Elfique Primitif - où tout commença

HISTOIRE INTERNE

Longtemps ils habitèrent dans leur première demeure près de l'eau sous les étoiles, et ils marchèrent sur la Terre en admiration; et ils commencèrent à faire un langage et à donner des noms à toutes les choses qu'il apercevaient. Ils se nommèrent eux-mêmes les Quendi, ce qui signifie ceux qui parlent avec des voix...

En créant un langage, les Premiers-nés d'Ulúvatar s'identifièrent eux-mêmes comme les Incarnée, enfants de l'Unique: "L'invention d'une lambë [langage] est le caractère principal d'un Incarné, observa Pengolodh le sage de Gondolin (WJ:397). En fait "les nouvellement éveillés inventèrent beaucoup de nouveaux beaux mots, et beaucoup d'artifices de langages malins" (WJ:422). Le langage fait par les premiers Elfes à Cuiviénen allait avoir un immense impact sur l'histoire linguistique de la Terre-du-Milieu. Habituellement appelé Quendien Primitif, c'est le tout premier ancêtre de tous les langages Elfiques, y compris le Quenya et le Sindarin. Même les langages qui ne descendent pas directement du QP ont emprunté des mots à l'Elfique, comme documenté dans les articles sur l'Adûnaïque, le langage Humain, le Nanique et même l'Orkish et le Noir Parler. WR:159 et PM:63 citent Faramir disant que "tous les langages des hommes dans le monde sont des descendants de l'Elfique". Le seul langage d'Arda qui n'ait rien à voir avec l'influence Elfique peut être le Valarin. Dans la pratique, le Valarin serait aussi le seul langage antérieur au Quendien Primitif. En réalité, Aulë avait inventé le Khuzdul pour les nains longtemps avant la venue des Elfes, mais puisqu'Ilúvatar n'a pas voulu que les Nains viennent avant ses Premiers-nés, les Nains dormaient toujours quand les Quendi se réveillèrent.

A quoi ressemblait-elle, la langue que les Elfes formèrent dans les premières années d'innocence près des eaux étoilées de Cuiviénien? Nous connaissons beaucoup de sa phonologie et des méthodes de dérivation; nous connaissons moins sa structure grammaticale précise. A en juger par les langages dérivés qui sont communément tenus pour descendre du langage original, nommément le Quenya et le Telerin, le QP était un langage de cas; une terminaison allative -da est explicitement mentionnée par Tolkien (WJ:366). Concernant le style général du langage primitif, la vaste majorité des mots avaient deux ou trois syllabes et se terminaient par une voyelle. Dans VT39:6, Tolkien dit que en Quenya, "toutes les consonnes finales ont probablement perdu une voyelle, si l'on considère les origines lointaines Quendiennes ". (Quelques unes des formes "reconstruites" de Tolkien's se terminent cependant pas une consonne, mais il y en a peu, et toutes les formes astérisquées ne décrivent pas nécessairement la plus ancienne étape du langages.)

Les fréquentes longues voyelles finales sont très caractéristiques de l'Elfique Primitif, ex: dans lindâ "son doux" ou ndorê "pays". Dans des mots tri-syllabiques, la première et la seconde voyelle sont habituellement identiques (ex. karani "rouge"), et dans un certain nombre de cas, la voyelle finale est aussi la même, mais longue (ex. eredê "semence", galadâ "arbre", kyelepê "argent", ñgolodô "Noldo"). selon VT39:6, les voyelles en hiatus ne se trouvent pas en position médiane dans le langage primitif ; là où elles apparaissent en Quenya une consonne a été perdue.

 

 

Tolkien's reconstructed forms nonetheless include at least one hiatic combination, ie, iê; in light of his later concept we may assume that this represents even earlier *ihe or *iñe with a middle consonant that was later lost.
          In the Silmarillion Index, Christopher Tolkien refers to Quenya as "the ancient tongue, common to all Elves, in the form that it took in Valinor". However, the style of "the ancient tongue" in many respects differed markedly from later Quenya, and generally the word Quenya should not be applied to it at all. True, the sound-changes that separate Primitive Quendian from classical High-Elven are so tidy and straightforward that a speaker of Quenya might, with a little training, have been able to understand the primitive language without actually "learning" it as a foreign tongue. But even so, the primitive language would sound pretty outlandish to him, and he would hardly recognize it as a mere variant of his own tongue. Still, the fact remains that the Noldor held that Quenya was the language "most nearly preserving the ancient character of Elvish speech" (WJ:374). Actually the most conservative language seems to be the Telerin of Aman, as least as far as phonology is concerned - but then Telerin was sometimes thought of as a dialect of Quenya, though the Teleri themselves held it to be a separate language.
          Tolkien distinguished two stages of primitive Elvish. The very first stage, as noted above, was Primitive Quendian. This was the ancestor of all Elvish tongues in the world (except, possibly, any products of wholesale a priori language construction, if the Elves engaged in such sports...as we know, some humans do! However, it is said in the Silmarillion, chapter 17, that "all the languages of the Quendi were of one origin"). In the Etymologies, only a few of the asterisked forms are explicitly identified as Primitive Quendian (atar, atû, dêr/der-, khalatirnô, mâ3/ma3-, and taurâ; see the entries ATA, NI1, NÊR, TIR, MA3, TÂ/TA3). Nonetheless, most of the asterisked forms must be assumed to represent the most primitive stage of the language. The next stage was Common Eldarin, the ancestor of all the Eldarin (as opposed to Avarin) tongues, including Quenya and Sindarin. Common Eldarin would be the language spoken by the Elves that followed Oromë and embarked on the March from Cuiviénen to the Sea, or rather the language they developed during the March. In the Etymologies, only three words are explicitly identified as (Common) Eldarin (mahtâ-, ndæ^r, wa, see MA3, NDER, WÔ...unfortunately the computer can't place a circumflex above æ). However, a number of Common Eldarin forms are given in WJ and PM.
          It might be helpful to know approximately how much time the periods involved represent. In WJ:5-6, a chronology in Valian years is set out. In WJ:20 we are told that 365 "long years of the Valar" equal "well nigh...three thousand and five hundred years of the Sun", sc. one Valian year is about nine and a half solar year. Using this figure, we get the following result: After the Elves awoke by the mere of Cuiviénen, they dwelt in peace for about 280 solar years (Valian Years 1050 to ca. 1080). Then they were found by the spies of Melkor and afflicted by them. About fifty more solar years passed, after which the Elves were found by Oromë in the Valian year 1085. The Separation of the Quendi into Eldar and Avari eventually followed, apparently in the year 1105, about 190 solar years later. (The impression one might get from the Silmarillion text, that the Separation occurred within weeks or months after Oromë's finding the Elves, turns out to be wholly wrong.) So from the Elves awoke until the Separation, well over five hundred solar years elapsed, ample time for developing a complete language - but still not very long by Elvish standards. (Cf. Legolas' words in LotR2/III ch. 6: "Five hundred times have the red leaves fallen in Mirkwood in my home since [the Meduseld was built], and but a little while does that seem to us." Half a millennium was not perceived as a very long time by the Elves.)
          The March from Cuiviénen to the Sea lasted well over two and a half century of solar years (Valian Years 1105-1132). During this time, the Marchers turned Primitive Quendian into Common Eldarin. Then the Vanyar and Noldor went over the sea, and about this point Common Eldarin evidently became Old Quenya, as the pre-record period in the evolution of Elvish was nearing its end. In Beleriand, Common Eldarin (or the Common Telerin dialect of it) started to evolve towards Sindarin.
          Millennia later, it was also in Beleriand that the exiled Noldor started to study comparative linguistics and reconstruct the primitive language: "It was...the contact with Sindarin and the enlargement of their experience with linguistic change (especially the much swifter and more uncontrolled shifts observable in Middle-earth) that stimulated the studies of the linguistic loremasters, and it was in Beleriand that theories concerning Primitive Eldarin and the interrelation of its known descendants were developed." - PM:342.

EXTERNAL HISTORY

There seems to have been no point where Tolkien's Elvish languages existed in a historical vacuum, with no history of change and evolution behind them. Notes Christopher Tolkien, "Those languages were conceived, of course, from the very beginning in a deeply 'historical' way... Every élément in the languages, every élément in every word, is in principle historically 'explicable' - as are the éléments in languages that are not 'invented' - and the successive phases of their intricate evolution were the delight of their creator... They image language not as 'pure structure', without 'before' and 'after', but as growth, in time." (LR:341) The diachronic aspect was very important to Tolkien; he seems to have been immensely attracted by the grand vision of an entire language family growing, changing, evolving, branching out in various directions. When discussing Elvish words, he very often introduces primitive stems and asterisked "ancestral" forms; he clearly felt that a word needs a history! (See for instance Letters:382-383.) Of course, this necessitated the construction (or at least sketching) of a primitive language, an ultimate ancestor for all the successive stages, since Tolkien could not well extend the history of the Elvish languages indefinitely into the past - especially when he thought of Elvish history as having one definite beginning in time and space, the awakening of the Quendi by the mere of Cuiviénen. All forms of Elvish had to be descendants of "Cuiviénenian".
          Already in the very first Elvish wordlist, the "Qenya Lexicon" of 1915, the words were derived from "primitive roots" (as in the Etymologies). These stems provide glimpses of a proto-language that seems somewhat inspired by the proposed reconstructions of Indo-European, the hypothetical tongue that most European and some Oriental languages descend from. For instance, Tolkien included stems involving syllabic N and L, such as SNKN and FLKL (LT2:341, where there are dots under the N's and L's to indicate that they are syllabic). Apparently starting from the same primitive stems, Tolkien two years later derived a new Elvish language, cognate with "Qenya" - Gnomish, a Celtic-sounding Elvish language that after thirty years of revisions and changing conceptions "ultimately became that of the type called Sindarin" (PM:379).
          The syllabic consonants of the "Proto-Elfin" of 1915 were gone twenty years later, when the Etymologies was written. Nevertheless, some ideas about the primitive language go back to the very beginning. For instance, there is the notion that many words originally began with nasalized explosives nd, mb, ng (that are reflected in the form that word has following the definite article in the Celtic-sounding language: Gnomish Golda "Gnome, Noldo", i Ngolda "the Gnome"; similarly Sindarin Golodh "Noldo", i Ngolodh "the Noldo"). In the Etymologies, quite a few "reconstructed" primitive words are given, allowing us to get a relatively good impression of the primitive language as Tolkien had now come to think of it.
          Of course, Tolkien's prime interest lay in the later Elvish languages, especially Quenya and (Noldorin >) Sindarin. Even in the Etymologies, primitive Elvish remains a somewhat shadowy entity whose prime function is to clarify the relationship between the various branches of Elvish and serve as the historical basis of them all, rather than being an "art-language" in itself. Like everything else, Tolkien's ideas about the ultimate derivation of certain words were subject to revision. For instance, both SD:419 and the Etymologies (stem TYUL) agree that the Quenya word for "mast" is tyulma. But selon to the Etymologies, tyulma comes from primitive tyulmâ, while SD:419 has it that tyulma descends from primitive kyulumâ. Both of these would yield tyulma in Quenya, so there is no discrepancy concerning the sound-changes - but Tolkien's ideas about the ancestral form changed over time. A similar case is Quenya findë "tress, braid of hair": does it come from primitive spindê (The Etymologies, stem SPIN) or from phindê (PM:362)? Such indecision goes back to the very beginning: Discussing the earliest "Lexicons", Christopher Tolkien notes that "in some cases it seems clear that the word was 'there', so to speak, but its etymology remained to be certainly defined, and not vice versa" (LT1:246). But Tolkien's general ideas about the primitive language, as exemplified in the Etymologies, seem to have gotten into shape in the 1930s and did not undergo substantial revisions later. For instance, in what Christopher Tolkien calls "a very late note" - evidently meaning that it dates from the seventies - the primitive form of Aulë's title "world-artificer" is given as mbartanô (LT1:266). This seems to be the same kind of Primitive Elvish as the reconstructed forms in the Etymologies, written forty years earlier. In any case, the historical development of Quenya and Sindarin had become "minutely refined" in the last years of Tolkien's life, so he must then have held in mind a quite clear image of their common ancestral language. - PM:367.
          It should be noted that one early idea was rejected later: the notion that the Elves did not invent language on their own, but learnt Valarin from Oromë (LR:168). As we have seen, Tolkien later decided that the Elves were alone for centuries before they were found by the Valar.
          Needless to say, the distinction between "recorded" Elvish words and "unattested" forms is pure fiction. Tolkien's "reconstructed" forms are just as authoritative as the vocabulary of the "attested" languages: Even if someone could come up with a more plausible reconstruction of Primitive Quendian than Tolkien did, it would still have to be rejected! In this essay, primitive words "reconstructed" by Tolkien himself are not asterisked - though Tolkien usually does asterisk them, cheerfully playing his Game.

