The question of nd or n(n)

We continue our not-so-exiting series of articles about the differences between the "Noldorin" of the Etymologies and Tolkien's later Sindarin by looking into the behaviour of the cluster nd. In "Noldorin" and Sindarin alike, it is supposed to become nn (or even simple n) in certain positions, but it would seem that Tolkien changed the relevant rules more than once, and the material is quite inconsistent. For the benefit of Sindarin lexicographers, we will review the corpus and try to suggest how it can be regularized.

We are so lucky that in LotR, Tolkien set out some rules for how nd/nn was supposed to behave in Sindarin. The relevant passage in Appendix E goes like this:

nd became nn usually, as Ennor 'Middle-earth', Q. Endóre, but remained nd at the end of fully accented monosyllables such as thond 'root' (cf. Morthond 'Blackroot'), and also before r, as Andros 'long-foam'. This nd is also seen in some ancient names derived from an older period, such as Nargothrond, Gondolin, Beleriand. In the Third Age final nd in long words had become n from nn, as in Ithilien, Rohan, Anórien.

With these rules in mind, let us go over the words in the "Noldorin" corpus that would be affected by the development Tolkien here sketches, and in some cases consider later Sindarin material as well.

1. The behaviour of final nd in monosyllables

The passage just quoted needs some exegesis, but one thing is clear: the combination nd remained unchanged "at the end of fully accented monosyllables". However, this is not true of the "Noldorin" of the Etymologies. In a great number of cases, the Etymologies chronicles a "Noldorin" change from nd to nn even at the end of monosyllabic words, where this change should not occur in Sindarin according to the explicit information given in LotR. Thus, Tolkien is seen to have changed his mind on this point. When updating "Noldorin" to Sindarin we must therefore ignore the development nd > nn in monosyllables.

In the following cases, the Etymologies lists both the "older" form with nd intact (which would be the only Sindarin form) and the later form where nd had become nn. Here we must simply ignore the phonological change and adopt the older form in our Sindarin wordlists: and > ann "long" (ÁNAD/ANDA), band > bann "duress, prison" (MBAD), brand > brann "lofty, noble, fine" (BARÁD), bund > bunn "snout, nose, cape" (MBUD), chwand > chwann "sponge, fungus" (SWAD; the Sindarin form or spelling should rather be hwand, which form is actually mentioned in the same entry), chwind > chwinn (only defined as the adj. corresponding to the verb chwinio "twirl, whirl, eddy", so it must mean *"twirling, whirling") (SWIN; the Sindarin spelling should be *hwind rather than chwind), fend > fenn "threshold" (PHEN), gwend > gwenn, which is both a noun "maiden" (WEN/WENED) and the older past tense of the verb gwedi (read evidently *gweði = *gwedhi) "bind" (WED), gwind > gwinn "blue-grey, pale blue or grey" (WIN/WIND; it is uncertain whether Tolkien rejected this word or not), hand > hann "intelligent" (KHAN), ind > inn "inner though, meaning, heart" (ID), lhand > lhann "wide" (LAD, cf. LAT; because of another revision we must read *land rather than lhand in Sindarin), lhind > lhinn "air, tune" (LIN2; again, read *lind rather than lhind in Sindarin), lond > lonn "path" (mentioned under AK, but derived from the stem LOD), nand > nann "wide grassland" (NAD), nend > nenn "watery" (NEN), pend > penn "declivity" (PEN/PÉNED), rhind > rhinn "circle" (RIN; because of another revision we must read *rind rather than rhind in Sindarin), rhond > rhonn "cave" (ROD, likewise read rond for rhond in Sindarin - the Sindarin form rond is attested in WJ:414, though the same source indicates that Tolkien had altered its derivation, now referring it to a stem RONO instead of ROD), thind > thinn "grey, pale" (THIN), thlind > thlinn "fine, slender" (SLIN; because of another revision we must read *lhind, not thlind, in Sindarin), tond > tonn "tall" (TUN), tund > tunn "hill, mound" (TUN). In addition to these, there are a few forms that Tolkien struck out in the Etymologies: dend > denn "sloping" (DEN, deleted and replaced by PEN) and glind > glinn "pale blue" (GLIND).

Another, much shorter list covers the cases where Tolkien only mentioned the "older" form with nd intact, not including any later "Noldorin" form with nn that we would have to ignore anyway. The forms in question are gwend "bond, friendship" (only mentioned in the entry WEN, WENED but said to be derived from WED - thus the word gwend has a third meaning in addition to the two mentioned above), lhand "open space, level" (LAT, cf. LAD - because of another revision we must read *land in Sindarin), lhend "tuneful, sweet" (LIND; again, read *lend in Sindarin), grond "club, mace" (RUD; this word occurs both in LotR and the Silmarillion), hwand "sponge, fungus" (SWAD; in Sindarin, this form is to be preferred to both chwand and chwann mentioned in the same entry).

