Stet and An Untranslatable Poem.

“To cheer you up in the practice of your profession,” John Cowan sent me a link to Stet, a collection of intemperate responses by authors to publishers who have committed editing outrages, e.g.:

In all the proof that has reached me, windrow has been spelled window. If, in the bound book, windrow still appears as window, then neither rain nor hail nor gloom of night nor fleets of riot squads will prevent me from assassinating the man who is responsible. If the coward hides behind my finding, I shall step into Scribner’s and merely shoot up the place Southern style.” — American author Gordon Dorrance (1890-1957), note to his publishers

It did indeed cheer me up, and in browsing around Futility Closet (which I should visit more often) I found (in the Language category) An Untranslatable Poem, about a six-word Portuguese poem by Cassiano Ricardo called “Serenata sintética” that “is so embedded in its language that García Yebra found himself unable to convey it in another tongue”:

“In this short poem, phonemic form is everything,” write Basil Hatim and Ian Mason in Discourse and the Translator. “The words themselves are evocative: a small town with ‘winding streets’ (rua torta), a ‘fading moon’ (lua morta) and the hint of an amorous affair: ‘your door’ (tua porta). But their impact is achieved almost solely through the close rhyme and rhythm; the meaning is raised from the level of the banal by dint of exploiting features which are indissociable from the Portuguese language as a code.

“García Yebra relates that he gave up the attempt to translate the poem even into Spanish, a language which shares certain phonological features with Portuguese.”


  1. An untranslatable poem in English is Ralph Hodgson’s “Eve,” which doesn’t even work in American. Its rhythmic effects are totally dependent on the vowel and syllable lengths of RP.

  2. I don’t find the poem a failure in AmE, at least not my own AmE: I read it out loud to myself. The problematic rhymes are was/grass, a/lay, syllable/fell, maid/said, vain/again. The first is not a perfect rhyme in either AmE (STRUT/TRAP) or RP (LOT/BATH). The second might be a rhyme in either accent if stress fell on a, which would put it into the FACE lexical set, but it doesn’t; everyone has the choice of an unnatural pronunciation or a slant rhyme. The third is schwa/DRESS in both AmE and RP, again not a perfect rhyme, and again the same choice. The fourth is FACE/DRESS in AmE, but FACE/FACE in RP, so this is a true discordance. However, Americans have had to get used to this pronunciation of again in order to read poetry at all. (When I say “AmE” I exclude the Eastern New England accent, which distinguishes TRAP and BATH.)

    The question of vowel length escapes me: the modified Scottish Vowel Length Rule that prevails in AmE puts it out of conscious awareness. If someone can explicate where length matters in this poem as distinct from vowel quality, I’ll be happy to comment.

  3. Oops. By “the fourth” I meant “the fifth” (vain/again). The true fourth is FACE/DRESS in both accents. Behold the monster Haplography in his native haunts!

    I checked my pronunciations against some dictionaries as well as the one in my head: for AmE, ODO and Collins for BrE.

  4. @John Cowan: In my experience, modern RP speakers tend to say “again” with DRESS. The only times I can recall hearing it with FACE, it’s been from Canadians.

    While we’re on the topic, though, it’s always puzzled me that modern(ish) songwriters – not generally inclined to use old-timey sight rhymes – still persist in rhyming “again” with FACE words, and that the singers never seem to use the alternative pronunciation of “again” to make the rhymes work. For example, in “Singin’ in the Rain”, the song rhymes “again” with “rain”, but Gene Kelly sings it with /ɛ/. In “I’ve Got a Feeling”, the Beatles rhyme “again” with “train”, but likewise use /ɛ/. I know I’ve heard this in several other songs, although they’re not coming to me at the moment.

  5. Songwriters probably use rhyming dictionaries, and singers probably don’t use any dictionaries.

  6. Off topic, but I was struck by Dorrance’s use of “merely” in “merely shoot up the place”. I could see myself using “simply” that way, but not “merely”, nor “only”, FTM (unless the “merely” or “only” was itself meant jocularly).

    Presumably, Dorrance did not intend to actually shoot anybody. But he doesn’t seem to have been using “merely” in a broadly comedic sense of, “What? I merely shot up the place!”