(The ultimate experiment in Tolkienian linguistics: Teach Primitive Quendian to a few thousand people and place them on a remote continent all by themselves. Then come back a millennium or two later and check if their descendants have developed languages similar to Quenya and/or Sindarin.)
         

PRIMITIVE QUENDIAN: BASIC PHONOLOGY

The sounds

The voyelles (monophthongs) of Primitive Quendian were a, e, i, o, u, short and long. The long voyelles are usually marked with a macron by Tolkien; here we will use a circumflex instead: â, ê, î, ô, û. As noted above, the frequent long final voyelles are very characteristic of primitive Elvish. (However, the final voyelles are sometimes - but not always - shortened if the word appears the final élément in a compound; compare tûrô "lord" with -turo in Spanturo "cloud-lord"; see also WJ:403 concerning khînâ "child" becoming -khîna. The plural terminaison -î remains long, though: kala-kwendî "Calaquendi".) The primitive diphthongs were apparently ai, au, ei, eu, iu, oi, *ou, ui. (*Ou is not explicitly given in any primitive word "reconstructed" by Tolkien, but it is implied to have existed, though it seems to have merged with au very early. Cf. Quenya nausë "imagination"; the root is given as NOWO, so originally, the initial syllable of this word must have been *now- = *nou-.) Combinations like âi may be taken either as "long diphthongs" or as â followed by i (two distinct syllables); we cannot know precisely what Tolkien intended. In VT39:11, Tolkien states that "Eldarin (and probably Primitive Quendian) exhibited a marked preference for diphthongs terminaison in i".
          As for the consonants, an interesting table of "The original Quendian consonants" was published in VT46:28. Here, Tolkien set out the entire consonant system as a grid organizing the consonants into three "series": the p-series, t-series and k-series.

  • First there are the three "Voiceless stops" that give their noms to their respective "series": p, t, and k.

  • Next Tolkien listed the corresponding "Voiceless aspirated stops", ph, th, and kh. This would be p, t, and k followed by an h-like sound, somewhat as in English uphold, outhouse, elkhorn. (The "normal" p, t, and k of Primitive Elvish can thus be inferred to have been unaspirated in all positions, as in French or Russian; in English these consonants become aspirated initially before a voyelle, though most speakers are little conscious of this.) Notice that Quendian ph, th, and kh were perceived as unitary consonants, not analyzed as clusters p + h etc.; in VT46:28 Tolkien actually used the single Greek letters phi, theta and chi to denote them. These Greek letters originally denoted aspirates, only later becoming spirants (sc. ph = f, th as in English think and kh like the "ch" of German ach - a similar development from aspirates to spirants is also observed when comparing Primitive Elvish to the later Elvish languages, Quenya and Sindarin).

  • Next are listed the voiced stops b, d, and g (the voiced counterparts of p, t, and k). Below these voiced consonants Tolkien mentioned the "nasal-stop groups" mb, nd, and ng, listing them in brackets. It is not altogether clear whether these are to be regarded as unitary consonants, nasalized versions of b, d, and g, or as b, d, and g with a distinct nasal élément intruding before them. The fact that Tolkien refers to these combinations as "groups" (hence consonant clusters?) probably hints that the latter interpretation is correct.

  • Next come the "Nasal sonants" in their own right: m of the p-series, n of the t-series and ñ of the k-series (if we use the symbol ñ for "ng" as in English ring, like Tolkien often does). Incidentally, the above-listed "nasal-stop group" ng would properly be ñg.

    Finally, Tolkien listed "Oral sonants, or weak voiced spirants". These are the semi-voyelle w of the p-series, the lateral/vibrant sounds l and r of the t-series, and the semi-voyelle j (= English "y" as in you) of the -series. Together with j Tolkien also listed 3, the spirant counterpart of g (like the Orkish gh).

    Tolkien added: "Outside the tripartite system [sc. the p-, t, and k-series] was the one voiceless spirantal sound s. To some extent it belonged to the t-series." He also remarked that this s had a voiced "variety" z, but this was an allophone of s rather than an independent phoneme. The actual primitive forms cited suggest that z occurs primarily where s comes into contact with a following d, as in the "root" MIZD (LR:373, compare the apparently related root MISK listed on the same page). EZDÊ must be taken as a Common Eldarin form, in light of what Tolkien says in WJ:403.

    We notice that ignoring the sibilant s, the primitive language had no other spirant than 3, and even this sound Tolkien often reconstructs as h instead (in LR:360, the original stem yielding Quenya ho "from" is given as 3Ô, while in WJ:368 the corresponding stem is given as HO). Since we are dealing with a reconstructed form of Elvish, the exact quality of this sound is of little importance.
              As for the semi-voyelle j, one detail of spelling must be remembered: When editing the Etymologies for publication, Christopher Tolkien changed J to Y, ex. KUY, DYEL where his father actually wrote KUJ, DJEL (see LR:346). This was done with good intentions, since many speakers of English would misunderstand the letter J, thinking that it referred to the English "dzh"-sound. We retain this revised spelling when referring to the basic stems listed in the Etymologies (in capital letters), but otherwise we henceforth restore Tolkien's original spelling in the actual word-forms mentioned in the Etymologies, ex. njadrô instead of nyadrô (therefore, the reader should not be confused when njadrô is derived from a stem NYAD, since the letters j and y in all cases refer to the same sound). In the essay Quendi and Eldar, where many reconstructed forms occur, Tolkien also used j rather than y, and here Christopher Tolkien left his father's spelling alone when editing the essay for publication. We also use j in primitive words where it seems that Tolkien did employ the letter y, to have a uniform spelling. - Some surmise that j and w often do not stand for independent semi-voyelles, but merely indicate that the preceding consonant is palatalized or labialized, respectively. If so, such palatalized and labialized sounds could be counted as independent phonemes in Primitive Quendian; yet their absence from the table of primitive consonants in VT46:28 may indicate that we are really dealing with consonant clusters terminaison in j and w.
              In the Etymologies, Tolkien in a few cases changed w to v, the stems WAY, WEY becoming VAY, VEY. Does this mean that he considered introducing v as a primitive sound, as distinct from b or w? The sound v does not fit the phonology very well: As noted above, Primitive Quendian possesses virtually no spirants if we disregard the sibilant s; even the spirant 3 is often reconstructed as h instead. Perhaps v as a distinct phoneme in Primitive Quendian was just a passing idea; it does not occur in the table in VT46:28.

    Initial clusters

    The largest group of initial clusters begin in s: sj-, sk-, skj-, skw-, sl-, sm-, sn-, sp-, sr-, st-, sw-.
              Some initial clusters may be considered simply nasalised stops: mb, nd, ñg. Already in the Gnomish Grammar (1917), Tolkien speaks of "words beginning with nasalized-explosives nd, mb, ng (a fairly numerous class originally)" (Parma Eldalamberon #11, p. 7). As noted above, it is not entirely clear whether these are to be perceived as unitary nasalized consonants or as clusters of a nasal + another consonant.
              A number of clusters end in one of the two semi-voyelles. In J: dj, gj, kj, khj, ndj, ñgj, nj, tj (and sj, skj already mentioned). In W: gw, ñgw, kw (and skw, sw already mentioned). It would seem that kw already before the Separation merged into a single labio-velar sound q that remained in Quenya (later spelt qu), while it very early became p in the dialect of the Teleri - still so in Sindarin and the Telerin of Aman. Some would interpret kw as a single labio-velar rather than a consonant cluster right from the start. (Tolkien's earliest "proto-Elfin" - the 1915 stuff - included stems like QORO; see LT1:264. Here, Q does stand for a labio-velar sound. See also ereqa in the wordlist below.) As mentioned above, some would indeed take several of the initial "clusters" involving a single consonant + -w or -j as merely a convenient way of spelling labialized and palatalized series of consonants; if so they would not really be clusters at all. Yet the absence of such consonants in the table in VT46:28 may indicate otherwise.
              SD:419 mentions a primitive word with initial hj (or hy, as it is there spelt). Is this a genuine cluster h + j, or simply hy as in Quenya, a unitary sound like German ich-Laut?

    Stress

    In the Etymologies, a number of reconstructed primitive words include an accent that apparently marks the stressed syllable (here we use italics instead of an accent mark). Most of the words are marked as accented on the first syllable: abarô (abaro), alâkô, balâ, balâre, Banâ, banjâ, bata (batâ), belek, belekâ, berja, boron-, b'ras-sê, orku, pheren, telesâ, ûbanô (see wordlist below for the meaning of the words). Other words are apparently stressed on the penultimate syllable: baradâ, ontâro, berekâ, morokô, turumbê. Yet other words are stressed on the final syllable: barasâ, barjâ, barnâ, battâ, khalnâ, tambâ. From these examples it is clear that in Primitive Elvish, accent was not determined by the form of the word (as is generally the case in Quenya and Sindarin). The words belekâ, berekâ and barasâ have the same number of syllables and exactly the same distribution of consonants and voyelles (short and long), but they are not stressed on the same syllable. There seems to be no certain way of predicting which syllable receives the accent in Primitive Elvish; we just have to take Tolkien's word in this matter. Some stems in the Etymologies, like MORÓK, are marked with an accent to indicate which syllable is stressed - and this is reflected in the derived word morokô. The stem MORÓK just happens to be accented on the second o, and that's it.
              It may be noted that there is no connection between accent and long voyelles. One might think that the frequent long final voyelles were accented, but there seems to be no such rule. In alâkô, the one short voyelle is also the one that is accented. Unlike the present writer, the early Elves apparently did not find it difficult to pronounce long voyelles that were wholly unaccented.
             

    Possible phonological restraints

    As far as can be told, the oldest form of Elvish had few phonological restrains, especially compared to the far more well-defined (and hence restrictive) phonologies of Quenya and Sindarin.
              At the very oldest stage, it may be that all words had to end in a voyelle: We have already quoted VT39:6, where Tolkien states that in Quenya, "all final consonants had probably lost a voyelle, if remote Quendian origins were considered". Yet, as we also indicated above, a few of Tolkien's "reconstructed" forms do end in consonants. Maybe these forms are not meant to reflect the very oldest stage.
              While there seems to be no correlation between long voyelles and stress, it may be that long voyelles cannot occur before consonant clusters. Two long voyelles in sequence (hiatus) may also be impossible.
              selon to VT47:35, describing the situation in Common Eldarin, the semi-voyelle w could not occur before the corresponding voyelle u, and similarly y (= j) could not occur before i.
              Whereas the primitive language did allow initial mb-, nd-, ñg-, the corresponding unvoiced combinations mp-, nt-, ñk- apparently could not occur at the beginning of words. Thus as an elaboration of the root TAM we find NDAM rather than *NTAM, apparently an impossible form (LR:390, 375).

