Reconstructing the Sindarin Verb System

The Reasoning Underlying the Suggested Conjugation

The Present (or Aorist?) Tense
The Past Tense
The Future Tense
Finite Verbs with Endings Attached
The Imperative
The Present (Active) Participle
The Perfect (or, Past Active) Participle
The Past (Passive) Participle
The Gerund
Special Verbs

Appendix A: The Past Tense System Revised?
Appendix B: A Distinction between Aorist and Continuative Tense?
Appendix C: Sindarin Phonology Revised?
Appendix D: "3rd Person" or "Personless" Forms: a Question of Terminology


While the grammar of J.R.R. Tolkien's invented languages generally appears to be relatively simple, one of the more complex features would seem to be conjugation of the Sindarin verb. No explicit description of the Sindarin verb system has ever been published, and while such descriptions are said to exist, we must presently reconstruct Tolkien's intentions from the available examples. This is a complex puzzle, requiring some insight on his general system and at least a basic understanding of the underlying phonological evolution he envisaged.

Several years ago, I had the privilege of acquainting myself with David Salo's interpretation of the Sindarin verb system. Though David's own writings on the subject still have not been presented to the public, he graciously let me present his conclusions in my own Sindarin article on Ardalambion (his results happily replacing some rather inferior attempts of my own at deciphering the inner workings of the Sindarin verb). Later I wrote out a full list of virtually all known Sindarin verbs, a "Suggested Conjugation" demonstrating how these verbs might - perhaps - be conjugated in all the principal forms. Praised by some, this work has also received a fair amount of criticism, since no distinction is made between relatively certain (or even attested) forms and more hypothetical formations. I should emphasize that the Suggested Conjugation was in no way intended as a prestigious "scholarly" work setting out irrefutable conclusions about the Sindarin verb system. It is, as I hoped the title would sufficiently well indicate, a mere suggestion and a help for people who want to write in Sindarin, so that they can treat the verbs in a way which is at least consistent - and which is certainly compatible with the attested examples we do have. I do think it is about the best reconstruction anyone can do, based on what is currently available. While we would all like to be able to conjugate Sindarin verbs with confidence and certainty, such confidence would require the publication of much more material, and while we must cling to the hope that such material will one day appear, its publication seems to be rather less than imminent. For the moment, and in all likelihood for years to come, we must try to do with what we have.

The grammatical rules underlying the forms presented in the Suggested Conjugation are set out in my Sindarin article, but though I cite many examples from the primary sources, the whole presentation may have a slightly "dogmatic" air, presenting a set of rules without fully documenting how they have been deduced from Tolkien's own examples. Writing this section of the article I was, in effect, describing David Salo's finished reconstruction of the entire system rather than documenting the reasoning behind the reconstruction as such. I make no apologies, for the subject is somewhat complicated, and in an article that aims to provide prospective writers with some guidelines rather than setting out a strictly academic or scholarly analysis, I believe a straightforward presentation is preferable. However, I have long wanted to supplement this unified presentation with an article that does refer back to the primary sources and sets out much of the underlying reasoning. Such an article would also be the right place to make a clear distinction between relatively certain conclusions about the Sindarin verb, and more tentative theories which rely on a specific interpretation of very few attested examples (like the Perfect Participle).

The sources: As students of the Sindarin are well aware, the Celtic-theme language of Tolkien's mythos underwent many revisions, both as regards its internal history and its structure and phonology. In the 1917 Lexicon finally published in Parma Eldalamberon #11, the Celtic-theme language manifested as Gnomish, the language of the "Gnomes" (Noldor). While the Lexicon lists quite a few verbs and in many cases also cites past-tense forms, the verb system we glimpse seems rather different from that of Sindarin, and no forms from the Gnomish Lexicon are considered in our present study.
         Parma #13 published a number of "Early Noldorin Fragments", including some writings on the Noldorin verb. "Noldorin" represents an intermediate stage between the earliest form of "Gnomish" and Tolkien's later Sindarin, and the language described in Parma #13 has clearly taken major strides towards LotR-style Sindarin. However, the sketches of the verb system appearing in Parma #13 seem quite tentative, recording the creative process itself, not confidently describing a system Tolkien had already invented and decided upon. Nonetheless, there seem to be certain vague similarities between these "Noldorin" verbs and the system used in Tolkien's later Sindarin, and a few references are made to the Parma #13 material below. However, this "Noldorin" system as a whole is so different from that of Sindarin that we can in no way rely on it to fill gaps in our knowledge of Sindarin as it manifests in LotR and later writings.
         Seemingly, the picture changes completely when we reach the next major conceptual stage: the "Noldorin" of the Etymologies (LR:347-400). By the standards of the field of Tolkien-linguistics, this work provides very many Noldorin verbs, and also quite a few conjugated forms. Except for certain trivial phonological details, the Noldorin verbs of the Etymologies seem quite compatible with the samples of Sindarin that we have. In other words, it seems that virtually all of this late Noldorin material and the verbs attested in Sindarin proper can be ordered into a coherent system, as if from Etym onwards, Tolkien maintained the basic system. Undeniably, it is the Noldorin material from the Etymologies rather than the scarce examples of Sindarin proper that is the main source for this study. This late form of Noldorin does seem so similar to Sindarin that some would risk relying on it regarding features of the Sindarin verb system that are presently unattested in the LotR texts and later samples (such as the infinitives in -o or -i).
         Finally we have "Sindarin proper": samples of Sindarin which Tolkien published under the name of Sindarin in LotR, as well as all post-LotR samples of the main Celtic-theme language. As he was writing the LotR appendices, Tolkien finally decided that this language was not after all the tongue which the Noldor brought from Valinor to Middle-earth, but the vernacular of the native Elves of western Middle-earth (the Sindar or Grey-elves). All the major texts of our Sindarin corpus are listed near the beginning of my regular Sindarin article. Many of them are referred to by name below: The King's Letter (maybe actually very late Noldorin, but whether or not this text predates the Noldorin/Sindarin revision of the internal history of the language is mainly an academic question), Lúthien's Song, A Elbereth Gilthoniel and Gilraen's linnod. For references, see my main Sindarin article.
         Of course, all samples of "Sindarin proper" may not necessarily belong to the same conceptual phase. Certainly Tolkien did not suddenly stop developing his languages once he had published LotR, though he tried to keep them compatible with the samples he had published. Certain post-LotR ideas about the phonology of Sindarin that were published in VT42:27 certainly do not fit the system we will outline here (see Appendix C), but neither do they fit all other samples of Sindarin proper. Tolkien evidently kept experimenting, and our task here is not to map every conceptual development, but to reconstruct a system that fits as much as possible of the available material.

The Infinitive

No certain example of an infinitive is attested in Sindarin proper. The King's Letter appears to use gerunds in place of infinitives. For all we know, Tolkien may have dropped the kind of infinitives we will proceed to discuss below, deciding that Sindarin uses gerunds instead.

The Noldorin of the Etymologies does exhibit special infinitive forms, and though the conceptual validity of the relevant forms in Tolkien's later Sindarin is not entirely certain, these features of Noldorin certainly deserve attention. In Etym-style Noldorin, two infinitive endings appear. One of them is -i, which descends from Old Noldorin -ie (notice that ON trenarie is explicitly called an "inf." form in the entry NAR2); thus these forms are in origin related to the Quenya gerundial or infinitival ending -ie (UT:317). Examples of Noldorin infinitives in -i: blebi "to beat" (root PALAP), degi "to slay" (NDAK; the ON inf. ndakie is given), deri "stop" (DAR), echedi "fashion" (KAT), esgeri "cut round, amputate" (OS), egledhi "go into exile" (LED; read -dl- for -gl- in Sindarin; the ON form et-ledie is given), gedi "catch" (GAT), giri "shudder" (GIR), gwedi "bind" (WED; this verb should evidently read *gwedhi), hedi "hurl" (KHAT), lhefi "lick" (LAB; in Sindarin-style phonology and spelling, read *levi), medi "eat" (MAT), nestegi "insert" (STAK; ON stem nestak- given), ortheri "conquer" (TUR), teli "to come" (TUL), tegi "to lead" (TUK), thribi "to scratch" (SRIP; in Sindarin-style phonology, read *rhibi), tiri "watch" (TIR), treneri "recount" (NAR2). In the case of blebi, degi, teli, tegi and thribi, Tolkien's English glosses explicitly includes the English infinitive marker "to". Also notice that treneri is said to be descended from trenarie, and as noted above, the latter is explicitly called an "inf." form (entry NAR2).
          The verbs that receive the infinitive ending -i seem to be so-called "basic" or "primary" verbs, that is, verbs representing a primitive root with no derivational ending suffixed to the root itself (though some prefixed element may be present, like or- in ortheri). As we shall see, these verbs prefer -i not only as an infinitive ending but also as a connecting vowel before pronominal suffixes, so (post-Tolkien) researchers have also referred to them as I-stem verbs.
          It should be noted that the infinitive ending -i causes umlaut, so that where the ending is added to a verb stem including the vowel a or o, it changes to e (the umlaut product of o was originally ö, pronounced as in German, but this ö later merged with e). So using examples from the list above, the Old Noldorin infinitive ndakie yields Noldorin degi (not *dagi), and the verb tol- "come" (cf. Sindarin imperative tolo, VT44:21) has the infinitive form teli (archaic *töli). In many cases, the o of the verbal stems as they appear in Noldorin/Sindarin is not original, but has been altered from primitive u: The verb tol- itself comes from a root TUL-. However, original o behaves in exactly the same way when umlauted: The infinitive esgeri "[to] amputate" is apparently meant to have developed from *os-skarie, literally "round-cutting", the prefixed element representing the root OS "round, about". Another form of the same verb is given as osgar, and the stem of the verb is also best cited as osgar-. This example (osgar- vs. the infinitive esgeri) also demonstrates that the umlaut a/o > e can be carried through all the syllables of a longer verb stem. However, a prefix is not necessarily umlauted if it is still recognized as such. In the case of ortheri "to conquer" or literally *"to overpower" (entry TUR), the prefix or- "over" is not affected by the umlaut (not **ertheri).
          English dictionaries list verbs by their infinitive form, but this is not a happy solution in Sindarin: The umlaut turning a/o into e also erases information needed to conjugate a verb correctly. Provided with an infinitive form like deri "to stop", we cannot know whether the underlying verb stem is dar-, **der- or **dor-. It is actually dar-, and this stem appears unchanged in, say, the imperative daro! (see Etym, entry DAR), but if we knew only the infinitive deri, the imperative might just as well be **dero or **doro. So for lexical purposes, the verbs that in the infinitive appear as blebi, echedi, egledhi (Sindarin *edledhi), esgeri, deri, gedi, hedi, lhefi (S. *levi), medi, nestegi, ortheri and tegi are better listed as blab-, echad-, egledh- (S. *edledh-), osgar-, dar-, gad-, had-, lhaf- (S. *lav-), mad-, nestag-, orthor-, tog-. Another term for this kind of verbs could thus be consonant stems, since their stems end in a consonant rather than a vowel. One rule for constructing Sindarin (or at least Noldorin) infinitives may therefore be stated as follows: "To stems ending in a consonant, add the suffix -i, and if a or o appears in the verbal stem, change this vowel to e." In the case of gad- "catch", Tolkien himself listed both this stem-form and the infinitive gedi (GAT). Certain other verbs are listed only as stem-forms, with no mention of the infinitive form, e.g. hab- "to clothe", rhib- "to flow like a [?torrent]", nag- "bite", gonod- "count" (KHAP, RIP, NAK, ). We can tell with some confidence that in Etym-style Noldorin at least, the infinitives of these verbs would be *hebi, *rhibi (in Sindarin, read r- for rh-), *negi, *genedi. The latter verb could also be *gonedi to the extent go- was still recognized as a distinct prefix and therefore not umlauted (go-nod- is literally "together-count" = count up).
          The vowels i and e are not affected by the umlaut; thus the verb tir- "watch" has the infinitive tiri, and the initial syllable of echad- "fashion" is unaltered in the infinitive form echedi (TIR, KAT). Cf. also redi "to sow" from the root RED, though this should evidently read *reði = *redhi (many other examples show that original post-vocalic d becomes dh in Noldorin/Sindarin, and the past tense of the same verb is even given as reðant = redhant).

In addition to the infinitive ending -i, Etym-style Noldorin also has the infinitive ending -o, which ending is more common. Indeed most Noldorin verbs listed in the Etymologies are quoted with the infinitive ending -o. The fact that these forms really are infinitives is apparently confirmed by the entry THÊ: Here a verb thio "to seem" is listed, and immediately afterwards we are told that thia means "it appears". Though the English glosses differ somewhat, it would seem that thia is the present (or aorist) tense of the verb that in the infinitive appears as thio.
          The verbs that have infinitives in -o are seen to be mostly derived verbs: These verbs do not represent a naked root, but are formed from a root by suffixing some derivational ending, most often original - or -. In Quenya these endings come out as -ya or -ta. In Noldorin/Sindarin the former ending becomes -ia and the latter -da, -tha, -ta or -na, depending on the phonological environment. Much rarer are the original endings - (> Noldorin/Sindarin -na), *- (> la), - (> -ra) or simple *-â (> a). These verbs may be termed A-stems, since they all end in -a. In the Etymologies these verbs are usually (but not consistently) cited with the infinitive ending -o attached, which ending displaces the final -a. Especially interesting in this regard is the entry LEK. Here a verb is first cited as lheitho "release" (with infinitive ending -o), but then Christopher Tolkien cites "a slip accompanying these etymologies", where the same verb is quoted as leithia without the ending -o attached. (The initial variation lh- vs. l- is inconsequential, reflecting Tolkien's indecision regarding one phonological detail; in Sindarin he settled on l-.) Examples of derived verbs with the ending -o attached:

berio "to protect" (Etym, entry BAR)
anno "to give" (ANA1)
gwesto "to swear" (WED)
gwatho "to soil, stain" (WA3)
mudo "[to] toil" ()
harno "to wound" (SKAR)
glavro "to babble" (GLAM)
bauglo "to oppress" (MBAW)
Of these, berio is an example of an original - stem (the primitive form baryâ- is cited), whereas anno, gwesto, gwatho and mudo demonstrate the various possible outcomes of original - stems (primitive *antâ, *wedtâ, wa3tâ and *môtâ; the form wa3tâ was cited by Tolkien himself). The verbs harno, glavro and bauglo demonstrate the much rarer endings *-, *- and *-; these verbs may be referred to *skarnâ, *glamrâ, *mbauklâ. The original suffix *- seems to be extremely rare as a verb ending; it is apparently attested in the word bauglo only.
          In terms of diachronic phonology, -o is not really an infinitival ending that is attached to a verb and "displaces" a final -a, though the rule may be so formulated for pedagogical purposes (as we did above). In origin, the final -o seen in the infinitive forms of this class of verbs is the final -a of the stem: It is simply another phonological outcome of the original final -â. For instance, Tolkien referred the infinitive matho "to handle" to Old Noldorin matthô-be. Here an explicit infinitive ending -be is present, added to matthô-, which in turn comes from the primitive - stem ma3-tâ (these forms are given in Etym, entry MA3, and the ending -be also occurs repeatedly in Etym - see the section "The Verb" in my Old Sindarin article). In matthô-be, the original long â has already become long ô, the normal outcome of original long â in Old Noldorin. On the way to classical Noldorin, the infinitive ending -be was worn down to nothing, and -ô eventually became a final -o (matthô-be > matho). But while it was still present, the infinitive ending -be had provided a phonological environment where original -â could come out as Noldorin -o, so a distinct infinitive form of the A-stems still existed in the later language. Presently we cannot say exactly how much of this scenario still survived when Noldorin had became Sindarin.
          As noted above, the infinitive ending -i seems to be typical of the "basic" or non-derived verbs, representing an original root with no derivational suffix added. However, there are a few verbs that may seem to be of this kind and yet receive the alternative infinitive ending -o: brono "[to] last, survive" (BORON), dravo "to hew" (DARÁM), faro "to hunt" (SPAR), galo "to grow" (GALA), garo "[to] hold, have" (3AR), naro "[to] tell" (NAR2), sogo "to drink" (SUK), tobo "[to] cover" (TOP), thoro "[to] fence" (THUR; the verbs galo, garo, thoro are cited with a final hyphen, "galo-" etc., but despite the hyphen, these forms seem to be infinitives in -o rather than some kind of curious "O-stem" verbs). As we have just demonstrated, the infinitives in -o historically descend from -â where it was followed by the infinitive marker -be. Therefore, our first assumption would have to be that these verbs are examples of the simplest kind of derived verbs, the ones that originally only showed the simple ending -â. Such simple A-stems do exist in the language. For instance, from the root LUT we have lhoda "float", the infinitive form of which is almost certainly *lhodo (in Sindarin, read l- for Noldorin lh-). Should we then assume that brono, dravo, faro etc. are the infinitive forms of simple A-stems *brona-, *drava-, *fara- and so on? The infinitive form faro "to hunt" Tolkien indeed referred to Old Noldorin (s)pharóbe, and the (s)pharó- part of this infinitive form most likely goes back on a simple A-stem *sparâ-. The infinitive naro "[to] tell" Tolkien also referred to an Old Noldorin form with the ancient infinitive marker -be: naróbe (see Etym, entries SPAR, NAR2 - naróbe is most surprisingly translated "he tells a story" though all other evidence suggests that it is rather an infinitive "to tell a story"; the gloss provided should not be taken literally). The two first syllables of the form naróbe suggest an original simple A-stem *narâ-.
          So some of the verbs listed above must indeed be examples of the simplest kind of A-stems. Others may have a more complicated history behind them. The same entry that lists the form brono "[to] last, survive" also provides an Old Noldorin infinitive bronie of the same meaning. The latter should have produced Noldorin (*bröni >) *breni, which would indeed be the most typical infinitive of a consonant stem bron-. When we see brono instead, this may be taken as an indication that the very frequent infinitive ending -o has started to spread from the A-stems (where it is historically justified) to the less numerous consonant stems (where -o can be present only by analogy with the true A-stems).
          In the entry GAR, Tolkien mentions a verb gar- "hold, possess": clearly a "basic" verb, or consonant stem. Then he struck out this entry, but the root GAR- apparently remained conceptually valid, for it is also mentioned under 3AR. Here an infinitive garo "hold, have" is mentioned. If the consonant stem gar- listed under GAR remained valid even though the entry as such was deleted, then garo would seem to be another example of a basic verb with an infinitive in -o rather than -i. It should also be noted that after mentioning the infinitive garo, Tolkien also cited the form gerin "I have, hold". As we will demonstrate below, a verb with infinitive in -o would normally be expected to have a present-tense form in -a, or when followed by the ending -n "I", -o-. So based on the infinitive form garo, the 1st person present tense "I have/hold" would normally be expected to be **garon. The actual form gerin is rather what we would expect if this were a consonant stem gar-, and it would seem that this is precisely what it is: This verb is merely exceptional in forming its infinitive in -o rather than -i.
          A comparable case could be sogo "to drink" (SUK). A verb with such an infinitive would normally be expected to appear as **soga in the (endingless) present tense, but Tolkien actually mentioned the form sôg, which looks like the present tense of a consonant stem sog-. This could be cited as possible evidence that that by analogy, the infinitive ending -o is beginning to spread from the A-stems to the consonant stems, but it is also possible that it is rather the form sôg which is irregular here, and that sogo "to drink" is the regular infinitive of an A-stem *soga- (so in my Suggested Conjugation). It is even possible that a present-tense form like sôg is regular for a short A-stem like *soga- (see below concerning the "Mixed Conjugation"), and then sôg would prove nothing at all. We cannot know.
          In VT44:30, commenting on a post-LotR Sindarin text, Bill Welden cites a verb caro, which he glosses as "to do". While the form caro does occur in the text he is commenting on, it is there an imperative, not an infinitive. If Welden, a member of the team editing Tolkien's linguistic manuscripts, hereby confirms that caro existed in Sindarin as an infinitive "to do", it would be highly interesting for several reasons. For one thing, this would be the first attestation of -o as an infinitive ending in Sindarin proper; otherwise there is no attestation of it post-dating the Noldorin of the Etymologies. Moreover, car- "do, make" is properly a consonant stem, as indicated by the pl. present (or aorist?) form cerir also occurring in the text Welden is commenting on (VT44:22): If this were an A-stem verb, we would rather see *carar here. If caro is indeed the Sindarin infinitive form of car-, this would be another example of -o as the infinitive ending spreading to the consonant stems. Car- is not listed as a Noldorin verb in Etym (though the stem KAR- "make, do" does appear there), but the infinitive of such a verb would most likely be *ceri - compare, for instance, deri as the infinitive of dar- "stop, halt" (root DAR). Therefore, my list of suggested conjugations of all Sindarin verbs currently suggests *ceri as the infinitive "to do". Unfortunately, Mr. Welden has not demonstrated any will to clarify this matter.

So to summarize: In the Noldorin of the Etymologies, A-stem verbs have infinitive forms in -o, whereas consonant stems typically have infinitive forms in -i (causing umlaut in the stem it is added to, so that a/o becomes e). However, a few consonant stems seem to form their infinitive forms in -o instead, possibly by analogy with the more numerous A-stems.

Finite verbs: We will now proceed to discuss the various finite forms of the Noldorin/Sindarin verb. In this initial survey, we will mainly (though not exclusively) focus on the verb as it appears with no endings attached. The pronominal endings finite verbs can receive, and how they influence the shape of the verb itself, will be discussed in a special section below.

The Present (or Aorist?) Tense

Sindarin A-stem verbs (with infinitives in -o) are seen to have present-tense forms identical to the stem of the verb itself; i.e., the present tense ends in -a. In the LotR corpus we have penna as the present-tense form "slants" in A Elbereth Gilthoniel (cf. the interlinear translation in RGEO:72). There is little reason to doubt that the stem-form of this verb is likewise penna-, a derivative of the root PEN, WJ:408. (Plausible primitive forms are *pentâ- or *pennâ-; both of these would come out as penna- in Sindarin.)
          With the plural ending -r to indicate a plural subject, the present tense of cuino "to be alive" is found in the relative sentence contained in the name Dor Firn i guinar "Land of the Dead that live" (mentioned in the entry KUY in Etym). The lenition of c to g, caused by the preceding relative pronoun i, is incidental; for our purposes we may cite the form as [c]uinar. There is no reason to doubt that this present-tense form would appear as cuina if the plural ending were removed; this form is indeed attested in yet another experimental name of the Land of the Dead that Live, Gwerth-i-Cuina (WJ:132). This form of the name seems abnormal for several reasons, but apparently it does confirm cuina as the present tense of cuino "to be alive".
          The Etymologies seems to support this reconstruction. No A-stem verb is cited in a form that is explicitly identified as the present tense, but there is the pair of forms already cited: thia "it appears" vs. thio "to seem" (entry THÊ). Apparently thia is the present tense and thio the infinitive. Another possible example of a present-tense form is gwinna, listed in the entry WIN. This entry was eventually struck out, but gwinna was a verb glossed "it fades" or "advesperascit" (Latin for "evening approaches"). This could be a genuine present tense in -a. Of course, many verbs in -a that are listed in the Etymologies cannot be seen as present-tense forms; Tolkien merely wrote down the basic stem-form of the verb (e.g. doltha "conceal", entry DUL - though the present-tense form "conceals" is evidently likewise *doltha). More commonly, he cites A-stem verbs as infinitive forms in -o.
          As for consonant stems, they have no ending in the 3rd person sg. present tense. Where the stem of the verb consists of a single syllable, this one syllable is however seen to be lengthened in the present tense, and in accordance with the normal spelling of Sindarin, a long vowel in a monosyllabic word is marked by a circumflex. Thus the present tense of tol- "come" is tôl "comes" (WJ:254). The present tense of ped- "speak" is [p]êd, attested in lenited form bêd in VT41:11. Immediately after listing the infinitive lhefi "to lick" in the entry LAB in Etym, Tolkien also mentioned an undefined form lhâf; this is most likely the 3rd person sg. present *"licks". In Sindarin, we must read *levi for lhefi (assuming that this infinitive formation is still valid), but the endingless present tense should still have final -f: *lâf. This is simply because final -v is spelt -f in the Roman orthography for Sindarin; *lâf is therefore to be pronounced "lâv", and would be so spelt in Tengwar. This extra complication only occurs in the "transcribed" Roman spelling of Sindarin, but it must be observed in the case of consonant stems in -v. (However, *lav- "lick" seems to be the only attested example of such a verb.)
          In the case of polysyllabic verbal stems, the lengthening of the vowel seen in forms like tôl "comes" does not occur. Thus the verb osgar- "cut around, amputate" (inf. esgeri) has the "3[rd person] sg." present tense osgar, mentioned in the entry OS in Etym. The verb orthor- "master, conquer" (inf. ortheri) likewise has the "3 sg" form orthor, listed in the entry TUR. So we must assume that the 3rd person sg. present tense of a verb like gonod- "count up, reckon" is simply *gonod.
          In the entry SUK, the "3[rd person] sg. [present tense]" of the verb sogo "[to] drink" is given as sôg. Here sog- behaves as a consonant stem, despite having the infinitive sogo instead of *segi. In the same place, the verb garo "[to] have, hold" is said to have the 1st person present tense gerin, pointing to *gâr as the corresponding 3rd person sg. form (if we assume that gar- is a well-behaved consonant stem except for the fact that its infinitive is garo instead of **geri). The question remains how to treat the other verbs that may look like non-derived verbs (consonant stems) and yet receive the infinitive ending -o: brono "[to] last, survive", dravo "to hew", faro "to hunt", galo "to grow", naro "[to] tell", sogo "to drink", tobo "[to] cover", thoro "[to] fence". As we argued above, at least some of these are evidently "simple" A-stems, originally derived by adding the short ending -â to the root. If they behave like the longer A-stems (e.g. penna "slants") they would end in -a in the present tense, so that, say, the infinitive faro "to hunt" would correspond to the present-tense form *fara "hunts". In our Suggested Conjugation it is assumed that garo "[to] hold, have" is an "irregular" infinitive form (irregular because it takes the ending -o instead of -i), and that otherwise, this verb should be treated as a consonant stem gar-. This, it may be assumed, is what Tolkien wanted to indicate by explicitly mentioning the form gerin "I hold, have" here. In the case of other verbs of comparable shape (brono, dravo etc.), my Suggested Conjugation is based on the assumption that these verbs are now treated like regular A-stems, so that they have present-tense forms in -a: *brona "lasts", *drava "hews" etc. Of course, this assumption cannot really be substantiated, and other interpretations equally compatible with the scanty evidence are certainly possible. For instance, brono "[to] last" should perhaps rather appear as *brôn in the present tense, just like the 3rd sg. present tense of sogo "to drink" is given as sôg (see entry SUK). The view taken in the current version of our Suggested Conjugation is that sôg is an irregular form and that this is why Tolkien cited this form separately, but obviously there is no way of being certain. It is difficult to know whether we can feel free to dismiss as "irregular" an example that does not readily fit the "regular" patterns when these patterns are reconstructed from very few examples - some of which could themselves be explicitly mentioned because they are irregular, for all we know.

Finally it may be noted that what we here call the present tense of consonant stems seems to be cognate with Quenya aorist forms. For instance, the form tôl "comes" apparently corresponds to Quenya tule; indeed tule would still be the form of the verb at the stage of the linguistic evolution that Tolkien in the Etymologies calls Old Noldorin (more or less = the "pre-historic Sindarin" of later sources). As we will discuss below, consonant stems show the connecting vowel -i- before endings, as do Quenya aorists: Quenya tulin "I come" corresponds to Noldorin (and probably Sindarin) telin of the same meaning. It is not currently known whether Sindarin makes a distinction between the present/continuative tense and the aorist (like Quenya túla "is coming" vs. tule "comes"). A fuller discussion of this problem will be found below. If there is such a distinction, the "present-tense" forms of consonant stems that we discuss here are probably what Tolkien would rather have termed aorists.