Then we have the most challenging group: the cases where Tolkien only mentioned the "later" form with nn, but where we can deduce that there must have been an older form that had nd - which is the form we must use in Sindarin. In some cases, "Noldorin" nn should however be nn in Sindarin as well (see below concerning crann, donn), so we cannot mechanically alter final nn to nd everywhere. We must find out whether nn in any given case descends from nd or not.

These are the "Noldorin" words that should evidently have nd rather than nn in Sindarin: cann "bold" (KAN; Tolkien mentioned the primitive form kandâ), cunn "prince" (the "stem" KUNDÛ seems to be an actual reconstruction of the primitive word; cf. also the Quenya cognate cundu), gonn "great stone, rock" (cf. the form of the stem GOND; the Sindarin form gond is mentioned in the Silmarillion Appendix, cf. also gondram "hewn stone" in Etymologies itself, entry DARÁM), ionn "son" (YÔ, YON, cf. the Quenya cognate yondo - if we are not to consider Sindarin *iond as the word for "son" superseded by iôn, WJ:337), lhonn "narrow path, strait" (LOD; cf. the compound Aglond in the entry AK; the Sindarin form lond - defined "land-locked haven" - is mentioned in the Silmarillion Appendix, entry londë, though Christopher Tolkien curiously adds lonn as a parenthetical alternative), ninn "slender" (cf. the stem NIN-DI and the Quenya cognate nindë), pann "courtyard" (PAD, cf. the Quenya cognate panda "enclosure" - but the homophone pann "wide", derived from the stem PAT, should be pann in Sindarin as well, since this word comes from primitive patnâ rather than *pandâ).

The adjective rhinn "circular" (RIN) we would normally update to Sindarin *rind, which would make it a homophone of the "Noldorin" noun rhind > rhinn, Sindarin *rind, mentioned in the same entry. However, the question has been raised whether the adjective rhind may not be a misreading for *rhend (Sindarin *rend) in Tolkien's manuscript, which would eliminate the homophones: The noun rhind would seem to be the cognate of Quenya rindë "circle", whereas the adjective also given as rhind seems to correspond to Quenya rinda "circular": primitive *rindê and *rindâ, respectively. *Rindâ would be expected to produce "Noldorin" *rhend, not rhind, because the final *-â would cause A-umlaut and change original *i to e (for a similar noun/adjective-pair, see the entry DEM: the primitive noun dimbê yields "Noldorin"/Sindarin dim "sadness", whereas the primitive adjective dimbâ "sad" comes out as dem with an umlauted vowel). Yet we cannot be certain that Tolkien intended the "Noldorin" adjective rhind to be wholly cognate with Quenya rinda, so it is difficult to say whether we should let the Sindarin word for "circular" be *rind or *rend.

In the case of the "Noldorin" adjective donn "swart, swarty" (DUN), some subtle reasoning is required to determine whether it should remain donn or be altered to *dond in Sindarin. The stem DUN could have yielded a primitive adjective *dundâ. Compare Tolkien's own "reconstruction" kandâ from KAN, yielding "Noldorin" cann "bold" (Sindarin *cand). Likewise, *dundâ would yield "Noldorin" donn, but Sindarin *dond. On the other hand, donn could also descend from a primitive form *dunnâ (*dun-nâ, the latter syllable being a well-attested adjectival ending). Compare the given primitive form k'rannâ "ruddy" from the stem KARÁN, yielding "Noldorin" crann; this would be the same in Sindarin since nn does not here descend from earlier nd but had been nn all along. So if "Noldorin" donn descends from *dunnâ, the word should have this form in Sindarin as well. Besides the stem DUN itself, the only clues we have to the origin of donn are the Doriathrin and Danian (Nandorin) cognates, quoted in the same entry as dunn and dunna, respectively. Doriathrin seems to preserve primitive nd finally; the entry NAD lists Doriathrin nand "field" as the cognate of Quenya nanda "watered plain" (both from primitive *nandâ, also yielding "Noldorin" nand > nann, in Sindarin only nand). So if donn descended from *dundâ, we should expect to see **dund rather than dunn as the Doriathrin cognate. Hence donn must come from *dunnâ instead (the original nn being unchanged in Doriathrin dunn). Thus, donn rather than **dond would also be the Sindarin form of the word; this word never contained nd.

So far monosyllabic words. It should be understood that if these words occur in compounds or receive an ending so that they are no longer monosyllabic, the shift from nd to nn takes place in Sindarin as well. For instance, LotR (volume 1, the chapter A Knife in the Dark) mentions ann-thennath (evidently meaning "long-shorts") as an Elvish form of verse. This word seems to include and "long", here turning into ann because the word does not appear as a monosyllable, but as part of a compound. (See also below concerning andaith = *ann-daith "long mark".)

