Edited with introduction, glossaries, and additional notes by Carl F. Hostetter
Tolkien's texts copyright ©1998 The Tolkien Trust
The following is that section of Appendix D, subtitled "*Kwen, Quenya, and the Elvish (especially Ñoldorin) words for Language", to the c. 1959-60 essay Quendi and Eldar that Christopher Tollden excluded from that essay in The War of the Jewels (see WJ:359). The placement of this section in App. D is indicated by Christopher at the top of p. 396 of that volume.
This section comprises most of eight of the typescript pages of the latest version of Quendi and Eldar. These typescript pages have been emended in a very few places by Tolkien in ink, chiefly in correction of typographical errors. In this edition, these changes have all been incorporated silently. Tolkien also added one note to the typescript in red ball-point pen (see Author's Note 5 below). In this edition, Tolkien's text has been reorganized slightly in the matter of notes. As throughout Quendi and Eldar, Tolkien at points interrupts the main text with notes, typed on the line following the notation mark, even where this interrupts a sentence (cf. WJ:359). Christopher's practice in editing Quendi and Eldar of collecting Tolkien's notes at the end of the essay, and distinguishmg them from editorial notes by referring to them in the text with Note 1, Note 2, etc. in parentheses, has been adopted here.
The typescript of Quendi and Eldar was preceded by a complex of manuscript materials, clipped together into two small bundles by Tolkien. These probably did not long precede the typescript, as one of the sheets in the second bundle is a calendar page for November 1959. Among the fìrst bundle are two manuscript versions of what would become Appendix D of the typescript, both of which have been inserted into the bundle between sheets numbered 18 and 19 by Tolkien. The earlier of these versions, titled "Appendix: Noldorin words for Language" and occupying four sides labelled "18(a)" through "18(d)" by Tolkien, differs greatly from the typescript version, and is given in full below, after the typescript version. The intermediate manuscript version, occupying nine sides labelled "a" through "i" by Tolkien, has little textual value, as the typescript follows it very closely, but variants of interest are given in the editorial notes. Some of these variants occur in the portion of App. D published on p. 394 of The War of the Jewels, and can most conveniently be cited here: in the manuscript the stem from which Q. tengwe 'indication, sign, token' was derived is given as *TEÑE, and the intermediate form as *teñgwe; the Quenya phrase translating 'phonetic signs' is given as hlonaiti tengwi; and a tengwesta employing phonetic signs is called a hlonaite tengwesta.
Two editorial glossaries of the Elvish forms encountered in the typescript and in the fìrst manuscript version have been supplied after each text, following Tolkien's notes, as a convenient place for citing further information relevant to them from other texts (especially The Etymologies and the published portions of Quendi and Eldar) and for most of the specifìcally linguistic editorial commentary.
I am grateful to Christopher Tolkien for providing these texts for publication in Vinyar Tengwar, and to Christopher Gilson, Arden Smith, and Patrick Wynne for their assistance in preparing this edition.
[This text immediately follows the last paragraph on p. 395 of The War of the Jewels]
At this period the Loremasters did not, of course, know of the Dwarvish iglishmêk, and were still in Aman limited to the examination of the Eldarin dialects and gestures, enlarged by some acquaintance with the language of the Valar (see note below).[1] They were, however, impressed by the analogy of silent gesture-signs, the component movements of which could be seen; and this much affected their earlier analyses and descriptions of their own language, which thus tended at fìrst to pay more attention to the physical movements made in speaking than to the audible effects, considering the speaker rather than the hearer.
Thus according to the earlier Loremasters a lambe must be analysed into a number of tengwi, which could be used alone or in combinations. In a hwerme more than one visibly distinct gesture could be combined to convey a simple meaning (Note 1). Yet any one such movement might suffìce alone to convey a meaning. So it was in speech. Though it was much more complex and organized and normally employed combinations of lingual or other movements, it could still use for one "word" a phonetic element that could be regarded as simple or uncombined. In this way the early analysts came to use, in dealing with speech, the term tengwe not for a "word", even if separate and uncompounded, as a sign or token of its meaning, but for the separable phonetic movements of which words were composed.
In their views of what constituted an elementary phonetic movement or tengwe they appear also to have been influenced by their theories concerning the origins of Elvish speech, and the way in which the normal "stems" or word-bases had been built up from simpler beginnings. In any case they at fìrst analysed the component tengwi of Elvish speech as being each of the actual or supposed basic consonants followed by one of the basic vowels which (as they said) "coloured" it. The total number of quante tengwi, or full signs, making up a tengwesta was, therefore, the number of its basic consonants multiplied by the number of its basic vowels.
