Saturday, 26 March 2016

Sources of Tolkien’s language-making

A tentative, speculative talk given to the Oxford Tolkien Society, Taruithorn, on 13 November 2015

When I last had the pleasure of addressing this society, my subject was ‘Tolkien and the Aesthetics of Philology’ and my aim was to look at Tolkien’s particular pleasure in language, on which, it seems to me, much of his theory of language was based. As part of this I looked at the languages whose sound gave him particular pleasure and how this was reflected in Tolkien’s invented languages. This required a rather rushed survey of the vocabulary of Elvish and its putative sources in, or influences from, real-world languages. I would like to come back to this subject and look at it in a bit more detail.

There are perhaps two reasons to do this. First, I think most of us are intrigued by the unique magic of Tolkien’s Elvish languages. They do seem literally to cast a kind of spell of aesthetic pleasure. How did he make them do it?

And secondly, there is the possibility of gaining insights into the art of inventing languages for oneself. Tolkien said in A Secret Vice that he believed that inventing languages was a more widespread activity than was generally realized. In fact I have recently heard that the talk was originally entitled A Hobby for the Home. The topic of invented languages has come to the fore with the popularity of The Game of Thrones with its fully developed languages Dothraki and Valyrian. Their developer, David Peterson, has spoken publicly about how he developed them and recently published a book on the subject, The Art of Language Invention.

I’m going to start by looking at the nature of the connection, if there is one, between  Elvish and Indo-European languages.

The Indo-European theory

In the early 1990s, three pioneers of Elvish studies, Christopher Gilson, Carl Hostetter, and Patrick Wynne devoted a series of articles in the magazine Vinyar Tengwar (VT) to the relationship between Tolkien’s invented languages and real-world languages. 

Note: I am aware that a great deal has been written on the subject by them and others since then. I am simply using these articles as a starting-point for presenting my own views, and I am indebted to them for stimulating thought about this topic.

Hostetter and Wynne wrote (VT 17):

The purpose of this column is to examine words and other linguistic features of Tolkien’s Secondary-World languages that have apparent cognates and analogues in the languages of the Primary World.

They were particularly interested in the idea that Tolkien coined Elvish words and roots to be the ancestors of Proto-Indo-European (hereafter PIE) words and roots. In VT 18 they said:

Tolkien intended these Secondary-World words to be the common ancestors of the Primary-World forms.

We need to recognize that although our authors (who I shall not hereafter distinguish, but lump all three together) were amateur linguists, they were well aware of the principles of historical linguistics. They knew that you can’t just seize on any random resemblance between words and posit a historical connection: you have to respect the fact that, barring some kind of regular contextual influence, sound X in the parent language will always become sound Y in the daughter language, and won’t, in some identically shaped words, become sound Z. So in Proto-Indoeuropean, for example, the sounds d, g, as in Latin edo, ager, always become t, k, as in English eat, acre, in Primitive Germanic, and don’t randomly become any other sounds.

In his background writings, Tolkien invents a human language Taliskan [HME V. 179 and following]. Our authors focus on a Taliskan word widris which they connect with the IE root *weid- ‘to see’, from which come, among others, Latin video ‘I see’and English wit.

In later discussion of Taliskan widris our authors suggest a connection between it and Noldorin idher ‘thoughtfulness’. Tolkien’s Etymologies tell us that this comes from an earlier form *idre (the dh sound in Noldorin goes back to Primitive Elvish (PE) d), and ultimately from an Elvish root ID-. From this, on their theory,  it follows PIE d in the root *weid- ‘to see’ is meant by Tolkien to be derived, via Taliskan, from PQ d, i.e. with no change of articulation. Similarly the Elvish root ROD- ‘cave’ is related, they think, to PIE *red- ‘scrape, scratch, gnaw’, once again with d preserved.

The PIE *ter- ‘to bore, pierce’ they link to an Elvish root TER-, TERES- pierce; hence t goes to t with no change. In another place they connect  PIE *pei- ‘to be fat’ with a root in the Qenya Lexicon PIWI which gives a word píwe fatness (this does not seem to appear in the Etymologies). So we can deduce PE p > PIE p, with no change of articulation. Again they connect PIE *ker- ‘to cut’ with  an Elvish root KÁRAK ‘sharp fang, spike, tooth’;  so k becomes k with no change.

Moving to something a bit more complicated, they connect PIE bha- ‘to shine’ with Elvish PHAY- ‘radiate’, Qenya faina ‘emit light’. This suggests that the rather strange PIE sounds bh, dh, and gh should match up with PQ PH, TH, and KH. But this runs them into a contradiction, as later they state emphatically that  Tolkien ‘deliberately intended  [the root] PHIR- to be cognate with Old English firas’, which means ‘mortals, humans’. But Old English f goes back to PIE p, which, on their theory, should also be p in PE, whereas PHIR- ought to become **bhir- in PIE.

Later on again they relate the Elvish base ȜAR ‘have, hold’ Q harya ‘possess’, Noldorin ardh ‘realm’, to PIE gher- ‘grasp, enclose’, with an extension gherdh- ‘grasp, encircle fence round’, from which comes English yard. So here we have PIE gh coming from the PE guttural Ȝ, and not from, say, KH. Meanwhile they state:

It is likely then that Tolkien intended some genetic relationship between Elvish *angai- ‘torment’ and the IE root angh- ‘tight, painfully constricted, painful’

In this case PIE gh would come from an Elvish g (though perhaps the presence of the preceding nasal has some modifying influence).

Damaging to this theory, in my view, is their suggestion that Taliska, the name of the language which is supposedly the ancestor of PIE, contains an element tal- which they connect with Old Norse tali ‘teller’. But this tal- is a Germanic root, going back, as they themselves state, to PIE *del-. It makes no sense to have Taliska wid- giving PIE *wid- at the same time as Taliska tal gives Germanic tal—or even less plausibly, PIE *del-.

They run into the same problem when they say that Tolkien ‘probably intended’ the Elvish root KUY- ‘as the original source of’ IE *gwei- live. We’ve already had  Elvish K giving PIE k: it can’t at the same time give g, let alone a labialized g. A similar problem arises when they connect  the Elvish root TUK- ‘draw, bring’ with PIE *deuk- ‘to lead’. And, lastly, in discussing the famous connection between Black Speech nazg ‘ring’ and Irish nasc, they plausibly relate the latter to a PIE root *ned- ‘bind, tie’, from which English net also comes. But then they link the root *ned-­ to the Elvish roots NAT- ‘lace, weave, tie’,  and NUT- ‘tie, bind’. Here again we have a voiceless stop in Elvish giving rise to a voiced stop in PIE. They confidently say

nazg was a ‘rule-driven development’ from the base NAT-, just as nasc is a rule-driven development from ned-

But I don’t think the middle premise, that ned- is a ‘rule-driven’ development from NAT-, holds water. We simply can’t have both the voiced and voiceless stop series giving rise to voiced stops in PIE while the voiceless stops sometimes give rise to voiceless stops.

Actually I am not out to rubbish their ideas completely. I think that their hearts were in the right place but their heads were too literal. I think that they are probably right when they suggest that Tolkien intended many of the resemblances between Elvish and real-world words. The relationships within the Elvish languages are of course ‘rule-driven’. But in my view the resemblances between the Elvish languages and real-world languages are not.