The Past Tense

In two cases, Tolkien in the Etymologies cites the past tense form of A-stem verbs (the verbs themselves being cited as infinitives in -o): ortho "[to] raise", pa.t. orthant (ORO) and tirio "[to] watch", pa.t. tiriant (TIR). Unlike tiriant, the form orthant is not explicitly identified as a past tense form, but it is difficult to imagine what else it can be (see entries SUK and WED for analogical forms in -ant which Tolkien did explicitly call past tense forms). The examples orth[a]- vs. orthant and tiri[a]- vs. tiriant would seem to indicate that A-stem verbs form their past tense in -nt. One may of course ask whether we can feel free to generalize from only two examples: Perhaps Tolkien explicitly mentioned these past tense forms because they are somehow irregular! However, it should be noted that the Etymologies also lists the verb teitho "[to] write" (TEK). Here no past tense form is cited; if it had been in any way irregular or special, we may assume that it would have been mentioned. Yet the past tense of this verb turns up in LotR, in the Moria Gate inscription: Celebrimbor o Eregion teithant i thiw hin, "Celebrimbor of Hollin drew [or, wrote] these signs". This provides some basis for assuming that the past tense of regular A-stems is indeed formed with the suffix -nt. There is other indirect evidence supporting this conclusion: Adding -nt to an A-stem of course produces a word in -ant, and we shall see that some verbs have past tense forms in -ant that cannot be historically justified; they must rather be seen as analogical. Indirectly, this suggests that past tense forms ending in -ant are very common in the language; by overwhelming analogy they have even started to spread to verbs where they did not originally belong. If the very numerous A-stems regularly receive the ending -nt in the past tense, such forms would indeed be frequent. Finally, the form prestannen "affected" given in the entry PERES seems to be the past (passive) participle of the verb presto "to affect" given in the same entry. As we will later demonstrate, Sindarin past passive participles are formed by adding -en to the past tense form. Remembering that in Etym-style Noldorin, the cluster nt becomes nn between vowels, we may discern a past tense form *prestant underlying the past participle prestannen.

This shorter consonant stems are another story. Where they end in any of the voiced stops -b, -d, or -g, they form their past tense by infixing the homorganic nasal before the consonant. Thus we have the infix m before b and n before d. Before g we have the sound of ng as in English song, which sound Tolkien sometimes represented as ñ. However, in the relevant class of past tense formations the infix ñ is simply written n in the normal Roman orthography of Noldorin/Sindarin (since before a velar sound, the correct pronunciation comes naturally to speakers of English; compare the pronunciation of n in English think). There is one more complication: Following a vowel, the voiced stops -b, -d and -g historically descend from unvoiced stops -p, -t, -k (or in Sindarin orthography, -c). Where the nasal infix intruded before these unvoiced consonants, and no other ending followed, their original unvoiced quality persisted. Thus consonant stems ending in -b, -d and -g have past tense forms in -mp, -nt and -nc, respectively. For instance, the verb ped- "say, speak" (cf. imperative pedo in the Moria Gate inscription) descends from pre-historic Sindarin *pet-, and the nasal-infixed past tense reflects this older form: "Said, spoke" is pent (cited in the so-called Turin Wrapper, shown by C. F. Hostetter to D. Salo on October 6th, 1996). Compare the Quenya cognates: quet-, pa.t. quente.

Examples of nasalized past tense forms from the Etymologies:

hant as the pa.t. of of had- "hurl" (inf. hedi, entry KHAT)
trevant as the pa.t. of trevad- "traverse" (inf. trevedi, BAT)
echant as the pa.t. of echad- "fashion" (inf. echedi, KAT)
nestanc as the pa.t. of nestag- "insert" (inf. nestegi, STAK)
rhimp as the pa.t. of rhib- "to flow like a [?torrent]" (RIP; read r- for rh- in Sindarin)

The form rhimp listed immediately after rhib- in the entry RIP is not explicitly said to be a past tense form, but it can hardly be anything else, given the wording of this entry. (Rhimp intrudes between rhib- and rhimmo, apparently two synonymous verbs "to flow like a [?torrent]"; the partly illegible gloss follows only after rhimmo. Apparently rhimp is best taken as a form of the verb rhib- listed immediately before, and then it can only be the past tense.)
          In the entry NDAK, the Old Noldorin infinitive ndakie "to slay" and the corresponding nasal-infixed past tense ndanke are listed. Ndakie is shown to yield later Noldorin degi "to slay". The later form of ndanke is not given, but it would be *danc, underlying the attested past participle dangen "slain" (< early Lindarin *ndankênâ; compare Tolkien's "reconstructed" past participle form tháurênâ in the entry THUR).
          One past tense form sog- "drink" (inf. sogo) is given as sunc in the entry SUK. This entry-head explains why the stem-vowel seemingly shifts from o to u in the past tense form: The original root indeed had the stem-vowel u, and while this vowel in very many cases became o in Noldorin/Sindarin, this shift did not occur before nasals. Contrast, for instance, two derivatives of the stem LUT "float": the verb lhoda of the same meaning, and the noun lhunt "boat" (in Sindarin, read l- for lh-). So actually it is the past tense sunc that preserves the original quality of the stem-vowel, whereas it has changed to o in most other forms of the verb (like the infinitive sogo). However, since Tolkien in at least one post-LotR source cited the Elvish root for "drink" as SOK rather than SUK (VT39:11), it may not necessarily be a mortal sin to let the past tense be *sonc instead of sunc. (The "U-stem" verbs are listed and further discussed in the section Special Verbs below.)
          There is also one example of nasal infixion in a verb the stem of which originally ended in -d: From the root WED comes the verb gwed[h]- "bind" (inf. gwed[h]i given) with past tense gwend or gwenn. (The latter is meant to be an assimilated form of gwend. In Tolkien's later Sindarin we can perhaps only have gwend, for in Appendix E to the LotR, Tolkien commented on the change from nd to nn and wrote that it did not occur "at the end of fully accented monosyllables".) I cite the verb gwed[h]- in this fashion because in the entry WED, it appears as gwedi "to bind". However, this must be an error for *gwedhi-, whether Tolkien or the transcriber is to be blamed: Many other examples show that following a vowel, original d - as in the root WED - becomes dh (sc. the sound of English th as in these clothes, sometimes expressed by the special letter ð in Tolkien's writings; "gwedi" is possibly a misreading for *gweði in Tolkien's manuscript). Only in the nasal-infixed past tense form gwend, where the intruding nasal shields the original d from the preceding vowel, can it escape the change d > dh. An alternative past tense form of this verb is indeed given as gweðant, or in normalized spelling gwedhant.
          In the case of polysyllabic verbs ending in -dh, the corresponding past tense would end in -nd in older or "middle" Sindarin, but according to the system of regularization adopted in my articles, -nd at the end of polysyllabic words appears as -nn (see the article The Question of ND or N(N) here). Hence our Suggested Conjugation presents *nelenn (for older *nelend) as the past tense of neledh- "go in". For all we know, *neledhant may also be a possible past-tense form in late Sindarin.
          The past tense form gwend seems to be the only good example we have of the past tense form of a verb derived from a root originally ending in a voiced stop, which is then preserved following the infixed nasal. The Noldorin verb lhefi "to lick" may be updated to Sindarin as *levi, assuming that the genitive forms in -i remained conceptually valid (notice that the letter f is meant to represent [v] also in Tolkien's spelling of the Noldorin form). The original root is LAB, so lhefi/*levi would come from older *labie. If this verb lab- formed its past tense by means of nasal infixion, the resulting form would be *lambe, early Sindarin *lamb (Noldorin *lhamb). However, the final cluster -mb early became -mm, -m, so in classical Sindarin the past tense of *lav- "lick" may be *lam. Our Suggested Conjugation lists this form, but it is not certain. (In Quenya the corresponding past tense is not formed by means of nasal infixion; the quite different formation láve "licked" appears with a prefix in Namárië. If láve reflects Common Eldarin *lâbê, and the corresponding Sindarin past tense also descended from this form, "licked" would translate into Grey-elven as *law!) Except for the one verb *lav- "lick" from the root LAB, there does not appear to be any attested consonant stems ending in -v derived from original -b (and even *lav- itself is only "attested" as a thoroughly updated form of Tolkien's Noldorin verb lhefi, though as noted above, the letter f is surely meant to express the sound of [v] here). However, if the Quenya verbs tyav- "taste" and lav- "allow" have Noldorin/Sindarin cognates *cav-, *dav-, the past tense forms of these verbs could plausibly be *cam, *dam (archaic *camb, *damb), these forms representing nasal-infixed versions of the original roots KYAB, DAB.
          Consonant stems ending in -v derived from original -m (instead of original -b) would likewise have past tenses in -m, though in their case there would be no archaic forms in -mb. No consonant stem in -v from original -m is attested, though there is no reason to assume that they could not exist in the language. Their likely behaviour is indirectly attested in the case of dravo "to hew" from the root DARÁM; this verb has the past tense dram[-]. (It is only attested with a pronominal suffix: drammen *"I hewed"; without the suffix -n "I" this would be *dram, since double -mm is simplified to -m finally.) Dravo itself is not a consonant stem (notice infinitive in -o rather than -i), but as I shall argue below, this verb belongs to a class of verbs that seem to behave like consonant stems in the past tense.
          The past tense forms of consonant stems in -r and -l is only indirectly attested. We would expect stems in -r to have past tense forms in -rn, e.g. *tirn as the past tense of tir- "watch, guard"; *tirn is the obvious cognate of the Quenya past tense tirne (see entry TIR in Etym; this is the stem tir- + the past tense ending -ne, in Noldorin/Sindarin worn down to -n). The past tense *tirn underlies the attested past participle [t]irnen "watched, guarded", lenited dirnen in Talath Dirnen "Guarded Plain" (so in the published Silmarillion; variant Dalath Dirnen in Etym, entry TIR).

NOTE: Regarding the verb "watch", the entry TIR- in the Etymologies reads, in part: "N tiri or tirio, pa.t. tiriant." Here, tiri would seem to be the infinitive form of a consonant stem tir-, whereas tirio is the infinitive of an A-stem *tiria-. It is my opinion that tiriant is intended as the past tense of tirio only, whereas the past tense of tiri is simply not stated. It could thus be *tirn, as theorized above. Carl F. Hostetter argues that tiriant is intended as the past tense of both tir- and the synonymous A-stem tiri[a]-, but it is safe to say that if tiriant is to be the past tense of a primary verb like tir-, such a preterite could not be a historically justified form.

          For the past tense of consonant stems in -l we must likewise fall back on a past participle: the form hollen that forms part in the name Fen Hollen mentioned in LotR. It occurs in the chapter The Siege of Gondor in Book V: "They came at length to a door... Fen Hollen it was called, for it was ever kept shut save at times of funeral." The name thus seems to mean "Closed Door", this translation occurring later in the chapter (as for the element fen "door", compare fennas "doorway" in the Moria Gate inscription). Hollen "closed, shut" would seem to be the lenited form of [s]ollen, and removing the participial ending -en points to *soll as the past tense of a verb *sol- "close". The form *soll also makes sense within the diachronic framework, for an earlier past tense form *solne (with the old past tense marker -ne) would indeed come out in this way: As for the cluster *ln becoming ll, compare a word like ndulna "secret" yielding Noldorin doll "obscure" (Etym, entry NDUL).
          For the past tense of consonant stems in -n there are no examples, direct or indirect. However, it is generally assumed that the past tense of (say) Quenya cen- "see" would be *cenne, since adding the past tense ending -ne to a stem like cen- would not result in any impossible cluster. The Sindarin form corresponding to Quenya *cenne would be *cenn. So as long as there is no other evidence, it may be assumed that consonant stems in -n have past tense forms in -nn, the last -n being a remnant of the past tense ending -ne. (Incidentally, the verb cen- does seem to appear in Sindarin as well as in Quenya; the gerund cened is attested in a compound in RS:466.)
          Some consonant stems are seen to have acquired analogical past tense forms, constructed simply by adding -ant to the root. The historically correct past tense of gwed[h]i "to bind" would be gwend, which form is given in Etym (WED), but there it is also said that gwend was later replaced by gwedhant. Another example is the verb sog- "drink"; its past tense form was originally sunc, but the form []sogant was also used (SUK).

NOTE: In the entry SUK as printed in LR, the past tense forms of sog- are cited as "sunc, asogant". The form "asogant" however violates Noldorin/Sindarin phonology: between vowels, s would become h. Moreover, the prefix a- seems quite superfluous, has no parallels anywhere and can hardly be accounted for in historical terms. Likely, Tolkien actually meant to cite the past tense forms as "sunc, and sogant", alternatively "sunc, or sogant". However, the conjunction "and" or "or" may be only a small doodle in his manuscript, and the transcriber appears to have read it as "a", perceiving this vowel as part of the following word so that the reading "asogant" arose. I here treat the form sogant as attested, but it will be cited as []sogant, as above.

In addition to the A-stems and the consonant stems, there may seem to be a small sub-group of verbs that share some characteristics of both. This group consists of the simplest A-stems, where the short ending -a (not a longer ending like -da/-tha/-na or -ia) has been added to the original root. Furthermore, there must only be a single consonant before this ending, so a verb like erch[a]- "prick" would not be included (inf. ercho in the entry EREK-; this entry-head suggests that this verb is to be derived from *er'kâ-).
          One example of the kind of verb we are discussing would be dravo "to hew", derived from the root DARÁM: primitive *d'ramâ-. The present tense of such a verb can hardly be anything but *drava, but what about the past tense? According to the rules so far formulated, it would be **dravant, but in the relevant entry in Etym, Tolkien indicated that the actual form is dram[] (attested with the ending -n "I" attached; drammen seems to mean *"I hewed"). The form dram[], dramm- would descend from *d'ramne-, with the old past tense ending -ne added directly to the root; the short ending -a does not appear in the past tense. Another example: In the entry NAR2, Tolkien mentions a verb which in the infinitive appears as naro "tell", evidently the infinitive of a simple A-stem *nara-. But the past tense is evidently not **narant; Tolkien cited the Old Noldorin past tense narne, lacking the middle vowel of the infinitive form naróbe (whence naro). Tolkien did not mention the later Noldorin/Sindarin form of narne, but it would come out as *narn, as if it were the past tense of a consonant stem **nar- (with infinitive **neri instead of the actual form naro). Compare our argument above, that we have indirect evidence suggesting that the past tense of the primary verb tir- is *tirn.
          These verbs, though properly A-stems, seem to form their past tense as if they were consonant stems: The short ending -a simply drops out in the past tense. These simplest A-stems may seem to constitute what I call a mixed conjugation in my Sindarin article, since they are treated as consonant stems in the (endingless) past tense, but otherwise (presumably) behave like other A-stems. When endings are to be added to the past tense form, the verbs belonging to the mixed conjugation seem to prefer the same connecting vowel as other A-stems (namely -e- instead of -i-); see below.
          To the Mixed Conjugation may also be assigned a few verbs that are derived from nouns or adjectives by means of the old derivational ending -, which (in the Noldorin infinitive form) has evolved into -do. For instance, the verb tangado "to make firm" (TAK) seems to be in origin a causative formation incorportating the original cognate of Quenya tanka "firm" (the Quenya cognate of this Noldorin verb would be *tankata-). Other verbs of this shape are gannado "play a harp" and lhathrado (Sindarin *lathrado) "listen in, eavesdrop" (ÑGAN, LAS2). We assume that the forms cited are the infinitives of A-stems like tangad[a]-, gannad[a]- etc. Though the final -da (infinitive -do) of these verbs is the descendant of a longer derivational ending -, these verbs are similar to the simplest A-stems (root + original -â) in that both categories of verbs end in a single, short vowel + a single consonant + a final -a (or in the infinitive, -o). We therefore theorize that these verbs would also form their past tense by dropping the final -a and treating the remaining part of the verb as a consonant stem. Thus, the past tense of verbs like tangado and gannado would likely be *tangant, *gannant rather than ?tangadant, ?gannadant. In the case of the verbs lhimmid "[to] moisten" and nimmid "to whiten", Tolkien explicitly listed the past tenses nimmint, lhimmint (LINKWI, NIK-W-). The only remaining mystery is why these verbs are cited as seeming consonant stems lhimmid, nimmid and not as infinitives *nimmido, *lhimmido (or as pure A-stems *nimmida-, *lhimmida-). There can be no doubt that these two verbs by their derivation are entirely similar to a form like tangado. For instance, nimmid "whiten" is derived by adding the causative ending - to the old adjective ninkwi "white" listed by itself in the entry NIK-W-; also compare the Quenya cognate ninqitá- (in later spelling ninquitá-). If the ending -o does belong to a verb like tangado, it is difficult to understand why it does not belong to these two verbs as well. In my Suggested Conjugation, it is assumed that the forms "nimmid" and "lhimmid" simply represent an elliptical annotation of the full forms *nimmido, *lhimmido (somewhat like the cognate of Quenya linga- "hang, dangle" is cited as gling in the entry LING; this must be elliptical for *glinga-). Thus, I list these verbs as A-stems *nimmida-, *limmida- (with l- for lh- because of the normal updating to Sindarin phonology). Possibly pannod "fill" (stem KWAT) also belongs to this group of verbs, but for Neo-Sindarin purposes the synonym pathro can be used; the latter would seem to be the infinitive of a regular A-stem, and it is easier to predict its various tense-forms. (As for pannod "fill" itself, some would also connect it to infinitives in -od occurring in Tolkien's pre-Etym versions of Noldorin, like tangod "to fix" in PE13:131. If so we may be dealing with a leftover form from an earlier conceptual phase.)