  7. Well, I suppose it could be the OED’s sense 2a. “Absolutely, entirely; quite, altogether” (rare after 1800), though that seems unlikely.

  8. The phrase “hides behind my finding” looks odd to me too. What finding? I take it that it means “hides so that I cannot find him”, in which case “merely” means something like “in default of anything better,” as in “Lacking a weapon, I merely bit him.”

  9. Yes, you must be right about “merely” (and of course you’re right that it’s an odd phrase).

  10. I’ve also often wondered about the dead convention by which again is made to rhyme with FACE words. The DRESS pronunciation was good RP already in Lewis Carroll’s day:

    ‘But oh, beamish nephew, beware of the day,
    If your Snark be a Boojum! For then
    You will softly and suddenly vanish away,
    And never be met with again!’

    He offered large discount—he offered a cheque
    (Drawn “to bearer”) for seven-pounds-ten:
    But the Bandersnatch merely extended its neck
    And grabbed at the Banker again.

  11. In Kelly’s version of “Singing in the Rain,” I notice a difference in the vowel qualities between the two times that “again” comes up as a rhyme for “rain.” The first time, when the lines fall right after each other in sequence, Kelly seems to do a bit to disguise the fact that the words don’t rhyme. The second time, there’s an instrumental sequence between “rain” and “again” and the words obviously do not rhyme.

    I was going to add that this song could provide an interesting example for studying how the different interpretations of the lyric-writer and performer could interact. Since Kelly was also co-director of the film and the lyricist Arthur Freed produced the film, the influence of others on the interpretation of the song might have been fairly limited. However, I recall now that the song was already a standard by the time the film was produced, so earlier renditions of the song may have significantly influenced how Kelly’s performance was done.

  12. “Singin’ in the Rain” first appeared in Hollywood Revue of 1929. The singer was Cliff “Ukulele Ike” Edwards, later known as the voice of Jiminy Cricket. He definitely uses the DRESS vowel.

    You can judge for yourself at this site:

  13. Here’s an example of a basically RP speaker using the FACE vowel at around 3:54:

    My grandparents had this routine on a 78 RPM record, and my father had it memorized, so it became a part of our family lore. One instance, when it comes to questions of pronunciation, “It all depends on which university you’re from” (7:15). Some other examples of linguistic humour are scattered throughout the sketch. “What a common office-boy!”

  14. That was delightful, thanks!

  15. @maidhc: That gives two pronunciations of “lignum vitae” (“depends on which university you’re from”), but neither represents how the rare wood enthusiasts that my father hung around with twenty years ago pronounced it. They said VI-tai, although I can’t find that pronunciation in any dictionaries, British or American.

  16. They said VI-tai, although I can’t find that pronunciation in any dictionaries, British or American.

    I’m having trouble interpreting your phonetic spelling; “VI” looks to me like /vay/ (which I would spell “VYE”), but “tai” looks to me like /tay/ (“TYE”), and presumably you wouldn’t have spelled the same vowel two different ways. Could you elaborate?

  17. Sorry, try VEE-tai.

  18. Ah. That would be reconstructed classical Latin pronunciation, except for using [v] for /w/ (spelled “v”).

  19. The New Oxford American Dictionary (3 ed., Oxford University Press, 2010) has it as the second pronunciation: “lignum vitae /ˈlignəm ˈvīˌtē, ˈvēˌtī/.”

  20. @John Cowan: I wasn’t primarily thinking about the rhymes but rather the syllable lengths, which produce a definite meter (long, long, short-long-long long (3) long short-short long) in RP and just a mess in American. It’s a very clever musical experiment that only works in its own variety of English. This is also the reason that Robert Bridges etc.’s attempts at real English quantitative verse fail in American.

  21. And of course by “American” I mean “the way I talk.” Your mileage may vary.

  22. *Sigh* Me again. First, I apologize for not reading your last paragraph carefully, or my post would have been more to the point. I think I addressed it anyhow, though.

    Also notice how the quantitative/musical meter I described breaks down in the final stanza, reflecting the sense.