    VARIOUS PARTS OF SPEECH AND THEIR INFLECTION

    Nouns: The primitive language distinguished at least three numbers: singular, dual and plural. The singular was apparently the basic form of the noun, as in most languages. The dual was formed with the terminaison -û, seen in besû "married pair" (BES), lasû "ears" (pair of ears, two ears of one person) (LAS2) and peñû "pair of lips" (VT39:11 cf. 9). If the use of this dual élément corresponds to its use in old Quenya, as outlined by Tolkien in Letters:427, this primitive dual applied only to two things belonging together as a pair, not to two things only casually associated.
              The normal plural terminaison was -î, the origin of Quenya -i (as in Quendi) and the i-affection seen in Sindarin plurals (like annon "gate", pl. ennyn, because a and o were assimilated to the Old Sindarin plural terminaison -i, later lost, and became e and y, respectively). Quendi descends from kwendî, the pl. of kwende (WJ:360); note that the short final -e is displaced by the plural terminaison. The frequent long final voyelles are apparently not normally displaced, but the plural terminaison -î is shortened to -i when added to a long voyelle: The pl. of Lindâ "Linda, an Elf of the Third Clan" (WJ:380) is given as Lindâi (WJ:378), not **Lindâî. It seems that these combinations of a long voyelle + i tended to become normal diphthongs in -i, like âi > ai in this case; the pl. of Lindâ is also given as Lindai (WJ:385). In SD:302 the pl. of ornê "tree" is likewise given as ornei, not *ornêi (the earlier form?) However, sometimes the plural is formed directly from a naked stem instead of being added to the final voyelle; thus, the pl. of balâ "Vala" is balî, formed from the stem BAL, instead of **balâi, **balai. (In Quenya, the form Vali, from balî, is still an alternative to Valar as the pl. of Vala. It is seen in the nom Valinor, the land or people of the Vali.)
              Another primitive plural terminaison, mentioned in the Etymologies under 3O, was -m. How and where it was used is not clear. It may have been used to indicate plurality after case terminaisons and enclitic particles. This -m is apparently the origin of the plural terminaison -n seen in some of the Quenya cases, such as the terminaison -ssen for locative plural (singular -ssë). The prepositional élément jô, jo- "together" (of more than two) is also given as jôm, jom- (WJ:361). It may be that this has something to do with the plural terminaison -m. Vinyar Tengwar #42 provides a little more information about it. Tolkien referred to this terminaison as "being certainly an ancient plural indicator in Common Eldarin" and cited the example lepem "fingers" (p. 26). Yet the normal plural indicator must have been the terminaison -î, directly reflected by -i in Quenya, Telerin and Old Sindarin.
              It would seem that the primitive language had at least some cases; Tolkien mentions an allative terminaison -da (WJ:366). The accusative found in archaic Quenya, formed by lengthening the final voyelle of words (cirya "ship" > ciryá), may suggest that at an earlier stage, there was an accusative terminaison that consisted of some guttural sound. When it was lost, the previous voyelle was lengthened (or remained long) in compensation: ?kirjâ3 > ciryá; contrast *kirjâ > cirya. However, some of the numerous case terminaisons in Quenya may be particles that were later suffixed; we know that the genitive terminaison -o descends from an originally independent particle 3o or ho, "from". Indeed the distinction between case terminaisons and enclitic particles may have been vague or absent in the earliest forms of Elvish. Interestingly, Tolkien states that "prepositional" éléments were normally "attached" (= suffixed?) to noun stems in QP; this was their "usual place" (WJ:368). It would seem that in QP, the "prepositions" normally acted as postpositions instead. (Real prepositions must have become dominant in Common Eldarin, since they occur in both Quenya and Sindarin.)
              Verbs: There isn't too much we can say about the verbal system in the primitive language. Some frequent verbal terminaisons, such as -jâ and -tâ (whence Quenya -ya, -ta) can be identified; see "Derivation in Primitive Elvish" below. WJ:415 suggests that in the primitive language, the past tense was marked by "the 'augment' or reduplicated base-voyelle, and the long stem-voyelle". Thus, the stem KWE "say, speak" had the past tense ekwê (the e of KWE being prefixed as an augment and the original e being lengthened to ê). The stem KAR "make, do", which stem might probably just as well be given as *kara, similarly has the past tense akâra "made, did". Similarly, we must assume that the past tense of kiri "cut" was *ikîri (my reconstruction), and so on. In the later languages, the prefixed stem-voyelles live on in the Quenya perfects, while they also appear in one class of Sindarin past tenses (akâra yielding Sindarin agor).
              In Quenya, past tenses are often formed with the terminaison -në (ex. orta- "raise" > ortanë "raised") or by nasal infixion + final -ë (ex. tac- "fasten", pa.t. tancë). Nasal infixion is also found in Sindarin past tenses (ex. sogo "to drink" > pa.t. sunc). Since the past tenses involving nasals occur in both Quenya and Sindarin, they must go back to at least Common Eldarin. No primitive form of the Quenya past tense terminaison -në is mentioned by Tolkien in the published material; if it existed, it would probably have been *-nê. Some of the nasal-infixed past tenses may simply be due to such an terminaison being added directly to a stem, whereupon the n and the final consonant of the stem were transposed. For instance, Sindarin sunc "drank" (Quenya *suncë, not attested) could be derived from, say, CE *sunkê < QP *suknê, sc. the stem SUK "drink" with the past tense terminaison *-nê. But this is speculation and requires shifts like kn > nk, that do not regularly occur; it may be better to assume that the nasal-infixed form *sunkê is original.
              In Quenya, a form held to be the aorist is formed with the terminaison -ë, that changes to -i when any terminaison is added. In the primitive language, this must have been -i everywhere (since final short -i became -ë in Quenya, but remained unchanged when not final).
              One of our very few examples of a primitive present tense is uljâ *"pours", the source of Sindarin eil "it is raining" (see ULU). May this argue the existence of a primitive present-tense terminaison -â, the source of the Quenya terminaison -a? In WJ:372, Tolkien refers to the "the [present?] tense stems in -â". It would seem that the terminaison -â is "invisible" when added to a verb already terminaison in -â, for the verb uljâ certainly shows the frequent verbal terminaison -jâ. Note, however, the form mâtâ "is eating", stated (in VT39:7) to be the "continuous form" of a stem mata- or MAT "eat" (VT39:7, 11; LR:371). With this compare Letters:427, where Tolkien states that Quenya palantír comes from a primitive form palantîrâ (or palantîra), and that this word includes a "continuative stem of TIR watch, gaze at". Clearly the "continous form" mâtâ "is eating" relates to MAT just like this "continuative stem" tîrâ *"is watching" relates to TIR. It seems that from basic verbal stems (with no terminaison like -jâ or -tâ) a continuous form corresponding to English "is ...-ing" can be derived by lengthening the stem-voyelle and adding the terminaison -â. The Quenya descendants of these forms (not attested in this case: *máta, *tíra) are clearly what is often called the present tense. Since mâtâ is translated "is eating", it seems that the continuative stems could function as finite verbs already in the primitive language.
              We have one example of a primitive perfect, nomly the form awâwiiê given in WJ:366. It would seem to be formed by lengthening and prefixing the stem-voyelle and adding the suffix -iiê. In Quenya, the terminaison has become -ië, but otherwise perfects are still formed in the same way.
              How other forms of the verb were constructed in primitive Elvish, we don't know. The infinitive terminaison -ië is found both in Quenya and Old Sindarin ("Old Noldorin"), so it must go back to at least Common Eldarin. Its primitive form may have been -iê (perhaps attested buried in the word luktiênê, see below). The Quenya and Sindarin future tense terminaisons, -uva and -tha, are evidently not cognate - perhaps suggesting that one or both are innovations with no counterparts in the primitive language.
              It may be noted that the primitive language had no inflectional imperative; instead the independent imperative particle â, variable in place, was used in conjunction with a verbal stem (WJ:365).
              Pronouns: Our knowledge of the primitive pronominal system is far from complete. A first person stem NI "I" is given in the Etymologies (LR:378 s.v. NI2), and ni is still found in Quenya (while the origin of the Sindarin word for "I", im, is obscure). The Quenya terminaison -mmë for (exclusive) "we" and the corresponding Sindarin terminaison -m argue the existence of a primitive 1st person plural pronoun including the sound m in Common Eldarin at the latest. Tolkien speaks of de and its variant le as "pronominal éléments in the 2nd person" (WJ:363). Quenya tye "you" (as object, "thee") and the Sindarin terminaison -ch *"you" seem to indicate that there was also a 2nd person élément including the sound k (since Quenya tye in light of the Sindarin terminaison must be assumed to descend from *kje, while Sindarin -ch represents older *-kk-). Concerning the 3rd person, the demonstrative stem TA "that" is relevant (it yields Quenya ta "that, it"). Quenya te "them" (and "they"?) may descend from unstressed *tai, sc. ta "that [one]" with a plural terminaison: *"those". The 3rd person was apparently primarily associated with another demonstrative stem, S-. Under this stem, the Etymologies lists sû or su (or sô/so) as an evidently primitive pronoun "he", while "she" is sî or si (or sê, se). Here, reference is also made to "-so inflexion of verbs" and the corresponding feminine "-se inflexion", evidently meaning that these pronouns were attached to verbs to express that "he" or "she" was the subject of the verb. Whether these inflections occurred already in the most primitive language is not clear.
              Other parts of speech: An example of an adverb is provided by the word akwâ, selon to WJ:415 "an extension or intensification of *kwâ, used adverbially" (Quenya aqua "fully, completely, altogether, wholly"). Another example is hekwâ "leaving aside, not counting, excluding, except", stated to be both adverb and preposition (WJ:364-5). This is formed from the "adverbial élément" HEKE, HEK, meaning "aside, apart, separate" (WJ:361, 364). No special adverbial terminaison, like English -ly, is known; the terminaison -wâ seen in the word hekwâ is also an adjectival terminaison (see "Derivation" below). - One "primitive negative élément" is mentioned in WJ:370: bâ "no!" (also abâ, aba), expressing refusal, not denial of facts. Otherwise, words based on the stem LA "no, not" or the negative stems GÛ, MÛ and their prefixed variants UGU, UMU were evidently used to form negations. - It is not known whether there were articles in the primitive language; this may be doubtful. The source of the article i "the" in Quenya and Sindarin, nomly the stem I, is stated to be a "deitic particle" (LR:361). So while Quenya i alda and Sindarin i 'aladh mean "the tree", primitive *i galadâ evidently meant "that tree" instead. Later, the meaning of i was weakened from "that" to "the" (perhaps already in Common Eldarin, since Quenya and Sindarin share this article). The Romance languages got their definite articles just like this: Their ancestor Latin had no word for "the", but the meaning of Latin demonstratives (typically ille, illa) was weakened to produce articles like la or el.
             