NOTE: By its form, the verb athrado "to cross, traverse" (RAT) may also seem to belong to the same class as nimmid- etc., but athrado is evidently rado "to make a way" with the prefix ath- "across" (cf. AT(AT)), so the original derivational ending - is not here present. Nonetheless, since rado may seem to belong to the Mixed Conjugation by its form, the past tense would probably follow the same pattern: *athrant rather than ?athradant.

Past tenses in -AU-: Another small sub-group of verbs are the past tense forms involving the diphthong au, sometimes only indirectly attested because it turns into a monophthong o in some positions. Among the explicit examples we have daul as the surprising past tense form of doltha "conceal". Daul seems to be an A-infixed form of the original root DUL, whereas doltha must descend from *dultâ-. Sometimes verbs in (original) - lose this ending in the past tense; compare Quenya onta- having the past tense óne, coexisting with the more regular form ontane (Etym, entry ONO). It should be noted that Tolkien marked the form daul as archaic or poetic, so in "modern" usage, the verb doltha would perhaps have the more regular past tense *dolthant.
          The past tense of thoro "[to] fence" is evidently *thaur, formed by A-infixion of the original root THUR. The form *thaur is only indirectly attested: it underlies the past participle thoren, which Tolkien referred to tháurênâ. In the participial form thoren, the diphthong au becomes o because of the added syllable. (In the entry DUL, the past passive participle corresponding to the past tense form daul is likewise given as dolen, evidently < *daulênâ.)
          In other cases, Noldorin/Sindarin au does not originate by A-infixion (as in the case of daul, *thaur vs. the original roots DUL, THUR). Rather it descends from original long â. Gilraen's linnod has onen for "I gave". The underlying endingless form "gave" is probably *aun, which is simply the cognate of the Quenya past tense áne (attested in Tolkien's early material: QL:31). Compare, say, naur "flame" as the Noldorin/Sindarin cognate of Quenya náre (Etym, entry NAR1). When endings are added, *aun becomes on-, as in onen "I gave". Compare daul vs. dolen. (Indeed the latter word could by its form evidently be the 1st person past tense "I hid" as well as the past participle "hidden", and onen could likely be the past participle "given" as well as the 1st person past tense "I gave".) - In some editions of LotR, the first word of Gilraen's linnod has a long vowel: ónen. This ó would maintain the prosodic length of the underlying diphthong au. However; the diphthong au does not always produce long ó when it turns into a monophthong; cf. a word like caun vs. its plural form conath (PM:362). Therefore, the form onen with a short o is not necessarily erroneous.
          The verb anno "to give" occurs in Noldorin (ANA1), and it is now also attested in Sindarin proper; the imperative anno occurs in Tolkien's Sindarin Lord's Prayer (VT44:21). The past tense form *aun seems irregular; normally we would expect anno to have the past tense **annant. However, this form becomes rather clumsy when further endings are added ("I gave" would be **annannen! - or contracted **annen). Therefore, it is not surprising that Tolkien went for the less common past tense formation *aun instead (< *ânê). It may be noted that anno "to give" is the cognate of Quenya anta, so this is another case of a -ta verb dropping this ending in the past tense, the pa.t. form being constructed directly from the root instead. Compare one Quenya example already mentioned: onta- with past tense óne, formed from the root ONO. Quenya anta- "give" [= Sindarin inf. anno] with past tense áne [= S. *aun] would have exactly the same relationship to the root ANA1, though in this case neither the Quenya nor the Noldorin past tense is listed in Etym.
          Finally we may consider the verb trenar- (inf. treneri "tell"), the past tense of which is given as trenor (NAR2). It seems likely that this form represents older *trenaur, and that the diphthong *au descends from an original long â (cf. *aun vs. Quenya áne). At the end of a polysyllabic word, au is quite regularly reduced to o. Ultimately, the past tense (*trenaur >) trenor is likely meant to descend from *trenârê. This past tense formation may be compared to one Quenya form occurring in Galadriel's Lament: unduláve "down-licked" (i.e. "covered") as the past tense of undulav-: the primitive form would be *undulâbê. Notice the lengthening of the stem-vowel of the original roots LAB "lick" and NAR2 "tell, relate"; such lengthening seems to be characteristic of this class of past tense formations.
          Besides trenor as the past tense of trenar-, the entry NAR2 in Etym also mentions a form printed as "trener". I cannot make sense of this form. Original a could become e by umlaut, but there is nothing to cause umlaut here. Possibly this "trener" is a misreading for *trenar in Tolkien's manuscript. (As we shall also have to mention later, the vowels e and a are sometimes difficult to distinguish in Tolkien's handwriting. Compare the entry KHAL2, where the transcriber tentatively reads one Noldorin word as "orchel" but admits that the e is uncertain. The Old Noldorin form orkhalla indicates that it should read orchal, and this form later turned up elsewhere: WJ:305, note 48.) Older *trenârê could perhaps produce both trenor and *trenar just like the old form katwârâ yields both cadwor and cadwar (KAT). However, *trenar would normally be expected to be the (3rd singular) present or aorist tense of the consonant stem trenar-, not the past tense. I would recommend the past tense trenor to writers, and let the problematic form "trener" rest in peace.

The Future Tense

The future tense is not attested in Etym-style Noldorin, and all the relevant evidence comes from Sindarin proper. There are only two or three attestations of the future tense, but what evidence there is would indicate that it is formed with the ending -tha. The only attestation of this ending with no further additions comes from the King's Letter: anglennatha "with approach". This is normally taken as an A-stem anglenna- "to approach" + the future-tense ending -tha. The verb linnathon "I will chant" (RGEO:72, occurring untranslated in A Elbereth Gilthoniel in LotR) is almost certainly an A-stem *linn[a]- "sing, chant" (compare linnon *"I sing" in Lúthien's Song, The Lays of Beleriand p. 354). If we drop the pronominal ending -n, the form linnathon would appear as *linnatha; see below concerning the shift from -a to -o- before the pronominal ending -n "I".
          The form estathar occurs in the King's Letter in the phrase i sennui Panthael estathar aen, translated "who ought to be called Fullwise" (SD:126). It is my opinion that in isolation, the form estathar would mean *"they will call", and that the literal meaning of the whole phrase is something like *"whom they preferably would call Fullwise" (sennui could be an adverb *"really, preferably" and aen a mode particle indicating that we are dealing with a hypothetical situation rather than an objective future event: "would" rather than "will"). However, the exact interpretation of this phrase has always been controversial (see the next section, in the discussion of the plural ending -r). Most people would however agree that estathar points to *estatha as the simplest future tense of a verb esta- "name, call" (attested in Quenya; see the entry ES in Etym).
          It is unfortunate that all the attested forms of the future tense seem to be derived from A-stems. It is not clear how the ending -tha would be added to consonant stems (if they indeed receive the same ending). In some cases it could be added to a stem without creating impossible clusters, e.g. ?cartha as the future tense of car- "make, do". However, since the th of the ending -tha most likely represents an older cluster (*tt or *ts?), impossible clusters would indeed have existed at an earlier stage if an ending beginning in such a combination had been added directly to consonant stems. In all likelihood, some connecting vowel is regularly slipped in between the consonant stem and the ending -tha. As will be shown below, in the present (or aorist) tense the vowel -i- is added to consonant stems before pronominal endings are suffixed. One educated guess might therefore be that the same connecting vowel is also inserted before the future-tense ending -tha. Where possible, the vowel -i- would umlaut the stem-vowel of the verb, just like -i as an infinitive ending does (see above). Of course, -i as an infinitive ending is by its origin quite distinct from this connecting vowel, but if the future-tense form of this class of verbs is formed like we have theorized, the grammatical rule could for pedagogical purposes be stated like this: The future tense of verbs with infinitives in -i is formed by adding -tha to the infinitive (irrespective of whether the Noldorin infinitives in -i are really conceptually valid in Sindarin). For instance, the future tense of dar- "stop, halt" (infinitive deri) would be *deritha. Notice how the stem-vowel of dar- would be umlauted.
          The forms given in my Suggested Conjugation are based on this assumption, but we cannot be sure. Another plausible suggestion is that the future-tense ending would appear in the longer form -atha when added to consonant stems; if so the future tense of dar- would rather be *daratha (with no umlaut of the stem-vowel). Ivan Derzhanski apparently assumes that the future tense of tir- "watch" is *tiratha (lenited and suffixed dirathar in the title of an article he published in VT38: Peth i dirathar aen, apparently intended to mean "A word that should be observed". He imitates the syntax of the controversial i sennui Panthael estathar aen-phrase from the King's Letter, a risky operation since no one can be quite sure what most of the individual words mean, but the underlying future-tense form *tiratha may not be equally controversial.)  

Finite Verbs with Endings Attached

Finite verbs may receive endings for pronouns or number. We will survey the known endings; unfortunately they do not add up to a quite complete pronoun table, and we cannot be sure that all of them really belong to the same conceptual stage.  
          The ending -n for 1st person sg: In Sindarin as in Quenya, -n is the pronominal ending "I". This is the best attested of all the pronominal endings. In Sindarin proper it is attested in the words nallon "I cry" (Sam's Invocation), linnon *"I sing" (Lúthien's Song), linnathon "I will chant" (A Elbereth Gilthoniel), onen "I gave" and ú-chebin "I do not keep" (Gilraen's linnod). There is also avon "I won't" (WJ:371). In the Noldorin of the Etymologies, the same ending -n may be observed in gerin "I hold, have" (3AR); it probably also occurs in the untranslated forms drammen *"I hewed" (DARÁM), hennin *"I hurled" (KHAT) and sogennen *"I drank" (SUK; as we will argue below, the latter form should probably read *sogannen). The Old Noldorin ancestor of this pronominal ending, namely -ne, occurs in the form yurine "I run" (YUR). This -ne would presumably come from *-ni at an even older stage; compare the stem NI2 "I" also listed in Etym.  
          The ending -m for 1st person pl: The ending -m is seen to signify "we". In our attested examples, this is an exclusive "we", not including the person(s) addressed. Quenya makes a distinction between inclusive and exclusive "we", and at the time our Sindarin examples date from, the relevant Quenya ending was -mmë (though Tolkien later changed it to -lmë). It seems highly likely that the -m of our Sindarin examples is intended to be the direct cognate of the Quenya exclusive ending -mmë: avam "we won't" (WJ:371), gohenam "we forgive" (VT44:21). Of course, it is also possible that the distinction between inclusive and exclusive "we" was not observed or maintained in Sindarin. It may be noted, however, that such a distinction did exist in pre-Etym Noldorin: at one stage, *"we slew" was dengim if "we" is exclusive, but [deng]int if "we" is inclusive (PE13:130). The form dengim, at least, may well be conceptually valid in Sindarin as well.  
          The ending -ch for 2nd person sg: In an Elfling post of January 22, 2002, Carl F. Hostetter wrote: "Charts can be found showing -ch as 2nd sg." (see As he immediately added, the ending -ch did not always have this meaning (it apparently denoted the 1st pl. inclusive "we" in an earlier conceptual phase, replacing the even earlier ending -nt which Tolkien used in the form [deng]int mentioned above). However, the ending -ch is currently best known to writers as the ending for sg. "you" (so used in Movie Sindarin), and it is so treated here.  
          The plural ending -r (possibly also covering the pronoun "they"): It seems that just as in Quenya, Sindarin verbs having plural subjects normally receive the ending -r. It is attested in the various forms of the name "Land of the Dead that live": Dor Firn i guinar (Etym, entries KUY, PHIR), Gwerth-i-guinar (WJ:71), Dor Gyrth i chuinar (Letters:417). Irrespective of the various mutations of the initial consonant of the verb, these name variants all point to [c]uinar as the plural present-tense form of cuino "to be alive" (Etym, entry KUY). More recently, the ending -r also turned up in Tolkien's Sindarin rendering of the Lord's Prayer, in the line di ai gerir úgerth = literally *"those who do [i.e. commit] sins" (VT44:30). Here, gerir is clearly the lenited form of cerir, this unmutated form being attested in a variant reading (VT44:22). It is a form of the verb car- "do, make"; the imperative caro occurs in the same text.
          It is possible that the ending -r is not merely a plural marker but can also function as the pronoun "they", so that [c]uinar could mean *"they live" and cerir would cover *"they do". One phrase from the King's Letter may confirm this. While [c]uinar and cerir are examples of the present (or aorist) tense, the ending -r is added to what looks like a future-tense form in the phrase Perhael i sennui Panthael estathar aen in the King's Letter (translated "Halfwise who ought to be called Fullwise", SD:126). Because of this translation, some have speculated that -r is here some kind of passive ending, so that estathar means *"will be called" (see, for instance, Ivan Derzhanski's "Bold Hypothesis 1" in VT38:10). There is some evidence for -r as a medio-passive ending in one variant of Quenya (or rather Qenya). I tend, however, to believe that -r is here the same plural ending that is also attested elsewhere, so that estathar rather means *"they will call". The following aen may be a mode particle indicating that the preceding verb describes an unreal or wished-for situation, so that the phrase estathar aen is not a simple indicative "they will call", but rather "they would call". Sennui is perhaps an adjective (cf. the ending -ui) used as an adverb; its meaning here could be something like "preferably". The whole phrase would then mean, literally: *"Halfwise whom they [= people in general] preferably would call Fullwise". This interpretation would confirm that -r can imply "they". (Tolkiens rendering "Halfwise who ought to be called Fullwise" is then seen as a non-literal, idiomatic translation.)

