Thursday, April 14, 2011

Umlaut and Tolkien

I suppose the first question, for many of you, is what is umlaut? This is a term people like me throw around a lot, often without stopping to consider the confusion among non-philologists. “Non-philologists”, I suppose, is another way of saying, 99.999% of the human race. ;)

Put simply [1], umlaut is a phonological process whereby the pronunciation of a vowel is influenced by the vowel (or semivowel) in the subsequent syllable. This sound change comes in many different flavors, some more common than others. In the Germanic languages, umlaut frequently refers to a more specific sound change where vowels are raised or fronted [2] under the influence of i or j in the following syllable. For these reasons, when speaking of Germanic umlaut, the synonymous terms “i-mutation” and “fronting” may be encountered (you will sometimes also see “palatal umlaut”). This sound change occurred in all the Germanic languages except for Gothic. [3]

As for why one vowel changes under the influence of another, there are two basic views. Randolph Quirk and C.L. Wrenn may have summarized it best: “The generally accepted phonetic explanation […] is that the high front i or j palatalised the preceding consonant and that this in turn pulled the vowel of the stem towards its own position, raising or fronting it. […] This theory may be called ‘mechanistic’, because it is based entirely on the assumed workings of the speech-organs. An alternative explanation is that in pronouncing the back vowel in the root-syllable the speaker unconsciously allows his mind and his tongue to ‘anticipate’ the i or j that is to come in the immediately succeeding syllable, […]. This is a ‘mentalistic’ or psychological theory of i-mutation. The orthodox view of articulatory influence through the consonant is a theory of attraction and assimilation, while the mentalistic view is one of anticipation.” [4]

Since some eyes may be glazing over at that, let me make this a little more plain: i is basically the highest, frontmost vowel there is. It’s so high and so fronted, that it can’t help but pull other vowels toward its point of articulation; not to do so would put a much greater strain on the speech process, and if there’s one sure thing we can say about the speech process, it’s that it’s lazy. It will always take the path of less resistance and least strain on the speech-organs.

Perhaps a few examples would help to make i-mutation clearer. Let’s consider Old English gold “gold”, and observe how the process worked. OE gold was originally *guld (cp. Old Norse gull, and Gothic *gulþ, attested only in the dative singular, gulþa). The suffix used to form the adjective “golden” is still clear in Modern English. We should have expected very early OE *gulden, which mutated by umlaut into gylden “golden”, the u “fronting” into the corresponding short front vowel, y. Subsequently, under the operation of a different sound change, the vowel is the noun, *guld, was lowered, giving us gold. In Modern English, the signs of umlaut in “golden” are long gone, but they were quite clear in OE gylden.

How about another? Think about Modern English “old”, “older”, “oldest”. Do you see where I’m going with this one? In more archaic English, of the type Tolkien often used to represent the speech of Gondor and Rohan, we see the forms, “old”, “elder”, “eldest”. Here umlaut survived into Modern English — for a while. Let’s have a look at the antecedent forms.

In Old English, these were eald, ieldra, ieldest. So, hmm, where’s the i or j we need to account for the fronting of the diphthong ea to ie …? You have to go further back. The original comparative and superlative suffixes in Proto-Germanic were *–izo, *–isto — there’s our vowel! By the time of Primitive West Germanic, the comparative had rhotacized to *–iro [5], while the superlative remained unchanged. By the time of Primitive Old English, this would have given us first eald, *ealdira, *ealdist, which would in turn have mutated by umlaut into the recorded forms, eald, ieldra, ieldest. These are the early West Saxon forms. In the later “classical” West Saxon of around the year 1000, these had eroded into eald, yldra, yldest. In the Mercian dialect (Tolkien’s favorite and mine), the situation looks at once more familiar: Old Mercian ald, eldra, eldest.