2. The behaviour of nd between vowels

After noting that "nd became nn usually", Tolkien quoted Ennor "Middle-earth" as an example (contrasting it with the more conservative Quenya form Endórë, where original nd persisted). We must therefore conclude that intervocalic nd normally turned into nn. Tolkien added: "This nd is also seen in some ancient names derived from an older period, such as...Gondolin." The implication is that Gondolin (instead of *Gonnolin) is not really an exception to the rule he has just set out; it is merely a name that persisted in archaic form - probably helped by conservative spelling. Interestingly, the Etymologies quotes Gondobar "Stone of the World" as an alternative name of Gondolin, and does mention Gonnobar as what is evidently a later form (GOND). The Etymologies also record a change of intervocalic nd to nn in a handful of other words: andabon > annabon "elephant" (MBUD), findel > finnel "(braided) hair" (SPIN), gandel > gannel "a harp" (ÑGAN/ÑGANAD), (tindumh >) tindu > tinnu "dusk, twilight, early night (without moon)" (TIN). People writing in Third Age Sindarin should use the forms with nn. The monosyllabic words discussed in section 1 above, that in Sindarin should end in -nd rather than -nn, would also change it to -nn- if some ending is added to them so that -nd becomes intervocalic rather than final. We deduced that the Sindarin word for "son" should be *iond rather than ionn as in "Noldorin", but even in Sindarin the collective plural is ionnath, attested in the King's Letter (SD:129). While a form *iondath must have existed earlier, it would be archaic in Third Age Sindarin.

Tinúviel "daughter of twilight", originally a kenning of the nightingale but also known as the name Beren gave to Lúthien, evidently represents a "modernized" form or spelling. The entry TIN in the Etymologies quotes a very old form as Tindúmhiell (where mh = nasalized v, later becoming normal v). The same entry makes reference to how Tindúmhiell then became Tinnúviel > Tinúviel (the double consonant evidently being simplified immediately before the accented vowel).

The place-name Tindobel "starlit village" (so in the entry PEL(ES) - the entry TIN has the variant Tindubel "twilit city") includes the same element tindu > tinnu "twilight" as Tin(n)úviel. If we were to "regularize" it to Third Age Sindarin, we would therefore have to read *Tinnubel or *Tinnobel, but since this is evidently a name descending from the First Age, it would perhaps persist in archaic form like the other old names Tolkien mentioned in LotR Appendix E.

It should be noted that intervocalic nd normally becomes nn only where the whole cluster is perceived to occur within one morpheme. The change does not occur where the n and the d simply happen to occur in sequence in a compound, the first element ending in -n and the second beginning in d-. One prominent example from LotR is the name Gondor, which is gon(d) "stone" + dôr "land". LotR also provides Baranduin as the original Elvish name of the Brandywine River; this is baran "brown, swart" + duin "river". Gandalf's epithet Mithrandir or "Grey Pilgrim" includes randir "pilgrim, wanderer", which is the verbal stem ran- "wander" + dîr "man", here used as an agental ending. No forms **Gonnor, **Barannuin, **Mithrannir are mentioned or implied to exist. (There is, however, the curious example Ennor for "Middle-earth"; this is en(d)- "middle" + dôr "land", and would earlier be Endor - attested as a Quenya form. Phonetically, this example seems to be entirely parallel to Gondor, so it is strange that Endor did not persist unchanged.)

However, some words from the Etymologies "ought" to have nn for nd in Third Age Sindarin as far as phonology is concerned: Under GOND we have Gondost *"Rock-city" as a name of Gondolin, plus two compounds derived from it, namely Gondothrim *"The people of Gondost" (Gondost-rim "Gondost-folk" becoming Gondothrim) and Gondothrimbar *"The land of the people of Gondost". In these words, nd belongs to a single morpheme and occurs between vowels, which should normally trigger the change nd > nn. Perhaps nd in these words would persist because of analogy with the more well-known name Gondolin (though we have already noticed that Etym does record a change from nd to nn in Gondobar > Gonnobar, yet another name of Gondolin).

We should also have a closer look at the name Gondolin itself. When pointing out that "nd is also seen in some ancient names derived from an older period", Gondolin was the only example of intervocalic nd that Tolkien went on to list (the other two examples involving the atypical survival of -nd finally in "long" names: Nargothrond and Beleriand). Of the origin of the name Gondolin, it is said in the Silmarillion Appendix, entry gond, that "the name of the hidden city of King Turgon was devised by him in Quenya as Ondolindë (Quenya ondo = Sindarin gond, and lindë 'singing, song'); but it was known always in legend in the Sindarin form Gondolin, which was probably interpreted as gond-dolen 'Hidden Rock'." Thus, the n and the d of Gondolin belonged to one morpheme as far as Turgon's original intentions were concerned (gond = Quenya ondo, rock or stone). But the folk etymology gon(d)-dolen would make the n and the d belong to separate morphemes, evidently making the word less prone to become *Gonnolin. (Incidentally, the same folk etymology would assist the loss of the final d in Gondolind = Ondolindë. When final nd survived in the two other examples of "ancient names" that Tolkien mentioned along with it, Nargothrond and Beleriand, we may well ask why it was simplified to -n in Gondolin(d).The association with dolen "hidden" would explain this, since this word never ended in -nd.)