  23. I still don’t see it. First of all, I don’t know if a short vowel followed by a coda consonant is meant to count as a short or a long syllable. If the latter, then “Wading in” is long-long-long, whereas your meter predicts long-long-short. If you are looking only at vowel lengths, then “Eve with her” is long-short-short, again not long-long-short.

  24. I must admit I’ve never been able to see how quantity is supposed to work in English.

  25. It depends on whose English. In RP, the length of a vowel is locked to its (qualitative) identity: the LOT vowel is short whereas the PALM vowel is long, for instance. In Scottish English and American English, vowel length is allophonic: seat has a short vowel, seed a long one, both FLEECE. In Australian English, length is phonemic: cup has [a], carp has [aː].

  26. This Canuck generally uses the FACE pronunciation, but can outomatically use the DRESS pronunciation if rhyme requires. To me it’s a flexible word. Flexibility in language reflects a flexible mind, which to me is preferable to a rigid one.

  27. -auto- of course.

  28. I don’t blame anyone for not knowing the word “windrow”, but in what possible context would “window” appear to be a reasonable correction?

    More embarrassingly, I had to text my sister to ask what we call windrows in Danish. I haven’t used the word in a decade, I think. It turns out to be one of the few words that still shows regional variation. I guess since most people rarely have reason to talk about them (like me).

  29. Hi, Sili, since we’re all urban nowadays it’s understandable that windrow is not in use. But according to google translate you will get the word slåttersträng in Swedish and skår in Danish. I hope you don’t mind my using the letter ‘å’:

    Byens navn er Aabenraa, uden svenske boller paa (the town’s name is Aabenraa, without Swedish balls upon).

    On the other hand the Norse vindue (window) is absolutely opaque to modern Swedes (unlike to Danes and Norwegians). It’s 100% substituted by fönster, German Fenster.

  30. I promise not to go on about this, but yes, I was being hasty and oversimple (as seems to be my wont) when I implied that “Eve” has a strict quantitative meter. I’m pretty sure, though, that the poem derives a very complex musical effect from the interplay of RP vowel and syllable lengths, and that much of the “poetry” disappears when the poem is read in Scottish or American English.

    This musical tour de force is the only reason I can see for the survival of this poem, which expresses the Fall of Man in a trope of child seduction and manages to trivialize both.

  31. It’s kind of a pity that English lost *eyethirl, *eyetril (parallel with nostril ‘nose hole’) when it picked up Norse vindaugr ‘wind’s eye’. Windows nowadays are much more for looking through than they are designed to let the wind in.

  32. David Marjanović says:

    German Fenster

    Itself of course from Latin fenestra, but such an early loan that it got merciless stress on the first syllable – something we haven’t done in a thousand years.

  33. Stefan Holm says:

    stress on the first syllable – something we haven’t done in a thousand years..

    What about Leutnant – 16th c. borrowing. But maybe you explicitly meant borrowings from Latin?

  34. The vowel reduction in Leutnant already happened in French, I think.

  35. In the early 16th century? I dunno, but maybe marie-lucie can help out. It’s an odd word anyway: why Leu-?

  36. From the German Wikipedia article on Leutnant: It is borrowed from French “lieu tenant”, which in turn comes from medieval Latin ”locum tenens”, governor or representative (locum). In pre-revolutionary France the Lieutenant du Roi was the representative of the king in the army or in a fortress. Napoleon after his Spanish campaign gave the title Lieutenant de l’Empereur to his appointed vice king of the Iberian peninsula. At least no vowel reduction in spelling.

    In the German army a Leutnant (Sw. “löjtnant”) however became the lowest (or one of the lowest) ranked officers. The change from “lieu-” to “leu-”/”löj-” is just a phonetic adjustment to Germanic. As for the reduction of the “e” maybe someone wondered what people (Leute) would make of a word like “Leute-nant”?

  37. The change from “lieu-” to “leu-”/”löj-” is just a phonetic adjustment to Germanic.

    Do you have any other examples of that phonetic adjustment?