NOTE: In one experimental form of the name of the "Land of the Dead that Live", Gwerth-i-Cuina, the ending -r is missing from the verb even though "the Dead" is plural (WJ:132). It is, however, present in other variants of this name. We may also consider one sentence published in Tolkien: Artist & Illustrator which appears to belong to much the same time-dialect as Etym-style Noldorin: Lheben teil brann i annon ar neledh neledhi gar godrebh, evidently = "five feet high [is] the door and three can go through together". Here the verb gar appears to be the word meaning "can". If we assume a verb *gar- "be able" (in Etym, a verb of the same form means "hold, have"), we might expext the form *gerir or possibly *garar here, with the plural ending attached to go with the plural subject neledh "three". Or is a group of three, indicated by the simple cardinal neledh "3", somehow perceived as a unit? If gar is a singular verb, we should perhaps read *gâr with a long vowel (to go with such forms as tôl, [p]êd etc.)

The four endings -n "I", -m "we", -ch "you" and -r ?"they" can presumably be added to any finite verb. However, the question remains what vowel they should be preceded by.
          As demonstrated above, A-stem verbs end in -a in the present (or aorist) tense, and so do future-tense verbs in -tha. However, the attested examples linnon *"I sing/chant", nallon "I cry" and linnathon "I will chant" indicate that an -a immediately preceding the ending -n is changed to -o- when this pronominal ending is added: We do not see **linnan, **linnathan. However, no such change occurs before the ending -m, as is evident from the example avam "we won't" in WJ:371 (which source also mentions avon "I won't" with the expected change from -a to -o- before -n).
          One attempt to explain why -a becomes -o- before the ending -n "I", but not before -m "we", goes like this: The final vowel of the A-stems was originally a long -â. Before endings including only a single consonant, such as -n, this vowel could remain long and regularly evolved into Sindarin o (via au and long ó). However, before endings including a consonant cluster, the original long -â was shortened to -a- which simply remains -a- in Sindarin. The ending -m most likely represents an older cluster -mm- (or it would have turned into v following a vowel), so before this ending original -â was shortened to -a- and remained -a- in Sindarin. Hence avon "I won't", but avam "we won't". Since the ending -ch "you" (sg.) must descend from an earlier cluster (likely -kk-), we would expect to see -a- before this ending as well: *avach "you won't". (An alternative ending for "you", -g, is also reported to occur in Tolkien's manuscripts. If this is to descend from a single -k-, we would see *avog with -o- as an alternative word for "you won't".)
          This successfully explains some things, but it remains unclear why -a does not become -o- before the plural ending -r as well, e.g. [c]uinar rather than **cuinor as the pl. present tense "live" (as in Dor Firn i guinar). There is no reason to believe that this -r is simplified from an earlier cluster, as is the ending -m "we". Some have suggested that this ending somehow entered the language at a late stage and therefore did not influence its phonological surroundings in the way we would expect. While some minor obscurities remain, we must conclude that a verb like linna- "chant" would most likely go like this in the present/aorist tense: linnon "I chant", *linnam "we chant", *linnach "you chant" and *linnar "(they) chant". In the future tense, the final vowel of the ending -tha would behave in the same way as the final -a of an A-stem in the present tense. So besides the attested form linnathon "I will chant", we would expect to see *linnatham "we will chant", *linnathach "you will chant", *linnathar "(they) will chant". Even ignoring the question of a possible distinction between inclusive and exclusive "we", at least one pronominal ending is missing: plural "you". One educated guess could be that it is *-l, since some variants of Quenya seem to use -lle for this meaning. If this *-l descends from a double -ll- (cf. the Quenya ending), we would likely see *linnal and *linnathal (rather than *linnol and *linnathol) for "you chant", "you will chant".

The present or aorist form of consonant stems is seen to include the vowel -i- before endings. (When they appear without endings, as in the 3rd person singular, these are the verbs that have their stem-vowel lengthened, e.g. tôl "comes".) For instance, from *heb- "keep" we have [h]ebin for "I keep" (attested in negated form ú-chebin "I do not keep" in Gilraen's linnod; the prefix ú- "not" causes lenition of the following consonant). This -i- umlauts the stem-vowel in the preceding syllable(s), just like the infinitive ending -i discussed above, so that the stem-vowels a and o both become -e-. Therefore, the verb car- "do, make" appears as cerir (not **carir) in the Sindarin Lord's Prayer, with -i- as a connecting vowel before the ending -r (VT44:22). Before the primitive root KHEP "keep" was published in VT41:6, it was indeed unclear whether the Sindarin verb "keep" was heb-, **hab- or **hob-, since all of these would become hebi- before endings and only the 1st sg form [h]ebin was attested. A verb like car- "do" would be expected to go like this with endings attached: *cerin "I do", *cerich "you do", *cerim "we do", cerir "(they) do". Tol- "come" would go like this: *telin "I come", *telich "you come", *telim "we come", *telir "(they) come" (compare teli as the umlauted infinitive form "to come", listed in the entry TUL in Etym).
          The verb garo is said to have the 1st person present-tense form gerin "I have, hold" (Etym, entry 3AR; we interpret "garo-" as an infinitive in -o, despite the final hyphen). Above, we saw the form gerin as evidence that this verb is properly a consonant stem (cf. gar- in the deleted entry GAR), and that only the infinitive garo instead of the "regular" form **geri is exceptional. The verb sogo "to drink", which is said to correspond to a 3rd sg (present tense) sôg, may likewise be treated as a consonant stem except for the infinitive form. With endings, sog- would then appear as *segi-, e.g. *segin "I drink". But it is also possible, and maybe more probable, that sogo "to drink" represents a short A-stem sog[a] of the "mixed conjugation". This is the view taken in our Suggested Conjugation, so that "I drink" would be *sogon. (Before the other endings we would see *soga-, and the 3rd sg. form sôg would then be irregular, since if this is indeed an A-stem we would perhaps expect to see **soga even where the verb appears with no ending. Or maybe Mixed Conjugation verbs do regularly drop the final -a in the endingless 3rd sg present tense, contrary to the system currently assumed in our Suggested Conjugation which dismisses sôg as irregular?)

The endings -n, -m, -ch, -r can evidently also be added to past-tense forms of verbs; there are at least some attestations of -n "I". Gilraen's linnod has onen for "I gave". As we have argued above, the endingless past tense form "gave" is evidently *aun, becoming on- in polysyllables. The form onen suggests that the connecting vowel -e- is used before endings in at least some past-tense forms. Other examples from the Etymologies seem to confirm this. In the entry DARÁM one past tense of dravo "to hew" is given as drammen, and while this is not explicitly translated *"I hewed", the pronoun -n "I" seems to be included. As we argued above, the endingless past tense "hewed" would be *dram or technically *dramm (final -mm is reduced to -m in orthography), from older *dramne with the same past tense marker -ne that is also common in Quenya. The form drammen *"I hewed" regularly evolves from *dramnene (for the pronominal ending -ne "I", compare Old Noldorin yurine "I run", entry YUR). There is little reason to doubt that besides drammen *"I hewed" we could also have *drammem "we hewed", *drammech "you hewed", *drammer "(they) hewed".
          Further evidence for -e- as a connecting vowel is provided by the entry SUK in the Etymologies. After listing the form []sogant, past tense of sogo "drink", Tolkien cited a word printed as sogennen (presented as a paranthetical variant of []sogant, which in turn is explicitly identified as a past-tense form). This sogennen is likely the past tense []sogant with the ending -n "I" attached, and a connecting vowel -e- is seen to appear before this ending. (Between vowels, -nt- regularly turns into -nn- for phonological reasons.) However, we would expect []sogant + -e-n to produce *sogannen, not "sogennen" as printed. In all likelihood, "sogennen" is simply a misreading for *sogannen in Tolkien's manuscript; we have already discussed other cases of likely confusion of a with e in Tolkien's handwriting. In this essay, this form is henceforth cited as sog[a]nnen. If we have correctly interpreted this form as *"I drank", this is an important example, demonstating that the (apparently very frequent) past tense forms in -ant appear as -anne- when endings are appended. With endings a 3rd person sg. form like orthant (pa.t. of ortho "raise", ORO) would then turn into *orthannen "I raised", *orthannech "you raised", *orthannem "we raised", *orthanner "(they) raised".
          However, the Etymologies provides one divergent example, where another connecting vowel appears. In the entry KHAT in Etym, Tolkien mentioned the verb that in the infinitive appears as hedi "[to] hurl". Two past tense forms are listed: hant and hennin (actually in the opposite order). Since the stem-form of the verb can be cited as *had- (the infinitive ending -i causes umlaut, hence the inf. form hedi), the form hant is simply the endingless (3rd person sg.) past tense, formed by nasal infixion as explained above. Hennin would seem to be hant *"hurled" with the ending -n attached, hence "I hurled". But in the form hennin, the connecting vowel -i- is used before -n, and hennin can then be derived from hant by observing the normal rules of umlaut a > e caused by a following i (as in the infinitive hedi and the present/aorist tense *hedin) and remembering that intervocalic -nt- becomes -nn-. The form hennin is however still surprising, for -i- as a connecting vowel does not seem to be historically justified. In the diachronic perspective, we would expect -e- to be the more or less universal connecting vowel in the past tense. This "connecting vowel" is simply the vowel all past tense verbs originally ended in, lost in endingless forms (where it was final), but preserved before endings. In Quenya, it appears that all past tense forms still end in the vowel -e, also in forms similar to the Noldorin verb before us. The verb quetin "I say" has the past tense quenten "I said" (both are attested with a prefix in WJ:370, 371). Phonologically, we would expect these forms to correspond to Noldorin/Sindarin *pedin and *pennen, respectively. The form *pedin, while not directly attested, is indeed what we would see according to the rules reconstructed above. However, the attested form hennin suggests that "I said" would not be *pennen, but rather *pennin. How can this be explained?
          To use the attested example hennin *"I hurled", there can be little doubt that historically speaking, the past tense should have been **hannen (< Old Noldorin *khante-ne). The historically unjustified form hennin apparently arose by analogy with the present (or aorist) tense, where the connecting vowel -i- is indeed used: According to the system reconstructed above, the verb had- "hurl" would have the 1st person present tense *hedin. Notice that the verbs which in the past tense preserve the historically "correct" connecting vowel -e- are also verbs that do not show -i- in the present tense. The attested past-tense forms drammen *"I hewed" and onen "I gave" would (according to our reconstruction) correspond to present-tense forms *dravon "I hew" and *annon "I give". In these present-tense forms there is no connecting vowel -i-, and the past-tense forms are then seen to retain the historically justified connecting vowel -e-. It would seem that A-stems (including the simplest A-stems of the "Mixed Conjugation") use the connecting vowel -e- when endings are to be added to the past tense form, whereas the primary verbs use the connecting vowel -i-, introduced by analogy with the connecting vowel the primary verbs use in the present tense.
          For a long time, the word hennin from the entry KHAT in the Etymologies was the sole example of a past-tense form with -i- as its connecting vowel, and I was somewhat skeptical about generalizing a rule affecting very many verbs based on only one single example (though this situation is far from unique in Tolkien-linguistics!) Now it turns out that the verb *dag- (inf. degi "to slay", NDAK), the endingless past tense of which would be *danc (Old Noldorin ndanke given), had the extended past tense form dengin *"I slew" already in pre-Etym Noldorin (PE13:130). Some of the other pre-Etym forms listed also seem compatible with later Sindarin, like dengim *"we slew" and [deng]ir *"(they) slew". These were the forms predicted in our Suggested Conjugation. Though the verb conjugations presented in PE13 (including the table providing the examples just cited) are mostly not compatible with Tolkien's later system, it would seem that having -i- as a connecting vowel in the past tense of some Noldorin/Sindarin verbs was indeed a long-standing idea: The form hennin does not represent a brief flicker in Tolkien's evolving conception.
          We conclude, then, like this: In the past tense, the most frequent connecting vowel before endings is -e-, but the verbs that have -i- as their connecting vowel in the present (or aorist) tense also show this vowel in the past tense, evidently by analogy. This vowel -i- causes umlaut in the past tense as well as in the present tense; thus the vowel of the verb *had- "hurl" would turn into -e- in the present tense (*hedin "I hurl") as well as the past tense hennin *"I hurled"; compare the attested infinitive hedi. The vowel o would likewise become e (for archaic ö), so while the verb "run" seems to be nor-, the past tense "I ran" would evidently manifest as *nernin (< *nörnin).
          When expanding a past tense form with the connecting vowel -e- or -i- before adding an ending, the past tense stem must often be somewhat modified. The forms sog[a]nnen and hennin vs. the endingless variants []sogant, hant demonstrate how -nt turns into -nn- when this cluster becomes intervocalic. Intervocalic -nd would also turn into -nn-, so the old past tense gwend *"bound" (WED) would with endings appear as *gwenni-.
          Similarly, -nc would become -ng- (as in English song, with no audible g). Sunc as the older past tense of sogo "drink" would presumably appear as *sunge- before endings (*sungen "I drank", *sungech "you drank" etc.) Cf. also dengin *"I slew" vs. endingless *danc "slew", if we dare to bring pre-Etym forms into the discussion (PE13:130). If there are any verbs that in the past tense would end in -ng (formed by nasal-infixing original roots in -G), it should be noted that this -ng came to be pronounced as a simple nasal - as in English sing, without a distinct [g].
          Past-tense forms that end in -m or -mp when no ending is present would show double -mm- before endings. As I have argued above, the form drammen *"I hewed" would likely be *dram without the ending -(e)n "I". Interestingly, Tolkien in the entry DARÁM indicated that there also existed an endingless form dramp (marked as poetic). This form cannot be historically justified (the root would have had to be **DARÁP instead), and it evidently arose by analogy: The group -mp would also turn into -mm- when intervocalic, so the form drammen as such might just as well be formed from an endingless variant dramp. A genuine past tense in -mp is rhimp, pa.t. of rhib- "to flow like a [?torrent]" (rhimp representing a nasal-infixed form of the root RIP). According to the system here reconstructed, this would before endings appear as *rhimmi-, e.g. pl. *rhimmir. (In Sindarin, read r- for Noldorin rh-.)
          When expanding an endingless past tense form, one must also bear in mind that the diphthong au occurring in some past tense forms turns into o when a monosyllabic word becomes polysyllabic. As noted above, the form onen "I gave" is probably formed from an endingless variant *aun "gave". (At the stage sometimes referred to as Middle Sindarin, "I gave" would still be *aunen.)  

