ] < [wɛ] survives mainly in the nationality names and imperfect and conditional verb endings that you you mention, and in craie, monnaie and a few like them (historically croie, monnoie, etc.) The change [wɛ] > [wa] came later, and the change in spelling oi > ai later still.

    Similar effects account for the restoration of /r/ in -ir, -oir, -eur but not -er, -ier (in earlier French all these /r/s had been lost); for profit rather than proufit and per contra troupe rather than troppe (all four of which had been in use earlier); for the restoration from the spelling of the final consonants of fils, net, sens; for the preservation of /ar/ < /er/ in English only in clerk, sergeant, derby, varmint and the words where the spelling now supports it (farm, star etc.); for the near-universal restoration of English aspirate initial /h/ where it had universally been lost; and even for the irregular am, are, is in Standard English, where many dialects have leveled it. Historical linguists are trained to think and say “Dialect mixture” in these cases even when there is no actual evidence for it, but that’s because we tend to underestimate the effects of standardization even on the spoken language.

  76. 1435. Your spelling was a bit off.

  77. David Marjanović says

    Oh, thanks! 1367, if you read on. 🙂

    and in craie, monnaie and a few like them

    Also connaître and thus connaisseur…?

    the near-universal restoration of English aspirate initial /h/ where it had universally been lost

    It hadn’t been lost in Scotland and northern England. Allegedly, Sir Walter Scott made fun of “the eclipsed manner of the Queen’s English”.

    Conversely, the King James Bible consistently uses an, mine etc. before h.

  78. True, O King. I should have said “Standard English and related southern dialects”. And yes, connaître < OF con(n)oistre < L cognōscere.

    The Great Vowel Shift of French is as striking as that of English: of the seven stressed vowels of Proto-Western-Romance, namely /aɛeiɔou/, four (possibly five) diphthongized (OF /e/ < PWR /a/ was apparently originally distinct in quality from unstressed /e/, as it does not rhyme or assonate with it), /u/ became fronted, and only /i/ was left intact. Fortunately for all, the French GVS happened before the orthography was established, and so the spelling reflects it, if not most of the changes that have happened since.

  79. Now I’m trying to imagine reading French if its orthography had been established before the shift. Yikes!

  80. No worse, surely, than Icelandic. See the Strasbourg Oaths:

    Pro Deo amur et pro christian poblo et nostro commun saluament, d’ist di in auant, in quant Deus sauir et podir me dunat, si saluarai eo cist meon fradre Karlo, et in adiudha et in cadhuna cosa si cum om per dreit son fradra saluar dist, in o quid il mi altresi fazet. Et ab Ludher nul plaid nunquam prindrai qui meon uol cist meon fradre Karle in damno sit.

    Si Lodhuuigs sagrament quæ son fradre Karlo iurat, conseruat, et Carlus meos sendra, de suo part, non lostanit, si io returnar non l’int pois, ne io, ne neuls cui eo returnar int pois, in nulla aiudha contra Lodhuuig nun li iu er.

    This is a compromise between Latin orthography and pre-Old French pronunciation, so it’s not quite what an unshifted orthography would be like but it is suggestive.

  81. If that was what French looked like today, you’d have to know to read poblo as /pœpl/ and nostro as /notr/ and Deus as /djø/ and fradre as /frer/, etc., and you think that’s no worse than Icelandic?

  82. Hmmmm…..I have never heard that Galicia (you know, in the northwest of Spain) was previously Austro-Hungary, now Ukraine. What’s up with that? That seems really incorrect. I have not researched it, though. I just know
    that Galicia is in Spain and always has been.

    And to think that I ended up here when Googling “chifferini,” which I am cooking tonight. When I was reading from this very interesting website, my husband said, “I have a “Language Hat” app available on my phone. What???
    I love words in many languages. Now I’m hooked. I’ll be back!

  83. There are lots of places whose names go back to (or appear to go back to) the Gauls: Galicia in Spain, Galicia in Eastern Europe, Gaul itself, and Galatia in what is now Turkey (see the Epistle to the Galatians in the New Testament). There may be more.

  84. January First-of-May says

    There are two different regions named Galicia. One of them is, and had been for the last few centuries, in the northwest of Spain. The other is now in the west of Ukraine, previously USSR, previously (between WWI and WWII) Poland, previously Austria-Hungary, and before that Poland-Lithuania (and I’m not confident enough in the geography and history involved to go further); historically the name comes from the Galicia-Volhynia principality.
    It is apparently uncertain whether the names of the two regions are related to each other (but, as far as I understand it, most of the current theories say they aren’t).

    Incidentally, I immediately recognized “cippus” as meaning a kind of column, because I’ve read a lot about Roman coins featuring an “inscribed cippus”… this refers to the “1000th anniversary of Rome” issue, mostly (apparently the emperor ruling at the time made a bunch of special coin types commemorating that event, which mostly featured said inscribed cippus).

  85. The Slavic Galicia takes its name from the Ukrainian town of Halych; there seems to be some disagreement as to whether the town’s name comes from the Celts or from some Slavic word. Most of the region is in Ukraine now, but part of it is in Poland.

    The Spanish Galicia is definitely Celtic-related; it’s also rendered differently in Latin, as Gallaecia.

  86. George Gibbard says

    Galicia in Spain was formerly Latin Callaecia, so something has happened to the initial stop. Portugal was originally a county within the kingdom of Galicia. Its name derives from Portus Cale, the Latin name of Porto. Wikipedia says some people claim the names Cale and Callaecia are related. If it was generally thought so in the past, the intervocalic voicing in the name “Portugal” could have influenced Callaecia > Galicia. It appears unlikely that these names are related to Gallia etc.

  87. Another Gothic word in Lest Darkness Fall is timrja, glossed ‘comrade’; however, Wikt defines it as ‘carpenter’ (cf. Zimmermann). I assume that de Camp’s eye skipped from one line in an English-Gothic glossary to the previous line. However, I can find no such Gothic word: OE gaedeling means ‘comrade’, but its Gothic cognate is used to translate ‘cousin’ in the NT.

  88. David Marjanović says

    Me on December 5, 2011:

    But, at least according to the Pffft, people can’t even agree on whether Lombardic is an East Germanic language or an Old High German dialect!

