Similarities between natural languages and Tolkien’s Eldarin

Roman Rausch

Sep. 24th 2005

’Well, I’m a philologist,’ said Lowdham, ’which means a misunderstood man.’
SD:233
A chance-meeting, as we say in Middle-earth.
UT:326

nanwen

Contents

Introduction

It is […] idle to compare chance-similarities between names made from ’Elvish tongues’ and words in exterior ’real’ languages (Let:297)

This is what Tolkien wrote in 1967, answering to a complete misinterpretation of The Lord of the Rings. And yet it is obviously not that idle because questions concerning these similarities do not cease to occur today and people are always trying to compare existing words to Elvish forms. In most cases, however, there is no relation at all and if there actually is one, it is usually completely different from what people think of it at the outset.

Thus the aim of this article is to show many interesting similarities of this kind which have been found so far, just for the sake of curiosity; to settle the misunderstandings concerning them; and finally it is an attempt to explain them in the context of Tolkien’s legendarium as far as it can be done – whether a particular case is pure coincidence, some kind of influence, or intention.

1  Of similarities between languages in general

At first: Chance similarities between unrelated languages are likely. But how likely? To get a satisfying answer we must apply some mathematics right from the beginning.

According to different estimates there are 2500-6500 different languages in the world, though the vast majority of them is spoken by far less than a million people. The number varies because it is difficult to determine whether a dialect or stage of development can already be considered a new language, and it usually depends on political rather than linguistic reasons. And if one includes all the dead languages, the number will grow further.

At the beginning of many foreign language courses you can find some statistics – in order to comprehend newspapers and complicated issues of a language, you willl need to know about 4000-5000 words, but a person of average education knows about 10000 words in his mothertongue. However, the whole lexical size may vary and reach 200-500 thousand words and even more, up to a million, if one adds up the technical vocabulary. This also depends on the history of a language, English e.g. has taken over words from more than 100 languages and thus has a tremendous lexical size.

But if we assume just 10000 lexemes (= minimal units, words as concepts in the mind disregarding flexion etc.) for a language, an average value of 4500 languages and multiply them, we will get 45 · 106 lexemes used on Earth today.

The human mouth is able to produce a lot of sounds, but comparatively few of them are actually used in speech because there has to be an audible difference which can be heard clearly. And again, the exact amount of phonemes (the smallest contrastive units in the sound system) of a language varies widely. Japanese for example distinguishes 26 consonant and 5 vowel phonemes, while Russian has about 40, and some rare languages distinguish even more than 100 phonemes. I think that 30 is a good average value, especially if we are concerned with audible similarities, as many unique phonemes of a language will be in some way similar to already existing ones. To these we will add 5 vowel phonemes.

[UPDATE : The World Atlas of Language Structure (http://wals.info/) determines 22 ± 3 consonant and 5−6 vowel phonemes as average from a sample of c. 560 languages. This is less than I guessed, but still close enough and will not change the main point.]

How long is an average word of a language? Here again it matters how much phonemes are distinguished. Languages with a limited phonology have to even it out by word length. An average Finnish word is for example 11 graphemes (signs, characters) long, a Japanese one 7 (in romaji transcription), while English has an average word length of about 5 graphemes (see part 3). However, it often happens that a phoneme is represented by several graphemes (e.g. ph in English for [f]) and vice versa (English ’tide’ for [thai̯d]), there are also graphemes without pronunciation (same example: in ’tide’ the ’e’ is not pronounced), so the average word length in phonemes will be a slightly different one. But at least we get the scale.
So let us assume an average word consisting of 7 phonemes – 4 consonants and 3 vowels. We get 304 · 53 = 108 different lexemes which can be produced. So altogether we have about 50 million existing lexemes and 100 million possible lexemes, which means that if we invent a new word, the probability that we find an already existing one with an identical shape equals 50%, disregarding the semantics. But since one can be sure that mankind has already occupied all the lexemes most pleasant to the ear, a completely new word would probably sound awful.

Regarding the semantics, we have to simplify that as well. Assuming that every lexeme of the given 10000 has an exact cognate in another language, we can say that any lexeme will have 4500 different shapes in other languages. What about the probability that it shares shape and meaning with some of them, at least approximately? At first we have to define ’approximately’.
Assuming a semantic leeway of 3 (which means that if a lexeme has a closely related meaning to another, e.g. house-dwelling-homestead or mound-hill-knoll, it will be accepted as a match), a variation of not more than one consonant (e.g. t-d) and not more than one vowel (e.g. a-o), we will get 4500 · 3 · 2 · 2 = 54 · 103 candidates for a match.
The probability of a match will be 54 · 103 divided by the lexical size, n = 45 · 106 lexemes. This results in p = 1.2 · 10−3. Not much, but given the large pool of words we can define a variable M: ’the quantity of matches’ and would naturally expect E(M) = n · p = 54000 such matches all over the world.
Furthermore, probability theory tells us that we can expect the true number to be in a 3σ interval around E(M) with a probability of 99.7%. We get: σ = √n p (1−p) = √45 · 106 · 2 · 10−3 · 0.998 = 300.

So with a probability of 99.7% we would expect between ∼ 53000 and ∼ 55000 matches all over the world, which is approximately 0.12% of the world’s vocabulary. Applying the same method to the exact matches only we’d expect between ∼ 4300 and ∼ 4700 of them, which is approximately 0.01% of the world’s vocabulary.

Of course this is just a very rough calculation. Not every phoneme is equiprobable; if we imagine an m somewhere in the middle of a word, it is very likely that a homorganic consonant – another m, a p or a b will follow; words with 4 consonants in a row and 3 following vowels will never occur; some languages compensate the shortness of their words and a limited phonology by different tone heights; there are a lot of loanwords; related languages share many lexemes; some agglutinating languages express whole sentences in one word, and so on. But even having taken all this into consideration, we will discover lots of new aspects of inaccuracy. This remains a common problem when applying mathematics to reality – we can calculate a dice fairly exactly, but in order to simulate more complicated events we could adjust our accuracy ad infinitum.

But the main point should be clear, we shall find a lot of similar words if we look for them thoroughly enough. The problem is that not many people speak even ten of those thousands of languages and if we want to make a computer find the matches, an input of millions of words in phonological transcription has to be done. So all these matches are usually found by chance, but we would be surprised if there were not any at all or too few.

And indeed, a lot is known today. For example, I do not think that German kaufen ’to buy’ was under any influence of Japanese kau, just as Japanese namae [namaε] is very similar to German Name [’na:mə], English name. The Akkadian third person feminine pronoun šī is basically identical to English she. The name of the Greek giant of the storms Typhon in English pronunciation is by chance very similar to Chinese tái fēng, Japanese taifū ’typhoon’.
This list can be further continued, but it is not the purpose here.

2  Of how dvergrinn Gandalfr became a Maia, or:
The projection of a certain linguistic situation onto another

Now back to Tolkien. I dislike allegory whenever I smell it — he said once in an interview for the BBC in 1971. And on other occasions he kept repeating that his works were not meant to be interpreted that way (and this includes linguistic allegories as well!), although there were lots of attempts to do it during his lifetime and unfortunately they have not ceased until today. Tolkien’s reaction? He was puzzled, and indeed sometimes irritated (Let:297).

If one opens the Völuspá, a part of the poetic Edda, one of the most important sources of legends in Norse mythology, one may stumble upon the stanzas 10-13 and especially upon the following lines [5]:

Veigr ok Gandálfr
Vindálfr
Þráinn
Þekkr ok Þorinn
Þrór Vitr ok Litr

’Liquor’ and ’Wand-Elf’,
’Wind-Elf’, ’Craver’,
’Clever’ and ’Daring’,
’Burgeoning’, ’Smart’ and ’Colour’

This is a list of dwarf names which will be very familiar to a reader of Tolkien’s prose. The ending -r indicates nominative singular of masculine nouns in Old Norse, and it assimilates to a final -n. Leaving this ending out we recognize the stem Gandalf- (formed out of gand-r ’wand, staff, cane’ and álf-r ’elf’) and also Thorin- (þora ’to dare’. Other dwarf names from the Völuspá are:

Compare them to Tolkien’s dwarves in the ’Hobbit’: Bifur, Bofur, Bombur, Nori, Fili, Kili, Fundin, Dwalin, Dori, Ori. There is also Durin, a legend in both worlds. In the Norse mythology he is one of the two firstly created dwarfs, in Tolkien’s epos Durin the Deathless was the eldest of the Seven Fathers of the dwarves (LotR App.A part III).

Now what’s happened here? A secret link to the Vikings or a joke? Certainly not, although I have heard such opinions. This is just a device of Tolkien’s linguistic construction. He took some words from our real world deliberately and left them unchanged, for the simple reason that an English reader should not feel as if being in a totally alien world.
Both The Hobbit and The Lord of the Rings are written from the hobbits’ point of view. And the hobbits spoke a modern dialect of the common tongue Westron. So wherever Westron occurred it was translated or substituted by English. And to some extent the different variations of Westron have been rendered by the different variations of English. In Gondor and Rohan a more archaic and elevated language was in use, in comparison to the Shire:

’Farewell, my hobbits! May we meet again in my house! There you shall sit beside me and tell me all that your hearts desire: the deeds of your grandsires, as far as you can reckon them …’
The hobbits bowed low. ’So that is the King of Rohan!’ said Pippin in an undertone. ’A fine old fellow. Very polite.’
(LotR III end of ch.8)

But being an exceptional philologist Tolkien went even beyond it, and so the Rohirric language which was akin to Westron, but very archaic, has been rendered by Old English, which has about the same relation to modern English. One has to be aware of the fact though, that Rohirric is not Old English. It is just a translation, Rohirric would have sounded to a Westron speaker just as Old English sounds to an English speaker, there is nothing more behind it than that.