  38. Stefan Holm says:

    No, Hat, I can’t think of any. What I had in mind was, that /iø/ after a consonant wasn’t part of the regular German phonetic inventory, while ‘eu’ is very common (neu, treu, heute, teuer, Feuer, Freude, Deutsch) and as such probably was attractive.

    In Swedish (and German) there actually are a few instances of preserved French ‘ieu’. One is ingenieur/ingenjör (engineer). Another is adjö (Ger. adieu), from French à Dieu (‘to God’) used as polite goodbye or farewell. But the difficulties with that can be shown by the forms from already early 16th century ade or adde in both German and Swedish.

    A Scandinavian variety of the umlaut, the so called breaking, gave us combinations of consonant(s)+j. But we never felt comfortable with that. So preceding d, t, g, k, h, l and s have all merged with the ‘j’ into fricatives. Ljöd (sounded) is thus pronounced /jø:d/, tjöt (howled) is /çø:t/ and sköt (shot, the verb), where we also have dropped the j in spelling, is /xø:t/ Even the aforementioned adjö never comes otherwise than /a’jø:/.

  39. ‘eu’ is very common (neu, treu, heute, teuer, Feuer, Freude, Deutsch) and as such probably was attractive.

    But it doesn’t sound at all like what it replaced. I remain unconvinced. Why not /letnant/?

  40. David Marjanović says:

    What about Leutnant – 16th c. borrowing.

    Oh. Oops. That’s an exception. The deformation that allowed the folk etymology of the first part must have helped, I guess.

    Or maybe I should rather drastically reduce my estimate from 1000 to 300. :-] 1000 was exaggerated anyway, what I had in mind is more like 700: the borrowings into Middle High German did mostly get initial stress. (An example is French aventure > MHG aventiure > NHG Abenteuer with assimilation to Abend “evening” and teuer “expensive, dear”. On the other hand, the half-borrowed compound suffix -ieren, which is highly productive with French and Latin roots and even a bit with native ones, is stressed, leaving lots of verbs stressed on the third or fourth syllable.)

    But it doesn’t sound at all like what it replaced.

    Postulate, for the sake of the argument, a metathesis that brings the word closer to German phonotactics: [liø]- ~ [ljø]- to [løi̯]. Then open the can of worms that is the pronunciation of German eu through time and space: descended from MHG /yː/, it must have been [œʏ̯] just after the “New” High German Diphthongization struck*, but since then it has moved on: in the ancestry of the standard, the first part was backed and the second more or less unrounded, leaving us with [ɔɪ̯] (more rounded versions are found in the north, where the allophone range of /ɪ/ includes [ʉ]; that must be why it’s often transcribed as [ɔʏ̯]). Other varieties feature [ɛɪ̯]**, [ɶ̝]***, [ui̯]… and that’s just the Bavarian dialects, or at least most of them.

    Alternatively, maybe the spelling was borrowed instead of the pronunciation (and the -i-, which would be phonotactically bizarre in German, dropped soon after). That may be harder to imagine, but then, all the -ier endings (Bombardier, Grenadier, Kanonier…; also Portier; usually not Bankier, a later borrowing) do have spelling pronunciations in -|ˈiːr|.

    When did the -t of lieutenant fall silent?

    * …which it hasn’t in High Alemannic, let alone Low German. It’s similar to the diphthongizations in the Great Vowel Shift of English.
    ** Merger with MHG ei.
    *** A phoneme distinct from /œ/.

  41. David Marjanović says:

    Der Leutnant von Leuthen befahl seinen Leuten, so lang nicht zu läuten, bis der Leutnant von Leuthen seinen Leuten das Läuten befahl.

  42. Postulate, for the sake of the argument, a metathesis

    Oh, I can postulate a dozen things before breakfast — I had plenty of training as a budding Indo-Europeanist. (“Sound changes have no exception, except for all these examples we have special explanations for…”) I simply don’t find that particular postulation all that plausible. Anything’s possible, of course.

  43. marie-lucie says:


    The French word is lieutenant, literally and still recognizably ‘placeholder’ (as in the Latin phrase) and as far as I know has never been anything else since its creation. If any language omitted the i (pronounced [j]), it is not French, but another, borrowing language in which the sequence [jö] was not permissible (at some point in its evolution).