Thus √mata "eat" was composed of tengwe ma + tengwe ta; but this conjunction and order constituted a distinct quetta or word, having no necessary connexion in meaning with, say, √maka (in which the second tengwe only was changed), nor with √tama (in which only the order was changed).
But with regard to Quenya as it was, this analysis had still to account for consonant groups and fìnal consonants (that is, for consonants without any following or colouring vowel); and for vowels occurring alone, especially initially. The Ñoldor at this time already knew much concerning the history of their own language, and it was partly in the light of this knowledge that they dealt with these two points.
With regard to vowelless consonants, they held these in every case to have "lost" a following vowel, and they called them rakine tengwi "stripped" or "deprived signs". For this purpose it was not necessary to distinguish between true "loss" and "omission", that is, between the "unintended" phonetic disappearance of sounds in the course of speaking and of linguistic transmission in time, and the suppression or rejection of sounds in the course of conscious invention and the construction of more complex words. (Eventually the loremasters regarded this distinction as of great impor-tance.)
With regard to vowels without preceding consonants, they proceeded similarly. They had in Quenya a few vocalic monosyllables, as i and ú; and also a few dissyllables that were vocalic, such as ëa "it is" or öa "away". There were also a not inconsiderable number of dissyllabic words that, though they had an initial consonant, had no medial, such as töa "wood" or lie "people". But most diffìcult to accommodate to their theory of basic structure were the many "stems" that had no initial consonant. Beside the pattern which they regarded as normal: XaXa (in which X stands for any basic consonant, and a for any vowel), as for example √mata, there were many of the type -aXa, such as √ara.[2] The Loremasters asserted that these vowels must have "lost" their preceding consonants; but since they held that a consonant was a more vital part of a tengwe than the colouring vowel, they called such unaccompanied vowels penye tengwi, that is "lacking" or "inadequate signs".[3]
The assumption of loss, in these two cases, was not, however, due merely to theory. It was in many words the actual historical explanation of the occurrence of vowels without a preceding consonant (in the case of medial hiatus probably always the explanation); it was also often the true explanation of the contact of consonants without intervening vowels, while all fìnal consonants had probably lost a final vowel, if remote Quendian origins were considered. The Loremasters were aware of this. Some consonants had been lost within the recorded history of Quenya in Aman; or their former presence could be detected by the examination of Quenya and by a comparison with the Telerin dialect. Many contemporary fìnal consonants had following vowels in older periods, or were clearly related to words that still had following vowels.
For instance, Mandos from older Mandosto beside osto from older ostŏ (the shortening and subsequent loss of fínal vowels in the second elements of compounds being a frequent feature of the earlier stages of Quenya).[4] The former presence of intervocalic ñ, later lost in Quenya, could be detected by consideration of the relations between tëa "indicates" and tenge "indicated", tengwe "sign", and comparison with ëa "exists" beside engwe "thing". The former presence of initial g could be detected by the comparison of, say, Q. alda "tree" with T. galla,[5] and the process of loss be deduced from the spelling in the old Rúmilian script ʒalda (using an initial sign which was known by tradition among the loremasters to have represented the open backspirant).
At this period, therefore, the loremasters represented the penye tengwi as dependent upon a sign for a lost consonant; and for this purpose they used the sign in Rúmilian for old [ʒ] : ;[6] though they did not intend by this to assert that the missing consonant was always due to a reduction of g, or that the loss had only occurred since the divergence of Quenya and Telerin. There were many cases in which Quenya and Telerin agreed in having no initial consonant.
With regard to the penye tengwi,[7] however, the early loremasters still had difficulty with the diphthongs. They were obliged indeed to include among the basic vowels the diphthongs ai, au (calling them "double" or "blended colours"). For it was plain that in many cases ai, au existed as modifìcations or "strengthenings" of simple i, u, which could not reasonably be explained by consonantal loss, whereas all other diphthongs arose only from the "deprivation" of the vowel that had once followed after consonantal j and w. For example, tuile "spring" in relation to *TUJU "sprout, bud". This was analysed as tu-yu-le. ai and au could of course also arise in the same way, and were then similarly analysed: e.g. taina "stretched, elongated" from *TAJA "stretch", and taure "forest" from *TAWA "wood";[8] these were analysed as ta-ya-na and ta-wa-re. But raika "crooked" from *RIKI "twist", and nauka "stunted" from *NUKU, were analysed as rai-ka and nau-ka.
This theory of loss was thus in very many cases historically justifìed, and was a reasonable account of how word-shapes had been invented, or constructed by combination, and of how they had been modified in transmission. But it was in many points erroneous historically. The initial consonant-groups, for instance, were certainly not, in most cases, produced by vocalic loss; neither was the important feature in Quenya of medial nasal-infixion, as in mante "ate" in relation to √mata. The vowels also had plainly in the formative period been employable by themselves as signifìcant elements; and the fact that there were few independent words of purely vocalic structure, though vowels were largely used as modifying additions to basic stems, was due simply to the small number of vowels.