Coming back to the dwarves - the internal reason for the taking of names in a Mannish language is mentioned in PM:304:
For reasons which neither Elves nor Men ever fully understood they would not reveal any personal names to people of other kin, nor later when they had acquired the arts of writing would allow them ever to be carved or written. They therefore took names by which they could be known to their allies in Mannish forms.
So they took names in the language of the Dale, where another archaic dialect of Westron was spoken, substituted by Old Norse this time.

This was not the conception from the very beginning. When writing The Hobbit Tolkien did not think a lot about these names, he wrote it as a fairy tale after all, there are just very few Elvish names mentioned in the book. Of course, his Silmarillion epos did already exist at this stage, but he had problems to publish it. Instead, a sequel of The Hobbit was demanded and so he had to accept the Norse names and to weave them somehow into his myths, to expand the world of The Hobbit and make it compatible to the legends of the first and second age.
Compare the following extract of the letter he wrote to G. E. Selby on 14 December 1937:

I don’t much approve of The Hobbit myself, preferring my own mythology (which is just touched on) with its consistent nomenclature – Elrond, Gondolin, and Esgaroth have escaped out of it – and organized history, to this rabble of Eddaic-named dwarves out of Völuspá, newfangled hobbits and gollums (invented in an idle hour) and Anglo-Saxon runes. (PM:70-71)

But the explanation with the language of the Dale has now to be accepted.

In The Lord of the Rings the languages which had no relation to Westron were kept in original, but some adaptation has been made here as well. Sindarin is for example written in a transcription which comes very close to Welsh (like the diphthong au which is written aw finally in Sindarin; and always this way in Welsh).
Tolkien states about it: […] changes have been deliberately devised to give it a linguistic character very like (though not identical with) British-Welsh: because that character is one that I find, in some linguistic moods, very attractive; and because it seems to fit the rather ’Celtic’ type of legends and stories told of its speakers. (Let:144)

Similarly, the role of Quenya in Arda is very much like the one of Latin in (medieval) Europe. Therefore its spelling has been ’latinized’:
The archaic language of lore is meant to be a kind of ’Elven-latin’, and by transcribing it into a spelling closely resembling that of Latin. (Let:144)

As one can see this projection of the linguistic situation of Arda is in no way a basis for allegories. The Rohirrim are not meant to represent the Old English in culture and Tolkien’s legendarium is not a big parable.
The ’dwarves’ of my legends are far nearer to the dwarfs of Germanic [legends] than are the Elves, but still in many ways very different from them. (Let:297)

Since this is in fact not a similarity between Tolkien’s language and another, we will quickly move on to the next point. I recommend to read the chapter On Translation in The Lord of the Rings, appendix F, where this literary device is extensively described.

3  Looking for Kvenland, or:
A statistical comparison of Finnish and Quenya

The Elvish languages have not been devised by Tolkien out of nothing (already knowing any language this would be impossible), he used some already existing human languages with which he associated certain feelings as a help for his creation. About Quenya he says:

Actually it might be said to be composed on a Latin basis with two other (main) ingredients that happen to give me ’phonaesthetic’ pleasure: Finnish and Greek. (Let:144)

So Quenya has Finnish, Greek and Latin as ’ingredients’ (though Finnish is clearly the main influence), while Sindarin (as already mentioned in a quotation above) is based on Welsh. That’s a kind of junction to the modern world and will naturally lead to some similarities. Quenya and Sindarin of course remain distinct languages with a lot of differences to their ’models’, but a close phonology will always raise the probability of a word match.

This can be exemplified mathematically by comparing Quenya and Finnish. Both languages have a fairly limited phonology, which makes calculations very easy.

So let’s take Quenya as the point of reference. Its phonology allows only 17 phonemes to stand initially:
k (q), f, h, l, m, n, p, r, s, t, v, w, j and the unvoiced hl, hr, hw. Note that k is written c in a latinized spelling and the combination qu or q is actually kw in pronunciation. I will use k below. Furthermore, the palatal approximant j is always written y in Quenya, but since the same letter represents a vowel in Finnish, I shall substitute j for it even in Quenya for this calculation.
Medially Quenya favours certain consonant clusters, the most common ones being: nt, nd, mp, mb, nk, ng, nn, mm, tt, st, lt, ld, ll, rt, rd, ss. And only these will be used for the calculation.
There are 6 diphthongs in Quenya: ai, au, oi, ui and eu, iu, but since the latter two occur only in a few words, we shall take only the first four.

At the end of a word Quenya does only permit the vowels a, e, i, o, u or the dentals l, n, r, s, t.

This should be enough to make a comparison. I was able to extract the Quenya words from Helge Fauskanger’s Etymologies index (as the matching lists between Quenya and Finnish refer to The Etymologies only) and compile them as a single string of characters excluding affixes. Then I wrote a simple computer program to analyze this string in regard to the given sound combinations (it has also been used to establish the average word length in part 1). The same has been done with a word list of ∼ 94 000 Finnish words.

Furthermore I have assumed that mp-mb, nt-nd, nk-ng, lt-ld and rt-rd are close enough, so that e.g. Quenya mp matches Finnish mb and vice versa (even though ng represents in Finnish a geminate velar nasal [ŋŋ]) and that the initial Quenya consonants hl-l, hr-r, hw-w-v can be seen close enough to be counted as one sound (in fact, initial w becomes v in Third Age Quenya). All long vowels have been treated as short.

It is also necessary to note that Finnish does not distinguish the voiced b, d, g from unvoiced p, t, k very clearly and they often sound quite similar, so I could have matched initial Quenya p with initial Finnish b. However, Finnish – as well as Quenya – does not permit voiced stops to stand alone, so that words where they nevertheless occur are loanwords, and are not of Finnish origin. The Finnish vowels y [y] and ö [ø] have no Quenya cognates and have not been assigned to any other vowel. However, the Finnish vowels ä [æ] and e [e] have been assigned to Quenya e.

Finally, I have got the following results (the relative frequencies given):

INITIALQuenyaFinnish
a0.09740.0663
e, ä0.04520.0363
i0.02880.0263
o0.03270.0309
u0.02650.0183
k0.06700.1428
f0.03580.0052
h0.06310.0611
l, lh0.09190.0680
m0.07630.0604
n0.09420.0295
p0.03120.0927
r, rh0.03890.0477
s0.05370.0911
t0.12380.0890
v, w, hw0.0545 0.0761
j0.03270.0285
sum1.00.97



MEDIALQuenyaFinnish
nt, nd0.12930.0506
mp, mb0.03660.0270
nk, ng, ŋŋ0.06070.0278
nn0.00620.0380
mm0.00230.0230
tt0.00780.1203
st0.03270.1394
lt, ld0.04130.0221
ll0.01790.0769
rt, rd0.01640.0168
ss0.04360.1221
ai0.03970.0741
oi0.00390.0712
ui0.01400.0223
au0.01870.0313
sum0.390.68



FINALQuenyaFinnish
a0.39560.2585
e, ä0.30220.1328
i0.02260.1051
o0.08720.0224
u0.01010.0229
l0.02340.0003
n0.08800.2637
r0.05760.0007
s0.00550.0317
t0.00780.0954
sum1.00.93

Very interesting observations can be drawn from these statistics. Thus, 97% of the Finnish words begin and 93% end just like in Quenya. The favoured Quenya consonant clusters actually occur only in approximately 39% of the Quenya words, but in 68% of the Finnish words, so that most Quenya words have single intervocalic consonants. But Tolkien has already told us that Quenya is meant to be less consonantal (Let:144). The frequency of certain phonemes is, however, usually very different, so that the languages are not that similar after all. Indeed, Harri Perälä states in his essay Are High Elves Finno-Ugric? that about one third of Quenya’s vocabulary in the Etymologies is incompatible with Finnish [1]. Notable is the final -a, more than one third of the words end by this sound in Quenya and one fourth in Finnish.

So another conclusion from this table is the most probable shape of a match. Judging by the given values, it might have a final -a or -e/ä, maybe -nt- or -nd- medially and k- or a- initially.

In order to estimate the probability we have to multiply all horizontal values and add them up afterwards. Thus we get the probability

The probability that a word will match in all the three positions will be pimf = pi pm pf = 0.0002924. This is tiny, but remember that we have taken just two languages and restricted our pool of words in comparison to the calculation in part one. Without the phonetic similarities this probability would have been without a doubt even lower. Thus, having analyzed 1284 entries from The Etymologies, we would expect 0.0002924 · 1284 = 0.38 matches without a semantic leeway. Using the formula p(X=k) = (n! / (k! (nr)! )) p(1−p)(nk) we can estimate the probability of a certain number of matches (this is our X). Inserting pimf and n = 1284 into the formula we get:

So, even with a close phonology it is most likely to get zero matches. But how many are there in reality? Without any semantic leeway there are three ([2], p.18):

Note, however, that a word like tië has not been thought of when defining the matching conditions, so in fact we have just two exact matches (and ’exact’ means here: fitting into these previously defined conditions). But even so we get just a probability for such a coincidence of ∼ 5% - it’s rather unlikely, which may be a hint that Tolkien took some of these words directly from Finnish, perhpas their sound already fitted the meaning for him just as the word ond did; see below. It must be said in addition that antaa has more meanings in Finnish and ruskea is a Lappish loan ([2], p.19).

