This hymn is the longest Sindarin text in LotR, found near the end of the
chapter "Many meetings" (LotR1/II ch. 1). The hobbits are in the house of
Elrond and leave the Hall of Fire: "Even as they stepped over the threshold a
single clear voice rose in song... [Frodo] stood still enchanted, while the
sweet syllables of the elvish song fell like clear jewels of blended word and
melody. 'It is a song to Elbereth,' said Bilbo. 'They will sing that, and other
songs of the Blessed Realm, many times tonight.' " In Letters:278, Tolkien
calls it a "hymn-fragment", suggesting that what we have is only one stanza of
many.
The hymn is nowhere translated in LotR, except for the words galadhremmin
Ennorath that are interpreted "tree-woven lands of Middle-earth" in the
second footnote in Appendix E. However, Tolkien provided a translation of this
song in RGEO:72, followed by some illuminating comments. This is the main
source for this article.
The hymn to Elbereth (that in RGEO:70 has a Tengwar superscript Aerlinn in Edhil o Imladris, *"Hymn of the Elves of Rivendell"):
A Elbereth Gilthoniel,
O Elbereth Star-kindler
silivren penna míriel
(white) glittering slants down sparkling like jewels
o menel aglar elenath!
from [the] firmament [the] glory [of] the star-host!
Na-chaered palan-díriel
To-remote distance far-having gazed
o galadhremmin ennorath,
from [the] tree-tangled middle-lands,
Fanuilos, le linnathon
Fanuilos, to thee I will chant
nef aear, sí nef aearon!
on this side of ocean, here on this side of the Great
Ocean!
In RGEO, Tolkien compared this hymn to the invocation uttered by Sam "speaking in tongues" in Cirith Ungol (LotR2/IV ch. 10: "then his tongue was loosed and his voice cried in a language which he did not know"...). We follow his example and will analyze this short utterance here as well. Notes Tolkien in Letters:278, "Though it is, of course, in the style and metre of the hymn-fragment [A Elbereth Gilthoniel], I think it is composed or inspired for his [Sam's] particular situation".
A Elbereth Gilthoniel o menel palan-diriel, le nallon
O Elbereth Starkindler from firmanent gazing afar, to thee I
cry
sí di-nguruthos! A tiro nin, Fanuilos!
here beneath death-horror! O look towards me, Everwhite!
Tolkien's own translation of these texts (rather free and florid):
[The hymn:] "O! Elbereth who lit the stars, from glittering crystal
slanting falls with light like jewels from heaven on high the glory of the
starry host. To lands remote I have looked afar, and now to thee, Fanuilos,
bright spirit clothed in ever-white, I here will sing beyond the Sea, beyond
the wide and sundering Sea."
[Sam's invocation:] "O! Queen who kindled star on star, white-robed from
heaven gazing far, here overwhelmed in dread of Death I cry: O guard me,
Elbereth!" Another translation, more literal, is given in Letters:278: "O
Elbereth Starkindler from heaven gazing-afar, to thee I cry now in the shadow
of (the fear of) death. O look towards me, Everwhite."
The hymn: A "o", here evidently used as a vocative particle. Elbereth the normal Sindarin name of Varda. The element el- means "star", while bereth according to RGEO:74 means "spouse", used of the spouse of a king, hence coming to mean "queen". Varda is both the Queen of the Valar and the spouse of Manwë; in Letters:282 Elbereth is translated "Star-lady". Why is bereth is not lenited to *vereth in Elbereth, though the second element in a compound would normally be lenited? Tolkien addressed this question in MR:387: It is because the element el- "star" was originally elen, as in Quenya, and so we have older Elenbarathi yielding Elmbereth, simplified to Elbereth, older lmb becoming lb instead of lv. Note that the word Elbereth is not directly related to Quenya Varda "Lofty, Sublime" (the Quenya form of Elbereth would have been something like *Elenvarsi, while the Sindarin cognate of Varda would have been *Baradh or possibly *Bradh, but there is no evidence that these forms were in use as names of the Starqueen). Gilthoniel "Star-kindler": Gil "bright spark, star" (as in Gil-galad "Star of Radiance") + thoniel "kindler". In MR:388, the latter element is said to come from a stem than, thân "kindle, set light to" + iel "a feminine suffix corresponding to male -we". (Sindarin th cannot undergo any lenition and is therefore unchanged when gil- is prefixed.) In Letters:278, Gilthoniel is translated "Starkindler", but Tolkien added a note: "in the past tense: the title belongs to mythical pre-history and does not refer to a permanent function". So somehow thoniel is marked as past tense "one having kindled" instead of "one who is kindling (now)". If we see it as a participle, displaying the same ending as in palan-díriel "having gazed far" later in