Tolkien and suggestiveness

What theory can we substitute for the ‘rule-driven’ relationship? I think that we have to go back to the fact that Tolkien led with the imagination. In everything he invented, he started with a vision, a word, a situation that excited his creative imagination; he then brought in his ‘rule-driven’ intellect when a historical explanation was necessary.

In VL 19 our authors say:

It is hardly due to mere ‘effect’ that we, like Frodo, find many of the words of the Eldarin tongues to be ‘vaguely familiar’, since Tolkien carefully constructed his languages to be seen as cognate with the Indo-European tongues.

While I do not agree with the second part of this assertion, I entirely agree that we find many Elvish words to be ‘vaguely familiar’: I think that the ‘vague familiarity’ is an ‘effect’, and a very clever effect. Effects are what Tolkien does, brilliantly. The languages are carefully constructed and they feel somehow cognate to real-world ones, but in reality the degree of resemblance varies widely. For the ‘rule-driven’ theory, which I have shown above does not seem to work well, I would substitute an approach to Tolkien’s linguistic creations based on the ideas of suggestiveness or (in Tolkien’s terminology) fitness. The roots and words he invented had to suggest to him, and to any audience who might eventually experience them, appropriate meanings.

So, looking again at the cases above, I would argue that the ID- root was chosen by Tolkien because it suggests ‘inwardness’. The list of words derived from this roots is Quenya íre ‘desire’, írima ‘desirable’, indo ‘heart, mood’, Noldorin inn ‘inner thought’, idhren ‘thoughtful’, idher ‘thoughtfulness’. These suggest such real-world words as English inner, intimate, image, innate, inkling, and Greek idea ‘idea’, which is actually from the PIE root WID-, and perhaps also idios ‘one’s own’.

Or, taking the root TER-: Quenya tere ‘through’, Noldorin trî ‘through’ are vaguely reminiscent of Latin trans ‘through’. The Elvish root KÁRAK ‘sharp fang, spike, tooth’, seems almost too obviously meant to be suggestive, if not echoic, with its derivatives Quenya karakse ‘jagged hedge of spikes’, karka ‘tooth’, karkane ‘row of tooth’ (which to me suggests English carcanet ‘necklace’), Noldorin carag ‘spike, tooth of rock’ (which is very similar to Welsh carreg ‘rock’), carch ‘tooth, fang’.

Again, the Elvish base ȜAR ‘have, hold’ Q harya ‘possess’, Noldorin ardh ‘realm’ in my view is partly, anyway, devised to suggest certain real-world items. In particular it has a derivative Doriathrin garth ‘realm’, suggesting the Old Norse garðr in Miðgarðr ‘Middle-earth’. 

As regards the angai- ‘torment’ root: in the Etymologies, ANGA- is a root meaning ‘iron’ (Quenya anga, adjective angaina; Noldorin angren – compare Angrenost ‘Isengard’).

The above suggestions may look a little unimpressive in isolation, but I’ll come back to this theme with more evidence later.

Early invented words

I’d like to look now at some of the different kinds of invented word and root in Tolkien’s inventory, starting with the extensive vocabulary of roots and derivative words which we find in Tolkien’s Qenya Lexicon of 1915, published in 1998 in Parma Eldalamberon No. 12. The first group I’d like to look at what I call the ‘q-words’. This is a series of roots and words in the Qenya lexicon which have close resemblances to real-world languages. However, what is interesting about them is that they resemble Germanic q-words, not PIE q-words.

We have already mentioned the Etymologies root KUY-, in the QL KOYO ‘have life’ (Qenya koiva ‘awake’): this resembles PrGermanic *kwiw- (Gothic qius) ‘alive, living’ , rather than the latter’s PIE etymon *gwei- ‘live’. 

QETE: qet- ‘speak, talk’ clearly resembles the Germanic word *kweþan speak, say; although the medial consonants fail to match up, for if qet- had been borrowed  into Proto-Indo-European, **kwet- would have given Germanic **hweþan, whereas if it had been borrowed into Germanic, the reflex should be **kwetan

QIMI, with derivative  qin (qim-) ‘woman’ resembles Gothic qino woman;  however  in the Elvish word, the n is only a secondary development, while in Gothic the /i/ is not original in this word, but comes from Germanic /e/, from PIE /e/.  

qalme death (QALA die),  cf. OE cwealm death. 

QELE- perish, die, cf. OE cwelan die. 

qelu a well, spring (QEL + U), cf. German Quelle source. 

And possibly

qoro- choke (QOŘO and QOSO), cf. Middle English qverkin choke. 

qarda bad (QRŘR or QARA ?), cf. Middle English qued bad.

Why are there so many q- words in Qenya which resemble Germanic /kw-/ words?  The use of the letter q without u for the sound /kw/ in Qenya words is probably derived from the spelling of Gothic. A distinct possibility, therefore, is that that the q- words are carried over from the Gothic-based language which Tolkien was inventing just before the Legendarium came into being and the Elvish languages were conceived. We know that certain Germanic-based items came into Elvish from Gothic; notably, as John Garth has shown,  miruvor, which is based on Gothic midu ‘mead’ and woþeis ‘sweet’. Note that in midu the root vowel i which carries over into Elvish is not the original PrGermanic sound, but was developed in Gothic, just as in qin ‘woman’ already mentioned (the PrGermanic vowel was e).  What I’m saying is that there was no effort here to make the Elvish resemble PrGermanic, as if it were its ancestor; Tolkien retained the actual, recorded, Gothic phonology of each word because, I hypothesize, that was what pleased his ear.

There are several other words, not with initial q, in which Germanic, and probably Gothic specifically, underlies an early Qenya form. One is MATA eat, cf. Gothic mats food (from PrGermanic matiz). The PrGermanic root is usually taken to be the same as that of OE metan measure, and therefore not to do with eating. This suggests again that when it was invented, the concern was with a Gothic-based language rather than a hypothetical base for Indo-European.  Another is  lauke ‘vegetable’; compare  Germanic *lauka- (English leek). A third example is the root TIŘI ‘stick up’ with the derivative tinda ‘spike’. It’s hard not to believe that the latter is a straight adaptation of Old English tind ‘spike’ (which is used in LR in Tindrock), and that the root TIŘI is back-derived from it (in Qenya, intervocalic ř is from the same source as d supported by a previous nasal).

What all these cases have in common is that the Elvish voiceless stop phonemes /p/ /t/ /k/ /kw/ correspond to the equivalent PrGermanic voiceless stops, whereas, if PIE were derived from PQ, these sounds would have become fricatives by the time they reached Germanic (compare Latin pater, English father, Latin tres, English three, Latin cor, English heart, Latin quod, English what). To put it the other way round, if Tolkien had intended to construct Elvish proto-forms for these words they would have had to have /b/ /d/ and  /g/ (*mada, *lauga, *dinda).

I’d now like to look at Elvish inventions where there seems to be something more complicated going on.

First of all, there are cases where the resemblance between the Elvish word and the real-world word is between a derived and developed form in each language. What I mean by this is that where one very ancient language is derived from another very ancient language, you would expect the resemblances to be hidden beneath layers of sound change and general attrition. Whereas when you have forms that are evidently highly developed (with formative suffixes and so on) that resemble each other closely, the implication is that the one is a simple borrowing of the other, and genetic relationship is absent. For example, you need a complex series of changes on both sides of the family tree to explain how Latin cor and English heart, or Greek odonta (accusative) and English tooth are cognates descended from PIE, whereas it’s obvious that English cordial and odontology are straightforward recentish borrowings from Latin and Greek and there is no ancient lineage involved.