  44. David Marjanović says:

    I simply don’t find that particular postulation all that plausible.

    Spelling-pronunciation it is, then. ;-)

    in which the sequence [jö] was not permissible

    More likely the problem was the /lj/ cluster lacking a syllable boundary through it.

  45. marie-lucie says:

    More likely the problem was the /lj/ cluster lacking a syllable boundary through it

    Yes, that may be more plausible.

  46. Stefan Holm says:

    Yes, the initial cluster consonant+j seems to be it. C.f. the Gmc ‘love’ word: The verb has been reconstructed as *lubojan. After syncope and umlaut it’s attested in OHG as liubi and in ON as liuba. But since we disliked the cluster /lj/ the Germans chose to make it a plain vowel lieben (to love) while the Swedes got rid of the initial ‘l’, so that written ljuv is pronounced /jʉ:v/ (lovely, sweet). C.f. Gothic liufs (beloved, dear).

    English love appears to come from the corresponding noun *lubu and didn’t have to bother with a disturbing ‘i’/‘j’. From the noun we also have Lob in German and lov in Swedish, both meaning ‘praise’.

    And for the sake of it – Swedish does accept the initial clusters ‘bj-‘, ‘mj-‘ and ‘nj-‘ but just in a handful of inherited, although common, words (e.g. bjuda, björk, björn, mjöd, mjöl, mjölk, njure, njuta – bid, birch, bear, mead, fluor, milk, kidney, enjoy respectively).

  47. English love appears to come from the corresponding noun *lubu and didn’t have to bother with a disturbing ‘i’/‘j’.

    OE lufu noun, lufian verb. The modern spelling with “o” is artificial, because “luue” would read as “lwe”; similarly in son, some (too many consecutive minims).

    From the noun we also have Lob in German and lov in Swedish, both meaning ‘praise’.

    OE lof. The last word of Beowulf is lofgeornost ‘most desirous of praise’, literally ‘love-yearn-est’, a characterization of the hero. French praise displaced lof in the 14C.

  48. Trond Engen says:

    Stefan Holm: Swedish does accept the initial clusters ‘bj-‘, ‘mj-‘ and ‘nj-‘ but just in a handful of inherited, although common, words

    bjʉ:k “American car brand”
    mjʉ:sikel “theatre piece with songs” (borderline)
    njʉ:s “Usenet”

  49. Stefan Holm says:

    Sorry, bad English on my side. I meant to say that we have very few inherited words with initial consonant+j. But foreign ones of course! The Asian sea vessel djonk we try to pronounce as written and New York we start with a /nju:/. We aren’t that rigid: ‘New’ is in our language ‘Ny’. And York is (as you surely know, Trond) the viking Jorvik from ON jor = horse and vik = bay. So ‘Nya Hästviken’, (the)New Horse Bay – I don’t think so.

  50. And just one universe away, there’s Nya Sverige, where they speak real Swedish, not that Riksmål-contaminated stuff you mostly use in Sweden. Nya Sverige is part not only of the (eastern) North American League but of the globe-spanning Skandinaviske Riksfællegsskap, which stretches from Tsingdav in the East to Nýja Ísland (also a NAL province) in the West.

  51. David Marjanović says:

    But since we disliked the cluster /lj/ the Germans chose to make it a plain vowel lieben (to love)

    Nah. OHG iu~io, from Proto-Germanic *iu and PIE *eu, must have been a diphthong stressed on the first part, [ɪʊ̯~ɪɔ̯] or thereabouts; it merged into ie in MHG (from OHG ie~ia) before the so-called New High German Monophthongization turned that into [iː] in Central German. It’s still a diphthong in Upper German: my dialect has [ɪɐ̯] (into which ir has more recently merged) in tief and lieb (with OHG io from PIE *eu) as well as in Spiegel and Ziegel (from Latin /eː/, preserved unchanged in Low German).