The Imperative

By the standards of this field, Sindarin imperative forms are very well attested. In the LotR itself there is tiro "look towards" in Sam's "inspired" cry in Cirith Ungol (translated in Letters:278 and RGEO:72). The Moria Gate inscription reads, in part: Pedo mellon a minno, "say friend and enter". Gandalf's invocation before the doors contains two imperatives: 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). The imperative edro! "open!" (which Gandalf also spoke in isolation) Tolkien added to the Etymologies, entry ETER. The form lasto also occurs in Lúthien's Song in The Lays of Beleriand p. 354, for which we have no Tolkien-made translation, but evidently it is the imperative "listen!" here as well. On their way into Lórien, the one Elf halted the Fellowship with the word daro!, and the entry DAR in Etym confirms that this is the imperative of a verb "stop, halt". Yet other imperatives appear in the text of the the LotR itself in the Cormallen Praise: Cuio i Pheriain anann!... 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)!" It would seem that cuio means simply "live!", though Tolkien here rendered it " long". Eglerio, occurring twice, is the word rendered "glorify!"
          The battle-cry of the Edain of the North, given in UT:65, contains two imperatives: Lacho calad! Drego morn! "Flame Light! Flee Night!" The interjection elo, identified as "an exclamation of wonder, admiration, delight" (WJ:362), is in origin an imperative "look! see!" (WJ:360). There is also ego "be off!" (WJ:365) In VT44:21, in Tolkien's translation of the Lord's Prayer, we have what would seem to be the imperatives of the verbs tol- "come", car- "do", ann[a]- "give" and díhen- "forgive": tolo, caro, anno, díheno. (One could also include no as the imperative of the verb "to be", if no aer i eneth lín is interpreted as "be holy your name" - but since no is the sole possible attestation of any form of the verb "to be", nothing more can be said of it.)
          The material is thus entirely consistent: all known imperative forms are seen to take the ending -o. Referring to the form tiro, Tolkien noted that this imperative 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. This is confirmed by the corpus itself: One Elf cried daro! "halt!" to the entire Fellowship as they were entering Lórien (see LotR1/II ch. 6), but in, say, the Lord's Prayer, a form like díheno "forgive!" is directed to one person only (the Father).
          It should be noted that the ending -o is added to both A-stems and consonant stems. The verb that in the infinitive appears as anno "to give" is the cognate of Quenya anta- and may therefore also be cited as ann[a]-; its imperative form is seen to be anno (VT44:21). The infinitive and the imperative of A-stem verbs thus coincide in form (unless, of course, Tolkien abandoned the Etym-style infinitives in -o in Sindarin proper). On the other hand, consonant stems have distinct forms, since the infinitive has the ending -i (causing umlaut if possible) and the imperative the ending -o (not causing umlaut). One example from the Etymologies (entry DAR) is the consonant stem dar- "halt", which has the infinitive form deri and the imperative form daro! Cf. also teli "to come" (TUL) vs. its impreative form tolo (VT44:21), and tiri "[to] watch" (TIR) vs. the imperative tiro (Letters:427, RGEO:72). But as we have indicated, the conceptual validity of the Etym-style infinitives in -o and -i is not beyond doubt in Sindarin proper.
          In the case of drego "flee!" (UT:65) it is impossible to tell whether this is the imperative of a consonant stem *dreg- or a longer (simple) A-stem *drega-.
          Tolkien's comments on the form ego! "be off!" throw some light on the origin of the imperative ending -o. This ending represents an originally independent imperative particle â, "though this now always followed the verb stem and had become an inflexion" (WJ:365). This derivation indicates that the imperative ending was -au in older Sindarin. Then it became long -ó (also mentioned in WJ:365) and finally short -o. The fact that this was originally an independent particle also explains why it was not lost in Sindarin, as all final vowels were at a certain stage.  

The Present (Active) Participle

As far as I can see, there is no attestation of the active participle of an A-stem verb in Sindarin proper. Such participles are however attested in the immediately preceding conceptual phase, the Noldorin of the Etymologies, and in all likelihood Tolkien carried the same formations over into Sindarin.
          In the entry GLAM-, we have the verb glavro "to babble", evidently the infinitive of an A-stem *glavra-. Then the word glavrol is listed, glossed "babbling". The English gloss is unfortunately ambiguous: is this "babbling" as a verbal noun (i.e., "the act of babbling") or "babbling" as an (adjectival) participle? Another couple of words may throw some light on this: In the entry SWIN, a verb chwinio "to twirl, whirl" is listed, evidently the infinitive of an A-stem *chwinia-. The same entry also lists the form chwiniol. This Tolkien glossed "whirling", but luckily he also supplied the additional glosses "giddy, fantastic" to cover the extended meanings of this word. This demonstrates that chwiniol is indeed an adjectival word, not a verbal noun, and then glavrol is probably also "babbling" as a participle rather than a noun.
          It may also be noted that already in Tolkien's early (pre-Etym) Noldorin, the verb glathra- "to polish" had the active participle glathrol, explicitly identified as such (PE13:126, 129). This would confirm the pattern glimpsed in the Etymologies.
          Forms like glavrol, chwiniol (in Sindarin, read *hwiniol) also make good sense within the diachronic scenario. In Quenya, the ending forming active participles is -la, well known from the "Last Ark" (Markirya) poem. It would descend from primitive *-. In Quenya, the verb corresponding to Noldorin chwini[a]- is hwinya-, "swirl, eddy, gyrate"; both come from primitive *swinjâ-. In Quenya, we would expect the participle of hwinya- to be *hwinyala, for primitive *swinjâ-lâ. In Noldorin, this form evolves into chwiniol (Sindarin *hwiniol) because the long -â- in the second-to-last syllable turns into -o- (via Middle Sindarin -au-). Compare the old 1st person present (or aorist) tense *lindâ-n- producing Sindarin linnon *"I sing/chant". We may conclude that in Sindarin, the active participle of A-stem verbs is formed by changing the final -a to -o and adding -l, as in glavrol "babbling" from glavr[a]- "babble".
         The King's Letter form edregol, translated "in especial", looks like a participle. Since the prefix ed- may mean "forth" or "out" (see entry ET in Etym), it is possible that edregol literally means something like *"outstanding(ly)" = in especial. If so, an A-stem verb *edrega- "stand out, be special" may be inferred. It is, however, too hypothetical to be included in our Suggested Conjugation list of Sindarin verbs.
          The participles of consonant stems are more poorly attested. In pre-Etym Noldorin, such verbs could also receive the ending -ol, e.g. madol as the active participle of mad- "to eat" (PE13:129, 131). Such forms are difficult to justify historically; maybe Tolkien meant -ol to have spread to the consonant stems by analogy with the more numerous A-stems. Intriguingly, the form madel is listed as the "ON" (Old Noldorin) participle "eating" in PE13:131, though this is not the same language as the "Old Noldorin" of the Etymologies (which is far more archaic and much closer to Quenya). We shall here argue that madel is the likeliest form of the participle "eating" in Tolkien's later Sindarin as well.
          In the LotR itself, the participle of tir- "to watch, gaze" appears as [t]iriel (lenited -diriel in the compound palan-diriel, "gazing afar" in Sam's invocation in Cirith Ungol, cf. the interlinear translation in RGEO:72). Does [t]iriel as the participle of the verb tir- indicate that in Sindarin, consonant stems regularly have active participles in -iel?
          Maybe not. As is very often the case in Tolkien-linguistics, the diachronic perspective is enlightening. Consonant stems are fond of -i- as their connecting vowel, and we would expect a verb like, say, cab- "leap" from a stem KAP to have the primitive participle *kap-i-lâ "leaping". (Compare such a Quenya form as itila "twinkling, glinting", PM:363.) If we carry *kap-i-lâ through the normal sound-shifts, we would expect the Noldorin/Sindarin outcome to be *cabel. (Compare the attested gerundial form of the same verb: cabed from *kap-i-tâ.) Likewise, *mat-i-lâ "eating" should produce Sindarin *madel, and as pointed out above, the form madel is actually attested as "ON" (Old Noldorin) in PE13:131. In other words, consonant stems would be expected to have active participles in -el. The attested participle [t]iriel "gazing" does indeed contain -el, but there is an -i- before it. Can this extra vowel somehow be accounted for?
          Probably. One curious feature of Sindarin is that when an ending is added to a syllable including the vowel i, this vowel is in some instances "echoed" before the ending. For instance, the derivational ending -as (attested in many words) is suddenly lengthened to -ias in the word ínias "annals", derived from în "year" (entry YEN in Etym). The augmenting ending -on (RGEO:73) is likewise well attested in this form, e.g. in annon "great gate" and aearon "[great] sea", but when it is added to sîr "river", the resulting form is not **Síron, but Sírion (name of the great river of Beleriand). The "extra" i turning up in the middle of this name is not the remnant of some original ending that is otherwise lost: The older form of sîr is cited as síre rather than *siri in Etym, entry SIR (where the name Sirion is also mentioned). It seems that in certain environments, an "echoic i" turns up before endings that are added to stems with the stem-vowel i. This seems to occur following the consonants r, l, and n. Perhaps Tolkien's idea is that these consonants were at some stage palatalized following i. Later, the resulting palatal consonants *rj, *lj, *nj turned into ri, li, ni in front of endings, the palatal quality of the consonant splitting from it and manifesting as a distinct vowel i.
          Interpreting [t]iriel "gazing" in this light, this form may not argue against the theory that primary verbs would regularly have participles in -el (< -ilâ), e.g. *madel "eating" (which, though attested in pre-Etym Noldorin, must still be asterisked as a Sindarin form). This is the system used in our Suggested Conjugation. Phonology suggests that the longer ending -iel would turn up only when this participial ending is to be added to verbal stems in -ir, -il- and -in-. Actually all the relevant verbs found in the published corpus have stems in -ir: besides tir- "watch, gaze" itself, they are fir- "die, fade", gir- "shudder", and glir- "sing, recite poem" (cited as "glin" in Etym, entry GLIR, but as this root indicates, "glin" is a misreading for the infinitive form *gliri). Their active participles would then be *firiel, *giriel, *gliriel (so in our Suggested Conjugation).  