So in the Qenya Lexicon we find TALA ‘support’ with a derivative  Qenya talante ‘scales’, which shows a remarkable similarity to Greek talanton ‘balance’: on the hypothesis that PIE derives from PQ we would have the colossal coincidence that a Qenya derivative and a Greek derivative were formed independently in almost exactly the same way, using a suffix –nt-, giving words with very similar meanings. Therefore I do not believe Tolkien was thinking here that the ultimate PIE form from which the Greek word comes had Qenya parentage. 

Another example is Q. enwe ‘name’ and Welsh enw ‘name’. Tolkien must have intended the resemblance, but he of course knew that the PIE ancestor of the Welsh word (related to English name, Latin nomen, etc.) was entirely unlike any possible root of the Qenya word.  I think he was just playfully making an Elvish shaped word out of a Welsh word that he liked by adding an e

A third case is Q. imbe ‘hive’, imbile ‘swarm’, which resembles the OE word ymbe ‘swarm’. Again, the phonetic shape of the OE word implies that whatever its ancestor was, its ancestor probably could not have been the Qenya word or its forerunner. And notice how the Qenya derivative words resemble developed words in a range of languages that Tolkien liked: Greek, Welsh, Old English. 

And finally, here’s an example that takes us out of Indo-European altogether: AWA ‘burn, be parched, yellow, warm’ has a derivative aurinka ‘sunlit, sunny’; compare Finnish aurinko ‘sun’. The simplest explanation for this resemblance is that the Finnish word pleased Tolkien and struck him as manifesting ‘fitness’ of sound to meaning, so he adapted it for Elvish. I don’t think that behind it there lies any sophisticated theory of descent involving a Finnish–Elvish connection.

Philological reverse engineering

Just now, speaking of Qenya tinda ‘spike’, I suggested that the Qenya root was ‘back-formed’ from the ostensible derivative . In several  of the examples we have given so far, I think there is strong evidence of ‘reverse engineering’, i.e. Tolkien starting out by adopting or adapting a word in  a real-world language that he liked, and then generating from it a root and often a set of other derivative words.  Tolkien explicitly admitted to doing this, though perhaps didn’t make it clear to what extent he did it:

You may, for instance, construct a pseudo-historical background and deduce the form you have actually decided on from an antecedent and different form (conceived in outline). [‘A Secret Vice’, in The Monsters and the Critics, p. 211–12]

He also of course refers to deriving words from roots, or at least from older words:

or you can posit certain tendencies of development and see what sort of form this will produce. [ibid.]

But given Tolkien’s love of inventing a story to explain an existing phenomenon  (‘To me a name comes first and the story follows’ (Lett. 165)), it is only to be expected that he would do this kind of reverse engineering within invented languages. 

Here’s a further example, this time involving Old Norse: VANA-, with derivative,  e.g., vane ‘fair, lovely’; proper name Vanar, Vani, meaning ‘the Valar’. It seems hard to doubt that (1) this was suggested originally by Old Norse Vanir,  the name of a tribe of gods to whom the ‘beautiful’ gods Freyr and Freyja belonged, (2) the real-world Vanir were identified with the invented Valar, and (3) because the latter were, of course, beautiful, the van- root acquired the meaning ‘beautiful’.

We can go a bit further with this. By means of this process of reverse engineering Tolkien seems sometimes to have constructed a group of two or even more related Elvish words that resemble  a set of real-world words which are not related to each other at all. Let me give you some examples.

KALA ‘shine golden’: derivatives kalle, kalleva, kalwa may be intended to relate to Greek kalos ‘beautiful’, ‘good’; in any case, the derivative  kalende ‘special day’ resembles Latin kalendae; while kalumet ‘lamp’ looks like English calumet, though that means ‘pipe’ (ultimately from Latin calamus ‘reed’). The semantic imprecision of the latter resemblance (although after all, a pipe and a lamp both contain a glowing fire) makes one feel that Tolkien is more interested in a kind of echo-formation or allusive or suggestive word than in semantic accuracy.

MAHA ‘grasp’: the derivative ‘hand’ suggests Latin manus (though this could be disputed); the derivative  makte ‘hold, power’ strongly suggests Old English mæht, German macht power (Germanic *maht-); the derivative maksima ‘powerful’ suggests Latin maximus; and the derivative  mapalin ‘plane, sycamore’ suggests English maple (Old English mapel-treow).

VARA: the derivative varya ‘different’ suggests Latin varius but the derivative var ‘or’ suggests Finnish vai ‘or’.

A long time ago, Christopher Tolkien pointed out that some of the resemblances between invented words and real-world words seemed almost a joke. There are two examples which he picked out, as being pseudo-explanations that can hardly be taken more seriously than the ridiculous explanation of the word ‘golf’ from Golfimbul in chapter 1 of The Hobbit.

One was SAHA & SAHYA ‘be hot’with derivative sahóra ‘the South’; this seems jocular in its resemblance to Sahara, which in reality comes from Arabic. The other is NENE ‘flow?’, derivative  nénuvar ‘pool of lilies’; (again as Christopher Tolkien comments)  this remarkably resembles English nenuphar ‘waterlily’, which is really from Persian, from Sanskrit nilutpala ‘blue lotus’. In neither case could Tolkien have been trying to make a connection with Primitive Indo-European or indeed Germanic. 

Another example I can add is the PQ root POL-I [glossed ‘?’], with a derivative polenta ‘oatmeal’,  which surely can hardly be meant other than to resemble English polenta from Italian polenta ‘barley porridge’, even though the cereal involved is different. And notice that at least one of these roots continued to be productive in later versions of the Elvish languages: the root NENE underlies Sindarin nen ‘water’, seen in the place names Nen Hithoel and Nenuial.

Etymological gap-filling

To return once more to the Vinyar Tengwar articles, an idea put forward there was that some of Tolkien’s invented roots were designed to explain gaps in real-world etymology. So, for example, in a discussion of TUMPU- ‘hump’, the authors relate this, convincingly indeed, to regional English tump, meaning ‘hillock’ or ‘mound’. They say:

Tolkien was providing his own answer to a linguistic mystery
the similarity..does not mean that Modern English comes from Noldorin. Rather, the similarity is the natural result of these languages descending from a common ancestor.
It is again the last assertion that I would dispute. If English tump is a Germanic word, its PIE etymon would have to be something like *dṃb-, and this is unlikely to have descended from Primitive Qenya TUMPU-. 

I do think Tolkien sometimes decided to patch holes in the etymological tapestry, but light-heartedly rather than seriously. For example, French toucher (English touch) has no recorded Latin etymon; a hypothetical Latin *toccare has been posited as their etymon: Tolkien’s PQ TOKO ‘feel with hand’ looks as if it might be inspired by this. Similarly there is no recorded Latin etymon for French developper (English develop); a hypothetical post-classical Latin *velup- or *volup- has been posited to explain them: Tolkien’s PQ has the root VELU- ‘unroll’, whose derivative velupantie  rather resembles this hypothetical Latin stem. He seems to be saying ‘perhaps the mysterious unrecorded etymon really existed, among the Elves’.