the hymn (as opposed to present tense palan-diriel "gazing far" in Sam's invocation), it should have a long vowel in the past tense. Since the stem is given as than-, thân- in MR:388, not *thon-, we are evidently to understand that long á (â) became o (via au). Many parallels show this to be the case; for instance, Sindarin Anor "Sun" comes from anâr- (LR:378, stem ANÁR). silivren "glittering (white)". The -ren part is an adjectival ending, while siliv- is a Sindarized form of Quenya silima, Fëanor's name for the special crystalline substance that he devised, using it to make the Silmarils. Tolkien noted that the word silivren "would recall to Elvish minds the silmarils and describe the stars as crystalline forms shining from within with a light of mysterious power" (RGEO:73). Here, silivren is apparently used adverbially, describing how "the glory of the star-host" (see below) is 'slanting down'. penna a verb "slants", here with the ending -a that A-stem verbs show in the present tense. It must be derived from a nasal-infixed form of the stem PED "slope, slant down" (WJ:375). The subject of this verb seems to be aglar elenath "the glory of the star-host"; see below. míriel adj. "sparkling like jewels" (compare mîr "jewel", Quenya mírë). Míriel looks almost like a participle, but for various reasons it must rather be taken as an adjective (with a long vowel it should have meant "having sparkled like jewels" if it were a participle, but this is clearly not the meaning, so we must rather assume that it simply preserves the long vowel of mîr). Here, míriel (like silivren) is used adverbially, describing how "the glory of the star-host" (see below) is 'slanting down'. o "from"; menel "firmament, high heaven, the region of the stars" (according to RGEO:72 a borrowing from Quenya). aglar "glory", elenath "(the) star-host", "(all) the (visible) stars of the firmament". In WJ:363, Tolkien states: "êl, pl. elin, class plural elenath. An archaic word for 'star', little used except in verse, apart from the form elenath 'all the host of the stars of heaven'." In RGEO:74-75, Tolkien explains that the ending -ath "was used as a group plural, embracing all things of the same name, or those associated in some special arrangement or organization" (RGEO:74-75). The whole phrase aglar elenath "(the) glory (of the) star-host" is an example of the Sindarin "genitival" construction that simply involves the juxtaposition of two nouns, the possessed followed by the possessor: "(the) glory (of the) star-host". Cf. aerlinn in Edhil "hymn (of) the Elves" in the superscript; cf. also the inscription on the gate of Moria: Ennyn Durin Aran Moria "Doors (of) Durin King (of) Moria". Na-chaered "to a remote distance": na- "to" + *haered "remote distance" (compare the Quenya adjective haira "remote, far"); *haered is lenited to chaered following the prepositional element na-. palan-díriel "having gazed far": palan- "afar, abroad, far and wide", an element borrowed from Quenya (occurring in palantír, "that which watches from afar"); -díriel lenited form of tíriel "gazing, watching", participle of tir- "watch". It is lenited as the second part of a compound. According to Tolkien, palan-díriel with a long stem-vowel (í) means "having gazed far away" (in the past), while palan-diriel as in Sam's invocation, with a short i, means "gazing away" (now). The distinction is past vs. present. - The stem tir- "watch" is of course the same as in Quenya palantír. Indeed the whole phrase palan-díriel is intended to suggest "having looked into a palantír", since this is a hymn sung by Elves that had been on a "pilgrimage" to the Emyn Beraid to look towards the Blessed Realm using the Seeing Stone there. - One may ask why the final n in palan, when prefixed to tíriel, does not cause nasal-mutation - sc. n + t becoming th, as when the underlying phrase *in tiw "the runes" manifests as i thiw in the Moria gate inscription. Instead of nasal mutation, the t of tíriel undergoes lenition (soft mutation) to become d, producing palan-díriel. Tolkien addresses this question in Letters:427: "palan-tîriel should phonetically > -thíriel...but grammatically before actual forms of verbs, the soft mutation only was normally used in later S[indarin], to avoid the confusion with other verb stems". o "from". galadhremmin "tree-tangled": galadh "tree" + remmin "tangled". This is an adjective derived from rem "mesh"; see the second footnote in Appendix E to LotR. Remmin is the plural form of this adjective; the singular form would be *remmen. It is pl. to agree with ennorath, translated "middle-lands" in RGEO:72. Actually it is Ennor "Middle-earth" (*en- "middle" + nor "land"; Quenya Endor, Endórë) with the collective ending -ath