  52. David Marjanović says:

    the can of worms that is the pronunciation of German eu through time and space: descended from MHG /yː/

    I spent the last 2 days being confused by Swabian. :-) Maybe eu has two origins – MHG /yː/ as in neu, and the umlaut of the io~iu diphthong as in deutsch –, and conservative variants of Swabian distinguish these…

  53. David Marjanović says:

    That’s what I get for getting confused again while writing a comment and editing incompletely: I’ve been spending the last 2 days being confused by Swabian. :-þ

  54. Stefan Holm says:

    Isn’t it standard theory that deutsch is the same word as Icelandic þjóð (folk, people) as in Svíþjóð (Sweden)? In that case ‘eu’ also has the origin ‘io/iu’. But why is the name of your country in Scandinavian Tyskland (in Icelandic Þýskaland)? It‘s consistent with the ’y → eu’ change as in Ger-Swe neu-ny, teur-dyr, steuren-styra, feuer-fyr, (be)deuten-(be)tyda.

    On the other hand the Icelandic Wikipedia article about Germany says, that Þýskaland er kallað Þjóðverjaland í gömlum íslenskum bókum (Germany is called Þjóðverjaland in old Icelandic books). And of course I know next to nothing about die schwäbische Dialekte.

  55. David Marjanović says:

    Isn’t it standard theory that deutsch is the same word as Icelandic þjóð (folk, people)

    Oh yes (also Middle English thede); but if that pesky diphthong was subject to umlaut, OHG diutisc provided the environment for it, and that might explain something, I guess.

    The þjóð root survives in modern German as an element in names, namely Diet- (or Lower-Than-That German Diede-) with the expected ie.

    The High Alemannic dialects have /yː/ for every eu/äu.

    teur [...] steuren [...] schwäbische

    teuer (two syllables: the second is [r̩] in Alemannic, [ɐ] elsewhere), steuern (ditto), schwäbischen because adjective declension is rather malicious. :-)

  56. Jim Parish says:

    On the CanE pronunciation of “again”, another data point: the Stan Rogers song “The Mary Ellen Carter” rhymes “again” with “brain”, both with the FACE vowel. Rogers was, of course, Canadian.

  57. At the beginning of the 16C, the English-speaking thede accidentally dropped the word in the River Thet, after which it rolled down to the North Sea where it has remained ever since.

    (Actually, the Thet is named for the nearby thetford ‘public ford’ in the town of that name. Thetford is the former capital of the East Anglian kings; it’s on the Little Ouse (< the Celtic reflex of *wed-), the border between Norfolk and Suffolk. Presumably the ford was so called because it was freely open to all.)

  58. Jim Parish says:

    Um. Actually, listening to it again, maybe he does use the DRESS vowel in “again” after all. It’s a little hard to tell, on the recording I have.

  59. marie-lucie says:

    … the Mary Ellen Carter rise again

    I have heard the song many times on Canadian radio (though not recently) and I don’t think the singer rhymes again exactly with train. I think the vowel he uses is intermediate in height and length between that of brain and that of dress. Singing tends to distort speech sounds somewhat, and its rhythm often affects the stress pattern of the wirds uttered and the relationship of stress to length in vowels. In the song again is found in stressed position at the end of a line, and the singer might pronounce the second vowel differently (perhaps at least shorter) in the middle of a line or in the course of his normal speech.

  60. marie-lucie says:

    the wirds : the words, of course

  61. Stefan Holm says:

    Sorry, David, about teuer, steuern and schwäbische. They were all due to sloppiness. I have no problems per se with strong and weak declension of adjectives. They are fully alive and kicking in my native tounge following the same rules as in German. (Actually I copied der schwäbischen Dialekte from an external source, changed genitive ‘der’ to nominative ‘die’ and having done that hastily misinterpreted ‘Dialekte’ as singular).

  62. On the different reflexes of Germanic *eu /iu in German – IIRC, OHG “iu” before high vowels in the next syllable is reflected by /y:/ (written “iu”) in Middle High German, which in Standard Modern High German became the diphthong written “eu” that is discussed above by David. In other environments it became the falling diphthong /iə/ that was then monophthongized to /i:/ in Standard Modern High German, but is still written “ie” and, as David says, it is still a diphthong in many (mostly?) Southern German dialects. Besides that, “eu” is also the result of umlauted MHG /u:/ (in that case it is often written “äu”).

