209 From a letter to Robert Murray, SJ.

[Murray wrote to Tolkien asking if 'I could pick your brains about "holy" words'. He wanted to know Tolkien's views on the original meaning of, and relationships between, the various words for 'holy' in the Indo-European languages.]

4 May 1958

These problems concerning the 'original' meanings of words (or families of formally connected words) are fascinating: strictly – that is: alluring, but not necessarily by a wholesome attraction! I often wonder what use (except historical: knowledge or glimpses of what words have meant and how they have changed in fact so far as ascertainable) we gain by such investigations. It is practically impossible to avoid the vicious circle of discovering from word-histories, or supposed histories, 'primitive' meanings and associations, and then using these for tracing histories of meaning. Is it not possible to discuss the 'meaning' now of 'sanctity' (for instance) without reference to the history of the meaning of the word-forms now employed in that meaning? The other way round seems rather like describing a place (or stage in a journey) in terms of the different routes by which people have arrived there, though the place has a location and existence quite independent of these routes, direct or more circuitous.

In any case in an historical enquiry we are obliged to deal simultaneously with two variables each in motions that are independent fundamentally, even when affecting one another 'accidentally': the meanings and associations of meaning are one, and the word-forms another, and their changes are independent. The word-form can go through a whole cycle of change, until it is phonetically unrecognizable without measurable change of meaning; and at any moment without any change in phonetics 'the meaning' of a 'word' may change. Quite suddenly⁶³ (as far as the evidence goes) yelp which meant 'to speak proudly', and was especially used of proud vows (such as a knight vowing to do some dangerous deed) stopped meaning that and became used of the noise of foxes or dogs! Why? At any rate, not because of any change in ideas about vaunts or animals! It is a long way from ὀδοντ- to tooth, but the changes of form have not much affected the meaning (nor has tine the equivalent of dent- moved very far).⁶⁴

We do not know the 'original' meaning of any word, still less the meaning of its basic element (sc. the pan it shares with or seems to share with other related words: once called its 'root'): there is always a lost past. Thus we do not know the original meaning of θέος or deus or god. We can, of course, make some guesses about the formation of these three quite distinct words, and then try to generalize a basic meaning from the senses shown by their relatives – but I do not think we shall necessarily by that way get any nearer to the idea 'god' at any actual moment in any language using one of these words. It is an odd fact that English dizzy (olim dysig) and giddy (olim gydig) seem related to θέος and god respectively. In English they once meant 'irrational', and now 'vertiginous', but that does not help much (except to cause us to reflect that there was a long past before θέος or god reached their forms or senses and equally queer changes may have gone on in unrecorded ages). We may, of course, guess that we have a remote effect of primitive ideas of 'inspiration' (to the 18th C[entury] an enthusiast was much what an Anglo-Saxon would have called a dysiga!). But that is not of much theological use? We are faced by endless minute parallels to the mystery of incarnation. Is not the idea of god ultimately independent of the ways by which a word for it has come to be?⁶⁵ whether through √dh(e)wes (which seems to refer basically to stirring and excitement); or √d(e)jew (which seems to refer basically to brightness (esp. of the sky)); or possibly (it is a mere guess) √ghew cry, – god is originally neuter and is supposed to 'mean' that which is invoked: an old past participle. Possibly a taboo-word. The old deiwos word (which produced dīvus, deus) survives only in Tuesday.⁶⁶

If he has to tackle such a word as holy, the old-fashioned philologist (such as I am) looks first at the history of the form. According to rules laboriously elaborated (and I think certainly valid within limits⁶⁷) he will say what it improbably formally related to. But he cannot wholly escape the quicksand of semantics. Before he proposes a relationship (that is an actual historical nexus of change) between holy and other words in the same language (or in other believed to be related to English) he will want both a phonologically possible kinship, and some 'possible kinship' in sense. All the time he will be uneasily aware of two things found in linguistic experience: (1) that there seem always to have been 'homophones', or 2 (or more) phonetically indistinguishable elements that possessed distinct senses and are therefore 'different words',⁶⁸ like I[ndo]-E[uropean] stems men 'stick out', and men 'think'; and (2) that semantic change is sometimes violent, and in the dark past may have operated without leaving evidence of its occurrence. For instance the formal equivalence of √sequ in Greek ἕπομαι and Latin sequor (and other languages) meaning 'follow' is exact with Germanic sek^w – stem of a verb : but this means 'to see'. Which is to have most weight: the form or the sense? He cannot decide finally on the evidence; though fiddling in an amateur way with 'semantics' he can make the sense-jump seems less impossible than it looks at first, by referring to the uses of 'follow'= 'understand', and to the fact that I-E words for see (as indeed our see) often mean, or the same 'bases' may mean, 'know', 'understand'. (This is particularly true of the √WID base: Latin video has its exact equivalent in O.E. witian 'watch, guard'; but Ϝοἰδα (= Latin vīdī) in O.E. wāt 'wot', 'I know'.) But probably, if he finds Germanic salwo- (our sallow) and Latin salvus (saluos), he will decide that there is no bridge between 'dirty yellow' and 'safe and sound'; so that either some thing is wrong with the phonological equation, or that he is dealing with 'homophones'. (There is always also the possibility that either sallow or salvus did not descend from a common antiquity – words can be invented, or borrowed and may closely resemble older words in either case.) The formal equivalent (the only known one) of our harp is Latin corbis. (The Romance arpa etc. are borrowed from Germanic.) But the poor philologist will have to call on some archaeological expert before he can decide whether any relationship between 'harps' and 'baskets' is possible – supposing Gmc. harpō always meant 'harp' or corbi-s always meant 'wicker basket'! corbīta means a fat-bellied ship.