Here’s a really bizarre instance, which you may think fanciful. The English f-word is of uncertain etymology. If it had a PIE etymon, this would be *pug- or *puk-. In the QL we find the root PU(HU) ‘generate’, which has a verbal  derivative  pukta [púke]: could this be a jocular etymon for the f-word?  I only suggest this because, remarkably, the root PU(HU) has another derivative, puntl, against which Tolkien has written, then crossed out, mem. vir., an abbreviation for ‘membrum virile’; and this word has a strong resemblance to Old English pintel, modern English pintle ‘penis’. Perhaps this is absurd, but it’s as if Tolkien were humorously inventing an etymological link between two unexplained (and in truth unconnected) English items from the sexual vocabulary. (And incidentally puhta ‘coitus’ survived into later Quenya.)


In the examples given above, I have suggested resemblances between  Qenya words and real-world words from a number of different languages. These are not isolated cases; many more can be listed from a number of languages. In the case of Latin, I think it was Tolkien’s first port of call for generating new roots. Here’s a list of over two dozen; they may not all be convincing resemblances, but several are to me (and some have already been pointed out by other commentators):

AFA                                         apta- ‘to open’                              aperire ‘to open’
ARA ‘be dry’                                                                                arere ‘to be dry’
KARPAR ‘pluck’                                                                          carpere ‘to pluck’
KNRN                                     kanda- ‘blaze’                              candere ‘to shine’
                                                kandóra ‘bright  dawn’                 candor
KOLO ‘to strain through’                                                             colare ‘to strain’
KOLO                                     kolman ‘peak, summit’                 culmen ‘peak, summit’
KUMU ‘heap up’                                                                          cumulus ‘heap’
TYOSO- ‘cough’                                                                           tussis ‘cough’
LAQA ‘catch’                          lak (laq-) ‘snare’                           laqueus ‘snare’
LARA                                      laru ‘fat, grease’                            laridum ‘fat, grease’
MINI                                        minwa ‘small’ minu- ‘make less’  minus, minuere
OSO                                         oksa ‘joint’                                     axis
OWO                                       oa ‘wool’, ue ‘fleece’                     ovis ‘sheep’
--                                              pekte ‘comb of a cock’                   pecten ‘comb for hair’
PIWI                                         piwe ‘fatness’, pingua ‘fat, rich’   pinguis ‘fat’
QASA- ‘shake’                                                                               quassare ‘shake’
RIPI                                          rípa ‘bank of stream’                    ripa ‘bank of river’
SARA                                       sara ‘saw’                                     serra ‘saw’
SILI                                          siliqa ‘flinty’                                 silex, silic- ‘flint’
TORO                                       toro- ‘bake’                                  torrere ‘scorch’
VIKI                                         viksa ‘sticky’                                viscidus, viscosus ‘sticky’

LAVA                                       lava- ‘lick’, lambe ‘tongue’         lambere ‘lick’
LIQI ‘flow’                                                                                   liquere
NYAŘA ‘relate’                                                                           narrare ‘narrate’
OŘO                                         óre ‘dawn’                                  aurora ‘dawn’
TEŘE                                       teret ‘borer’ teste ‘small worm’  teredo ‘boring mollusc’
URU                                         uru ‘fire’                                     urere ‘burn’

And I think Greek was also drawn on:

KERE-                                    kere, keres ‘earthenware’               keramos ‘pottery’
TYURU- curdle                                                                             tureuo ‘make cheese’
                                                tyur ‘cheese’                                  turos ‘cheese’
MALA ‘crush’                        malaqa ‘soft’                                 malakos ‘soft’
NEHE                                     nekte ‘honey’                                 nektar ‘nectar’
NEME                                     neme- ‘I sew’, nemba ‘thread’      neo ‘I spin’, nema ‘thread’
OLO ‘tip’                                óleme ‘elbow’                               ólemé ‘elbow’
--                                              pelekko ‘axe’                                pelekus ‘axe’
PIQI                                         píqa ‘bitter’                                  pikros ‘bitter’
PURU ‘consume by fire’         pur ‘a fire’                                    pur ‘fire’
SALA-                                     salme ‘harp-playing’                    psallo, psalma
SRKR                                       sarko ‘flesh’                                sarx, sark- ‘flesh’

ELE ‘drive, push’                     elin ‘I drive’                                elauno ‘I drive’
KAYA ‘lie, rest, dwell’                                                                 keimai ‘I lie’
KEME ‘soil’                             kemi ‘earth’, kemen ‘soil’           khamai ‘on the earth’
LO’O                                        lóte ‘flower’                                lotos ‘lotus’
MAKA-                                    makil ‘sword’                              makhein ‘fight’, makhaira ‘sword’
MELE                                       mel- ‘to love’                              mele ‘my dear’
MI                                             mir ‘one’                                     mia ‘one’
NERE                                        ner ‘man, husband’                    aner ‘man, husband’
NO- be born                              nore ‘native land’, nosta ‘birth’ nostos return home
TEL + U                                    telu ‘end’                                   telos ‘end’
TURU ‘am strong’                    túranu ‘king’                             turannos ‘absolute monarch’

As was Finnish:

KANTAN ‘play harp’                 kantele ‘harping’                       kantele (stringed instrument)
KAVA                                         kava- ‘dig’                                 kaivaa ‘dig’
PO- [i.e. póme ‘north’]               póya (not glossed)                     pohja, pohjola ‘north’
SOVO, SOWO wash                  saune ‘bath’                               sauna ‘sauna’
TADA                                         tar ‘fence’, tarwa ‘enclosure’    tarha ‘enclosure’
TUN + TƏ                                  tunto ‘quickness of perception’  tunto ‘sense, feeling’

ANA                                         anta- ‘gives’                                antaa ‘give’
KULU ‘gold’                                                                                kulta ‘gold’
PIKI ‘thin, little, small’                                                                pikku ‘little’
TEHE pull?                               tie ‘road’                                     tie ‘road, path’
TULU-                                      tulu- ‘come’                                tulla ‘come’
TUPU-                                      tupu- ‘roof’                                 tupa ‘cottage’

Old English, in my opinion, contributed to a number of items, several of which I've already mentioned:

Old English
--                                                ea, earen ‘eagle’                             earn ‘eagle’
LESE ‘collect’                                                                                  lesan ‘collect’
MAKA-                                     makse ‘net’                                     max ‘net’
SIQI ‘sigh’                                                                                        sícan ‘sigh’
TUŘU & TUSO ‘kindle’          tunda- ‘kindle’, tundo ‘firewood’   tendan ‘kindle’, tynder ‘tinder’
GWELE                                    wele- ‘it boils, bubbles’                   weallan ‘boil’
                                                  welwe, welme                                  wielm ‘boiling’, wiell ‘spring’
TIŘI  ‘I stick up’                       tinda ‘spike’                                   tind ‘spike’

HERE ‘rule’                              heru ‘lord’                                      hearra  (or Old Saxon herro)
IBI ‘to swarm’                           imbe ‘hive’, imbile ‘swarm’          ymbe ‘swarm’
MAWA- ‘cry, bleat’                   máwe ‘gull’                                   mǽw ‘gull’
QALA ‘die’                               qalme ‘death’                                cwealm ‘death’
QELE- ‘perish’                          qelet ‘corpse’                                cwelan ‘die’ (cf. Fr. squelette?)