If you speak any German, this should look equally familiar, as the Modern German forms are alt, älter, ältesten. As you can see, in German, the vowel experiencing umlaut is still written as the same original letter, but a diacritical mark is placed over it to indicate the umlaut, and the pronunciation is indeed raised or fronted (from a to e).

As the last example shows, umlaut is still very much with us today. It’s typically associated with German, from which the process takes its name (umlaut is um “after” + laut “sound”). A couple more examples from German: Frau, but Fräulein. And schön from Old High German scóni. But though usually thought of in connection with German, it’s still present in English too.

Here’s something I’ve been building up to. Ever wonder why it’s Anglo-Saxon but English? Or why it’s Anglia, but England? Had I begun with this question, you might have been scratching your heads, but now you know the answer: the change from a to e is umlaut! In Old English, the angle were the Angles (as in Angles, Saxons, and Jutes), but the adjectival form of their ethnonym was englisc (originally *anglisc, acted upon by umlaut). And this is where we come to Tolkien. You may have wondered whether I’d live up to that promise, so dense has been the discussion up to now! Hopefully, you’re all still with me.

Tolkien was far more expert than I in matters of Germanic sound laws. He owned books with impressive titles like Laut und Formenlehre Altgermanischen Dialekte [“Sound and Morphology in the Old German Dialects”], and he read them in their original languages and made annotations and corrections in their margins. He was better versed in umlaut than I will ever be and would surely have found plenty to niggle at in my explanations of it. It should come as no surprise, therefore, that he worked examples of umlaut into his fiction (as he did so many other philological elements). I can think of three instances in The Lord of the Rings. If I’ve successfully communicated the basic idea behind umlaut, can you think of any? Pause here to put on your thinking cäp.

Here’s a hint: just like Angle/English, the examples of The Lord of the Rings are topo/ethnonymic.

So. Just as we have Angle, but English, Tolkien gives us Dunland, but Dunlending and Dunlendish, where the i in the final syllable causes the a in the second to be raised to e. This is as straightforward a case of umlaut as you could wish for. Interestingly, the words dún-land “down, hilly land” and dún-lendisc “hilly, mountainous” are attested in Old English, as are uppe-land and up-lendisc, both pairs clearly demonstrating umlaut in the real world. Returning to Middle-earth, another example from Tolkien follows the same pattern: Sunlands, but Sunlending, each used only once in the novel, in reference to the far southern regions of Harad.

And finally, again in connection to Harad and the Sunlands, what about the curious Shire word, Swertings? “Swertings we call ’em in our tales; and they ride on oliphaunts, ’tis said, when they fight,” Sam tells Gollum [6]. This is a little less obvious, but it must be the umlauted form of *Swartings, derived from swarthy, a word Tolkien often uses of the Harad-folk. The word swart or swarthy comes from Old English sweart “black” (Mercian *swart), cp. Old Norse svartr “black” and Modern German schwarz. In Old Norse, there is a proper name Svertingr, which probably carries a swarthy meaning and shows umlaut from svartr; likewise, probably, for the Swerting in Beowulf, though we can’t really say much about him. Also in Old English, swertling was used to gloss the Latin ficedula, a small passerine bird, dun or drab (or swarthy) in color. Today, ficedulae are Old World flycatchers of the order Passeriformes, but Bosworth/Toller supposed that swertling might be the titlark, a bird of the same taxonomic order, but different in family, genus, and species.

There you have it. Both real-world and Middle-earth examples, side by side. Put there, in fact, by one of the most gifted Germanic philologists the world has ever seen. Should we be surprised? Of course not! Is it interesting? Well, I certainly think it is, and I hope you agree. :)


More, and lengthier, notes than usual

[1] As complicated as this must sound to a lay reader, believe me, I have simplified it. The whole process is made more difficult by the fact that the i or j (especially the latter) frequently disappeared by the time the words in question were being set to parchment. Other processes of sound change might subsequently alter the vowels of the stem, inflexions, or both. Inflexions may have been lost entirely. Exceptions may have preserved root vowels where we would have seen umlaut. And so on. But at its simplest: i-mutation is the raising or fronting of a root vowel under the influence of i or j in the following syllable.