    COMMON ELDARIN

    As noted above, Common Eldarin (CE) is the next stage of Primitive Elvish. This is the language of the original Eldar as distinct from the Avari, the tongue developed from Primitive Quendian during the two and a half centuries the March from Cuiviénen to the sea lasted, and hence the last common ancestor of Quenya and Sindarin.
              In PM:342, Tolkien actually states, "When the Eldar arrived in Aman and settled there they had already a long history behind them...also their languages had been elaborated and changed and were very different from their primitive speech as it was before the coming of Oromë." No drastic changes are reflected in the material that has been published to far, however. In many cases, QP words would be unchanged in CE; note that ñgolodô (Noldo) is said to be both CE (WJ:379) and QP (WJ:381). The plural terminaison was still -î, as in elenî "stars" (WJ:360).
              As mentioned in the introduction, there are only three forms in the Etymologies that are explicitly identified as "Eld" = Eldarin, evidently meaning Common Eldarin: mahtâ- "to handle", ndæ^r "bridegroom", wa "together" (see MA3, NDER, WÔ). These three are derived from earlier ma3tâ-, ndêro and wo, forms that must necessarily be QP. A number of other Common Eldarin forms are found in WJ and PM, as well as in Vinyar Tengwar #42. Some phonological developments may be observed. The change of stressed wo to wa is explicitly mentioned in Etym (under WÔ). In QP ma3tâ > CE mahtâ we are evidently to understand that the sound 3 (sc. spirant g, selon to Christopher Tolkien) became unvoiced by assimilation to the following t, if the spelling "ht" in mahtâ represents German ach-Laut + t, as it does in the Quenya form mahta-. Forms like the verb wahtâ- "to soil, stain" and the noun wahsê "stain" from the stem WA3 must therefore be taken to be Common Eldarin for Primitive Quendian *wa3tâ-, *wa3sê. (Note that in wa3râ "soiled, dirty", 3 is unchanged, because there is no unvoiced consonant following.)
              The main change seems to have affected the short final voyelles. Original -a, -e and -o disappeared; for instance, QP abaro "recussant" yielded CE abar (WJ:371), while QP kwene "person" became CE kwên (WJ:360 - the QP word kwende seems to be unchanged in CE, though). selon to most sources, long -â, -ê, -ô were unchanged, as were -î and -û - though VT42 cites some CE forms in which the long final voyelles may seem to have become short already, ex. daira "large, great" or netere "nine". Whatever the case may be, it may have been at the CE stage original short -i and -u turned into -e and o, a change reflected in Quenya. The change of short final -i to -e is also reflected in Old Sindarin, so it would seem that this change occurred in Common Eldarin, the ancestor of both languages. As the CE word kwên as compared to QP kwene demonstrates, the voyelle of the new monosyllabic words could be lengthened (but not in the plural form kwenî, where the word was not monosyllabic - this is still reflected in Quenya quén, pl. queni instead of **quéni).
              Another change was that "medial h was very early lost without trace in CE", the noun enclitic -hô "from" becoming -ô, the origin of the Quenya genitive terminaison -o (WJ:368). This would seem to support what we argued above: that in mahtâ-, the letter H actually stands for ach-Laut. This stronger "H" wasn't lost (still present in Quenya mahta-).
              Some difficult consonant clusters changed into more pronouncable combinations in Common Eldarin "and possibly earlier", sc. already in QP (WJ:416). In WJ:416, the shift bm > mb is mentioned, QP labmê "tongue" (language) becoming lambê in CE at the latest. In the Etymologies, we find double forms like stabnê, stambê "room" (STAB); may this suggest a similar change bn > mb, perhaps at the CE stage? (But in this case, the bn form must have survived alongside the new mb form, since "Old Noldorin"/Old Sindarin still has stabne.) We know from WJ:403 that the combination sd was assimilated to zd in Common Eldarin, esdê "repose" becoming ezdê. (The stem EZDÊ in the Etymologies must therefore be understood as an Common Eldarin form; not all the heads of the entries in Etym represent primitive roots. EZDÊ < esdê is itself a rearranged form of the basic stem SED "rest".) While s became voiced to z before d, it seems that d was devoiced to t before s, primitive sjadsê "cleft, gash" becoming sjatsê (SYAD). Perhaps the change ds > ts also occurred at the Common Eldarin stage.
              Common Eldarin may have introduced some new diphthongs derived from e, o. Later Loremasters were "inclined to the opinion that...ae, ao [produced by A-infixion] were not primitive developments, but comparatively late and due to the analogy of ai : i, and au : u" (see VT39:9-10). "Comparatively late" may mean at the Common Eldarin stage rather than in Primitive Quendian. The fact that words that originally must have contained the diphthongs ae, ao are found both in Quenya and in Telerin strongly supports this conclusion. In Quenya, these diphthongs became é and ó, respectively; in Telerin they both became á. As an example, Tolkien mentions Quenya méla "loving, affectionate", Telerin mála. Both of these are derived from an A-infixed form of the stem MEL "love", implied to be *maelâ (not explicitly given). See VT39:10.
              Common Eldarin was not an entirely uniform structure; already on the March, there were dialects. At a very early stage, perhaps even before the Separation, the Teleri shifted original kw (> Quenya qu) to p, a change that is still reflected in Sindarin and the Telerin of Aman (like Quenya quár "fist" corresponding to Sindarin paur, Telerin pár; all of these descend from primitive kwâra, PM:318). In PM:401, Pengolodh points out that "the Quendi were sundered also in speech: the Avari from the Eldar; and the Teleri from the other Eldar".

    THE STEM AND ITS MODIFICATIONS

    When we are dealing with primitive Elvish, the concept of the stem, root or base must be clearly understood. Already in his very early "Qenya Lexicon" of 1915, Tolkien stated that "roots...are not words in use at all, but serve as an elucidation of the words grouped together and a connection between them" (LT1:246). The root or stem is a somewhat abstract skeleton containing a basic meaning, and in the process of derivation this skeleton is fleshed out to produce actual words developing its meaning. For instance, the general idea of youth is apparently contained in the stem NETH - Tolkien simply wrote "young" to suggest its meaning - but this is not to say that the primitive Elvish word for "young" was neth. NETH is just the basis of actual words like nêthê "youth" or nethrâ "young" (whence Quenya nésë, nessa). The Etymologies is for the most part a list of such stems followed by some of the actual words that they yielded in various languages. (However, some of the entry-heads in Etym seem to be complete words in themselves, such as RAMBÂ "wall" or TINKÔ "metal".) The vast majority of primitive Elvish words consist of a stem combined with an terminaison; these terminaisons are explored in detail below.
              "My father wrote a good deal on the theory of sundokarme or 'base-structure'," Christopher Tolkien informs us (LR:343). However, little more than his own summary of his father's ideas is available to us: "Very briefly indeed, the Quendian consonantal 'base' or sundo was characterised by a 'determinant voyelle' or sundóma: thus the sundo KAT has a medial sundóma 'A', and TALAT has the sundóma repeated. In derivative forms the sundóma might be placed before the first consonant, ex. ATALAT" (WJ:319). It would seem, then, that the "base" consists of consonants (like K-T in KAT) plus a "determinant voyelle" (in this case A) that can move around and be reduplicated - but since it has to be present somewhere, the Quendian base structure is not a "Semitic" system with purely consonantal roots, as in Khuzdul. This is more like the Adûnaic system: consonantal roots that are associated with a certain "Characteristic voyelle" that can be inserted in various places, but has to be present in all derived words - or stems with the same consonants would become impossible to tell apart.
              The system that eventually crystalized at Cuiviénen featured a "basic structure" with a preference of stems of "the pattern X-X(-), with a fixed medial consonant...such as *Dele, *Heke, *Tele, *Kala, *Kiri, *Nuku, *Ruku, etc. A large number of monosyllabic stems (with only an initial consonant or consonant group) still appear in the Eldarin tongues; and many of the dissyllabic stems must have been made by elaboration of these" (WJ:392). When Tolkien speaks of "the pattern X-X(-)", he evidently means "consonant-voyelle-consonant(-voyelle)". Usually, the first and the second voyelle are identical; indeed it doesn't seem to matter whether the stems for, say, "follow" and "lick" are given as KHILI and LABA (WJ:387, 416) or simply as KHIL and LAB, as in the Etymologies (LR:364, 367). In a similar manner, the stem for "pour" is given both as ULU (LR:396) and as UL (WJ:400). Likewise, stems like the ones Tolkien lists as examples - *Dele, *Heke etc. - could probably just as well be given as DEL-, *HEK- etc. (DEL is actually found in WJ:363). The latter system seems to be employed in the Etymologies (note that Etym has KAL where WJ:392 gives *Kala-); the stem ULU instead of UL is one of the few exceptions. In Etym, the suffixed stem-voyelles may have been dropped simply to save space. But when Tolkien in WJ:392 mentions "monosyllabic" stems, he seems to be referring to stems with no consonant following the first voyelle (like KWE, NA), so that the voyelle cannot be suffixed.
              While Tolkien in WJ:392 speaks of a "large number" of such monosyllabic stems, they are relatively rare in our corpus. On the same page in WJ, Tolkien speculates that a stem KWE referring to vocal speech may have existed at the most primitive stage, but was later expanded to KWENE and KWETE, thus being adapted to the system that had evolved in the meantime. In the Etymologies, most of the stems listed consist of three éléments: an initial consonant or a consonant group, a voyelle, and one consonant following the voyelle (ex. BAL, SPAN). In some cases, there is no initial consonant (ex. EL), but there are very few stems of one syllable that lack the final consonant, such as NÂ "to be" (LR:374 s.v. NÂ2). As noted above, the latter seems to be the kind of stems Tolkien calls "monosyllabic" in WJ:392 (and not stems like KWEN, EL, DEL [WJ:361-363], that can readily be turned into polysyllabic stems by suffixing the stem voyelle: KWENE, ELE, DELE [WJ:360]). Of the more than six hundred stems listed in the Etymologies, less than thirty have this "monosyllabic" structure, and several of them are not the stems of verbs, nouns or adjectives, but prepositions, particles, prefixes and the like. (Some stems originally ended in a guttural consonant written as 3, but lost it and had the voyelle lenghtened in compensation: DO3 > DÔ and evidently TA3 > TÂ. Perhaps stems like THÊ, THÛ, YÔ are to be understood as later forms of *THE3, *THU3, *YO3, not given.) In accordance with this, Tolkien stated that "the later view [of the Loremasters in Middle-earth] was that in fact 'full stems' (meaning noun-adjective or verb stems) were actually by the end of the common development of primitive Quendian seldom if ever monoconsonantal" (VT39:11). "Monoconsonantal" is a better term for this kind of stems than "monosyllabic".
              The stem for "bite" is a good example of how a stem can be modified to produce the basis for new words. No less than four varieties of it are found in the Etymologies. First there is NAK, apparently the most basic form, with the simple meaning "bite". The stem NDAK "slay" is evidently to be understood as a strengthened form of NAK, the strengthening of the initial consonant symbolizing the intensified meaning. Another variant of NAK prefixes the stem-voyelle to produce ÁNAK, a stem yielding words for "jaw", the body-part used for biting (Quenya anca, Sindarin anc, both from primitive ankâ, in turn derived from ÁNAK; see NAK). A fourth possible variant is NAYKA with an infixed Y (and the stem-voyelle suffixed); this is called an "elaboration" of NAK. This "elaborated" stem seems to mean basically *"biting" and hence *"painful"; it yields words like Quenya naicë, Sindarin naeg "(sharp) pain". We will now have a closer look at the various ways of manipulating a stem.

    Stem voyelle prefixed: In the entry I in the Etymologies, Tolkien explains that i is an "intensive prefix where i is base voyelle". He mentions ITHIL "Moon" as an example; this is derived from a stem (or "base") THIL "shine silver" (see SIL). INDIS "bride" as a nom of the goddess Nessa comes from NDIS "woman"; the voyelle-prefixed variant i-ndise is called an "intensive form". Cf. also WJ:318, where Quenya and Sindarin estel "hope" is said to be a stemvoyelle-prefixed derivative of a stem STEL "remain firm".
              In a number of cases, voyelle-prefixed versions of a stem are given as separate entries in the Etymologies. Sometimes, the stress moves to the new first syllable; sometimes the original stem-voyelle retains the accent. ÁLAK "rushing" is derived from LAK2 "swift". ÁNAK "jaw" from NAK "bite" has already been mentioned. ANÁR "sun" is stated to be a derivative of NAR1 "flame, fire". (In the Silmarillion appendix, entry nár, Christopher Tolkien mentions (a)nar as "the same ancient root" that yielded words for both fire and sun.) AYAN "holy" is derived from YAN of similar meaning. ELED "go, depart, leave" connects with LED "go, fare, travel". ÉNED "centre" comes from NED of similar meaning. ERÉD, yielding words for "seed", is derived from RED "scatter, sow". ÓLOS "dream" is connected to LOS "sleep". ÓROM, the stem that selon to the Etymologies is the source of the nom of the Vala Oromë, comes from ROM "loud noise, horn-blast" (but Tolkien later rejected this as an Elvish folk etymology). It has been suggested that ÓROK, the stem that the Elvish words for Orc are traced to in the Etymologies, is connected to ROK-, the stem for "horse". While this may seem semantically strained, ROK- may originally have referred to the steed of "the dark Rider upon his wild horse" that afflicted the Elves at Cuiviénen, evidently some servant of Morgoth (Silmarillion ch. 3). Hence the strengthened stem ÓROK could be used of other evil creatures. (However, Tolkien seems to have dropped this idea and decided to derive the Elvish words for "Orc" from a stem RUKU instead; see WJ:389.) The negative stems GÛ, MÛ have prefixed variants UGU, UMU. Slightly more complex is the derivation of AKLA-R *"brillance" from KAL "shine" and OKTÂ "war" from KOT "strive, quarrel"; here the stem-voyelle is prefixed as usual, but also lost in its normal position, and other terminaisons are introduced. Other examples of words where the stem-voyelle is removed from its normal place between the first and second consonant of the stem to be prefixed instead include esdê "repose" from SED "rest" (see WJ:403), the above-mentioned ankâ "jaw" from NAK "bite" and ostô "fortress" from the stem SOTO- "shelter, defend" (see WJ:414 for the latter). Cf. also the agental formation edlô from DEL, DELE "walk, go, proceed travel" - but also edelô with the stem-voyelle of DEL intact. In WJ:363, Tolkien says that the word edlô displays "loss of sundóma" (stemvoyelle), and so, obviously, do words like esdê, ostô, ankâ. The stem RUKU is said to have variant forms uruk- and urk(u). It is perhaps impossible for monosyllabic stems like KWA (having to do with completion) to appear without their stem-voyelle in its normal place, but it may still be prefixed, as in the derivative akwâ (selon to WJ:415 "an extension or intensification of *kwâ, used adverbially" - Quenya aqua "fully, completely, altogether, wholly").