In light of the above, quite a few "Noldorin" words from the Etymologies that display intervocalic nd can be accepted into Sindarin as they are, since they represent compounds where the first element ends in -n and the second element begins in d-: The river-name Baranduin, or Branduin, has already been discussed (BARÁN). We also find a number of masculine names or agental forms that end in -dir "man": the names Brandir (BARAD, DER) and Handir (KHAN), plus the common noun rhandir "wanderer, pilgrim" (RAN - read Sindarin randir because of another revision, and cf. Mithrandir in LotR).

Another group of words where we should not change "Noldorin" intervocalic nd to nn in Sindarin are compounds where d is actually a lenited form of t. An example occurring both in the Etymologies, in LotR (Appendix A) and in the Silmarillion is Idril's epithet Celebrindal, Silverfoot: This is a compound of the adjective celebren (celebrin-) "[made] of silver" and tâl "foot", the latter appearing as -dal in this position. It seems that this could never become **Celebrinnal. (In the Etymologies, entry KYELEP/TELEP, Tolkien changed Celebrindal to [C]elebrendal [lenited Gelebrendal], but that change can be ignored in Sindarin.)

Another example occurring both in LotR and in the Etymologies is the eagle-name Thorondor "King of Eagles" (THOR/THORON), which is thoron "eagle" + taur "king", the latter here manifesting as -dor. In Etym, this name also appears as part of the place-name Cilthorondor (KIL) or Cil-thorondor (THOR/THORON), apparently meaning "Thorondor's Cleft".

The word andeith or "long mark" (a diacritic indicating long vowels), mentioned in the entry TEK in Etym, reappears in its Third Age Sindarin form andaith in LotR Appendix E. This is and = ann- "long" + *taith "mark" (earlier teith, the form given in Etym), the initial t of the latter word being lenited to d in the compound andaith = *ann-daith.

The word mindon, which in the Etymologies is defined both "tower" (MINI) and "isolated hill" (TUN) also contains a lenited t: the ancestral form is variously given as minitaun and minitunda. Another word incorporating a lenited t is muindor "brother", with its "analogical pl." muindyr: the final element comes from the stem TOR "brother". In these words, nd should stay in Sindarin as well.

The name Lhúndirien or Luindirien is said to mean "Blue Towers", another name of the Eredluin or "Blue Mountains" (LUG2, spelt Ered Luin in the published Silmarillion). The -dirien part of this name would seem to be what is translated "towers"; cf. Quenya tirion "watchtower" (TIR). Thus we have yet another lenited t, but whether these "Noldorin" names of the Ered Luin would still be valid in Tolkien's later Sindarin is quite doubtful. (In any case, the lh of Lhúndirien must become simple l- in Sindarin; moreover, -dirien as the plural of *-dirion would be a strange formation in Sindarin: read *-diryn???) If I were to write in Sindarin, I would refer to the Blue Mountains as the Ered Luin, period.

This only leaves us with a handful of "Noldorin" words with nd between vowels. The name Glorfindel (LÁWAR/GLÁWAR, GLAW-[R], SPIN, PHIN) occurs in the LotR itself, so we cannot well alter it to *Glorfinnel. It may be noted, though, that in one late source Tolkien wrote that the use of this name "in The Lord of the Rings is one of the cases of the somewhat random use of the names found in the older legends, now referred to as The Silmarillion, which escaped reconsideration in the final published form of The Lord of the Rings. This is unfortunate, since the name is now difficult to fit into Sindarin" (PM:379). We are not told precisely what was "wrong" with the name Glorfindel (that is, why it did not fit Tolkien's late vision of Sindarin very well), but part of the problem may well have been that a name of this shape ought to have become *Glorfinnel by the late Third Age. The simplest solution would seem to be that it was simply an archaic First Age form, preserved or revived by its reincarnated owner (since Tolkien did decide that Glorfindel of Rivendell was the same person as Glorfindel of Gondolin way back in the First Age).

A couple of other names from the Etymologies, Borthandos (BOR, KHAN) and Findabar/Findobar (PHIN, MBAR) we may also see as archaic. Discussing whether this should be *Borthannos or *Finnabar/Finnobar in Third Age Sindarin is not very interesting as long as there is no evidence that these names were used in later ages at all.

Finally we have Eredlindon "Mountains of Lindon (Ossiriand)" (ÓROT, LIN2, LUG2; the published Silmarillion employs the two-word spelling Ered Lindon). The interesting element is of course Lindon; would it be pronounced *Linnon in later ages? In the entry LIN2, Tolkien actually wrote: "Lindon, Lhinnon Ilk[orin] name of Ossiriand". The idea is probably that Lindon is the proper Ilkorin form, whereas Lhinnon is an "Noldorin" adaptation. Lhinnon would correspond to Sindarin *Linnon, which might seem to be encouraging if we want to change nd to nn. But since Ilkorin as the indigenous tongue of Beleriand was replaced by Sindarin when Tolkien revised his mythos in the early fifties, there whole linguistic scenario underwent great upheavals that must also be taken into account. In the post-LotR source Quendi and Eldar, the name Lindon is presented as being neither Ilkorin nor "Noldorin"; Tolkien now explained it as a borrowing from Nandorin (Silvan Elvish): "The Sindar...adopted the names Lindi and Lindon, giving them the forms Lindil (sg. Lindel)...and Lindon" (WJ:385). Thus, the Sindarin form evidently was and ever remained Lindon. The same source lists Glinnil as an inherited Sindarin term for the Lindi or Lindarin Elves, so the change of intervocalic nd to nn had taken place in genuinely Sindarin words at the time the name Lindon was borrowed from Nandorin: The word was not fully adapted, perhaps deliberately, if it continued to be recognized as a borrowing.

