Sunday, February 10, 2013

Did Tolkien coin the plural “dwarves”?

John Rateliff recently wrote about Roger Zelazny’s conspicuous use of the incorrect plural, dwarves, suggesting that he borrowed this form directly from Tolkien. “‘Elves’”, he says, “[Zelazny] might have gotten from old tradition (or from Dunsany, who influenced everyone who came after), but ‘Dwarves’ is Tolkien’s own invention.” This got me thinking it was time to share some thoughts on this particular plural. (And this, by the way, is not the first time I’ve written about strange plurals.)

First of all, I think we have to consider the possibility that Zelazny did not borrow this from Tolkien at all, but rather formed the incorrect plural on his own, on the same model as calf / calves, wolf / wolves, hoof / hooves, elf / elves, etc. These plural forms are all correct, so it is a natural “mistake” to model the plural of dwarf on the same rule. Some plurals of this type (calves, knives, thieves, lives, wolves) are stubbornly holding onto their original plural forms, but in many cases the correct plurals are being ousted by “normalized” forms, as in hoofs, roofs, turfs.

How did we end up with these seemingly irregular plural forms? The answer is, they aren’t irregular at all; we’ve simply modified their spelling to match their pronunciation. Let’s take a quick look at one example, wolf. The Old English antecedent is wulf, a strong a-stem noun whose nominative plural is written wulfas, but pronounced /wulvɑs/. The letter v was foreign to the Old English alphabet, but it became more and more common after the Norman Conquest, and during the Middle English period, the plural was being spelled both wulfes and wulves, the former gradually giving way to the latter. So, for instance, around the middle of the twelfth century, we see “nu ich eow sende swa swa lamb betwux wulfes” (Luke x.3), but a century later, give or take, we find “suich wulves hit hadde tobrode” (l. 1008, The Owl and the Nightingale, a poem of which Tolkien made a special study of the vocabulary as an undergraduate). And by the fourteenth century, the word was beginning to look almost modern: “Þe wolves dra3eþ uorþ þe children þet byeþ uorkest and wereþ his uram oþre bestes” (“Ayenbite of Inwyt”, MS Arundel 57).

So, as you can see, the plural wolves, spelled with a v is not really so strange. It would be no different, really, if we were to spell the plurals of cat and dog, cats and dogz; after all, the final sound is different. The reason this orthographic change could occur so easily in the wolves class of plurals is that English spelling had not yet ossified, nor would it begin to until after the arrival of the printing press and moveable type.

So, bringing this back to Tolkien, is it possible he merely extrapolated or mistaken a likely plural, as any one of us might have done? Well, it’s certainly less likely, if only because of Tolkien’s philological training. That is to say, he would have known better. He used the form dwarves from the beginning of his mythography, as far back as The Book of Lost Tales (circa 1916–1920). He commented on the use of dwarves in his letters (and elsewhere), noting that the proper plural form ought to have been dwarrows, something he might have learned around 1919–1920 during his work on the Oxford English Dictionary. In the first edition of the OED (then called the NED), this note is offered (edited slightly for ease of reading): “The plural dweorgas became dwerwhes, dwerwes, dwerows, dwarrows […]. Parallel forms appear in barrow, burrow, berry [etc.] from OE beorg ‘hill’, and borough, burrow, bury [etc.] from bur3 ‘town’.” So, even dwarrows is not without precedent, however odd it may look to us today. But if the Old English word was dweorg, how do we explain the final f sound? Again, not so strange. Just consider: how do we pronounce the gh in enough? And British zythophiles still spell it draught, but here in America we now write draft. There’s no mystery about it. It’s a perfectly mundane phonological process.

As pertains to the question with which I framed this post, all I’ve suggested so far is that Zelazny might not have borrowed the spelling dwarves from Tolkien, arriving at it through a natural mistake; and moreover, Tolkien himself might have “invented” it in the same way. But now, consider that Tolkien might have seen this spelling somewhere and been the one to borrow it himself.

Others, it transpires, have used dwarves before, and it’s possible Tolkien knew it. In 1916, right around the time Tolkien would have been starting work on The Book of Lost Tales, the American-Scandinavian Foundation (with Oxford University Press) published a new translation of the Prose Edda. The translator, Arthur Gilchrist Brodeur, uses the plural dwarves consistently throughout, as here: “The dwarves had first received shape and life in the flesh of Ymir, and were then maggots; but by decree of the gods had become conscious with the intelligence of men, and had human shape” (p. 26). This story of the origins of the Norse dwarves obviously sheds some light on Tolkien’s own story in the Quenta Silmarillion, written not so many years after Brodeur’s work first appeared. I don’t know whether Tolkien saw this particular translation, but it was right in his wheelhouse, so I don’t doubt the likelihood. And it arrived on bookshelves right around the time Tolkien first began using the incorrect plural dwarves, an interesting coincidence if nothing else.