    A year ago I came across a paper that argued pretty convincingly, based mostly on a few words that have not so far shown up in West or East Germanic, that Lombardic was North Germanic. That would actually explain quite a few things, including not only the origin story, but also why, according to another paper, the cognate of bide is spelled PIDH (in scratches that seem to be Latin letters) on a sword – with the High German consonant shift sensu lato, but without the West Germanic [ð] > [d] (> High German [t]) shift.

    I’ll try to dig up both papers tomorrow.

  89. John Cowan says

    Woh. A variety of North Germanic that underwent the High German shift? *Akisomsalsa!

  90. David Marjanović says

    Here’s a brief paper on Lombardic as North Germanic (followed by the same reformatted as what appear to be very, very badly made slides for a conference presentation). Too bad examples for the absence of West Germanic consonant lengthening are not given; but that’s not the only argument.

    The paper on the sword with the Lombardic inscription is here.

    The good news is that the inscription is not just scratched in; I had confused this sword with a bunch of others. The inscription is professionally engraved and inlaid with silver; there is no doubt that it reads:

    + I H I N I N I h V I L P I D H I N I h V I L P N +

    The bad news is that the paper is in German. 🙂 The author segments the inscription as


    and interprets that as ih inni ni hwīl pidh, inni hwīl p[i]n. The lack of the 1sg ending *-u on ih […] pidh is explained by the following vowel-initial word; things like quidu ih > quidih “say I” are attested in OHG. The lack of I in PN is interpreted as a real syncope, which would be massively anachronistic, but I don’t think it needs any linguistic explanation – maybe the engraver just missed it among all the vertical lines. The lack of *-a on hVIL, which would be expected at the very least in the accusative governed by *bīðan, is referred to the fact that a few originally endingless nominatives, including (h)wīl, are undeclined in fixed phrases in OHG, even though they otherwise appear with -a in the accusative and even the nominative. inni, later inne, is an OHG adverb meaning “inside”, ni ~ ne the OHG negation.

    The whole thing would mean something like “I inside [ = the sword in its sheath] bide not the time, I inside am the time” or perhaps “I myself am the time on the inside”, implying “I get to decide when you die”.

    If interpreted as OHG, the forms inni and ni with -i instead of -e should be older than about 900. The same holds for the preserved h before /w/, which, however, was already quite unstable in the earliest Upper German sources around 750. The HG consonant shift (ih) should be younger than 650 or something, the two p (for *b) should be younger than about 700 according to the cited reference. The final n instead of m should be younger than about 800. All in all the author says 8th century.

    The one thing that doesn’t fit any of this, neither the timeline nor the identification as recognizably OHG at all, is the DH; it is “beyond doubt” that it cannot represent a *þ (OHG th ~ dh ~ đ ~ d). The author postulates that the West Germanic change from [ð] to [d] (which is presupposed by the further shift in OHG to [t]) was an areal phenomenon that happened to reach Lombardic last. This is where the other paper comes in.

    Within known Lombardic, initial *þ stays th “into the 8th century” and then turns into t; in all other attested positions it becomes d around the same time. That makes it look even more Scandinavian (and West/North Frisian, I admit).

    The ugly news is that the sword was found in 1921 by a treasure-digger in the ruins of the fortress of… Pernik in Bulgaria. He promptly sold it to the archeological museum in Sofia, but the exact circumstances of the find are nonetheless unknown. For how the sword got from Italy to Bulgaria before the end of the 12th century, when the fortress was destroyed, the author suggests either that the crusades were involved, or alternatively points out the fact that the Franks got into a war with the Bulgars in 818–832 not long after taking over Italy in 774; Lothar, son of Louis the Pious, governed in Italy since 822 and then participated in the war with the Bulgars in 830.



  91. John Cowan says

    ON agi ‘awe’, duly grimmed, and of course the Vulgar Latin loanword. In short, ‘Awesomesauce!’

  92. Trond Engen says

    + I H I N I N I h V I L P I D H I N I h V I L P N +

    ih inni ni hwīl pidh, inni hwīl p[i]n

    “I inside bide not the time, I inside am the time”


    First, I’m impressed that it’s possible to get that much out of it. With all those I’s I would have dismissed tham as ornamental lines. Now I think the text must have been composed with that visual effect in mind. Maybe that also explains the endingless forms.

    Second, the fall of the endings may also be regular in this language. The lack of the accusative ending on hwil seems well explained from OHG. I’ll add that “fixed phrase” here goes a long way towards grammaticalization as a temporal adverb. For pidh we need to invoke ON: inf. bita, 1sg pres. bít with no ending.

    Third, I’m not convinced about the details. It seems to me that there’s a parallellism between opposing statements in the first and the last clause, a common element in Skaldic poetry. My Lombardic is a bit rusty, but how about something like

    ih inni nihwīl pidhi, ni hwīl p[i]n

    “I inside nowhile bide, nor time [something]

    “I (to go) inside (the flesh) never wait, nor (there) time stay”?

    Here I have made myself a new problem by lashing an i onto pidh. It could be another error by the engraver, maybe influenced by the all-but-1sg subjunctive.

  93. David Marjanović says

    First, I’m impressed that it’s possible to get that much out of it. With all those I’s I would have dismissed tham as ornamental lines.

    Compare the earlier interpretation cited in the paper and dismissed: IHesus. IN Ihesu Nomine. IHesus. VIrgo. Laus Patris Ihesu Domino cHristo. IN IHesu VIrgo. Laus Patris Nostri.

    There are sword inscriptions that are abbreviated beyond recognition almost like that and have similar contents. But still.

    For pidh we need to invoke ON: inf. bita, 1sg pres. bít with no ending.

    Interesting. That’s got to be another verb, though.

  94. Trond Engen says

    Ouch. Yes, I meant bíða “(a)wait” and 1sg bíð, but managed to mangle it. Bíta is “bite”.

    I also meant to say something about Lombardic as North Germanic.

    The evidence is very slim, but that goes for any categorization. The commonalities with North Germanic are shared retentions, while those with High German are shared innovations. OTOH, diagnostics based on shared innovations is of limited value within a continuum, and especially a dynamic one like Migration Era Germanic (see also Iranian, all eras). One suggestion could be that the Lombardic core was North Germanic, but as the nation acquired new members on its way south, the NG element was reduced to a substrate. Another could be that Lombardic represents an otherwise lost middle between North and (Continental) West Germanic, pushed east and south by expanding Limes tribes.