as in elenath above, hence referring to the different lands of Middle-earth as a group. Fanuilos is translated "Ever-white". There are three elements: Fân, fan- means "veil", but also connotes the Quenya cognate fana, used of the physical shapes that the Valar put on when presenting themselves in visible form. Ui means "ever", while -los must be a lenited and reduced form of gloss, "snow-white". The whole, Tolkien explains, therefore means "bright (angelic) figure ever white (as snow)" (RGEO:74). A slightly different explanation is given in Letters:278; here the element fan- is also said to mean "white": "Everwhite is an inadequate translation, as is equally...show-white... The element ui (Primitive Elvish oio) means ever; both fan- and los(s) convey white, but fan connotes the whiteness of clouds (in the sun); loss refers to snow." le "to thee", a pronoun borrowed from Quenya (in which language it was probably not a dative form, but rather an accusative and/or nominative; in High-elven, "to thee" would probably be dative *len or allative *lenna). It is possible that le can also be used as an accusative in Sindarin and was originally borrowed as such. linnathon "I will sing": Stem linna- "sing" + the future-tense ending -tha "will" + the ending -n "I", that causes a preceding a to become o. Compare nallon in Sam's invocation below; cf. also linnon *"I sing" in The Lays of Beleriand p. 354. nef is a preposition "on this side of". It is pronounced nev and is so spelt in Tengwar writing, but Tolkien had a strange horror of final orthographic v's, so when writing Sindarin with our letters he used the letter f instead, as in English of ("ov"). The "correct" spelling is used when v is not final, as in the name Nevrast "Hither Shore" in the Silmarillion. aear "ocean, sea" (Quenya ëar as in Eärendil). In Tolkien's later conception, aear must be seen as a lenited form of gaear, lenited because it follows the preposition nef (just like *haered is lenited to chaered following na). But the evidence is that when Tolkien originally wrote the hymn, he thought of aear as a complete word, not as gaear with the normal lenition G > zero. In the Etymologies, the word for "sea" had been oear, derived from a stem AYAR/AIR. Tolkien later revised the sound-changes so that this would become aear instead. But in an essay dating from about 1960, Tolkien derived the Sindarin word for "sea" from a stem GAYA instead, so now it became gaear (WJ:400). Then he reverted to the stem AYAR yielding Sindarin aear, as is evident from a letter he wrote in 1967 (Letters:386). Then he changed his mind again; in a text written at least one year later, the Sindarin word for "sea" is once again gaear (PM:363), and as far as we know, this was his final decision. (However, I wouldn't be surprised if it turns out that a note with some doodles was found near his deathbed: "AYAR sea; Q ëar, S aear.") sí "here", nef "on this side of" again, aearon must be seen as a lenited form of gaearon, sc. gaer with what Tolkien calls an augmenting suffix (RGEO:73). If gaer is simply "ocean", gaearon is "Great Ocean" (RGEO:72). Just as in the case of aear, there is little doubt that Tolkien thought of aearon as a complete word in itself, not a lenited form, when he first wrote the hymn. (If we don't see aear, aearon as lenited forms, they must be explained as variants of gaear, gaearon influenced by Quenya ëar.)
Then there is Sam's invocation in Cirith Ungol:
First sentence: A Elbereth Gilthoniel, o menel palan-diriel, le
nallon sí di-nguruthos! "O Elbereth Starkindler, from heaven gazing
far, to thee I cry now beneath the shadow of death." A Elbereth
Gilthoniel "o Elbereth Starkindler" as in the hymn (the first edition of
LotR had o Elbereth instead of a Elbereth; this was an
error that Tolkien later corrected, see Letters:278). o "from". menel
"firmament" as in the hymn above. palan-diriel "gazing far", same
elements as in palan-díriel "having gazed afar" above, but here
the stemvowel is not lengthened (i, not í), and this
somewhat flimsy device indicates that this participle is present tense "gazing
far" instead of past tense "having gazed afar". (Early editions of LotR
actually read palan-díriel with a long vowel instead of
palan-diriel; this is an error, according to a footnote in RGEO:72.)
le "to thee" as in the hymn. nallon "I cry", evidently a verbal
stem *nalla- "cry" with the pronominal ending -n "I". For some
reason, this ending always causes a preceding -a (for present tense as
here, or as part of the future-tense ending -tha) to change to
-o, hence "I cry" is nallon rather than **nallan, just
like "I will sing" is linnathon rather than **linnathan (see
above). Cf. also linnon *"I sing" in The Lays of Beleriand p.