    A-infixion: In some cases, a new voyelle A is inserted into a stem, turning the stem-voyelles i, u into diphthongs ai, au. The stem SLIW "sickly" yields the adjective slaiwâ "sickly, sick, ill" (contrast another derivative, slîwê "sickness", that does not show infixion). A-infixion is also seen in the word taun ?"hill" from TUN (see MINI). From the stem MIL-IK *"greed" is derived Mailikô, a nom of Melkor. Yet other examples from the Etymologies are thausâ "foul" from THUS and taurâ "mighty" from TUR. In WJ:337, Tolkien derives maikâ "sharp, penetrating, going deep in" from a stem mik "pierce". Besides, the Quenya word nauta "bound" derived from NUT points to a primitive form *nautâ (not given); likewise, Sindarin glaer (glær) "long lay" from GLIR must descend from *glairê (cf. Quenya lairë). In the essay Quendi and Eldar, naukâ "ill-shapen, *short" is derived from a stem NUKU "stunted". This is called an "adjectival formation" (WJ:413); note that maikâ, naukâ, slaiwâ, taurâ, thausâ are also adjectives. A-infixion is also found in the abstract khaimê "habit" form KHIM "adhere" (that also yields the adjective khîmâ "sticky" without A-infixion - as if to keep things from getting too predictive!) Furthermore, one of the "ancient forms" of the stem RUKU (having to do with "fear", the source of the Elvish words for Orc) is given as rauk- (WJ:415). Using examples from Quenya, Tolkien explained that words formed by A-infixion "were mostly 'intensive', as in rauko 'very terrible creature' (*RUK); taura 'very mighty, vast, of unmeasured might or size' (*TUR). Some were 'continuative', as in Vaire 'Ever-weaving' (*WIR)" (VT39:10). - Whether A-infixion ever produced diphthongs ae, ao from simple e, o, just like this process produced ai, au from i, u, was a matter of debate. Fëanor held that such forms had indeed occurred in the Primitive Quendian, but as mentioned above, later Loremasters were "inclined to the opinion that this ae, ao were not primitive developments, but comparatively late and due to the analogy of ai : i, and au : u" (see VT39:9-10).

    I/Y-infixion: This seems to be rarer than A-infixion. It is stated that the stem NAYKA *"painful" may be an "elaboration" of NAK "bite"; NAYKA yields Quenya words in naic-. The stem WAIWA "blow" is apparently an I-infixed variant of WAWA, that in turn seems to be a reduplicated form of WÂ. In VT39:11, Tolkien indicates that "desiderative" formations often show i-infixion; see below.

    Nasal infixion: Stems could be modified by infixing a nasal before the second consonant of the stem, m before b and p, and n otherwise (except possibly ñ before w, see below). Thus, the stem DAT "fall down" has a nasal-infixed variant DANT. LAK1 "swallow" becomes LANK-, yielding words for "throat". One of the "ancient forms" of the them RUKU is nasal-infixed: runk- (WJ:415).
              Nasal-infixion is not uncommon in the derived words. For instance, TUG yields tungâ "taut, tight", and ronyô "chaser, hound of chase" comes from a stem ROY "chase" (LR:384 s.v. ROY1). In some cases, it is hard to tell whether seemingly nasal-infixed forms are actually due to later metathesis. Quenya sambë "room" is said to descend from primitive stabnê, stambê. The latter would seem to reflect a nasal-infixed form of the stem STAB, but Tolkien's wording can also be interpreted to mean that the oldest form was stabnê derived from STAB simply by adding an terminaison, and that the cluster bn later underwent metathesis to become *nb > mb. Alternatively, Tolkien may have meant to say that it was impossible to tell whether the ancestral form of Quenya sambë was stabnê or stambê. Another such double form is found under SYAD: sjadnô, sjandô "cleaver" = sword. Whatever the case, the stem PAT yields both patnâ "wide" and the nasal-infixed form pantâ "open", words that were seemingly distinct also originally, so it would seem that nasal-infixion did occur also in the primitive language.
              There is one example of ñ-infixion before w: liñwi "fish" from the stem LIW.

    Strengthening, fortification, reinforcement, enrichment: These are Tolkien's terms for certain changes that stems sometimes undergo. For instance, RUKU also appear as a "strengthened stem" gruk- (WJ:415); in this case the "strengthening" consists of a g-prefix. A prefix s is seen in s-rot- "delve underground, excavate, tunnel" as compared to the simpler stem rot (PM:365; groto in WJ:414 is apparently a g-prefixed variant). (selon to VT39:11, later loremasters held that the original strengthening of r initially was dr rather than gr; the latter was modelled on the frequent variation l / gl.) Fëanor is said (in VT39:9) to have cited examples of initial strengthening involving "the relations between initial st- and s-, or t-; gl- and l-; ky- and kw- and k-". Another "frequent initial enrichment" (WJ:413), that particularly impressed Fëanor, is turning b, d, g into nasalized plosives mb, nd, ñg. This could be called nasal prefixion, the initial version of the nasal infixion discussed above. However, initial n, like d, may be strengthened to nd, and m can similarly become mb (changes that can also occur in the middle of words, see below). Perhaps initial ñ could be strengthened to ñg (no examples). In LR:377, the stem ÑGYÔ, ÑGYON "grandchild, descendant" is suggested to be related to YÔ, YON "son", suggesting that Y- can be strengthened to ÑGY-.
              The stem DORO "dried up, hard, unyielding" yields QP ndorê "dry land" by initial enrichment d > nd (WJ:413). The stem NDER "bridegroom" is said to be a "strengthened form of der" (LR:375), sc. the stem DER "man". NDUL, yielding words meaning "dark, dusky, obscure", comes from DUL "hide, conceal". MBAD "duress, prison, doom, hell" is a strengthening of BAD "judge". MBUD, the stem that yields words for "nose", comes from BUD "jut out". MBAR "dwell, inhabit" is said to be related to BAR, though it is not clear how they connect semantically (the probable original meaning of BAR is given as "raise"). Concerning the strengthening N > ND and M > MB, there is the stem NDIS ?"bride", said to be a "strengthening" of NIS "woman" (LR:375). The stem NDÛ "go down, sink" comes from NÛ, an apparently prepositional stem yielding such words as "down" and "under". We have already mentioned NAK "bite" > NDAK "slay". The stem MASAG "knead" connects with MBAS of similar sense; presumably they are both elaborations of a simpler root *MAS. (Note, however, that there are many stems with initial MB, ND that cannot be matched with any corresponding stem in B-/M- or D-/N-. In such cases, we must assume that the nasalized stop is "original".)
              Similar changes can also occur in the middle of words. Kwende "elf" is derived from a stem KWENE by "primitive fortification of the median n > nd" (WJ:360). Cf. also some words in the Etymologies, like tundu "hill, mound" from TUN. The Quenya verb tamba- "knock, keep on knocking" vs. the simpler verb tam- "tap" indicates that a fortification m > mb has taken place (stem TAM). Tolkien explains that Lindâ "Linda, Teler-elf" is derived from the primitive stem LIN by "reinforcement of the medial N and adjectival -â" (WJ:382). Common Eldarin eldâ, "an adjectival formation 'connected or concerned with the stars' ", would seem to be derived after the same pattern and includes a medial fortification l > ld (stem EL, ELE); this is not found initially.
              In the middle of words, the "median" could also be doubled: Grottâ "a large excavation" is an "intensified" form (WJ:415) of grotâ "excavation" (WJ:414). Concerning the stem for "horse", ROKO, it is said that this is actually an "older simpler form of the stem, found in some compounds and compound noms, though the normal form of the independent word 'horse' had the fortified form rokko" (WJ:407). As we see, rokko is "fortified" by doubling the middle consonant of ROKO. The word battâ "trample", with "medial consonant lengthened in frequentative formation" (LR:351), provides us with an example of a "fortified" verbal stem: The basic stem BAT means "tread", and the fortified stem symbolizes the repetition of the action by lengthening the middle consonant. For the semantic change, compare Quenya tam- "tap" vs. tamba- "knock, keep on knocking" mentioned above.

    Extension: Some stems have special "extended" forms made by suffixing the stem-voyelle (as in DELE compared to DEL - in Quenya, this is called ómataina or "vocalic extension") and adding a final consonant, usually n, k, t, or s. In the Etymologies, the stem BORÓN is said to be an extension of BOR "endure" (when accented on the second syllable it is a verbal form of the noun-stem bóron-). A similar extension involving a final n is provided by the stem EL, ELE yielding Common Eldarin elen "star" (said to represent an "extended base", WJ:360; compare Old Sindarin toron "brother" from TOR; cf. also the pairs PHER/PHÉREN "beech" and THOR/THORON "eagle").
              Among the "ancient forms" of the stem RUKU (having to do with fear) are rukus and rukut (WJ:415). Could the extended stems with ómataina followed by t be what Tolkien refers to as "the so-called kalat-stems" in WJ:392? Kalat looks like an extended form of KAL, the stem having to do with "light". If so, yet another example may be the stem ÓROT "height, mountain", that is apparently extended from the more basic stem ORO "up; rise; high". Here we see how the extended form develops the meaning of the more basic stem (the other examples of extended stems are not separately glossed). Double stem-forms in the Etymologies, like LEP/LEPET "finger" or ESE/ESET "nom" seem to exemplify the same phenomenon. A certain example is arat-, that in PM:363 is said to be "an extended form of the stem ara- 'noble' ". When the stem NA "to be" yields Quenya nat "thing", this may reflect a similar t-extension.
              There are some possible extensions with final -k, like OTOK "seven" from OT. Perhaps NÁYAK "pain" is connected with NAY "lament", while KIRIK (whence Quenya circa "sickle") is definitely extended from KIR- "cut, cleave" (not defined in the Etymologies, but see kir- in the Silmarillion Appendix; cf. also KIRIS "cut" as noun - another expanded form). LEPEK is given as an extension of LEP "five" (also LEPEN). Cf. also MIL-IK *"greed", evidently an extension of a simpler stem *MIL (whence Quenya mailë by A-infixion).
              Extensions involving final -s (cf. rukus and KIRIS above) include OT/OTOS "seven" (also OTOK already mentioned), THEL/THELES "sister", TER/TERES "pierce", PHAL/PHÁLAS "foam" (plus the variant SPAL/SPALAS); cf. also KYEL(ES) "glass". The stem NIS "woman" is said to be "elaborated from INI" (see NDIS); perhaps NIS should rather be derived from the simple stem NÎ "woman", of which INI must be a voyelle-prefixed version. (For the shortening of the long stem-voyelle in the voyelle-prefixed variant, compare the negative stems GÛ vs. UGU and MÛ vs. UMU.) Tolkien speculates that THUS ?"evil-smelling" is related to (extended from?) THÛ "puff, blow". The latter examples indicate that "monosyllabic" stems (stems with no final or "medial" consonant) can be expanded by adding the final consonant -n, -t, -s directly to the original stem-voyelle; the voyelle cannot be reduplicated finally because there is no consonant to which it can be suffixed. (But apparently the stem-voyelle can be reduplicated following the new consonant after the consonant has been added; cf. Tolkien's reference in WJ:392 to the stem "*KWE, of which *KWENE and *KWETE were elaborations".)
              Note that there are some stems that seem to be polysyllabic right from the start. For instance, KYELEK "swift, agile" can for semantic reasons hardly be an expanded form of KYEL "come to an end".
              It should be noted that Tolkien sometimes uses the term "extended stem" also with reference to stems with a prefixed stem-voyelle (see above), when the voyelle is still present in its normal place.
              Regarding the use of the sundóma (stem-voyelle) in such triconsonantal stems, Tolkien noted that "more than two insertions of the sundóma were not necessary in the Common Eldarin system" (hence we see two A's in palat-, palan- as extensions of PAL, VT47:8). However, the first sundóma following the initial consonant could be "omitted and replaced by an extruded sundóma initially, in which case the vocalizing was ap'lata" (VT47:13). Here the ' indicates the place of the "omitted" stem-voyelle; this is not phonetic loss. In VT47:13, Tolkien also cited the example aklara (the primitive form underlying Quenya alcar, Sindarin aglar "glory, splendour"); this is thus ak'lara, a reworked form of triconsonantal KALAR as an extended form of the simple root KAL having to do with light.