We go further and apply a semantic leeway of 10, so that each word can match another one with the identical shape and one of 10 related meanings. We would expect 10 times more matches, or 3.8. The probabilities of the possible variations are now:

And examining the languages the following matches have been found ([2] p.18):

And one could add:

The combinations -ny- and -nn- were not counted as a match, but even if we had done this, -ny- is not very frequent in Quenya (neither is Finnish -nj-), so that it would not have changed the results fundamentally.

So together with anta and ruska there are 5 or 6 matches, much better within our limits of expectation. And as you can see ’spear-head, point’ is distanced from ’thumbtack’ fairly enough, one has to go through words like spear, stake, spike, pointed, to stab, to pierce, spine, nail, pin before one arrives at something like ’thumbtack’.
Furthermore these words confirm our assumption in regard of the shape of a match, they all have a final -a; medial -nt- and initial k-, a- do also occur.

From a mathematical point of view we may guess that Tolkien has taken (or was strongly influenced by) one or two of the ’exact’ matches listed below. But otherwise it is phonetic similarity which inevitably leads to vocabulary resemblances above the usual average.

An interesting fact: Making the same formal analysis of Helge Fauskanger’s Quettaprama Quenyallo, a wordlist which also includes words from other writings as well as Qenya vocabulary from the Lost Tales. But the distributions hardly differ from those of The Etymologies and for the analyzed 2818 words I got pi = 0.0636, pm = 0.0263, pf = 0.1611 and pimf = 0.0002695 (instead of 0.0002924) which shows that Q(u)enya’s phonology did not change too much in the course of time.

An almost identical match is found beyond The Etymologies and cannot be taken into account here (intervocalic -n- is involved):

Finally, I will complete the list, giving the remaining matches originally compiled by the Finnish linguist Rautala ([2] p.18), but the matching is here either initial-medial or medial-final:

Q. et(e)- prefix ’forth, out’ (Etym:ET-)
F. eteen ’forward, to the front’, etu- prefix ’front-’

Q. hala ’(small) fish’ (Etym:KHAL-/SKAL-, VT45:20)
F. kala ’fish’

Q. lapsë ’babe’ (Etym:LAP-)
F. lapsi ’child’

Q. si- ’this, here, now’ (Rgeo:67, LR:47, SD:310, VT43:34)
F. se ’that, the’; siinä ’therein, in that, in it, in there, there’

Q. tereva ’fine, acute’ (Etym:TER-)
F. terävä ’sharp’

Q. tundo ’hill, mound’ (Etym:TUN-)
F. tunturi small mountain (in Lapland)

Q. tul- ’come’ (infinitive tulë) (Etym:TUL-, WJ:368, LR:47, SD:246)
F. tulla ’come’

Q. varya- ’protect’ (Etym:BAR-)
F. varjella ’guard, preserve, protect’

[Note: The list of the referred article includes Q. gandel/gannel ’harp’ — F. kannel (Finnish psaltery-like national instrument); but since Quenya does not permit an initial g-, this is a mistake; gandel/gannel is Noldorin.]

For a similar comparision between Noldorin/Sindarin and Welsh and Irish see 8.

4  Why Abraham had never been at the Misty Mountains, or:
Tolkien’s technique of composing new vocabulary

With all the new languages Tolkien needed a lot of new words. He derived them from roots using different endings. This process is very well exemplified in The Etymologies, a part of the Lost Road, the fifth volume of the History of Middle-Earth (roots and their derivates underwent external changes, but this will be regarded later).
For example he thought of a stem MOR-, having to do with blackness or darkness and devised from it words like Q. more ’black’ or mordo ’shadow, obscurity, stain’. This stem is also present in Sindarin place names like Mordor ’black land’, morannon ’back gate’ or Moria ’dark abyss’.

Looking into the Bible one might find the following passages interesting (my underlining):

Then God said, ”Take your son, your only son, Isaac, whom you love, and go to the region of Moriah. Sacrifice him there as a burnt offering on one of the mountains I will tell you about.” (NIV Genesis 22:2)

Then Solomon began to build the temple of the LORD in Jerusalem on Mount Moriah, where the LORD had appeared to his father David. (NIV Chronicles 3:1)

Moriah sounds almost identical to Moria (’Moriah’ is stressed on the ultimate syllable, ’Moria’ on the first) and indeed a connection between the two was suggested during Tolkien’s lifetime. His answer was:
Internally there is no conceivable connexion between the mining of Dwarves, and the story of Abraham. I utterly repudiate any such significances and symbolisms. My mind does not work that way. (Let:297)

Then he explained the external influence which had led to the word. He had once read a Scandinavian tale where the place name Soria Moria Castle occurred. After some time had passed the only thing he remembered was just the sound-sequence moria and since it also alliterated with ’mine’ he took it into his legendarium. However, the main reason for this adaptation was the presence of the stem MOR- which he had devised at first. So the mine Moria neither has connection to the Bible, nor actually to the Scandinavian castle. And this technique of adapting pleasant sound forms must not be confused with the projection described in part 2, one always has to distinguish whether a word is pure Elvish or just ’translated’ into an existing language.
The ’source’, if any, provided solely the sound-sequence (or suggestions for its stimulus) and its purport in the source is totally irrelevant (Let:297)
So even if the source of a certain word had been the Koran or any other book, it would not have had any impact on the internal storyline or Tolkien’s intention. His books were not meant to criticize the present and there are no parallels between the Third Age and the 20th century. He describes his intention as following:

Nobody believes me when I say that my long book is an attempt to create a world in which a form of language agreeable to my personal aesthetic might seem real. But it is true. An enquirer (among many) asked what the L[ord of the] R[ings] was all about, and whether it was an ’allegory’. And I said it was an effort to create a situation in which a common greeting would be elen síla lúmenn’ omentielmo, and that the phase long antedated the book. (Let:205)

As I have already argued in the first part, the multitude of lexemes in the world makes a completely new invention almost impossible. Furthermore the human mind does not work like a computer and cannot create random sound combinations according to the phonetic structure of the mother tongue. Therefore, if a new word is needed, it is much easier to form a new compound. But a language would be overloaded with too many compounds. Thus, loanwords are taken from other languages more frequently and a virtually new word is created, e.g. ’biology’ instead of ’life-science’. Besides, ’science’ is already of Latin origin… So it is not very surprising that there is a similar process when creating new words for an artificial language. Someone who is given such a task is likely to take some known words and modify them or extract some sound combinations in order to compose them anew in a different way. Tolkien concedes such an influence:

Nonetheless one’s mind is, of course, stored with a ’leaf-mould’ of memories (submerged) of names, and these rise up to the surface at times, and may provide with modification the bases of ’invented’ names. (Let:324)

However, it might be interesting to know about the exact source of influence. Tolkien has commented on a lot of similarities:

4.1  Gondor

There is a city in Ethiopia called Gondar. Tolkien states that this might have been an influence, due to the prominence of Ethiopia in the Italian war […] But no more than say Gondwana-land (Let:324). Then he recalls the actual influence – when he was eight he read a statement that not much was known concerning the languages of the natives inhabiting Britain even before the Celts, but one word might be ond ’stone’. He found that the sound of this word was well suited for its meaning.
Gondor means ’Stone-land’. Compare also Q. ondo ’stone’. In an early conception of The Lord of the Rings the land which later became Gondor had indeed the name Ond (TI:400).

4.2  Rohan

An ancient French aristocratic family is called Rohan as well (though differently pronounced in modern French of course) and it is still a place name in Brittany, located near Redon. Tolkien knew about this family and liked the name. However, the stem ROK- and the Primitive Elvish derivative rokkō ’swift horse for riding’ (yielding Q. rocco and S. roch) were already devised at that time (Let:297) and could be used to produce the word Rohan, by adding the suffix -and, later simplified to -an. The sound ch [x] was weakened to h [h] in the speech of Gondor (Let:144).
Tolkien reminds once more about the absence of allegories: Nothing in the history of Brittany will throw any light on the Eorlingas.

4.3  Endor

There is a story in the Bible where Saul talks to a spirit by the invocation of a witch (NIV 1 Samuel 28). She is called ’the witch of Endor’. But there is no connection to Q. Endor, S. Ennor ’Middle-earth’ at all, it’s pure coincidence. Endor is composed out of en(ed) ’middle’ and (n)dor ’land(mass)’ (Let:297). Tolkien recognized the similarity, but left the word unchanged.
Compare also the following:
Within Issachar and Asher, Manasseh also had Beth Shan, Ibleam and the people of Dor, Endor, Taanach and Megiddo (NIV Joshua 17:11)
There is probably not much more connection between Dor and the Elvish element (n)dor.
But Endor/Ennor involves a connection of another kind. It is a direct translation of Old English middangeard, Old Norse Miðgarðr, commonly referred to as ’Midgard’. In Norse mythology this is the dwelling-place of men, compared to Asgard – the dwelling of gods, Hel – the underworld, and a lot of other places. Once there was also an ’Asgard’ in Tolkien’s translation, but he decided to keep only ’Middle-earth’. See the earlier mythological stages in the discussion of the Lost Tales below.