  63. David Marjanović says:

    is still a diphthong in many (mostly?) Southern German dialects

    All of them southern (Alemannic and Bavarian), yes.

    What confuses me is the distinction some Swabian dialects apparently make, where some eu words and some ie words have some kind of [ʊɪ̯]* while the other eu words have merged their diphthong with ei from MHG /iː/.

    * The article says this comes from metathesis of iu, but diphthongization of [yː] seems more likely to me.

  64. The article itself seems a little confused in some places. ;-) But looking at some of the examples, like the cognates of Standard German neu (both nui and nei), makes me wonder whether this is (a) dialect mixture (i.e. in some variants of Swabian the outcome is ui, in others ei, and in the Sawbian koine both results became mixed up, or (b), keeping in mind that the outcome ei is the same as for MHG /i:/, the go back to a MHG /y:/ that had been de-rounded before diphthongization set in, while ui goes back to /y:/ that had not yet been de-rounded. Or you can combine (a) and (b) to assume that some variants of Swabian had de-rounding before diphthongization and others after.

  65. In my view, historical linguists are too quick to shout “Dialect mixture!” whenever they see an unconditioned phonemic split. Sometimes that explanation is right, as in the distinction between French français and danois: we have clear evidence of competing [ɛ] and [wɛ] dialects ([wɛ] > [wɑ] came later), and standardization did freeze a mixed result in place; similarly, the vocalism of English bury and busy do come from well-documented dialects other than the one which led to the standard (and provides the standard spelling). But most of the time, the dialects and their mixture are entirely hypothetical, and in those cases the explanation is ignotum per ignotius: rather than having to explain why some words became ei and some words ui, you instead have to explain why some words were drawn from one dialect and some from another, a demand that generally goes unvoiced, or if voiced, is answered with Schlamperei.

  66. Stefan Holm says:

    Well, in Swedish we have a quite trivial explanation for one such ‘dialect mixture’: When the old declension system collapsed the Geat dialect in former weak masculine nouns kept the -e nominative ending while the Swee dialect did the same with the -a accusative one. So today we have:
    mosse (bog) but mossa (moss)
    flotte (raft) but flotta (navy)
    ande (spirit, ghost) but anda (spirit, breath)
    make (husband) but maka (wife – originally ’one of a pair’)
    Or with identical meaning ände / ända end, timme / timma hour.

    In most cases though only one form survived in written language (maybe on a 50-50 scale). That has simply been traced back to whether the compiling 18th or 19th c. grammarian happened to be a Geat or a Swee. Couldn’t it also in German be a question of person, which word happened to slip into the dictionaries – from Goethe and onwards?

  67. Sure, for Standard German, but we are talking about Swabian, where there was no standardization. Dialect mixture does typically underlie any standard language. It’s the use of the notion where there is no other evidence for dialects and where there is no resultant standard that I object to. “Every word has its own story.”

  68. But you have things like “dialect koine” or “generic dialect X”, and what’s described in the Wikipedia article is if not the former, then the latter. Otherwise, you’d have to go into describing the dialect of each single Swabian village and neighbourhood. At least, I assume, that when the article has several variants for “neu”, that these are not in general free variation for each individual speaker of Swabian, but that these are geopraphical or sociolectal variants. I also don’t like the use of “dialect variation” as linguistic handwavium, but if one has that kinf of Variation in the same word and no clear indications of a sound law causing the different outcomes, that’s at least something to be considered.
    And perhaps someone, somewhere, already has investigated this and has come to a clear and simple solution; I’m just not that interested in Swabian to search for it. ;-)

  69. No notice of my “unvoiced/voice” play on words? ~~ sob ~~

  70. David Marjanović says:

    when the article has several variants for “neu”, that these are not in general free variation for each individual speaker of Swabian, but that these are geopraphical or sociolectal variants

    It says so explicitly somewhere.

    No notice of my “unvoiced/voice” play on words? ~~ sob ~~

    Perhaps because Swabian, being Upper German, lacks voiced plosives or fricatives. :-þ

Speak Your Mind