3. The behaviour of nd before r

Tolkien further stated that nd remained "also before r, as Andros 'long-foam'." This seems to agree quite well with the "Noldorin" material in the Etymologies. We have the eagle-name Lhandroval (RAM), that reappears in its Sindarin form Landroval in the LotR itself, gondrafn or gondram "hewn stone" (DARÁM), and pendrad or pendrath "passage up or down slope, stairway" (PEN/PÉNED). An exception would seem to be the word anrand "cycle, age" (RAD); the primitive form randâ quoted in this entry is obviously the origin of only the second part of anrand. If the prefixed an- is a form of and "long", just as in Tolkien's example Andros above, we may ask why this word does not appear as *andrand. Another exception seems to be the name Finrod (PHIN, RAUTÂ), where the first element represents find "hair"; why not *Findrod? According to a note of Tolkien's published in VT41:9, the name Find-raud > Findrod > Finrod shows "loss of the medial d before a following d, as in the actual Sindarin name Thinrod 'noble member of the Thindrim (Sindar)'." The word anrand can then be explained in the same way: it was altered from *andrand because of the following d in the element -rand, overriding the normal rule that nd is preserved before r. (But the word pendrad "stairway" [PEN/PÉNED] does not fit this system...it should have been *penrad instead, and in our Sindarin dictionaries we shall probably have to use this emended form. However, writers can simply use pendrath of the same meaning and avoid the problem.)

4. The behaviour of final nd in polysyllabic words

Perhaps we should read *anrann or *anran as the current form of anrand in Third Age Sindarin, for there remains the question of what eventually happened to nd at the end of the second element of a compound. After observing that nd "remained nd at the end of fully accented monosyllables such as thond 'root' (cf. Morthond 'Blackroot')", Tolkien added that "in the Third Age final nd in long words had become n from nn, as in Ithilien, Rohan, Anórien". Well, just how long are "long words"? Tolkien's example Rohan, the final n of which represents older nd, would seem to indicate that a two-syllable word is long enough for this reduction to take place. Yet Tolkien in the same paragraph quotes a word of similar length, Morthond "Blackroot", with no similar reduction to *Morthon. While thond "root" may be a "fully accented monosyllable" by itself, Morthond is neither monosyllabic nor stressed on the syllable ending in -nd. In LotR, this is presented as the contemporary Third Age name of a river, so we cannot well assume that it is particularly "archaic" (cf. LotR, chapter 2 of Book Five, The Passing of the Grey Company, where Elladan tells Gimli: "We have descended from the uprising of the Morthond...you will not need to ask hereafter how comes its name: Blackroot men call it.") Indeed LotR would seem to indicate that there were "contemporary" words even longer than two syllables where a final -nd persisted instead of turning into -nn and then -n, cf. the Merethrond or "Great Hall of Feasts" mentioned in Book Six, chapter 6 (Many Partings); this is mereth "feast" + rond "(great) hall".

Yet our corpus does contain examples of the development from nd via nn to n at the end of polysyllabic words. To first quote a post-LotR source, in RGEO:70 we have aerlinn as a word meaning something like *"holy song" or hymn. (It occurs in the Tengwar superscript Tolkien placed above his transcription of the chant A Elbereth Gilthoniel: Aerlinn in Edhil o Imladris, usually taken to mean something like *"Hymn of the Elves of Rivendell". By another suggestion aerlinn means *"sea-song", since this hymn was sung by Elves that had returned from a pilgrimage to Emyn Beraid near the ocean.) The final element of aerlinn is apparently *lind, the Sindarin form of the word that in "Noldorin" had appeared as lhind "tune" (LIN2).

In the "Noldorin" of the Etymologies, we can find similar examples. The entries DO3/DÔ and LIN2 give a word for "nightingale" as dúlind, dúlinn, dúlin(n), thus confirming the development nd > nn > n in "long words" that Tolkien speaks of in Appendix E (though a two-syllable word would seem to be "long" enough). Furthermore, the entries AM2 and PEN/PÉNED together provide all the members of a trinity ambend > ambenn > amben as the word for "uphill" (the second element being a lenited form of pend "declivity"). In the entry DUL we also have all three forms explicitly recorded in the case of the name "Gondolind, -inn, -in", sc. Gondolind > Gondolinn > Gondolin. The entries KHOR and ID also show how the name Húrin evolved via Húrinn from Húr-ind (apparently meaning something like *"vigorous mind").