4.4  endings -an(d), -en(d), -(n)dor

An enchanted forest from the saga of king Arthur is called Brocéliand or Brocéliande. There is the Labrador peninsula in northeastern Canada.
The Elvish endings owe something to such names (Let:297). However, Tolkien keeps emphasizing: […] but is perfectly in keeping with an already devised structure of primitive (common) Elvish (C.E.), or it would not have been used.
An interesting fact however is that the very first name of what was later called Beleriand had been indeed Broceliand (LB:157), so originally this was more than just ’owing’; but when Tolkien wrote the letter in 1967 he was not expecting that his notes and works which he had made some 40-50 years ago would be published one day. Other forms he toyed with were: Broseliand, Golodhinand, Noldórinan, Geleriand, Bladorinand, Belaurien, Arsiriand, Lassiriand, Ossiriand (LB:160).

4.5  Erech

This is the hill in Gondor on which the Men of the Mountains swore an oath to assist Isildur and broke it later. There was a city state called Erech in southern Mesopotamia, perhaps better known as ’Uruk’
Tolkien admits that Erech might have been an echo (Let:297), as he had knowledge about the Sumerian city. He invented the name while writing the passage when Aragorn departs upon the Paths of the Dead (LotR V, ch. 2,9), so the ancient place might have been linked to the ancient story of Isildur subconsciously. But according to his words it is more probable that he was influenced by the important element ER (in Elvish) = ’one, single, alone’ (Let:297) since the theme of this passage was Aragorn’s separation from Gandalf and his courage was put to the test.
Note that Erech could be Sindarin, but does not have any meaning in this language, being probably of pre-númenorean origin. This fits into the nomenclature of Gondor, described in VT42, similar names without a meaning are Adorn, Ciril, Eilenach. They were probably just composed out of different Sindarin sound patterns by Tolkien, without a strict derivation from roots (for example Ciril – the root KIR- ’cut’ is known as well as the feminine ending -il < CE *-illē, but together they do not make a suitable sense); so subconscious influence by already known words should have been much more probable in this case. There is also a river Erui and a beacon Erelas beginning with Er- (VT42:8-13).
Also notable are several names of the first men entering Beleriand as described in chapter 16 of the Silmarillion: Marach, Malach, Amlach, Imlach – all ending in -ch – which may be related to these pre-Númenorean names. They clearly hint at a Semitic language, having the same radicals M-L(R)-CH. It must be an ancestor of Adûnaic, the language of Númenor, which is meant to have a faintly Semitic flavour (SD:240). If so, one tends to think that this is one of Tolkien’s devices again – the languages of the early civilizations in the Middle East (as partially described in the Bible) are all Semitic, as well as those of the early men of Arda. Both have such a detail in common as the frequent ending -ch (e.g. Enoch, Antioch occuring in the Bible).

4.6  nazg

This means ’ring’ in Black Speech (cf. Nazgûl ’ring-wraith’). In Irish Gaelic, nasc means ’band, tieband, collar’ (thus in fact only loosely related) and is written nasg in Scottish Gaelic. Here again, subconscious influence might have been the case, as it is admitted by Tolkien:
It is thus probable that nazg is actually derived from it, and this short, hard and clear vocable, sticking out from what seems to me (an unloving alien) a mushy language, became lodged in some comer of my linguistic memory. (Let:297)

4.7  Númenor, númen

Latin has nūmen ’divine touch, divine presence’ and plural nūmina ’divine powers by which gods act their will’. This is just pure coincidence of sounds (not matching in meaning at all). C. S. Lewis spelt it **numinor and Tolkien explains:
His [Lewis’] spelling numinor is a hearing error, aided, no doubt, by his association of the name with Latin nūmen, nūmina, and the adjective ’numinous’. Unfortunate, since the name has no such connexions, and has no reference to ’divinity’ or sense of its presence. It is a construction from the Eldarin base NDU ’below, down; descend’; Q. núme ’going down, occident’; númen ’the direction or region of the sunset’ + nóre ’land’ as an inhabited area. (Let:276)

We can also observe that Tolkien tried to alter his vocabulary when similarities became too obvious or inconvenient:

4.8  Arnor

This means ’royal land’ and should have had the Sindarin form *Ardor, but Tolkien writes that he desired to avoid this form and used Arnor as a blending with Q. arnanóre (Let:347), apparently because it reminded too much of English ardo(u)r.

4.9  û

This is a negating element in the Eldarin languages, e.g. the prefix ú- in Sindarin (as in ú-chebin ’I do not keep’, LotR App.A) as well as in Quenya (e.g. úvanima ’not fair, ugly’ (VT39:14)). Tolkien wrote (date not explicitly given, but it’s a very late note):
It is not necessary to avoid at all costs similarities with known European languages – Eldarin is deliberately devised to resemble them in style – but here the resemblance either to Greek û or to the unrelated Norse ú, as a prefix, is too close. (VT42:33)
But he could not make up his mind how to solve this, introducing and abandoning several solutions. It is unclear whether û was influenced by the real world languages. Its occurrence in the earliest sources (û – ’negative prefix with any part of speech (GL:73)), at a time when Tolkien was learning Greek, makes it quite possible, see also the discussion of the Lost Tales below.

There are other examples where we could at least risk some speculations:

4.10  orch

There is a clear parallel to Old English orc which occurs in the plural form orcneas in Beowulf (l.112), probably the main point of Tolkien’s study and research as a philologist of the English language. However, in his linguistic essay Quendi and Eldar he notes:
The word used in translation of Q urko, S orch, is Orc. But that is because of the similarity of the ancient English word orc, ’evil spirit or bogey’, to the Elvish words. There is possibly no connexion between them. The English word is now generally supposed to be derived from Latin Orcus. (WJ:389)
Latin Orcus means ’hell, underworld’ or ’the God of the underworld’ (Pluto), in a poetic sense also ’Death’.
Although he clearly denies a connection to Old English, one can hardly believe that there is no connection at all. It is obvious, however, that this denial is not so harsh and that Tolkien throws in a ’possibly’. Looking at the external development of the term orc in the Elvish languages, its trace can be already found in the earliest stage of 1915-1917.

The Qenya Lexicon gives ork (orq-) ’monster, demon’ and the Gnomish Lexicon orc ’goblin’ (LT1:304, GL:63). In the process of external evolution Qenya experienced a lot of changes and was later spelled ’Quenya’, while Gnomish became Noldorin and then Sindarin. In The Etymologies (1937-38) the words took the shapes Q. orko and N. orch (stem ÓROK-). The above quotation from Quendi and Eldar (1959-60) shows the latest forms: Q. urko and S. orch (the stem is now given as RUKU-).
Looking at this progression we might wonder whether Tolkien just ’took’ this word from Old English in the beginning without much thinking and then transformed it on purpose, to make it less similar.

4.11  elda, el-

This is a difficult case. The word sounds relatively similar to its meaning elf and we look once again into its development.
The Gnomish Lexicon has Egla ’a being from outside’ (GL:32), name of the fairies given by the Valar and largely adopted by them; a related element is eg ’far away, far distant’. Early Qenya has Elda without etymological notes (LT1:287-288, QL:35). A note in LT2:345 confirms their relation: Rúmil said these names Egla and Elda were akin, but Elfrith cared not overmuch for such lore and they seem not over alike.

This conception remained unchanged in the earliest draft of the Quenta Silmarillion (described in The Shaping of Middle-Earth). But The Etymologies give: ELED- ’Star-folk, Elf’, Q. Elda, N. Eledh while Egla became Doriathrin. Rejected versions were: base ELED- ’go, depart, leave’, Elda/Eledh meaning ’departed’ and base EDE-, EDEL- ’precede, come forward’ with edela/eleda ’firstborn’. In Quendi and Eldar a complicated situation is described, the stems EL- ’star’ and DEL- ’to go, proceed’ are merged, so that Elda is either ’departed’ or ’a person connected or concerned with the stars’, the Sindarin form is now Edhel (WJ:363-364) while Eglain, Egladhrim (sg. Eglan) are just a tribe of the Sindar, the stem is given as HEK- (WJ:379-380).

This is an opposite development to the situation of orc – Tolkien obviously liked the sound of Elda and tried to preserve it, revising the etymology of the word. Egla was moved into background in the slightly changed form Eglan. But Sindarin kept also the element el(l)- in compounds, e.g. elvellon ’elf-friend’ (WJ:312).
The similarity to elf is quite obvious and Tolkien must have been aware of it. It is not that obvious, however, when one looks at the ancient forms of the word ’elf’: ON álfr, OE ælf; OHG Alb, Alp; the suggested origin is the Indo-Germanic stem *albhos- ’white’ (compare Latin albus), though the precise etymology is unknown.

Nothing certain can be said about whether Tolkien intended his Elvish word to be of similar sound. An interesting remark can be found in Let:114: Eldalië (or Elves, by a not very accurate translation). He obviously means that the mythological elves have not much in common with the Eldar, at least in the later stage. In his earliest writings he called them very often ’fairies’ and intended to implement his legends into the mythology of the real world, so it is possible that this was originally indeed an adaptation from natural languages (see the discussion of the Lost Tales below) which became so established that he did not want to change it anymore.
For a further comparison of the mythological elves with the Eldar and other similarities (like Elessar ’elf-stone’ – Albstein) see [4].