And a number of other items may owe something to Gothic, apart from those already discussed:

--                                                marma ‘sand’                              malma ‘sand’
MAYA                                       mai, máye ‘too much’                 mais ‘more’
MIMI                                        mimbe ‘flesh’                               mims ‘flesh’
SITI, SIT-U                               sito ‘habit, custom’                     sidus ‘habit’
ULUN(T) (ulump)                    ulun(t), (-mp) ‘camel’                 ulbandus ‘camel’

MULU ‘grind fine’                   mul(d) ‘fine powder’                   mulda ‘dust’
TULUK-                                   Q. tulunka ‘steady, firm’    
                                                  Gnomish tulg-                             tulgus ‘firm’
UL-                                           ulban(d-) ‘monster’                     ulbandus ‘camel’

If I am on the right track, in most of these cases Tolkien was borrowing elements from real-world languages because they seemed to him to have an appropriate or fitting sound, or because, to him at least, they recalled the original word enough to suggest the meaning he wanted them to have. However, I think that he also coined lexical items that suggested or alluded to their meaning in a much looser way than this. The suggestiveness or allusiveness is therefore more subjective, and other people can certainly challenge it. But here are some examples that work for me:

AWA                                        avar ‘wealth’, avarna ‘very rich’            avarice
FILI-                                         filma ‘fine hair, line’                               film, filament, file
HUNTO                                   Gnomish funt ‘elephant’                        (ele)phant
KAWA ‘stoop’                         kaure ‘timidity’                                      cower
--                                              kitya ‘to tickle’                                       (regional) kittle
KLTL-                                      kilt ‘a tuck’                                            to kilt up
TYETE- ‘give suck’                tyetse ‘a teat’                                         teat, tit
LAPA ‘enfold’                                                                                        to lap
--                                               lattin, lattulis ‘window’                        lattice
LI + ya                                      liante ‘tendril’                                      liana
LINI-                                         lint ‘fluff’                                             lint
MUKU  ‘cacare’                                                                                    muck
NRRŘ                                       nar(d) ‘fragrance’                                nard
OYO- ‘salve, rub, oil’               oinalis ‘ointment’                               ointment, anoint
POKO ‘tuck away’                    poko ‘bag’                                          (archaic)  poke
--                                                senna ‘red-brown’                               sienna ‘reddish-brown paint’

AVA- ‘go away, depart, leave’    avande!  ‘get hence!’                     (archaic English) avaunt
FANA-                                         fanta- ‘swoon’,                              phantom, phantasy
                                                    fantl ‘vision, dream’
KASA                                         kasien, kasqar(in)  ‘helm(et)’          casque (a kind of helmet)
NOHO                                         nóla ‘head, hill’                               noll ‘crown of head’,  knoll
TLTL                                           tilt- ‘make slope’                             to tilt

Here are some oddments that seem to relate to languages we don’t normally associate with Tolkien:

ENU ‘God Almighty’                                                                      Semitic En
MANGA ‘lacking’                   manka ‘few’                                   French manquer ‘lack’
--                                               lapatte ‘rabbit’, lapatl ‘leveret’      French lapin
--                                               rue ‘rest, stillness’                          German Ruhe ‘rest’
--                                              veliki- ‘great’                                   Russian velikiy ‘great’
--                                               pe ‘mouth’                                      Hebrew pe ‘mouth’
 (U)NQ(U)N                             ank ‘loop’                                       Ancient Egyptian ankh

And here, finally, are the items that, as argued above, seem to reflect unrelated words in several real-world languages.

MILI                                         milin ‘grain of seed’                 Latin milium ‘millet’
                                                  milt ‘semen’                              milt ‘soft roe’
NOSO-, NOTO-                       nos ‘wetness’                             German naß ‘wet’
                                                 noswe ‘wet wind (SW)’             Greek notos ‘south-west  wind’

ALA                                         alda ‘tree’, alalme ‘elm’               elm, Latin ulmus, perhaps alder
--                                              alqa ‘swan’, Gnomish alfa            Welsh alarch, ON alpt
--                                              atta, Atar ‘father’                          Gothic atta, Irish athair
KALA ‘shine golden’              kalende ‘special day’                     Latin kalendae
                                                 kalumet ‘lamp’                               calumet ‘pipe’
                                                 kalle, kalleva, kalwa                      Greek kallos ‘beautiful’
KRN (Gnomish grintha)         karne ‘red’                                     carmine, carnage, L. carn- ‘flesh’
MAHA ‘grasp’                         ‘hand’                                       Latin manus
                                                makte ‘power’                                OE meaht, Gothic maht-
                                                maksima ‘powerful’                       Latin maximus
                                                mapalin ‘plane, sycamore’            maple
MANA                                    mane ‘good’,  Manwë                   Latin Manes, Algonquian manitou,
                                                                                                      Austronesian mana ‘power’
MORO-                                   morna ‘black’                               Moor, mourn, Gk mauros
MULU ‘grind fine’                 mul(d) ‘fine powder’                     Gothic mulda ‘dust’
                                                mulmin ‘mill’                                Latin molina, OE mylen
OŘO                                        oro- ‘rise’, oro ‘hill’                     L. orire ‘raise’, Gk oros ‘mountain’
TALA (support)                       talas ‘sole’                                   L. talus ‘ankle-bone’ (talon)
                                                 tala- ‘bring’                                 L. tollere ‘bear  away’
                                                 talante ‘scales’                            Gk talanton ‘balance’
VEŘE                                      veru ‘husband’                             OE wer ‘man, husband’
                                                vesta ‘state of marriage’               L. Vesta ‘goddess of household’

The case of orthanc

I’d like to say more on the subject of how the sound or shape of a word fascinated Tolkien, sometimes for years, inciting him to do something with it. Elsewhere [Peter Gilliver, Edmund Weiner, and Jeremy Marshall, ‘The Word as Leaf: Perspectives on Tolkien as Lexicographer and Philologist’ in Stratford Caldecott and Thomas Honegger (editors), Tolkien’s The Lord of the Rings: Sources of Inspiration (Zurich and Jena: Walking Tree Publishers, 2008), 75] I have put forward the idea that a number of Old English words which can all be found between pages 288 and 307 in Wright’s Anglo-Saxon Grammar (1908) lingered in Tolkien’s consciousness, from the time when he encountered them as a student or earlier, demanding to be used in some way, and were all eventually exploited in LR. This list includes the word orthanc ‘skill, intelligence’. It was not until the writing of LR that the other words all ‘had something done about them’. But orthanc was clearly a sequence of sounds that Tolkien liked so much that he used it, in complete disregard of its true meaning, in early Gnomish. In the vocabulary in Parma Eldalamberon 13 164/2 we find 

orthanc without break or cleavage, united, continuous, unbroken, etc.  pl. oerthainc, erthainc. [cf. ENF privative prefix ur- ‘without, -less; thanc ‘split, cloven, forked’ ENF also has the homonym orthanc ‘masked’, listed under ur-.] 

and later,  in Parma Eldalamberon 13 156/1 
orthanc masked, pl. oerthainc = idanc
Here the only thing Tolkien has kept from the Old English base word is its formation from a prefix or- or ur-. This is very important evidence for the way words appealed to Tolkien. He might use them in an authentically etymological way, as he later did when he said that Orthanc, in the language of Rohan, meant ‘cunning mind’(though even that invested it with a dimension of meaning that it doesn’t have in Old English). Or, he might make an etymological pun, as when he took English waybread and made it the name of a kind of food instead of that of a plant, by reinterpreting its second element. Or thirdly, he might simply use the attractive shape of the word in a new way, as he originally did with orthanc. (Though both these earlier meanings, ‘without break’ and ‘masked’, attributed to Elvish orthanc, have a curious appropriateness to the later conception of the tower and its occupant.)

