Sindarin

INTERNAL HISTORY

Sindarin was the main Eldarin tongue in Middle-earth, the living vernacular of the Grey-elves or Sindar. It was the most prominent descendant of Common Telerin, Common Telerin itself branching off from Common Eldarin, the ancestor of Quenya, Telerin, Sindarin and Nandorin. "The Grey-elven was in origin akin to Quenya," Tolkien explains, "for it was the language of those Eldar who, coming to the shores of Middle-earth, had not passed over the Sea but had lingered on the coasts in the country of Beleriand. There Thingol Greycloak of Doriath was their king, and in the long twilight their tongue...had become far estranged from the speech of the Eldar from beyond the Sea" (LotR Appendix F). Though Sindarin is said to be the best preserved of the Eldarin tongues of Middle-earth (PM:305), it is nonetheless the most radically changed Elvish language we have any extensive knowledge of: "The language of the Sindar had changed much, even in unheeded growth as a tree may imperceptibly change its shape: as much maybe as an unwritten mortal tongue might change in five hundred years or more. It was already ere the Rising of the Sun a speech greatly different to the ear from [Quenya], and after that Rising all change was swift, for a while in the second Spring of Arda very swift indeed" (WJ:20). The development from Common Eldarin to Sindarin involves much more radical changes than the development from CE to Quenya, or to the Telerin of Aman. Tolkien suggested that Sindarin "had changed with the changefulness of mortal lands" (LotR Appendix F). This is not to say that the changes were chaotic and unsystematic; they were definitely regular - but they dramatically changed the general sound and "music" of the language. Some prominent changes include the final vowels being dropped, the unvoiced stops p, t, k becoming voiced b, d, g following a vowel, the voiced stops becoming spirants in the same position (except g, that disappeared altogether) and many vowels being altered, often by assimilation to other vowels. According to PM:401, "the development of Sindarin had become, long before the arrival of the Ñoldorin exiles, mainly the product of unheeded change like the tongues of Men". Commenting on the great changes, PM:78 remarks that "it was a fair tongue still, well fitted to the forests, the hills, and the shores where it had taken shape".
By the time the Noldor returned to Middle-earth, nearly three and a half millennia after their separation from the Sindar, the classical Sindarin language was fully developed. (Indeed it seems to have entered a more stable phase, despite Tolkien's statement that change was swift after the rising of the Sun: the changes that occurred during the next seven thousand years, until Frodo's day, were small indeed compared to the swift development in the previous three thousand years.) In the First Age, there were various dialects of Sindarin - the archaic language of Doriath, the western dialect of the Falathrim or "Shore-people" and the Northern dialect of the Mithrim. Which of these was the basis of the Sindarin spoken in later Ages is not known with certainty, but the tongue of the Falathrim seems the best candidate, since Doriath was destroyed and what very little we know about North Sindarin suggests that it differed from the Sindarin of Frodo's day. (The name Hithlum is North Sindarin; see WJ:400.)
The Noldor and the Sindar were not at first able to understand one another, their languages having drawn too far apart during their long separation. The Noldor learnt Sindarin quickly and even started to render their Quenya names into Grey-elven, for "they felt it absurd and distasteful to call living persons who spoke Sindarin in daily life by names in quite a different linguistic mode" (PM:341). Sometimes the names were adapted with great care, as when Altariel must have been tracked back to its (hypothetical) Common Eldarin form *Ñalatârigellê; starting with this "reconstruction" the Noldor then derived the Sindarin form that would have appeared in Sindarin if there had actually been an ancient name *Ñalatârigellê: Galadriel. The names were not always converted with such care. The prominent name Fëanor is in fact a compromise between pure Quenya Fëanáro and the "correct" Sindarin form Faenor ("correct" in the sense that this is what primitive *Phayanâro would have become in Sindarin, if this name had actually occurred in Common Eldarin in ancient times). Some names, like Turukáno or Aikanáro, were simply Sindarized in sound, though the resulting forms Turgon and Aegnor were pretty meaningless in Grey-elven (PM:345). Many of the name-translations took place very early, before the Noldor had sorted out all the subtleties of Sindarin - therefore the resulting names "were often inaccurate: that is, they did not always precisely correspond in sense; nor were the equated elements always actually the nearest Sindarin forms of the Quenya elements" (PM:342).
But the Noldor, ever ready linguists, soon achieved full mastery of the Sindarin language and sorted out its precise relationship to Quenya. Twenty years after the coming of the Noldor to Middle-earth, during the Mereth Aderthad or Feast of Reuniting, "the tongue of the Grey-elves was most spoken even by the Noldor, for they learned swiftly the speech of Beleriand, whereas the Sindar were slow to master the tongue of Valinor" (Silmarillion ch. 13). Quenya as a spoken tongue was finally abolished by Thingol when he learnt that the Noldor had killed many Teleri and stolen their ships to get back to Middle-earth: "Never again in my ears shall be heard the tongue of those who slew my kin in Alqualondë! Nor in all my realm shall it be openly spoken." Consequently "the Exiles took the Sindarin tongue in all their daily uses" (Silm. ch. 15). It seems that Thingol's edict merely accelerated the process; as noted, many of the Noldor spoke Sindarin already.
Later, mortal Men appeared in Beleriand. Appendix F in LotR (and UT:216) informs us that "the Dúnedain alone of all races of Men knew and spoke an Elvish tongue; for their forefathers had learned the Sindarin tongue, and this they handed on to their children as a matter of lore, changing little with the passing of the years". Perhaps it was the Dúnedain that stabilized the Sindarin language, at least as used among themselves (UT:216 states that Sindarin spoken by mortal Men otherwise "tended to become divergent and dialectal"). Whatever the standard of Mannish Sindarin might have been in later ages, back in the First Age "the most part of [the Edain] soon learned the Grey-elven tongue, both as a common speech among themselves and because many were eager to learn the lore of the Elves" (Silmarillion ch. 17). Eventually, some Men knew and spoke Sindarin just as well as the Elves. The famous lay Narn i Chîn Húrin (as it is properly spelt) was made by a Mannish poet by the name of Dírhavel, "but it was prized by the Eldar, for Dírhavel used the Grey-elven tongue, in which he had great skill" (UT:146. On the other hand, the people of Haleth did not learn Sindarin well or with enthusiasm; see UT:378). Túrin learnt Sindarin in Doriath; one Nellas "taught him to speak the Sindarin tongue after the manner of the ancient realm, older, and more courteous, and richer in beautiful words" (UT:76).
The Elves themselves continued to use Sindarin throughout the First Age. In a Noldo-colony like Gondolin one might have thought that the Noldor would have revived Quenya as their spoken language, but this appears not to have been the case, except in the royal house: "For most of the people of Gondolin [Quenya] had become a language of books, and as the other Noldor they used Sindarin in daily speech" (UT:55). Tuor heard the Guard of Gondolin speak first in Quenya and then "in the tongue of Beleriand [Sindarin], though in a manner somewhat strange to his ears, as of a people long sundered from their kin" (UT:44). Even the Quenya name of the city, Ondolindë, always appears in its Sindarized form Gondolin (though this is a mere adaptation and not "real" Sindarin; primitive *Gondolindê should have produced **Gonglin, if the word was inherited).
Many speakers of Sindarin perished in the wars of Beleriand, but by the intervention of the Valar, Morgoth was finally overthrown in the War of Wrath. Many Elves went to Eressëa when the First Age was ended, and from now on Sindarin evidently became a spoken tongue in the Blessed Realm as well as in Middle-earth (a passage in the Akallabêth, quoted below, indicates that the Númenóreans held converse with the Eressëans in Sindarin). The Valar wanted to reward the Edain for their sufferings in the war against Morgoth and raised an island out of the sea, and Men, following the Star of Eärendil to their new home, founded the realm of Númenor.
Sindarin was widely used in Númenor: "Though this people used still their own speech, their kings and lords knew and spoke also the Elven tongue, which they had learned in the days of their alliance, and thus they held converse still with the Eldar, whether of Eressëa or of the westlands of Middle-earth" (Akallabêth). The descendants of the people of Bëor even used Sindarin as their daily speech (UT:215). Though Adûnaic was the vernacular for most of the Númenórean population, Sindarin was "known in some degree to nearly all" (UT:216). But times later changed. The Númenóreans started to envy the immortality of the Elves, and eventually they turned away from their ancient friendship with Aman and the Valar. When Ar-Gimilzôr "forbade utterly the use of the Eldarin tongues" in the 3100s of the Second Age, we must assume that even the Bëorians dropped Sindarin and took up Adûnaic instead (UT:223). The story of the folly of Ar-Pharazôn, Sauron's cunning "surrender", the total corruption of the Númenóreans and the Downfall of Númenor is well known from the Akallabêth. After the Downfall, the surviving Elf-friends set up the Realms in Exile, Arnor and Gondor, in Middle-earth. PM:315 states: "The Faithful [after the Downfall]...used Sindarin, and in that tongue devised all names of places that they gave anew in Middle-earth. Adûnaic was abandoned to unheeded change and corruption as the language of daily life, and the only tongue of the unlettered. All men of high lineage and all those who were taught to read and write used Sindarin, even as a daily tongue among themselves. In some families, it is said, Sindarin became the native tongue, and the vulgar tongue of Adûnaic was only learned casually as it was needed. The Sindarin was not however taught to aliens, both because it was held a mark of Númenórean descent and because it proved difficult to acquire - far more so than the 'vulgar tongue'." In accordance with this, Sindarin is stated to have been "the normal spoken language of Elendil's people" (UT:282).
Among the Elves themselves, Sindarin crept eastwards in the Second and Third Age and eventually displaced some of the Silvan (Nandorin, Danian) tongues. "By the end of the Third Age, the Silvan tongues had probably ceased to be spoken in the two regions that had importance at the time of the War of the Ring: Lórien and the realm of Thranduil in northern Mirkwood" (UT:257). Silvan was out, Sindarin was in. True, we get the impression from LotR1/II ch. 6 that the language used in Lórien was some strange Wood-elven tongue, but Frodo, the author of the Red Book, got it wrong. A footnote in LotR Appendix F explains that in Frodo's day, Sindarin was indeed spoken in Lórien, "though with an 'accent', since most of its folk were of Silvan origin. This 'accent' and his own limited acquaintance with Sindarin misled Frodo (as is pointed out in The Thain's Book by a commentator of Gondor)". UT:257 elaborates on this: "In Lórien, where many of the people were Sindar in origin, or Noldor, survivors from Eregion, Sindarin had become the language of all the people. In what way their Sindarin differed from the forms of Beleriand - see [LotR1] II 6, where Frodo reports that the speech of the Silvan folk that they used among themselves was unlike that of the West - is not of course now known. It probably differed in little more than what would now be popularly called 'accent': mainly differences of vowel-sounds and intonation sufficient to mislead one who, like Frodo, was not well acquainted with purer Sindarin. There may of course also have been some local words and other features ultimately due to the influence of the former Silvan tongue." Standard Sindarin, with no "accent", was evidently spoken in Rivendell and among Círdan's people in the Havens.
But by the end of the Third Age, the Elves were fading away in Middle-earth, no matter what tongue they spoke. The rule of Mortal Men, the Second-born of Ilúvatar, was about to begin. Tolkien notes that at the end of the Third Age there were more Men who spoke Sindarin or knew Quenya than there were Elves who did either (Letters:425). When Frodo and Sam met Faramir's men in Ithilien, they heard them speak first in the Common Tongue (Westron), but then they changed to "another language of their own. To his amazement, as he listened Frodo became aware that it was the Elven-tongue that they spoke, or one but little different; and he looked at them with wonder, for he knew then that they must be Dúnedain of the South, men of the line of the Lords of Westernesse" (LotR2/IV ch. 4). In Gondor, "Sindarin was an acquired polite language and used by those of more pure N[úmenórean] descent" (Letters:425). The talkative herb-master of the Houses of Healing referred to Sindarin as the "noble tongue" (LotR3/V ch. 8: "Your lordship asked for kingsfoil, as the rustics name it, or athelas in the noble tongue, or to those who know somewhat of the Valinorean [= Quenya]...").
How Sindarin fared in the Fourth Age we shall never know. Like Quenya, it must have been remembered as long as the realm of Gondor endured.

Designations of the language

"Sindarin" is the Quenya name of this language, derived from Sindar *"Grey ones" = Grey-elves; it may be (and is) translated Grey-elven. What Sindarin was called by its own term is not known with certainty. It is said of the Elves in Beleriand that "their own language was the only one that they ever heard; and they needed no word to distinguish it" (WJ:376). The Sindar probably referred to their own tongue simply as Edhellen, "Elvish". As noted above, the herb-master of the Houses of Healing referred to Sindarin as the "noble tongue" (while "the noblest tongue in the world" remains Quenya, UT:218). Throughout LotR, the term usually employed is simply "the Elven-tongue", since Sindarin was the living vernacular of the Elves.

EXTERNAL HISTORY

In 1954, in Letters:176, Tolkien stated that "the living language of the Western Elves (Sindarin or Grey-elven) is the one usually met [in LotR], especially in names. This is derived from an origin common to it and Quenya, but the changes have been deliberately devised to give it a linguistic character very like (though not identical with) British-Welsh: because that character is one I find, in some linguistic moods, very attractive; and because it seems to fit the rather 'Celtic' type of legends and stories told of its speakers". Later, he found that "this element in the tale has given perhaps more pleasure to more readers than anything else in it" (MC:197).
A Welsh- or Celtic-sounding language was present in Tolkien's mythos from the beginning. This language was originally called Gnomish or I·Lam na·Ngoldathon, "the tongue of the Gnomes (Noldor)". Tolkien's original Gnomish dictionary, dating from about 1917, was published in Parma Eldalamberon #11 and turns out to be a very comprehensive document, with thousands of words. Many Gnomish words are also found in the appendices to LT1 and LT2. Parma also published a (never completed) Gnomish grammar. But though Tolkien put much work into this language, it was in effect rejected later. In PM:379, in a late document, Tolkien refers to Gnomish as "the Elvish language that ultimately became that of the type called Sindarin" and notes that it "was in a primitive and unorganized form". Some of the central concepts of Gnomish grammar, in particular certain consonant mutations, were later recycled in Sindarin. A number of Gnomish vocabulary items also survived into Sindarin, unchanged or in recognizable forms. Even so, Gnomish was really a wholly different language, though it had a phonetic style somewhat similar to that of Sindarin (lots of ch's and th's, and most words end in a consonant!) An important feature of Sindarin, the umlaut or affection of vowels, reportedly first appears in grammars written by Tolkien in the twenties. But only in the thirties, with the Etymologies, did a language really close to LotR-style Sindarin emerge in Tolkien's notes. This was however called "Noldorin", for like its predecessor Gnomish it was conceived as the language, not of the Sindar, but of the Noldor - developed in Valinor. At this stage, Quenya was thought of as the language of the "Lindar" (later: Vanyar) only. Only as late as when the appendices to LotR were being written did Tolkien abandon this idea, and turned Noldorin into Sindarin. Quenya now became the original language of both the Vanyar and the Noldor - the latter simply adopted Sindarin when they arrived in Middle-earth. It "turned out" that the Celtic-sounding language of Tolkien's mythos was not, after all, their own tongue (though in the annals of Middle-earth, they certainly came to be the most prominent users of it). It did not originate in the Blessed Realm of Valinor, but was an indigenous tongue of Middle-earth.
In the former conception, the native Elves of Beleriand spoke a language called Ilkorin, that Sindarin in effect displaced when Tolkien made this revision (Edward Kloczko has argued that some elements of Ilkorin were maintained as the northern dialect of Sindarin; his article is appended to my own treatise about Ilkorin). Tolkien's decision to fundamentally revise the history of the Celtic-sounding language of his mythos was probably a happy one, making the linguistic scenario much more plausible: Surely it was difficult to imagine that the Vanyar and the Noldor could have developed two languages as markedly different as Quenya and "Noldorin" when they lived side by side in Valinor. Turning "Noldorin" into Sindarin took care of that problem; now the two branches of Elvish could develop wholly independently during the long ages their speakers lived in absolute separation from one another.
The "Noldorin" of the Etymologies is not entirely identical to Sindarin as it appears in LotR, since Tolkien never stopped refining and altering his invented languages. But many of the differences that separate "Noldorin" from LotR-style Sindarin are happily regular, Tolkien adjusting some details of the evolution from Primitive Elvish. Therefore, most of the "Noldorin" material can quite easily be updated to agree with the linguistic scenario of LotR. A number of words must be subtly altered; for instance, the "Noldorin" diphthong oe should rather be ae in Sindarin. One example involves Belegoer as a name of the Great Ocean (LR:349, 352); this form Tolkien later changed to Belegaer - so on the map of the published Silmarillion. Another change has to do with the consonants lh- and rh-; where they occurred in "Noldorin" many examples show that Sindarin should have simple l- and r- instead. Thus, we can deduce that a "Noldorin" word like rhoeg ("wrong", LR:383) should rather be raeg in Sindarin - though the latter form is nowhere explicitly attested. It has been suggested that the "Noldorin" of the Etymologies, with its various peculiarities, can be equated with the "somewhat strange" dialect of Sindarin that the Noldor spoke in Gondolin (UT:44). In this way we could even account for its being called Noldorin rather than Sindarin. However, it is also possible that Tolkien would have considered "Noldorin" wholly obsolete to the extent it differs from his later vision of Sindarin.

ELEMENTARY PHONOLOGY

Sindarin phonology is less restrictive than that of Quenya. Many consonant clusters are allowed in all positions, while initial and final clusters are virtually absent in Quenya. The sounds ch (German ach-Laut, NOT "tsh" as in English church) and th, dh ("th" as in think and this, respectively) are frequent. Tolkien sometimes used the special letter eth (ð) to spell dh, and occasionally we also see the letter thorn (þ) instead of th. However, we will here use the digraphs, as in LotR. The unvoiced plosives p, t, c never occur following a vowel, but are lenited (see below) to b, d, g. Note that as in Quenya, c is always pronounced k (standard example: Celeborn = "Keleborn", not "Seleborn"). At the end of words, f is pronounced v, as in English of. (In Tengwar spelling, a word like nef is actually spelt nev.) R should be trilled, as in Spanish, Russian etc. The digraphs rh and lh represent unvoiced r and l (but sometimes these combinations may actually mean r + h or l + h, as in Edhelharn - not surprisingly, our alphabet cannot represent Sindarin quite adequately).

Sindarin has six vowels, a, e, i, o, u and y, the last of which corresponds to German ü or French u as in Lune (pronounce ee as in English see with rounded lips as when you pronounce oo, and you've got it). Long vowels are marked with an accent (á, é etc.), but in the case of stressed monosyllables the vowels tended to become especially long and are marked with a circumflex: â, ê etc. In HTML one unfortunately cannot place a circumflex above the vowel y. To avoid ugly spellings like my^l ("gulls", WJ:418), we here use an accent instead (the relevant words occurring in this article are býr, thýn, fýr, rýn, mrýg, mýl, 'lýg and hýn - ideally these should have had a circumflex instead). This is not very critical: In Tengwar writing, no distinction is made between long and super-long vowels; the use of circumflexes instead of accents in monosyllables is merely an extra complication Tolkien introduced in his Roman orthography for Sindarin (evidently to make it abundantly clear how the words are to be pronounced).
The Sindarin diphthongs include ai (as in English aisle, NOT as in mail), ei, ui (as "ooy" in too young) and au (as in German Haus, or as "ow" in English cow). At the end of words, au is spelt aw. There are also the diphthongs ae and oe, with no English counterparts; Tolkien actually suggests substituting ai and oi if you don't care about such details (indeed he sometimes anglicized Maedhros as "Maidros", but anyone reading this document probably does care about the details). Ae and oe are simply the vowels a, o pronounced in one syllable with the vowel e (as in English pet), just like ai and oi are a and o pronounced together with i. Somewhat confusingly, in Tolkien's writings the digraph oe is sometimes also used to signify umlauted o, apparently the same sound as German ö (actually we often prefer the spelling ö in this article, to avoid confusion). By the end of the Third Age, ö had merged with e (that's why the Grey Mountains appear as Ered Mithrin and not Öröd Mithrin on the Map to LotR!), but we still need to refer to this sound when discussing archaic Sindarin.

THE CORPUS

Important samples of Sindarin in LotR include:

# Glorfindel's greeting to Aragorn: Ai na vedui Dúnadan! Mae govannen! (LotR1/I ch. 12). The first words are not translated, but probably mean *"Ah, at last, Westman!" Mae govannen means "well met" (Letters:308).

# Glorfindel's cry to his horse: Noro lim, noro lim, Asfaloth! (same chapter). Untranslated; evidently meaning *"run fast, run fast, Asfaloth!" (Variants of this line have been transferred to Arwen in the Peter Jackson movie, since the movie-makers dropped the Glorfindel character.) The name of the horse cannot be interpreted, but seems to include loth "flower".

# Gandalf's fire-spell: Naur an edraith ammen! Naur dan i ngaurhoth! The first part literally means, according to TI:175, "fire be for saving of us". (Actually there seems to be no word meaning "be".) The second part must mean *"fire against the werewolf-host!" (Cf. Gandalf's remark the morning after the wolf-attack: "It is as I feared. These were no ordinary wolves.") (LotR1/II ch. 4)

# Gandalf's invocation before the Moria Gate: Annon edhellen, edro hi ammen! Fennas nogothrim, lasto beth lammen! "Elvish gate open now for us; doorway of the Dwarf-folk listen to the word of my tongue" (LotR1/II ch. 4, translated in RS:463). An earlier variant of the invocation is found in RS:451.

# The inscription on the Moria Gate itself: Ennyn Durin Aran Moria: pedo mellon a minno. Im Narvi hain echant: Celebrimbor o Eregion teithant i thiw hin. "The Doors of Durin, Lord of Moria. Speak, friend, and enter. I, Narvi, made them. Celebrimbor of Hollin [Eregion] drew these signs."

# The song A Elbereth Gilthoniel / silivren penna míriel / o menel aglar elenath! / Na-chaered palan-díriel / o galadhremmin ennorath, / Fanuilos le linnathon / nef aear, sí nef aearon (LotR1/II ch. 1). It is translated in RGEO:72 and means roughly, "O Elbereth Starkindler, white-glittering, sparkling like jewels, the glory of the starry host slants down. Having gazed far away from the tree-woven lands of Middle-earth, to thee, Everwhite, I will sing, on this side of the Sea, here on this side of the Ocean" (my translation based on Tolkien's interlinear rendering). An earlier variant of the song is found in RS:394. (The hymn is quite similar to Lúthien's Song [untranslated] in The Lays of Beleriand p. 354: Ir Ithil ammen Eruchîn / menel-vîr síla díriel / si loth a galadh lasto dîn! / A Hîr Annûn gilthoniel, le linnon im Tinúviel.)

# Sam's "inspired" cry in Cirith Ungol: A Elbereth Gilthoniel o menel palan-diriel, le nallon sí di-nguruthos! A tiro nin, Fanuilos! "O Elbereth Star-kindler, from heaven gazing afar, to thee I cry now in [lit. beneath] the shadow of death. O look towards me, Everwhite!" (translated in Letters:278 and RGEO:72).

# The praise received by the Ringbearers on the Fields of Cormallen (LotR3/VI ch. 4): Cuio i Pheriain anann! Aglar'ni Pheriannath! ... Daur a Berhael, Conin en Annûn, eglerio! ... Eglerio! This is translated in Letters:308 and means "may the Halflings live long, glory to the Halflings... Frodo and Sam, princes of the west, glorify (them)! ... Glorify (them)!"

# Gilraen's linnod to Aragorn in LotR Appendix A: Ónen i-Estel Edain, ú-chebin estel anim, translated "I gave Hope to the Dúnedain; I have kept no hope for myself".

Outside LotR, the most important source - indeed the longest Sindarin text we have, and the longest prose text in any Elvish tongue - is the King's Letter, a part of the Epilogue to LotR, that Tolkien later dropped. It was finally published in SD:128-9: Elessar Telcontar: Aragorn Arathornion Edhelharn, aran Gondor ar Hîr i Mbair Annui, anglennatha i Varanduiniant erin dolothen Ethuil, egor ben genediad Drannail erin Gwirith edwen. Ar e aníra ennas suilannad mhellyn în phain: edregol e aníra tírad i Cherdir Perhael (i sennui Panthael estathar aen) Condir i Drann, ar Meril bess dîn; ar Elanor, Meril, Glorfinniel, ar Eirien sellath dîn; ar Iorhael, Gelir, Cordof, ar Baravorn, ionnath dîn. A Pherhael ar am Meril suilad uin aran o Minas Tirith nelchaenen uin Echuir. (The names Elessar Telcontar are Quenya; the Sindarin translation of Elessar, Edhelharn [Elfstone], occurs in the text.) This translation is given in SD:128: "Aragorn Strider the Elfstone [but the Elvish text reads "Elessar Telcontar: Aragorn Arathornson Elfstone"], King of Gondor and Lord of the Westlands, will approach the Bridge of Baranduin on the eighth day of Spring, or in the Shire-reckoning the second day of April. And he desires to greet there all his friends. In especial he desires to see Master Samwise (who ought to be called Fullwise), Mayor of the Shire, and Rose his wife; and Elanor, Rose, Goldilocks, and Daisy his daughters; and Frodo, Merry, Pippin and Hamfast, his sons. To Samwise and Rose the King's greeting from Minas Tirith, the thirty-first day of the Stirring [not in the Elvish text:], being the twenty-third of February in their reckoning." The words in the parenthesis ("who ought to...") are omitted from the translation in SD:128, but cf. SD:126.

Other samples of Sindarin include:

# Voronwë's uttering when he saw the Encircling Mountains around the realm of Turgon: Alae! Ered en Echoriath, ered e·mbar nín! "Alae [= ?behold]! [The] mountains of Echoriath, [the] mountains of my home!" (UT:40, translated in UT:54 note 19.)

# Gurth an Glamhoth!, "death to [the] din-horde", Tuor cursing the Orcs in UT:39 (cf. UT:54).

# The battle-cry of the Edain of the North, given in UT:65: Lacho calad! Drego morn! "Flame Light! Flee Night!"

# An exclamation of Húrin's: Tôl acharn, "Vengeance comes", also in the form Tûl acharn (WJ:254, 301).

# The Sindarin names of the certain Great Tales in the Silmarillion, the Nern in Edenedair or *"Tales of the Fathers of Men", given in MR:373: 1) Narn Beren ion Barahir, "Tale of Beren son of Barahir", also called Narn e·Dinúviel, "Tale of the Nightingale". 2) Narn e·mbar Hador *"Tale of the house of Hador" including Narn i·Chîn Hurin "Tale of the Children of Hurin" (also called Narn e·'Rach Morgoth "Tale of the Curse of Morgoth") and Narn en·Êl "Tale of the Star" (or Narn e·Dant Gondolin ar Orthad en·Êl, *"Tale of the Fall of Gondolin and the Rising of the Star").

# A sentence published in VT41:11: Guren bêd enni "my heart (inner mind) tells me".

# An incomplete translation of the Lord's Prayer, published in VT44:21, 22: Ae Adar nín i vi Menel / no aer i eneth lín / tolo i arnad lín / caro den i innas lin / bo Ceven sui vi Menel. / Anno ammen sír i mbas ilaurui vín / ar díheno ammen i úgerth vin / sui mín i gohenam di ai gerir úgerth ammen. In a more-or-less literal translation, this is apparently: "O my [sic!] father who [is] in heaven, / be holy your name / let your kingdom come / make ?it [happen,] your will / on Earth as in Heaven. / Give to us today our daily bread / and forgive us our wrong-doing / like us who forgive those who do wrong-doing to us."