4.12  Arda, ardh, garth

Q. Arda ’The Realm’ is the name for the Earth, while it can also be used as a common noun meaning ’region’ (WJ:402,413), S. ardh is translated as ’realm’ (Etym:ƷAR-).
This sounds very similar to the various forms of ’Earth’ in the Germanic languages, e.g. OE eorþe, Dutch aarde, Gothic aírþa ( = [ε]), OHG erda, modern German Erde. Incidentally, ’earth’ is [ʔarð] in Arabic.

It is also interesting that the Elvish stem is ƷAR- or GAR- ’have, hold’; we have thus Dor. garth ’realm’ and from a similar stem GARAT-: S garth ’fort, fortress’. This causes another similarity: Compare ON garðr ’yard, courtyard, fence’, garth ’small piece of enclosed ground’, a word from northern and western English dialect (c. 1340), OE geard ’enclosure, enclosed place, court, dwelling, home, region, land’, modern English garden, German Garten, Welsh gardd ’garden’ (dd = [ð]), Gothic gards ’house’, Russian город ’town’, archaic град; a common word in other Slavonic languages as well.

It is also remarkable that this word is used in place names, see e.g. ON Holmgarðr ’settlement on the island’ = Russian Новгород ’new city’ and Dor. Garthurian ’Fenced Realm’ (= Doriath).

Another interesting detail: When Galadriel gave to Sam a box with earth from her orchard, she said:
Here is set G for Galadriel, but also it may stand for garden in your tongue. (LotR II ch. 8)
So the Westron word for ’garden’ must also begin with g-, although it is otherwise unattested.

Carl Hostetter and Patrick Wynne have already concluded in VT32 that English garden is thus ultimately of Eldarin descent. I think that the evidence is quite obvious and it is highly unlikely that Tolkien as a good philologist has been just influenced subconsciously without recognizing this similarity.

4.13  The Problem of ROS-

This is how a chapter of The Peoples of Middle-earth is titled (p.367 ff.). Late notes from the last years of Tolkien’s life are mentioned there, in which he discusses a possible reinterpretation of two homophone elements ros meaning ’foam’, as in Elros ’star(lit) foam’ and referring to red or red-brown hair, as in Amros ’top-russet’ (< CE russā). The problem was:
It is also unfortunate that the first [homophonic element] appears too reminiscent of Latin rōs [’dew’] or Greek drosos, and the latter too close to well-known modern European ’red’ words: as Latin russus, Italian rosso, English russet, rust, etc. However, the Elvish languages are inevitably full of such reminiscences, so that this is the lesser difficulty. (PM:368)
In The Etymologies one finds ROS- ’distil, drip’, Q. rossë ’fine rain, dew’.
One could add here the Germanic king Barbarossa ’red-beard’. The father of Nerdanel also had a red beard, very exceptional for Elves. It was him from whom Amros (Amras), Amrod and Maedros have inherited their ruddy complexion. However, as it can be seen from the quotation above, these similarities did not disturb Tolkien much, he seems to have regarded them as coincidental.

5  No rule without: Some exceptions

5.1  Eärendil

In the text above a quotation was given:
The ’source’, if any, provided solely the sound-sequence (or suggestions for its stimulus) and its purport in the source is totally irrelevant (Let:297)
But I have cut this passage. The whole sentence reads:
[…] and its purport in the source is totally irrelevant except for the case of Earendil.

This word has its origin in Old English (Anglo-Saxon), Tolkien discovered it already in 1913. He had to learn Greek and Latin, but studied Anglo-Saxon in his free time. In Crist, a collection of Anglo-Saxon religious poems by the poet Cynewulf, he was astonished by the following lines:

éala éarendel engla beorhtast
ofer middangeard monnum sended

Hail Éarendel, brightest of angels
above the middle-earth sent unto men

There is nothing one can add to Tolkien’s own words:
I was struck by the great beauty of this word (or name), entirely coherent with the normal style of A[nglo]-S[axon], but euphonic to a peculiar degree in that pleasing but not ’delectable’ language. Also its form strongly suggests that it is in origin a proper name and not a common noun. This is borne out by the obviously related forms in other Germanic languages; from which amid the confusions and debasements of late traditions it at least seems certain that it belonged to astronomical-myth, and was the name of a star or star-group. (Let:297)

Then he imagined that this star was Venus which announced the sunrise, and he adapted the word into his mythology, as a person becoming a star – not just like that, but giving him an Elvish etymology. At first he left the form unchanged as Eärendel, associating it with Qenya ea, earen ’eagle’, but without giving an explanation of the ending. He just stated that the name was thought to be woven of a secret tongue of Gondolin (LT1:287, LT2:270-271). His ship was called Earum which seems to be an Elvish adaptation of OE. Earnhama ’Eagle-coat, Eagle-dress’ written at its side (LT2:264,281). It looks like Tolkien was toying with an Anglo-Saxon etymology or at least interpretation of the first part of the name.
But later he found explanations for both parts, changing the name slightly to Eärendil; and so eär became ’Sea’ in Quenya (stem AYAR-) while the suffix -(n)dil meant ’to love, to be devoted to’ (Let:273).

Thus Eärendil is not a mere sound borrowing, only in one case Eärendil will reference to its source cast any light on the legends or their ’meaning’ – and even in this case the light is little. (ibid.)

5.2  Atalantë

And yet we may doubt whether it is really the only exception. Another similar case might be Atalantë ’the Fallen’ (Etym:TALAT-, also in ’Akallabêth’), one of the names of Númenor. This very close to the famous Atlantis which also shares a lot of features with Númenor. Tolkien was aware of this and used in fact the term Númenor-Atlantis several times (e.g. Let:151).
A verse in LR:47 shows us another form of ’fallen’: ataltane, but the name of the island did not take this form. Furthermore, during the same period Tolkien devised the Noldorin words atlaud, aclod ’sloping, tilted’, atlant ’oblique, slanting’, atlanno ’to slope, slant’ (< TALÁT-, VT46:17). This is quite peculiar because Noldorin belongs to the ’Welsh branch’ of Tolkien’s linguistic creation which means that it has a system of consonant mutations similar to Welsh, just like its predecessor Goldogrin (Gnomish) and its follower Sindarin. Since the stem TALÁT- receives an additional prefix A-, we would rather have expected a mutation t > d of the initial consonant. Even if this lack can be explained by some internal reason, one might get the idea that Tolkien just wanted to preserve a sound which would still be very similar to ’Atlantis’. Finally, these three words are the only ones in the material of The Etymologies and later Sindarin in which a medial combination -tl- occurs.

And yet a footnote to the letter 257 tells us: It is a curious chance that the stem talat used in Q[uenya] for ’slipping, sliding, falling down’, of which atalantie is a normal (in Q) noun-formation, should so much resemble Atlantis.
Curious chance? Well, one can hardly believe that, even if these are Tolkien’s words. It rather seems that this is a subtle device to underline the role of Númenor-Atlantis, which is a commonly known mythological story in Europe. Tolkien was trying to put a lot of such ancient, typical and recognizable elements into his narratives, which is eventually one of the main reasons for their popularity.

5.3  Avallónë

A third pun of the same category seems to be Avallónë, a city on the eastern coast of Tol Eressëa which clearly resembles the Celtic Avalon, a blessed island in the west to which king Arthur departed to be healed from his wounds.
The meanings of the two words are however different, Avalon is supposed to be related to Welsh afal [’aval] ’apple’. In The Etymologies we can find Avalóna as an alternative name for Tol Eressëa (AWA-, LONO-). An addition reads: A-val-lon, and Avallon is used in the ’Lost Road’ with the meaning ’hard by Valinor’ (= near Valinor), compare the early Qenya name Arvalion, Arvalin ’nigh Valinor’ (LT1:289), later only Arvalin ’outside Valinor’ (Etym:AR2- ’outside, beside’) also G. ar- ’beside, along with, compared with, etc.’ (GL:10) which seems to be related to ar ’and’. In Lowdham’s report on the Adunaic language (SD:413-440) Quenya is always called Avallonian.
So we have two interpretations based upon two divisions of the compound: ava-lóna (< *awa-lōnā) and a-val-lóne (< *a-bal-lōnē, with a- ’near’, or simple sundóma prefix?). The latter contains the otherwise so well-known element val-, as in Vala.
But the similarities do not stop yet: The Welsh name for Avalon is Annwn (with w = [u]) which is so close to S. annûn ’west, sunset’ (Etym:AM-, NDŪ-). For the discussion of their relation see Lambengolmor messages #4-6.

5.4  Danas

And there might be a fourth pun of this kind. From The Etymologies we get to know that the Green-elves called themselves Danas (DAN-), probably connected to NDAN- ’back’, as they turned back during the great march of the Elves from the east, thus Q. Nandor. Another interpretation (but not necessarily a contradicting one) is that there was an elf named Dân who let away a numerous people (WJ:13), they became the Nandor.
But in Celtic mythology Tuatha De Danann ’the people of the goddess Dana (Dan/Danu)’ was a term for fairies who are in particular masters of magical craft and live in underground dwellings – or, simply – for the elves.
We know from another example that the name of a mythological god was used as one of a king in Arda, namely Ing, Ingwë (see part 7 below), so the same might have happened with Dân/Dana.