Another possible influence appeared much earlier. In 1866, George Webbe Dasent published The Story of Gisli the Outlaw, a translation of the 13th-century Norse Gísla saga Súrssonar. Dwarves — yes, I use Tolkien’s spelling too! — do not play any central role in the saga, but Dasent had cause to mention them once in his book, right at the outset of the saga, in reference to “Graysteel”, a sword that “will bite whatever its blow falls on, be it iron or aught else; nor can its edge be deadened by spells, for it was forged by the Dwarves” (p. 4). As most of you will recall, Tolkien knew Dasent’s work and quoted his famous “bones of the ox” metaphor in the essay “On Fairy-stories”. There seems a reasonable chance Tolkien saw this use of the incorrect plural form, although I admit we don’t know whether a single usage would have been conspicuous enough to catch Tolkien’s eye.

There are other antecedent uses of dwarves too. For example, in F. York Powell’s Old Stories of British History (Longmans, Green, & Co., 1885), a book based loosely on British and Scandinavian mythology and folklore. Google books has scanned a copy from the Bodleian Library, which provides the illustration at the top of this post. I have no basis to assume Tolkien was familiar with this work, but he might have been. For another historical example, though one I doubt Tolkien knew, unless from the OED, see Gilliver et al., The Ring of Words: Tolkien and the Oxford English Dictionary, pp. 104–8.

The upshot is, we don’t know whether Tolkien coined this word or borrowed it. There was a long previous history of its use, and I’ve given two examples Tolkien could very well have known. There are likely others. Even if Tolkien didn’t borrow the spelling deliberately, he might have done so accidentally. Or he may have chosen it for other reasons, quite independent of anybody else. We don’t really know, but — a bit like seeing hobbit in the Denham Tracts — it catches the eye to see the spelling for which Tolkien is often given credit in the earlier Scandinaviana.


  1. If the list you've found is exhaustive, I would say that it was still conceivable that Tolkien had not seen any of these, but I assume that other examples could be added, each of which reduces the probability that Tolkien had never seen this spelling.

    This is slightly different from his coinage of ‘Hobbit’ where we know of only a single prior use in a rather obscure publication (even if about folklore), which, in my estimate, makes it more likely that he had never seen this than that he had.

    I will abstain from giving the formulas for how each known example influences the likelihoods involved (it's simple applications of Bayes' theorem anyway) :-) but merely note that I do not recall any claim by Tolkien to have coined this particular plural form (as he claimed to recall the moment of coining ‘Hobbit’).

  2. I think it would be more historically accurate to say that the plural forms calves, elves, selves, shelves, halves, hooves, knives, lives, wives, leaves, sheaves, loaves, scarves, thieves, wharves, wolves all have irregular singulars, for their immediate relatives invariably have /v/: for lives we have the verb live in all its forms and the adjective live, leaving life as the outsider; similarly with the verbs shelve, halve, thieve (but not the more modern purely denominal verbs knife, wolf). I think this is a complete list for Standard English, since staff and stave have split into separate nouns with plurals staffs and staves.

    Dwarf is as interesting for its rare beginning as its unusual ending: see the comments to this Languagehat post (search on the page for "The term nanism exists"), followed up at this one (search on the page for "I'll try to hijack this one"). Sometimes the comments (and Hat is the first to say so) are more interesting there than the posts.

  3. I think the finding of examples cuts both ways, to some extent. On the one hand, finding plausible examples (as I think at least both Dasent and the Edda-translation are) gives us a specific model that Tolkien might have followed. On the other hand, the more often we see people interested in northern antiquities come up with this plural, the more plausible it becomes that Tolkien also just came up with it. Kind of like how an easily repeatable event is poor evidence as a shared change in historical linguistics.

    One thing that seems interesting in this case is that it's particularly hard to tell. There are enough models that Tolkien might have been following someone else, but Dwarves doesn't seem to have been such an established tradition that we have to assume he did.

    John Cowan, what time depth to do you mean by 'historical'? The voiceless forms are generally the older, and the examples you mention the plural and verbal forms share voicing because both morphological categories provided the conditions for voicing. I'd think that if the singulars are to be regarded as 'irregular', it's really from a synchronic perspective, or at best a historical one dating from the full phonemicization of /v/ and /f/. (The adjective live is a bit more complicated, but it amounts to the same thing: a morphological context providing a following vowel, the condition for medial voicing.)

    1. I grant that at the Indo-European level, where the relevant sound is a stop, it is unvoiced. But as soon as we get to Proto-Germanic, we have a fricative whose voicing is unspecified. It is just as valid to say that the underlying form is voiced and that devoicing occurs under specified conditions (e.g., being initial or in the coda) as the converse.