# A sentence from the so-called "Túrin Wrapper": Arphent Rían Tuorna, Man agorech?, probably meaning *"And Rían said to Tuor, What did you do?" (Compare agor "did" in WJ:415. The full contents of the Túrin Wrapper will "soon" be published and discussed in Vinyar Tengwar...or so Carl F. Hostetter wrote in TolkLang message 21.09 back in 1996.)

THE STRUCTURE OF SINDARIN

The most distinctive feature of Sindarin as a language is probably the complex phonology, Grey-elven often relying on phonological features such as umlauts and mutations instead of affixes to express various grammatical ideas. We shall have to touch on such matters quite often in our attempt to survey the structure of Sindarin.

1. THE ARTICLES

Like Quenya, Sindarin has no indefinite article like English "a, an"; the absence of a definite article indicates that the noun is indefinite: Edhel = "Elf" or "an Elf".

The definite article, "the", is i in the singular: aran "king", i aran "the king". These examples might just as well be Quenya. In an untranslated text in The Lays of Beleriand p. 354 we find the phrase ir Ithil. If this means *"the moon", it would seem to indicate that the article takes the form ir before a word in i- (to avoid two identical vowels in hiatus). However, since this theory was first advanced a new relevant example has been published. The Sindarin Lord's Prayer includes the phrase i innas lin "your will" or literally *"the will of yours". Here we do have i, not ir, even though the next word begins in i-. Moreover, the word for "Moon", Ithil, seems to count as a proper name in Sindarin, so we would not expect it to take any article at all. Some therefore think the ir of the phrase ir Ithil is not a variant of the definite article "the", but has another meaning.

Unlike Quenya (and English), Sindarin has a special plural form of the article, in. "Kings" is erain (formed from aran by vocalic umlauts, see below); "the kings" is in erain.

In both the singular and the plural, the article may appear as a suffix appended to prepositions. This suffix has the form -n or -in. Thus the preposition na "to" becomes nan "to the". Ben "in the" or more literally *"according to the", a word occurring in the King's Letter, seems to be a preposition be "according to" - not attested by itself - with the suffix -n for "the". (This be would be the Sindarin cognate of Quenya ve "like, as".) The preposition nu (or no) "under" becomes nuin "under the" (as in Dagor-nuin-Giliath "Battle under the Stars", a name occurring in the Silmarillion, chapter 13). When the article occurs in the form -in, it may trigger phonological changes in the word it is appended to. Or "over, on" turns into erin "on the", the vowel i umlauting o to e (via ö; "on the" must have been örin at an earlier stage). The preposition o "from, of" appears as uin when the article is suffixed, since in Sindarin earlier oi becomes ui (cf. Uilos as the cognate of Quenya Oiolossë). One might think that the ending -in added to prepositions corresponded to the independent article in for plural "the", so that words like erin or uin would be used in conjunction with plural words only. But the King's Letter demonstrates that this is not the case; here we find these words used together with singulars: erin dolothen Ethuil "on the eighth [day] of Spring", uin Echuir "of the Stirring" (month-name). Presumably -n, -in suffixed to prepositions represents an oblique form of the article that is used both in the singular and the plural. - In some cases, the normal, independent article is used following an independent preposition, just as in English: cf. naur dan i ngaurhoth *"fire against the werewolf-host" in one of Gandalf's firespells. Dan i "against the" is not replaced by a single word, sc. some form of dan "against" with the article suffixed. Perhaps some prepositions just can't receive a suffixed article, or perhaps it is optional whether one wants to say nan or na i(n) for "to the", erin or or i(n) for "on/over the", uin or o i(n) for "of/from the". We don't know.

The genitival article: Sindarin often expresses genitival relationships by word order alone, like Ennyn Durin "Doors (of) Durin" and Aran Moria "Lord (of) Moria" in the Moria Gate inscription. However, if the second word of the construction is a common noun and not a name as in these examples, the genitival article en "of the" is used if the noun is definite. Cf. names like Haudh-en-Elleth "Mound of the Elf-maid" (Silmarillion ch. 21), Cabed-en-Aras "Deer's Leap", *"Leap of the Deer" (UT:140), Methed-en-Glad "End of the Wood" (UT:153) or the phrase orthad en·Êl "Rising of the Star" in MR:373. Cf. also Frodo and Sam being called Conin en Annûn "princes of the West" on the Field of Cormallen. (This genitival article sometimes takes the shorter form e; cf. Narn e·Dinúviel "Tale of the Nightingale", MR:373. See below, in the section about consonant mutations, concerning the various incarnations of this article and the environments in which they occur.) Only infrequently does the normal sg. article i replace e(n)- in genitival phrases, but in the King's Letter we have Condir i Drann for "Mayor of the Shire". But in the plural, the normal pl. article in is normally used even in a genitival construction, cf. Annon-in-Gelydh "Gate (of) the Noldor" (UT:18), Aerlinn in Edhil *"Hymn (of) the Elves" (RGEO:70, in Tengwar writing). However, there are examples of the explicitly genitival article en being used in the plural as well: Bar-en-Nibin-Noeg, "Home of the Petty-dwarves" (UT:100), Haudh-en-Ndengin "Hill of Slain", or *"of the Slain Ones" (Silmarillion ch. 20). This seems to be less usual, though.

In many cases, the articles cause the initial consonant of the following word to change. These phonological intricacies are described below, in the section about consonant mutations. The article i triggers lenition or soft mutation of the following noun; see below. The final n of the article in is often swallowed up in a process called nasal mutation; the n disappears and the initial consonant of the noun is changed instead. On the other hand, the nasal of the suffix -n or -in, "the" appended to prepositions, apparently persists - though it seems to trigger what we tentatively call mixed mutation in the following word.

The articles are also used as relative pronouns; cf. Perhael (i sennui Panthael estathar aen) "Samwise (who ought to be called Panthael)" in the King's Letter, or the name Dor Gyrth i chuinar "Land of the Dead that Live" (Letters:417 - this represents *Dor Gyrth in cuinar, an example of nasal mutation. Dor Firn i Guinar in the Silmarillion ch. 20 employs singular i as a relative pronoun even though Firn is plural; the reading Dor Gyrth i chuinar from a very late letter (1972) is to be preferred).

It will be noted that Tolkien sometimes, but not always, connects the Sindarin articles to the next word by means of a hyphen or a dot. This is apparently optional. In this work, when not quoting the sources directly, we connect the genitival article e, en "of the" to the next word by means of a hyphen (since it would otherwise often be hard to tell apart from the preposition ed, e "out of"), but we do not hyphenate the other articles.

2. THE NOUN

#

In the fictional timeline, the Sindarin noun originally had three numbers: singular, plural and dual. However, we are told that the dual form early became obsolete except in written works (Letters:427). On the other hand, a so-called class plural developed, coexisting with the "normal" plural; see below.

As in most languages, the singular is the basic, uninflected form of the noun. Tolkien noted that the Sindarin plurals "were mostly made with vowel-changes" (RGEO:74). For instance, amon "hill" becomes emyn "hills"; aran "king" becomes erain "kings". The consonants remain the same, but the vowels change. There are a few English nouns that form their plurals in a similar way: man pl. men, woman pl. women (pronounced "wimen"), goose pl. geese, mouse pl. mice etc. Yet English usually relies on the plural ending -s. In Sindarin, the situation is the opposite: the trick of changing the vowels is the usual way of forming the plurals, and only a few words display some kind of ending in the plural. The rules for these vowel-changes are the same for both nouns and adjectives (the latter agree in number), so we will also quote adjectives among the examples as we explore the Sindarin plural patterns. Ultimately, the vowel-changes go back on so-called umlaut phenomena. Umlaut (in origin a German term literally meaning something like "changed sound") is an important feature of Sindarin phonology; the Sindarin term for this phenomenon is prestanneth, meaning disturbance or affection. It has to do with one vowel "affecting" another vowel in the same word, making it more like itself, in linguistic terms assimilating it. The umlaut relevant for the plural formation Tolkien referred to as "i-affection" (WJ:376), since it was a vowel i that originally triggered it. Tolkien imagined that the primitive Elvish language had a plural ending *-î, still present in Quenya as -i (as in Quendi, Atani, Teleri etc). This ending as such did not survive into Sindarin, but there are clear traces of its former presence, and these "traces" have themselves become the indicator of plurality in Grey-elven. When the plural form of, say, fang "beard" (as in Fangorn "Treebeard") is feng, this is because the a was affected by the old plural ending *-î, -i while the latter was still present. In the most primitive form of Elvish, the word for "beard" appeared as spangâ, plural spangâi; by the stage we call Old Sindarin, this had become sphanga pl. sphangi. The former yielded "Classical" Sindarin fang, but the plural sphangi became feng, the original vowel a drifting towards the quality of the plural ending -i before the ending was lost - and so in the later plural form feng we have e as a kind of compromise between (the original vowel) a and (the lost ending) i. (It may be that there was an intermediate stage that had ei, hence ?feing.)
SINDARIN PLURAL PATTERNS
When "affected" or "umlauted", the various vowels and diphthongs undergo different changes. The precise environment and the phonological history must sometimes be taken into account to determine how the word would appear in the plural. We will list the vowels by their "normal" or unaffected forms.

# The vowel A: An a occurring in the final syllable of a word usually turns into ai in the plural. This also applies when the final syllable is also the only syllable, sc. the word is monosyllabic (in such words we often see long â). The example we used above, fang pl. feng instead of **faing, is somewhat atypical (see below); otherwise this pattern is relatively well attested:

tâl "foot", pl. tail (singular in LR:390 s.v. TAL; the plural tail is attested in lenited form -dail in the compound tad-dail "bipeds" in WJ:388)
cant "shape", pl. caint (singular in LR:362 s.v. KAT; for the pl. form cf. morchaint = "dark shapes, shadows" in the Silmarillion Appendix [entry gwath, wath]; this is mor "dark" + caint "shapes", c here becoming ch for phonological reasons)
rach "wagon, wain", pl. raich (cf. Imrath Gondraich "Stonewain Valley" in UT:465)
barad "tower", pl. beraid (Silmarillion Appendix, entry barad)
lavan "animal", pl. levain (WJ:416)
aran "king", pl. erain (LR:360 s.v. 3AR)

NOTE: In the "Noldorin" of the Etymologies, a in a final syllable often comes out as ei instead. Hence we have adar "father" pl. edeir (entry ATA), Balan "Vala" pl. Belein (BAL), habad "shore" pl. hebeid (SKYAP), nawag "dwarf" pl. neweig (NAUK), talaf "ground, floor" pl. teleif (TAL). Same thing in monosyllables: Dân "Nandorin elf", pl. Dein (NDAN), mâl "pollen" pl. meil (SMAL), pân "plank" pl. pein (PAN), tâl "foot" pl. teil (TAL). But as demonstrated above, the plural form of tâl had become tail in Tolkien's later Sindarin (lenited form -dail in tad-dail in WJ:388). Likewise, the Sindarin plural of adar is seen to be, not edeir as in the Etymologies, but edair (as in Edenedair "Fathers of Men", MR:373 - this is a post-LotR source). The Silmarillion Appendix, entry val-, also confirms that in Sindarin the plural form of Balan "Vala" is Belain, not Belein as in the Etymologies. It seems that in all the examples just listed, we should read Sindarin ai for "Noldorin" ei in the plural forms. In one case at least, evidence from the Etymologies agrees with the patterns observed in later Sindarin: the already-quoted example aran "king" pl. erain (not *erein) in the entry 3AR. (For erain as the Sindarin plural, compare the name Fornost Erain "Norbury of the Kings" occurring in LotR3/VI ch. 7.) Interestingly, Christopher Tolkien notes that in the Etymologies, the group of entries that 3AR belongs to was "struck out and replaced more legibly" (LR:360). Perhaps this was after his father had revised the plural patterns that otherwise persist in Etym. PM:31, reproducing a draft for a LotR Appendix, shows Tolkien changing the plural of Dúnadan from Dúnedein to Dúnedain. It seems that the older "Noldorin" plurals in ei are not conceptually obsolete; they may be seen as archaic Sindarin: In certain environments, the change ei > ai occurred also within the imagined history, so Dúnedain could indeed have been Dúnedein at an earlier stage. It seems that Tolkien decided that ei in the final syllable of a word (this also goes for monosyllables) became ai, but otherwise remained ei. Hence we have teithant for "drew" (or *"wrote") in the Moria Gate inscription, and this teith- is related to the second element -deith of the word andeith "longmark" (a symbol used to mark long vowels in writing, LR:391 s.v. TEK). Yet the word andeith from the Etymologies instead appears as andaith in LotR Appendix E, since ei was here in a final syllable. Teithant could not become **taithant because ei here is not in a final syllable. Other words confirm this pattern. As indicated above, the normal plural of aran is erain, but erein- is seen in the name Ereinion "Scion of Kings" (a name of Gil-galad, PM:347/UT:436). Evidently the plural form was erein in archaic Sindarin, later becoming erain because ei changed to ai in final syllables, but in a compound like Ereinion the diphthong ei was not in a final syllable and therefore remained unchanged.

In words of a particular shape, a in the final (or only) syllable becomes e instead of ai. In the plural forms, a may first have become ei as usual, but then the final element of the diphthong was evidently lost (before ei turned into ai) leaving only e that simply remained unchanged later. MR:373 indicates that the plural form of narn "tale" is nern, not **nairn or **neirn, though the latter may have occurred at an earlier stage. It seems that we have e rather than ei/ai before ng as well; the Etymologies provides the example Anfang pl. Enfeng (not **Enfaing) for "Longbeards", one of the tribes of the Dwarves (LR:387 s.v. SPÁNAG). WJ:10, reproducing a post-LotR source, confirms that the plural Enfeng was still valid in Tolkien's later Sindarin. Following the example of fang "beard" pl. feng it would seem that the plural of words like lang "cutlass, sword" (for "Noldorin" lhang, LR:367), tang "bowstring" or thang "need" should be leng, teng, theng.

NOTE: In the Etymologies, there are further examples of "Noldorin" plurals where a in a final syllable becomes e instead of ai or ei. We have adab "construction, building" pl. edeb (TAK), adar "father" pl. eder besides edeir (ATA), Balan "Vala" pl. Belen besides Belein (BAL), falas "beach, shore" pl. feles (PHAL/PHALAS), nawag "dwarf" pl. neweg besides neweig (NAUK), rhofal "pinion" pl. rhofel (RAM) and salab "herb" pl. seleb (SALÂK-WÊ). However, in the case of these words there seems to be little reason to believe that the e-plurals would still be valid in Tolkien's later Sindarin. At least two of these "Noldorin" plurals - eder and Belen - clash with the attested Sindarin plurals edair and Belain. It seems, then, that we can feel free to replace also edeb, feles, neweg, rhofel, seleb with Sindarin edaib, felais, newaig, rovail, selaib, though the latter forms are not directly attested (notice that "Noldorin" rhofal "pinion", pl. rhofel, must become roval pl. rovail if we introduce Sindarin phonology and spelling). - Another "Noldorin" case of an a > e plural is rhanc "arm" pl. rhenc (RAK). The singular must become ranc if we update it to LotR-style Sindarin, but should the plural be renc or rainc? The Sindarin example cant "shape" pl. caint (see above) seems to indicate that a before a cluster consisting of n + an unvoiced stop becomes ai in the plural; hence "arms" should probably be rainc in Sindarin.

In one word at least, earlier ei stays unchanged and does not turn into ai even though it occurs in a final syllable. According to UT:265, the plural form of alph "swan" is eilph; it would seem that ei is unchanged before a consonant cluster beginning in l. (Earlier, in the "Noldorin" of the Etymologies, the word for "swan" was spelt alf, and its plural was given as elf: LR:348 s.v. ÁLAK; for the plural form, cf. hobas in Elf *"Haven of Swans" in LR:364 s.v. KHOP.) In accordance with the example eilph, the Sindarin plural of lalf "elm-tree" should probably be leilf, though the "Noldorin" plural listed in the Etymologies was lelf (LR:348 s.v. ÁLAM).

In a non-final syllable, a becomes e in plural forms, as is seen in some of the examples already quoted: aran "king", pl. erain; amon "hill", pl. emyn; lavan "animal", pl. levain. This does not only go for the vowel in a second-to-last syllable as in these examples; it can be carried through a longer word as well, a in any non-final syllable turning into e. This goes even if a occurs several times: According to WJ:387, the word Aphadon "Follower" becomes Ephedyn in the plural. LR:391 s.v. TÁWAR indicates that the adjective tawaren "wooden" has the plural form tewerin. In MR:373 we have Edenedair for "Fathers of Men", the plural of a compound Adanadar "Man-father" (adan "man" + adar "father"). Here we see a in the final syllable becoming ai, but in all three non-final syllables, a becomes e. Of course, the plural of adan would be edain (well attested) if the word occurred by itself, since the second a would then be in the final syllable. But in the compound Adanadar it is not, and so we see Eden- in the plural.
# The vowel Y: As far as we can imagine, this vowel (long or short) cannot change in the plural. A word like ylf "drinking-vessel" (WJ:416) in all likelihood covers plural "drinking-vessels" as well; there simply isn't anything the umlaut can "do" with such a vowel, just like it cannot change the vowel i. We lack any explicit example of a word with the vowel y occurring both in the singular and the plural, but in WJ:418 we find Bar-i(n)-Mýl for "Home of the Gulls". Likely the word for "gull" is mýl in the singular as well (this would be the case if it is derived from the stem MIW "whine" in LR:373, though a quite different "Noldorin" word for "gull" is there given - quite different because the forms listed there, Quenya maiwë and "Noldorin" maew, clearly reflect an a-infixed stem *MAIW-).

# The diphthong AU: In the "Noldorin" of the Etymologies, words containing this diphthong are seen to have plural forms in ui:

gwaun "goose", pl. guin (LR:397 s.v. WA-N)
naw "idea", pl. nui (LR:378 s.v. NOWO)
rhaw "lion", pl. rhui (LR:383 s.v. RAW)
saw "juice", pl. sui (LR:385 s.v. SAB)
thaun "pine-tree", pl. thuin (LR:392 s.v. THÔN)

However, it seems that this is one feature of "Noldorin" that did not survive into Tolkien's later Sindarin: In UT:148 we have Nibin-noeg as a name of the Petty-dwarves, and the final element is obviously a plural form of naug (cf. Naugrim as a name of the Dwarvish race, found in the Silmarillion). So in Sindarin, au turns into oe in the plural. In the plural forms of the "Noldorin" words listed above, we should apparently read oe instead of ui if we update them to later Sindarin. ("Noldorin" rhaw pl. rhui would become Sindarin raw pl. roe, but thaun "pine-tree" Tolkien apparently changed to Sindarin thôn; cf. Treebeard singing about Dorthonion and Orod-na-Thôn in LotR2/III ch. 4; the Silmarillion Index explains that Dorthonion means "Land of Pines". In the Etymologies, thôn had been an "Ilkorin" word. The pl. of thôn as a Sindarin word is presumably thýn.)

NOTE: The diphthong au, when occurring in an unstressed syllable in the second element of a compound, is often reduced to o, but presumably it would still become oe in the plural. Hence the plural form of a word like balrog "demon of might" (where the -rog part represents raug "demon") is presumably belroeg - unless analogy prevailed to produce ?belryg.

# Other diphthongs: For the most part we lack wholly good examples, but if our understanding of general Sindarin phonology holds water, the diphthongs ae, ai, ei, ui do not normally change in the plural (except that ai in one special category of words normally becomes plural î; see below). As in the case of the vowels i and y, there just isn't much the umlaut can "do" to these diphthongs, so a word like aew "bird" probably covers "birds" as well. For the diphthong ui, at least, we have attested examples: The adjective "blue" is seen to be luin both in the singular and the plural (see note below). The numerous adjectives in -ui also seem to be unchanged in the plural; in the King's Letter we have i Mbair Annui for "the Westlands" or literally *"the Lands Western", where the adjective annui "western" must be plural to agree with "lands". Unfortunately this adjective is not otherwise attested, but there is no reason whatsoever for believing that its singular form would be any different (compare annûn "west" - and as noted above, there are many other adjectives in -ui).
NOTE: In a phrase like Ithryn Luin "Blue Wizards" (UT:390) the adjective luin "blue" must be plural to agree with "wizards". It might be thought that luin is the plural form of lûn, which is what we would get if we were to make a Sindarin update of the "Noldorin" word for "blue", namely lhûn (LR:370 s.v. LUG2). As indicated above, long û in a final syllable becomes ui in the plural, so everything seems to fit: luin could be the plural form of lûn. What kills this seductively promising theory is the name of the mountain Mindolluin, "Towering Blue-head" (translated in the Silmarillion Index). Here, there is no reason for the adjective "blue" to be plural, so luin has to be the singular/basic form as well. There is also Luindirien "Blue Towers" in WJ:193; at the beginning of a compound, the word for "blue" would be expected to appear in its more or less basic form, not inflected for plural. It should also be noted that the same entry in the Etymologies that gives "Noldorin" lhûn (> Sindarin ?lûn) as the word for "blue", also gives lúne as the corresponding Quenya word. In Namárië in LotR, the adjective "blue" is luini instead (this is a plural form, from the phrase "blue vaults"; the singular is probably luinë). So while in the Etymologies the words for "blue" had been derived from a primitive form lugni (stem LUG2, LR:370) producing Quenya lúne and "Noldorin" lhûn, Tolkien must later have decided that the primitive form was something like *luini yielding Quenya luinë and Sindarin luin. Bottom line is that luin "blue" seems to cover both singular and plural, indicating that the diphthong ui undergoes no change in the plural. The fact that the adjective annui "western" is both sg. and pl. points in the same direction.

Special ai-plurals
As indicated above, it seems that the diphthong ai is normally unchanged in the plural. However, in one small group of words, ai becomes either i (usually long î) or more rarely ý in the plural. For instance, the plural form of the noun fair "mortal man" is given as fîr (WJ:387, where the sg. fair is quoted in archaic form feir). The plural forms in î (i) occur where ai in the singular forms ultimately arises from i or e being influenced by y later in the word. The example just quoted, fair or archaic feir, comes from an Old Sindarin form similar to the Quenya cognate firya (in late OS perhaps firia; see skhalia- in the wordlist appended to the Old Sindarin article). We must assume that other words sharing a similar phonological history would form their plurals in a similar way, though in most cases these plurals are not explicitly mentioned in Tolkien's published material. The nouns and adjectives in question are cai "fence" (pl. cî), cair "ship" (pl. cîr), fair "mortal man" (pl. fîr), gwain "new" (pl. gwîn), lhain "lean, thin, meager" (pl. lhîn), mail "dear" (pl. mîl) and paich "juice, syrup" (pl. pich, notice short i). The "Noldorin" word sein "new" pl. sîn (LR:385 s.v. SI) could become Sindarin sain pl. sîn, but it seems that Tolkien changed the Sindarin word for "new" to gwain pl. gwîn as just listed (notice that the same entry in the Etymologies that provides Noldorin sein also gives sinya as the corresponding Quenya word for "new", but in later sources, the Quenya adjective "new" is vinya - apparently the cognate of gwain).

NOTE: In "Noldorin", lhain pl. lhîn appeared as thlein pl. thlîn, the primitive (sg.) form being quoted as slinyâ (LR:386 s.v. SLIN). One revision separating "Noldorin" from Sindarin is that while primitive initial sl- became thl- in N, it becomes lh- in S. We alter the word in accordance with Tolkien's revised phonology. Thlein can be more directly adapted as lhein, but such a form would be archaic in Frodo's day, the current form being lhain instead. Similarly, paich "juice, syrup" actually appears as peich in the Etymologies (LR: 382 s.v. PIS); this "Noldorin" form is not conceptually obsolete, but can be seen as archaic Sindarin. This is also the case with ceir "ship" (LR:365 s.v. KIR); the form cair in LotR-style Sindarin is attested (cf. the footnote in LotR Appendix A explaining that Cair Andros means "Ship of Longfoam"; see also PM:371). - The word cair provides an example of another peculiar property of this group of words: when they occur as the first element in compounds, ai is reduced to í-, as in the name Círdan "Shipwright". However, ai remains unchanged if such a word is the final element of a compound; hence gwain "new" appears as -wain in the Sindarin name of the month of January, Narwain (evidently meaning "New Sun" or "New Fire"; compare Quenya Narvinyë).

In three words, where ai represents ei from even older öi (spelt "oei" by Tolkien), the plural forms should probably show the vowel y, ý, though we lack explicit confirmation in Tolkien's published papers. This theory is based on the fact that the first part of the archaic diphthong öi represents o or u in the original stem, and the umlaut product of these vowels is y, just as in cases where the older vowel-sound still survives in Sindarin (as in orch "Orc" pl. yrch). The words in question are 1) fair adj. "right" or noun "right hand" (pl. fýr, stem PHOR, cf. Quenya forya), 2) rain "slot, spoor, tract, footprint" (pl. rýn, stem RUN, cf. Quenya runya) and 3) the related word tellain "sole of foot" (pl. tellyn, since the final element -lain is actually assimilated from rain < runya, cf. the archaic form talrunya quoted in LR:390 s.v. TAL, TALAM). In the "Noldorin" of the Etymologies, these words appear as feir (the older form "foeir" = föir is also mentioned), rein (older röin) and tellein (older form tellöin not mentioned but clearly intended). Notice that while fair can mean both "right (hand)" and "mortal man", the different derivations make for distinct plurals: fýr in the former case and fîr in the latter.

Monosyllables later becoming polysyllables
(but perhaps still behaving as monosyllables for the purpose of plural formation)
This is something that is not directly addressed in Tolkien's published writings, but then almost nothing of his grammatical writings is available to us. However, our general understanding of the evolution of Grey-elven seems to strongly suggest that certain groups of nouns would behave in somewhat unexpected ways in the plural - though this is perfectly justified when the underlying phonological history is taken into account.

One important change that occurred in the evolution of Sindarin was that final vowels were lost. Hence an old word like ndakro "battle" later became ndakr. In early Sindarin, this word appeared as dagr. Another example is makla "sword" later appearing as makl, early Sindarin magl. We must assume that the plural of words like dagr, magl was formed after the same pattern as other monosyllables of comparable shape, like alph "swan", pl. eilph. So the plurals "battles" and "swords" would presumably be deigr, meigl (this would be before ei in a final syllable normally become ai).

What complicates matters is that words like dagr and magl were eventually changed. The final r, l came to constitute a separate syllable, so that for instance magl was pronounced mag-l just like English "eagle" is pronounced eeg-l. Later, these syllabic consonants turned into full-fledged normal syllables as a vowel o developed before them: Dagr (dag-r) turned into dagor and magl (mag-l) became magol. (Incidentally, the latter word was apparently often replaced by megil, which must be an adapted form of the Quenya word for "sword", namely macil.) The plurals deigr, meigl would presumably undergo the same process to become deigor, meigol (and the late change ei > ai in final syllables would never occur simply because ei was no longer in the final syllable). From a synchronic point of view, this results in what looks like irregularities: Normally, singular words like dagor and magol would be expected to have plural forms degyr, megyl, since o in the final syllable normally becomes y in the plural (e.g. amon "hill" vs. emyn "hills"). But in cases like dagor or magol, the o intruded relatively late and seems to be younger than the umlaut o > y; hence such newly developed o's would - presumably - remain untouched by the umlaut. If Tolkien did not imagine that analogical leveling bulldozed these "irregularities" out of existence, all two-syllable words where the second syllable contains a secondarily developed o must still be treated as monosyllables as far as plural formation is concerned. The o must be left alone and the vowel in the "second-to-last" syllable must be treated as if it were the vowel in the final syllable, which is precisely what it used to be.