6  ’Quettar i ataren ar i amilen’, or:
The universality of pa-pa and ma-ma

There are two groups of words where matches are not a big surprise.

The first are the so-called onomatopoeic words, a rendering of animal sounds using human speech. And so we have e.g. English crow, German Krähe, Japanese karasu, Finnish korppi, as well as Q. corco and S. corch (Etym:KORKA- << KARKA-); all containing the radicals k-r.

Some words are formed in resemblance of a characteristic sound of a certain item. Most probably we can count Exilic Noldorin dramb, dram(m) ’a heavy stroke, a blow (e.g. of axe)’ (Etym:DARÁM- ’beat, hew’) and English drum under this category.

The second group consists out of the words for ’mother’ and ’father’. Children all over the world say very similar things when they are learning how to speak – they are calling their parents using the simplest vowel [a] with some basic consonants. These words are usually diminutive, but sometimes have developed into regular words.
Apart from Indo-European mama and papa there are for example Turkish ata ’father’, ana ’mother’, Japanese haha ’(my) mother’ (actually deriving from *papa), Gothic atta ’father’, Quechua tayta ’father’ and many more. Curiously, Turkish and Albanian baba ’father’ turns out as a derogatory term for older women in Russian as well as in Japanese.

And thus we have the Elvish stem ATA- with Q. atar (or #Átar (VT43:13)), diminutive atto, atya; S. adar, diminutive ada all meaning ’father’; the stem NAN- with S. naneth, diminutive Q./S. nana; AM- with Q. amil, amme; and finally diminutive emme, emya all meaning ’mother’ (Etym, VT47:10,26).

7  How many Eldar were driven out by Rome’s legions, or:
Why the Lost Tales are different

Tolkien’s legendarium is not wholly remote from the real world. It takes place in its imaginative history, just in a different stage of imagination. But looking for direct links you will not find a lot. There is just the whole shape of the world, the general outlines of the continents and the identical sky – sun, moon and the same constellations. And what is The Lord of the Rings all about? It is not ’about’ anything but itself (Let:165).
However, Tolkien’s works underwent a lot of changes during his lifetime. He describes his original motivation as following:

Also – and here I hope I shall not sound absurd – I was from early days grieved by the poverty of my own beloved country: it had no stories of its own (bound up with its tongue and soil), not of the quality that I sought, and found (as an ingredient) in legends of other lands. There was Greek, and Celtic, and Romance, Germanic, Scandinavian, and Finnish (which greatly affected me); but nothing English, save impoverished chap-book stuff.
[…] Do not laugh! But once upon a time (my crest has long since fallen) I had a mind to make a body of more or less connected legend, ranging from the large and cosmogonic, to the level of romantic fairy-story-the larger founded on the lesser in contact with the earth, the lesser drawing splendour from the vast backcloths - which I could dedicate simply to: to England; to my country. (Let:131).

The writings of these English ’neo-legends’ have been published in the two first tomes of The History of Middle-earthThe Book of the Lost Tales I and II respectively. This first stage needs a special treatment if we want to understand linguistic similarities.

One main difference to the later works is their close relation to other real-world mythologies, especially the Norse. We can observe a lot of direct adaptations. Just a few examples, like this later rejected outline:

After Úrin (later: >> Húrin) is released from the captivity of Melko (later: >> Melkor), he travels with some companions to the caverns of the Rothwarin (later: >> the caves of Nargothrond) where he finds the great treasure of the dragon Glorund (later: >> Glaurung). It is guarded by a single dwarf called Mîm. Úrin demands the treasure for himself, Mîm resists and is slain by Úrin, but speaks a curse upon the gold before his death. Then Úrin and his companions, not paying any heed to this curse, take the gold and bring it to the halls of Linwë (later: >> Thingol), throw it before his feet and Úrin accuses the king of weakness which led to the death of his son, Túrin. His intention is to leave without the gold after that, but his companions do not want to abandon it there, and a fight breaks out between them and the guards of Linwë. Úrin’s band is slain and Linwë casts the cursed gold into a deep pool except for the Ring of Doom, which Tolkien changed to the Necklace of the Dwarves (Nauglafring, later: >> Nauglamir) (LT2:137-138).
On another note, is also said that whoever eats a dragon’s heart will know all tongues of Gods or Men, of birds and beasts (LT2:86).

Compare it to the Germanic Siegfried legend: Siegfried slays the dragon Fafnir, defeats the dwarf Alberich and acquires the cursed treasure of the Nibelungs, especially the most costly Ring of the Nibelungs. But being cursed, the treasure (for which a lot of blood is shed) brings only misfortune and in the end it is cast into the Rhine by Hagen.
The Edda version of this legend, which is more concerned with the slaying of the dragon, describes that Sigurd (= Siegfried) roasted Fafnir’s heart on a spit. Having tasted its blood he became able to understand the speech of birds (Fáfnismál:31-38, see also LT2:125-126).

Oromë fashions a long rope with the help of Vána which connects two mountain peaks of Valinor and Middle-earth as a bridge.
Now because it glistens most marvellously in the slanting rays of the Sun, and when the rains of heaven moisten it it shines most magically therein and the gold light breaks upon its dripping cords to many hues of purple, green, and red, so do men most often name it the Rainbow […] (LT1:239).
In Norse legends the rainbow has the same mythical explanation – it is a bridge connecting Midgard and Asgard. And Valinor is translated as ’Asgard’ (LT1:316). Similarly the Valar Manweg (later: >> Manwë) and Tulkas are identified with the Germanic gods Odin (god of wisdom, runes and magic with two ravens who report to him everything that happens in the world) and Thor (master of rain and lightning; strongest among the gods) (LT2:295).

All this has been rejected in the later outlines of the Silmarillion with the exception of Midgard – Middle-earth. Tolkien’s later legends involve just very faint similarities, some well-known archetypical motives. But the enormous depth of his legendarium needed a lot of time to evolve, and at its very beginning he was of course more strongly inspired by other sources. Consider that the Lost Tales were written by Tolkien at the age of 23-25. But not only the legends were similar to those of the real world, the whole story itself had a setting in the history of Earth.
Those tales were explicitly put into another framework. There is a sailor called Eriol or later Ælfwine (’elf-friend’) who visits the island of Tol Eressëa and listens to the narratives of the last Elves who live there. His story is, according to Christopher Tolkien, the most difficult part of the earliest form of the mythology (LT2:282) and will therefore be compressed here. In later writings he is mentioned by the Quenya name Elendil ’elf-friend’ and is the one who passes on the teachings of Pengolodh, but he is not given an internal story in this kind of detail anymore:

According to the first outline there is a man called Ottor Wǽfre who lives on the island of Heligoland and his sons are Hengest and Horsa (mythical leaders of the Anglo-Saxon invasion of England in the 5th century). After the death of his father he sets out to reach ’the unknown island’ Tol Eressëa and finds it somewhere in the Atlantic Ocean. The Elves live there, retreated after the battles against Melko (>> Melkor). Wǽfre visits the Cottage of the Lost Play in the city of Kortirion where he is being told all the legends of the former ages and writes them down. The Elves call him Eriol or Angol. Later Tol Eressëa is moved from the west and anchored at the coast of Middle-earth. Ossë tries to drag it back and a part is split, becoming the isle of Íverin (Ireland), while the other part becomes England. The age of men has begun and the Elves begin to fade, becoming small and translucent until they are invisible to the eyes of most men; and thus ultimately become ’ fairies’. Only those men who are friendly to the Elves (i.e. Eriol, his family and thereby all their descendants - the English) have the true tradition of the fairies (LT1:14-15, LT2:298-299).

According to the later Ælfwine story, England is not Tol Eressëa, but the two are simultaneously in existence. The last Elves live in England. The ruler of the island is called Ing or Ingwë (a human being, not yet a Vanya – this mysterious name is taken from an Old English runic inscription, see LT2:311) and he is a friend of the Elves. Eärendel comes to England and tells Ingwë of Tol Eressëa. Ingwë sets out to find the island, but is hindered by Ossë and driven to the east (probably to the Baltic region) where he becomes king of the Ingwaiwar (’friends of Ingwë’), the ancestors of the Anglo-Saxons. He teaches them about the Elves and this knowledge is inherited. England suffers a lot of invasions by several peoples of hostile men, which causes more and more Elves to depart to Tol Eressëa on the Atlantic Ocean; and they name the places there after the places in England. The last invasion by the Anglo-Saxons is friendly to the Elves. Finally, Ælfwine (’elf-friend’) from the house of Ingwë sets out to find Tol Eressëa and is successful. The Elves speak Old English there and tell him the legends of the first ages. They call him Lúthien which means ’friend’ (and has yet nothing to do with the elf-maid Lúthien Tinúviel) while England is called Luthany ’The Isle of Friendship’ (LT2:314-315).

In a slightly later account of the languages it is even said directly that ’fading Telerin’ was still spoken at the Southern and Western shores of England and Wales (PE14:61) and that Ilkorin (the language of those Elves who did not reach Valinor) was still present in Ireland, England, Wales and Scotland; and that a related form was even spoken in Scandinavia, the lands bordering the North Sea and English Channel (there are indeed many similarities with Germanic languages, see below). The Ilkorindi were said to be lingering over the whole of Europe up to the western regions of Russia (PE14:62).