In a number of other cases, the Etymologies records only the "original" form in -nd and the "final" form in -n, skipping the intermediate form in -nn. They are Aglond > Aglon "defile, pass between high walls" (AK, said to be a place-name), Banwend > Banwen "Vána", name of a Valië (spelt Vana in the Etymologies) (BAN), Brethiliand > Brethilian "Forest of Brethil" (BERÉTH), iðrind > iðrin "year" (RIN), moerilind > merilin "nightingale" (TIN; the change from ö [here spelt "oe"] to e evidently occurred during the same period that saw -nd become -n), othlond > othlon "paved way" (LOD), tuilind > tuilin "swallow" (TUY), Uinend > Uinen name of a Maia, the wife of Osse (UY - but this name Tolkien would later re-explain as a borrowing from Valarin, with no Elvish etymology: WJ:404), and finally ulund > ulun "monster, deformed and hideous creature" (ÚLUG).

In the case Lhothland > Lhothlann "empty and wide", name of a region (LAD, read Loth- for Lhoth- in Sindarin), we have the older form in -nd and the next stage in -nn recorded, but not the "final" form *L(h)othlan.

In a few cases only the forms in -nn are mentioned, but not the original form in -nd or the later form in -n, such as dadbenn "downhill, inclined, prone" (but the full development of the counterpart ambenn "uphill" is recorded, see above). Angolonn "Land of the Gnomes (Noldor)" is another example; the Quenya counterpart Ingolondë quoted in the same entry (ÑGOLOD) points to a primitive form *Ñ·golondê (with syllabic ñ). The oldest "Noldorin"/Sindarin form would be *Angolond, the "final" form *Angolon.

In the case of the name Túrin, the older form Túrinn is recorded (insofar as the name is given as "Túrin(n)" in the entry ID), but the oldest form *Túrind is not mentioned. (Contrast the full series Húrind > Húrinn > Húrin above, including the same final element ind "inner thought, meaning, heart".)

Then we have a whole series of cases where only the "oldest" form in -nd is recorded, while no mention is made of any later forms in -nn or eventually just -n: anrand "cycle, age" (100 Valian Years) (RAD), atland "sloping, tilted" (TALÁT), Baragund (masc. name) (BARÁS, KUNDÛ), camland "palm of hand" (LAD), Celebrond "Silver-mace" (RUD), Elulind (personal name, ?"Sky-singer") (3EL), Elrond "starry-dome", "vault of heaven", name of Eärendil's (Earendel's) son (EL, ROD, 3EL), Felagund (masc. name, in Etym suggested to mean *"Cave-prince", though Tolkien would later reinterpret it) (KUNDÛ, PHÉLEG), talagand "harper" (ÑGAN/ÑGANAD, also as masc. name Talagand). There is also the name Nargothrond (or Narogothrond), "Fortress of Narog" (NÁRAK, OS, ROD), that Tolkien in LotR Appendix E used as an example of an ancient name that still retained -nd in the Third Age.

The latter category may be compared to genuinely Sindarin (not "Noldorin") words found in the post-LotR essay Quendi and Eldar: In WJ:414 we have Hadhodrond as the Sindarin name of Khazad-dûm or Moria (this is Hadhod, a Grey-elven adaptation of Dwarvish Khazâd "Dwarves", plus rond "cave"). On the same page we also have othrond for "underground stronghold" (ost "fortress", reduced to oth- before r-, plus rond "cave"). Tolkien did not write *Hadhodron(n), *othron(n).

That completes our survey of the diverse zoo of attested forms. So what is the poor "regularizer" to do, attempting to crystallize a standard spelling for people interested in writing in Third Age Sindarin?

Of all the proper names so far mentioned, we should probably leave them more or less alone. Certainly it is strictly inconsistent to have (say) Baragund next to Húrin. Within the imagined history, while people still said Baragund, they also said Húrind - and when the latter name was eventually reduced to Húrin, Baragund was similarly reduced to *Baragun. However, the inconsistency in spelling may actually be quite in accord with the internal history Tolkien imagined for the Silmarillion (where these names occur). Observes Christopher Tolkien in his Foreword, "My father came to conceive The Silmarillion as a compilation, a compendious narrative, made long afterwards from sources of great diversity (poems, and annals, and oral tales) that had survived in agelong tradition." Sources of various age and written by many different authors would certainly display considerable differences in spelling - some quoting the names more or less as they had actually been pronounced in the First Age, others representing them as they had come to be pronounced in contemporary Sindarin. We know that Tolkien imagined differing spellings; in the Etymologies, entry WEN/WENED, he pointed to the fact that the names Morwen and Eleðwen "show no -d even in archaic spelling" as evidence that they could not descend from **Morwend, **Eleðwend. Hence, the "archaic spelling" did reflect older -nd where it had once been present, and the implication is that there was also a "modern spelling" that left out the final -d, since it was no longer pronounced anyway. But writing can also be quite conservative, especially in the case of proper names. So when you read, say, Baragund in the Silmarillion, you can take it as an "archaic spelling" of a name that was actually pronounced *Baragun in the Third Age - whereas Húrin, Túrin are samples of a "revised" or "modern" spelling, taking into account the fact that no one actually said Húrind, *Túrind anymore. (Or perhaps the actual "revised" Tengwar spelling was rather Húrinn, Túrinn; see below regarding the Tengwar spelling of aerlinn. If so, names like Húrin and Túrin are simply transcribed in accordance with the Third Age pronunciation, just like Tolkien in LotR used the transcription Noldor instead of Ñoldor or Ngoldor because initial ñ [like ng in king] had come to be pronounced n in the Third Age - he thus ignored the distinction ñ/n that was still upheld in Tengwar writing. Ignoring a solely orthographic distinction between single n and double nn would be even less dramatic.)