The adjectives and nouns in question are: badhor "judge" (pl. beidhor if the theory holds - otherwise it would be analogical bedhyr), bragol "sudden, violent" (pl. breigol; this adjective also appears as bregol, pl. presumably brigol), dagor "battle" (pl. deigor), glamor "echo" (pl. gleimor), hador "thrower, hurler" (pl. heidor), hathol "axe" (pl. heithol), idhor "thoughtfulness" (unchanged in the pl.; luckily a noun with this meaning normally will not require a pl. form), ivor ?"crystal" (unchanged in the pl.), lagor "swift" (pl. leigor), maethor "warrior" (unchanged in the pl.), magol "sword" (pl. meigol), magor "swordsman" (pl. meigor), nadhor "pasture" (pl. neidhor), nagol "tooth" (pl. neigol), naugol "dwarf" (pl. noegol), tadol "double" (pl. teidol), tathor "willow" (pl. teithor), tavor "knocker, woodpecker" (pl. teivor), tegol "pen" (pl. tigol). Perhaps gollor "magician" also belongs on this list (pl. gyllor rather than ?gellyr).

NOTE: Some other peculiarities about this group of words may also be noted here. In (older?) compounds, the newly-developed o does not appear, and the final vowel that has otherwise disappeared, is sometimes preserved. Hence magol, that descends from primitive makla, may appear as magla- in a compound. LR:371 s.v. MAK lists Magladhûr for "Black Sword" (magol "sword" + dûr [lenited dhûr] "black, dark"). If one of these words is prefixed to an element beginning in a vowel, the original final vowel does not reappear, but the newly-developed o is not found: LR:398 s.v. TAM indicates that tavr (also spelt tafr) "woodpecker" retains that form in the compound Tavr-obel, Tavrobel *"Woodpecker-town" - though tavr became tavor as an independent word. Similarly, LR:361 s.v. ID indicates that the word "idher" (misreading for idhor?) "thoughtfulness" appears as idhr- in the name Idhril. - It is possible that in late Sindarin, analogy to some extent prevailed, this group of words being treated like any other. Before the collective plural ending -ath (see below), we would not expect to see the subsequently developed vowel o. For instance, we would expect the collective plural of dagr "battle" to be dagrath (not attested), unaffected by the fact that dagr had later become dagor when it occurred as a simplex (by itself). Yet in UT:395, 396 we find, not dagrath, but dagorath, though there can be little doubt that the latter is a historically unjustified form: R was not final or syllabic in dagrath, so no o would develop in front of it, and dagorath must be formed on analogy with the simplex dagor. This is all the more surprising when another attested form, the collective plural of nagol "tooth", is what we would expect: Naglath (WR:122). A form ?nagolath paralleling dagorath is not found. (The simplex nagol is not attested, but Tolkien undoubtedly imagined a primitive word *nakla "instrument for biting" = "tooth" [cf. the stem NAK "bite", LR:374], this *nakla becoming *nakl and then *nagl > *nagol in Sindarin.) There is also Eglath "The Forsaken" as the name of the Sindar, this collective plural reflecting the primitive (singular) form hekla or heklô (WJ:361; we don't know whether this also yielded an independent sg. form in Sindarin; if so it would be egol for earlier egl, the normal pl. being igl and later igol). A form ?Egolath nowhere occurs (and would be just as surprising as if the attested compound Eglamar "Land of the Forsaken Elves" suddenly were to appear as *Egolmar instead). Are we to assume, then, that Tolkien forgot his own rules when he (twice) wrote dagorath instead of dagrath in UT:395, 396? Rather we may imagine that there were several variants of Sindarin around. In a "purer" or more "classical" style, the collective plurals of words like dagor, nagol would perhaps be the historically correct forms dagrath, naglath, but in a more "colloquial" or "informal" style, forms like dagorath, nagolath may have come into use by analogy. We may speculate that in the form of Sindarin that preferred dagorath to dagrath, the historically justified plural deigor would also be altered to degyr, the umlauts following the more normal pattern. Interestingly, the name Dagorlad "Battle Plain" occurring in LotR gives away that dagor does not become ?dagro- as the first part of a compound, reflecting the earlier form ndakro (contrast examples quoted above: magol "sword" becoming magla- reflecting primitive makla in the compound Magladhûr, and tavor "wood-pecker" occurring in archaic form tavr in the compound Tavrobel). So again, analogy with the simplex form is at work. Perhaps Dagorlad would have been ?Dagrolad if the compound had been older, coined already in the really good old days when the Elves still said something like *Ndakro-lata (final vowel uncertain). Instead Dagorlad was clearly pieced together from dagor "battle" and -lad "plain" later. A late compound "Sword-Black" would presumably be, not Magladhûr, but simply Magoldhûr, and "Woodpecker-village" as a late compound could well be Tavorobel rather than the attested form Tavrobel.
Certain other cases of monosyllables turning into polysyllables involves, not a new vowel intruding before a consonant as in dagr > dagor, but a consonant turning into a vowel. Most of the examples involve older -w becoming -u. Before the stage where the final vowels were lost, some words ended in -wa (typically adjectives) or -we (typically abstracts). When the final vowels disappeared, only -w was left of these endings. For instance, the word for "craft" or "skill" that appears in Quenya as kurwe (curwë), which would also be the Old Sindarin form of the word, came out as curw in early Sindarin. We must assume that in the plural this would become cyrw, a perfectly regular form according to the rules set out above. But as indicated in LR:366 s.v. KUR, curw later became curu: Final -w following another consonant turned into a vowel -u, the semi-vowel becoming a full vowel. Presented with a noun like curu, it would be tempting to let it go like tulus "poplar-tree", pl. tylys - hence curu pl. cyry. In an older version of this article, I noted: "But the latter, if it occurred at all, would be an analogical form. The historically justified plural of curu can only be cyru, the older pl. cyrw turning into cyru just like the older sg. curw turned into curu." However, it now turns out that the analogical plural form cyry was indeed listed by Tolkien in the Etymologies (VT45:24), though it was omitted from the entry KUR as printed in LR.

The attested example cyry may indicate that Tolkien meant the analogical plural forms to have superseded the historically justified ones, at least in the class of nouns with final -u derived from earlier -w. Here are the words that are affected; we will indicate what both the historically justified plural and the analogical alternative would be: anu "a male" (historically justified plural form einu, but analogically eny), celu "spring, source" (hist. pl. cilu, analog. cily), coru adj. "cunning, wily" (hist. pl. cyru, analog. cery), curu "skill, cunning device, craft" (hist. pl. cyru, attested analogical pl. cyry), galu "good fortune" (hist. pl. geilu, analog. gely), gwanu "death, act of dying" (hist. pl. gweinu, analog. gweny), haru "wound" (hist. pl. heiru, analog. hery), hethu "foggy, obscure, vague" (hist. pl. hithu, analog. hethy), hithu "fog" (unchanged as a hist. pl., whereas the analogical pl. form would be distinct: hithy), inu "a female" (again the historically justified pl. would be unchanged, whereas the analogical pl. would be iny), malu "fallow, pale" (hist. pl. meilu, analog. mely), naru "red" (hist. pl. neiru, analog. nery), nedhu "bolster, cushion" (hist. pl. nidhu, analog. nedhy), pathu "level space, sward" (hist. pl. peithu, analog. pethy), talu "flat" (hist. pl. teilu, analog. tely), tinu "spark, small star" (the hist. pl. would be unchanged, the analogical pl. would be tiny). In the historically justified forms, we let words with the stem-vowel a have plural forms in ei rather than ai, again assuming that these words became disyllabic before ei turned into ai in final syllables (that is, when this change occurred, the syllable in which ei was found was no longer final because -w had already become -u, constituting a new final syllable). Hence anu : einu, gwanu : gweinu etc. However, if Tolkien had decided to go for the simpler analogical forms, these extra complications are transcended.

NOTE: In the Etymologies, the later stage where final -w became -u is often not explicitly recorded. There is curu besides older curw (entry KUR) and naru besides older narw (NAR1), but otherwise only the older forms where -w still persists are listed: Thus we find anw (3AN), celw (KEL), corw (KUR), galw (GALA), gwanw (WAN), harw (SKAR), hethw / hithw (KHITH), inw (INI), malw (SMAL), nedhw (NID), pathw (PATH) and tinw (TIN) instead of anu, celu, coru etc. as above. These later forms are not directly attested in Tolkien's papers. It may be that as far as the "Noldorin" of the Etymologies is concerned, Tolkien still had not decided once and for all that -w in this position did become -u; this idea just pops up in a couple of places. Yet we needn't hesitate to introduce the later forms in -u if we are aiming for the kind of Sindarin exemplified in LotR and the Silmarillion. Notice that in Etym, it is said that the "Noldorin" form of the Quenya name Elwë would have been *Elw, marked with an asterisk since it was not actually used in "Exilic" in this form (LR:398 s.v. WEG). However, in Chapter 4 of the published Silmarillion the scenario is another. "Noldorin" has now become Sindarin, and not only is there a Sindarin form of Elwë, but it is also Elu rather than "Elw" as in the Etymologies: "Elwë's folk who sought him found him not... In after days he became a king renowned... King Greymantle was he, Elu Thingol in the tongue of that land [Beleriand]." Here we are clearly to assume a development Elwë > Elw > Elu. It seems wholly justified, then, to alter (say) celw "spring, source" to its later form celu (to go with Elu), even though the form celu as such is not explicitly attested. A parallel case is provided by the name Finwë; again the Etymologies states that the "Noldorin" form would be *Finw, but that no such form was in use (LR:398 s.v. WEG). A much later, post-LotR source agrees that there was no Sindarin form of Finwë, but if this name "had been treated as a word of this form would have been, had it occurred anciently in Sindarin, it would have been [not Finw, but] Finu" (PM:344). If "Noldorin" Finw would have corresponded to Sindarin Finu, we can also conclude that "Noldorin" gwanw would correspond to Sindarin gwanu. - The word talu "flat" listed above actually appears as dalw (not **talw) in the Etymologies, but listed immediately after dalw is dalath "flat surface, plane, plain" (LR:353 s.v. DAL), occurring in the name Dalath Dirnen "Guarded Plain" (LR:394 s.v. TIR). However, Tolkien later changed dalath to talath; in the published Silmarillion, the "Guarded Plain" in Beleriand is called Talath Dirnen instead. In accordance with this revision, we also alter the related "Noldorin" word dalw "flat" to Sindarin talw > talu. We may still accept (dalw >) dalu - and for that matter dalath - as valid side-forms.

There are also a few cases of final -gh (spirant g) turning into a vowel. One example is provided by LR:381 s.v. PHÉLEG, where a word fela "cave" is derived from Old Sindarin (or "Old Noldorin") phelga. Since final vowels were lost following the Old Sindarin stage, fela is not a case of an original final -a surviving into later Sindarin. What Tolkien imagined seems to be this: Old Sindarin phelga naturally became phelg when the final vowels went. Then stops turned into spirants following the liquids l, r (UT:265), so that phelg became phelgh (or felgh, since the shift ph > f occurred at about the same stage). However, gh in no case survived into the Sindarin of Frodo's day; initially it was lost with no trace, but in this position it was vocalized: Felgh turned into fela. The plural of felgh had evidently been filgh formed according to the normal rules (cf. e.g. telch "stem", pl. tilch - LR:391 s.v. TÉLEK). The plural form filgh then became fili, the vocalization of earlier gh here being i rather than a (perhaps g > gh was somehow palatalized by the lost Old Sindarin plural ending -i that also caused the umlaut, biasing the subsequent vocalization towards i). It matters little precisely how we imagine the development: in any case, the end result is the peculiar pair fela pl. fili, for older felgh pl. filgh.

Fela pl. fili is the only known case of Tolkien explicitly mentioning both the singular and the plural of such a pair. There are, however, two or three other words that share a similar phonological development. The word thela "point (of spear)" derives from a stem STELEG (LR:388), and while Tolkien lists no primitive forms, we are probably to assume a Primitive Elvish form stelgâ (final vowel uncertain) turning into Old Sindarin sthelga and later (s)thelgh, the plural form of which would be (s)thilgh. The singular then yields the attested Sindarin form thela (wholly parallel to fela); the unattested plural "spear-points" must be thili (to go with the attested plural fili).

There are also a very few adjectives. An adjective thala "stalwart, steady, firm" is in LR:388 s.v. STÁLAG is derived from Old Sindarin/"Noldorin" sthalga. The unattested intermediate form would be (s)thalgh pl. (s)theilgh, following the normal pattern of (say) alph "swan", pl. eilph. We must assume that the plural form of thala is theili. A similar case would be tara "tough, still", stated to represent Old "Noldorin"/Sindarin targa (LR:390); again the unattested intermediate form would be targh. The plural form of this adjective could be teirgh, which would presumably produce Sindarin teiri. There is one other possibility: As already mentioned, it seems that ei was at one stage simplified to e before a consonant cluster beginning in r (hence we have nern rather than neirn > nairn as the plural form of narn "tale"). If this happened before the final gh of the plural adjective teirgh became a vowel so that the cluster disappeared, the form would turn into tergh, in later Sindarin teri. Presently we cannot say for sure whether teri or teiri is the best plural form of tara, since we do not know in what exact sequence Tolkien imagined the sound-shifts involved to have taken place; I would probably use teiri.

Expanded plurals
This is a group of words that seem to be longer in the plural than in the singular. Historically speaking it would be more accurate to turn the perspective around and speak of "reduced singulars", for in this case, the shape of the word that underlies the plural form gives a better impression of the primitive word than the current singular form does.

In WJ:363, êl is said to be an (archaic) Sindarin word for "star". According to the rules set out above, based on patterns like hên "child" pl. hîn (WJ:403), we would expect the plural form to be **îl. However, WJ:363 also informs us that the actual plural of êl is elin. Here it might seem that a plural ending -in is present. This, however, is not really the case. By comparing these words to their Quenya cognates elen pl. eleni one may begin to suspect what is really going on. Eleni would also be the plural form used in Old Sindarin, eventually yielding Sindarin elin: the plural ending being lost like all final vowels, but leaving its mark on the word by umlauting the second e to i. But one thing that occasionally happened in Old Sindarin was that consonants at the end of words might drop out. The n of the plural form eleni was "safe" because it was shielded by the plural ending following it, but the singular form elen was apparently reduced to ele, though this form is not explicitly mentioned by Tolkien. Later, final vowels were lost, leaving just el, and later still, the vowel of a monosyllable of this shape was lengthened, producing Sindarin êl. Hence we are left with the curious couple êl pl. elin in Third Age Sindarin. In the case of another, similar couple, nêl "tooth" pl. nelig, the Etymologies lists the Old "Noldorin"/Sindarin forms nele pl. neleki, confirming that the explanation sketched above is correct: By comparing the singular nele to the stem NÉL-EK (LR:376) we understand that the final consonant has dropped out. (In Common Eldarin, nele had evidently still been *nelek, which form directly underlies Quenya nelet listed in the same place - High-Elven phonology doesn't permit final -k, so it became -t instead.) Hence we have singular *nelek > nele > *nel > Sindarin nêl, but plural neleki (still used in Quenya) > umlauted *neliki > later *nelik with loss of final vowel > Sindarin nelig.

Other words that behave in a similar way:

# ael "pool, mere", pl. aelin (updated from "Noldorin" oel pl. oelin, LR:349 s.v. AY; we have Aelin-Uial for "Meres of Twilight" in the Silmarillion)
# âr "king", pl. erain (but the full singular aran seems to be more usual than shortened âr)
# bór (or better bôr) "steadfast, trusty man; faithful vassal", pl. beryn (LR:353 s.v. BOR, where the pl. occurs in "Noldorin" form berein, beren; we update it to its probable Sindarin form. Cf. the "Noldorin" plural geleidh "Noldor" corresponding to Sindarin gelydh. - The entry BOR indicates that the plural of bór later became býr, formed on analogy with the reduced singular; writers should probably use býr.)
# fêr "beech-tree", pl. ferin (LR:352 s.v. BERÉTH, cf. LR:381 s.v. PHER; the latter source indicates that this word for "beech-tree" was later replaced by brethil - which word would be unchanged in the pl.)
# ôr "mountain", pl. eryd or irregular ered (but as in the case of âr above, the full singular orod is apparently more common than reduced ôr; LR:379 s.v. ÓROT lists two "Old Noldorin" singulars, full oroto or reduced oro; in the later language these would come out as orod and ôr, respectively, but actually the only singular listed is orod - descended from unreduced oroto.)
# tôr "brother", pl. teryn (LR:394 s.v. TOR; we update the plural form from "Noldorin" terein. However, the same entry in the Etymologies indicates that this word for "brother" was normally replaced by muindor pl. muindyr, or - when "brother" is used in the wider sense of "male associate" - gwador, the "Noldorin" plural of which was gwedeir; read gwedyr in Sindarin.)
# thôr "eagle", pl. theryn (LR:392 s.v. THOR; again we update the plural from "Noldorin" therein. - This entry in the Etymologies indicates that the unreduced singular thoron was also in use)

In addition to the above, there are a few words that belong to the same category even though the plural forms have no final consonant; pêl "fenced field" pl. peli, ôl "dream" pl. ely and thêl "sister" pl. theli. What has happened is simply that an original final consonant h, lenited from s at the Old Sindarin stage, has dropped out in the plural forms: The relevant stems are given as PEL(ES), ÓLOS and THELES in the Etymologies. In the first of these entries, pêl "fenced field" is demonstrated to come from pele (LR:380), which given the stem-form PEL(ES) is understood to be a reduced from of *peles (cf. the Quenya cognate peler, clearly meant to come from *pelez < *peles). The plural of the old form pele is given as pelesi, and it is further stated that this became pelehi ("peleki" in LR:380 is a transparent misreading of Tolkien's manuscript; for s becoming h like this, cf. barasa > baraha in LR:351 s.v. BARÁS). Just as in one case referred to above, neleki becoming nelig, the plural pelehi became *pelih - but in this case the now final consonant was so weak that it was lost to produce the plural form peli, creating the false impression that Sindarin occasionally employs a plural ending similar to Quenya -i.

NOTE: Several of forms quoted above are somewhat regularized. Pêl "fenced field" actually appears as pel in LR:380 s.v. PEL(ES); according to the phonology we can reconstruct from many other examples, the vowel definitely ought to be long. The omission of the circumflex in the form pel must be a mere mistake, whether Tolkien himself or the transcriber is to be blamed (perhaps the singular was confused with the plural peli, in which form the e should be short). - The plural form of ôl "dream" is given as elei in LR:379 s.v. ÓLOS; in Sindarin we should evidently read ely, as suggested above. This is a case wholly parallel to "Noldorin" geleidh corresponding to Sindarin gelydh as the word for Noldor (sg. golodh): In both cases "Noldorin" ei derived from o in the singular corresponds to Sindarin y (cf. also the corrected/updated plurals suggested above: Sindarin beryn, teryn, theryn where the "Noldorin" of the Etymologies actually has berein, terein, therein). - One other form is also regularized: In the Etymologies, the plural of thêl is not theli as suggested above, but thelei (LR:392 s.v. THEL, THELES). Why a word thêl derived from a stem THELES should behave any differently in the plural than a word pêl derived from PELES is difficult to understand, so if the plural is peli in the latter case, we may feel free to emend the plural of thêl from thelei to theli. The plurals theli and attested peli fit the general system better: The plurals represent the full stems THELES and PELES, except for the detail that the final -s was later lost (after becoming -h), and as usual, e in a final syllable becomes i in the plural (as in Edhel "Elf" pl. Edhil, WJ:377). Hence the pl. of *peles ought to be *pelis, and removing the lost final consonant we arrive at the attested plural peli; in light of this, the pl. of *theles ought to be *thelis > theli rather than "thelei". If we were to keep the plural thelei (in which case we would have to alter peli to pelei for the sake of consistency), we must take into account Tolkien's post-Etym discovery that ei in a final syllable eventually became ai, which would land us on thelai, pelai as the rather outlandish plurals of thêl, pêl in late Third Age Sindarin. So all things considered, it seems better to regularize thelei to theli in accordance with the attested example peli rather than going the other way. (In the case of thelei/theli "sisters" writers can happily avoid the problem; LR:392 s.v. THEL indicates that the more normal word for "sister" was muinthel pl. muinthil, or - where "sister" is used in the wider sense of "female associate" - gwathel pl. gwethil.) - Another plural in -ei is "Noldorin" tele "end, rear, hindmost part", pl. telei (LR:392 s.v. TELES). As far as the singular is concerned, the development differs somewhat from that which produced thêl from the stem THELES; notice that in tele, the last vowel of TELES is still in place (it has not become **têl to parallel thêl). The primitive form of tele is given as télesâ (the accent marks stress only). In "Old Noldorin", this would have become telesa > teleha (not explicitly given in Etym but compare primitive barasâ "hot, burning" producing "ON" barasa > baraha, LR:351 s.v. BARÁS). Later the final vowels were lost, hence teleha > teleh, but eventually the weak final consonant -h also dropped out, leaving tele only (and the new final vowel was not lost; the stage where such loss occurred had already passed). But what about the plural form telei? It is difficult to tell precisely what kind of development Tolkien envisioned. The "Old Noldorin" plural of teleha is not mentioned but should have been telehi (cf. for instance poto "animal's foot", pl. poti, LR:384 s.v. POTÔ). Later, we would expect the final i to umlaut the e in the second-to-last syllable, telehi becoming telihi; then final vowels and later final h are lost, which ought to leave us with teli as the plural form. So how did Tolkien come up with telei instead? Are we to assume that at the telehi-stage, h dropped out so that the vowels e and i came into direct contact and formed a diphthong telei? But this would be inconsistent with the example referred to above: the plural form pelehi becoming peli instead of **pelei. It seems that when updating "Noldorin" tele pl. telei to Sindarin, it is best to read tele pl. teli. Again, the plural form telei cannot be kept as it is in any case, since in Sindarin ei in a final syllable becomes ai.

Plurals in -in
There are a few words that seem to display a genuine plural ending -in, though the origin of this ending would be unclear; conceivably Tolkien imagined it to be invented on the analogy of such examples as êl pl. elin, where (as demonstrated above) no genuine ending is present.

What may be the best example involves a loan-word, Drû "Wose", the name of one of the Drúedain or "Wild Men"; the Sindarin term was based on their native word Drughu. According to UT:385, one Sindarin plural of Drû was Drúin. Perhaps this extraordinary plural somehow marks the word as a loan; it is not inflected according to the normal pattern (that would have landed us on **Drui as the plural form).

On the fields of Cormallen (LotR3/VI ch. 4), the Ring-bearers were hailed as Conin en Annûn, and according to Letters:308, this means "Princes of the West". Assuming that Conin "princes" contains the plural ending -in, it could be the plural form of ?caun (since by adding -in, constituting a new syllable, au becomes o in the polysyllabic environment thereby arising). This ?caun could in turn be a Sindarized form of Quenya cáno "commander" (PM:345), which would again be a loan-word rather than a "native" Sindarin word (PM:362 mentions a quite distinct inherited word caun, meaning outcry or clamour). If conin "princes" is not the plural of *caun, it could be the plural of an otherwise unknown word *conen, but this looks like an adjective rather than a noun.

The name Dor-Lómin occurring in the Silmarillion is interpreted "Land of Echoes" in LR:406. The Silmarillion Appendix lists a word lóm "echo", though nothing is said about what language this is supposed to be. Is lómin the plural form of lóm? We must carefully distinguish various stages in Tolkien's conception. The Etymologies lists a word lóm "echo" (LR:367 s.v. LAM), but this is Doriathrin, not "Noldorin" > Sindarin. In Doriathrin (one dialect of the Ilkorin language whose place in the mythos would later be usurped by Sindarin), there is indeed a plural ending -in, so lómin could be Doriathrin for "echoes". Yet in the entry in the Etymologies just referred to, the name obviously corresponding to Dor-Lómin in the Silmarillion appears as Dorlómen instead. Dorlómen is said to be, not Doriathrin, but a "Noldorinized" form of the true Doriathrin name Lómendor. The first element is not a plural form at all, but a Doriathrin adjective lómen "echoing". This may provide a clue to how Tolkien would later have interpreted the name. When he had made Sindarin the language of Beleriand, dropping "Ilkorin", he still made references to the peculiar North Sindarin dialect, and the name Dor-Lómin seems to fit what little is known about it (m is not opened to mh > v following a vowel; cf. the North Sindarin name of Oromë being Arum rather than Araw [for *Arauv] as in standard Sindarin: WJ:400). One educated guess may be that in the post-LotR period, Tolkien interpreted Dor-Lómin as meaning literally "Echoing Land", lómin being the North Sindarin adjective descending from older *lâmina. In standard Sindarin, the adjectival ending would be -en in the singular and -in only in the plural, but this may not be true of this dialectal form of the language. If lómin is really an adjective, it is of course irrelevant for a discussion of Sindarin plural formation.

Singulars derived from plurals
In the vast majority of cases, the singular must be considered the basic form of the noun, from which the plural is derived. However, there are a few cases where it is actually the plural that is the basic form, and the singular is derived from it. Historically, fileg "small bird", pl. filig, is such a case. The stem PHILIK (LR:381) came out as filig in Sindarin, but since so many plural forms have i representing singular e in the final syllable (e.g. Edhil as the pl. of Edhel "Elf"), the word filig was taken as such a plural form and a singular was made according to the normal pattern: Fileg. Since the stem was PHILIK, such a singular was wholly unjustified historically; it is, as Tolkien noted in the Etymologies, an "analogical singular" only. The pair fileg pl. filig, being fully adapted to the normal patterns, of course presents no extra problem for people studying Sindarin synchronically. But the Etymologies indicates that the singular could also be filigod, where the ending -od is in effect a "singular ending", producing the most peculiar pair filigod pl. filig. Another, similar case, involving another "singular ending", is lhewig "ear", pl. lhaw. (Cf. the hill Amon Lhaw in LotR, "Hill of Hearing" or literally *"Hill of Ears", mentioned near the end of the chapter The Great River in Volume 1.) The plural lhaw is explained to represent an old dual form denoting a pair of ears, or as Tolkien wrote, "ears (of one person)" (LR:368 s.v. LAS2). The singular lhewig "ear" is in turn derived from this plural or dual form. A similar "singular-from-dual" formation in -ig is gwanunig "twin", derived from gwanûn "pair of twins" (WJ:367).