The matter is in fact much more complicated, but for this analysis it is important to know that:

What about the languages during that time? Christopher Tolkien remarks that it is immediately obvious that an already extremely sophisticated and phonetically intricate historical structure lies behind the languages at this stage; but it seems that (unhappily and frustratingly) very little indeed in the way of phonological or grammatical description now survives from those days. I have found nothing, for instance, that sets out even in the sketchiest way the phonological relations between the two languages (LT2:282).
So the linguistic situation is already fairly complex at that time, up to such details as poetic words (e.g. G. culu for ’gold’ or other red and yellow metals) and archaic forms like the name Laigolas (LT1:308). The Noldoli (later: >> Noldor) made a pun out of the similar names Laigolas ’green-leaf’ and Legolast ’keen-sight’, while Legolas was a blending of the two (ibid.). Two lexicons of Qenya and Goldogrin with the longest word lists Tolkien ever wrote down are put together with the Lost Tales. Patrick Wynne argues in his short essay Are Goldogrin and Qenya ”primitive”? that the two are definitely not ’primitive’ in the sense of ’too simple’ [6].

The point of this lengthy excursus is to clarify the framework of the languages – imagine an authentic linguistic situation applied to a world where the Elvish tongues coexist with known human languages, where the Elves inhabit England and have close contact to the Anglo-Saxons. Imagine a young Tolkien who is looking for ideas, motives and inspirations in the real world. All this will lead to similarities and loans on a large scale.

7.1  England

At first there is Eriol or Angol himself. The name Angol is derived from G. ang ’iron’ and ôl ’cliff’ (LT1:288, GL:19,62) and he was named after the regions of his home (LT1:14) which means that the name of the people of the Angles suddenly gets an Elvish explanation and thereby also the term ’England’, derived from it.
An earlier conception explains ’England’ as Ingilnórë, named after Ingil, son of Ingwë. In a previous version it is called Inwinórë after Inwë (nórë ’land’) (LT1:294). Some additional names of Inwë are: Inweg, Inwithiel, Im (GL:51).
Tolkien took this name from real world mythology, and as it had an obscure etymology he invented an Elvish one, from the root INI- ’small’ (LT:294, QL:42). The Germanic god Ing (Yngve, Ingui), later Freyr is responsible for fertility. It is notable that Freyr means ’lord’ (compare Gothic fráuja, OE. freá ’master, lord’). In the later sources Ingwë is translated as ’chief’, also in the form Ingwë Ingweron ’chief of chiefs’ (PM:340), derived from the stem ING- ’first, foremost, high, noble, lofty’ (Etym, VT45:18); inga ’top’ (VT47:28), thus the Vanyar are the ’high-elves’.
So Ing, Ingwë must be added to the prominent name list of Eärendil, Atalantë, Avallónë (and possibly Danas) in part 5.

7.2  Ireland

Similarly Ireland is called Íverin and its Goldogrin name is Aivrin or Aîvrien (GL:18), in Qenya Īwerin, Iverindor (LT2:352, GL:18). The Celtic root *Iveriu yields Eriu, the name of a Celtic fertility goddess and also the first element in OE Írland, Íraland; the Irish name Éire is related as well. But Tolkien seems to have made up an Elvish origin of the root *Iveriu, cf. G. ivrin ’fertile’. With the ending -(n)dor ’land’ it is transparently the ’fertile land’.

7.3  Warwick

There is the city of Kôr in Valinor and Kortirion on Tol Eressëa, both contain Kôr < Qorā < Guorā, which seems to have something to do with roundness, referring to the hill on which Kôr was built; the second element tirion means ’tower’. But Kortirion is (or later becomes) Warwick (old form: Wæring) and thus Tolkien derives the element War- (Wær-) from the same root, while the real origin of Warwick is not certain (LT2:297,334). It may be that Tolkien imagined an Elvish influence here, closest is the Goldogrin form Gwâr (GL:47).
In the fantasy adventure novel She by Sir Henry Rider Haggard (1856-1925) there occurs the city of Kôr in Africa built on top of an isolated mountain, just like the Elvish Kôr is built upon a lonely hill. This is where Tolkien most probably had got the sound sequence from (cf. the footnote in LT2:335).
In addition Kôr was called Tûn, OE tún means ’an enclosed dwelling’, compare modern English town. In the later Silmarillion Túna is the hill of Tirion, but without any relation to Old English, being derived from pure Elvish TUN- ’hill, mound’ instead.

7.4  Rome

At the end of ’The Fall of Gondolin’ the disaster is compared to similar falls of some cities of the men, namely Bablon, Ninwi (Ninwë), Trui and Rûm, obviously being phonetic adaptations of Babylon, Nineveh, Troy and Rome (Rōma) (LT2:196-197,203). Remarkable is also the mentioned term Rúmhoth ’Romans’ (LT2:299) with the suffix -hoth ’folk, people’, compare orchoth (LT1:304).
Another name for Rome is G. Magbar (LT2:321), evidently containing the root MAKA- having to do with ’sword, slay, battle, slaughter’ (LT1:298, QL:58) and bar ’home’ (LT1:287, GL:21).

7.5  various similarities

But apart from frequent geographical puns Tolkien’s word creation was obviously based on real words in a lot of cases. Christopher Tolkien observes:
It is noteworthy that my father introduced a kind of ’historical punning’ here and there (LT1:282), and he mentions some of these right away:

The root SAHA- ’be hot’ yields Q. Sahóra ’the South’ (QL:81), G. Sacha ’fiery day’ (GL:66) and the similarity to Sahara is quite obvious.

NENE- yields nēnuvar ’pool of lilies’ (QL:65) — modern French nénufar ’water-lily’

G. hôr means ’old, aged, ancient (only of things still existing)’ (GL:49). Compare OE har ’grey, old (and grey-haired)’, Middle English hor and modern English hoary ’ancient, extremely old’.

Similarly G. rûm ’secret (whisper)’ (in LT1:282; GL:66 gives ’a secret, mystery’) — OE rún ’whisper, confidence, secret, mystery’ (a ’rune’ is thus a ’secret sign’ and so in all Germanic languages: ON. rún, Gothic rúna ’secret, mystery’).

Stem HERE- ’rule’ — German Herr ’lord, master, ruler’, OE hér, hér- ’noble, sublime’

These words were completely rejected later with the exception of HERE-; it has been changed to KHER- with the derivates Q. heru, S. hîr ’master’.
But such similarities can be further explored:

Q. Ostor means ’the East, the Sun when she issues from her white gates’ and osto are ’the gates of the Sun’ (LT1:304, QL:70-71) which is a clear analogy to English east, German Ost(en), similar in other Germanic languages. The stem at the early stage is ORO- ’steepness, rising’, probably being from *OSO- (otherwise it cannot lead to medial -st-) and also yields words like órëa ’of the dawn, Eastern’ or oronto ’Sunrise’ (ibid.). There is however also a separate stem OSO- yielding G. ostor ’enclosure, circuit of walls’, ost ’enclosure, yard, town’, Q. os(t) ’house, cottage’ and others (GL:63).
At the later stage the root ORO- remains with derivates like Q. óre ’rising’, but oron (pl. oronti) from the extended form OROT- becomes ’mountain’ while Q. osto, S. ost are assigned to the stem OS- ’round, about’ and have the meaning ’city, town with wall around’. All the words for ’east’ are now derived from RŌ-, a variation of ORO-. Tolkien might have thought here of wiping out the similarities.

The word Silmaril (pl. Silmarilli) exists already in Early Qenya as well as the element sil which is connected to white or silvery light (never abandoned or changed). It reminds immediately of the word silver, compare also Sindarin silivren ’white-glittering’ (LotR II, ch. 1) and the archaic English word silvern which Tolkien used as well (LT2:360). The Goldogrin form is given as silubrill-, silum(b)aril- (LT1:306), Silubrilt or Silobrilt (GL:67) and is a close match to Gothic silubr ’silver’.

The Great Sea between Valinor and Middle-earth is called Haloisi Velikë in the Lost Tales (LT1:292). In Russian великий means ’great’ (-ий being the nom. masc. sg. ending). This is a surprising language to find a match in, since the vast majority of Russian words is not compatible with Qenya or Goldogrin at all.
Probably because of this similarity Tolkien rejected this word. Hence the stem BEL- is, according to The Etymologies, not found in Q – anymore. The Goldogrin form beleg ’mighty, great’ however survived into Sindarin (compare Beleg Cúthalion in the Silmarillion), Tolkien might have judged it to be distinct enough.

G. falcon, changed to falchon means ’a great two-handed sword, twibill’ (LT2:348, GL:33). The falchion is known today as a broad-sword. Nothing like this occurs in the later sources.

G. cwed- ’say, tell’, cweth ’word’, Q. qet- ’speak, talk’, qent ’word’ are derived from the root QETE- (LT2:356, GL:28, QL:77). Compare it to OE cweðan ’say, speak’, Gothic qiþan ’say, tell, speak’ (-an is the infinitive ending, so the stems are cweth-, qiþ-; note that Qenya and Gothic q, OE and Goldogrin cw represent the same consonant cluster [kw]. See also the archaic English past tense quoth and the word quote, Spanish cuento ’story, short story, tale’.
This has never been wholly abandoned. In Quenya, quet- ’say, speak’ with the new spelling has been kept (VT41:11,15), ’word’ has become quetta (WJ:391). But the later Sindarin forms strongly differ from those of Goldogrin due to the new change kw > p, thus #ped- ’speak’ (LotR II, ch. 4, Moria gate inscription), peth ’word’ (ibid., Gandalf’s invocation; also in Etym:KWET-).