[2] Fronting and raising aren’t the same thing, though they’re closely related. Each vowel, like all speech sounds, is articulated at a certain location somewhere in the speech cavity, somewhere from the lips to the glottis (front to back), from the soft palate to the lower jaw (top to bottom). Fronting means that the articulation of a vowel moves from the back of the speech cavity toward the front (e.g., fool to foot to fur); while raising means a vowel moves from the bottom toward the top of the speech cavity (e.g., frond to friend to frill). Try pronouncing these groups of words and pay attention to how your tongue moves inside your mouth: forward with the first group of words, then upward with the second group.

[3] Actually, it might have occurred in Gothic, but two problems: the vast bulk of the Gothic we have is from the 4th century, which predates the umlaut process; and if umlaut did occur in Gothic, we don’t have any later texts that would show evidence of it. One would think it should have occurred in Gothic, and this has occasionally been alleged by scholars, but we have no clear evidence of it in the surviving corpus. By the way, i-mutation isn’t an exclusively Germanic process — there are examples in the Romance languages as well — but it was much, much more prevalent in the Germanic language family than in others.

[4] Quirk, Randolph and C.L. Wrenn. An Old English Grammar. 2nd ed. London: Methuen, 1957, pp. 153–4.

[5] Rhotacism is another sound change in the Germanic family, whereby Proto-Germanic z became a rhotic, or r-like sound. Like i-mutation, this occurred in all the Germanic languages except Gothic (er, maybe; see note 3). Example: PG *deuzom gave Gothic *dius (attested in dative plural diuzam), preserving the z; but this was rhotacized throughout the rest of the family: Old Norse dýr, Old Frisian diar, Old Saxon dior, Old High German tior, Old English déor “wild animal (> deer)”. Why should the z and r sounds be related? Ask Antonín Dvořák!

[6] In draft, it was Gollum, not Sam, who called the Haradrim Swertings.

28 comments:

  1. That explanation should be published! Before today, I didn't really "get" what i-mutation was (and I'd read three different textbook explanations of it). Thank you Jason:)!

    ReplyDelete
  2. Interesting that the 'Dun' in Middle-earth's Dunland refers to the colour when the hill sense is attested in OE, though Dunland was hilly.
    "Dunland and Dunlending are the names that the Rohirrim gave to them, because they were swarthy and dark-haired."
    Was Tolkien making a pun? I guess it's safe to say that he knew of the OE.

    ReplyDelete
  3. Thank you, Lillyput90. I’m glad it made sense. As I worked on this, I kept starting down this or that side road of explanation, elaboration, and exception, then stopping myself with a warning to stay as close to the main point as possible. As you can see, even that cannot be done in few words! If I’d tried to drill down deeper, the post could have been twice as long. :)

    Robert, yes, Tolkien would have known both very well indeed. In OE, it’s one of those substantive differences due to a short versus a long vowel: dun “dun (the color)” versus dún “hill, down, dune”. Another example, and a better known, is god “god” versus gód “good”.

    And I wouldn’t put it past Tolkien to have been punning here. That would have been just like him!

    ReplyDelete
  4. Returning to Middle-earth, another example from Tolkien follows the same pattern: Sunlands, but Sunlending, each used only once in the novel, in reference to the far southern regions of Harad.

    I remember squire noting, as he worked on his article on "The South" in Tolkien, how "Nomenclature of The Lord of the Rings" indicates that "Sunlending" does not in fact refer to Harad, but to Anórien. The name appears in LOTR when a poet of Rohan describes Théoden leading "six thousand spears to Sunlending, / Mundburg the mighty under Mindolluin".

    ReplyDelete
  5. Ah, you’re right; my mistake! Thanks for pointing this out, Brig.