The Perfect (or, Past Active) Participle

Sindarin has (at least) one participle with no direct English equivalent. Wrote Tolkien in RGEO:73, "Actually the form used in the hymn [that is, A Elbereth Gilthoniel] is palandíriel (past part.), 'having gazed afar'." In Letters:427, he likewise defines this form as a "past participle 'having gazed afar'." While the term "past participle" is often used to refer to passive participles, the translation Tolkien offers indicates that he is here discussing a past active participle. In the pre-Etym "Noldorin fragments" published in Parma #13 there are references to a "past participle active" (PE13:131), though its form in this early version of "Noldorin" is hardly valid in Sindarin proper. We will here use the alternative term perfect participle to refer to participles expressing the idea of "having done" something.
          Palandíriel (spelt palan-díriel in LotR) incorporates a lenited form of the verb tir- "watch, gaze, look (towards)". Removing the prefix palan- "far and wide" and undoing the lenition of t to d leaves us with [t]íriel as the participle "having gazed, having watched". Only the lengthened stem-vowel (í rather than i) distinguishes this form from [t]iriel "watching", discussed above. (In RGEO:73, Tolkien quite explictly contrasts the "past [active] part[iciple]" palandíriel with the "present participle" palandiriel; in a footnote he points out that the latter has a "short dir" instead of the long -dír- of the past active participle.) Above we theorized that in [t]iriel, the ending proper is just -el, an extra -i- turning up before it merely because of the phonological environment. However, [t]íriel "having gazed" may well contain a genuine longer ending -iel, which would appear in this form no matter what the verbal stem it is added to looks like. The form [t]íriel vs. the verbal stem tir- would seem to suggest that the perfect participle of a verb of this shape is formed by lengthening the stem-vowel and adding -iel. This process is reminiscent of how the perfect tense of such a verb is formed in Quenya: the stem-vowel is lengthened and the ending -ie is added. The perfect form of tir- would in Quenya be *tírie or augmented *itírie, "has watched". Quite likely, Tolkien meant the ending -iel seen in Grey-elven perfect participles to be related to the Quenya perfect ending -ie; the final -l could be a remnant of the old participial ending *- (manifesting as -la in Quenya and as -l in Noldorin active participles, e.g. glavrol "babbling" discussed above).
          Besides [t]íriel, we may have one more attested perfect participle: the second element of the name Gilthoniel. This name is translated "Star-kindler" (gil- meaning "star"), but Tolkien in some places insists that the name Gilthoniel is somehow a past tense form. In Letters:278 he writes that Elbereth Gilthoniel means "Elbereth Starkindler (in the past tense: the title belongs to mythical pre-history and does not refer to a permanent state)". In RGEO:72, the interlinear translation provided reads simply "O Elbereth Star-kindler", but on the same page, Tolkien also translated Elbereth Gilthoniel as "Elbereth who lit the stars" (emphasis added). May Gilthoniel contain a perfect participle so that it means, literally, *"(the one) having kindled stars"? According to MR:388, the original root "kindle" was than- (or thân-). If we assume the existence of a Sindarin verb *than- "to kindle", its perfect participle would indeed be *thóniel according to the system here reconstructed (shortened as -thoniel in a compound like Gilthoniel)? The lengthening of the stem-vowel, if it is indeed comparable to the lengthening seen in Quenya perfect forms, would be so ancient that it predated the change from long â to long ó in the Noldorin/Sindarin branch (this change is explicitly mentioned in the Etymologies, entry THÔN). Thus older *thâniel- derived from a root than- would come out as *thóniel in Sindarin. Likewise, primary verbs with the stem-vowels e and o would show í and ú (respectively) in the perfect participle, since this is how old long ê and ô came out in Sindarin. Thus the following system emerges:
car- "make", perfect participle *córiel "having made"
heb- "keep", perfect participle *híbiel "having kept"
nor- "run", perfect participle *núriel "having run"
As for the of change of -e- to -í (< old long ê), the verb orthel- "roof, screen above" must be especially watched: When -thel- becomes -thíl- in the perfect participle, the í here emerging would be expected to umlaut the prefix or- to er-, so that "having roofed" would be *erthíliel (older *örthíliel).
          Primary verbs with the stem-vowel i simply show long í in the perfect participle, as in the attested form [t]íriel from tir-, since long î remained unchanged at the stage that saw long â, ê, ô becoming ó, í, ú, respectively.
          This reconstructed system is used in my Suggested Conjugation. It should be noted, though, that in MR:388 Tolkien states that the ending -iel occurring in the name Gilthoniel is a mere "feminine ending" not a participial ending. But since he elsewhere insists that the word Gilthoniel describes one who has kindled stars "in the past tense" (Letters:278), it remains tempting to connect this -thoniel with the "past [active] participle" [t]íriel (Letters:427). It may be that Tolkien, true to his habit, kept reinterpreting forms even after they had been published. It seems likely that at least sometimes (!), Tolkien interpreted the name Gilthoniel as containing a perfect participle "having kindled". Anyway, the form [t]íriel cannot possibly include any feminine ending; this must be seen as a perfect (past active) participle.
          As discussed above, primary verbs with stems in -ir (and -il, -in, if there are any) seem to take the long ending -iel (instead of just -el) in the present participle, as an extra -i- echoing the stem-vowel turns up before the simple participial ending -el. In the case of these verbs, only vowel-length distinguishes the present and the perfect participles, as demonstrated by the attested pair [t]iriel "watching" vs. [t]íriel "having watched". Notice, then, these distinctions:
*firiel "dying, fading" vs. fíriel "having died, having faded"
*giriel "shuddering" vs. gíriel "having shuddered"
*gliriel "singing, reciting" vs. *glíriel "having sung, having recited"
According to the system here reconstructed, other primary verbs would not come so close to coinciding in the perfect participle, e.g. *madel "eating", but *módiel "having eaten". If *thóniel does mean "having kindled", this form would confirm that all perfect participles take the ending -iel; here the i does not echo the stem-vowel (as could conceivably be the case in [t]íriel). Contrast the simple (present) active participle "kindling", which by our reconstruction would be *thanel (< *thanilâ).
          Since [t]íriel and possibly -thoniel (for *thóniel) are the only attested examples of perfect participles, and both of these seem to be formed from consonant stems, we cannot be certain how the far more numerous A-stem verbs would behave in this form. However, above we theorized that the ending -iel may originate from the same perfect ending -ie that is well known from Quenya + a remnant of the ancient participial ending -. If these assumptions hold water, we may look to Quenya for further evidence: how is the ending -ie added to A-stems in that language?
          Basically, the ending -ie seems to displace a final -a. Quenya lantie "fell" and lantier "they fell" (LR:56, 47) are forms of the verb lanta- "to fall" (cf. entry DAT, DANT in Etym). While Tolkien in this case translated the Quenya forms using the English past tense, "fell" rather than perfect "has/have fallen", these forms do seem to be examples of the perfect tense in -ie. (As for the translation provided, Tolkien did state "the forms of past and perfect became progressively more closely associated in Quenya" - WJ:366.) If the Quenya perfect ending -ie is indeed related to the ending -iel manifesting in Sindarin perfect participles, we may theorize that -iel, when added to A-stem verbs, would displace the final -a of the verbal stem. So for a verb like trasta- "to harass" (TARAS), our Suggested Conjugation presents the perfect participle *trestiel "having harassed". Here, there can be no lengthening of the original stem-vowel, since it is followed by a consonant cluster. We would not expect the such lengthening before the single consonants th and ch either, since they descend from older clusters, e.g. bath[a]- "trample" from battô (BAT): These clusters were still present at the time when the lengthening would have occurred.
          Where the stem-vowel cannot be lengthened (and thus made immune to umlaut), the i of the ending -iel would be expected to umlaut the vowel of the preceding syllable, hence *trestiel vs. trasta-. The vowel o would likewise turn into e (archaic ö), e.g. *elthiel (older *ölthiel) "having dreamt" from the verb oltha- "to dream". The Perfect Participle of bath[a]- "trample" would be *bethiel "having trampled"; lach[a]- "flame" would correspond to *lechiel "having flamed".
          In the case of the A-stems of the "Mixed Conjugation", where there is no consonant cluster following the stem-vowel, the characteristic lengthening of this vowel (seen in our attested example [t]íriel "having watched") could presumably take place. We would then expect to see the same change a, e, o > ó, í, ú, respectively, as we predict in the basic verbs. The verb groga- "feel terror" (WJ:415) would then correspond to *grúgiel "having felt terror"; aphada- "follow" would correspond to *aphódiel "having followed" (< *ap-pâtiel-).
          A special challenge is provided by the verbs that originally ended in -, becoming -ia in Noldorin/Sindarin. Here the I-umlaut caused by the ending has neutralized the vowel of the root to which the ending was added. Whether this original root had a, o, u, the Noldorin/Sindarin verb would show e. Examples from the Etymologies include berio "to protect" from BAR, erio "[to] rise" from ORO, and delio "to conceal" from DUL: The ancestral forms of these verbs, irrespective of the infinitive ending -o, can be given as baryâ, *oryâ and *dulyâ (Tolkien himself cited the form baryâ). If the original root had e, the Noldorin/Sindarin verb would preserve it unchanged. There seems to be no attested example, but if Quenya perya "[to] halve" (root PER) had a N/S cognate, it would certainly be *peria- (Etym-style infinitive *perio). Yet the root of a verb of this shape could also be *PAR, *POR, or *PUR, since the first vowel of the ending -ia neutralizes the preceding stem-vowel anyway. (Archaic Sindarin would however show ö rather than e where the root had the vowel o or u: the older form of delio is given as "doelio" = dölio [DUL], and erio would likewise descend from *örio [ORO].)
          The problem is that the ending -iel denoting the perfect participle would likely displace the entire ending - > -ia (just like the perfect ending -ie seems to displace -ya in Quenya), and the stem-vowel can then be lengthened just as in the case of primary verbs. Being long, it would not be umlauted. We must then refer back to the original root to know what vowel should appear in the perfect participle. Where the original stem-vowel was a, the perfect participle would have ó. For instance, beri[a]- "protect" would correspond to *bóriel "having protected" (stem BAR; participle derived from *bâriel-). On the other hand, where the original stem-vowel was either o or u, the Sindarin perfect participle would be expected to show ú, e.g. eri[a]- "rise" vs. *úriel "having risen" (stem ORO, participle derived from *ôriel-). Similarly with deli[a]- "hide, conceal" vs. *dúliel "having concealed", because the original stem was DUL. Before Middle Sindarin ö merged with e in Classical Sindarin, these verbs were easier to distinguish from the verbs in -ia that were derived from roots with the stem-vowel a, like beri[a]- above. Yet already in Middle Sindarin there would be no distinction between verbs in -ia derived from roots with the stem-vowel A and similar verbs derived from roots with the vowel E, if any exist. Even so, their Perfect Participles would likely differ. If we do assume *peria- as the cognate of Quenya perya "halve", the perfect participle "having halved" would likely be *píriel, for *pêriel-.
          In the case of the verb genedia- "reckon", the prefix ge- is an umlauted form of go- (for which see entry in Etym). The -nedia part of the word represents older *notyâ, so the original stem-vowel o has also been umlauted. The perfect participle we would form from *-nôtiel-, and here the prefix would not be umlauted: *gonúdiel "having reckoned".
          Where the verb contains a diphthong, like breith[a]- "break out suddenly", nuitha- "stunt" or nautha- "conceive", the view presented in the Suggested Conjugation is that it undergoes no change in the perfect participle (breithiel, nuithiel etc.) The only exception may be where ei represent older a, e.g. teili[a]- "to play" and seidia- "set aside" from the roots TYAL, SAT (see VT42:19-20 regarding the latter). This is an exceptional development; normally we would expect teli[a]-, *sedia- with e rather than ei (and in the case of teili[a]-, Tolkien did list teli[a]- as well). Possibly the perfect participles should be formed from *tyâliel- and *sâtiel-, hence *tóliel "having played" and *sódiel "having set aside". If so, gleina- "bound, enclose, limit" (< *(g)lanyâ) should likewise have the perfect participle *glóniel (< *(g)lâniel-). This verb is somewhat strange; as the cognate of Telerin glania- (given in the same source), we would expect *gle(i)nia- with the full ending -ia (VT42:8; compare the editor's comments in VT42:28, note 13). Yet this uncertainty may not influence the form of the perfect participle.
          In the case of two-syllable verbs in -uia or simply -ia, like pui[a]- "spit", thui[a]- "breathe" and thi[a]- "seem" (PIW, THÛ, THÊ), we assume that the ending -iel would partially merge with the verbal stem itself, resulting in forms like *puiel, *thuiel, *thiel.
          Such reasoning underlie the forms presented in the Suggested Conjugation, yet it should be reiterated that [t]íriel "having watched" and maybe -thoniel ?"having kindled" (in Gilthoniel) are the sole attested examples of the Sindarin perfect participle. We try to work out the implications of Tolkien's general system, but by necessity, the "rules" here presented represent a mountain of extrapolation suspended by a hair of known Tolkien material. The Perfect Participle is definitely one of the most poorly attested features of Sindarin grammar.
          Some have assumed that the apparent lack of a distinct perfect tense in Sindarin can be worked around by using constructions involving the perfect participle, e.g. *i Edhel túliel "the Elf [is] having-come" for "the Elf has come". The construction as such is plausible enough, but remains completely unattested.

Note on plural forms of active participles: In Etym-style Noldorin, as well as in Sindarin proper, there is currently no evidence whatsoever regarding any plural forms of the active participles (present or perfect). Regular Noldorin/Sindarin adjectives do agree in number, and the participles are basically adjectival. So would a participle like glavrol "babbling" have a distinct plural form, used to describe more than one "babbling" individual? A word of this shape would have the plural form *glavroel given its phonological history (-ol, representing older *-aul and still older *-âl-, would pluralize as -oel, and there would be no umlaut in the rest of the word). Participles in -el and -iel would have plural forms in -il, with I-umlaut also in the preceding syllable or even thoughout the entire word. In pre-Etym "Old Noldorin" (not the same as the ON of Etym) Tolkien did mention madel pl. medil as forms of the participle "eating" (PE13:131). These forms could be conceptually valid in Sindarin as well (representing pre-historic Sindarin *matila and *matili, in turn from *matilâ and *matilâi at the oldest stage). Yet we cannot be sure whether Sindarin active participles agree in number at all. Quenya participles in -la, cognate with the Sindarin ending, do not seem to agree in number in the late Markirya poem.

The Past (Passive) Participle

In the entry DUL in Etym, Tolkien lists the past tense daul "hid", and then he goes on to mention the "p.p." (past or passive participle) dolen, "hidden". If we remember that au tends to become o in polysyllabic words, these examples suggest that the past passive participle is formed by adding -en to past tense forms, and this theory seems to fit the material very well. The form dangen "slain" would then be formed from *danc, past tense of degi (stem *dag-) "to slay" (the older past tense ndanke is mentioned in the entry NDAK). In accordance with the established phonology, the -nc of *danc becomes -ng- (technically -ññ-) between vowels, hence dangen rather than **dancen. (Compare, for instance, the verb tangado "make firm", the first element of which is a cognate of Quenya tanca "firm"; see entry TAK.) The word prestannen "affected" would seem to be a past passive participle by its gloss, and it can successfully be explained as being formed from *prestant, the past tense of the verb presto "to affect" (PERES): Intervocalic -nt- becomes -nn-; compare Quenya kentano "potter" with its Noldorin cognate cennan (TAN). If, as we theorized above, most A-stems end in -ant in the past tense, past participles in -annen will be very common. (If sog[a]nnen in the entry SUK is not the past tense []sogant "drank" with the pronominal ending -n "I" attached, it may be the passive participle "drunk". Tolkien does not clearly gloss this form as such. In any case, it appears that in the case of A-stems, the sg. passive participle and the 1st person sg. past tense would coincide in form.)
          Above we cited the participles [t]irnen and [s]ollen (lenited forms in Talath Dirnen "Guarded Plain", Fen Hollen "Closed Door") as indirect evidence for the past-tense forms *tirn "guarded" and *soll "closed". These participles would suggest that primary verbs in -r and -l have past-tense forms in -n and -l, respectively, such past-tense forms underlying the attested participles.
          Like regular adjectives in -en (like malen "yellow", pl. melin - entry SMAL), passive participles have plural forms in -in, combined with normal I-umlaut throughout the word. Thus we have dangen "slain", pl. dengin (as in Haudh-en-Ndengin "Hill of Slain", to quote the spelling used in the Silmarillion). There is also ("Eboennin" =) Ebönnin as the plural form of Abonnen "After-born" (WJ:387). Incidentally, Ebönnin would be archaic or "Middle" Sindarin, before ö merged with e; in Frodo's day the Elves would say *Ebennin.
          Though the past participles in -en seem to behave exactly like the numerous adjectives in -en as regards plural formation, it may be noted that these endings are not really cognate. In the entry THUR in Etym, Tolkien calls thoren the "pp." (past/passive participle) of thoro "[to] fence". He derives thoren from older tháurênâ, which seems to be an old past tense form *tháurê with the adjectival or participial ending - added to it (as we argued above, the contemporary past tense of thoro "[to] fence" would be *thaur). Thus, the vowel of the ending -en is in origin the final -ê originally occurring in all past-tense forms (still preserved as -e in Quenya), whereas the -n is what is left of the ending -. On the other hand, the common adjectival ending -en represents older -ina (at the oldest stage -inâ), the final -a umlauting the -i- to -e- before it was lost. For instance, Tolkien derived malen "yellow" from older malina, in turn descending from smalinâ (Etym, entry SMAL). Interestingly, Quenya passive participles typically receive the ending -ina, and one might think that the Sindarin ending is a direct cognate, but this does not appear to be the case. Neither are Quenya passive participles formed from past-tense forms.