NOTE: The endings -od, -ig, -og used to form singulars from plurals can also be used to form so-called nomina unitatis, words denoting one distinct part of something larger, or words denoting a single entity within a collective. Indeed this is probably their proper function. WJ:391 provides a good example. There was a Sindarin word glam "din, uproar, the confused yelling and bellowing of beasts". Since bands of Orcs could be very noisy, the word glam "alone could be used of any body of Orcs, and a singular form was made from it, glamog". Hence we have glamog as a word for "Orc", an individual member of a glam or body of Orcs as a collective. In such a case one cannot well say that glam is really the plural form of glamog (it would be like asserting that "troop" is the plural form of "trooper"); perhaps glamog could itself be the basis of a plural form ?glemyg. Another, similar case is the word linnod, nowhere explicitly explained but used in LotR Appendix A: "[Gilraen] answered only with this linnod: Onen i-Estel Edain, ú-chebin estel anim [I gave Hope to the Dúnedain, I have kept no hope for myself]." So what, really, is a linnod? Knowing that -od is an ending used to form nomina unitatis, as in filigod from filig above, linnod can be recognized as such a formation, transparently based of lind "song" (*lindod naturally becoming linnod since Sindarin phonology does not permit intervocalic -nd- in unitary words; this group can only occur in compounds, such as Gondor "Stoneland"). So a linnod is some kind of unit within a song, and the example provided indicates that it means a verse, a single line in a song. Again it makes little sense to say that linnod is the "singular" form of lind (as if this word for "song" must be considered a plural just because a song is made up of verses). Rather we must see linnod as a derived noun, an independent word for "verse" that can probably have its own plural linnyd "verses". (In the case of Gilraen's linnod it seems clear that her particular "verse" was not part of a longer song; it was just a verse or very short poem in its own right.) Nouns in -ig seem to denote specifically one out of a pair, as in the examples quoted above: gwanunig "a twin" from gwanûn "pair of twins", or lhewig "an ear" besides lhaw "pair of ears". Again one may discuss whether gwanûn, lhaw are really the "plural" forms of gwanunig, lhewig; the latter forms simply denote one out of a couple.

The first element of compounds
One example quoted above, Edenedair "Fathers of Men" or literally *"Man-fathers" (MR:373) is transparently the plural of a compound Adanadar "Man-father" (adan + adar). Here we see the umlaut carried through the whole word, all the a's in non-final syllables becoming e's, just as if this were a unitary word. Yet it would probably have been permissible to use the plural ?Adanedair as well, leaving the first element of the compound unaffected and umlauting just adar "father" (to edair). In WJ:376, Tolkien makes a note about the plurals of orodben "mountaineer" and rochben "rider" (actually compounds orod-ben "mountain-person" and roch-ben "horse-person"). The i-affection occurring in the plural was originally carried through the whole word, resulting in the forms örödbin and röchbin (spelt "oeroedbin" and "roechbin" in WJ:376; this would have become eredbin and rechbin in the Sindarin of Frodo's day, though Tolkien does not mention these later forms). However, Tolkien further noted that "the normal [sc. un-umlauted] form of the first element was often restored when the nature of the composition remained evident"; therefore the plural of rochben could also be rochbin, the umlaut only affecting the vowel of the final element -ben "person", while roch "horse" is unchanged. (The implication is that the plural of orodben "mountaineer" could similarly be orodbin with orod "mountain" in its normal form, though the form orodbin is not mentioned in WJ:376.) In the compound Edenedair the first element has not been restored, but as already mentioned, a form ?Adanedair would probably have been equally permissible.

THE CLASS PLURAL
Beside the normal plural, Sindarin also has a so-called Class plural, or a collective plural. In RGEO:74, Tolkien states that "the suffix -ath (originally a collective noun-suffix) was used as a group plural, embracing all things of the same name, or those associated in some special arrangement or organization. So elenath (as plural of êl, [irregular] pl. elin) meant 'the host of the stars': sc. (all) the (visible) stars of the firmament. Cf. ennorath, the group of central lands, making up Middle-earth. Note also Argonath, 'the pair of royal stones,' at the entrance to Gondor; Periannath, "the Hobbits (as a race)," as collective pl. of perian, 'halfling' (pl. periain)." The King's Letter provides more examples: sellath dîn "his daughters" and ionnath dîn "his sons", referring to all of Sam's sons and daughters as groups. In some cases, -ath seems to have a longer form -iath. WJ:387 gives firiath as the class plural of feir "a mortal" (normal plural fîr); cf. also the "collective pl." form giliath "stars" in LR:358 s.v. GIL (as in Osgiliath, "Citadel of the Stars"). In earlier versions of this article, we explained this i intruding before -ath is a remnant of an earlier y that is here preserved (earlier firya "mortal", gilya "star"). This may be correct in the case of the words firiath and giliath, but it may seem that the longer ending -iath appears whenever the Class Plural ending is to be added to a word that has the stem vowel i: This vowel is echoed in the ending.
If the ending -ath is added to a noun ending in -nc or -m, they would for phonological reasons change to -ng- and double -mm-, respectively, whereas final -nt and -nd would both become -nn-: The class plurals of words like ranc "arm", lam "tongue", cant "shape" and thond "root" would evidently be rangath, lammath, cannath, thonnath, respectively. Also remember that since the sound [v] is spelt f only finally, it would be spelt as it is pronounced - simply v - if any ending is appended. Hence the class-plural of a word like ylf "drinking-vessel" must be written ylvath.
In some cases, other endings than -ath seem to be used, such as -rim "people"; in WJ:388, Nogothrim is said to be the class plural of Nogoth "Dwarf". Yet another ending is -hoth "folk, host, horde", cf. Dornhoth "the Thrawn Folk", another Elvish term for Dwarves. The Silmarillion Appendix (entry hoth) states that this ending is "nearly always used in a bad sense" and mentions the example Glamhoth "Din-horde", an Elvish kenning of Orcs. The one who first called the Snowmen of Forochel Lossoth (for *Loss-hoth, loss = "snow") evidently did not like them. In Letters:178, Tolkien explains that while the normal plural of orch "Orc" is yrch, "the Orcs, as a race, or the whole of a group previously mentioned would have been orchoth" (for *orch-hoth, evidently). It could be discussed whether forms like Nogothrim and Lossoth are really "plural" forms or simply compounds: Dwarf-folk, Snow-horde. Words with the "collective" ending -ath are seen to take the plural article in, so they are evidently considered plurals. Words in -rim and -hoth seem to behave in the same way; cf. the name Tol-in-Gaurhoth "Isle (of) the Werewolves"(Silmarillion ch. 18, where the name is translated simply "Isle of Werewolves"). In Letters:178, Tolkien does state that "the general plurals [italics mine] were very frequently made by adding to a name (or a place-name) some word meaning 'tribe, host, horde, people' " - namely the endings we have been discussing here. So it would seem that from a grammatical point of view, the forms employing these endings really are to be considered plurals, not compounds.
THE UNINFLECTED CASES
As far as we can tell from what has been published, the Sindarin noun is not inflected for a great number of cases, as in Quenya. Their common ancestral tongue of Quenya and Sindarin was apparently a case language, but in Sindarin the relevant endings have been lost (though traces of them may be found in some words - for instance, ennas "there" must once have ended in a locative ending similar to Quenya -ssë). Grey-elven depends on prepositions instead of case endings. It is noteworthy, though, that Sindarin nouns can be used as genitives without changing their form. We have already quoted the Moria Gate inscription as an example of this: Ennyn Durin Aran Moria, "Doors of Durin, King of Moria", the names Durin and Moria functioning as uninflected genitives: of Durin (or Durin's), of Moria (or Moria's). To say "X of Y" or "Y's X" you simply juxtapose the words: X Y. The King's Letter provides more examples: Aran Gondor "King (of) Gondor", Hîr i Mbair Annui "Lord (of) the Western Lands", Condir i Drann "Mayor (of) the Shire". Tolkien noted that these uninflected genitives probably descended from "inflexional forms" (WJ:370). At an earlier stage, Sindarin probably had the same genitive ending -o as in Quenya, but it was lost together with the other final vowels. (Doriathrin Sindarin sometimes shows a genitive ending -a, as in Túrin's epithet Dagnir Glaurunga "Glaurung's Bane"; cf. also Bar Bëora for "the House of Bëor" in WJ:230. The origin of this ending is very unclear, and it is apparently not used in standard Sindarin.)
Sometimes one or both of the nouns in a genitive phrase is somewhat shortened: Double consonants may be simplified; compare toll "isle" with tol in a name like Tol Morwen "Morwen's Isle" (WJ:296). Long vowels may be shortened; compare dôr "land" with dor in Dor Caranthir "Caranthir's Land" (WJ:183). But such shortening is not necessary to produce correct Sindarin; cf. Hîr rather than Hir in the phrase Hîr i Mbair Annui "Lord (of) the Western Lands" in the King's Letter.

Not only the genitive, but also the dative can be expressed by a Sindarin noun that does not in any way change its form. This is evident from the first part of Gilraen's linnod in LotR Appendix A: Onen i-Estel Edain, "I gave Hope to the [Dún]edain". The indirect object, or dative object, is clearly Edain - but it shows no inflectional ending, nor is there anything corresponding to the preposition "to" in Tolkien's English translation. The dative is apparently expressed by word order alone. This construction may be compared to English "I gave the Edain Hope", again with no preposition or inflectional ending - but while English in such a case inserts the indirect object before the direct object, Sindarin has the indirect object following the direct object.

- - -

The Sindarin noun, as well as other parts of speech, is often subjected to certain regular changes of the initial consonants. To these we must now turn our attention.

3. THE CONSONANT MUTATIONS

In Sindarin, the initial consonant of words often undergo certain changes, so that the same word may appear in different shapes (words beginning in a vowel are unaffected). These changes are termed mutations, with a series of subcategories (soft mutation, nasal mutation etc.) Consider two completely distinct words like saew "poison" and haew "habit". One mutation rule dictates that s in certain grammatical contexts becomes h. The article i "the" is one of the triggers of this mutation, so if we prefix it to saew to express "the poison", the result is not **i saew. "The poison" must be i haew instead. Though haew also means "habit", a competent user of Sindarin would not misunderstand i haew (thinking it means "the habit" instead of "the poison"). For in the same position where s becomes h, the mutation rule also dictates that h becomes ch. So if we combine haew "habit" with the article i, we would get i chaew for "the habit", the words still being distinct. However, it is obvious that there is here considerable room for confusion if one does not understand the Sindarin mutation system. It is all too easy to imagine some naive student seeing the combination i haew in a text and then looking up haew instead of saew in his wordlist - wrongly concluding that i haew means "the habit" instead of "the poison", since it does not occur to him that haew is merely the form the word saew takes in this particular position. It is quite impossible to use a Sindarin wordlist properly unless one understands the mutation system; in some cases the wordlist would be downright misleading.
We will attempt to describe the various mutations, as well as they can be reconstructed. The actual evidence being scanty, we must in many cases fall back on our general understanding of Sindarin phonology to fill the gaps. What follows is based on a thorough analysis (mainly conducted by eminent Sindarist David Salo), but future publications may well prove it wrong in some respects. However, the most frequent mutations (soft and nasal) are relatively well attested, so that we can reconstruct the rules with some confidence.
I. SOFT MUTATION
The most frequent mutation, it is also known as lenition (= "softening"). The name reflects the fact that by this mutation, "hard" or unvoiced sounds like p or t become "softened" (or lenited) to voiced b, d, while original b, d are further "softened" to spirants: v, dh. We will describe the effects of the soft mutation before discussing in detail where it occurs, but it may be noted that lenition typically occurs after particles ending in a vowel when such a particle immediately precedes a word and is closely associated with it, such as the definite article i (singular "the"). In Letters:279, Tolkien comments upon the lenition c > g and notes that it is used "after closely connected particles (like the article)". The phonological background for this phenomenon is not very difficult to understand. In the evolution of Sindarin, many consonants changed following a vowel; for instance, c became g and t became d (compare Sindarin adar "father" with the primitive word atar, still preserved in Quenya). What happened was that particles like prepositions and articles immediately preceding a word became so closely associated with the word itself that the whole phrase of particle + main word was perceived as a kind of unity. Hence a word like tâl "foot", when occurring in a phrase like i tâl "the foot", was subjected to the same rule that turned a unified word like atar into adar: There is a vowel preceding the t, so it has to turn into d - and while tâl remained as the word for "foot", "the foot" is henceforth i dâl instead (see LR:298 concerning this example). See below concerning the various uses of the soft mutation; while describing the mutations themselves, we will use the changes occurring after the definite article i as examples.

The soft mutation turns the plosives p, t, c into voiced b, d, g; original b, d become v, dh, while g disappears altogether. (It should be noted that the mutations here described for b, d, g only apply when these sounds are derived from primitive b, d, g. Sindarin initial b, d, g may also derive from mb, nd, ñg, and in such cases, the lenited forms differ. See the section "The development of nasalized stops" below.)

pân "plank" > i bân "the plank"
caw "top" > i gaw "the top"
tâl "foot" > i dâl "the foot"
bess "woman" > i vess "the woman"
daw "gloom" > i dhaw "the gloom"
gaw "void" > i 'aw "the void"

Note: G originally turned into the back spirant gh, but this sound later disappeared (i ghaw becoming i 'aw). To indicate that a g has been lenited to zero, one may use an apostrophe ' as in this example, but Tolkien's writings are inconsistent on this point. In UT:390 we have Curunír 'Lân for "Saruman the White", the apostrophe evidently indicating that the second word (the adjective "white") is glân when not mutated. Cf. also galadh "tree" > i 'aladh "the tree" in LR:298 (there spelt galað, i·'alað). But in the Silmarillion we have names like Ered Wethrin "shadowy mountains", wethrin being a lenited form of gwethrin, the plural form of the adjective gwathren "shadowy" (compare gwath "shadow", LR:396 s.v. WATH). Perhaps a spelling equivalent of Ered 'Wethrin would actually be used in Tengwar writing, Tolkien sometimes dropping the apostrophe in names occurring in his narratives.

These consonants evidently undergo the same mutations if they form part of clusters:

blabed "flapping" > i vlabed "the flapping"
brôg "bear" > i vrôg "the bear"
claur "splendor" > i glaur "the splendor"
crist "cleaver" (sword) > i grist "the cleaver"
dring "hammer" > i dhring "the hammer"
gloss "snow" > i 'loss "the snow"
grond "club" > i 'rond "the club"
gwath "shadow" > i 'wath "the shadow"
prestanneth "affection" (disturbance) > i brestanneth "the affection"
trenarn "tale" > i drenarn "the tale"

The consonants h, s and m are lenited to ch, h and v, respectively:

hammad "clothing" > i chammad "the clothing"
salph "soup" > i halph "the soup"
mellon "friend" > i vellon "the friend" (also spelt i mhellon)

It will be noticed that b and m both become v when lenited. In a few cases, ambiguity may arise. Consider two adjectives like bell "strong" and mell "dear"; only context can decide whether i vess vell means "the strong woman" or "the dear woman". (In Sindarin, an adjective normally follows the noun it describes, and in this position, the adjective is lenited.) The mutation product of m is sometimes spelt mh instead (as in the King's Letter, SD:128-9: e aníra ennas suilannad mhellyn în, "he wishes there to great his friends"). It seems that in Third Age Sindarin, this mh was no longer pronounced any differently from v, though the distinction may have been upheld in Tengwar writing. Earlier, mh was evidently a distinctly nasal variant of v, that may also be termed "spirant m". Compare LotR Appendix E, in the discussion of the Runes: "For (archaic) Sindarin a sign for a spirant m (or nasal v) was required."

The sound hw (unvoiced w, like English wh in dialects where it is still kept distinct from w) probably becomes chw in mutation position:

hwest "breeze" > i chwest "the breeze"

(In the "Noldorin" of the Etymologies, this sound is chw in all positions, also where the word is not lenited, but it seems that Tolkien revised this.)

The unvoiced spirants f, th, the nasal n and the liquids r, l are unaffected by the soft mutation:

fend "threshold" > i fend "the threshold"
thond "root" > i thond "the root"
nath "web" > i nath "the web"
rem "net" > i rem "the net"
lam "tongue" > i lam "the tongue"

The behavior of the unvoiced liquids rh, lh in mutation position is somewhat uncertain. The view presented in earlier versions of this article was that they turn into normal voiced r, l. This was based primarily on the example rhass "precipice", with article i rass (LR:363 s.v. KHARÁS). However, this is probably "Noldorin" rather than Sindarin. One of the revisions Tolkien did when he turned "Noldorin" into Sindarin affected the sounds rh, lh. In "Noldorin", they were descended from normal r, l in the primitive language, where these sounds occurred initially. However, Tolkien later decided that primitive initial r, l were unchanged in Sindarin, a primitive word like lambâ "tongue" yielding Sindarin lam (WJ:394; contrast earlier "Noldorin", where this word had been lham instead: LR:367 s.v. LAB). The sounds rh, lh still occur initially in Sindarin, but in this language they are derived from primitive initial sr-, sl- (e.g. srawê > Sindarin rhaw, MR:350), not simple r-, l-. This new derivation must be taken into consideration when we make our educated guess about how Sindarin rh, lh behave in mutation position. Basically, the soft mutation corresponds to how certain consonants develop following vowels. Medial primitive sr, sl became thr, thl, e.g. "Noldorin" lhathron "listener, eavesdropper" (Sindarin lathron?) from primitive la(n)sro-ndo (LR:368 s.v. LAS2). So perhaps this is also what the soft mutation of rh-, lh- would produce, though we lack examples:

rhaw "flesh" > i thraw "the flesh" (primitive *i srawê)
lhûg "dragon" > i thlûg "the dragon" (primitive *i slôkê)

The uses of the soft mutation: The soft mutation has a variety of uses. It occurs after a series of particles, prepositions and prefixes, the example we have used so far - the definite article i - being only one of these particles. Typically, we are talking about particles that either end in a vowel or did end in a vowel at an earlier stage. A preposition like na "to" triggers the same mutations as the article i, for instance na venn "to a man" (unmutated benn). In the hymn to Elbereth (A Elbereth Gilthoniel) we have the phrase na-chaered "to-remote distance" (see RGEO:72 for translation), haered "remote distance, the remote" undergoing soft mutation to become chaered. (For haered as the unmutated form, compare the name Haerast "Far Shore" mentioned in the Silmarillion Index; see the entry Nevrast.)

We know or deduce that soft mutation occurs after the following particles and prefixes:

- the prefix and preposition (?) ab "after, behind, following, later" (since this was earlier apa, as in Quenya)
- the preposition adel "behind, in the rear (of)" (since this was probably *atele in Old Sindarin)
- the preposition and prefix am "up, above, over" (cf. Quenya amba); the soft mutation is attested in compounds like ambenn "uphill" (am + a lenited form of pend, penn "declivity")
- the prefix ath- "on both sides, across" (older *attha)
- the prefix athra- "across" (cf. a word like athrabeth, "debate", the second element being a lenited form of peth "word")
- the preposition be "according to" (perhaps also "as, like", since it must correspond to Quenya ve)
- the adverb/prefix dad "down" (cf. dadbenn "downslope", which is dad + a lenited form of pend, penn "declivity")
- the preposition di "under, beneath"
- the prefix go-, gwa- "together" (possibly also used as an independent preposition "with")
- the preposition na "to, towards; at; of; with, by"
- the preposition nu (no) "under"
- the preposition trî "through" and the corresponding prefix tre-
- the negative element ú-, u- "not" or "without", used as a prefix, e.g. ú-chebin *"I do not keep" in Gilraen's linnod (compare unmutated hebin "I keep"). Cf. also such a word as ubed "denial" (u + ped, the latter being the stem of the verb "say", hence ubed = "no-saying").

The sentence guren bêd enni "my heart tells me" (VT41:11) incorporates a lenited form of the verb pêd "tells". This example seems to indicate that a verb immediately following its subject is lenited. This is not the case if the verb comes before the subject, as in the sentence tôl acharn "vengeance comes" or literally *"comes vengeance" (WJ:254; notice that tôl is not here lenited to dôl). Some are skeptical of the rule that a verb is lenited even where it does immediately follow its subject. We are told that in one version of the so-called Turin wrapper, the wording Rían pent *"Rían said" occurs; here the verb pent "said" is not lenited (to bent), even though it does immediately follow its subject. Tolkien surely experimented with different systems over the years, or there may be something special about the phrase guren bêd enni that causes pêd to appear in lenited form bêd here. At least it seems certain that a verb is not lenited where it does not immediately follow its subject, as is evident from the Moria Gate inscription: Celebrimbor o Eregion teithant [not: deithant] i thiw hin "Celebrimbor of Hollin drew these signs". Perhaps it makes some difference that the phrase o Eregion "of Hollin" here intrudes between the subject and the verb, perhaps not. It would be interesting to know whether "Celebrimbor drew" would translate as Celebrimbor deithant or Celebrimbor teithant - or maybe both are possible.

In Sindarin, adjectives (including participles) following the noun they describe are usually lenited. In Sindarin, an adjective normally does follow the noun it describes; you say "isle green", Tol Galen, instead of "green isle". Galen is here the lenited form of calen "green". Another example of the same is the name Pinnath Gelin "Green Ridges" or literally "Ridges Green", gelin being a lenited form of celin, in turn the plural form of calen (plural to agree with "ridges"). The name Talath Dirnen "Guarded Plain" ("Plain Guarded") contains a lenited form of the past participle tirnen "watched, guarded" (cf. the verb tir- "watch, guard"). Eryn "wood" + morn "dark" produces Eryn Vorn "Dark Wood" (UT:436, 262). Dor Dhínen "Silent Land" ("Land Silent") includes the lenited form of dínen "silent" (WJ:333, 338). There are, however, quite a few attested cases where soft mutation fails to take place in such combination. The name Dor Dhínen just mentioned also appears as Dor Dínen in a number of texts (so in the published Silmarillion). From LotR we also remember the Rath Dínen or "Silent Street" in Minas Tirith; we might have expected *Rath Dhínen instead. (However, the form Barad-dûr instead of *Barad-dhûr for "Tower-Dark" may be explained by the fact that the words are here practically a compound, as indicated by the dash - though the second element of compounds are often lenited as well, see below.) Cases of d where we would expect dh may in some instances be explained (away) as inaccurate transcription on Tolkien's part, since he sometimes substituted d for dh simply because he found the latter digraph "uncouth" (UT:267). However, we cannot easily explain cases like Cú Beleg rather than *Cú Veleg for "Great Bow" (beleg "great"; for "great bow" cf. the song Laer Cú Beleg or "Song of the Great Bow" mentioned in the Silmarillion, chapter 21). Another example is the name Nan Tathren, "Vale of Willows" or literally "Vale Willowy"; we might have expected *Nan Dathren instead. We probably have to assume that the discrepancies are simply due to the fact that there were many variants or dialects of Sindarin; the rules for where soft mutation occurs differed somewhat from dialect to dialect. (I would advise people writing in Sindarin to let adjectives lenit in this position, though, since this seems to be the main rule.)
When a word is used as the second element of a compound, it often undergoes changes similar to the effects of the soft mutation. Tolkien stated (in Letters:279) that "the initials of words in composition" are lenited (he used the example Gil-galad, that represents *Gil-calad "Starlight"; cf. unlenited calad "light" in UT:65 - another explanation of the element galad is given in PM:347, though). In RGEO:73, Tolkien mentions the "the S[indarin] change of medial t > d": in the hymn to Elbereth we have palan-díriel for *palan-tíriel "far-seeing" (compare the verb tir- "watch, see, guard").
Other examples include compounds like Calenhad "Green Space" (calen "green" + sad "place, spot", UT:425), Elvellyn "Elf-friends" (El = reduced form of the word for "Elf" + mellyn "friends", WJ:412) or Nindalf "Wetwang" (a compound of nîn "wet" and talf "flat field", see A Tolkien Compass p. 195). The uninformed have sometimes assumed that a name like Gildor means "Star-land", sc. that the final element is the same as in country-names like Gondor, Mordor etc., but "Star-land" does seem like a strange name for a person. The final element of Gildor is actually taur "king, master", blended with an identical adjective meaning "lofty, noble". In Gildor, t becomes d by lenition, and unaccented au becomes o. The name is better interpreted "Star-lord".
The negative adverb avo, that is used with an imperative to express a negative command, causes soft mutation of following verb: caro! "do (it)!", but avo garo! "don't do (it)!" Avo may also be reduced to a prefix av-, still followed by the same mutation: avgaro means the same as avo garo. See WJ:371.
A noun is also lenited if it appears as the object of a verb, even if there is no article preceding it. Hence, Sindarin has an "accusative" of sorts. Notice one sentence from the King's Letter: ennas aníra i aran...suilannad mhellyn în, "there the king wants...to greet his friends", mhellyn being the lenited form of mellyn "friends" (and a variant spelling of vellyn as in Elvellyn "Elf-friends" above). The word "friends" is lenited as the object of the verb "greet". One wonders if the lack of lenition was the reason why Gandalf misunderstood the inscription on the Gate of Moria: Pedo mellon a minno, "say 'friend' and enter". Gandalf, as we recall, at first thought it meant "speak, friend, and enter". Normally, mellon should presumably have been lenited as the object of pedo "speak" (*pedo vellon), but the ones who made the inscription had evidently ignored the normal lenition rules and given the word mellon in exactly the form it had to be spoken for the doors to open. (Of course, we don't know exactly how the "magic" or para-technological mechanism behind the doors worked, but it must have been some kind of artificial intelligence responding to the sound-sequence M-E-L-L-O-N only.) Perhaps it was because of this Gandalf did not at first understand that mellon was the object of pedo "say, speak" and took it to be a vocative instead: "Speak, o friend!" It may be that the form of Sindarin used in this inscription did not use the lenition of m to mh/v at all, but actually there is a variant of the Moria Gate inscription where the tengwar seem to read pedo mhellon instead of pedo mellon. (See J. R. R. Tolkien: Artist & Illustrator, p. 158.)
It was formerly thought that the conjunction a "and" caused soft mutation (a view that was also reflected in some of the earliest versions of this article). This was because of the phrase Daur a Berhael "Frodo and Samwise" in LotR3/VI ch. 4: One correctly observed that Berhael "Samwise" is a lenited form of Perhael and rashly concluded that it was the preceding conjunction a that caused the mutation. However, the Moria Gate inscription has a minno, not **a vinno, for "and enter". Since mellon "friend" fails to lenit to vellon in the same inscription, one might think that the inscription is in a form of Sindarin that does not use the lenition m > v. However, as mentioned above, an alternative form of the inscription occurs in J. R. R. Tolkien: Artist & Illustrator p. 158. In this version, the word mellon is lenited (mhellon/vellon) - but the word minno following the conjunction still shows no lenition, once and for all burying the theory that a "and" triggers the soft mutation. Why, then, is Perhael lenited? The context must be taken into consideration. The whole sentence goes: Daur a Berhael, Conin en Annûn, eglerio! According to Letters:308, this means "Frodo and Sam, princes of the west, glorify (them)!" There is not actually any final pronoun "them" in the Sindarin sentence, as indicated by the parentheses. The object of the verb eglerio "glorify" is of course "Frodo and Sam", and being objects, these names are lenited. The sentence is simply a rearranged form of *eglerio Daur a Berhael, Conin en Annûn "glorify Frodo and Sam, Heroes of the West". Hence, it is not only the name Perhael that is lenited (to Berhael); we must assume that Daur is also a lenited form, the unmutated version being Taur. (According to LR:389 s.v. TÂ, TA3, "Noldorin"/Sindarin had an old adjective taur "lofty, noble", used in "ancient titles"; this would be a fitting honorary epithet for Frodo.) - As the example Daur a Berhael, Conin en Annûn "Frodo and Sam, heroes of the West" indicates, lenition is not carried through an entire phrase when the latter part merely stands in apposition to the first. The main words, Taur and Perhael, are lenited - but the phrase Conin en Annûn "heroes of the West", that merely stands in apposition to Daur a Berhael, is not (hence no "Gonin en Annûn" instead). Cf. also an example like i Cherdir Perhael, Condir "the Master Samwise, Mayor" from the King's Letter: Herdir "master" is lenited because of the article preceding it (actually it would have been lenited even without the article, since this phrase is also the object of a verb), but here, the name Perhael "Samwise" and his title Condir are not subjected to soft mutation, since they stand in apposition to Herdir (hence no "i Cherdir Berhael, Gondir"). So the rule is that when several words stand in apposition, only the first of them undergoes mutation (and this probably goes for all the mutations).