354, stem *linna-. sí "here", as in the hymn; also
translated "now" (Quenya sí "now"). di-nguruthos "beneath
death-horror". So it is spelt in RGEO:72; LotR has di'nguruthos with an
apostrophe instead of a hyphen. The "normal" (that is, unmutated) form of
-nguruthos would be *guruthos. The first element of this compound
is plainly guruth "death" (LR:377, stem ÑGUR). According
to the Etymologies, this should mean "death as a state or abstract"
rather than "act of dying" (that is gwanw or gwanath), but Sam
would be concerned with the possibility of his own "act of dying" rather than
death as an abstract, so it seems that guruth here takes on the meaning
of gwanw (which, by the way, should rather be *gwanu in LotR-style Sindarin). The final element in *guruthos is evidently the same as
in delos "abhorrence", where del- represents the stem DYEL
"feel fear and disgust" and the -os part is equated with gos
(LR:355), g leniting to zero in the compound. Gos in turn
connects with the stem GOS/GOTH "dread", whence Quenya
ossë "terror" - which word is also the name of the Maia
Ossë, according to the Etymologies. An unused Sindarin
cognate of the name Ossë is given as *Goss, so we can
conclude that gos, goss means "terror, *horror", giving
*guruthos the meaning "death-horror". This is how Tolkien translated it
in RGEO:72, while Letters:278 has "the shadow of (the fear of) death"; no
word meaning "shadow" is actually present. The prefixed element di' or
di- is translated "in" in Letters:278, but the more literal translation
in RGEO:72 seems to indicate that it actually means "beneath". This prefixed prepositional element
is somehow responsible for the fact that *guruthos here manifests as
nguruthos. There are two possibilities. Since guruth comes from a
stem ÑGUR, initial ng- reflecting the original initial nasalized stop would appear following closely related particles ending in a vowel, such as the article (*i nguruthos "the death-horror"). There may be a preposition di "beneath" that behaves like the article i "the" in this respect. On the other hand, the preposition may also be *din, and the final n causes nasal mutation of the initial g of *guruthos. Does the ' of di'nguruthos suggest that the final n of *din has disappeared, being swallowed up in the nasal mutation? (If so, a more standard spelling would have been simply *din Guruthos; compare in Gelydh, not i'Ngelydh, for "the Noldor".) Following the apostrophe, we might have expected a space clearly separating the words (*di' nguruthos), but there seems to be no space in the text in LotR.
Second sentence: A tiro nin, Fanuilos! "O look towards me,
Everwhite". A "o", not here used as a vocative particle as in a
Elbereth, but rather as a particle emphasizing the following imperative:
tiro "look" or "watch": stem tir- with the normal imperative
ending -o. According to Letters:427, this imperative in -o covers
all persons (no distinct plural form: one Elf cried daro! "halt!" to the
entire Fellowship as they were entering Lórien; see LotR1/II ch. 6).
Early editions of LotR read tíro with a long í, but
according to RGEO:72 this is an error. nin is translated "towards me".
The first element of this pronoun is clearly identical to Quenya ni "I";
the final -n could be what remains of the primitive element na
"towards" after the loss of the final vowels (cf. the stem
NÂ1 in the Etymologies and the na of
na-chaered "to remote distance" in the hymn). This would give nin
the meaning "I-towards" = "towards me". Fanuilos "Everwhite" again.
Note on the mutations following the preposition o "from,
of":
As this hymn illustrates, many Sindarin prepositions trigger lenition,
also known as "soft mutation", of the following word. *Haered becomes
chaered following na, and gaear/gaearon becomes
aear/aearon following nef. But what about the preposition
o "from, of"?
In the phrase o Imladris "from Rivendell" in the superscript there is
of course no mutation since words beginning in a vowel cannot undergo any such
change. But it is remarkable that there is no lenition in the phrases o
menel and o galadhremmin ennorath. We might have expected **o
venel and **o 'aladhremmin ennorath instead, the words following
o being lenited. This does not happen. Why?
Notes Tolkien in WJ:366-367, "The preposition o [is] the usual word
for 'from, of'... As the mutations following the preposition o show, it
must prehistorically have ended in -t or -d. Possibly, therefore,
it comes from *aud... [The preposition o] is normally o in
all positions, though od appears occasionally before vowels, especially
before [words beginning in] o-."
We must assume that before voiced sounds like the m of menel
and the g of galadh, the final d of aud was simply
assilimated to a similar sound, *aum m- and *aug g- later being
simplified to o g- and o m- as in o galadh, o
menel. Therefore, words beginning in m or g (and probably
other voiced sounds like d, b, l, r, n) are
unchanged following the preposition o. But since Tolkien refers to "the
mutations following the preposition o" in WJ:366, something interesting
sometimes does happen following o. We have no direct examples of what he
means, but what we generally know about Sindarin phonology and its evolution
suggests that o triggers spirant mutation of nouns beginning in
the unvoiced plosives p-, t-, c-. Before a noun like
perian "halfling, hobbit", the d of aud would be
assimilated to the following p, so that aud became *aup.
Then the double p of *aup perian, like all other double
p's, became a single spirant ph (= f) in Sindarin. "From
(a) hobbit" would therefore probably be *o pherian. Similarly, c
and t would become spirants ch and th following the
preposition o. (In a poem published in Tyalië
Tyelelliéva #11, David Salo translated "from Celos" as o
Chelos.)