    So, Sunlending is a direct calque of Anórien (with the Sindarin element anor “sun”) and not a reference to Harad. In spite of the error, it is still quite clear that umlaut is in operation in the Englsh (~ Rohirric) word “Sunlending”. It is a bit strange that the Rohirrim would call Anórien Sunlending when the Shire preserved Sunlands, a word clearly related, for Harad. It is also atypical to see the –ing suffix used of a place, as opposed to a people (cp. Eorling, Beorning, Barding).

    ReplyDelete
  6. It is also atypical to see the –ing suffix used of a place, as opposed to a people

    What makes it actually unusual is that it should be combined with -land, because -ing in English toponyms is mostly found in combination with personal names as patronymic or possessive (other than some instances where it means "meadow", which can hardly be the case here) and then turned into a toponym, either by itself or by the addition of yet another suffix. It can also be adjectival, but then one might have expected the construction to be the other way around, as in Stoningland (= Gondor). In RC:541 Sunlending is translated as "sun-land-people" (that is "sunland" + -ing), and all in all this seems to be the best interpretation. Anyway, Sunlending fits the alliterative verse there better than just Sunland.

    Sunlending could also be read (more poetically) as OE "landing-place of the sun", with lending as in Scippelending, Cnarlendding (see Ekwall, English Place-names in -ing, pp. 24-5). But then it wouldn't be a literal translation of Anórien, and one may suppose that Tolkien would have mentioned this in his note.

    [Sunnlendinga-fjórdungr, BTW, was one of the traditional divisions of Iceland, where sunn(r)- means "south". But of course Tolkien in the Nomenclature explicitly denied this could be the intended meaning.]

    ReplyDelete
  7. Hlaford, great comment as usual!

    And yes, OE –ing, in masculine nouns, usually signals “son, people, clan of” or “associated with”, which applies to the other Tolkien examples I have given (e.g., Beornings). In feminine nouns, it’s equivalent to –ung, and has a gerundive function (e.g., bodung “preaching”, rǽding “reading”).

    I checked PE17 for any further comment on Sunlending, but there is nothing more there. In RC, is the offered translation Wayne and Christina’s, or Tolkien’s? One cannot tell, but I think it’s the former. It wouldn’t be people (plural), in any case, would it? Such would be, following Tolkien’s own models elsewhere, either *Sunlendings or *Sunlendingas.

    Thanks very much for pointing out the Icelandic example of this word. I was unaware of it, but thus prompted, I checked my copy of Zoëga’s Old Icelandic Dictionary (which Tolkien himself knew), and found sunn-lendingr “a man from the south of Iceland”, and an adverb, sunnan-lands, “in the south part of the country”. Again, Tolkien says he had “sun”, not “south”, in mind (though these are related in the Germanic languages), but these are still nice finds that further demonstrate the umlaut process. :)

    ReplyDelete
  8. "sun-land-people" is not given in the Nomenclature, which is where the note is taken from. It must be a guess, but with all its difficulties it's the best I can think of. To refer to the people, Sunlendingas might be preferable in that context, since the Rohirrim usually use Eorlingas, Helmingas as opposed to CS Beornings, Entings etc.

    Sunnlendinga-fjórdungr and other derivatives are mentioned in Cleasby-Vigfusson s.v. sunnr, a work Tolkien was well acquainted with. BTW, today I realised that fjórdungr means just "quarter" and it's cognate to Farthing - another toponym with the -ing ending. It doesn't show umlaut today, but it did in ME ferthing. Tolkien stated that he based his "Farthings" on thriding (Nomenclature), but isn't it possible that he was aware of the Icelandic divisions too?

    ReplyDelete
  9. Metaphony is actually quite common in the Romance languages, just not in the standard ones, although Spanish ie ue began as metaphonic forms of VL e o, but now have become universal. Many varieties of Italian mark number by metaphony: see "Metaphony Revisited" by Andrea Calabrese, which contains many useful examples.