  95. Lars (the original one) says

    What is this pin? Cognate to German bin (or at least to PIE *bʰu-)? The article seems to gloss something as ‘am’ and I don’t see what else it could be.

  96. Trond Engen says

    That’s my understanding too. I glossed it as “[something]”, since I wanted to keep it open whether it’s “be” or “dwell”. Glossed over, as it were. My shot at a translation probably works better with a form of ON búa, but bin is a more straightforward reading.

  97. “Bide” versus “bite” works in some modern Germanic languages. It could have been a pun in Lombard as well.

  98. Lars (the original one) says

    The bit about ‘final n instead of m‘ must be about the PN word, can *pim be a form of búa too? (Well, circular argument really, I guess the authors think the word should have had -m before 800 because they think it’s a form of ‘be’).

  99. David Marjanović says

    The commonalities with North Germanic are shared retentions

    Other than the two words, right?

    What is this pin? Cognate to German bin […]?

    That’s the idea.

    can *pim be a form of búa too?

    You tell me. In German at least, only “be” and “have” ever have 1sg forms with (*)-m instead of (*)-u that I know of.

    Funnily enough, if there’s a cognate of búa there and it means “dwell”, that’s an argument against the language being German, where bauen means “build, construct” instead! Only the noun Bau “burrow, den” retains some shade of the “dwell” meaning, and even that can mean “building” or “construction” (both act and result). “To live in a place” is wohnen, already wonên in OHG.

  100. Trond Engen says

    The whole thing is odd. German bin is the 1sg present of búa merged with the cognate of Eng, am. If Lombardic is supposed tp be NGmc, there should only be búa in uncontaminated form, which mean no n in sight. The ON 1sg present was (with what I assume is i-umlaut from the lost present suffix *i), which might perhaps do the job, but I don’t think we can evoke that much NGmc development in Lombardic if it moved south around the 1st century CE.

  101. Lars (the original one) says

    Only the noun Bau — how about Bauer? Danish bonde is from ON búandi, but we also still have the verb itself whose present participle is boende. (In udeboende børn, for instance). (The senses ‘dwell’, ‘prepare’ and ‘build’ are given for the ON word, but only ‘dwell’ survives in Danish. ‘Build’ is covered by the extended bygge < *būwijaną).

    Oh, and PGer *būaną is actually a zero grade of PIE *bʰuH-, or from the zero grade at least, I missed that. According to Wiktionary, ‘to be’ was e-grade *beuną (with lots of suppletion). So if *beuną and *būaną was a doublet with senses split between ‘be’ and ‘dwell’, wouldn’t it be more parsimonious to say that G bin is based on *beuną (even though the 1s present may have been the same)?

  102. Trond Engen says

    Me: NGmc development in Lombardic if it moved south around the 1st century CE

    I meant the first century BCE. At that time Germanic as we know it must have been fairly homogenous. Östen Dahl has argued that Proto-Norse was brought to Scandinavia from the south along with Runic script from 200 CE or so. If so, the first attested homeland of the Lombards at the lower Elbe could actually be in the cradle of North Germanic.

    Lars: So if *beuną and *būaną was a doublet with senses split between ‘be’ and ‘dwell’, wouldn’t it be more parsimonious to say that G bin is based on *beuną (even though the 1s present may have been the same)?

    Yes, possibly, but I don’t think we know that there was such a doublet anywhere outside the suppletive paradigm of ‘to be’. For the independent verb meaning “dwell, build” there were different developments in the different branches. Different enough to make me wonder if it may have been two verbs originally.

    In WGmc it became weak. OE būan – būe – būde – gebūn. HG bauen – baue – baute – gebaut.

    In North Germanic it stayed (or became) a messy class 7 strong verb. Wiktionary has a nice but sanitized overview. *bʰew-H- developed into several stems, which were moved around in the paradigm. One was the zero grade (or “long u-grade”) *bʰū- that forms the infinitive stem bú- and with i-imlaut the present stem , another *bʰeww- that gave the stem bjugg- and with i-umlaut bygg-. The two latter are mostly found in past and plural forms, but were occasionally used elsewhere in the paradigm.

    Gothic is confusing in its tidyness. Wiktionary has both a weak and a storng paradigm, and for some reason the root diphtong is au throughout both.

  103. Trond Engen says

    Instead of two verbs, maybe it was less than one. It’s suppletive with the *h₁es- verb in Celtic and Balto-Slavic. It’s only the presumed 0-grade that can be fairly safely projected back to Core IE. I think we may see a, eh, perfectum tantum *bʰū-H- “have existence, form, place”. This was built out with different analogical paradigms in the daughter branches.

  104. Lars (the original one) says

    Well, according to Cowgill, Ringe concurring, at the “West IE” stage *bʰúh₂-t was a root aorist ‘became’ and didn’t form present or perfect stems. So analogy would have free play.

    By Wikt, Latin had present fio ‘become / am made’ whose perfect (< aorist) was originally fui, but the latter was moved to the paradigm of esse (and fio got an analytic passive perfect from facio so it’s called semi-deponent, but in fact its active forms have passive meaning instead of vice versa). Cf fiat lux (3rd sg pres subj act).

  105. Trond Engen says

    Me: *bʰū-H-


    ““have existence, form, place””

    In a complete and timeless and unchanging way as opposed to *h₁es- “be (temporarily)” and *h₂wes- “stay (temporarily)”. It will unite all the meanings of the Sanskrit root noun भू (bhū́) except 1. “the act of becoming, arising”. This is the first semantic extension:

    “come into existence, form, place” (> “grow (intr.)”)

    This is all over Greek φῠ́ω. The active forms can also be used transitively meaning “produce, bring forth, beget, etc.” This is the second semantc extension:

    “bring into existence, form, place”

    Indo-Iranian and Balto-Slavic instead made an analogical o-grade to derive a regular causative.