This was the situation when Fëanor, early in his career, turned his attention to matters of language and writing. It is said that he soon advanced far beyond the loremasters of his time. He made collections of all the available lore, oral and written, concerning Quenya in earlier days, and studied in detail its relations with Telerin. He is said also, being then in his youth before the days of his discontent, to have learned "more than any other of the Eldar in Arda" of the language of the Valar. This he got mostly from Aule (Note 2), and so enlarged his view by experience of a tongue wholly different in sounds and structure from his native language. But Fëanor soon turned to other matters; and in any case his primary interest was in writing, in its practical and its decorative aspects rather than as an accurate phonetic transcription. Not that he was without interest in phonetic analysis. He was indeed superior in this department to any of his predecessors; and the alphabet, or alphabetic system, that he devised[9] provided the means of expression for many more individual sounds than those that actually existed in Quenya or Telerin. Though being primarily made for their expression, it was naturally largely conditioned by the phonetic character and range of these languages.
In the mode that he propounded for the practical representation of Quenya he made use of the syllabic analysis of his predecessors, already embodied in the older Rúmilian script, but he did this chiefly for the sake of compactness and brevity. The basic "letters" were consonants, and vowels were indicated by diacritic signs, usually written above the preceding consonant (that is, according to the older terminology, indicating its "colour"). Where a vowel had no preceding consonant use was made of the device already mentioned, by which the vowel signs were attached to the letter . But this no longer had in the Fëanorian system any consonantal value, and became merely a "carrier" for convenience in writing. Fëanor indeed repudiated the theory that penye tengwi[10] were always due to consonantal loss.[11] (Note 3)
Fëanor, therefore, in spite of the usual mode of spelling, held that vowels were each independent tengwi, or word-building elements, though different in functions. Vowels he called óma-tengwi or ómear'y and consonants ñáva-tengwi or ñávear.[12] That is, those chiefly dependent on resonance of the voice, and those chiefly dependent on the movements in the mouth (including the lips)[13] (Note 4). For ñávear he later substituted the invented word patakar, taking p, t, k (as in his alphabet) to represent the chief positions of consonantal contact or friction.[14]
Fëanor actually devised "for the Loremasters" separate independent letters for the vowels, distinct from the tehtar.[15] This quanta sarme or "full writing" was indeed mainly used by the Loremasters for special purposes, until later in Middle-earth the Fëanorian letters were applied to other languages, such as Sindarin, in which the diacritic method of indicating vowels was inconvenient.[16]
Among Fëanor's other opinions or discoveries two may be mentioned. Both proceeded from his conception of the process which he called "strengthening" (antoryame). He said that it was plain from an examination of the inter-relations among stems, and among the derivatives of any one stem, that the "word-builders" had deliberately enriched or "strengthened" this or that component sound (according to its own character), for emphasis, or merely for differentiation. The simplest cases were those in which a sound had simply been lengthened: as in the relation of *mātā (the stem of the continuous form "is eating") to the supposed stem *mata. Where the vowel was lengthened the process had been disregarded; but Fëanor held that it was not different in principle from such cases as *grottā beside *grotā from the stem *(g)roto,[17] or *lassē, Q. lasse "leaf", in which the medial consonant had been lengthened. It was absurd to analyse these as go-ro-to-ta, go-ro-ta, la-sa-se. Their true relations to the simplest forms were *rot > +RoT-ā; *las > laS-ē. (Note 5)
In gr- as a variation of r- Fëanor saw a case of another method of strengthening: the construction of what he called "blends" (ostimi). That is, the running together of two elements, that could be analysed phonetically (without reference to intention or effect) as separate gestures or movements, into a combination that had and was intended to have a unitary effect and significance. Other examples that he cited were the relations between initial st- and s-, or t-; gl- and l-; ky- and kw- and k-. But he was particularly impressed by the nasal combinations, notably mb, nd, ng, and pointed out that these last were enormously frequent in Quenya as strengthenings of both m, n, ñ and b, d, g, being in fact preferred to the simple lengthenings mm, nn, ññ, and far more frequent than bb, dd, gg. For example, he said that rondo "vaulted hall" was not ro-no-do, nor even ron+do, but in relation to the stem *ron was roNo with "blend" (ostime) or (consonantal) diphthong (ohlon) used for the strengthening of n.
Fëanor used the same terms, ostime or ohlon, for vocalic diphthongs....
widez2