    Differentiation: As noted above, the long forms of stems with a final voyelle usually involve simple repetition of the stem-voyelle: DEL > DELE, KAL > KALA etc. But there are some rare cases where another final voyelle, -U, turns up. In WJ:411, Tolkien mentions a stem TELE "close, end, come at the end" and adds that "this was possibly distinct from *tel-u 'roof in, put the crown on a building'... But *telu may be simply a differentiated form of *TELE, since the roof was the final work of a building." It would seem that variant or "differentiated" stems could be made by modifying the final voyelle.
              Except for TELU, the evidence for such stems is usually indirect. The stem KEL "go, run (especially of water)" clearly has a longer form KELU. (The Index to Unfinished Tales, entry Celos, actually mentions a root kelu- "flow out swiftly".) The longer form turns up in Quenya celumë "stream, flow" (but not in celma "channel"). The Ilkorin word for "river", celon, is derived from what seems to be an expanded form in -n: "kelu + n", hence *kelun (LR:363). A similar case seems to be Quenya cotumo "enemy" from KOT, KOTH: the middle u has to come from somewhere. There are also some Quenya stems in -u, such as nicu- "be chill, cold (of weather)" (WJ:417) or hlapu- "fly or stream in the wind" (MC:223). But how they relate to "differentiated" stems like TELU, if they do at all, is far from clear. A new clue (of sorts) appeared in VT46:8: "On -u- suffix frequent in Q[uenya] after el, al, see Q structure: cf. kelu, telu, smalu, etc." If "Q structure" refers to any specific written essay, it has never been published. Smalu appears in the Etymologies (entry SMAL) as a word for "pollen, yellow powder". Maybe we must simply await the publication of whatever essays Tolkien wrote about "Q structure"...

    Variation: There seems to be some variation between certain similar consonants, such as T/TH/D, and also between TH and S. In the Etymologies, there is obviously a connection (confirmed by Tolkien's own cross-reference) between the stems PAT, having to do with openness, and PATH, yielding words like "Noldorin" pathw "level space" (Classical Sindarin *pathu). It is also suggested (in LR:393) that THIN, yielding words for "grey", may be a variant of TIN "emit slender (silver, pale) beams". Similarly, there is obviously a connection between the stems DAL "flat", LAD *"wide" and LAT "lie open". The stems SIL, THIL "shine" are said to be variants, and a similar variation S/TH is seen in the pairs GOS/GOTH "dread" and KHIS/KHITH "mist, fog". Variation SP/PH is seen in SPAL/SPALAS, variant of PHAL/PHÁLAS "foam". KAR "make, do" seems to have a variant KYAR "cause", and under the stem KEL "go, run" we find references to KYEL "come to an end, run out" and KWEL "fade, fade away, wither". The variation between different semi-voyelles (Y/W) that is seen in the pair KYEL/KWEL is also found in KHAW as compared to KAY "lie down"; in the Etymologies, KHAW is equated with the latter stem. This also provides an example of variation K/KH; cf. also RIK(H) "jerk, sudden move". Under TAM "knock" there is a reference to NDAM "hammer, beat"; the latter seems to display both "initial enrichment" with a prefixed nasal and variation T/D. Primitive Elvish evidently did not allow *NT as an initial combination, so it had to become ND instead. This principle may also explain the second élément of the nom Moringotto (MR:194), said to be the Quenya form of the nom Morgoth, the Black Enemy: If the -ngotto part suggests a primitive form *ñgottô "enemy", this word could be referred to the stem KOT, KOTH "strive, quarrel" by assuming that the "initial enrichment" of K- is *ÑG- rather than ?ÑK-. The "initial enrichment" of P- would likewise be *MB- rather than ?MP-, though we have no examples.
              Variation P/T is found in the stems PIK and TIK; both of these evidently have to do with smallness. Under TIK, Tolkien made a cross-reference to PIK. Variation between T and D is seen in the pair TING, DING, but since these words are simply onomatopoeic, such variation is to be expected.
              selon to WJ:363, there was "some evidence" that variation between D and L occurred in Primitive Quendian, "a notable example being de/le as pronominal éléments in the 2nd person". In late QP, the combination GL appeared as an initial variation of L (WJ:411, cf. VT39:11).
              Variation between different voyelles is much rarer, but BEL "strong" is tentatively compared by Tolkien to to the stem BAL (whence balâ "Power, god, Vala"), and under NAT "lace, weave, tie" Tolkien made a cross-reference to NUT "tie, bind".
             

    DERIVATION IN PRIMITIVE ELVISH

    In Primitive Elvish, nearly all words can be split into a stem followed by a derivational terminaison, and we will here attempt to list these suffixes. In the primitive language, the stem and the terminaison are usually easy to distinguish, while the border between them is often blurred by sound-changes in the later languages. For instance, primitive sukmâ "drinking-vessel" is easily split into a stem SUK "drink" with the terminaison -mâ denoting an implement - but in Quenya, that has shifted original km to ngw, the resulting word sungwa can no longer be analyzed as easily. (Despite examples like this, the original terminaisons are usually recognizable in Quenya, with shortening of the final voyelles: -mâ normally appears as -ma. Much of what is said below still holds true for the direct Quenya descendants of these suffixes, but in Sindarin, the original terminaisons are much worn down and sometimes even replaced with new terminaisons.)
              It should be noted that the second, reduplicated voyelle of the stem, the ómataina or "vocalic extension", is often not included when an terminaison is added to produce an actual word. There are definitely some words where the second voyelle persists, as when ULU "pour" yields ulumô *"pourer", but often it disappears. In WJ:416, a stem NUKU "stunted" is given, but in the derivative nuktâ- "stunt", the second U of NUKU is not included. On the other hand, the ómataina may sometimes turn up in the derivatives even when the stem is given in the shortest form, as when the noun tjulussê "poplar-tree" is derived from TYUL "stand up (straight)"; this noun is actually based on the ómataina-form *TYULU.
              The second voyelle of two-syllable stems like GÓLOB or STÁLAG may also be omitted in the actual words that are derived from them; these stems manifest as golb- and stalg- in the derivatives golbâ "branch" and stalgondô "hero, dauntless man". (See also below concerning laik-wâ from LÁYAK.)
              There are also words where the first voyelle drops out when it is unaccented: for instance, the stem BERÉK yields b'rektâ- "break out suddenly" and KARÁN yields k'rannâ "ruddy" (but from the same stems come berekâ "wild" and karani "red" with the first voyelle of the stem intact). This loss of unaccented stem-voyelles is most often seen in the original forms of Sindarin words and may be thought to be a phenomenon that occurred after the oldest stage, in Common Telerin, so that b'rektâ-, for instance, represents earlier *berektâ-. But in at least one case, a form where an unaccented voyelle has been omitted is seen to underlie a Quenya word: ráca "wolf" descterminaison from primitive d'râk, stem DARÁK. Primitive *darâk- with the first voyelle intact would have yielded Quenya **laraca instead. So in some cases at least, the unaccented voyelle must have disappeared in Common Eldarin at the latest.
              In the case of two-syllable stems with a final consonant, this consonant and the final voyelle may also change places when an terminaison is added: thus the stem ÚLUG manifests as ulgu- in the word ulgundô "monster".
              Note that in actual words, j as the final consonant of a stem invariably becomes i before a consonant, merging with the stem-voyelle to produce a diphthong in -i (as when the stem TUY - or TUJ - yields the word tuimâ "a sprout, bud" - for *tujmâ). Similarly, w becomes u before a consonant, as when TIW yields tiukâ "thick, fat" (for *tiwkâ). If a two-syllable stem loses its second voyelle in a derivative form, and the middle consonant is a semi-voyelle that is thus brought into direct contact with the final consonant, the semi-voyelle may again merge with the preceding voyelle to form a diphthong (as when the stem LÁYAK manifests as laik-, for *layk- or *lajk-, in the derivative laik-wâ "green"). Interestingly, VT45 reveals that in this and many other cases, Tolkien in his original manuscript marked voyelles arising from semi-voyelles with a special diacritic (a small bow under the letter). To some extent, such voyelles may still be regarded as consonants.
              Sometimes, but not always, j becomes i also before voyelles, as when DAY (DAJ) yields daiô "shade" - but contrast nâje "lament" from NAY (NAJ). In the former case, the i is marked with the diacritic discussed above (see VT45:8), evidently indicating that it may also be regarded as a semi-voyelle. Hence there may not be a significant "discrepancy" after all.

    Most primitive words ended in a voyelle, sometimes short but often long. The voyelle may be a complete terminaison in itself or part of a longer terminaison. No hard-and-fast rules can be formulated as to what the different final voyelles denote; at most there are certain tendencies. Very generally speaking, words with final A are often verbs or adjectives, and if they are nouns, they denote concrete things more often than substances or intangibles. Words in E are usually nouns and tend to denote abstracts or substances rather than simple, tangible objects. Words in I are often colour-adjectives; if they are nouns they usually denote female beings. Words in O are for the most part nouns and typically denote animate (male) beings; very often such words have an agental meaning. Words in U are relatively rare; they are nearly always nouns and typically denote either male beings or body parts.