    The suppletive paradigms of Celtic and Balto-Slavic are similar in that *h₁es- is used in the present forms and *bʰu-H- in the past and perfect. The two aspectual verbs found themselves merged when aspect had been reinterpreted as tense. Germanic instead uses *h₂wes- in the preterite. Reinterpretation of the perfective as a preterite was impossible in Germanic since those forms already had the specified meaning “dwell”. West Germanic invented an e-grade for the participles, North and East Germanic instead suppleted *h₂wes- all over. Gothic seems to have generalized the o-grade of *bʰu-H- and narrowed the scope to “dwell” only. North Germanic generally used the “long u-grade” — but messy in the details and with some residual o-grade forms in the dialects and in ancient forms like “farm, dwelling”. There’s also a funny duality in meanings with bu “prepare” seemingly reflecting “make, bring forth”.

    There are more branches, but Germanic was messy enough, so I think I’ll stop there.

    … except:

    Lars: By Wikt, Latin had present fio ‘become / am made’ whose perfect (< aorist) was originally fui, but the latter was moved to the paradigm of esse

    Yes, similar to the rest of Northwestern IE (or whatever). No fu- in the imperfect, though.

    Latin has a diverse set of reflexes. I’ll just note that the suffix -bus seems to point directly back at the original perfective.

  106. Lars (the original one) says

    Wikt tells me that perfective (< aorist) fuī < *fefuai and imperfective fīō < *fuiō — those are “Proto-Italic” but I don’t know whose since no sources are given; the point is that fīō does come from *bʰúh₂-. And future -bō too, it says.

    (I assume that by imperfect you meant the imperfective ‘present’ stem — I don’t know what fīēbam would be reconstructed as, but I’m pretty sure it would have had *fu- like the present).

  107. Trond Engen says

    I assume that by imperfect you meant the imperfective ‘present’ stem

    Yes I misread the table of inflections. I had “The suppletive paradigms of Italo-Celtic and Balto-Slavic” until your comment made me look closer at the Latin Wiktionary entry, thought I caught myself in an error of oversight and went on to make one.

    Anyway, *h₁es- and *bʰu-H- were incomplete and complimentary paradigms in PIE. With so much parallel-yet-different innovation in the daughter branches, it should be feasible to make a tree of shared innovations and learn something about the relationship between the branches post-PIE. Does PLat. *fefuai go back on the same proto-form as ON bjó, or are they both innovations?

  108. Trond Engen says

    It’s also interesting how Latin used *bʰu-H- to form a new verbal inflectional paradigm. How much of the paradigm of *bʰu-H- can be reconstructed from this alone?

  109. Lars (the original one) says

    The reduplication in *fefuai is unexpected (to me) if Italic perfects come from aorists and *bʰu-H- formed a root aorist in “West IE”. On the other hand PIE perfects were always reduplicated, I don’t know how to square that up.

    As for bjó, can it be the result of a reduplicated perfect? I checked Ringe and he only says that *būaną is from an ‘innovative present’ *bʰuh₂-ye/o- (Þórhallsdóttir 1993, 152-6); I have no idea if a reduplicated perfect could be formed by analogy so it included the *-ye/o- element > -j-.

    Þórhallsdóttir 1993 is Guðrún Þórhallsdóttir, The development of intervocalic *j in Proto-Germanic, that looks like it could be relevant to this exact question.

  110. David Marjanović says

    It is. I’m currently rereading this paper, which digs up the long forgotten change *-uj- > *-ij- in West IE and Balto-Slavic and finds that it explains a whole lot of things. I’ll report back when I’m done. Unfortunately, the paper is scanned in sideways – and that was done the wrong way around, so you have to scroll four pages down, then read one, scroll up, read the next, then scroll four down again!

    how about Bauer?

    That means “farmer”; literally one who “builds” crops on his land, a verb now disambiguated with a prefix as anbauen. I don’t know if a former “dwell” meaning is hidden in Nachbar “neighbour”, which is always explained as nah + Bauer, or if it really just means “the next farmer over”.

  111. Lars (the original one) says

    Actually, according to EWikt it goes back to PGer *būraz which is glossed as ‘dweller, inhabitant’, I thought it was a later formation from the verb and agentive -er. But your point stands, in modern German it feels so strongly connected to the anbauen sense that it does mean ‘agriculturalist,’ and that was what I was asking about.

    And yes, it’s the ancestor of G -bar and cognates all over West Germanic. But ON has nábúi instead, that looks like a different formation.

    (Old Danish did have forms like naabur, it’s not quite clear whether the ODS thinks that’s only because of the Low Germans — and we do have a later alternative form with the agentive suffix, en bybo and en byboer are interchangable, both ‘city dweller’. (Don’t be confused by the plurals byboer and byboere)).

  112. Trond Engen says

    I’m reading Jón Axel Harðarson: The morphology of Germanic from Klein, Joseph, Fritz (eds.): Handbook of Comparative and Historical Indo-European Linguistics (2017). From p. 934-935 (my emboldening):