NOTE: Tolkien revised the lenition rules repeatedly. One obsolete rule may be mentioned. As noted above, the genitive may be expressed by word order alone in Sindarin: Ennyn Durin Aran Moria, "Doors (of) Durin Lord (of) Moria". According to a rule that Tolkien later rejected, the second noun of such a construction is lenited. Therefore, the first draft of the Moria Gate inscription had the reading Ennyn Dhurin Aran Voria, with Durin and Moria lenited. Compare some genitive phrases from the Etymologies, LR:369: Ar Vanwë, Ar Velegol, Ar Uiar for "Day of Manwë", "Day of Belegol (Aulë)", "Day of Guiar (Ulmo)" (b and m leniting to v and g to zero). After the revision, the forms would presumably be *Ar Manwë, *Ar Belegol, *Ar Guiar instead.
II. NASAL MUTATION
While this may sound like something out of a horror movie (or out of Pinocchio), it actually refers to another important phenomenon in Sindarin phonology. Just like the article i for singular "the" triggers soft mutation, the article in for plural "the" triggers nasal mutation: Tolkien explicitly stated that "the nasal mutation...appears after the plural article in: thîw, i Pheriannath" (Letters:427 - it seems that Humphrey Carpenter editing this letter thought that "in" is here the English preposition rather than the Sindarin article in, since he does not use italics!) Other particles triggering nasal mutation would be the preposition and prefix an "for, to" and the preposition dan "against", also used as a prefix "re-".
The examples Tolkien used in Letters:427 quoted above, thîw and i Pheriannath, come from the Moria Gate inscription and from the praise the Ringbearers received on the fields of Cormallen. In the former we have i thiw hin for "these signs", literally "the signs these". (The shortening of thîw to thiw probably has something to do with the following hin "these" and doesn't have to be considered here.) Frodo and Sam were praised with the words aglar 'ni Pheriannath, "glory to the halflings" ('ni being short for an i "to the"). But why is the article i seemingly used in conjunction with these plural words, when we have already established that the word for plural "the" is in instead? Another anomaly seems to be that "letters" and "halflings" suddenly appear as thîw (thiw) and Pheriannath instead of tîw and Periannath, though these words are attested in LotR itself (Appendix B, the chronology of the Third Age, entry for 1050: "The Periannath are first mentioned in records..." - while in Appendix E reference is made to the "the Tengwar or Tîw, here translated as 'letters' "). Both of these problems are solved when we take into consideration the effects of nasal mutation: I thîw and i Pheriannath actually represent in tîw, in Periannath. The King's Letter has a Pherhael for "to Perhael (Samwise)"; this represents an "for" + Perhael. If we wanted to say in cirth = "the runes", this would manifest as i chirth. In terms of diachronic phonology, this whole phenomenon is easily explained. In Old Sindarin, p, t, k (c) following an n became aspirated, turning into an aspirate ph, th, kh. Compare an Old Sindarin word like thintha- "fade" (LR:392 s.v. THIN), undoubtedly representing even older *thintâ- with the common verbal ending -tâ. Hence we also had in tîw > i thîw (th here being aspirated t rather than a spirant þ). Later, the aspirates turned into spirants and the preceding nasal was assimilated to them, in effect disappearing (in þîw > iþ þîw, i þîw, normally spelt i thîw in Roman letters).
The nasal mutations of the unvoiced stops p, t, c are thus ph, th, ch. The initial clusters cl, cr, tr, pr probably behave in the same way as the simple stops when nasal mutation is due (so if we combine words like claur "splendor", crûm "left hand", trenarn "account", prestanneth "affection" with the preposition an "to, for", we might see a chlaur, a chrûm, a threnarn, a phrestanneth).
The voiced stops b, d, g behave differently when subjected to nasal mutation. They do not turn into spirants like the unvoiced stops. There has, however, been some confusion regarding their behavior. Earlier versions of this article presented the view that n + b, d, g produces mb, nd, ng. There is little doubt that this was indeed what Tolkien imagined at one stage. This is evident from the example Cerch iMbelain "Sickle of the Valar" in LR:365 s.v. KIRIK, clearly cerch "sickle" + in plural article "(of) the" + Belain "Valar". However, one late example indicates that Tolkien abandoned this "Noldorin" system in Sindarin. In WJ:185, we have Taur-i-Melegyrn for "Forest of the Great Trees". This is clearly taur "forest" + in plural article "(of) the" + beleg "great" + yrn "trees". (The word beleg is listed in the Silmarillion Appendix, there glossed "mighty".) Here, n + b is seen to produce m; by the same system, "Sickle of the Valar" would of course be Cerch i Melain (not, as before, Mbelain). By analogy, we have to conclude that n + d produces simple n, while n + g comes out as ng (a unitary sound as in English sing, sometimes spelt ñ by Tolkien, not this unitary sound followed by a distinct g, as in English finger):

in pl. "the" + Dúredhil "Dark Elves" = i Núredhil "the Dark Elves"
in pl. "the" + gelaidh "trees" = i ngelaidh (sc. i ñelaidh) "the trees"
in pl. "the" + beraid "towers" = i meraid "the towers"

Theoretically, we have long or double consonants here (innúredhil, iññelaidh, immeraid), though this is hardly reflected in pronunciation. But in the case of the prepositions an "to, for" and dan "against", that trigger similar mutations, it would be in keeping with Tolkien's general principles to mark this in spelling (though we lack exactly parallel examples):

an + Dúredhel "Dark Elf" = an Núredhel (rather than simply a Núr...) "for a Dark Elf"
an + galadh "tree" = an Ngaladh "for a tree" (provisory Roman spelling of añ Ñaladh, the equivalent of which would probably appear in Tengwar writing)
an + barad "tower" = am marad "for a tower"

It is desirable to keep the preposition an clearly separate from the conjunction a "and"; confusion could arise if we simply wrote a Núredhel, a marad (the first of which might be misinterpreted "and a Deep-elf").

Before some consonant clusters beginning in voiced stops, such as dr, gl, gr, gw, it may seem that no particular mutation occurs. In LotR Appendix A, we have Haudh in Gwanûr for "Mound of the Twins" (not **Haudh i Ngwanûr); cf. also Bar-in-Gwael "Home of the Gulls" (?) in WJ:418 (not **Bar-i-Ngwael). So combining an, dan, in with words like draug "wolf", glân "border", grond "club" or gwêdh "bond" may produce simply dan draug "against a wolf", dan glân "against a border", dan grond "against a club", dan gwêdh "against a bond" (definite plurals in droeg "the wolves", in glain "the borders", in grynd "the clubs", in gwîdh "the bonds"). Compare Tawar-in-Drúedain for "Forest of the Drúedain (Woses)" in UT:319; the initial dr is not changed by any visible nasal mutation, even though it follows the plural article in "(of) the". Cf. also the exclamation gurth an Glamhoth "death to (the) Din-horde (= Orcs)" in UT:39, 54, providing an attested example of an "to" followed by a word in gl-. It is, however, probable that the final n of dan, an, in would be pronounced "ng" (ñ) before words beginning with a cluster in g-, and perhaps also so written in Tengwar spelling.
The clusters bl, br may become ml, mr when subjected to nasal mutation, e.g. an "for" + brôg = a mrôg (or am mrôg) "for a bear", definite plural i mrýg "the bears". We have no examples, but general principles may suggest this.
Before m, the preposition an "to, for" appears as am; the King's Letter has am Meril for "to Meril [Rose]". Dan "against" would surely become dam in the same position (dam Meril "against Meril"). The plural article in appears as i when followed by m; WJ:418 has Bar-i-Mýl for "Home of the Gulls" (changed by Tolkien from Bar-in-Mýl with the n intact). Cf. also a phrase like Gwaith-i-Mírdain "People of the Jewel-Smiths", clearly representing ...in Mírdain. Before words in n, we would again see in reduced to i (cf. i Negyth for in Negyth "the Dwarves", WJ:338). The prepositions an, dan would be unchanged.
Before s, in is again reduced to i, as in Echad i Sedryn "Camp of the Faithful" (UT:153). The prepositions an "to, for" and dan "against" may appear as as, das before s- (e.g. as Silevril "for a Silmaril").
No examples show what nasal mutation does to initial r-. In Third Age Sindarin at least, n + r produced dhr (as in Caradhras = caran "red" + ras(s) "horn"). So perhaps, say, "against a horn", dan + rass, would produce dadh rass??? Definite plural idh rais "the horns", for in rais? But in First Age Sindarin, or at least in the Doriathrin dialect, we might see simply dan rass, in rais (compare the name of Thingol's sword Aranrúth "King's Ire", indicating that the change nr > dhr still had not occurred in his day).
Before l, the final nasal of the plural article in disappear. Compare Dantilais as a name of Autumn in PM:135; this is transparently Dant i Lais "Fall of the Leaves" (for Dant in Lais) written in one word as a pseudo-compound. The prepositions an, dan may appear as al, dal before a word in l-.
The behavior of unvoiced L and R, sc. lh, rh, can only be guesswork. An "for" + lhûg "dragon" or rhavan "wild man" may produce al 'lûg "for a dragon", adh 'ravan "for a wild man" (or, with in = plural "the", i 'lýg for il 'lýg "the dragons", but idh 'revain "the wild men"). The ' would indicate the loss of a consonant, the s of the original clusters sl-, sr- that yielded lh-, rh-. See under Mixed Mutation below concerning the attested (?) example e-'Rach.
Nasal mutation turns h into ch, as in Narn i Chîn Húrin "Tale of the Children of Húrin", i Chîn representing in Hîn (compare hên "child", pl. hîn). It should be noted that the form Narn i Hîn Húrin occurring in UT is wrong. In LR:322, Christopher Tolkien confesses: "Narn i Chîn Húrin...is so spelt at all occurrences, but was improperly changed by me to Narn i Hîn Húrin (because I did not want Chîn to be pronounced like Modern English chin)." (Cf. MR:373.) Before h > ch, the prepositions an, dan may simply be spelt a, da (a chên "for a child", da chên "against a child" - ach chên, dach chên would also be a possibility, but no unmutated Sindarin word begins in ch, so there can be no confusion with a hên "and a child").
The nasal mutation of hw may follow the same (hypothetical) pattern as lh, rh, e.g. an "for" + hwest "breeze" > a 'west "for a breeze".
The sounds th, f seem immune to all sorts of mutations. In pl. "the" + thynd "roots" would probably appear simply as i thynd; in the case of an "for" and dan "against" we might see ath thond "for a root", dath thond "against a root", or one might simply write a thond (and risk confusion with "and a root"), da thond. Likewise in > i before f (cf. i-Fennyr for in-Fennyr in LR:387 s.v. SPAN). An, dan might come out as af, daf before f; in this case, final f would actually be pronounced [f] rather than [v], despite Tolkien's normal orthographic conventions. Compare his use of ef as an assimilated form of ed "out of" before words in f-; see the section about the Mixed Mutation below.

III. MIXED MUTATION
"Mixed mutation" is not a Tolkien-made term; we don't know what he called it. In the published material, this mutation is nowhere explicitly referred to; we merely observe its effects in a some texts. Sometimes it is similar to soft mutation, sometimes to nasal mutation, and historically both mutations are probably involved - hence this mutation may be called "mixed" (but sometimes it differs from both soft and nasal mutation!)

No less than three examples of mixed mutation are found in one sentence in the King's Letter: erin dolothen Ethuil, egor ben genediad Drannail erin Gwirith edwen "on the eighth [day] of Spring, or in the Shire-reckoning on the second [day] of April". Here we have three examples of prepositions that incorporate the definite article in the oblique form -(i)n: twice erin "on the" (or "on" + in "the" > umlauted örin > later erin), plus ben, here translated "in the", but more literally "according to the" (be "according to" clearly being the cognate of Quenya ve "like, as"; hence ben genediad Drannail "according to the Shire reckoning"). Other prepositions incorporating the article in the form -in or -n, such as nan "to the", uin "from the, of the" and possibly 'nin "to/for the", would be followed by the same mutations (at least in the singular - in the plural we may see nasal mutation instead, cf. 'ni Pheriannath "to the halflings", for 'nin [= an in] Periannath). But what kind of mutations are we talking about?
Because of the -n we might expect something similar to nasal mutation, but the sentence from the King's Letter shows that this is not the case. Consider the phrases erin dolothen "on the eighth", ben genediad "according to the reckoning", erin Gwirith edwen "on April the second" (literally "on the April second"). The unmutated form of dolothen "eighth" is clearly tolothen (compare toloth "eight", LR:394 s.v. TOL1-OTH/OT). Yet we see no nasal mutation (**eri[n] tholothen), but rather a shift t > d that is similar to a soft mutation. But soft mutation would also lenit g to zero. Even so, genediad "reckoning" and Gwirith "April" are unaffected when preceded by ben, erin. (We know that the unmutated forms would also show g-; for genediad compare the verb gonod- "reckon" in LR:378 s.v. NOT, while the month-name Gwirith is mentioned in LotR Appendix D.) We do not see **erin 'enediad, **erin 'Wirith with regular soft mutation here.
The singular genitival article e, en "of the" is seen to trigger similar mutations. Consider some of the names of various tales listed in MR:373. In Narn e·Dinúviel, "Tale of the Nightingale", we see the same "soft mutation" t > d as in erin dolothen for erin tolothen (for the unmutated form of Dinúviel is of course Lúthien's well-known epithet Tinúviel). But again we see that no such soft mutation affects voiced plosives like b, d, g (cf. Gwirith, genediad remaining unchanged): MR:373 also lists Narn e·Dant Gondolin, "Tale of the Fall of Gondolin", where dant "fall" undergoes no mutation (we know that the unmutated form is also dant; compare Dantilais for *"Fall of the Leaves = Autumn" in PM:135; the stem is DAT, DANT "fall down", LR:354). We do not see **e·Dhant with soft mutation.
The origin of these "contradictory" mutations evidently have to do with soft and nasal mutation operating on different stages in the evolution of Sindarin. We needn't enter into the phonological intricacies here, but rather simply set out their effects as far as they can be reconstructed - for to a large extent, we have to rely on reconstruction.

The best-attested effects of the mixed mutation may be inferred from the examples given above. The unvoiced plosives p, t, c are voiced to b, d, g (pân "plank", caw "top", tâl "foot" > e-bân "of the plank", e-gaw "of the top", e-dâl "of the foot", and likewise erin bân, erin gaw, erin dâl for "on the plank/top/foot"). The voiced plosives b, d, g are unchanged (benn "man", daw "gloom", gass "hole" > e-benn "of the man", e-daw "of the gloom", e-gass "of the hole", and likewise erin benn "on the man" etc.) It is hardly necessary to point out that there is room for some confusion here, since the phonemic distinction between voiced and unvoiced plosives is neutralized in this position. Only the context can tell us whether, say, e-gost means "of the quarrel [cost]" or "of the dread [gost]".

Before the initial cluster tr-, we would probably see the full form of the genitival article (en), and the cluster tr itself would mutate to dr, e.g. trenarn "tale" > en-drenarn "of the tale". Original dr, as in draug "wolf", would behave in the same way, but here there is of course no visible mutation (en-draug "of the wolf"). The clusters pr and br may both come out as mr, and the article takes the short form e-: prestanneth "affection" > e-mrestanneth "of the affection", brôg "bear" > e-mrôg "of the bear". The cluster bl may likewise become ml-, as in blabed "flapping" > e-mlabed "of the flapping". Here the mixed mutation is similar to nasal mutation. The clusters cl-and cr- would behave more like tr-, being voiced (to gl-, gr-), but we would see only the short form of the article before them: claur "splendor" > e-glaur "of the splendor", crist "cleaver" (sword) > e-grist "of the cleaver". On the other hand, the long form en- is used before gl-, gr-, gw-, and these clusters undergo no change: gloss "snow" > en-gloss "of the snow" (compare Methed-en-glad "End of the Wood" in UT:153), grond "club" > en-grond "of the club", gwath "shade" > en-gwath "of the shade".

Before words in f-, the example Taur-en-Faroth would seem to indicate that the article appears in its full form en- (for this example, see the Silmarillion Appendix, entry faroth - Taur-en-Faroth does not seem to mean precisely "Hills of the Hunters", though). It is very uncertain how words in h-, l-, m-, th- would behave; possibly the genitival article would take the short form e-, and the initial consonant would undergo no change: e-hên "of the child", e-lam "of the tongue", e-mellon "of the friend", e-thond "of the root". Perhaps we would also have short e- before words in s-, but this consonant would probably become h-: salph "soup" > e-halph "of the soup". Before n- we have long en-; compare a name like Haudh-en-Nirnaeth "Mound of Tears", occurring in the Silmarillion. Before r- the genitival article may take the form edh- because of the dissimilation nr > dhr, e.g. edh-rem "of the net", but en-rem may also be permissible, at least in Doriathrin Sindarin.

This leaves only three initial sounds to be accounted for: all of them descended from clusters in s-, namely lh, rh, hw from primitive sl-, sr-, sw-. What effect does the mixed mutation have on unvoiced L, R, W? We have one possible attestation of such a mutation: The phrase Narn e·'Rach Morgoth "Tale of the Curse of Morgoth" in MR:373. This example indicates that 'rach is what the word for "curse" turns into when subjected to the mixed mutation. Unfortunately, this word is not otherwise attested, so we don't know for sure what the unmutated form would be. It has generally been assumed that this is a lenited form of *grach. But if so, analogous examples suggest that "of the curse" would be *en-grach. It may be, then, that the unmutated form is actually *rhach, primitive *srakk-, the ' of e·'rach marking the loss of this s (and/or the loss of its effect on the unmutated form, in which s, though no longer present as a distinct sound, has made the following r unvoiced: rh). If this is correct, we would expect the mixed mutation to have a similar effect on lh, hw, e.g. lhûg "dragon" > e-'lûg "of the dragon", hwest "breeze" > e-'west "of the breeze".

The prepositions that incorporate the article as -n or -in would trigger mutations similar to those just described for the genitival article en-, but there is apparently no variation between forms where n is included and "short" forms where it is omitted, paralleling the variation en/e: An n representing the article is always present. (Contrast erin dolothen and e·Dant; we don't see **eri·dolothen paralleling e·Dant or **en Dant paralleling erin dolothen.)
IV. STOP MUTATION
The term "stop mutation" does not occur in Tolkien's published writings on Sindarin, but a reference to this mutation (by this name) does occur in one of the first entries of the "Gnomish Lexicon" of 1917 (see Parma Eldalamberon #11). In later material, there is one brief reference to what could also be termed stop mutation. In WJ:366, we read: "As the mutations following the preposition o ['from, of'] show, it must prehistorically have ended in -t or -d." Unfortunately, the Professor told us nothing more about these mutations. Our few examples of o occurring in actual texts would seem to indicate that nothing happens to an m or a g following this preposition (o menel "from heaven" and o galadhremmin ennorath "from the tree-tangled lands of Middle-earth" in the hymn to Elbereth, + o Minas Tirith "from Minas Tirith" in the King's Letter), and o also has this form before vowels (o Imladris "from/of Rivendell" in RGEO:70, in Tengwar writing; cf. also Celebrimbor o Eregion "Celebrimbor of Hollin" in the Moria Gate inscription). Tolkien further noted concerning the development of the primitive preposition et "out, out of" in Sindarin: "[It] retains its consonant in the form ed before vowels, but loses it before consonants, though es, ef, eth are often found before s, f, th." We will use ed to illustrate the mutations caused by the final stop, as well as they can be reconstructed. Due to lack of examples, much of what follows must remain hypothetical extrapolation.

Before a vowel, Tolkien informs us that we see the basic form ed (e.g. ed Annûn "out of [the] West"). But before consonants, ed appears as e, but the following consonant would often change. If we can trust our understanding of the phonological evolution of Sindarin, the unvoiced stops t-, p-, c- would turn into spirants th-, ph-, ch- (the clusters tr-, pr-, cl-, cr- likewise become thr-, phr-, chl-, chr-):

pân "plank" > e phân "out of a plank"
caw "top" > e chaw "out of a top"
taur "forest" > e thaur "out of a forest"
claur "splendor" > e chlaur "out of splendor"
criss "cleft" > e chriss "out of a cleft"
prestanneth "affection" > e phrestanneth "out of affection"
trenarn "tale" > e threnarn "out of a tale"

On the other hand, the voiced plosives b-, d-, g- (occurring alone or in clusters bl-, br-, dr-, gl-, gr-, gw-) would undergo no change: Compare o galadhremmin ennorath "from the tree-tangled lands of Middle-earth" in the hymn to Elbereth; the word galadh "tree" is unchanged.

barad "tower" > e barad "out of a tower"
daw "gloom" > e daw "out of gloom"
gass "hole" > e gass "out of a hole"
bronwe "endurance" > e bronwe "out of endurance"
blabed "flapping" > e blabed "out of flapping"
dring "hammer" > e dring "out of a hammer"
gloss "snow" > e gloss "out of snow"
groth "cave" > e groth "out of a cave"
gwath "shadow" > e gwath "out of shadow"

The system here sketched refers to "normal" b, d, g; notice that where these sounds come from primitive mb, nd, ñg, they behave differently. See "The development of nasalized stops" below.

Words in m- and n- would not change, either:

môr "darkness" > e môr "out of darkness"
nath "web" > e nath "out of a web"

But h- and hw- may become ch- and w-, respectively:

haust "bed" > e chaust "out of a bed"
hwest "breeze" > e west "out of a breeze"

As for the form of ed before s-, f-, th-, we are told that "es, ef, eth are often found" (WJ:367) before these consonants:

sarch "grave" > es sarch "out of a grave"
falch "ravine" > ef falch "out of a ravine"
thôl "helm" > eth thôl "out of a helm"

However, Tolkien's wording "often found" rather than "always found" indicates that e sarch, e falch, e thôl would be equally permissible. The preposition ned *"in", that probably behaves like ed "out of", should probably not be nef (but rather ne) before a word in f-, since the spelling nef would cause confusion with the distinct preposition nef "on this side of". (There would be no confusion if it had not been for Tolkien's idea that final [v] is to be spelt f in his Roman orthography for Sindarin; nef "on this side of" is pronounced [nev], but nef as a form of ned would be pronounced [nef]. Ef, nef as forms of ed, ned should strictly speaking have been spelt eph, neph according to Tolkien's orthographic system, since they are pronounced [ef], [nef] - but in WJ:367, Tolkien himself uses the spelling "ef"!)

The unvoiced liquids lh, rh may behave like we have assumed that they do under the influence of soft mutation: turn into thl-, thr-. (It must be emphasized that this is speculation and at best a qualified guess, which goes for many of the possible effects of the stop mutation presented here. Of all the unattested forms, only the behavior of the unvoiced stops is relatively certain.)

lhewig "ear" > e thlewig "out of an ear"
Rhûn "East" > e Thrûn "out of (the) East"

As for normal, voiced l, r, the general principles of Sindarin phonology (as far as they can be reconstructed) may suggest that "out of" would here appear in its full form ed, despite Tolkien's statement in WJ:367 that the final stop is lost before consonants:

lach "flame" > ed lach (e lach?) "out of a flame"
rond "cave" > ed rond (e rond?) "out of a cave"

This hopefully covers the mutations caused by ed "out of"; ned *"in" would behave in the same way. The preposition o "from, of" causes the same mutations, but here the preposition itself does not change its form (no variation corresponding to ed/e). Tolkien noted, however, that o occasionally appears in the form od before vowels (WJ:367). As mentioned above, Tolkien himself used o Eregion "of Hollin" in the Moria Gate inscription and o Imladris for "from/of Rivendell" in RGEO:70 (in Tengwar writing). Od Eregion and od Imladris would apparently have been possible, but not necessary. However, Tolkien noted that od was more usual before o- than before other vowels, so (say) "from/of an Orch" should perhaps be rendered od Orch rather than o Orch to avoid two identical vowels in hiatus.

V. LIQUID MUTATION
This mutation represents a leap of faith. It is not mentioned, alluded to or directly exemplified anywhere in the published material; yet our general understanding of Sindarin phonology seems to demand it. If Tolkien adhered to his own rules (he did sometimes), we would expect liquid mutation in Sindarin.

We know that following the liquids l, r, Sindarin at one point changed plosives to spirants (UT:265, footnote); compare Telerin alpa "swan" with Sindarin alph, or Quenya urco "Orc" with Sindarin orch. This does not only happen in unitary words. The prefix or- "over", clearly separable, is seen to cause a similar change in the verb ortheri "master, conquer", literally *"over-power" (LR:395, where the stem is given as TUR "power, control"). There is little reason to doubt that or, also when appearing as an independent preposition "over, above, on", would trigger similar changes in the word that follows: Stops become spirants.

pân "plank" > or phân "above a plank"
caw "top" > or chaw "above a top"
tâl "foot" > or thâl "above a foot"
benn "man" > or venn "above a man"
doron "oak" > or dhoron "above an oak"

G originally turned into a spirant gh, but this sound later disappeared (marked by ' where it formerly occurred):

galadh "tree" > or 'aladh "above a tree" (archaic or ghaladh)

It does not matter whether the initial stop occurs by itself or as part of a cluster; it would still turn into a spirant under the influence of liquid mutation (tr- > thr-, pr- > phr, cl- > chl-, cr- > chr-, dr- > dhr-, bl- > vl-, br- > vr-, gl- > 'l, gr- > 'r, gw- > 'w).

M, like b, would probably turn into v when subjected to liquid mutation. This change is seen in unitary words; cf. primitive *gormê (Quenya ormë) "haste" yielding Sindarin gorf (LR:359 s.v. GOR; gorf is of course just Tolkien's way of spelling gorv, since final [v] is represented by the letter f). Hence:

mîr "jewel" > or vîr "above a jewel" (archaic or mhîr, where mh = nasalized v)

H- and hw- are probably strengthened to ch-, chw-, under the influence of liquid mutation:

habad "shore" > or chabad "above a shore"
hwand "fungus" > or chwand "above a fungus"

For the change h > ch, compare a word like hall "high" becoming -chal when or- is prefixed to produce a word for "superior, lofty, eminent" - orchal literally meaning over-high, super-high. ("Orchel" in LR:363 s.v. KHAL2 is a misreading; compare WJ:305.)

The unvoiced liquids lh, rh may become 'l, 'r, as we surmised is the case of nasal and mixed mutation:

lhûg "dragon" > or 'lûg "above a dragon"
Rhûn "East" > or 'Rûn "above (the) East"

The voiced liquids r, l would be unaffected by the liquid mutation:

rem "net" > or rem "above a net"
lam "tongue" > or lam "above a tongue"

The unvoiced spirants f, th, the nasal n and the sibilant s would not be affected, either:

fend "threshold" > or fend "above a threshold"
thond "root" > or thond "above a root"
nath "web" > or nath "above a web"
sirith "stream" > or sirith "above a stream"

SPECIAL CASES: The development of nasalized stops
There exists a subcategory of words in b-, d-, g- that needs to be watched, and that must be memorized separately. In the words in question, b-, d-, g- does not come from b-, d-, g- in the primitive language. Instead, they were originally nasalized stops mb-, nd-, ñg- (ñ representing the sound of ng as in English sing, and ñg therefore being pronounced like "ng" in English finger, with a distinct, audible g). In Sindarin, you cannot readily tell whether the initial consonant in a word like Golodh "Noldo" is a "normal" g, sc. one that was g all along, or whether it represents earlier ñg-. But it is important to know this, for when mutations are due, a word that originally began in a nasalized stop behaves quite differently from a word that had a simple stop all along. For instance, if the first consonant of Golodh had been a "normal" g, prefixing the article i would have produced i 'Olodh for "the Noldo" - g being lenited to zero because of the soft mutation triggered by the article. Cf. one example quoted above, in the section about the soft mutation: galadh "tree" > i 'aladh "the tree" (LR:298). But the g of galadh was a simple g also in the primitive language (where the word appeared as galadâ). The g of Golodh, on the other hand, was originally ñg; the word descends from primitive ñgolodô. When we prefix the article and thereby trigger soft mutation, the resulting form is actually not i 'Olodh, but i Ngolodh.