    The terminaison -â (or -a) occurs on many types of words, but most prominent is the adjectival terminaison -â, mentioned by Tolkien in WJ:382. Adjectives may be derived by simple suffixation, like mizdâ "wet" from the stem MIZD or telesâ "rear" from TELES. However, the terminaison is often combined with certain manipulations of the stem:
              - Medial fortifications like M > MB, N > ND, L > LD, ex. rimbâ "frequent, numerous" from RIM, kandâ "bold" from KAN, kuldâ "golden-red" from KUL.
              - Nasal infixion, ex. tungâ "taut, tight" from TUG; cf. also WJ:375, where Tolkien derives pendâ "sloping" from a stem PED "slope, slant down".
              - A-infixion, ex. thausâ "foul, evil-smelling, putrid" from THUS, taurâ "masterful, mighty" from TUR (cf. also maikâ "sharp" from MIK, WJ:337, and naukâ *"stunted" from NUKU, WJ:413).
              - I-infixion; this occurs in a small group of desiderative formations. For instance, the adjective meinâ "eager to go, desiring to start" comes from a stem MEN "go" (VT39:11). (Apparently this word could also be used as a verb "desire to go in some direction, make for it, have some end in view"; this is at least true for its Quenya descendant mína-.) Other examples are found in Quenya: maita "hungry" from the stem MAT "eat", and soica "thirsty" from SOK "gulp, quaff, drink" (primitive *maitâ, *soikâ, my reconstructions). See VT39:11.
              - Lengthening of stem-voyelle, ex. khîmâ "sticky, viscous" from KHIM, râba "wild, untamed" from RAB, dâla "flat" from DAL.
              - Stem-voyelle prefixed: askarâ "tearing, hastening" from SKAR "tear, rend" (in effect, askarâ becomes a kind of participle).
              Nouns in -â display much the same variation; in most cases, such nouns denote inanimate things. Some are derived by simple suffixation, ex. wedâ "bond" (WED) or golbâ "branch" (GÓLOB). Some show nasal infixion: kwentâ "tale" (from KWET "speak"), randâ "cycle, age" (RAD), kwingâ "bow" (KWIG). We also note cases where the stem-voyelle is lengthened, such as râmâ "wing" from RAM or kânâ "outcry, clamour" from KAN (see PM:361-362 for the latter example). Doubling of the final consonant in the stem is also found: rattâ, ratta "course, river-bed" from RAT, gassâ "hole, gap" from GAS. The word ankâ "jaw, row of teeth" is based on a rearranged form of the stem NAK "bite"; Tolkien actually wrote "an-kâ" as if to emphasize that the middle voyelle was lost. Whether the final -â is an independent terminaison or just the stem-voyelle suffixed and lengthened is difficult to say. The similar formation OKTÂ "war" from KOT "strive, quarrel" clearly displays an independent terminaison -â, since the stem-voyelle is here O.
              As noted above, there are many verbs showing final A, but then as part of the longer terminaisons -tâ or -jâ. The simple terminaison -a, -â is very rare on verbs. We note olsa- "to dream" from the stem ÓLOS. Long -â combined with medial fortification M > MB occurs in tambâ "to knock" (TAM); the final -â is marked as accented. So is the final voyelle of battâ "trample", with the "medial consonant [of the stem BAT, *BATA] lengthened in frequentative formation".
              In some verbal stems, the final -a is quite clearly just the stem-voyelle repeated, for instance stama- "bar, exclude" (UT:282) or glada "laugh" (PM:359). They are therefore irrelevant here.

    The suffix -dô is a (usually agental) terminaison that is preferred in the case of stems terminaison in N: ñgandô *"harper" from ÑGAN/ÑGANAD and lindô "singer" from LIN2. (In the latter entry in the Etymologies, only the later Quenya form lindo is listed, but the archaic form lindô is given as part of the compound tuilelindô "swallow", etymologically "spring-singer", in the entry TUY. Ñgandô is likewise attested only as a part of the word tjalañgandô "harp-player"; see TYAL, ÑGAN/ÑGANAD.) There is also the word ndandô "Nando, Green-elf", interpreted "one who goes back on his word or decision" (the Nandor were so called because they left the march from Cuiviénen; the stem DAN-, NDAN- indicates "the reversal of an action, so as to undo or nullify its effect", WJ:412). In ñgolodô "Noldo" (WJ:364, 380), the terminaison -dô follows the reduplicated stem-voyelle (ómataina) of the stem ÑGOL. In this word, -dô apparently does not have any agental meaning; it is simply a personal (masculine) suffix, indicating one that has the property denoted by the stem ÑGOL (wise, wisdom).
              The Common Eldarin word rondô "vaulted roof" does not contain the terminaison -dô; this is the stem RONO (not in Etym) with medial fortification n > nd (VT39:9, cf. WJ:414). Indeed we cannot be sure that words like lindô are not derived from LIN by means of a similar fortification and the simpler terminaison -ô (see below). The question does not have much practical interest.
              The terminaison -dô also appear in a nasal-infixed form -ndo or -ndô. In the word ulgundô "monster, deformed and hideous creature" from ÚLUG it does not seem to be agental, but is simply used to form a noun. In the words kalrondô "hero" (from KAL "shine") and lansrondo, lasrondo "hearer, listener, eavesdropper" (from LAS2 "listen"), the terminaison -ndo, -ndô seems to be suffixed to another masculine terminaison, -rô/-ro (see below). Tolkien actually wrote "lansro-ndo, lasro-ndo" to make this clear. See also -ondô.
              As the feminine counterpart of -dô we would expect -dê, and this terminaison may be attested in asmalindê "yellow bird, 'yellow hammer' " (SMAL). The terminaison -(i)ndê that here occurs may be seen as a nasal-infixed form of *-dê, paralleling -ndô from -dô. (In Quenya, -ndë can apparently be used of an inanimate as well as a female agent: cf. ulundë "flood" from ULU "flow".)

    The terminaison -ê, -e has several meanings, or rather a few specialized meanings as well as some very general ones. A number of words in -ê, -e denote abstract or intangible things; in such cases the stem-voyelle is often lengthened: nêthê "youth" (NETH), ñgôlê "Science/Philosophy" (PM:360), ñôle "odour" (ÑOL), rênê "remembrance" (PM:372), slîwê "sickness" (SLIW), tûrê "mastery, victory" (TUR). The stem-voyelle remains short in we3ê "manhood, vigour" (WEG), et-kelê "spring, issue of water" (KEL) and naje "lament" (NAY), while khaimê "habit" shows A-infixion instead of lengthening (KHIM). In the word esdê > ezdê "repose", the origin of the Quenya nom of the Valië Estë, the stem SED occurs in an alternative form ESD- (WJ:403). For -ê as an abstract terminaison, compare also the longer terminaisons -mê, -rê, -wê, that are often used to derive abstract words.
              Another group of nouns in -ê denote substances: khjelesê "glass" (KHYEL(ES) ), kjelepê "silver" (KYELEP), laurê "golden light" (LÁWAR/GLÁWAR), mazgê "dough" (MASAG), rossê "dew, spray" (Letters:282), slingê "cobweb" (SLIG); srawê "flesh" (MR:350); we may even include mizdê "fine rain" (MIZD).
              A feminine terminaison -ê, -e is seen in the word tawarê, taware "dryad, spirit of woods" (evidently fem., contrast masc. tawarô, tawaro) (TÁWAR). Cf. also bessê "wife" (BES), though this may contain a longer terminaison -sê, and the final voyelle in the pronoun sê, se "she" (stem S; also sî, si).
              A few nouns in -ê denote localities: ndorê "land" (NDOR, WJ:413), taurê "great wood, forest" (TÁWAR); we may also add et-kelê "spring, issue of water" already mentioned (KEL).
              However, the terminaison -ê also occurs in many nouns that seem to have nothing in common semantically. The terminaison -ê may be used alone (as in spinê "larch" from SPIN, tatharê "willow-tree" from TATHAR), but more often it is combined with some other manipulation of the stem, such as nasal-infixion (londê "narrow path" from LOD), lengthening of the stem-voyelle (rîgê "crown" from RIG), A-infixion (laibê "ointment" from LIB2), medial fortifications like M > MB or N > ND (rimbê "crowd, host" from RIM, spindê "tress, braid of hair" from SPIN) or doubling of the final consonant of the stem (lassê "leaf" from LAS1, b'rittê "gravel" from BIRÍT). Nîbe "front, face" shows short -e, but the stem-voyelle of NIB is lengthened. In some nouns, the terminaison -ê, -e may be analyzed as being simply the stem-voyelle suffixed and sometimes lengthened, ex. in eredê "seed", kjelepê "silver", ndere "bridegroom" (ERÉD, KYELEP, DER/NÊR). Adjectives like dene "thin and strong, pliant, lithe" (WJ:412) or verbal stems like dele "walk, go, proceed, travel" (WJ:360) should probably be analyzed in the same way; no actual derivational terminaison is present. The same is the case with the noun kwende "Quendë, Elf"; it is derived from the stem KWENE by medial fortification N > ND, not by any distinct terminaison -e (WJ:360).

    The terminaison -i occurs in a number of adjectives, many of which are colour-words. In the case of monosyllabic stems terminaison in N, it is always combined with the fortification N > ND: slindi "fine, delicate" (SLIN), thindi "pallid, grey, wan, pale or silvery grey" (THIN, WJ:384), windi "blue-grey, pale blue or grey" (WIN/WIND; windi was struck out). Ninkwi "white" combines the terminaison -i with nasal-infixion of the stem NIK-W. On the other hand, karani "red" (KARÁN) shows no extra modifications, just the terminaison. Yet another colour-adjective, lugni "blue" (LUG2), seems to contain a longer terminaison -ni that is attested in this word only. In ringi "cold" the terminaison may be the stem-voyelle suffixed. Mori is stated to be both the adjective "dark" and the abstract "darkness" (Letters:382; in the Etymologies, stem MOR, the gloss is simply "black"). This brings us over to nouns in -i. Some are abstract, such as rinki "flourish, quick shake" (RIK(H), note nasal-infixion). The word etsiri "mouth of a river" is in origin plainly the abstract "outflow(ing)" (ET, compare SIR). A few nouns in -i refer to periods of time: ari "day" (AR1) and dômi- "twilight" (DOMO).
              Some few denote substances: g-lisi "honey" (LIS) and pori "flour, meal" (POR); khîthi "mist, fog" may also be seen as a substance (KHIS/KHITH). In light of this, may liñwi "fish" (LIW, note nasal-infixion) be "fish" as a substance, as food, rather than "fish" as an animal? Only one word in -i refers to a single, concrete, tangible object: phini "a single hair" (PM:362 - this word is stated to be Common Eldarin rather than Primitive Quendian). In several of the examples above, including phini, the "terminaison" may also be the stem-voyelle suffixed (but obviously not in ari, dômi-, pori).

    A feminine terminaison -î is seen in the two words Barathî (BARATH), an early nom of Varda, and in târî "queen" (wife of a târo, "king"). The word târî is probably formed after târo, since there is no R in the stem TA/TA3 and the feminine equivalent of the masculine terminaison -rô, -ro seems to be properly -rê (as in weirê "weaver", WEY), not *-rî. For -î as a feminine élément, cf. also the pronoun sî, si "she" (stem S; also sê, se). (Note, however, that Tolkien later implied another etymology for Quenya Vairë; see weirê in the wordlist below.)
              The -î of the word îdî "heart, desire, wish" seems to be unconnected (an abstract terminaison, or just the stem-voyelle suffixed, or even a misreading for *îdê as the Quenya form írë may suggest?) The stem ID is not defined.

    An abstract/infinitive terminaison -ie is found in Quenya and Old Sindarin, and we would expect it to correspond to something like -iê in the primitive language. This terminaison may be attested in the word luktiênê "enchantress" (LUK), if this is *luktiê "enchantment" + the feminine terminaison -nê, hence *"enchantment-female". *Luktiê would be an abstract or verbal noun formed from *luktâ- "enchant" (my reconstruction, cf. Quenya luhta-).
              In gwa-lassiê "collection of leaves, foliage" from lassê "leaf", the terminaison -iê + the prefix gwa- "together" is used to form a collective (Letters:282).

    An adjectival terminaison -imâ occurs in the word silimâ "shining white", "silver" (as adj.) (SIL). This would be the origin of the Quenya adjectival terminaison -ima (often meaning "-able", but sometimes used in a more general sense). Alternatively we would have to explain silimâ as including the ómataina-form of SIL, nomly *SILI, followed by the terminaison -mâ; see below. But this terminaison is typically used to derive words for implements and is found on no (other) adjective, so it is better to assume an terminaison -imâ.

    The feminine terminaison -ittâ is mentioned in PM:345; this is the origin of Sindarin -eth. See also -otta, -otto.