    Most verbs of class VII must originally have shown ablaut alternations in their reduplicated preterites. The “strong” stem of the singular had o-grade of the root, whereas the “weak” stem of the dual and plural had zero-grade, cf. *re-rō-/re-r- (OIce. rera, rero/røro), *se-zō-/se-z- (OIce. sera, sero/søro), *le-lōt-/le-lt- (Go. lailot, OE [Angl.] leorton, dissimilated from *leltun), *re-rōþ-/re-rđ- (Go. -rairoþ,OE reordon) from *rō-je/a-‘row’, *sǣ-je/a- ‘sow’, *lǣt-e/a- ‘let’ and *rǣđ-e/a- ‘advise’. But it is not quite clear whether those preterites that had a-vocalism in the root forms of the singular, as did the corresponding presents, still maintained the zero-grade of the root in the dual and plural in late PGmc or had generalized the vocalism of the singular. In other words: We do not know for sure whether PNGmc and PWGmc had the pattern *he-ǥait- :*he-ǥit-,*he-ǥald- :*he-ǥuld- or *he-ǥait- :*he-ǥait-,*he-ǥald- :*he-ǥald- before the reduplicated forms were compressed or replaced by new ablaut forms (cf. OE heht and hēt from hātan ‘call, command, promise’ and leolc <*lelk- and lēc from lācan ‘swing, move, play’). The possibility of explaining some of the preterite forms with secondary ablaut as the regular outcome of reduplicated forms with a zero-grade root favors, however, the first alternative, cf., e.g., *e-uk- (OIce. iók) from *auk-i/a- ‘increase’, *be-ƀut-(> *beut- or *beft- in OE bēot and beoftun, beafton, cf. northern ME pret. and pp. beft in Cursor Mundi; for these forms cf. Hogg and Fulk 2011: 254; in the supposed preform *be-ƀut- [beβut] either the fricative ƀ or the vowel u was ejected.) from *baut-i/a- ‘beat’, and *he-ǥit- > *[xe-ʝit-] (> *heht- or *hej(i)t- >*heit- >*hē2t-) from *hait-i/a- ‘call, command, promise’. Later, the stem forms of the plural would have been extended to the singular. Preterite forms like *hleup- (OIce. hlióp,OE hlēop) from *hlaup-i/a- ‘leap, run’ or *lē2k- (OIce. lék,OE lēc) from *laik-i/a- ‘play’ are best explained as analogical formations. Jasanoff (2007: 265 ff.) explains plural stem forms like *heht- and *held-(for phonotactically impermissible **hegld-) as a result of the proportion *le-lōt-,*re-rōþ- :*le-lt-,*re-rđ- =*he-ǥait-,*he-ǥald- : pl. X.

    Lars: I thought [Bauer] was a later formation from the verb and agentive -er.

    My first instinct was the opposite of yours, that the Gmc agent nouns first meant “dweller”, and that the verb took on the meaning “cultivate” secondarily from the noun. This could be a late development, since the sense is directly attested only in CWGmc, but note Scand. bygg “barley” <- “crop, cultivation”.

    I now think the dual sense “dwell/farm (intr. = be a farmer)” could be inherited from PIE. It would be natural for mobile pastoralists to equate permanent settlement with farming. The transitive and and specific sense of “grow (a certain crop)” came later in Gmc.

    ON has nábúi instead, that looks like a different formation.

    Yes. ON búi is another deverbal noun meaning “dweller, inhabitant; farmer”.

  113. David Marjanović says


    Ah, so the Baur, 16th c. plural bawren spellings weren’t hypercorrect.

    it does mean ‘agriculturalist,’

    It does include pure livestock farmers.

    I now think the dual sense “dwell/farm (intr. = be a farmer)” could be inherited from PIE. It would be natural for mobile pastoralists to equate permanent settlement with farming.

    *lightbulb moment*

  114. Trond Engen says

    in the supposed preform *be-ƀut- [beβut] either the fricative ƀ or the vowel u was ejected.) from *baut-i/a- ‘beat’

    I didn’t spell out the relevance since I thought it obvious, but for good measure: Replace the *t with the unspecified laryngeal *H. “Ejection” of [β] and loss/assimilation of the laryngeal then leads straightforwardly to ON bjó. There are complications when we look at the ON plural forms, I think, but that will have to wait.

  115. Lars (the original one) says

    I don’t know that *būraz is specifically an agent noun, I don’t remember seeing a morpheme / root extension like that elsewhere in Pre-PGer — but before I found that I was assuming that Bauer had been formed like Da boer at a much later stage, with the *-sr- ‘professional’ morpheme that Piotr wrote about.

    So the North Germanic form would have been *bew at some point? Makes sense. I’m still not sure there’s an exact correspondence with PI *fefuai because that’s supposed to stem from an aorist.

  116. Trond Engen says

    ON bjó is quite explicite in its demands for an immediate pre-form *bew (or very similar). The difficult part for me was getting there from a presumed reduplicated form **bebawH. I don’t know if it ever was *bew, though. The loss of [β] could have happened while the rule *ew > *ju was still productive.

    Could PI *fefuai be metathesized from **fefiau or ultimately **fefeuh₂?

  117. Lars (the original one) says

    Could PI *fefuai be metathesized from **fefiau or ultimately **fefeuh₂? — I didn’t even think that far, I’m stuck on why a root aorist grows reduplication on the way to PI. There are lots of reduplicated thematic aorists in PIE, but bʰúh₂-t is not one of them. (EDIT: Checked Ringe, and it seems that root aorists were almost the only type in PIE but lots of them got reshaped in the daughter languages).

    The ON preterite should be from a perfect, presumably created at a later stage than “West IE,” and creating a reduplicated perfect does not strike me as strange. But even if Pre-PI had the same perfect, it would be lost and the *fefuai that we see built on a reshaped aorist.

  118. Trond Engen says

    Bjorvand & Lindeman in my usual ad-hoc translation:

    The ON bjó may compared with other IE perfects […] reflect a reduplicated perfect IE *bhe-bhów(H)-e > Gmc. *be-báwe > PN *be(ba)u > ON bjó, ref. ON fell < PN *fe(fa)ll [to the verb falle].

    It’s what happened through PN that has bothered me. Following Harðarson above, PGmc. would have had the two roots sg. ‘be-baw- and pl. *be-bu-. The latter would have developed to bjó and have been extended tp the singular.

    But that’s not what we actually see. The ON verb has different preterite stems in singular and plural: Wiktionary:

    1p sg bjó
    2p sg bjótt
    3p sg bjó
    1p pl bjoggum, bjuggum
    2p pl bjogguð, bjugguð
    3p pl bjoggu, bjuggu

    The -gg- must be from *-ww-, and I can’t see any other source for the gemination than the laryngeal: *be-buH- > *bewH- > *beww-. So what’s going on in the singular? It can’t be from *be-bówH-, and if it’s in some way analogical, it’s odd that it didn’t either come from the plural or strike the plural as well. Did the analogical leveling from the plural happen before the laryngeal was assimilated, so * be[β]úH- ~ *be[β]uH-´ > *beúH- ~ *bewH-´ > bjó ~ bjugg. But that means we’re bringing the loss of the laryngeal very close to ON.

  119. David Marjanović says

    So! I’ve finished reading the really thorough paper cited above on *-uj- > *-ij-, and obtained a one-page-per-page pdf from the source.