Highly interesting in this regard is WJ:5, where Tolkien assumes the perspective of an annalist writing in later ages: "Beleriand is the name of the country that lay upon either side of the great river Sirion ere the Elder Days were ended. This name it bears in the oldest records that survive, and it is here retained in that form, though now it is called Belerian." Thus the use of the form Beleriand is a deliberate archaism; in contemporary Sindarin the name was pronounced Belerian. Names like Baragund etc. would also be archaisms, whereas a spelling like Húrin reflects the pronunciation current in later ages; back in the First Age, Húrin would have called himself Húrind.

Apparently "archaic spellings" were still quite common in the Third Age, and in the text of LotR this is also reflected in names that were still in contemporary use, such as Morthond, Merethrond or Elrond. When such names were used in a foreign language like Westron, people would perhaps even be inclined to pronounce them exactly as written, reviving the final -d that had actually dropped out in contemporary Sindarin as spoken by the Elves. We have already decided to leave proper names alone in our regularized form of Sindarin, using whatever spelling Tolkien himself employed.

As for forms like othrond rather than *othron(n) occurring in Quendi and Eldar (WJ:414), it seems that Tolkien in many cases deliberately quoted a somewhat archaic form of Sindarin in this essay - perhaps because the ultimate derivation of the words was thereby less obscured. Notice that the same source also quotes words still including the vowel ö (spelt "oe", e.g. Eboennin in WJ:387), though ö had merged with e in late Third Age Sindarin. We have already quoted one example from the Etymologies that seems to indicate that the form of "Noldorin"/Sindarin that still had oe = ö, also preserved final -nd: moerilind as a word for "nightingale", only later becoming merilin (see TIN).

Othrond and moerilind/merilin bring us over to common words (as opposed to names), and in the case of such words we must decide upon some sort of standard spelling. Let us again recapitulate what Tolkien says in LotR Appendix E: "nd became nn usually...but remained nd at the end of fully accented monosyllables... In the Third Age final nd in long words had become n from nn, as in Ithilien, Rohan, Anórien." So the rules for how to treat original nd are as follows: 1) leave it alone at the end of accented monosyllables, 2) change it to simple n at the end of "long" words, 3) otherwise change it to nn. Then the only remaining obscurity is just how long a "long" word is. As we have already pointed out, Tolkien includes Rohan among his examples, which would seem to indicate that a word of two syllables is "long" enough (or we would see **Rohann). Then "long" simply means "polysyllabic". But the example aerlinn in RGEO:70 (a post-LotR source!) seems to indicate that a word of two syllables is not long enough for nn to be reduced to n. If we define "long" words as words of three or more syllables, the form aerlinn is just what we would expect according to the rules set out by Tolkien in Appendix E.

It may be that Rohan is a somewhat "bad" example. It is not pure Sindarin anyway, but a Gondorian/Westron adaptation of a Sindarin name (with h for ch; Appendix E also notes that ch "was weakened to h in the speech of Gondor, and that change has been recognized in a few names, such as Rohan, Rohirrim"). In light of the example aerlinn, we may ask: would the normal Sindarin form be Rochann, perhaps usually spelt Rochand because of the general conservatism in the spelling of proper names as opposed to common nouns (aerlinn being an example of the latter)? Just before discussing the development of nd, Tolkien wrote: "Note that consonants written twice, as tt, ll, ss, nn, represent long or 'double' consonants. At the end of words of more than one syllable these were usually shortened: as in Rohan from Rochann (archaic Rochand)." But what language is being discussed here, Sindarin or Westron? The example given, Rohan, is Westronized Sindarin. Still bearing the word aerlinn (archaic *aerlind) in mind, we may conclude that the actual Sindarin form was Rochann or archaic Rochand. At least that was how it was spelt (in RGEO:70 we have aerlinn attested in Tengwar writing!), though perhaps -nn at the end of words had come to be pronounced no differently from normal -n. UT:318 elaborates a little on the name Rohan:

[The] proper [Sindarin] form was Rochand...and [it was] spelt as Rochand, or Rochan...in the records of Gondor... In Rochand the Sindarin ending -nd (-and, -end, -ond) was added; it was commonly used in the names of regions or countries, but the -d was usually dropped in speech, especially in long names, such as Calenardhon, Ithilien, Lamedon, etc.