    Hlaford: There was a sound change /er/ > /ar/ in Early Modern English that was eventually partially reversed in the 19th century. Ferthing, like sterre and ferme, was respelled to match the new pronunciation, but servant and sermon were not, though we know that in the 18th century they were sarvant and sarmon. A few words retained the er spelling and the /ar/ pronunciation, as sergeant, clerk, derby; the last two were reverted to a spelling pronunciation in American English, though note the related proper names Clark, Darby. Finally, some words split into two, one with er and one with ar, as person/parson, university/varsity.

    ReplyDelete
  10. Metaphony is actually quite common in the Romance languages, just not in the standard ones, although Spanish ie ue began as metaphonic forms of VL e o, but now have become universal. Many varieties of Italian mark number by metaphony: see "Metaphony Revisited" by Andrea Calabrese, which contains many useful examples.

    Hlaford: There was a sound change /er/ > /ar/ in Early Modern English that was eventually partially reversed in the 19th century. Ferthing, sterre, ferme were respelled to match the new pronunciation, but servant, sermon were not, though we know that in the 18th century they were sarvant, sarmon. A few words retained the er spelling and the /ar/ pronunciation, as sergeant, clerk, derby; the last two were reverted to a spelling pronunciation in American English, though not the related proper names Clark, Darby. Finally, some words split into two, one with er and one with ar, as person/parson, university/varsity.

    ReplyDelete
  11. Thank you for these contributions, John. I look forward to reading that essay; Italian is one of my favorites. :)

    Nice points about /er/ > /ar/, of which clerk (still pronounced as clark, and so spelled in the personal name, as you note) is by far the most familiar example. And of course, the word cleric is another reflex of the same root. But the other examples you give are just as interesting. Thanks again.

    ReplyDelete
  12. Some more examples are the pairs shear/sharp, sherd/shard (synonymous), dear/dar(ling), steer/starboard, firm/farm, errant/arrant, and the name of the letter R.

    Indeed, the names of the letters are borrowings from Etruscan (through Latin and French) and are particularly juicy examples of pure sound-change, because they have no standard written forms to serve as a brake — they are maintained by oral tradition in the midst of almost entirely literate Anglophonia. The Etruscan convention was that the vowel letters were given their long sounds, the stops were suffixed with /e/ (except that K and Q were /ka/ and /ku/), whereas the non-stops were prefixed with /e/, thus /be/ but /ef/. French palatalization of velars before front vowels, final vowel lengthening in Middle English, and the Great Vowel Shift did the rest of the work, giving us the familiar /eɪ bi si/.

    When Latin lost /h/, the names of A and H became homonymous, and the new name /akka/ was created for H; this became /atʃə/ in Old French and eventually /eɪtʃ/ in Modern English. Of the new letters, J and V got their names by analogy, W from its shape, and Z by analogy in the U.S. and from zeta elsewhere.

    While I'm at it, here are the other English words thought to be of ultimately Etruscan origin

    ReplyDelete
  13. John, more great examples! I especially like your point about the name of the letter R. And a very good capsule summary of the letter names in general. I very much like how you put it: “no standard written forms to serve as a brake”. By the way, a fun piece of trivia: Q and W are the only two letters whose names in English are spelled without the letter in question: “cue” and “double-u”.

    ReplyDelete
  14. Technically, "double-u" does indeed have the letter in question in its name. It is as the name indicates a DOUBLE U and for the medieval period and well into the age of print, "w" was written uu, or vv since the pointed form we now know as a "v" was just as commonly used as the rounded form. Undoubtedly, one notes the similarity in shape between vv and w, the latter being simply two u letters in the pointed form written with a single pen stroke. Thus, double u does have the letter in question in its name.

    By the way, much Old English particularly early OE used the runic letter wynn to signify the "w" sound.