    First we get six pages on the declension of *u-stem adjectives in Germanic and Baltic and on the classic attempt to explain the oddity that many of their forms look like *ja-stems. Then there’s a page on the fact that Gothic -jan-verbs ( < *-je-presents) made from u-stem nouns lack this u. This was first noted by Schmidt (1883), who postulated *-uj- > *-ij- preceding the loss of intervocalic *j in Germanic. Then we get another page and a half on why Vedic doesn’t contradict the existence of this *u in the first place (basically, Avestan has u in those places), and the affirmation that *-uj- > *-ij- happened in Germanic and Baltic.

    The second section treats “the inflection of ‘be’ in East and West Germanic in detail. There are two verbs meaning “be”. One continues good old *h₁ésti; *h₁sénti:

    Gothic: im, is, ist, sijum, sijuþ, sind
    Anglian Old English: eam, earþ*, is, earun, sind, sindun

    therefore PGmc. *izmi, issi**, isti, ?, ?, sindi.

    * via earþ þu < *ear þu
    ** Given the OE r, I’d reconstruct *izi, with word-final devoicing in Gothic, but that’s all beside the point here.

    The other is fully preserved only in Old English:

    Late West Saxon: bēo, bist, biþ, 3 x bēoþ
    Anglian: bīom, bis, biþ, 3 x bīoþ

    therefore PGmc: *biō, bisi, biþi, ?, ?, bianþi.

    But Gothic had it once, too: there is an attested word for “but at the same time also”, bijandzuþþan, which is effortlessly segmentable as *bijands-uh þan, literally “being-and then”, with the present participle of this *bi- verb.

    Etymologically, that verb “obviously” belongs to PIE *bʰuH-, a root aorist meaning “grow”, later “be, come into being, become”. The Germanic present stem is most parsimoniously a *je-present: *bʰuH-jé-, which appears to have also given us Greek φύομαι “I grow” and Latin fiō “I become”.

    Then, Dybo’s law strikes (without ever being called such, “most recently Schrijver, 1991”): long vowels followed by resonants followed by the stress are shortened in Germanic, Italic and Celtic, *bʰūjé- becoming *bʰujé-.

    (Actually, add Holtzmann’s law, and it should become *bujji- at some point. This *-jj- probably turns out to be irrelevant, but I wonder about the Norse bygg- verb. But see below.)

    Then, we would expect:
    *bʰujṓ, bʰujési, bʰujéti; bʰujónti
    > *bujō, bujesi, bujeþi; bujanþi
    > PGmc. **buō, buisi, buiþi; buanþi,

    which is not what we can in fact reconstruct: as mentioned, that’s *biō, bisi, biþi; bianþi.

    *uj > *ij to the rescue:

    *bʰujṓ, bʰujési, bʰujéti; bʰujónti
    > *bijō, bijesi, bijeþi; bijanþi
    > PGmc. **biō, biisi, biiþi; bianþi.

    Now we’re almost there.

    We can get all the way there by taking into account that the *h₁es- verb became a clitic, i.e. lost its stress, sometime before Verner’s law: that’s the only explanation for the *z in *izmi and the *d in *sindi. This suggests that the *bʰuH-jé- verb also became a clitic “in Proto-West-Germanic times”. That would explain why bis and biþ are so often spelled with y in Late West Saxon, where unstressed e is routinely spelled y (hælynd, fædyr, wintrys < hælend, fæder, wintres “savior, father, winter’s”).

    (Lack of stress would also prevent Holtzmann’s law from operating. The Norse bygg- is not a copula, so maybe it’s the non-clitic doublet…)

    It also suggests that the *-ii- clusters should get reduced in some way.

    One possibility of what that might mean is a merger into a long vowel, of which endnote 18 says that it’s hard to tell if that would be regularly shortened under continued lack of stress or whatever. I guess it could also have been leveled out by analogy to the short *-i- in the rest of the paradigm.

    The other possibility is that the outcomes were *bjis, *bjiþ. And then, a known sound change would strike: loss of *j before unstressed *i in PWGmc., as exemplified by the distribution of WGmc. consonant lengthening in the “tell” verb:

    PGmc. *taljō, taljisi, taljiþi; taljanþi
    > PWGmc. *tallju, talis, taliþ; talljanþ
    > Old Saxon telliu, telis, telid; telliad;
    Old English ?, teles, teleþ; tellaþ;
    OHG zellu, zelis, zelit; zellent.

    So, this way, we’d get PWGmc. *biu, bis, biþ; bianþ.

    The we get 2 pages on Celtic, which famously has the same 2 verbs. There, too, *bʰuH- shows up with *i, and the traditional explanation contradicts the Gaulish attestations, which have i in the present stem (2pl imperative biiete attested once, < *bʰuH-jé-), but not elsewhere (3sg subjunctive buetid attested twice, < *bʰuH-e-).

    The author is an expert on Baltic, so we get 6 pages on that. There, too, there’s a past-tense copula with i in both Lithuanian and Latvian, which should go back straight to *i in Proto-(East?-)Baltic. We can’t blame Dybo’s law here, which didn’t operate in Balto-Slavic; but if we assume that not only *-uj- > *-ij- as shown by the *u-stem adjectives, but also *-ūj- > *-īj-, then we can probably feed *bīje-into the known process of PIE *-eje- > PBS with circumflex tone, “probably through the intermediate stage” *-ije-. Given that unstressed PBS circumflex was regularly shortened to Proto-Baltic *i, the attested forms follow.

    Without this shortening, we get the bi(-) copula of the OCS future conditional and past/irreal conditional. That, and a convincing attempt to explain away potential counterevidence, takes up the next 4 pages.

    The Latin fi- is not brought up again, not even in the endnotes. Is that just vowel reduction (-umentum ~ -imentum), or could *-uj- > *-ij- have operated in Italic, too, so potentially just twice in IE (once in West IE, once in Balto-Slavic)?

    Endnote 10 mentions Hittite parku- “high”, parkije- “to lift”, and wonders if that’s another independent occurrence of the same sound change.

  120. David Marjanović says

    Endnote, not footnote. (Ran out of edit time.)

    Now that I think of it, *-jj- would help ensure that the WGmc. reduction process would produce specifically *bjis, *bjiþ. But the Gothic bijandz- is unambiguous: *-jj- would have produced **biddjandz-.