(From this it is clear that the name Rohan occurring in a Sindarin context should probably be spelt Rochand, this being its "proper" form.) Again, Tolkien insists that the full development -nd > -nn > -n was "especially" characteristic of "long" words only, the examples here listed having three or four syllables.

It seems that we shall have to simply boldly establish some kind of rule for regularizing Sindarin words that agrees at least reasonably well with the examples Tolkien provided (we must realize that no single rule can explain all of his spellings). Suggested rule for standardized spelling of polysyllabic words originally ending in -nd: regularize in accordance with aerlinn and let earlier -nd manifest as -nn at the end of words of two syllables. Only in genuinely "long" words, three or more syllables, is -nn to be further reduced to -n. For us as potential users of Sindarin there are also certain practical considerations: If we were to regularize -nd/-nn to simple -n at the end of all polysyllabic words, we would have to memorize in each case whether -n represents older -nn or was simple -n all along, since original double nn would be preserved before (say) the collective plural ending -ath. So if we were to regularize aerlinn to *aerlin (to go with Rohan rather than Rochann), we would still have to remember that the collective plural should be *aerlinnath with the double n intact. With such a spelling we would be left to wonder whether a word like aran "king" represents older *arann, in which case the collective plural should be *arannath, or whether it was simply aran from the beginning, with a collective plural *aranath. (The two would be pronounced quite differently: accented ARanath and arANNath, respectively - the first of which is probably the correct form.) So for the sake of simplicity, it seems best to retain double nn in writing wherever we would not be stretching Tolkien's intentions too far by doing so. The example aerlinn provides us with a post-LotR Tolkienian precedence that we can appeal to when regularizing his older material. (But we cannot retain -nn in words of three or more syllables without stepping outside Tolkien's LotR-sanctioned spellings; hence we shall have to write perian rather than *periann for "halfling/hobbit", and we must just try to remember that the collective plural is periannath rather than **perianath! The nn of periannath probably does represent older *nd, so the word is relevant for this article.)

We will give a full list of suggested regularized spellings of the words (not including proper names) discussed here in section 4. The attested spellings - sometimes different, sometimes not - are given in brackets, along with the source (entry in the Etymologies, or book and page). Since the forms can thus be immediately compared, we do not asterisk our regularized spellings even where they do differ from Tolkien's spelling:

aerlinn *"holy song" or possibly *"sea-song" [aerlinn, RGEO:70]
ambenn "uphill" [am-bend, ambenn, amben, AM2 and PEN/PÉNED]
anrann (or andrann?) "cycle, age" [anrand, RAD - the element and- "long" would normally be preserved before r-, but d is lost because there is another d in the word (VT41:9) - however, this second d later disappeared in the change -nd > -nn, and it is unclear whether or not an- would then revert to and-.]
adlann "sloping, tilted" [atland, TALÁT - it is a mystery why t in atland is not lenited to d, as both "Noldorin" and Sindarin phonology would seem to demand; suspecting some kind of mistake we tacitly alter t to d.]
camlann "palm of hand" [camland, LAD]
dadbenn "downhill, inclined, prone" [dadbenn, PEN/PÉNED]
dúlinn "nightingale" [dúlind, dúlin(n), DO3/DÔ; dúlinn, LIN2]
idhrinn "year" [iðrind, iðrin, RIN - we prefer the digraph dh to the special letter ð, in accordance with the spelling used in LotR]
merilin "nightingale" [moerilind, merilin, TIN - since this word has more than two syllables, the final -nd is further reduced to -n after becoming -nn. Cf. perian, talagan below.]
othlonn "paved way" [othlond, othlon, LOD]
othronn "underground stronghold" [othrond, WJ:414]
perian "halfling" [perian, LotR, twice in chapter 8 of Book Five, The Houses of Healing]
talagan "harper" [talagand, ÑGAN/ÑGANAD]
tuilinn "swallow" [tuilind, tuilin, TUY]
ulunn "monster, deformed and hideous creature" [ulund, ulun, ÚLUG]

To this list we might add *mindonn "tower" or "isolated hill", if this is to be derived from minitunda as in the entry TUN in the Etymologies: Then we would have an older form *mindond, and Tolkien's spelling mindon would be just an alternative representation of the later form *mindonn. On the other hand, mindon is derived from minitaun in the entry MINI; if so the spelling mindon is proper enough, since according to this scenario this word never ended in anything but a single -n. Pick your choice...but if I needed a Sindarin word for "tower", I would use barad or minas instead, these words being attested in LotR and later sources. It may be that Tolkien in the end came to think of mindon as a Quenya rather than a Sindarin word (it occurs in the Markirya poem: atalantië mindoninnar or mindonnar, "upon fallen towers" [MC:222 cf. 215]).

Ardalambion Index