Already in Tolkien's earliest "Gnomish" language (ca. 1917), we find the idea that the original nasalized stops behave in a special way in mutation position. In the Gnomish Grammar of 1917 (published along with the Gnomish Lexicon in Parma Eldalamberon #11), the principle described is that the original nasalized stops were preserved when the article is prefixed. Hence we had for instance balrog "demon, balrog" > i mbalrog "the demon", dôr "land" > i ndôr "the land", Golda "Gnome, Noldo" > i Ngolda "the Gnome". Is this system still valid in Sindarin? In WJ:383, in an essay dating to ca. 1960, Tolkien indicated that the Sindarin word for Noldo was "Golodh (Ngolodh)". So the word Golodh sometimes appears as Ngolodh instead. In the essay in question, Tolkien did not clarify where the form Ngolodh would be used, but the variation Golodh/Ngolodh seemed to correspond to Gnomish Golda/Ngolda. Earlier versions of this article therefore presented the view that the soft mutation of b, d, g, where these sounds were nasalized in the primitive language, is mb, nd, and ng - the original nasalized stops being restored, or rather preserved, in this position.

However, a closer look at Sindarin phonology seems to indicate that it was rash to conclude that the "Gnomish" system was still valid in later Grey-elven (and demonstrates that Tolkien's early material must be treated with considerable skepticism if one wants to learn LotR-style Elvish, despite certain claims made by the editors that the publication of the Gnomish Grammar and Lexicon would throw more light upon Sindarin). The soft mutation corresponds to how certain consonants or consonant groups develop between vowels. It is triggered, among other things, by the negative prefix ú-. So if we prefix it to a verb like bartha- "doom", derived from the stem MBARAT, what do we get? The related word úmarth "ill-fate", where the same prefix occurs (though with a different shade of meaning), points unequivocally to *ú-martha for "does not doom". The soft mutation of b, where it represents primitive mb, is therefore m. The soft mutation of d derived from primitive nd would then be n. This largely corresponds to the development of the mb, nd medially, where they become m(m), n(n) - e.g. amar "earth" as the cognate of Quenya ambar, or annon "gate" corresponding to Quenya andon. What, then, about the attested form Ngolodh - apparently the soft mutation of Golodh? Is not the original initial cluster of primitive ngolodô preserved here, just as in Gnomish? Probably not; we are merely being confused by an unfortunate deficiency of the English alphabet, the absence of a single letter for the sound that often spelt ng, as in sing, thing. As already mentioned, Tolkien sometimes denoted this sound as ñ. This single, unitary sound ñ must be distinguished from ñ + g, which is what the spelling ng denotes in finger. It seems that in Sindarin Ngolodh, the initial ng is to be pronounced as in sing, sc. simple ñ with no audible g - whereas in Gnomish Ngolda, the spelling ng indicates a real cluster, pronounced as in English finger. Hence, the mutation products of g from primitive ñg are not really the same in Sindarin and Gnomish after all, and the treatment of b, d from mb, nd also differs.

bâr "land, home" (stem MBAR) > i mâr "the land, the home" (not i mbâr as stated in earlier versions of this article)
dôl "head" (earlier ndolo) > i nôl "the head" (not i ndôl)
Golodh "Noldo" (primitive ngolodô) > i Ngolodh "the Noldo" (sc. i Ñolodh, not i Ñgolodh with a real consonant cluster)

Update: Since I wrote the above, another relevant example has been published. Tolkien's incomplete Sindarin Lord's Prayer includes the words i mbas "the bread" (the unmutated word for "bread" being mas/mass, from a root MBAS). This kind of mutation is surprising in such a late text: For a moment at least, Tolkien seems to have revived the system he used in his very earliest "Gnomish" language. However, we have also had explicit confirmation of the system whereby b, d, g from primitive mb, nd, ñg are lenited to m, n, ñ (spelt ng), respectively: It turns out that such a system had come into place already in one variant of early "Noldorin"; see the table of mutations published in Parma Eldalamberon #13 p. 120. This table even provides explicit Tolkienian confirmation of one of the forms listed above, i mâr, still unattested when I originally wrote this article. This system does seem to fit the general phonology best. I would therefore write i mas, not i mbas, for "the bread" - irrespective of Tolkien's curious indecision in this matter.

Actual clusters, or nasalized stops, do arise when nasal mutation is due. The plural of bâr "land, home", bair, occurs in the King's Letter (SD:129), combined with the plural article in, and this combination is seen to produce i Mbair "the lands". So when in = plural "the" occurs before b or d representing mb, nd, the final n of the particle is dropped, but the original nasalized stop reappears. In the case of the other particles triggering nasal mutation, namely an "for" and dan "against", it may be convenient to let the final nasal of the particle remain in spelling; for instance, "for a land" (an + bâr) may be represented as am mbâr (an becoming am before m-), and likewise dam mbâr "against a land" (dan + bâr). Similarly an ndôl "for a head" and dan ndôl "against a head" (an/dan + dôl).
As for the nasal mutation of g from primitive ng, this would on the same principle be ng; so if we want to say "for a Noldo" (an + Golodh), we would expect an Ngolodh (actually añ Ñgolodh, with ñg like ng in English finger, with an audible g). This spelling, however, would create a problem. The nasal-mutated form of normal g (derived from primitive g, not ng) is also spelt ng (e.g. an + galadh = an ngaladh [sc. añ ñaladh] "for a tree"). Upholding the distinction between ñ and ñg is no problem in Tengwar writing, but when using our normal alphabet to write Sindarin, we have to use special solutions. The plural Gelydh, when combined with the article in, might have produced i Ngelydh (sc. i(ñ) Ñgelydh - the corresponding spelling would be used in Tengwar writing). But presumably to make it clear that the intended pronunciation is indeed i Ñgelydh and not i Ñelydh, Tolkien used the spelling in Gelydh instead (cf. place-names like Annon-in-Gelydh "Gate of the Noldor" mentioned in the Silmarillion). In this way - by keeping the n and the g clearly separate when the intended pronunciation is ñg rather than ñ - the distinction can be upheld. So "for a Noldo" or "against a Noldo" would also be simply an Golodh, dan Golodh (as if there is no mutation at all - but it should be realized that the proper or ideal spellings would be a(ñ) Ñgolodh and da(ñ) Ñgolodh, and that the corresponding spelling would be used in Tengwar writing). When in, dan or an precedes a word in g-, remember that the final n is pronounced ng as in sing.

NOTE: It is interesting to notice the different mutations affecting the collective plural gaurhoth = "werewolves" or "werewolf-host". Gaur "werewolf" comes from an ng-stem (ÑGAW "howl", LR:377). In the case of a collective plural like gaurhoth, it is optional whether one uses the singular article i or the plural article in. In one of Gandalf's fire-spells, naur dan i ngaurhoth! *"fire against the werewolves!", the singular article i is used, causing soft mutation: i ngaurhoth = i ñaurhoth. But in the Silmarillion, we find the place-name Tol-in-Gaurhoth "Isle of the Werewolves", where the plural article in is used in front of the same collective plural. The Roman spelling in-Gaurhoth here represents i Ñgaurhoth with nasal mutation triggered by the final nasal of in, exactly parallel to in-Gelydh = i Ñgelydh "the Noldor".

As for the mixed mutation of b, d, g from mb, nd, ng, the example Narn e·mbar Hador *"Tale of the house of Hador" indicates that it is similar to the nasal mutation, mbar "house" exemplifying the mixed mutation of bar (bâr) "house, home, land" (stem MBAR "dwell, inhabit", though this word is not listed in Etym, LR:372). Hence b, d, g again "revert" to original mb, nd, ng, and just like we have e-mbar for "of the house", we would see for instance e-ndôl "of the head", en-Golodh "of the Noldo" (provisory Roman spelling of e-Ñgolodh). But spellings like en-ndôl may also be permissible; compare a name like Haudh-en-Ndengin "Hill of Slain" occurring in the Silmarillion.
When the article appears as -n or -in directly suffixed to a preposition, as in nan "to the" (na "to" + -n "the"), this final -n does not seem to be assimilated in any way (at least this is not reflected even in Tengwar writing):

nan "to the" + bâr "house" = nan mbâr "to the house"
nan "to the" + dôl "head" = nan ndôl "to the head"
nan "to the" + Golodh "Noldo" = nan Golodh (provisory and not wholly satisfactory Roman spelling for nan Ñgolodh) "to the Noldo"

The stop mutation following prepositions like o "from/of", ed "out of" and ned "in" would produce forms similar to the mixed mutation above. The prepositions ed, ned would appear in the short forms e, ne (but e ñg-, ne ñg- unfortunately have to be represented as en g-, nen g- in Roman spelling; morphologically speaking, the nasal has nothing to do where orthography forces us to place it):

bâr "house" > e mbâr "out of a house"
dôr "land" > e ndôr "out of a land"
gorth "horror" > en gorth "out of horror" (provisory Roman spelling for what is properly e ñgorth - not to be confused with en-gorth "of horror")

The liquid mutation probably caused by the preposition or "over, above, on" would have no apparent effect on b-, d-, g- descended from primitive nasalized stops (while "normal" b-, d-, g- turn into spirants v-, dh-, '-):

bâr "house" > or bâr "above a house"
dôr "land" > or dôr "above a land"
Golodh "Noldor" > or Golodh "above a Noldo"

The words involved: The words with initial b, d, g representing primitive nasalized stops must be memorized, and we will attempt to list most of them. As an example of an actual mutation we use lenition (soft mutation); the other mutations are described above. Where the word in question is a verb and not a noun, I list the form it would have following the particle i when used as a relative pronoun ("who, which") rather than as the article "the"; since this is merely a secondary use of the definite article (also found in German), the following mutations are the same. So from bartho "to doom" we have for instance i martha "who dooms" or "the [one who] dooms" (verbs with infinitives in -o forming their present tense in -a; see the section on verbs below). In the plural, the plural article in is used as a relative pronoun, triggering nasal mutation (hence "dead who live" is gyrth i chuinar = ...in cuinar), so "who doom" or "the [ones who] doom" must be i mbarthar.

1: Mutation of B from primitive MB

The "trade" words derived from the primitive stem MBAKH:

bachor "pedlar" > i machor "the pedlar"
bach "article (for exchange)" > i mach "the article"

The "doom" pair from MBARAT:

barad "doomed" > i marad "the doomed [one]" (contrast the homophone barad "tower" > i varad "the tower")
bartho "to doom" > i martha "the [one who] dooms"

The "bread" pair from MBAS:

bast "bread" > i mast "the bread"
basgorn "loaf" > i masgorn "the loaf"

The "duress" group from MBAD and MBAW:

band "duress, prison" > i mand "the prison"
baug "tyrannous, cruel, oppressive" > i maug "the tyrannous (one)"
bauglo "to oppress" > i maugla "the [one who] oppresses"
bauglir "tyrant, oppressor" > i mauglir "the tyrant"
baur "need" > i maur "the need"

The "festive" group from MBER:

bereth "feast, festival" > i mereth "the feast" (but mereth > i vereth may be more usual, cf. Mereth Aderthad, not *Bereth Aderthad, for "Feast of Reunion" in the Silmarillion)
beren "festive, gay, joyous" > i meren "the festive [one]" (contrast the homophone beren "bold" > i veren "the bold [one]" - but since Tolkien evidently settled on mereth instead of bereth as the word for "feast", we should probably read meren instead of beren as the word for "festive")

And miscellaneous:

bâr "home, land" > i mâr "the home" (stem MBAR, but this word is not given in Etym)
both "puddle, small pool" > i moth "the puddle" (MBOTH)
bund "snout, nose, cape" > i mund "the snout" (MBUD)

2: Mutation of D from primitive ND

The "slaying"-group from NDAK:

daen "corpse" > i naen "the corpse"
dangen "slain" > i nangen "the slain (one)"
dagor (older dagr) "battle" > i nagor (i nagr) "the battle"
daug "(Orkish) warrior" > i naug "the warrior"

The "hammering" group from NDAM:

dam "hammer" > i nam "the hammer"
damma- "hammer" as verb ("damna" in LR:375 must be a misreading) > i namma "the (one who) hammers"

The "head" pair from NDOL:

dôl "head" > i nôl "the head"
dolt "round knob, boss" > i nolt "the round knob"

(These may be somewhat uncertain; David Salo argues that dôl behaves like a normal word in D, hence *i dhol. Compare the name of the mountain Fanuidhol.)

And miscellaneous:

dûn "west" > i nûn "the west" (NDÛ)
Dân "Nandorin Elf" > i Nân "the Nandorin Elf" (NDAN)
dangweth "answer" > i nangweth "the answer" (since the primitive form of the word is given as ndangwetha in PM:395; evidently the first element is to be equated with the stem NDAN)
daer "bridegroom" > i naer "the bridegroom" (NDER; the "Noldorin" form doer must be emended to daer in Sindarin.)
dess "young woman" > i ness "the young woman" (NDIS)
dôr "land" > i nôr "the land" (NDOR)
dortho "to stay" > i northa "the (one who) stays" (NDOR)
doll "dark" > i noll "the dark" (NDUL)

3: Mutation of G from primitive ÑG

The "harping" pair from ÑGAN:

gannel "harp" > i ngannel "the harp"
ganno "to play a harp" > i nganna "the (one who) plays a harp"

The "wolf" group from ÑGAR(A)M and ÑGAW:

garaf "wolf" > i ngaraf "the wolf"
gaur "werewolf" > i ngaur "the werewolf" (cf. i ngaurhoth in one of Gandalf's fire spells).
gawad "howling" > i ngawad "the howling"

The "wise" group from ÑGOL:

golu "lore" > i ngolu "the lore" (the "Noldorin" word golw must become golu in Sindarin)
golwen "wise" > i ngolwen "the wise (one)"
goll "wise" > i ngoll "the wise (one)"
gollor "magician" > i ngollor "the magician"
Golodh "Noldo" > i Ngolodh "the Noldo"
gûl "magic" > i ngûl "the magic"
Golovir "Silmaril, Noldo-jewel" > i Ngolovir "the Silmaril"

and finally the words for "death" and "horror":

gûr "death" > i ngûr "the death" (also guruth, i nguruth) (ÑGUR)
goroth "horror" > i ngoroth "the horror" (ÑGOROTH)

SUMMARY
We will list all the attested and surmised mutations in table form. In the first column, we list all Sindarin initial consonants and consonant groups alphabetically, in their "Basic" = unmutated form. The soft mutation is exemplified by the article i = singular "the". To make things more complicated than necessary, there are two columns for the nasal mutation. The mutations as such are exactly the same, but in the first column ("Nasal I") the examples given involve the plural article in, which is reduced to i in most cases. However, in the case of the prepositions an "to, for" and dan "against" it is in many cases preferable (and in harmony with the attested example am Meril "to Meril/Rose") to use assimilated variants of the prepositions instead of simply reducing them to a, da in spelling, though this happens in some contexts (cf. a Pherhael "to Perhael/Samwise" in the same source that provides am Meril). The column "Nasal II" suggests various forms of an. The mixed mutation is exemplified by the genitival article en- "of the", the stop mutation by the preposition ed "out of", and the liquid mutation by the preposition or "above, on". (Before a word beginning in a vowel, that cannot be mutated in any way, all of these particles would appear in their full forms, as just quoted: i ael "the pool", in aelin "the pools", an ael "for a pool", en-ael "of the pool", ed ael "out of a pool", or ael "above a pool".)

The mixed mutations described above follow the system seen in such phrases as e-mbar Hador "of the house of Hador" (MR:373) and possibly Taur e-Ndaedelos "Forest of the Great Fear" (mentioned in LotR Appendix F as a Sindarin name of Mirkwood). Bar-en-Danwedh "House of Ransom", a name mentioned in the Silmarillion and clearly incorporating a descendant of the stem NDAN, ought to be spelt Bar-e-Ndanwedh instead. Perhaps Tolkien thought this looked somewhat uncouth and used a spelling more palatable to his readers. The full form of the article en "of the" is seen in another name from the Silmarillion, Haudh-en-Ndengin "Hill of the Slain". Here, a descendant of the stem NDAK is present, and initial nd is restored following en "of the". According to the system sketched above, this ought to be spelt Haudh-e-Ndengin instead (cf. Taur e-Ndaedelos), while based on the example Bar-en-Danwedh, we ought to write Haudh-en-Dengin. We needn't be worried by this. If Sindarin had been an actual spoken language in a "medieval" age, just like Tolkien imagined, there is every reason to believe that such inconsistencies in spelling would be quite common - various scribes using their more or less "private" systems, there being no central authority or language academy that could establish a standardized spelling.

It is hardly necessary to reiterate that the system set out above varies from certain, attested forms to very tentative speculation and sheer guesses, with several shades of more or less plausible interpolation between these extremes. Complexas this system may seem, it may still be over-simplified. Some points may be commented on:

1) Thr, thl as the soft mutations of rh, lh are phonetically sound, but remain speculative. In one name mentioned in the Silmarillion, Talath Rhúnen "East Vale", or literally and with Sindarin word order "Plain Eastern", the adjective rhúnen "eastern" is not lenited in any way, though adjectives in this position usually are. It would not be wrong, then, to let adjectives in lh-, rh- remain unchanged when they stand in apposition to a noun. By analogy, neither would it be a great sin to let nouns in lh-, rh- remain unchanged when they stand as the object of a verb, though "accusatives" are normally lenited. When a word functions as the second element of a compound, the initial consonant usually undergoes changes comparable to soft mutation, but lh, rh seem to become l, r in this position. Compare Rhûn "East" with -rûn in he longer word Amrûn of similar meaning. If thr, thl do occur as mutations of lh, rh, they may most typically appear following particles ending in a vowel, such as the definite article i or the preposition na "to".
2) We list m, n, ng as the soft mutation of b, d, g representing primitive mb, nd, ñg, but in some cases it would seem that these sounds behave like "normal" b, d, g, so that the lenited variants are v, dh, and zero, respectively. One "Noldorin" example is Nann Orothvor "Vale of Black Horror" (LR:355 s.v. DUN), where orothvor ("horror-black") is a lenited form of gorothvor, the first element goroth "horror" representing the stem ÑGOROTH of similar meaning (LR:377). It is remarkable that even g representing primitive ñg lenits to zero in Orothvor. In Sindarin as opposed to "Noldorin", a noun in genitive position would not be lenited, so we would see Nan(n) Gorothvor without any mutation. But in Sindarin, lenition does occur in comparable positions, like when an adjective in apposition (following the noun) undergoes soft mutation. We are left to wonder whether an adjective like goll "wise" (< stem ÑGOL) would appear as 'oll or ngoll in this position; maybe both would be permissible. Above, we have listed nôl as the lenited form of dôl "head" (< stem NDOL), but in the name of the mountain Fanuidhol "Cloudyhead" (found in LotR itself and therefore decidedly Sindarin rather than "Noldorin"), lenition d > dh is seen. Would it then be permissible to use i dhôl rather than i nôl for "the head"? Had Tolkien decided that the stem was DOL, not NDOL as it had been in the Etymologies (LR:376)?
3) The lenition m > v is sometimes ignored. Contrast a name like Eryn Vorn "Dark Wood" (UT:436, 262, cf. morn "dark") with Ered Mithrin "Grey Mountains" on the Map to LotR, or Imloth Melui in LotR3/V ch. 8 - not translated but evidently meaning "Lovely Flower-Vale". In light of the example Eryn Vorn, we must assume that *Imloth Velui and *Ered Vithrin would have been equally possible - and conversely, if we can have Imloth Melui and Ered Mithrin, we can presumably have *Eryn Morn as well. Above we noted that one has to rely on the context to distinguish the lenited variants of two adjectives like bell "strong" and mell "dear"; e.g. to decide whether i vess vell means "the strong woman" or "the dear woman". But if the lenition m > v is ignored, we can have the unambiguous phrase i vess mell for the latter meaning.

4. THE ADJECTIVE

Typical adjectival endings are -eb, -en and -ui: aglareb "glorious" (< aglar "glory"), brassen "white-hot" (< brass "white heat"), uanui "monstrous, hideous" (< úan "monster") (AKLA-R, BAN, BARÁS). However, many adjectives have no special endings, and the word-form as such sometimes belongs to more than one part of speech. Morn "dark" can be both adjective and noun, just like its English gloss.

Adjectives agree with their nouns in number. It seems that adjectives form their plurals following patterns similar to the noun plurals, e.g. malen "yellow", pl. melin (SMAL). Note that the initial consonant of adjectives following the noun they describe is lenited (see above).

In PM:358, Aran Einior is translated "the Elder King". Einior is our sole example of the comparative form of the adjective; the uninflected form is iaur (seen in the name Iant Iaur "the Old Brigde"). The prefix ein- seems to be related to the Quenya superlative prefix an-. The prefix may not have the form ein- prefixed to any adjective; it seems to be umlauted by the following i.

It so happens that we may also have the superlative form of iaur "old"; during the Council of Elrond, the Sindarin name of Tom Bombadil was given as Iarwain, meaning "Eldest". The ending -wain would seem to be the superlative suffix. Why not *Iorwain, with the normal monophthongization au > o? (David Salo answers, "Because you are looking at the direct descendant of a form like *Yarwanya (perhaps, I am not sure of the exact form of the final element) in which the vowel was in a closed syllable." I don't feel much wiser, but then I am not so deep into Eldarin phonology as David is.)

5. VERBS

"The Sindarin verbal system is not fully understood - far from it." So began the section on the Verb in my original Sindarin article, and this is to a large extent true still. However, I have since had the opportunity to acquaint myself with David Salo's interpretations and theories regarding the Sindarin verb, and what follows owes very much to his work. David's theories do seem to make a great deal of sense. It must still be realized that we have desperately few examples to go on, and that many conclusions must remain tentative at this stage. To be sure, hundreds of verbs are listed in the Etymologies, but we have so little actual Sindarin text that we cannot always be sure how these verbs are to be conjugated. In Etym itself, Tolkien did sometimes list a few inflected forms of a verb next to the basic form, but his notes are extremely dense, and often it is not even made clear what the inflected forms are intended to mean. But if we try to generalize from our few examples, taking into account everything we think we know about Eldarin phonology, the evolution of Sindarin and the primitive verbal systemas it can be inferred from Quenya, we may arrive at something like the system we are going to sketch here. The details can certainly be argued. To make this readable, I will for the most part skip the complex deductions that underlie the following scenario. The actual evidence, as well as the reasoning underlying the reconstructed system presented here, are presented in a separate article. While I will claim that the evidence has been thoroughly examined, future publications may well blow parts of the system sketched below to pieces. Yet I think we can be reasonably sure of the general outlines.

General: There seem to be two main categories of Sindarin verbs. As in Quenya, we can speak of derived verbs and basic verbs. The first, and larger, class consists of verbs that were originally formed by combining a primitive stem with some ending, suchas *- nâ (Sindarin -na), *-jâ (Sindarin -ia), *-tâ (Sindarin -da/-tha/-ta/-na, depending on the phonological environment), *-râ (Sindarin -ra) or *-â (Sindarin -a). Since all of these end in -a, this class can also be termed the A-stems. The other, smaller class consists of verbs that come directly from a primitive stem with no suffixes. For instance, nag- "bite" is simply the naked stem NAK as it appears in Sindarin. Since this category of verbs have present-tense stems in -i-, they may also be termed I-stems.

Suffixes: In many forms, Sindarin verbs (derived or basic) take endings for number and person. Sindarin, like Quenya, adds the ending -r to verbs with a plural subject; cf. the phase gyrth i-chuinar "dead that live" in Letters:417 (cuinar "live, are alive", here incidentally in nasal-mutated form chuinar, being the plural of cuina "lives, is alive"). Other endings denote various persons. Known pronominal endings include -n for "I", -m for "we" and apparently -ch or -g for "you". It is possible that the plural ending -r can denote "they" as well as mere plurality. The verb cuina- "live" can evidently have forms like cuinon "I live" (for *cuinan), cuinam "we live", cuinach or cuinag "you live" and cuinar "they live". The 3rd person singular does not seem to have any ending by itself: cuina "(he, she, it) lives". The 3rd person singular can in some cases be considered the basic form to which the various endings are added to produce forms for other persons and numbers.

I. DERIVED VERBS

# The conjugation of the derived verbs (A-stems) seems to be fairly straightforward, for the most part involving simply a series of suffixes. Indirect evidence may suggest that Tolkien would have termed this class the "weak" verbs.

# The infinitive is formed with the ending -o, displacing the ending -a:

bronia- "endure" > bronio "to endure"
dagra- "make war" > dagro "to make war"
esta- "call, name" > esto "to call, to name"
ertha- "unite" > ertho "to unite"
lacha- "flame" > lacho "to flame"
linna- "sing" > linno "to sing"
harna- "wound" > harno "to wound"

# The (3rd person singular) present tense is identical to the A-stem itself:

bronia- "endure" > bronia "endures, is enduring"
dagra- "make war" > dagra "makes war, is making war"
ertha- "unite" > ertha "unites, is uniting"
esta- "name" > esta "names, is naming"
lacha- "flame" > lacha "flames, is flaming"
linna- "sing" > linna "sings, is singing"
harna- "wound" > harna "wounds, is wounding"

The plural or pronominal endings mentioned above are added to this form: broniar "(they) endure", broniam "we endure" etc. Notice that the ending -n for "I" causes the final -a to become -o instead: hence bronion "I endure", dagron "I make war" etc.

# The (3rd person singular) past tense of this class of verbs is in most cases formed with the suffix -nt:

bronia- "endure" > broniant "endured"
dagra- "make war" > dagrant "made war"
esta- "call, name" > estant "called, named"
ertha- "unite" > erthant "united"
lacha- "flame" > lachant "flamed"
linna- "sing" > linnant "sang"
harna- "wound" > harnant "wounded"

Again, plural or pronominal endings may be added, just like in the present tense. If so, the suffix -nt becomes -nne- before the ending follows:

broniant "endured" > bronianner "they endured" (also plural, e.g. in Edhil bronianner "the Elves endured"), broniannen "I endured", broniannem "we endured" etc.

For, say, "(they) sang" we would expect linnanner (since "sang" is linnant), but wherever "double nn" would occur, the verb is probably contracted: "(they) sang" may simply be linner.

# The future tense is formed by adding the suffix -tha to the stem:

bronia- "endure" > broniatha "will endure"
dagra- "make war" > dagratha "will make war"
esta- "call, name" > estatha "will call, will name"
ertha- "unite" > erthatha "will unite"
lacha- "flame" > lachatha "will flame"
linna- "sing" > linnatha "will sing"
harna- "wound" > harnatha "will wound"

Again, plural and pronominal endings can be added, following the same rules as in the present tense. As in the present tense, the ending -n for "I" causes the final -a to become -o instead: broniathon "I will endure" (linnathon for "I will sing" is actually attested in LotR). Otherwise, the final -a is unchanged: broniatham "we will endure", linnathar "they will sing" etc.