    The terminaison -jâ, -ja, -iâ, -ia has several meanings. It occurs on a number of adjectives: banjâ "beautiful" (BAN), erjâ "isolated, lonely" (VT42:4), kalarjâ "brilliant" (KAL), miniia "single, distinct, unique" (MINI), oijâ "everlasting" (OY), slinjâ "lean, thin, meagre" (SLIN), windiâ "pale blue" (WIN/WIND - it is uncertain whether Tolkien rejected the word windiâ or not). Wanjâ "fair, beautiful" is called an "adjectival derivative...from the stem WAN" in WJ:383, and Tolkien explicitly referred to -ja as a Common Eldarin adjectival élément (VT42:10). It also occurs in some ordinals that are said to be Common Eldarin: lepenja "fifth", otsôja "seventh" (VT42:26, 25). The word kwendjâ, the origin of Quenya, is explained as being an adjective meaning "belonging to the *kwendî, to the people as a whole" (WJ:360, 393). May this wording suggest that kwendjâ comes from *kwendî-â, sc. the plural form kwendî "Elves" + the adjectival terminaison -â?
              The verbal terminaison -jâ, -ja, -iâ is attested in the words barjâ- "to protect" (BAR), berja- "to dare" (BER), beujâ- "follow, serve" (BEW), ramja- "fly, sail; wander" (RAM), tjaliâ- "to play" (TYAL), uljâ "it is raining" (ULU). In the Etymologies, the word barjâ has a diacritic indicating that the terminaison -jâ (or its final voyelle) was accented (BAR). But we cannot conclude that this is always the case; berja "to dare" is marked as accented on the first syllable. (Adjectival -jâ is apparently not accented; cf. banjâ "beautiful".)
              There are only a few nouns in -jâ, -ja: galjâ "bright light" (KAL), gilja "star" (GIL), kegjâ "hedge" (UT:282), talrunja "sole of foot" (TALAM, RUN). Tolkien struck out winjâ "evening" (WIN/WIND). Wanjâ "Vanya" (Quenya pl. Vanyar, the first clan of the Eldar) is really an adjective "fair, beautiful", as noted above (WJ:380, 383). Tolkien also reconstructed the primitive form of Vanya as banjâ (BAN; cf. pl. "Banyai" in PM:402).

    Another adjectival terminaison is -kâ. In Letters:282, Tolkien mentions a "basis" LAY (also present in Quenya lairë "summer") that yields laikâ "green". Other examples include gajakâ "fell, terrible, dire" (PM:363), poikâ "clean, pure" (POY), tiukâ "thick, fat" (TIW); later -kâ became short -ka as in lauka "warm" (LAW). The terminaison -kô, attested only in the word tiukô "thigh" (TIW), would seem to be a nominalized form of -kâ (tiukâ "thick" > tiukô *"thick thing" = "thigh").

    The terminaison -la seems to mean little more than "thing" (or "person"); it is used as a noun-former. Tolkien defines hekla as "any thing (or person) put aside from, or left out from, its normal company" (WJ:361; stem HEKE "aside, apart"); this could be turned into a "personal form" heklô "waif or outcast" with the masculine terminaison -ô; see below. (There is also an adjectival form heklâ formed with the adjectival terminaison -â, discussed above.) In the Etymologies, -la is found in the noms of a number of implements where the terminaison -mâ (see below) could presumably have been used as well: makla "sword" from MAK "sword, fight with sword", tekla "pen" from TEK "write" (hence *"thing for writing"), and, with a nasal-infixed stem, tankla "pin, brooch" from TAK "fix, make fast". In the word magla (read *smaglâ?) "stain" from the stem SMAG- "[?to] soil, stain" the terminaison simply acts as a noun-former. (In the Etymologies, the Sindarin word mael that is referred to magla is glossed both "stain" as a noun and adj. "stained", but the adjective "stained" is presumably derived from *(s)maglâ with adjectival -â.) In one case, the terminaison -la is added, not directly to the root, but to another derived word: Sjatsela/sjatsêla "broadsword-blade", "axe-blade" includes the word sjatsê < sjadsê "cleft, gash" derived from the root SYAD "shear through, clear"; a sjatsêla is thus a *"thing used for making gashes".
              The adjective ndulla "dark, dusky, obscure" may not contain the terminaison -la; it is apparently formed from the root NDUL by "strengthening" the final consonant to double LL and adding the adjectival terminaison -â. Indeed the form in QP and CE must have been *ndullâ with a long final voyelle, for primitive ndulla would have yielded Quenya **nul (null-), but the actual Quenya form is nulla. Ndulla must be understood as being ancient Quenya (after the shortening of the original long final voyelles) rather than primitive Elvish.
              The terminaison -la combined with the adjectival suffix -â produces -lâ, as in heklâ mentioned above. This -lâ would seem to be the origin of the Quenya participial terminaison -la, Sindarin -l.

    The terminaison -lê is used to derive nouns that "seem properly to have been universal or abstract" (VT39:16); this also goes for its direct Quenya descendant -lë. In most attested examples it simply acts as a verbal noun terminaison. The root TUY "spring, sprout" yields tuilê "day-spring" or "spring-time"; the basic meaning would be simply *"springing, sprouting". Keglê comes from keg- "snag, barb" and would mean basically *"snagging, barbing", but abstracts often take on a concrete meaning, and in Sindarin cail (< keglê) means "fence" or "palisade" (UT:282).
              The -rille of silimarille "Silmaril" may be a verbal noun derived from RIL "glitter", so that rille means something like *"radiance, brilliance".
              The -le of nenle "brook" (NEN) may or may not be connected; if it is, the word would mean "watering". But this -le may also be a diminutive terminaison.
              How does ramalê "pinion, great wing (of eagle)" fit in? (RAM)

    The suffix -mâ is one of the most productive terminaisons. Tolkien points out that this suffix is frequent in the noms of implements (WJ:416). Hence the stem TAK "fix, make fast" may yield takmâ "thing for fixing", the origin of Quenya tangwa "hasp, clasp". SUK "drink" yields sukmâ "drinking-vessel". Another word of the same meaning, julmâ, is likewise derived from a stem meaning "drink" (WJ:416 - this is the origin of Quenya yulma "cup", known from Namárië). From the stem YAT "join" comes jatmâ, apparently meaning "bridge" or "joining" (Quenya yanwë). Note that the stem to which -mâ (-ma) is appended is not required to have a verbal meaning; kasma "helmet" comes from a stem KAS "head". Telmâ "hood, covering" comes from a stem (TEL/TELU) that is not defined, but apparently has to do with the top or canopy of something. (In the Etymologies, the final voyelle of telmâ has a diacritic denoting that it may be either long or short, so the variation -mâ vs. -ma is unimportant.)
              Some "implements" may even be body parts, such as nakma "jaw" from NAK "bite", or labmâ "tongue" from LABA "lick" (WJ:416).
              However, not all words in -mâ denote implements. Often the meaning of the terminaison -mâ is very general; it simply denotes an object somehow connected with the state or action denoted by the stem. Parmâ "book" comes from a stem PAR "compose, put together"; a parmâ is simply a "thing that is composed or put together". Sometimes -mâ denotes an impersonal agent, as in tuimâ "a sprout, bud" from TUY "spring, sprout" or tjulmâ "mast" from TYUL "stand up" (but in SD:419, the primitive form of Quenya tyulma is reconstructed as kjulumâ instead). In some cases, -mâ is used simply to derive concrete nouns, as in pathmâ "level space, sward" or sjalmâ "shell, conch, horn of Ulmo" (stems PATH, SYAL not defined). Similarly, skelmâ "skin, fell" comes from a stem SKEL that is not clearly glossed; it may mean "strip, strip bare" (cf. SKAL1). Quenya corma "ring" plainly represents a primitive form *kormâ (not reconstructed by Tolkien); the stem KOR means "round", so a *kormâ is simply a "round thing".
              Infrequently the terminaison -mâ may also denote a substance, as in wilmâ "air, lower air" from the stem WIL "fly, float in air", or sagmâ "poison" from SAG (stem meaning not given; perhaps "bitter").
              The terminaison -mâ also seems to occur in one adjective, silimâ "shining white", "silver" (as adj.) (SIL). But this is probably a longer adjectival terminaison -imâ; see above.

    The terminaison -mê is properly an abstract or verbal noun terminaison, much like English "-ing", as in julmê "drinking, carousal", from the stem JULU "drink" (WJ:416) or labmê "the action of *LABA", sc. a stem having to do with licking or moving the tongue (WJ:416). The nom of the Vala Oromë is really adapted from Valarin (an early Eldarin form was Arâmê), but in later ages the Eldar took the nom to mean "horn-blowing", wrongly supposing that it contained the verbal noun terminaison -mê (WJ:400).
              A number of other words are easily explained as abstract words that have taken on a more concrete meaning, as such words often do: rakmê "fathom" from RAK "stretch out, reach", tekmê "letter, symbol" from TEK "make a mark", tinmê "sparkle, glint" from TIN "to sparkle", tulukmê "support, prop" from TULUK (stem not defined but having to do with being firm or steadfast). Note that English "support" may have both an abstract and a concrete meaning (the act of supporting vs. a tangible prop), illustrating how abstracts and concretes are easily conflated. In one case, the terminaison -mê seems to be confused with -mâ; both telmâ and telmê (or telma, telme) "hood, covering" are mentioned by Tolkien when he etymologizes Quenya telmë "hood" (TEL/TELU). Once again, a full abstract "covering" takes on a concrete meaning: a hood, that should more properly be called a telmâ with the terminaison for implements.
              In a few cases, the terminaison -mê/-me occurs in the noms of substances: khithme "fog" (KHIS/KHITH), silimê "light of Silpion", also a poetic word for "silver" (SIL; this may actually be a nominalized form of the apparently adjectival terminaison seen in silimâ; see -imâ). In one word -mê simply denotes something intangible: do3mê "night" (DO3, see DOMO).

    The agental terminaison -mô is attested in the word Ulumô "Pourer, Ulmo" only (ULU). However, its Quenya descendant -mo is well attested and is stated to be an terminaison that "often appeared in noms or titles, sometimes with an agental significance" (WJ:400; here "the Pourer" as the meaning of Ulmo is said to be an Elvish folk etymology, for the nom was actually adopted and adapted from Valarin Ul(l)ubôz).

    The terminaison -nâ (-na) is very productive. In a few cases (khalnâ, barnâ under KHAL2, BAR) the final voyelle is marked as accented; perhaps this terminaison received the accent in primitive Elvish. Its function is to form adjectives: In UT:266, a word in -nâ is called as an "ancient adjectival form", while in WJ:365 another such word, heklanâ, is called an "extended adjectival form" (extended as compared to the shorter adjectival form heklâ, presumably). Examples include ku3nâ "bowed, bow-shaped, bent" (KU3 "bow"), magnâ "skilled" (MAG, under MA3), ndeuna "second" (NDEW "follow, come behind"), ornâ "uprising, tall" (UT:266), patnâ "wide" (PAT), pathnâ "smooth" (PATH), ragnâ "crooked" (RAG), sarnâ "of stone" (SAR, see STAR), ta3na ?"high, lofty, noble" (TÂ/TA3), tubnâ "deep" (TUB). This terminaison may well be added to stems that already have an adjectival meaning, such as k'rannâ "ruddy (of face)" from KARÁN "red" or mornâ "dark" from MOR "black" (see Letters:382 for mornâ; this derivative is not given in the Etymologies, though its Quenya descendant morna is).
              Sometimes the terminaison -nâ (-na) produces forms that may be considered past participles, as when DUL "hide, conceal" yields ndulna "secret" (or *"hidden, concealed"). Gjernâ "old, worn" may be seen as a past participle if the stem GYER means "to wear (out)", like a Quenya verbal derivative of this stem (yerya) does. Likewise, skelnâ "naked" comes from a stem (SKEL) that may mean "strip bare" (cf. SKAL1 ). Clearly participial are the forms skalnâ "veiled, hidden, shadowed, shady" from SKAL1 "screen, hide (from light)", skarnâ "wounded" from SKAR "tear, rend", and barnâ "safe, protected, secure" from BAR "uplift, save, rescue". We also note wannâ "departed, dead" from WAN "depart, go away, disappear, vanish" and khalnâ "noble, exalted" from KHAL2