  4. In "JRR Tolkien An Audio Portrait" there's an audio clip of Tolkien saying something like "Dwarves is really a mistake in grammar, I've tried to cover it up ... I Thought 'dwarf, dwarves' why not." Can anyone verify this? It's been a while since I listened to it. Of course, an interview, probably late in life, is not a very good source to go by. Tolkien could have easily forgotten, or made a mistake speaking off-the-cuff, or simplified/modified what he said to make it easily understandable or sound better.

  5. @Robert: I have never heard J.R.R. Tolkien: An Audio Portrait, so I can’t comment, but perhaps another reader will know. It would be pretty interesting if that is what he said (later in life).

    @John: Your links to LH did not come through. Could you try again? And @Nelson: I’ll leave you two to talk over the other linguistic issues you and John raised, as they aren’t central to my point. But please feel free to continue! :)

    @Troels: Yes, there are certainly other examples, including (I don’t doubt) many that I have not seen. Here are two more that I have seen, both of which are, once again, pretty close to Tolkien. The latter is especially plausible.

    (1) The Poetry of the Old Northern Tongue, from the Earliest Times to the Thirteenth Century. Ed. and trans. Gudbrand Vigfusson and F. York Powell. Vol. 1: Eddic Poetry. Oxford: Clarendon Press, 1883. Dwarves is used consistently throughout.

    (2) A Short Historical English Grammar. Henry Sweet. Oxford: Clarendon Press, 1892. Under “Accidence”, §314: “Nouns in –rf, such as dwarf, scarf, turf, wharf, made this change in Early MnE — dwarves, etc. — but they now generally keep the f in the plural — dwarfs, etc.”

    In addition to Sweet and his work being well known to Tolkien, this passage is actually highly relevant to the whole discussion, since it suggests that dwarves was the normal plural in Early Modern English. I probably should have included this in the main post, but I was unable to track down examples from the EME corpus to support Sweet’s claim quickly enough, so I decided against quoting it there. But Sweet probably doesn’t need my support. I think we can assume he was correct and had examples available. I’ll make a longer search at some point. The problem there is that dwarf is, after all, a relatively uncommon word. At least, before Tolkien. :)

  6. Another two attestations of "dwarves" in works Tolkien possibly or certainly knew:

    Dasent had also used "dwarves" in his own translation The Prose or Younger Edda (1842)

    Wright, English Dialect Dictionary (1889-1905). This in vol. II p. 5 gives "an attack of dwarves" (meaning convulsions) as a dialectal expression of unknown origin, in a "list of words for the present kept back from the want of further information". The source could have been W.G. Black Folk-Medicine (1883) p. 148: "The dog does not bulk so largely in folk-medicine as might have been expected. A cake of the 'thost' of a white hound baked with meal was recommended against the attack of dwarves (convulsions)." This of course brings to memory the Anglo-Saxon charm "against the dwarf". (M.L. Cameron, Anglo-Saxon Medicine (1993) p. 152: "Dweorh has almost always been translated as 'dwarf', which may be its primitive meaning, but there is ample evidence in other Old English medical texts that it also means 'fever', apparently fever accompanied by delirium or convulsive seizures", with examples.)

    All these are in There are many others examples, but I couldn't find any attestation before the 19th century.

  7. @Hlaford: Thanks very much for these additions! Especially for the citation from the EDD, which I checked (s.v. dwarf), but I doubt I would ever have noticed this one. But I think you meant p. v, since it’s in the front matter.

    No surprise to find another example from Dasent. I checked a couple of his other familiar works, without seeing an example. But since he used dwarves in Gisli, I’d have expected the same elsewhere. Nice to have that confirmed.

  8. By the way, I might say that the story is looking more and more like Tolkien probably borrowed this spelling (though maybe unconsciously). We can’t be certain without direct evidence, but the indirect evidence is mounting.

  9. You're absolutely right Jason, it's page v.

    Add to Sweet's Grammar this from Wright's An Elementary Historical New English Grammar (1924) p. 132:

    "Regular plurals are: calves, halves, knives, leaves, lives, loaves, shelves, thieves, wives, wolves, &c. New plurals formed direct from the singular are: beliefs, dwarfs beside earlier NE. dwarves, cliffs, hoofs beside earlier NE. hooves, roofs.

    Maybe one could trace a sort of tradition of writing "dwarves", if not in general usage, at least in relation to Scandinavian material. For example, W.C. Green in Translations from the Icelandic (1908). I think that most of the examples I saw were from books on Norse mythology & literature.

    (Cleasby-Vigfusson's Dictionary has dwarfs throughout.)