# The imperative is formed with the ending -o, displacing the final -a. In this class of verbs, the imperative is therefore identical to the infinitive (see above). The imperative in -o covers all persons (Letters:427); hence the form is the same no matter whether the command is directed to one person or to several people. One Elf cried daro! "halt!" to the entire Fellowship as they were entering Lórien; see LotR1/II ch. 6. (In Quenya, it is optional whether one wants to make a plural/singular distinction in the imperative; we don't know if this can be done at all in Sindarin.)

# The active participle (also called present participle) is an adjective derived from a verb, describing the condition one is in when carrying out the action denoted by the verb (if you sing, you are singing; therefore, singing is the participle of the verb "to sing"). In Sindarin, the active participle of derived verbs is formed by means of the ending -ol, displacing the final -a of the verbal stem:

bronia- "endure" > broniol "enduring"
glavra- "babble" > glavrol "babbling"
ertha- "unite" > erthol "uniting"
lacha- "flame" > lachol "flaming"
linna- "sing" > linnol "singing"
harna- "wound" > harnol "wounding"

(The example glavrol is attested, LR:358 s.v. GLAM; cf. also chwiniol "whirling" from chwinio "to whirl", LR:388 s.v. SWIN. In mature Sindarin, as opposed to the "Noldorin" of the Etymologies, we should probably read hw- for chw-.) It seems that the adjectival participles so derived do not have an explicit plural form, as most other adjectives do.

# There is also another active participle, that may be termed the perfective active participle. In meaning it is similar to the normal active participle in -ol described above, except that it does not describe the state of someone (something) that is carrying out the action of the verb; it describes the state of someone already having carried out this action. It seems to have the ending -iel, displacing the final -a of the stem (or in the case of verbs in -ia, this whole ending):

esta- "call, name" > estiel "having named"
hwinia- "whirl" > hwíniel "having whirled"
linna- "sing" > linniel "having sung"

In the case of the numerous verbs in -ia, parallel forms suggest that the stem-vowel should be lengthened, as in hwíniel from hwinia- above. (The verbs siria- "flow", thilia- "glister" and tiria- "watch" would presumably behave in the same way: síriel, thíliel, tíriel.) However, this has somewhat complicated consequences. If we dare to trust the phonological system we glimpse in Tolkien's works, we must often take into account what the original vowel in these verbs were.

Where the original, primitive root or stem had the vowel A, the perfective participle would show ó (representing long á, since earlier long á had become ó in Sindarin):

beria- "protect" (stem BAR) > bóriel "having protected"
gweria- "betray, cheat" (stem WAR) > gwóriel "having cheated"
henia- "understand" (stem KHAN) > hóniel "having understood"
pelia- "spread" (stem PAL) > póliel "having spread"
penia- "fix, set" (stem PAN) > póniel "having fixed, having set"
renia- "stray" (stem RAN) > róniel "having strayed"
revia- "fly, sail" (stem RAM) > róviel "having flied, having sailed"
telia- "play" (stem TYAL) > tóliel "having played"

Notice especially egleria- "glorify" (related to aglar "glory"), that may have the perfective participle aglóriel "having glorified".

Where the original stem had the vowel O or U, the perfective participle would show ú (representing long ó, since earlier long ó had become ú in Sindarin):

delia- "hide, conceal" (stem DUL) > dúliel "having hidden, having concealed"
elia- "rain" (stem ULU) > úliel "having rained"
eria- "raise" (stem ORO) > úriel "having raised"
heria- "begin suddenly" (stem KHOR) > húriel "having suddenly begun"

# (In archaic Sindarin, it was easier to keep this category apart from the one above, since these verbs earlier showed ö instead of e: dölia- etc. After ö became e, these verbs must be memorized.) The verb bronia- "endure" (stem BORÓN-) would likewise yield brúniel "having endured". Indeed it is a mystery why bronia- does not appear as *brenia-, archaic *brönia-; in all comparable cases, the ending -ia causes umlaut (cf. for instance delia-, older dölia-, from *duljâ- or later *dolja-).

Other derived verbs than the ones in -ia may show simple umlaut when the ending -iel is added (we cannot be sure of this). If so, the vowels a and o both become e (again, o became ö in archaic Sindarin, ö later merging with e):

awartha- "abandon" > ewerthiel "having abandoned"
banga- "trade" > bengiel "having traded"
dortha- "stay" > derthiel "having stayed" (archaic dörthiel)
edonna- "beget" > edenniel "having begotten" (archaic edönniel)

Verbal stem with the vowels e or i would not be affected by the umlaut:

critha- "reap" > crithiel "having reaped"
ertha- "unite" > erthiel "having united"

Verbs with a diphthong (ei, ui, ae, au etc.) would not change, either:

eitha- "insult" > pl. eithiel "having insulted"
gruitha- "terrify" > pl. gruithiel "having terrified"
maetha- "fight" > pl. maethiel "having fought"
baugla- "oppress" > pl. baugliel "having oppressed"

# The last of the participial forms we know anything about is the passive participle (or past participle), an adjective describing the condition of something or someone that is (or has been) exposed to the action of the corresponding verb: If someone sees you, you are seen; "seen" is therefore the passive participle of the verb "to see". "Seen" is actually irregular; in most cases, English forms its passive participles by means of the ending -ed (e.g. killed from kill). Sindarin normally forms its past participles with the adjectival ending -en added to the (3rd person singular) past tense. Since derived verbs form their past tenses in -nt, the corresponding passive participles end in -nnen representing -nten (Sindarin phonology demanding that the cluster nt becomes nn between vowels):

gosta- "fear exceedingly" > gostannen "feared, dreaded" (cf. gostant as the past tense of the verb)
egleria- "glorify, praise" > egleriannen "glorified"
eitha- "insult" > eithannen "insulted"
esta- "call, name" > estannen "called, named"
ertha- "unite" > erthannen "united"
gruitha- "terrify" > gruithannen "terrified"
harna- "wound" > harnannen "wounded"
maetha- "fight" > maethannen "fought"
baugla- "oppress" > bauglannen "oppressed"

As the past participle of linna- "sing" we might expect linnannen ("sung"), but as in other cases where "double nn" would occur, the form is probably simply contracted: linnen.

In form, the past participles coincide with the 1st person past tense: gostannen could also mean "I feared", egleriannen is also "I glorified" etc. The context must decide how the form is to be understood.

In some cases, where the corresponding verb is intransitive (sc. when it cannot normally take a direct object, e.g. "go"), the past participle may describe the state that the one performing the verbal action isin having completed it. For instance, one who goes will thereafter be gone ("gone" is the past participle of "go"). In a similar manner, the past participle of an intransitive verb like lacha- "flame" (lachannen) may perhaps be used to describe a fire having flamed. But in Sindarin, it may be better to use the perfective active participle instead (like lechiel in this case); see above.

Unlike the active participles (we think), the past or passive participle has a distinct plural form (used when the participle describes a plural word). This is formed by altering the ending -nnen to -nnin, combined with I-umlaut throughout the word. As usual, the effect of this is that the vowels a and o, where they occur, are altered to e (but again, e from o was actually ö in archaic Sindarin):

harnannen "wounded" > pl. hernennin
gostannen "feared, dreaded" > pl. gestennin (archaic göstennin)

Notice that the ending -a in the verbal stem itself, here the final -a of harna and gosta-, is also umlauted to e: In the plural, -annen always becomes -ennin.

The vowels e and i are not affected by the umlaut:

linnen "sung" > pl. linnin
erthannen "united" > pl. erthennin

Again, neither are various diphthongs (ei, ae, ui, au etc.):

eithannen "insulted" > pl. eithennin
gruithannen "terrified" > pl. gruithennin
maethannen "fought" > pl. maethennin
bauglannen "oppressed" > pl. bauglennin

For a similar reason, it may be that the plural past participle of the verb boda- "ban, prohibit" should be bodennin, NOT **bedennin with umlaut o > e, since this o represents an older diphthong au (compare the related word baw! "no! don't!")

# The last form of the verb that we know anything about is the gerund, actually a derived noun, the verbal action considered as a "thing". In English, gerunds are derived by means of the ending -ing, e.g. thinking from the verb think. In English, the gerund and the active participle cannot be distinguished by their form; both end in -ing. However, while the participle has an adjectival function, the gerund is a noun, and in Sindarin, the two are distinct also in form. All derived verbs, or A-stems, form their gerunds by means of the ending -d:

bronia- "endure" > broniad "enduring" (= the act of enduring, endurance)
nara- "tell" > narad "telling" (as in "the telling of a tale")
ertha- "unite" > erthad "uniting" (= union, as abstract)

(Cf. the Mereth Aderthad, Feast of Reunion, mentioned in the Silmarillion.)

It seems that gerunds are often used where English would have an infinitive instead. In the King's Letter (SD:129), Aragorn writes that he aníra...suilannad mhellyn în = "wishes...to greet his friends", literally "wishes greeting (of) his friends". It is indeed possible that Tolkien had decided to drop the infinitives in -o and -i (see below concerning the latter), replacing them with gerunds. The infinitives in -o and -i are not attested in any sources later than the Etymologies. This may not mean much, since our post-Etym evidence is very scanty, but I would generally use gerunds for English infinitives when writing in Sindarin.

NOTE: As mentioned above, the object of a sentence undergoes lenition (soft mutation). It should be noted that in the phrase aníra...suilannad mhellyn în = "wishes...to greet his friends" or literally "wishes...greeting (of) his friends", the object from a grammatical point of view would undoubtedly be the suilannad or "greeting". However, the logical object is mellyn "friends", and this is the word that is lenited (to mhellyn). The gerund suilannad is not lenited (to *huilannad). This strongly suggests that the gerund is here perceived as an infinitive, not as a noun that could be lenited as the object of a sentence; the lenition affects the logical object "friends" instead.

II. BASIC VERBS

# The conjugation of the basic, ending-less verbs (aka primary verbs) is somewhat more complex than that of the derived verbs. Tolkien may have thought of this class as the "strong" verbs; cf. WJ:415.

# The infinitive is formed with the ending -i (not -o as in the case of the A-stems above):

fir- "fade, die" > firi "to fade, to die"
gir- "shudder" > giri "to shudder"
ped- "speak" > pedi "to speak"
pel- "wither" > peli "to wither"
redh- "sow" > redhi "to sow"

This ending causes the vowels a and o to umlaut to e:

blab- "flap" > blebi "to flap"
dag- "slay" > degi "to slay"
dar- "stop, halt" > deri "to stop, to halt"
nor- "run" > neri "to run" (archaic nöri)
tol- "come" > teli "to come" (archaic töli)
tog- "lead" > tegi "to lead" (archaic tögi)

Some verbs inevitably coincide in the infinitive; for instance, can- "call, shout" and cen- "see" would both have the infinitive ceni. The context must decide which verb is intended. (But as suggested above, Sindarin would often use the gerund where English has an infinitive, and here the distinction is preserved: caned "shouting", but cened "seeing".)

# The present tense of these verbs is formed in two different ways. The third person singular, that requires no further ending, is the same as the verbal stem, but in the case of monosyllabic verbal stems, the vowel becomes long:

dar- "stop" > dâr "(he, she, it) stops"
fir- "fade, die" > fîr "fades, dies"
ped- "speak" > pêd "speaks"
tol- "come" > tôl "comes"

(These may also cover the English compound tenses: "is stopping", "is fading" etc., but we cannot be sure; see Note (i) below.) Attested examples include blâb as the present tense of blebi- "to flap" (LR:380 s.v. PALAP), and - with a clearer wording - the entry TUL- in LR:395, where tôl is translated "he comes", thus being clearly identified as the 3rd person singular of teli "to come". That the form itself is simply 3rd person and not necessarily "masculine" or even animate ("he comes") is apparent from another attestation, the sentence tôl acharn "vengeance comes" (WJ:254; according to WJ:301 Tolkien later wrote tûl acharn instead, but accepting this change would cause such an upheaval in the verbal system and the phonology that it is probably best ignored at this point). Acharn "vengeance" would not normally be referred to with the pronoun "he".

NOTE (i): Pêd as the present tense "speaks" is also attested (incidentally in lenited form: bêd) in VT41:11, where it is seen to correspond to the Quenya aorist quete. Whether Sindarin has an aorist tense distinct from the present tense is unclear; if so, forms like pêd are probably aorists: "speaks" as opposed to present tense "is speaking".

NOTE (ii): When final, v is spelt f. Therefore, the 3rd person singular present tense of lav- "lick" is lâf. In other forms, where the v is not final, it would also be spelt v (e.g. levin "I lick" - cf. below).

In the case of polysyllabic basic verbal stems (usually verbs with some prepositional element prefixed), there is no lengthening of the vowel, and the 3rd person singular present tense is identical to the verbal stem itself:

osgar- "cut around, amputate" > osgar "cuts around, amputates" (this form is explicitly mentioned in LR:379 s.v. OS)

In all present-tense forms except the 3rd person singular, some ending is required, as outlined initially. These endings are added to a form of the verb that is identical to the infinitive, hence with the ending -i and umlaut where the verbal stem has the vowel a or o (while i and e are not affected in any way):

dar- "stop, halt" > derin "I stop, halt", derir "(they) stop, halt" (with multiple subjects, e.g. in Edhil derir "the Elves halt"), derig/derich "you stop", derim "we stop"
fir- "fade, die" > firin "I die" etc. with the various endings
ped- "speak" > pedin "I speak" etc.
tol- "come" > telin "I come" etc.
osgar- "amputate" > esgerin "I amputate" etc.

This form has long been thought of as the perfect tense, which was also the view presented in earlier versions of this article. This was primarily because of Gilraen's linnod in LotR Appendix A: Onen i-Estel Edain, ú-chebin estel anim, translated in a footnote as "I gave Hope to the Dúnedain, I have kept no hope for myself" (emphasis added). However, in light of other examples, it may be best to see ú-chebin as a present-tense form (and translate "I do not keep [any] hope for myself"), assuming that Tolkien's perfect-tense translation "I have kept no hope for myself" is slightly free and makes a concession to natural English. (It used to be unclear what the basic form of ú-chebin is; removing the negative prefix ú- "not" and the soft mutation h > ch that it triggers, we are left with hebin "I keep". This could come from hab-, heb- or hob- "keep", the umlaut neutralizing the vowels in the form hebin. However, the stem KHEP "retain, keep" published in VT41:6 must be the root underlying this verb; hence the basic form is evidently heb-.)

# The past tense of basic verbs involves a nasal suffix or infix, though it is sometimes assimilated beyond recognition. We will first concentrate on the 3rd person singular forms, since the other forms,in turn, may be derived from them.

In the case of basic verbs in -r, an -n is simply suffixed to the stem (a remnant of a longer past tense ending -ne, still current in Quenya):

dar- "stop, halt" > darn "stopped, halted"
gir- "shudder" > girn "shuddered"
nor- "run" > norn "ran"

Verbal stems in -n probably behave in the same way (cen- "see" > cenn "saw"). As for verbs in -l, the ending -n is probably assimilated to it (pel- "wither" > pell "withered"). We lack examples, but extrapolations from Quenya would point in this direction.

When it comes to verbal stems ending in -b, -d, -g, -v, -dh, the nasal element denoting past tense would manifest as an infix instead of as a prefix. That is, it is not added to the final consonant of the stem, but inserted before it. This has some consequences that might surprise students not familiar with the evolution of Eldarin.In Sindarin, b, d, g, v, dh following a vowel descend from earlier p, t, c, b (or m) and d, respectively. But where the nasal infix intruded between the vowel and the consonant, this change could not take place: The infix "shielded" the consonant from the vowel that would otherwise cause it to change. Hence b, d, g seemingly reverts to p, t, c following the infix. Actually they do not revert; they simply never changed:

had- "hurl" > pa.t. hant "hurled" (original stem KHAT; this past tense is actually listed in LR:363)
cab- "jump" > pa.t. camp "jumped" (original stem KAP)
dag- "slay" > pa.t. danc "slew" (original stem NDAK; Sindarin c = k).

(It will be observed that the nasal infix, that most often manifests as n, is assimilated to m before p.) Presumably dh from earlier d likewise reverts to its original quality:

redh- "sow" > pa.t. rend "sowed" (stem RED)

One attested case is gwend (or gwenn) as the past tense of a verb gwedhi "to bind" (LR:397 s.v. WED-, where the infinitive is given as "gwedi", but this is surely a misreading for gweði = gwedhi; compare the related word angweð = angwedh). However, Tolkien noted that gwend was later replaced by gwedhant (spelt gweðant in LR), as if this were a derived verb *gwedha-; perhaps the past tenses in -nd were somehow disliked by the Elves (/by Tolkien). It may be that the past tense rend "sowed" (not directly attested in Tolkien's papers) was likewise replaced by redhant in later Sindarin.

Verbs of more than one syllable would have past tenses in -nn instead of -nd, if we dare to trust our reconstructed Sindarin phonology. There are only two such verbs known: neledh- "go in, enter" (pa.t. nelenn?) and edledh- "go into exile" (pa.t. edlenn?). The latter verb is not directly attested, but is reconstructed from "Noldorin" egledh- (LR:368 s.v. LED).

Verbs with final -v may also be slightly special. In most cases, post-vocalic v would come from earlier b, so certainly these verbs at one point ended in -mb (the nasal infix manifesting as m before b, just as before p). But final mb became simple m in Sindarin. (Cf. WJ:394, where Tolkien states that primitive *lambê "tongue" became lam in Sindarin, surely representing earlier *lamb. Compare the "Noldorin" form lham(b) in LR:367 s.v. LAB, that would correspond to Sindarin lam(b).) Hence, basic verbs in -v may have past tenses in -m, for -mb:

lav- "lick" > lam (for lamb) "licked" (the noun lam "tongue" is related and shares precisely the same phonological history)

As mentioned above, the forms so far derived are 3rd person singulars. Other forms are quite easily derived from them by means of the same endings that were mentioned above: -n "I", -m "we", -r "they" or just plurality, etc. The question is, what connecting vowel do we add between the verb and the ending? In terms of phonological history, we would definitely expect e: The Sindarin form corresponding to Quenya quenten "I said" would be expected to be *pennen. However, our one-and-only example points in a different direction, and this is one of the cases where we must generalize from one single form, with great consequences for an entire class of verbs. I would have liked to have other (and in particular later) examples, to make sure that this was not just a passing whim in Tolkien's evolution of "Noldorin"/Sindarin, or indeed a misreading on Christopher Tolkien's part.

The example in question is found in the Etymologies, LR:363, stem KHAT "hurl". Here we have a verb hedi, clearly the perfectly regular infinitive of had-, but then two forms explicitly identified as "pa.t." are listed: hennin and hant. The latter is transparently the 3rd person singular, "(he/she/it) hurled", formed from had- with a nasal infix according to the rules we have tried to sketch (indeed using this example). But hennin, with the ending -n that is known to mean "I", must be the 1st person past tense: "I hurled". The change nt > intervocalic nn is what we would expect on phonological grounds, but it is surprising that i is used as the connecting vowel before the pronominal ending is added. It would be tempting to dismiss hennin as an error for hannen, but the umlaut a > e is exactly what we would expect when there is an i following in the next syllable. We do know cases of confusion a/e and e/i in the texts produced by various editors trying to decipher Tolkien's handwriting, but to assume that Christopher Tolkien managed to misread two vowelsin the same word, and that the result just happened to beautifully comply with Sindarin phonology, may be assuming too much. It may be that JRRT imagined that forms like hannen had been reformed on analogy with the corresponding present-tense forms (in this case hedin "I hurl"), the connecting vowel i and therefore also umlaut being introduced in the past tense as well as the present: hannen > hennin.

Accepting this example, we must formulate this rule: All past tense forms of the basic verbs, except for the 3rd person singular, are formed by adding -i- and the appropriate ending to the 3rd person singular itself:

#

gir- "shudder" > 3 pers. sg. pa.t. girn "(he, she, it) shuddered" > girnin "I shuddered", girnim "we shuddered", girnig/girnich "you shuddered", girnir "(they) shuddered"

cen- "see" > 3 pers. sg. pa.t. cenn "(he, she, it) saw" > cennin "I saw" (etc. with the various endings)

As the example hant > hennin indicates, the connecting vowel i triggers the normal umlauts in the syllable before it, a and o both becoming e:

dar- "stop, halt" > 3 pers. sg. pa.t. darn "(he, she, it) halted" > dernin "I halted" (etc.)
nor- "run" > 3 pers. sg. pa.t. norn "(he, she, it) ran" > nernin (archaic nörnin) "I ran" (etc.)

The example hant > hennin also illustrates another phenomenon: Not all the final consonant clusters occurring in the past tense can remain unchanged when they are no longer final at all, but have become intervocalic because an ending has been added. The clusters -nt, -nc, -mp become -nn-, -ng-, -mm- instead:

ped- "speak" > 3 pers. sg. pa.t. pent "(he, she, it) spoke" > pennin "I spoke" (etc.)
dag- "slay" > 3 pers. sg. pa.t. danc "(he, she, it) slew" > dengin "I slew" (etc.)
cab- "jump" > 3 pers. sg. pa.t. camp "(he, she, it) jumped" > cemmin "I jump" (etc.)

The cluster nd, like nt, would become nn intervocalically:

gwedh- "bind" > 3 pers. pa.t. gwend "(he, she, it) bound" > gwennin "I bound" (etc.)

Final m (usually representing mb) would become double -mm-:

lav- "lick" > 3 pers. pa.t. lam "(he, she, it) licked" > lemmin "I licked" (etc.)

# As for the future tense, we must assume that the ending -tha is relevant also for this class of verbs, but obviously some connecting vowel is required. While we have no direct examples, analogy with other forms suggests that an i would be inserted before this ending. This i would cause the normal umlauts where appropriate. In short, the future tense of a verb of this class may be constructed by adding -tha to the infinitive form (see rules above):

dar- "halt" > inf. deri "to halt" > future deritha "will halt"
ped- "speak" > inf. pedi "to speak" > future peditha "will speak"
gir- "shudder" > inf. giri "to shudder" > future giritha "will shudder"
tol- "come" > inf. teli "to come" (archaic töli) > future telitha (archaic tölitha) "will come"

These (3rd person singular) future-tense forms may then be further modified with the normal endings, just as in the case of the derived verbs: telithon "I will come", telitham "we will come", plural telithar "(they) will come" etc. (As usual, -a becomes -o before the ending -n for "I", hence telithon rather than **telithan.)

# The imperative of the basic verbs is easily formed with the ending -o:

dar- "halt" > daro "halt!"
ped- "speak" > pedo "speak!"
tir- "watch, look" > tiro "watch! look!"
tol- "come" > tolo "come!"

Three of these are attested in LotR: An Elf halted the Fellowship with the command daro! when they were entering Lórien. Pedo "speak, say" is found in the Moria gate inscription (pedo mellon, which should be translated "say friend", though Gandalf at first took it to mean "speak, friend"). Sam speaking in tongues in Cirith Ungol used the phrase a tiro nin, Fanuilos! "o look towards me, Everwhite!" (a title of Varda); see Letters:278 or RGEO:72 for translation.

# The active participle of this class of verbs would probably take the ending -el (for older *-ila):

dar- "halt" > darel "halting"
ped- "speak" > pedel "speaking"
tol- "come" > tolel "coming"

However, where the stem vowel is i, this ending seems to be expanded to -iel (it may be that this only occurs following the consonants n, l, and r, and that the extra i materializing after them reflects their being palatalized by the vowel i occurring before them at an older stage of the language):

fir- "die, fade" > firiel "dying, fading"
gir- "shudder" > giriel "shuddering"
glir- "sing" (also "recite poem") > gliriel "singing" (or, "reciting")
tir- "watch" > tiriel "watching" (the only attested example - see below)

# The perfective active participle seems to have the ending -iel combined with lengthening of the stem-vowel. The vowel i would simply become long í:

fir- "fade, die" > fíriel "having died, having faded" (or simply "dead, faded")
glir- "sing" (/"recite") > glíriel "having sung" (/"having recited")
tir- "watch" > tíriel "having watched".

(It will be noticed that vowel-length alone distinguishes tiriel "watching" from tíriel "having watched". Compare RGEO:73, where Tolkien explains that while palan-diriel means "gazing far away", palan-díriel has a perfective meaning: "having gazed far away". In these words, -diriel/-díriel are simply lenited forms of -tiriel/-tíriel.)

This lengthening of vowels probably occurred so early that the subsequent changes affecting long vowels must also be taken into consideration. Earlier é, á, ó would be expected to manifest as í, ó, ú, respectively - reflecting a change that took place at the Old Sindarin stage:

mad- "eat" > módiel (for mádiel) "having eaten"
ped- "speak" > pídiel (for pédiel) "having spoken"
nor- "run" > núriel (for nóriel) "having ran"

It seems that neither of the active participles so derived (in -el and -iel) have distinct plural forms.

# The passive participle (or past participle) of this class of verbs can be constructed by adding -en to the 3rd person singular past tense form (see rules above):

dar- "stop" > 3 pers. sg. pa.t. darn "(he, she, it) stopped" > passive participle darnen "stopped, halted"
sol- "close" > 3 pers. sg. pa.t. soll "(he, she, it) closed" > passive participle sollen "closed" (the latter form being all that is attested of this verb: LotR refers to the Fen Hollen or "Closed Door" in Minas Tirith, hollen presumably being a lenited form of sollen)
tir- "watch, guard" > 3 pers. sg. pa.t. tirn "(he, she, it) watched, guarded" > passive participle tirnen "watched, guarded"

(The latter is attested in the Silmarillion, in the name Talath Dirnen "Guarded Plain": Dirnen is the lenited form of tirnen.)

Again, when another vowel comes to follow them, final -nt, -nc, -mp, -nd, -m become -nn-, -ng-, -mm-, -nn-, -mm-, respectively:

ped- "speak" > 3 pers. sg. pa.t. pent "(he, she, it) spoke" > passive participle pennen "spoken"
dag- "slay" > 3 pers. sg. pa.t. danc "(he, she, it) slew" > pass. part. dangen "slain" (attested in LR:375 s.v. NDAK)
hab- "clothe" > 3 pers. sg. pa.t. hamp "(he, she, it) clothed" > pass. part. hammen "clothed"
redh- "sow" > 3 pers. sg. pa.t. rend "(he, she, it) sowed" > pass. part. rennen "sowed"
lav- "lick" > 3 pers. sg. pa.t. lam "(he, she, it) licked" > pass. part. lammen "licked"

These passive participles in -en would have plural forms in -in, causing the normal umlauts: a and o both become e:

dangen "slain" > pl. dengin
hollen "closed" > pl. hellin (archaic höllin)

(Compare the Haudh-en-Ndengin or "Hill of Slain" mentioned in the Silmarillion; ndengin is a form of dengin.) As usual, the vowels e and i would not be affected in any way:

pennen "spoken" > pl. pennin
tirnen "guarded" > pl. tirnin

# Finally we have the gerund, the verbal noun, that may also be used to translate English infinitives (see above). The gerunds of basic verbs are easily formed with the ending -ed:

cab- "jump" > cabed "jumping" (as noun, = "a jump, a leap")
cen- "see, look" > cened "looking"
glir- "sing" > glired "singing"
tol- "come" > toled "coming"

The Sindarin verbs cab- "jump, leap" and cen- "see, look" are actually attested as gerunds only! According to the Silmarillion, the gorge where Túrin slew Glaurung was called Cabed-en-Aras or "Deer's Leap" ("Jumping-of the-Deer"). The verb cab- is obviously to be referred to the stem KAP "leap" listed in the Etymologies (LR.362), but it is not mentioned there. Cened "looking" occurs as part of the compound cenedril "looking-glass" in RS:466.