  121. Fixed the “footnote” thing, and thanks very much for doing the research and giving such a thorough report!

  122. Trond Engen says

    Yes, thank you indeed. I read this to mean that the West Germanic forms are entirely regular while the North Germanic forms are unexplained. Could it be that the change *-uj- > *-ij- happened only in the unstressed verb, i.e. after it became a clitic, affecting only the West Germanic forms meaning “be (copula)” and not North Germanic “dwell” and “prepare”?

  123. John Cowan says

    Given the OE r, I’d reconstruct *izi, with word-final devoicing in Gothic

    IIRC earþ is known to be irregular, so I wouldn’t base anything on that. (Can’t find a reference.)

    Celtic, which famously has the same 2 verbs

    It’s only in Old English and Welsh, though, that they are sorted out semantically: the /b/-forms are syncretically future and consuetudinal (the last as late as Chaucer). What is more, as Tolkien points out, the Northumbrian plural form is bi(o)ðun, whose ending has no Germanic etymology but is suspiciously like Welsh byddwn, byddwch, byddan(t). All of which screams “Sprachbund effect”.

  124. David Marjanović says

    Gothic is confusing in its tidyness. Wiktionary has both a weak and a storng paradigm, and for some reason the root diphtong is au throughout both.

    That’s just the regular Gothic shift of [i] > ai [], [u] > au [] before w, r, h an in open syllables.

    such a thorough report!

    The paper is so thorough on its 31 pages that it was hard to try to omit things.

    while the North Germanic forms are unexplained.

    Yup, they’re not even mentioned. Neither is bauen or its West Germanic cognates.

    Could it be that the change *-uj- > *-ij- happened only in the unstressed verb, i.e. after it became a clitic

    I don’t think I could explain the *u-stem adjectives or the -jan-verbs formed from *u-stem nouns that way.

  125. By Wikt,

    A fine oath for a Hattic.

  126. David Marjanović says

    I seem to have forgotten to insert the IPA: ai [ɛ], au [ɔ].

  127. Almost certainly there was no phonemic distinction in Gothic between [ɛ ~ ɛː] and [ɔ ~ ɔː], the long vowels appearing just in case another vowel followed. But Jacob Grimm, who devised the romanization of Gothic, wrote aí, aú for the former and ai, au for the latter for no apparent reason.

    What’s more, Grimm wrote ái, áu for what had been actual diphthongs /ai/, /au/ in Proto-Germanic. These digraphs appear only in native words: in Greek words the diphthongs are written aj, aw as in Pawlus, (Orthographically Wulfila used the rune ᚢ ur (or something that looked very like it) for /u/, Greek Υ for /w/, Greek Ι with a tail (like modern Latin J) for /i/, and Latin G for /j/.) So it’s probable that written ai, au were always /ɛ/, /ɔ/ in Wulfila’s day, and that the accents were Grimm’s attempt to write Gothic and Pre-Gothic at the same time.

    In addition, Wulfia wrote /iː/ as ei in imitation of Greek, but it also happens to coincide with the Proto-Germanic origin of many (not all) occurrences of /iː/ < /ei/, so more glee for Grimm. Length on /a/ and /u/ was phonemic but not marked in writing (a macro is used in Grimmanization); e, o are always long and tense.

  128. David Marjanović says

    the accents were Grimm’s attempt to write Gothic and Pre-Gothic at the same time.

    That’s the explanation I’ve encountered: Gothic for the use of comparative linguists.

    actual diphthongs /ai/, /au/ in Proto-Germanic

    Not that it matters much, but I think (following this idea and my poorly coherent further rambling on it) that Proto-Germanic didn’t have phonemic diphthongs, just closed syllables in /aj/ & /aw/ (& /ew/), and that diphthongs treated by the language as vowels, instead of as sequences of a vowel and a consonant, are a West Germanic innovation. That would explain why East and North Germanic have long consonants traceable to /jː/ and /wː/, while West Germanic has sequences of diphthongs and short /j/ or /w/ in their place.

    The same kind of phonological shift must have happened in Greek in late Mycenean times and the Greek Dark Age as part of the wholesale abolition of /j/ and /w/, allowing some to leave traces by metatheses-plus-vocalizations like *korwos > kouros and *korjanos > koiranos.

  129. the wholesale abolition of /j/ and /w/

    Alternately, Greek abolished syllable-initial glides but not syllable-final ones. I have even wondered if this was through some kind of fortition: some PIE *y give ζ after all, and also late Etruscan ‹F FH› for supposed /w f/ would really make more sense if this were something like /v f/. Cf. moreover the general Modern Greek reflex /v~f/ from ancient αυ ευ, which under your hypothesis would first require a shift from /a͡u e͡u/ back to /aw ew/.

    (We also find /v/ as the modern Tsakonian reflex of ϝ, but that doesn’t say much when nearly every language variety in continental Europe west of the Urals fronted *w to /β ~ v ~ ʋ/ over the last 2000 years anyway.)

  130. Roberto Batisti says

    @ David Marjanović

    *korwos > kouros is not metathesis, though, but (the third) compensatory lengthening – the digraph just stood for /o:/ here, but other dialects with 3rd CL write it as κωρος or κορος (ō implied). Cf. *kalwos > κᾱλός in the same dialects (East Ionic, etc.).

    *korjanos > koiranos *is* a metathesis, descriptively, though perhaps not a direct one but rather going through a phase *koʲɾʲ-.

  131. David Marjanović says

    Alternately, Greek abolished syllable-initial glides but not syllable-final ones.

    That would be weird, though. Has this happened anywhere else?

    But I don’t have a better explanation for αυ ευ. There are inscriptions where αυ is spelled AO, but they definitely date to after υ had become [y] everywhere else.

    3rd CL

    Oh yes, thanks.

    through a phase *koʲɾʲ-

    Most likely – this would have been what allowed the frontness to be interpreted as a feature of the preceding syllable in the first place.

  132. Has this happened anywhere else?

    Cantonese monophthongized all on-glides while maintaining the Middle Chinese coda consonants /j/, /w/, /ɥ/ as off-glides. Cantonese is usually called conservative because it has kept the full set of nine codas, including the above three and /m/, /n/, /ŋ/, /b/, /d/, /g/ (usually transcribed with voiceless stops, but they are unaspirated and often unexploded). But between monophthongization and reducing the 36 MC initials to only 19 (compared to 23 in Mandarin), plus in-progress /n/ > /l/ and /ŋ/ > zero sound shifts, giving only 17 in actual speech, “conservative” is far from being the mot juste.