  10. I can verify Robert's audio clip. I'm not familiar with J.R.R. Tolkien: An Audio Portrait either, but I found the same interview snippet among others attached to a CD set of the BBC Radio dramatization of The Hobbit, ISBN 978-1-60283-454-5. Here's a transcription, as well as I can make it out.

    Disc 5 of 5, Track 07, 3:59-4:52:

    But, uh, then of course, there’s a tremendous lot of revision. I typed the whole of that work out twice, and lots of it many times. On a bed, in an attic; because I couldn’t afford, the um, the cost of typing. There are some mistakes still. And also, some vital mistakes in grammar! [Laughing] From a professional English language--rather shocking, isn’t it? Well, there’s one where I use bestrode as the past participle [?] of bestride. [Laughs] Well, there’s a lot of things like that, yeah. Yes, there are some. And of course dwarves, originally a mistake in grammar of course, I’ve tried to cover it up, but…but it was just purely the fact that, uh, I have a tendency to, uh, to increase the number of these vestigial plurals in which there’s a change of consonant, like leaf/leaves. My tendency to make more of them than are, than are now standard. And I find I really thought ‘dwarf/dwarves, wharf/wharves, why not?’

  11. Thanks, Alyssa. That’s quite interesting! And as I said in the post, it’s a very natural “mistake” to make. We all have a tendency to do this, especially when we are younger and still mastering our cradle tongue.

  12. Alyssa, thanks for the verification and the long transcription. In "JRR Tolkien An Audio Portrait" the clip is in among narration with no information about it. Does your BBC CD give any information about the recording, like the date?

    1. There is no date or information given in my CD set, but after a little sleuthing in Scull and Hammond’s /Chronology/ and /Reader’s Guide/ I’m going to guess that the clip in question comes from an interview with Tolkien recorded by Irene Slade at 76 Sandfield Road on November 26, 1964, for the BBC program “Reluctant Olympians,” in the series /A World of Sound./ It could possibly also be from one of two interviews of Tolkien by Denys Gueroult for the BBC Sound Archives, the first an informal trial run recorded on the same day as Slade’s, the second a formal interview recorded on January 20, 1965 at the BBC studio in Oxford; or an audio track borrowed from the BBC film /Tolkien in Oxford/, filmed February 5-9, 1968, directed by Leslie Megahey.

      I think Slade fits the best because Scull and Hammond specifically note (II.822-824, “Recordings”) that portions of that interview appear in /J.R.R. Tolkien: An Audio Portrait./ They also state that some clips from /Tolkien in Oxford/ appear in /Audio Portrait/, but the one I could definitively identify in my own CD compilation as coming from /Tolkien in Oxford/ (Tolkien calling death an accident for every man, Scull and Hammond II.611, my CD Track 09 1:22-1:39) has a different sound quality from the dwarf/dwarves clip--scratchier background noise--so I think they come from different sources. My money’s on Slade for the part I transcribed above.

      After listening to a couple of clips of Brian Sibley on YouTube, I’m pretty convinced that my BBC CD set basically has an abbreviated version of Sibley’s /Audio Portrait/ tacked onto the dramatic /Hobbit/ production without any explanation or credit except that it’s a “bonus author interview.” The male voice providing commentary and transitions between the clips of Tolkien talking sounds an awful lot like the Brian Sibley I found on YouTube. So I guess I am (partly) familiar with /Audio Portrait/ after all!

      In reading through Scull and Hammond’s /Chronology/ I found something else interesting, a brief reference to a letter Tolkien wrote to a fan specifically about the word “dwarves,” on July 26, 1955 (Scull and Hammond I.462). It is not among those published in /The Collected Letters of J.R.R. Tolkien./ So if anyone is studying the chronological changes in Tolkien’s own recollection about whether “dwarves” was a mistake or a deliberate strategy, you could write to Scull and Hammond and inquire about the specific content of the 1955 letter. The published /Letters/ seems to show a trajectory in Tolkien’s explanation of his idiosyncratic spelling from “private bad grammar” and “I knew no better” (Letters #17 in 1937 and #25 in 1938) to vigorous defenses of it as an artistic choice (#156 in 1954, #236 in 1961); but if he reverted to calling it a mistake again in the Slade interview in 1964, it might be worth looking at the 1955 letter to see if there’s more evidence of Tolkien wavering.

      Technical question: How do I get italics? Blogspot doesn't want to let me make them, and I hate to keep inflicting my work-arounds on other commenters.

  13. Alyssa, thanks for these additions! I’ll have to inquire about the 1955 letter you mentioned! The italics is easier. You want to use the HTML tags for italics. I’m hoping you know what that means. If not, what you want to do is this: <i>Something in Italics</i> = Something in Italics.

    Give it a try!