The case of leminkainen

I should also like to consider the curious case of the entry in the Qenya lexicon for Lemin ‘five’ (HME I. 246). It reads: Lemin ‘five’ Lempe ‘ten’ Leminkainen ‘23’. Christopher Tolkien here comments ‘The choice of “23” suggests that this was my father’s age at the time’. But he omits to say, presumably because it was so obvious, that this word closely resembles the name of the Finnish hero Lemminkäinen. If this is not a coincidence, then the process of thought must be from identification on Tolkien’s part with the hero and the transfer of a form of his name to the number of Tolkien’s age at the time, from which the numbers were ‘back-formed’.  This doesn’t fully resolve the puzzle, for if lemin is ‘five’, what can kainen be? Later (in the ‘Etymologies’), when the number system is definitely decimal, kainen is ‘ten’ (Qenya lempe is then ‘five’). One could pursue the various possibilities of a counting system based on six, but that’s not important for the present argument. The main point I’m making is that Tolkien seems to have started with a name whose sound he liked and gave it a meaning and morphology that has nothing whatever  to do with its meaning in the real world.

Arising from this one might observe that Elvish numerals do not resemble Indo-European ones, and can’t be made to look as if they might be their remote ancestors. Numerals tend to be very stable down the ages, rarely being replaced by loanwords or new coinages, so if Tolkien had intended Elvish to be the ancestor of Proto-Indo-European, you would have expected him to construct his Elvish numerals in such a way that they might have developed into the Indo-European ones.

The Etymologies and afterwards

I’d now like to look at what had happened to this vast abundance of roots and derivatives  by the time of the Etymologies, around 1937–8, which is when the Elvish languages had to begin to crystallize into a stable form for use in LoR.

I haven’t got exact figures, so this all may be thought to be rather impressionistic. But I think that we can gain an insight which more precise figures would not greatly change. I have a list of about 220 roots from the Qenya Lexicon which, to me (and this is again admittedly subjective) look as if they were based in some way on real-world words. I have cited  these extensively in the lists I’ve already provided above. Of these, about 125–130 seem not to be reflected in any way in the later Elvish of the Etymologies. They have been dropped. (These are in the lists headed ‘I.’) Only about 60–65 of them are continued in some form in the Etymologies. (These are in the lists headed ‘II.’; the discrepancy from the overall total results from uncertainty about various items.) Of the first and larger group, some have been replaced, while others are semantic items that just don’t receive a lexical form in later Elvish. It means that a great many of the items closest to real-world words have been removed, for example kava ‘dig’, malaqa ‘soft’, maxe ‘net’, poko ‘bag’, pur ‘fire’.

So two-thirds have gone and one-third have remained. But what is very interesting is that many of these are central items, roots whose derivatives appear in the various quoted pieces of Elvish and in names, in LR and the later writings. Here is a conflation of my lists of Elvish roots and words that may be based on real-world words, with their later representation in the Etymologies and in Lord of the Rings Elvish.

ALA ‘plant,grow’             alda ‘tree’, alalme ‘elm’          now ÁLAM-: Q. alalme, S.? alve ‘elm’
--                                        alqa ‘swan’, Gnomish alfa      now ALAK-: Q. alqua, S. alph ‘swan’
ANA                                  anta- ‘gives’                           ANA(1) ‘give’: Q. anta- ‘give’, S. ónen ‘I gave’
--                                        atta , Atar ‘father’                  ATA-: Q. atar ‘father’
AVA- ‘go away, depart’     avande!  ‘get hence!’             AWA-: Q. av- ‘to depart’
ELE ‘drive, push’              elin ‘I drive’                          (perhaps)  ELED- ‘go, depart’
FANA-                               fanta- ‘swoon’,                      now SPAN- ‘cloud’: Q. fanya
                                          fantl ‘vision, dream’
HERE ‘rule’                      heru ‘lord’                              KHER- ‘rule’: Q. heru ‘lord’, S. hir
IBI ‘to swarm’                  imbe ‘hive’, imbile ‘swarm’    Q. umba ‘swarm’
KALA ‘shine golden’       kalende ‘special day’              KAL- ‘shine’ Q. cal(a)- ‘shine’, cala ‘light’
                                          kalumet ‘lamp’                        calma ‘lamp’
                                          kalle, kalleva, kalwa    
KASA                               kasien, kasqar ‘helm(et)’         KAS-: cár ‘head’,  cassa  ‘helmet’
KAYA ‘lie, rest, dwell’     kaime ‘dwelling’                      KAY-: caita ‘lie down’, kaima ‘bed’
KEME ‘soil’                     kemi ‘earth’, kemen ‘soil’         KEM-: kén (kemen) ‘soil, earth’,
KRN (Gnomish grintha)   karne ‘red’                                KARAN- red: Q carnë ‘red’,  S. caran
KULU ‘gold’                                                                      KUL- ‘gold’  > ‘golden-red’
LAVA ‘lick’                      lambe ‘tongue’                          LAB- ‘lick’, lavin ‘I lick’ lamba ‘tongue’
LIQI ‘flow’                                                                         LINKWI , linqe ‘wet’
LO’O                                lóte ‘flower’                              LOT(H) ‘flower’, Q. lóte, S. loth
MAHA ‘grasp’                  ‘hand’                                 now MAȜ- hand: Q.
MAKA-                            makil ‘sword’                           MAK- ‘sword; fight’ , makil  ‘sword’, S. megil
MANA                             mane ‘good’                             MAN- ‘holy spirit’ manu ‘departed spirit’
MAWA- ‘cry, bleat’         máwe ‘gull’                              maiwë ‘gull’
MELE                              mel- ‘to love’                           MEL- love: Q. mel- ‘love’, melda; S. mellon
MI                                    mir ‘one’                                  MINI- ‘stand alone, stick out’ mine ‘one’
MORO-                           morna ‘black’                           MOR- ‘black’, morna ‘gloomy, sombre’
MULU ‘grind fine’         mul(d) ‘fine powder’                 Q. mulë ‘meal, grist’, mulma ‘fine flour’
NERE                              ner ‘man, husband’                   NER- Q stem for PQ der- ‘man’; Q. nér
NO- be born                    nore ‘native land’                      NO- ‘beget’, nóre ‘country’
NOHO                             nóla ‘head, hill’                         NDOL : Q. nóla, S dol
NYAŘA ‘relate’                                                                 NAR- (2), nyar- ‘tell’
OŘO                                 oro- ‘rise’, oro ‘hill’                 ORO-, oron ‘mountain’, orta- ‘raise’; S. orod
OŘO                                óre ‘dawn’                                  now órë ‘rising’, anarórë ‘sunrise’
--                                      pe ‘mouth’                                 now PEG ‘mouth’ > Q. ‘lip’, peu ‘mouth’
PIKI ‘thin, little, small’                                                      PIK-: pikina ‘tiny’
QALA ‘die’                     qalme ‘death’                             KWAL- ‘die in pain’, qalme ‘agony’
QELE- ‘perish’                qelet ‘corpse’                             KWEL- ‘fade, wither’, kwelett- ‘corpse’
TALA (support)               talas ‘sole’                                 now TAL- ‘foot’, now tallune ‘sole’
TEHE pull?                      tie ‘road’                                   TEȜ- ‘line, direction’, tie ‘path, course,..way’
TEL + U                          telu ‘end’                                   tella ‘hindmost’ telle ‘rear’; Q. tele- ‘finish, end’
TEŘE                              teret ‘borer’                                TER-, TERES- ‘pierce’, tereva ‘fine’, ter ‘through’
TLTL                               tilt- ‘make slope’                        now TALAT- ‘to slope, lean’, talta ‘to slope’
TULU-                            tulu- ‘come’                               TUL- ‘come, approach’,  tulin ‘I come’
TULUK-                         Q. tulunka ‘steady, firm’            TULUK-, tulka ‘firm’
TUPU-                            tupu- ‘roof’                                TUP-; later top- ‘cover’, tópa ‘roof’; untúpa ‘covers’
TURU ‘am strong’         túranu ‘king’                              TUR- ‘power, control’: Q. tur- ‘govern’
URU                               uru ‘fire’                                     UR- ‘be hot’ úr ‘fire’       
UL-                                 ulban(d-) ‘monster’                    *ulgundo, Q ulundo ‘monster’
     root replaced by UGU- negative, ULUG-
(U)NQ(U)N                    ank ‘loop’                                   now UNUK-?, unqe ‘hollow’ etc.
VEŘE                             veru ‘husband’                            now BES- ‘wed’ / WED- ‘bind’, veru ‘married pair’
                                       vesta ‘state of marriage’              vesta ‘contract’