Q. lapatte ’hare’ (GL:52) reminds of Latin lepus ’hare, rabbit’ or of the (probably unrelated) French lapin ’rabbit’. It does not occur in later sources anymore.

Q. rinko ’disc, circle, orb’ (QL:80) and G. rinc ’circular (aj.); disc, rondure (n.)’ (GL:65) immediately remind of the word ring, German Ring ’ring, circle’, OE hring ’circle, circuit, cycle, orb, globe’, ON hring-r ’ring, circle’. The word has been altered, so that we find Q. rindë ’circle’, rinda ’circular’; N. rhind, rhinn ’circle’, rhinn ’circular’ in The Etymologies, from the root RIN- (earlier RINI-). These, on the other hand, resemble round, German rund.

Q. is translated as ’crescent moon’ and G. as ’bow, crescent; the waxing or waning moon’ (QL:49, GL:27, LT1:314). In Finnish kuu means ’moon’. In The Etymologies Q. becomes simply the ’bow’; N. ’arch, crescent’. The word ’moon’ is now formed by symbiosis with the stem RAN- ’wander, stray’ (whence also N. Rhân, Q. Rana ’Moon’ and Q. Rána ’the Wayward, the Wanderer’ = ’Moon’ in the Simarillion), thus N. cúran ’the crescent moon’.

The root QALA- is translated ’die’ (LT1:305, QL:76) with Q. qalmë ’death’, qalin ’dead’. Compare OE cwellan ’to kill, slay’, German quälen ’to torture’.
This root KWAL- ’die in pain’ with exactly the same derivatives is also found in The Etymologies, also KWEL- ’fade, wither’, qelet ’corpse’. Interestingly this is avoided in the late Markirya poem (MC:221-223) where loiko in line 20 is preferred (which in its turn similar to OE líc, Gothic leik [li:k] ’flesh, body, corpse’).

Qenya irmin ’the inhabited world – the whole of the created world, not only earth’ (QL:43) is derived from IRI- ’dwell’ < *IÐ-. The OE prefix irmen-, yrmen- and its ON cognate jörmun- denotes something huge and vast, especially Jörmungand, the great serpent encircling the world, and hence Jörmungrund ’the earth’, OE irmen-þeóde ’the peoples of the earth’.

Qenya QEL+U yields Q. qelu ’a well, spring, source’, qelu- ’well up’, qelume ’source, origin’, qĕlŭva ’original’ (QL:76). German Quelle means ’source, origin’, quellen ’to stream out’.

The roots QIV-, QIPI-, KOI̯-(VI) mentioned in GL:29 yield Q. koiva ’lively, living’, koivie ’liveness’; qîva ’awake’, qîvie ’liveness, awakening’, G. cwiv- ’am awake’. Compare Gothic qius ’lively’, fem. qiwa, Engish. quick in the old sense ’living, alive’, quicksilver ’mercury’, lit. ’living silver’, German keck ’jaunty’.

The roots SEHE, SE’E (QL:82) yield Q. ’eye, eyeball’, sehta ’to see’, sie ’sight, sense of sight’. Compare Gothic saihwan ’to see, take heed’, OE seon, German sehen ’to see’.

The root KALA ’shine golden’ produces words having to do with ’day’ or ’light’ and among them kalende ’a special day, festival’. Compare the Roman Kalends.

Q. maksima ’powerful, having possession of or authority over’ (QL:57) imediately reminds of Latin maximus ’greatest, largest, most powerful’, maxime ’greatly, exceedingly, to the highest degree, very’.

The Qenya root GWERE- ’whirl, twirl, twist’ gives ’werin(a) ’round’, ’werelin(d) ’whirligig’, werelinda ’twirling, pirouetting’ (QL:103), G. gwert ’a twist’, gwiril, cwiril ’spindle’ (GL:46) and so on. Compare ON verða, OHG werdan, German werden, Gothic waírþan ’to become’, lit. ’to turn into’, Russian вертеть ’spin, rotate’, all from PIE *wer- ’turn, bend’.

Qenya lilt- ’to dance’, lilt ’a dance’ (QL:55) looks suspiciously close to English lilt ’to sing or speak rhythmically and with fluctuating pitch’, ’to move in a lively springy manner’, as a noun ’(lively song or tune with a) well-marked rhythm or swing’.

Q. moc- ’to hate’ derives from MOKO ’hate’ (QL:62) and may be compared with mock, French moquer.

The name Luthany for ’England’ is taken from the poem The Mistress of Vision by Francis Thompson (1859-1907):

Where is the land of Luthany,
And where the region Elenore?
I do faint therefor.
(cf. also LT2:335)

It is not known, however, where Thompson has this name from. Tolkien gave Luthany the meaning ’friendship’, obviously due to the imagined bonds between the Elves and the people of Eriol/Ælfwine. He also reshaped this sound sequence into Elvish: Leithien (LT2:306) or Leithian (LT2:336) and transformed it into the name Lúthien or Lúsion translating it either as ’friend’ or as ’wanderer’ (LT2:307).
No etymology of these words is given (the words seem to have arisen after the lexicons) and for whatever reason Lúthien and Leithian venture very soon into the story of Beren and Tinúviel. The Lay of Leithian, a long poem about them, is translated ’Release from Bondage’ (LB), also leithian ’release’ in The Etymologies (LEK-), Lúthien is derived there from LUK- ’magic, enchantment’. The originally intended meaning of Leithian must have been ’Land of Release’, this is the translation of Dor Faidwen, the Goldogrin name for Tol Eressëa (LT1:286, GL:33).
Curiously the second word for ’England’ from Francis Thompson, Elenore is interpretable in Q(u)enya as ’star-land’ (but not yet at the earliest stage of the Lost Tales) with the elements el-/elen- ’star’ and -nor/-nóre ’land’ (ELEN-, NDOR-).

For the similarity between Q. purya- ’to set fire to’ (stem PUŘU) - Greek πυρ ’fire’ and related words see Lambengolmor messages #808-810.
For another one involving Early Noldorin, milk and oil, see the message #812.

However, one should never forget that these are just few examples out of several thousand words in the two lexicons. The vast majority of the vocabulary was still invented by Tolkien without any borrowing.
But overall I think that the picture is very clear – Tolkien’s languages went hand in hand with the world they lived in. There was much resemblance and inspiration in the early stage, but as the legends became more evolved in their details, as was the nature of the men and elves, as several essays on philosophical matters occurred, the legendarium became something new and could stand by itself as an epos. And so it was with the languages, they became very unique in style and phonology, in ways of expression, more sophisticated regarding links between each other and connections to the history of their speakers.

8  List of matches between Noldorin/Sindarin and Welsh & Irish

See separate page.

9  List of other matches

See separate page.

References

[A]
Etymologies: Index by Elvish words – language by language from Ardalambion compiled by Helge Fauskanger
http://www.uib.no/People/hnohf/index2/ind2s.htm
[B]
Quenya entries in The Etymologies as a single string (changes: long vowels shortened, spelling de-capitalized, qkw, hll, hrr, hww; -e appended to base verbs, hyphen removed from a-verbs, several words deleted)
[C]
Finnish wordlist used for comparison (changes: long vowels shortened, äe)
[1]
Harri Perälä. Are High Elves Finno-Ugric? http://www.sci.fi/~alboin/finn_que.htm
[2]
Jonas Rosén. Quenya and Finnish http://www.evilrobotics.com/quenya/
[3]
Mark Rosenfelder. How likely are chance resemblances between languages? http://www.zompist.com/chance.htm (My calculations owe a lot to this article, but I have tightened up the matching conditions.)
[4]
Leonid Korablev. The true Elves of Europe http://www.shelltown.net/~dangweth/elfsaga.html
[5]
Völuspá — translated by Henry Adams Bellows http://www.cybersamurai.net/Mythology/nordic_gods/LegendsSagas/Edda/PoeticEdda/Voluspo.htm
[6]
Patrick H. Wynne. Are Goldogrin and Qenya ”primitive”? http://www.elvish.org/Tengwestie/editorials/20040404.phtml
[7]
Andreas Andreou. Quenya: The Influence of the Greek language http://home.agh.edu.pl/~evermind/pdf/andreou.pdf
[8]
Joseph Wright. Grammar of the Gothic Language. 1910
[9]
Geir T. Zoëga. A Concise Dictionary of Old Icelandic
[10]
Gerhardt Köbler. Althochdeutsches Wörterbuch (4. Auflage). 1993
[11]
J. Bosworth and T. Toller. An Anglo-Saxon Dictionary. 1898/2003
[12]
Henry George Liddell and Robert Scott. A Greek-English Lexicon 1940, supplement 1996


update: Jan. 1st 2006 — various new matches added
update: Jan. 8th, 2012— several corrections, link to WALS, even more matches
update: Apr. 22nd 2012 — more matches added
update: Mar. 18th 2013 — more matches added

nanwen


This document was translated from LATEX by HEVEA.