    ReplyDelete
  15. Sure, that is one way to look at it, Larry. Perhaps I should have said, double-u no longer contains the letter it names, since I don’t think anyone still regards the letter w as a double u. A double v, if anything (as it is named in the Romance languages). At least if we are speaking of the language today.

    Of course, the other way to look at it is this: “d-o-u-b-l-e-u” does not contain the letter w anywhere in it. :)

    ReplyDelete
  16. Well, since there is no standard spelling, you could decide that the name of the letter Q is spelled "queue". :-)

    ReplyDelete
  17. Gothic and Spanish.

    Durmiendo, not dormiendo, unless I misremember. There is no fronting but there is lifting before an i/j in Gothic and there were Visigoths in Spain.

    Connexion, do you think?

    ReplyDelete
  18. Oh, Swedish also has äldre, äldst, and the noun ålder (age rather than "elde"), but the positive is replaced by "gammal". Gamling means "old man" in Swedish as much as in - I suppose - Icelandic

    ReplyDelete
  19. Hans-Georg, yes, it seems plausible that sort of a connection between Spanish and Gothic might have existed. It would take some poking around to try to substantiate it, but it’s a good hypothesis. The Visigothic influence in Spain was on its way out by the end of the 6th century, I believe. How early in Old Spanish do we start seeing these sorts of sound changes?

    ReplyDelete
  20. Seeing? Latin near Classical spelling covered up sound changes (except for misspelling tellers) up to around anno 1000 when it comes to Spain and Italy, just as French sound changes do not show before the Strassburg oaths.

    ReplyDelete
  21. In Sweden and Germany - where umlaut is a regular feature of the grammar (when were "lenger, lengest" eliminated for "longer, longest" in English, if they existed?) - if you have quit school with not too bad notes in grammar, even if you are no philologist, you know what Umlaut/Omljud means.

    However, in German, where every umlauted vowel except i for e or ie for eu (if I got that correct that they are such) is the non-umlauted vowel plus a trema above or an e after, often the word "Umlaut" is simply used for trema.

    "Ä"="A, Umlaut" or "Umlaut-A", and similar for "Ö", "Ü".

    In Sweden this is not so, and the word Omljud is therefore less known. We have Y instead of Ü for the umlaut of U ("ung, yngre"="jung, jünger"), and neighbouring languages have Æ and Ø for Ä and Ö. Also, we have Å, which is not an umlaut at all, but a vowel shift with a long ah going aw (extant long ah's were either short at the time or borrowed after, obviously).

    ReplyDelete
  22. So much for the 99,999% of humanity who are supposed to know nothing of Umlaut: Swedes and Germans would be a bit more than just 0,001% of the 7 billion (Swedes alone are more than 0,1%, and far fewer than Germans, plus the fact that Swedish is talked by other nations too: Finland Swedes, "Visconsin Svedes" ... et c.

    ReplyDelete
  23. In Modern English, the signs of umlaut in “golden” are long gone, but they were quite clear in OE gylden. - from which Modern English verb gild, pp/adjective gilded? gilt? when paper thin gold is added on top of other material. Right?

    ReplyDelete
  24. Hello again, Hans-Georg. I see your line of thinking with Modern English gild; however, it’s not really an example of umlaut in Modern English — certainly not directly. The word came to Modern English from Middle English gilden, in turn from Old English gyldan “to gild”, back formation from the adjective gylden “golden”. The latter exhibits umlaut, but the new verb gyldan does not. It inherits its vowel from the adjective. Middle and then Modern English, likewise. And notice that the adjective in Modern English lost the signs of umlaut.

    ReplyDelete
  25. Ah, if you do distinguish "exhibiting umlaut" from "inheriting a vowel from another word that exhibits umlaut" ... which is a moot point, I do not. But then again, I am more into diachronics than synchronics, if you like.

    ReplyDelete
  26. I wonder if "golden" comes directly from "gylden" or from a newer formation from "gold" or from Flemish "gulden" (if that is without umlaut, my Dutch is lousy).

    ReplyDelete