It is noticeable that a  number of these items have been modified. Some roots have been reshaped according to a different concept, so that many more of them end with a consonant rather than a vowel. More importantly, the meaning of several roots has been sharpened, for example from ‘flow’ to ‘wet’, from ‘perish’ to ‘fade or wither’, from ‘support’ to ‘foot’, from ‘am strong’ to ‘power, control’, or precise meaning has been assigned where it was not stated before. There have been some shifts, mainly in a more concrete direction, in the meaning of the whole group, such as the MANA group from a nebulous ‘goodness’ meaning to a specific ‘spirit’ one. And very significantly, most of the very blatant resemblances to real-world words have gone, for example kalende, kalumet, túranu, tilt, and ank.

But another important thing to note is that many of these formations have been slightly changed over time in such a way that the original real-world link has become opaque. Here are some examples: , which in early Qenya meant ‘mouth’, and strongly recalls the homonymic Hebrew synonym, is changed in later Quenya to mean ‘lip’, with a dual peu ‘the two lips, the mouth-opening’. The Finnish-like aurinka ‘sunlit’ disappears, but aurë ‘sunlight, day’ remains. The application of Vanar or Vani to the Valar is not maintained, but the root van- is kept, and is very productive, e.g. in the name of the Valië Vána, and in vanima ‘beautiful’. The Qenya velikë ‘great’, which closely resembles the Russian equivalent, is dropped, and the later root BEL-, which gives Noldorin beleg ‘great’, is said to be ‘not found in Quenya’ (HME V p. 352).

In the Etymologies, it is noticeable how the semantic and phonological development of the roots have together become more sophisticated. For example: NIB- ‘face, front’:—Noldorin nivra- ‘to face, go forward’, nivon ‘west’. Or TEK- ‘make a mark, write or draw’:— Quenya teke ‘writes’, tehta ‘a mark in writing’, tekil ‘pen’, tengwa ‘letter’, tengwe ‘writing’, tengwesta ‘grammar’, tenkele ‘writing system’tekko ‘stroke of pen.

Essentially, in the intervening period Tolkien has been crafting his invented languages further and further away from direct dependence on real-world ones. But this process must be seen as part and parcel of a wider improvement, as he refined both the grammatical structure and the underlying etymological development. In fact as Gnomish changed to Noldorin, he increased the similarity between the way Noldorin developed from primitive Elvish and the way Welsh developed from Old Brittonic, but this only served to increase a feeling of the authenticity of Noldorin. By contrast the philological development of Quenya becomes quite idiosyncratic, though wholly convincing.

Having said that, I would also maintain that Tolkien continued during this later period to coin roots and derivative words that were suggestive of real-word ones, possibly subtler than the older ones. Here are some examples of the new coinages:

ANGWA- ‘snake’             Q. angui (plural)                           Latin anguis ‘snake’
DEM- ‘sad, gloomy’         Ilkorin dim ‘gloom’                      English dim
DOȜ, DÔ-                         Noldorin dûr ‘dark’                       Welsh du ‘black’
DORON ‘oak’                  Noldorin doron, deren                   Welsh derwen ‘oak’
DUN- ‘dark’                     Doriathrin dunn ‘black’                 OE dunn ‘brownish black’
GÁYA- ‘fear’                   Old Noldorin gaia ‘dread’             Gothic (us)gaisjan ‘frighten’
INK-                                 Noldorin inc ‘guess, idea, notion’  English inkling
KHAG-                            Noldorin hauð ‘mound, grave’       ON haugr ‘grave mound’
LAN- ‘weave’                 W. lanya ‘weave’, lanwa ‘loom’     Latin lana ‘wool’
LIN- ‘pool’                      Noldorin lhîn                                   Welsh llyn ‘lake’
LUG-                               Q.  lunga ‘heavy’                             ON þungr ‘heavy’
MBAKH- ‘exchange’      Noldorin banc                                  English bank?!
ÑGARAM-                      Doriathrin garm ‘wolf’                   ON Garmr hound of underworld
NIB- ‘face’                      Doriathrin nef                                  OE neb ‘face’, ON nef ‘beak’
PAN- ‘fix in place’          Q. pano ‘piece of shaped wood’      English pane, panel
PÁRAK-                          Q. parka ‘dry’, Noldorin parch       English parch
PEN- , PÉNED                Q. pende ‘slope, declivity’              Latin pendere, French pente ‘slope’
PHAY- ‘radiate’               Q. faina- ‘emit light’                       Greek phainein ‘shine’
POTO- ‘animal’s foot’     Noldorin pôd                                   Dutch poot (related to) English paw
RAUTA ‘metal’               Q. rauta (originally ‘copper’)         Finnish rauta ‘iron’
SAM- ‘unite, join’                                                                   ON samna ‘gather, collect’
SIK-                                 Q. sikil ‘dagger’                              Latin sica ‘dagger’ , English sickle
SNAR- ‘tie’                                                                             OE sneara ‘snare’, Dutch snoer ‘cord’
THE- ‘look’                                                                             Greek theaomai ‘see’
THEL-, THELES- ‘sister’                                                       Greek thelus ‘female’
UB- ‘abound’                  Q. úvea                                            Latin uber ‘fertile’
YAG- ‘yawn, gape’         Noldorin ia ‘gulf’                            Latin hiare ‘yawn, gape’

However,  I am very well aware that many elements in Elvish cannot easily be related to anything obvious in the real-world languages that Tolkien knew. There are certain items that we know he felt from his childhood had the meaning he gave them, such as the elements lint-‘swift’ and gond- ‘stone’. But take, for example, the oath Et Earello Endorenna utúlien. Sinome maruvan ar hildinyar tenn’ambar-metta. How many of these words suggest anything in a language Tolkien would have known? We can account for the dor of Endorenna and the tul of utúlien. Et perhaps suggests Latin ex and English out. Ear, it has been suggested, owes its shape to Old English ear ‘sea’. The si of Sinome perhaps suggests the deictic s of Old English se, seo ‘that’. But what about ende ‘middle’, mar- ‘abide, dwell’, hilde ‘heir, follower’, ambar ‘world’, and metta ‘end’?