  133. David Marjanović says

    Spelling confusion between υ and β between vowels is attested all the way back to the 2nd century BC, says the first page of this teaching document.

  134. nearly every language variety in continental Europe west of the Urals fronted *w

    Yes, islanders verbalize very weirdly, even when they wend from the island to continental life again.

    But there are weirder places than Britain and Iceland, and sometimes the continent comes to the island rather than vice versa. The Urkers dialect of Low Saxon is spoken in Urk on the western tip of the North-East Polder in the IJsselmeer. It is separated by the whole expanse of the polder (which is entirely Dutch-speaking since it did not exist until 1942) from the nearest other Low Saxon speakers, and until 1939 when a bridge was built, Urk was an island. Although I don’t have detailed information on Urkers in any language I can read, sources agree that it is the most divergent Low Saxon variety, which is saying quite a bit considering how dialectal Low Saxon is.

  135. Stu Clayton says


    I was about to come down like a ton of bricks on that capitalization of the “j”. Caution prevailed, I first checked the German WiPe. Sez there in Dutch “ij” is a ligature.

  136. Yep, a ligature that isn’t actually ligated.

  137. Stu Clayton says

    Urkers for beginners

    As a bonus, here’s a posh English robot explaining what it’s all about:

    Urkers dialect

    It doesn’t have “Urk” in its vocabulary, so it spells it out each time: u-r-k. What a clever little robot it is !

    There’s a pronunciation audio at the beginning of the German WiPe article on Urk. The “r” sounds like a front flap.

  138. Trond Engen says

    John C.: It is separated by the whole expanse of the polder (which is entirely Dutch-speaking since it did not exist until 1942) from the nearest other Low Saxon speakers, and until 1939 when a bridge was built, Urk was an island.


  139. IJ laughed!

  140. Stu Clayton says

    As I’ve said before, somebody should do something about this Trond Engen person. His puns are always 10 times better than mine in my mother tongue.

  141. Trond Engen says

    Er is geen IJssel meer.

  142. Stu Clayton says

    And in Dutch too, but that’s ok since I have no ambitions in that respect.

  143. Probably Norway’s best Dutch-punning structural engineer.

    It’s a small but viciously competitive field.

  144. Yes, except that pronouncing IJ as /ai/ rather than /ei/ makes you sound like a post-millennial.

  145. Trond Engen says

    I know a couple of Dutch and Belgian structural engineers working in Norway, and I wouldn’t get into a a punfight with any of them — especially since my own Dutch is of the pretend type. If the line works at all, it’s because I ran it by my Dutch copy-editor, my very-soon-to-be-eighteen-year-old daughter, who has developed a thing for Belgians and Flemish. I’ll ask her about the postmillennialism, but in my mind I was mispronouncing English rather than Dutch. I don’t know what that makes me.

  146. David Marjanović says

    Yep, a ligature that isn’t actually ligated.

    Like ы, uppercase Ы.

  147. Belgians

    Batavians, Helvetians and Illyrians.


    Just belated protest about appropriation of this ancient Celtic tribal name by 19th century descendants of Gallo-Romans and Franks.

  148. January First-of-May says

    Well, they had to call themselves something, and if they couldn’t agree on much else anyway, an ancient name from the right area was about the only option in the first place. (It helps, of course, that there don’t seem to be any actual Belgae around.)
    I’ve seen “Flanoonistan” proposed elsewhere, and I suppose it’s an appropriate name for the modern dysfunction, though I don’t think it would have been particularly popular in the 19th century.

    Ironically enough, Tacitus reports that it was a tribe from roughly what is now Flanders that originally decided to call themselves Germans…

    [EDIT: in any case, Francia, Gallia, and Helvetia were already taken by other places, Illyria was from nowhere near the right region also already taken, and Batavia was either 1) taken by another place or 2) associated with a highly unpopular occupation.]

  149. Well, they had to call themselves something

    I suggest Lower country.

    Or even Lowest country (if Luxembourg doesn’t mind)

  150. Lower Country would lead to mix up with the neighbour to the North, wouldn’t it?
    And Lowest country would have been an untruthful claim.
    In any case, even if it’s mock indignation, I don’t really get what you’re protesting about. The country is in the right area and at least some of its population may be the descendants of the ancient Belgae, so it’s not an unreasonable choice.

  151. Like ы, uppercase Ы

    ‹å›, ‹ů›, ‹ä ö› (< ‹aͤ oͤ›)…

  152. I don’t think of base + diacritic(s) as a ligature. Indeed, with very few exceptions a diacritic that touches its base doesn’t count as a diacritic to Unicode. Exceptions are çedilla, ǫgǫnek, and Vietnamese hơrn. But beɬt and słash are just variants of plain old ell.

  153. I think of diacritics more as garniture than ligature. A handful of diacritical parsley helps the furrin medicine go down.

  154. Lower Country would lead to mix up with the neighbour to the North, wouldn’t it?

    If we adopt the Lowest Country name, in French, Netherlands will be called Les Pays-Bas and Belgium Le Pays Le Plus Bas.

    Very neat.

  155. {thinking} on the other hand, if we are to appropriate Roman provincial names for modern countries, maybe we should just rename the Netherlands.

    Germania Inferior, I believe, it was called.

    Fits the Dutch just perfectly.

  156. Stu Clayton says

    Low is not always a bad thing. Remember the limbo from Trinidad: “How low can you go?”

  157. Indeed, Early Modern Low German was a sensible language for sensible people, at least from the English/Frisian/Dutch/Scandinavian perspective. Non-Low German, who can understand that? Kipperish.

  158. David Marjanović says

    Also from other perspectives. A round of apocope that happened later has been compensated in some dialects by the mere abandonment of word-final fortition like in English or French, but in others it reappeared, and a third degree of vowel length developed in some of these, while others have gone for a pitch accent system which is, IIRC, upside-down from a Rhineland perspective…

Speak Your Mind