III. THE MIXED CONJUGATION

Some verbs that by their form would appear to be A-stems in effect sit on the fence between the two conjugations outlined above. An example is the verb drava- "hew". In most forms it is probably a well-behaved A-stem: infinitive dravo, present tense drava, future tense dravatha, imperative dravo, active participle dravol, gerund dravad. But in the past tense we would expect the form *dravant, which does not occur. In Tolkien's notes as reproduced in LR:354 s.v. DARÁM, he gave the 1st person past tense as drammen ("I hewed"), pointing to a 3rd person form dram ("he, she, it hewed"). (There was also an irregular past tense dramp, that we need not concern ourselves with here - see part IV below.) A past tense dram is precisely what we would expect if this were a basic verb, with stem drav- (infinitive **drevi) instead of drava- (inf. dravo). Another example is the verb nara- "tell" (infinitive naro, LR:374 s.v. NAR2); the Old Sindarin ("ON") past tense is given as narne, implying that the Sindarin past tense would be narn rather than **narant. In short: A number of A-stems form their (3rd person sg.) past tense as if the final vowel did not exist; the past tense is formed according to the rules of the basic verbs instead. Our few examples suggest that this group includes most verbs with a single consonant before the final -a, as long as this consonant is not th or ch (representing earlier consonant clusters). Ignoring the final vowel and employing the same rules as for the basic verbs, we would arrive at past tense forms like the following:

ava- "will not" > am "would not"
brona- "last, survive" > bronn "lasted, survived"
drava- "hew" > dram "hewed"
fara- "hunt" > farn "hunted"
gala- "grow" > gall "grew"
groga- "feel terror" > grunc "felt terror"
laba- "hop" > lamp "hopped"
loda- "float" > lunt "floated"
nara- "tell" > narn "told"
pada- "walk (on a track or path)" > pant "walked"
rada- "make a way, find a way" > rant "made a way, found a way"
soga- "drink" > sunc "drank"
toba- "cover, roof over" > tump "covered, roofed over"

(Concerning the shift o > u in groga-, loda-, soga-, toba- > pa.t. grunc, lunt, sunc, tump, see section IV below.)

A number of three-syllable verbal stems in -da must also be assigned to the mixed conjugation: aphada- "follow", athrada- "traverse", gannada- "(play a) harp", lathrada- "eavesdrop", limmida- "moisten", nimmida- "whiten" and tangada- "make firm": past tenses aphant, athrant, gannant, lathrant, limmint, nimmint, tangant, or with endings aphanne- etc. (The "Noldorin" past tense lhimmint, that would correspond to Sindarin limmint, is mentioned by Tolkien in LR:369 s.v. LINKWI.)

Long vowels would probably be shortened before the consonant cluster arising in the past tense:

aníra- "wish" > anirn "wished"
síla- "shine" > sill "shone"
tíra- "watch" > tirn "watched"

When further endings are to be added (to produce forms other than the 3rd person singular), the connecting vowel is here e, as the example drammen "I hewed" demonstrates.

NOTE: Since these verbs might seem to jump over to the I-stems in the past tense, we might have expected the connecting vowel i as in hennin "I hurled", hence **dremmin "I hewed", but this is not the case. This might support the theory that the connecting vowel i in the past tense arose on analogy with its use in the present tense (hedin "I hurl"). The verb drava- does not have i in the present tense (drava "hews, is hewing"), and hence does not show i in the past tense, either. Instead we find e, like we would expect on phonological reasons alone: drammen.

As usual, final -m, -nc, -nt, -mp becomes -mm-, -ng-, -nn-, -mm- between vowels:

drava- "hew" > dram "(he, she, it) hewed" > drammen "I hewed", drammem "we hewed", drammeg/drammech "you heaved", drammer "(they) hewed"
laba- "hop" > lamp "hopped" > lammen "I hopped" (etc. with the various endings)
loda- "float" > lunt "floated" > lunnen "I floated" (etc.)
soga- "drink" > sunc "drank" > sungen "I drank" (etc.)

The passive participle would be derived with the ending -en, just as in the case of normal basic verbs. Thus, as usual, the past participle is identical to the 1st person singular form, hence drammen could also be "hewed" as a participle, sungen is also "drunk" etc. These participles would have plural forms in -in (causing umlaut), in other words behaving just like the passive participles of normal basic verbs. See rules in section II above. (The umlaut product of u, where it occurs, would be y. Hence the plural form of sungen would be syngin.)

As noted above, these verbs probably have active participles in -ol, like normal A-stems (drava- "hew" > dravol "hewing"). The perfective active participle would presumably be formed according to the rules of the I-stems, as if the final vowel did not exist. Hence we would see the ending -iel combined with lengthening of the stem-vowel, í, ó, ú representing í, á, ó (drava- "hew" > dróviel "having hewed", soga- "drink" > súgiel "having drunk"). If the vowel is long already, we must assume that it simply stays long (síla- "shine" > síliel "having shone").

IV. IRREGULAR AND SPECIAL VERBS

# By adhering to the rules above, one would get most of the verbs right - assuming that the Sindarin verbal system has been correctly reconstructed. We are left with a happily quite small number of special cases. Some of them are readily explicablein terms of the phonological evolution Tolkien imagined, some may reflect the shifting whims of the Language-maker, some are downright weird - hopefully not implying that our reconstruction of the Sindarin verbal system is flawed with major shortcomings and failures to second-guess Tolkien's intentions.

# Original U surviving before a nasal: At one point in the evolution of Sindarin, original u in very many cases became o. For instance, the verb soga- "drink" comes from a stem SUK. However, this change did not occur before a nasal, like n or m. So if a verbal stem containing the vowel o < u has a past tense form involving nasal infixion, the original quality of the vowel would persist before this consonant. Hence, Tolkien noted that the 3rd person sg. past tense of soga- is sunc (LR:388 s.v. SUK, though the form sogant was also valid, the verb being transferred to the normal A-stem class). Other likely cases of the same phenomenon (not directly attestedin Tolkien's papers):

groga- "feel terror" > 3 pers. sg. pa.t. grunc (original stem RUK)
loda- "float" > pa.t. lunt (stem LUT)
nod- "tie, bind" > pa.t. nunt (stem NUT)
toba- "cover, roof over" > pa.t. tump (stem *TUP)
tog- "lead, bring" > pa.t. tunc (stem TUK)

NOTE: In the Etymologies, LR:378 s.v. NOT, the verb nod- is given as "nud-", but this would contradict everything we think we know about Sindarin phonology. The verb toba- [inf. tobo] is derived from a stem TOP in LR:379, in which case the past tense would be tomp, but the Quenya verb untúpa "covers" in Namárië in LotR suggests that Tolkien had decided that the stem was TUP instead, though a distinct stem TUP occurs in Etym.

Grunc, lunt, sunc and tump would appear as grunge-, lunne-, sunge-, tumme- before the normal plural/pronominal endings - grunger "(they) felt fear", grungen "I felt fear" etc. If the example hant > hennin (LR:363 s.v. KHAT) holds, we would in the case of nunt and tunc see the connective vowel i before the normal endings are added. This i would trigger umlaut u > y, so (with the normal change of intervocalic nt, nc to nn, ng) we would have for instance 1st person sg. nynnin "I tied" and tyngin "I led, I brought". (But groga-, loda-, toba- would belong to the mixed conjugation, with e rather than i as the connecting vowel, and hence no umlaut either: grungen "I felt fear", lunnen "I floated", tummen "I covered".)

The past participles of all the verbs we are dealing with here can be formed, quite regularly, by adding -en to the 3 sg. past tense (with the normal changes in final consonant groups when they become intervocalic instead):

groga- "feel terror" > pa.t. grunc > passive participle grungen
loda- "float" > pa.t. lunt > passive participle lunnen
nod- "tie, bind" > pa.t. nunt > passive participle nunnen
soga- "drink" > pa.t. sunc > passive participle sungen
toba- "cover, roof over" > pa.t. tump > passive participle tummen
tog- "lead, bring" > pa.t. tunc > passive participle tungen

And again, we would see umlaut u > y in the plural forms of these participles: gryngin, lynnin, nynnin, tymmin, syngin, tyngin. (A few of these verbs, "feel terror" and "float", may not normally have passive participles, though - since they are intransitive.)

But in the case of groga-, loda-, soga- and toba-, it may also be permissible to take the easier path and simply let them go as A-stems (Tolkien made an explicit note to this effect in the case of soga-). Hence we would have (3 pers. sg) past tenses grogant, lodant, sogant, tobant (-nt regularly becoming -nne- before endings), and past participles grogannen, lodannen, sogannen, tobannen (pl. gregennin, ledennin, segennin, tebennin - or archaic grögennin etc.)

# Impersonal verbs:A "Noldorin" impersonal verb bui "it is necessary, one must, one is compelled to" appears in the Etymologies; in Sindarin it would have to become boe. We have no examples, but it can probably be used in such contexts as boe maethad in Yrch "it is necessary to fight the Orcs". (English "must" may be expressed as "it is necessary for X to do Y": Boe 'nin Edhil maethad in Yrch "it is necessary for the Elves to fight the Orcs" = "the Elves must fight the Orcs"; boe anim baded "it is necessary for me to go" = "I must go".) Perhaps boe has no inflected forms for tense etc.; nothing of the kind is suggested in the Etymologies.

Another impersonal verb is elia- "rain". The "Noldorin" impersonal form expressing "it rains", namely oeil [= öil], later eil, is given in the Etymologies (LR:396 s.v. ULU). In Third Age Sindarin, the form would be ail. The past tense, denoting "it rained", could be aul or regular eliant. We may conjugate the verb like this: infinitive elio "to rain", present tense ail = impersonal 3 sg. form "it rains", past tense eliant or aul = impersonal 3 sg "it rained", future eliatha = "it will rain", imperative elio "rain!", participle eliol "raining" (perfective úliel "having rained"), gerund eliad "raining". A verb with this meaning would hardly have any passive participles. The form would be eliannen, or, if we use aul as the past tense, olen.

# Various irregular verbs: In Gilraen's linnod, the word onen (or in some editions of LotR, ónen) is used for "I gave". This would seem to be the irregular past tense of the verb anna- "give" listed in the Etymologies (if the verb were regular, the past tense should be **annant, with annannen > annen for "I gave"). Onen points to a 3rd person singular form aun "(he, she, it) gave", that could be regularly derived from an older past tense áne (compare óne as the past tense of the Quenya verb onta- "beget", LR:379 s.v. ONO; given the fact that Sindarin anna- corresponds to Quenya anta-, it is not implausible to assume that a past tense áne may have existed at some stage, and it is actually attested in QL:31). Aun becomes one- when an ending is to be added, the diphthong au being monophthongized to o. Suggested conjugation of anna- "give": Infinitive anno "to give", present tense anna "gives, is giving", irregular past tense: 3 sg. aun "gave" with endings one- (onen "I gave", onem "we gave" etc.), future annatha "will give", imperative anno "give!", active participle annol "giving", perfective participle óniel "having given", past participle onen (pl. onin) "given", gerund annad "giving". Notice that there is no umlaut in the plural form of the past participle (onin, not **enin for archaic **önin), because o from au is not umlauted like this.

In LR:375 s.v. NDAM, a verb damna- "to hammer" is listed, with a (3rd person sg.) past tense dammint. Both forms are positively weird. There can be no doubt that damna is a misreading for damma-, the form we would expect on phonological grounds; cf. mm in the past tense. The past tense "dammint" is very strange. We would definitely expect dammant. Where does the i in the past tense come from in the first place, and if it is to be there at all, why does it not cause the a to umlaut to e (i.e. demmint)? If we accept this past tense form (with endings damminne-), we would also have to use damminnen pl. damminnin as the passive participle. But personally I am strongly inclined to dismiss dammint as a misreading for dammant, in which case the verb would be perfectly regular.

The verb drava- "hew" would regularly have the past tense dram (with endings dramme-). According to LR:354 s.v. DARÁM, an irregular (3rd person sg.) past tense dramp was used in poetry - as if the verb were **draba- instead. This form was apparently used in addition to, not instead of, the regular past tense. With endings, dramp and dram would both appear as dramme- anyway (e.g. the 1st person pa.t. drammen that is mentioned in this entry in the Etymologies).

#

As mentioned above, the regular past tense of the verb gwedh- "bind" is gwend (with endings gwenni-), but Tolkien indicated that an irregular past gwedhant (as if this were an A-stem **gwedha-) came into use "later". The regular past tense had come to be regarded as archaic or poetic. When the change occurred, it may be that the passive participle "bound" was also altered from gwennen to gwedhannen. Presumably, the verb was still inflected as a regular "primary" verb otherwise (infinitive gwedhi, present tense gwêdh or before endings gwedhi-, future gwedhitha, imperative gwedho, active participle gwedhel, perfective participle gwídhiel). Perhaps the verb redh- "sow" underwent a similar development, so that the regular past tense rend was replaced by redhant?

The verb soga- "drink" would regularly have the 3rd person singular soga "(he, she, it) drinks", but LR:388 indicates that the 3rd person sg. is actually sôg (as if this were a primary verb sog-). When endings are to be added to produce other forms than the 3rd person sg., we may use the regular present-tense stem soga- (hence sogon [for **sogan] "I drink", sogam "we drink" etc.) The (3rd person sg.) past tense is either regular sunc (with endings sunge-) or irregular sogant (with endings soganne-); Tolkien indicated that both are valid. The passive participle "drunk" would then be either sogannen (pl. segennin) to go with the past tense sogant, or sungen (pl. syngin) if one prefers the past tense sunc. Hopefully, the verb soga- "drink" is otherwise a normal, well-behaved Mixed Conjugation verb, as the infinitive sogo (given in LR:388) would suggest. Hence future sogatha "will drink", imperative sogo "drink!", participle sogol "drinking" (perfective súgiel "having drunk"), gerund sogad "drinking" (as noun).

NOTE: The actual wording in LR:388 s.v. SUK is "N sogo, 3 sg. sôg, pa.t. sunc, asogant (sogennen)". Sogo is clearly the infinitive "to drink", sôg is identified as the 3rd person singular (present), and sunc is likewise identified as the (3rd person singular) past tense. However, asogant cannot be a correct reading of Tolkien's text. It is very difficult to understand where this a-prefix could come from, and moreover, such a prefix would in all likelihood cause soft mutation of the initial s, so that we would have the form **ahogant. What Tolkien actually wrote in his less-than-calligraphic handwriting must have been "sunc, or sogant", alternatively "sunc, and sogant" - a small doodle representing or or possibly and being misread as a by Christopher Tolkien, and prefixed directly to the following verb. The form sogennen must be the passive participle "drunk", but since the past participle is derived by suffixing -en to the past tense (nt regularly becoming nn between vowels), we must conclude that "sogennen" is a misreading for sogannen.

The verb thora- "fence" is stated to have the passive participle thoren "fenced" (LR:393 s.v. THUR). Thoren suggests a past tense thaur. The verb may go like this: Infinitive thoro "to fence", present tense thora "fences, is fencing", irregular (3rd person sg.) past tense thaur (with endings thore-, e.g. thoren "I fence, I am fencing"), future thoratha "will fence", imperative thoro "fence!", active participle thorol "fencing" (perfective thóriel "having fenced"), passive participle thoren "fenced" (pl. thorin), gerund thorad. Notice that the perfective participle is thóriel instead of **thúriel, and that there is no umlaut in the plural form of the participle thoren (pl. thorin, not **therin). As in the case of other verbs, this is because o, ó here represents the diphthong au.

The verb trenar- "recount, tell to the end" is stated to have the (3rd person singular) past tense trenor or trener (LR:374 s.v. NAR2). Regularly, we would expect **trenarn. The verb may go like this: Infinitive treneri "to recount", 3rd person present tense trenar "recounts, is recounting" (with endings treneri-, hence trenerin "I recount", trenirem "we recount" etc.), irregular past tense trenor or trener (with endings either trenori- or treneri-, hence trenorin "I recount" etc.; the alternative form trenerin would clash with the present tense), future treneritha "will recount", imperative trenaro "recount!", active participle trenarel "recounting" (perfective trenóriel "having recounted"), passive particle ?trenoren (plural trenorin) "recounted", gerund trenared "recounting". Notice the absence of umlaut in the form trenorin ("I recount" or the pl. form of the passive participle "recounted"). We would probably not find trenerin, since the o of trenorin may represent au (in turn derived from long á, a lengthened version of the vowel of the stem NAR2; trenor may reflect a primitive past tense *trenâr-).

In the entry MBAKH in the Etymologies (LR:372), we read: "Q[uenya] manka- trade; makar tradesman, mankale commerce. N[oldorin] banc, banga." What are we to make of this? Banga- is surely the "Noldorin" > Sindarin word corresponding to Quenya manka-, hence the verb "to trade". But what does banc mean? If banc is a form of banga-, it would most likely be an irregular 3rd person past tense: "(He/she) traded" (instead of regular bangant). Again assuming that the example hennin "I hurled" can be trusted, we would have bengi- before endings, e.g. bengin "I traded", bengir "(they) traded" etc. The passive participle would also be bangen (pl. bengin) rather than bangannen (pl. bengennin). But I will not rule out the possibility that banc is not intended as a form of the verb banga- at all; it could be a noun "trade", corresponding to (but not an exact cognate of) Quenya mankale.

# A possible revision of the system: One passage in the essay Quendi and Eldar from about 1960 suggests that Tolkien had carried out a major revision in one part of the Sindarin verbal system (WJ:415). Reference is made to

...a primitive past tense, marked as such by the 'augment' or reduplicated base-vowel, and the long stem-vowel. Past tenses of this form were usual in Sindarin 'strong' or primary verbs: as *akâra 'made, did' > S agor.

The new rules for the derivation of the past tense of primary verbs are quite easily reconstructed: The vowel occurring in the verb is prefixed, but in the verbal stem itself, a, e, o are altered to o, i, u, respectively (representing the "long stem-vowel" â, ê, ô, since the quality of such long vowels were changed in Old Sindarin). The vowel i would not change. The initial consonant would undergo soft mutation when a vowel is prefixed to it, p > b, t > d, c > g (hence agor from car-), b > v, d > dh, g > zero, m > v, s > h. (The consonants f, th would be unchanged.)

ped- "speak" > ebid "spoke"
tir- "watch" > idir "watched"
car- "do" > agor "made"
bad- "go" > avod "went"
dar- "halt" > adhor "halted"
gwedh- "bind" > ewidh (= e'widh) "bound".
mad- "eat" > avod "ate" (same as the pa.t. of bad-!)
nor- "run" > onur "ran"
sog- "drink" > ohug "drank"
fir- "die" > ifir "died"

Of course, this would contradict the earlier system glimpsed in the Etymologies, where, for instance, the past tense of gwedh- is explicitly given as gwend (or later gwedhant) instead of ewidh. Etym also has sunc and sogant rather than ohug for "drank". Moreover, pent instead of ebid for "spoke, said" is attested outside the Etymologies. We must await the publication of more material before we can determine to what extent Tolkien carried out this revision - whether this was really intended to be the new way of deriving the past tenses of primary verbs, fully obsoleting the earlier system that we have tried to reconstruct above. For the moment, I would accept agor as the past tense of car- "make, do", but otherwise largely continue using the "classical" system. Perhaps Tolkien's wording - that agor-type past tenses were "usual" rather than universal - implies that one could to some extent chose which way to form the past tense (it is clear from several texts that Tolkien imagined that there were many varieties and dialects of Sindarin). We may let car- "do" go like this: Infinitive ceri, present tense: 3 sg. câr "(he, she, it) does", with endings ceri- (cerin "I do", cerim "we do" etc.), irregular past tense agor "did" (before endings agore-, e.g. agoren "I did"), future ceritha "will do", imperative caro "do!", active participle carel "doing", perfective participle córiel "having done", passive participle coren (or carnen?) "done", gerund cared "doing".

# The question of umlaut in prefixes: A number of Sindarin verbs have a prefixed (typically prepositional) element. The question is, should the vowel in such prefixed elements be umlauted in the forms of the verb that require umlaut? Consider the verb aníra- "wish to, want to". This, as far as we can tell, consists of two elements: a verb íra- "wish" with the prefix an- "to", hence literally "to-want" = want to. The passive participle of this verb would probably be anirnen. But what about the plural? Anirnin or enirnin with umlaut throughout the word?

Tolkien's notes seem less than consistent. The verb osgar- "cut around, amputate" includes the prefixed element os- "around". The infinitive esgeri, listed in LR:379 s.v. OS, shows umlaut throughout the word (not *osgeri, the prefix being exempted from umlaut). On the other hand, the verb orthor- "master, conquer" (literally "over-power", with or- meaning "over") shows no umlaut in the infinitive, which is listed in LR:395 s.v. TUR as ortheri. If esgeri from osgar-, why not *ertheri from orthor-? Alternatively, if ortheri from orthor, why not *osgeri from osgar-?

Perhaps this is to some extent optional. WJ:379, dealing with noun plurals, suggests that the "affection" or umlaut was originally carried through the word, so that a compound like orodben "mountaineer" in older times had the plural oerydbin (= örydbin, classical Sindarin erydbin). But later, to the extent this word was recognized as a compound orod-ben "mountain-person", only the second element was umlauted in the plural: orodbin. So perhaps esgeri "to amputate" later became *osgeri instead, and perhaps ortheri represents earlier *ertheri.

Here are some verbs with prefixes and suggested conjugations.

With the prefix go- "together":

# govad- "meet, come together", infinitive gevedi, present tense gevedi- (add the appropriate ending, except in the 3rd person singular, that is govad), past tense gevenni- (3 sg govant), future geveditha, imperative govado, participle govadel (perfective govódiel), past participle govannen, gerund govaded
# gonathra- "entangle, enmesh", inf. gonathro, pr.t. gonathra, pa.t. gonathranne- (3 sg gonathrant), fut. gonathratha, imp. gonathro, part. gonathrol (perfective genethriel), pp. gonathrannen (pl. genethrennin), ger. gonathrad
# gonod- "count up, reckon, sum up", inf. genedi, pr.t. genedi- (3 sg gonod), pa.t. genenni- (3 sg gonont), fut. geneditha, imp. gonodo, part. gonodel (perfective gonúdiel), pp. gononnen, ger. gonoded
# genedia- "reckon", inf. genedio, pr.t. genedia, pa.t. genedianne- (3 sg genediant), fut. genediatha, imp. genedio, part. genediol (perfective gonúdiel), pp. genediannen (pl. genediennin), ger. genediad

(Notice that in the latter verb, go- appears in umlauted form in all forms except the \perfective participle gonúdiel "having reckoned". The closely related verbs gonod- and genedia- would have identical perfective participles.)

This group of verbs incorporating the prefixes ad- "re-" and an- "to" would probably not change them to ed- or en- where umlauts may be thought to occur, though we have no clear examples:

# adertha- "reunite", inf. adertho, pr.t. adertha, pa.t. aderthanne- (3 sg aderthant), fut. aderthatha, imp. adertho, part. aderthol (perfective aderthiel rather than ?ederthiel), pp. aderthannen (pl. aderthennin rather than ?ederthennin), ger. aderthad
# anglenna- "approach", inf. anglenno, pr.t. anglenna, pa.t. anglenne- (3 sg anglennant), fut. anglennatha, imp. anglenno, part. anglennol (perfective anglenniel (rather than ?englenniel), pp. anglennen (pl. anglennin rather than ?englennin), ger. anglennad
# aníra- "desire", inf. aníro, pr.t. aníra, pa.t. anirne- (3 sg anirn), fut. aníratha, imp. aníro, part. anírol (perfective aníriel rather than ?eníriel?), pp. anirnen (pl. anirnin rather than ?enirnin), ger. anírad

With the prefix os- "around":

osgar- "cut round, amputate", inf. esgeri, pr.t. esgeri- (3 sg osgar), pa.t. esgerni- (3 sg osgarn), fut. esgeritha, imp. osgaro, part. osgarel (perfective osgóriel), pp. osgarnen (pl. esgernin), ger. osgared

A long, clearly independent prefix like palan- "far and wide" may not show any umlauts:

palan-dir- "view far and wide", inf. palan-diri, pr.t. palan-diri- (3 sg palan-dir), pa.t. palan-dirni- (3 sg palan-dirn), fut. palan-diritha, imp. palan-diro, part. palan-diriel (perfective palan-díriel - hardly ?pelen-díriel), pp. palan-dirnen (pl. palan-dirnin, hardly ?pelen-dirnin), ger. palan-dired

5. PRONOUNS

Attested Sindarin pronouns include:

1st person sg: Independent pronoun im "I", also the ending -n; cf. also nin "me", genitive nín "my", also anim "for myself" (evidently an "for" + im "I, *me") and enni "to me".
2nd person sg: The ending -ch, assuming that agorech does mean *"you did"; cf. also the reverential dative pronoun le "to thee", said to be of Quenya origin (RGEO:73), and lín as the genitive "thy, your".
3rd person sg: E "he", genitive dîn "his" (this could probably also be spelt dín, cf. lín "your" above). The word den in the Lord's Prayer translation may mean "it" as object; if so, it could likely cover "him" as well (and then dîn possibly covers "its" as well as "his", since the words are obviously related).
1st person pl: Ending -m "we" (in avam "we won't", WJ:371), evidently men and mín as independent pronoun "we" or "us", also ammen "for us" or "of us" (for *an men; an "for, to", men = "us"?). "Our, ours" is vín.
2nd person pl: none found, unless -ch covers both sg. and pl. "you" (cf. PM:45-46)
3rd person pl: hain "them" (prob. also subject "they")

When added to a stem ending in -a, the pronominal ending -n "I" seems to change this vowel to -o; contrast avam "we won't" with avon "I won't" (WJ:371, ava = "won't"). Cf. also linnon "I sing" and linnathon "I will sing"; the stems are evidently linna and *linnatha, "sings" and "will sing" (hence *linnam "we sing", *linnach "you sing"?)

Though an independent word for "my" is given in UT:54 (nín), there also exists an ending -en that can express the same meaning. It is attested in the word lammen "my tongue" in Gandalf's invocation before the Gate of Moria (LotR1/II ch. 4; see RS:463 for translation). Compare the Quenya ending -nya "my". A second attestation of the corresponding Sindarin ending became available in July 2000, when a sentence including the word guren "my heart" was published in VT41:11, 15. Presumably Sindarin has other pronominal possessive endings as well, but only -en "my" has been published. Since Tolkien elsewhere uses independent pronouns for "my" and "his", it may be that he changed his mind back and forth as to whether Sindarin used endings or independent genitive pronouns.

In addition to the genitive pronoun dîn "his", the King's Letter also has în: The king wishes to greet mhellyn în phain, all his friends. Though în, like dîn, is translated "his" in English, it appears that this is actually a reflexive genitive pronoun, referring back to the subject of the sentence. In Sindarin there may be a distinction that is not regularly expressed in English. Two sentences like *i venn sunc haw în and *i venn sunc haw dîn would both translate as "the man drank his juice" in English, but the first means "the man drank his (own) juice", while the second means "the man drank his (someone else's) juice" (Norwegian mannen drakk saften sin vs. mannen drakk saften hans, if I may refer to my mothertongue).

Under the stem S- in the Etymologies, some "Noldorin" pronouns are listed, but whether they are valid in LotR-style Sindarin is not known: Ho, hon, hono "he", he, hen, hene "she"; ha, hana "it". The plurals are given as huin, hîn, hein, evidently meaning "they" referring to a group of men, women and things, respectively. Hein would later appear as hain because of regular sound-change; cf. the Moria Gate inscription: Im Narvi hain echant "I Narvi them [= the letters] made". Moreover, the "Noldorin" pronoun huin would appear as *hýn in (Third Age) Sindarin.