Back to Taliskan

And what about Taliskan? 

Taliskan is said in the Lammasethen to be “of Quendian origin”, learned by the forefathers of the Western Men from Danian Elves east of Eredlindon.
In B the statements concerning Taliska are not perfectly clear: the Western Men “learned of the Danas, or Green-elves”, and their language was “greatly influenced by the Green-elves”. In the third Tree of Tongues Taliskan is shown as deriving directly from Danian; cf. the addition to Lhammas A…: “But Taliska seems to have been derived largely from Danian.” (HME V. 197)

The Etymologies, in the same volume of HME, come from the same period of development. There are not many Danian words there, about ten. It’s striking that two of them closely resemble Old English words, namely Danian beorn ‘man’: compare Old English beorn ‘warrior’, and dunna ‘dark’: compare Old English dunn ‘dark brown’. The others do not manifest a close philological link with any real-world language, which would imply direct influence. Here’s the list:

ealc ‘swan’, alm ‘elm’, hrassa ‘precipice’, cogn ‘bow’, cwenda ‘elf’, meord ‘fine rain’, urc ‘orc’ (plural yrc), and swarn ‘perverse’.

But what one does notice is a close morphological and phonological resemblance to Old English. Three of these words ealc, beorn, and meord, manifest the ‘breaking’ of a short vowel, and meord also shows a rare sound-change occurring in an identically spelt Old English word, its root, like the Old English word, being mizd-. Swarn and cwenda with consonant plus /w/ clusters, cogn with final gn, and hrassa with initial hr- also resemble Old English.

It seems very likely that the suggestion, made by Gilson, Hostetter, and Wynne—that Taliska itself is formed from the Germanic root tal­- seen in tell, tale, etc., plus the suffix –isk-, used to form ethnic and linguistic names—is correct. If so, within the earlier Legendarium, the Elvish influence was imagined to be on Germanic languages rather than Primitive Indo-European.  It’s even possible that Taliskan was the Gothic-based language which Tolkien was developing just before he switched to inventing Elvish. However, all this is rather academic, since, once he got into the Middle-Earth of LR, Tolkien abandoned the idea of any link between real-world languages and Elvish.


To recapitulate: in inventing lexical items for his Elvish languages, Tolkien seems to have drawn on a wide range of lexical items in real world languages. It is difficult to make a case that the resemblances were intended to represent the direct and internally consistent derivation of Indo-European languages or even Germanic languages as a whole from Elvish. One might suggest that Tolkien at one time intended the resemblances to reflect borrowing by speakers of various human languages from Elvish, but if that was ever the case it was one of his ingenious post facto explanatory mechanisms. Instead what I am suggesting is that Tolkien was primarily guided by the sound of words, and by his idea of the fitness of their sounds to meaning. He attempted to create lexis that fitted its meanings by alluding to, echoing, or suggesting words that had the right kind of fitness; but this can only be argued to be partial. As the Elvish languages developed, they dropped many of the items that most crudely, or even ridiculously, echoed real-world items. But he also invented many other roots that seem to have no basis in the real world.

What can this tell us about language invention? Firstly, that we are bound to make words or roots that suggest to us words in real-world languages with similar meanings. Secondly, that for authenticity we need to distance our inventions from their sources, so that the suggestiveness is veiled. They ring a bell but the audience can’t quite place the bell. Thirdly, that sophistication in vocabulary is profoundly influenced by sophistication in the other linguistic areas: morphology, syntax, and phonology. And fourthly, that within groups of related words and roots we need to ensure that the implied semantic developments, changes, losses, gains, are themselves believable from what we know of real-world semantic change. But all this only matters if you are practising a ‘hobby for the home’.


  1. That was a very interesting read, thank you.

    I like the idea of ‘fitness’, though it may, of course, also open up for some more random grasping at accidental similarities (or should not believe Tolkien's dismissal of any connection between Elvish √ÞAW and Greek σαύρα?).

    A couple of random thoughts that came to me while reading:

    The VAN- may have ceased being applied to the Valar, but it was instead applied to the first tribe of Elves led by Ingwë – the connection here to Yngvi-Freyr, the Vana, seems almost to present itself to me, though I hesitate to suggest it a conscious choice by Tolkien.

    Driving a car in Finland, I noticed helpful signs giving the temperatures of road and air – tie and ilma. I could not help but think of Tolkien's Ilmen, and later, reading Tolkien's notes in PE17 to omentielmo (p.13) making tië a word for path / road, I was forcefully reminded of this again.

  2. I agree with all these points, and I should have thought to bring them in. As soon as I get my hands on the new edition of A Secret Vice, I'll be able to see how close to or wide of the mark I am!

  3. Are you in need of a Loan to pay off your debt and start a new life? You have come to the right place were you can get your loan at a very low interest rate. Interested people/company should please contact us via email for more

  4. "But I don’t think the middle premise, that ned- is a ‘rule-driven’ development from NAT-, holds water."

    I don't either, in that I don't think Tolkien worked out any such rules, or was ever concerned to do so (certainly I've never seen any evidence of it). But then, Pat and I weren't claiming anything quite so strong as this (as I think we made clear in our column). What we did claim (and I still do) is that many of Tolkien's bases/roots do bear more than chance resemblance to IE roots, and (in light of what is said about Taliska)* that this resemblance was intended to be seen as a due to a genetic influence of (at least) Taliska on the later IE languages (as opposed to mere borrowing into IE from Tolkien fictive languages).

    * "Yet other Men there were, it seems, that remained east of Eredlindon, who held to their speech, and from this, closely akin to Taliska, are come after many ages of change languages that live still in the North of the earth." I don't think it's at all controversial to claim that this implies a _genetic_ relationship.

  5. To expand on that a bit: In "external" historical terms, we know that Tolkien associated certain sounds with certain meanings, and that he often drew those associations from the languages he knew/encountered. That fact alone is enough to account for numerous resemblances between his roots/bases and those of especially the IE languages (with which he was most and most widely familiar). What we're saying, though, is that at least at one point Tolkien more than implied that his invented languages were _intended to be seen_ as underlying the IE languages, in a genetic relationship (as opposed to mere borrowings). In "Words and Devices" we use that as launching point and frankly justification to explore the many resemblances to be found in his languages, especially at the level of root/base forms and meanings, with various "real world" languages.

    And we did so for two chief reasons: 1) because _at the time_ (as Tolkien linguists of a certain age will recall) there was a very strong tendency to view Tolkien's languages as completely _de novo_ and having _no_ relationship, intended or otherwise, with those of the "real world" (and to assert the same); and 2) because it was fun and, we thought, instructive, to follow these threads through the "real world" language(s) that likely inspired Tolkien's personal associations.