NOTE: On October 19th, 2003, in a letter to the Elfscript list, Carl F. Hostetter writes regarding this explanation of the origin of the Noldorin/Sindarin past participles: "I disagree with this unqualified assertion. This -en may also have arisen from *-inâ, and thus be cognate with the Quenya past participial ending -ina exhibited by such relatively late Quenya forms as rákina 'broken', etc" (MC:223). However, the Quenya past passive participles are quite different in form. If we assume that, say, the verb *dag- "slay" (inf. degi from root NDAK, LR:375) had the Old Noldorin past participle **ndákina to go with a Quenya form like rákina, then the later form of the past participle would have been (**daugen >) **dogen, but the attested form is dangen! Moreover, in Quenya the ending -ina is also added to A-stem verbs, resulting in a diphthong ai as in hastaina "marred" (MR:254). If we tried to apply this system to a verb like prest[a]- "to affect", so that the Old Noldorin passive participle had been **prestaina, then the later form would have been either **prestoen (in Etym-style Noldorin) or **prestaen (in Sindarin). The attested form is prestannen! Presumably Hostetter will have to agree that the Noldorin/Sindarin passive participles cannot possibly be direct cognates of the Quenya formations; the N/S forms are unquestionably formed from past-tense verbs. But if I understand him correctly, he argues that the very ending -en may well be a direct cognate of the Quenya ending -ina. However, the form thoren, which Tolkien explicitly calls a "pp." (past/passive participle), he equally explicitly derived from tháurênâ (Etym, entry THUR). This indicates that in this case at least, the ending -en descends from -ê (old past tense ending) + (basically an old adjectival ending). As far as I can see, all attested Noldorin/Sindarin past passive participles can be successfully explained as being formed by the same pattern: we are dealing with old past tense forms to which the ending - was added. The consonant of the ending -en is what remains of - after the loss of final vowels, whereas the vowel of -en is the vowel all past tense forms formerly ended in (in Quenya, all preterites still end in -e). While it is true that *-inâ would also produce Noldorin/Sindarin -en (the final â umlauting the i to e before it was lost), I do not see the need to arbitrarily assume that this ending was formerly present when we have an attested pattern which is equally capable of explaining all known Noldorin/Sindarin past participles.

          To summarize, in a synchronic perspective the rules for how past passive participles are formed can be stated like this: The ending -en is added to the past tense. In accordance with general phonology, final -nc, -nt, -mp become -ng- (i.e. -ññ-), -nn-, -mm- respectively when intervocalic: the past tense forms *danc, *prestant, dramp (DARÁM) therefore correspond to the participles dangen, prestannen, *drammen (the form drammen is actually given in the entry DARÁM, but evidently this is the 1st person past tense "I hewed" rather than the passive participle "hewed"; however, the two would probably coincide in form). Though we have no explicit examples, general phonology tells us that past tense forms in -nd, -ng (sc. ñg, if there are any), and -m (older -mb) would also show -nn-, -ng-, -mm-, respectively, before the ending -en. For instance, gwend as one past tense of gwed[h]- "to bind" would correspond to the passive participle *gwennen "bound". Past-tense forms including the dipthong au show o in the passive participle (daul "hid" vs. dolen "hidden"). The plural forms show I-umlaut throughout the word, so that a becomes e (dangen pl. dengin) and o becomes ö, later merging with e (Abonnen pl. Ebönnen, but later *Ebennin). However, o derived from au would not be affected (so the pl. form of dolen "hidden" must be *dolin rather than **delin; compare daul as the past tense "hid").
          The example Abonnen "Afterborn" (pl. Ebönnin > *Ebennin) may also illustrate another rule. Ignoring the prefix ab- "after", the form onnen meaning "born" (or perhaps rather *"begotten") may well be the passive participle of *onna-, the otherwise unattested Sindarin cognate of Quenya onta- "beget, create" (LR:379 s.v. ONO; notice that the same stem is mentioned in WJ:387 where the form Abonnen is cited - Tolkien calls Abonnen a "participial formation" from this stem). If *onna- has the past tense *onnant (as it would have according to the rules we have tried to reconstruct), its passive participle would be *onnannen according to the rules suggested above, but it may seem that this cumbersome form is contracted by merging the two double nn's and omitting the vowel between them. Hence onnen as in Abonnen. A similar system is employed throughout our Suggested Conjugation; haplology in general seems to be a frequent phenomenon in Tolkien's Elvish. So though "chant" is linna- and the past tense "sang" would presumably be *linnant, we suggest *linnen rather than ?linnannen as the passive participle "chanted" (and also the 1st person past tense "I chanted", which by our reconstruction would coincide with the passive participle in form).

The Gerund

Sindarin gerunds function as verbal nouns (like their English counterparts in -ing). Basically abstract nouns, some of them may take on a more concrete meaning, like hammad "clothing" (apparently referring to the actual garments rather than "[the act of] clothing" someone). But being derived from verbs, gerunds are also able to take a direct object.
          Noldorin/Sindarin gerunds seem to end in either -ad or -ed. The former ending is used in the case of A-stems. The word hammad "clothing" just mentioned seems to connect with the verb ham[m]a- "clothe" (in LR:363 s.v. KHAP this verb is cited as "hamnia-", but this is transparently a misreading of Tolkien's manuscript; such a form would be quite incompatible with the established phonology of the language). The verb eitha- "to prick; to insult" is seen to correspond to a gerund or verbal noun eithad (WJ:365: "the gravest eithad" = insult). The toponym Cabed-en-Aras "Deer's Leap" occurring in the Silmarillion includes cabed "leap". This is obviously to be connected to the stem KAP- "leap" listed in Etym, and Sindarin evidently has a basic verb *cab- "to leap". There also seems to be a verb *cen- "see, look" (as in Quenya); its gerund cened appears, compounded, in the word cenedril "looking-glass" (RS:466). In the Noldorin of the Etymologies, we have [g]onoded "counting" as the gerund of gonod- "count up" attested in the compound aronoded "innumerable" (entry NOT). It seems that aronoded is literally "without (or, beyond) counting": In the entry AR2- reference is made to the prefix ar- "without" (compare arnediad, explicitly translated "without reckoning" in the same entry; this word is very similar, both in structure and meaning, to aronoded). The prefix ar- apparently causes lenition, hence the initial g- of [g]onoded lenits to nothing in aronoded.
          Why do A-stems have gerunds in -ad, whereas basic verbs apparently have gerunds in -ed? It seems that Sindarin gerunds descend from primitive forms in -ta (or at an even older stage, -). Where this ending was to be added to a basic verb, a vowel -i- was slipped in before the ending (as in the present tense, or possibly aorist, of the same class of verbs). Similar formations may function as infinitives in Quenya, as in karita i hamil mára "not to do what you judge good" (VT42:33). If the form karita also existed in pre-historic Sindarin, it would regularly evolve into classical Sindarin *cared (the original final -a umlauting the -i- to -e- before it was lost, and post-vocalic t becoming d). The attested forms cabed "leap, *leaping", cened "looking" and -noded "counting" may confidently be referred to older *kapita, *kenita, *notita. On the other hand, where the ending -ta was added to an A-stem, the A-umlaut could not change the final vowel of this stem (already identical to the vowel causing the umlaut), so all that happened later was the regular loss of the final -a and the voicing of -t to -d. Hence the gerunds of A-stems simply end in -ad, the -a being part of the verbal stem and the -d being what is left of the older ending -ta.
          It may be objected that in the entry ÑGAW in Etym, what looks like a basic verb gaw- "howl" is cited, and yet Tolkien in the same entry went on to list the word gawad "howling". This would come from *ñgawata. Why not *gawed for older *ñgawita, if gaw- is to be a basic verb? Possibly the very common ending -ad had begun to spread to basic verbs as well (by analogy). Compare the past tense ending -ant (properly belonging to A-stem verbs only, its vowel -a- being part of the verbal stem) beginning to spread to basic verbs, as discussed above. But it is also possible that gaw- is simply an incomplete annotation of the A-stem *gawa- "howl". This verb is so treated in our Suggested Conjugation, and the gerund gawad is the most important piece of evidence for this assumption.

It may also be noted that "gaw-" would be the sole known basic verb in -w, and if we try to conjugate it according to the normal system, some pretty outlandish forms would result. If the older past tense of *ñgaw- was formed by nasal-infixion, hence *ñganwe, we would have Middle Sindarin *ganw and Classical Sindarin *ganu! (For nasal-infixion before -w, compare the old Quenya past tense anwe vs. the root AWA, WJ:366.) I tend to believe that if Tolkien had something this exotic in mind, he would have made an explicit note about it in the Etymologies. In our Suggested Conjugation, gaw[a]- is assumed to belong to the Mixed Conjugation, the older past tense *ñgawne = *ñgaune yielding Sindarin *gaun, or with endings *gone-.

          The King's Letter contains several examples of gerunds or verbal nouns in -ad. Irrespective of the prefix ge-, the form genediad "reckoning" (used = "calendar") seems to be the gerund of the verb nödia ("noedia") listed in the entry NOT in Etym. In the early Fourth Age when the Letter is supposed to have been written, ö had long since become e (compare arnediad "without reckoning" in the entry AR2 in Etym itself; this form contains nediad as the gerund of *nedia-, later form of nödia-). Interestingly, gerunds are repeatedly used as infinitives in the text of the King's Letter: E aníra ennas suilannad mhellyn ín phain, "he desires to greet there all his friends", e aníra tírad i Cherdir Perhael "he desires to see the Master Samwise". One may think that this is literally *"he desires greeting [of] all his friends", *"he desires seeing [of] the Master Samwise" - as if the forms in -ad are really just verbal nouns. But this analysis would make the verbal noun the real object of the sentence, and then it should have been lenited (as the object of a Sindarin sentence regularly seems to be). We would have seen **e aníra huilannad mellyn ín phain (suilannad rather than mellyn being lenited). But since Tolkien actually left suilannad unlenited and lenited mellyn instead (as mhellyn = vellyn), we must probably conclude that suilannad here does function as a regular infinitive, and it is mellyn, as the logical object of the sentence, that is lenited. As noted above, the Sindarin gerunds are descended from forms in -ta, and Tolkien translated one such form as an infinitive in a Quenya phrase (karita "not to do, VT42:33). We must inevitably ask whether the infinitives in -i and -o that are exemplified in the Etymologies belong only to that conceptual stage, and are a feature of "Noldorin" that did not survive into Tolkien's thinking on Sindarin proper. In some ways, gerunds would be clearer and less ambigious than these infinitives: All the Noldorin infinitives in -o clash with the imperative forms of the same verbs (though they would hardly be very difficult to distinguish in context). As for the infinitives in -i, the I-umlaut of the stem-vowel which it causes may sometimes result in confusion, since this umlaut neutralizes both a and o to e (and leaves original e unchanged). For instance, is *ceni the infinitive of cen- "see, look" or can- "call"? If there exists a verb *con-, its infinitive would be *ceni as well. On the other hand, the gerunds would remain distinct: *cened "seeing" vs. *caned "calling" (and if there is indeed a verb *con-, its gerund would be *coned). So even apart from the uncertain status of the Noldorin infinitives in -o and -i in Sindarin proper, there may be good reasons for writers to use gerunds instead.
          Incidentally, suilannad "to greet" seems to include the gerund of anna- "give"; hence "to give (a) greeting" or "giving a greeting". Later in the text, suilad is used for "greeting"; this may seem to presuppose a simpler verb *suila- "greet". As for tírad, translated "to see", it is obviously formed from the stem TIR "watch", but tírad can hardly be formed from the basic verb tir- as such (*tirita should have produced *tired instead); rather tírad seems to presuppose a longer A-stem *tíra-, but one may not exclude the other. See Appendix B.
          Finally it may be noted that A-stem verbs in -ada seem to have gerunds, not in **-adad, but simply -ad (evidently by haplology). In the entry RAT- in Etym, Tolkien listed a verb athrado "to cross, traverse" (evidently an A-stem athrad[a]- with infinitive in -o). Then he mentioned the word athrad "crossing", which may in origin be a gerund or verbal noun connecting with this verb, though it was also used in a concrete sense: "ford".  

Special Verbs

We will discuss some verbs that may seem to behave in peculiar ways.

U-stems: Some verbs that in contemporary Sindarin show the vowel -o- are derived from stems that had U instead. For instance, the verb sogo "drink" comes (at least according to the Etymologies) from a root SUK-. At one stage, the short vowel u was changed to o in most positions. But for whatever reason, this did not happen before nasal consonants - and in Eldarin, the past tense is often formed by nasal infixion. Therefore, the past tense of sogo (< *sukâ-) is given as sunc (< *sunkê). The original quality of the stem-vowel is here preserved in the nasal-infixed past tense.
          We have already noted that Tolkien elsewhere (VT39:11) listed the stem for "to drink" as SOK rather than SUK, so maybe the past tense "drank" could also simply be *sonc. Yet the pattern exemplified by sunc may be assumed to be valid as such, and then it would also apply to the nasal-infixed past tense forms of other verbs of derived from stems that have the vowel U:

*tog- "lead, bring" > pa.t. *tunc (stem TUK)
nod- "tie, bind" > pa.t. *nunt (stem NUT)

The verbal stem tog- can be deduced from the 3rd sg. tôg listed in the entry TUK; otherwise Tolkien listed this verb in the infinitive form tegi, with an infinitive ending that has umlauted the stem-vowel to e. The verb "tie, bind" is actually listed as nud- in the entry NUT in Etym, but this must be a mistake, by Tolkien or the transcriber, for nod-, the form demanded by general phonology. In the entry , Tolkien explicitly states that the old stems "not- count, nut- tie coalesced in Exilic [Noldorin] *nod-"; then he explains that the verb "count" was distinguished by adding the prefix go- (hence gonod-). It would seem, then, that the verb "tie, bind" is nod- with no prefix.

If, as suggested by the example hennin *"I hurled" (KHAT), the past tense forms of basic verbs use the connecting vowel -i- before endings, interesting things would happen in the case of U-stem verbs. For, as is demonstrated by hennin vs. the endingless form hant *"hurled" (also listed under KHAT), the vowel -i- umlauts the stem-vowel. In these verbs - the only verbs to have past-tense forms preserving original -u- as the stem-vowel - the umlaut would turn this vowel into -y-. From *nunt "tied" and tunc "brought" we would thus expect *nynnin "I tied" and *tyngin "I brought" (final -nt, -nc regularly becoming -nn-, -ng- when intervocalic).
          Also some of the verbs that must be assigned to what we have called the "Mixed Conjugation" would preserve the original vowel -u- in the nasal-infixed past tense (indeed our attested example of this phenomenon, sog[a]- "drink", inf. sogo with past tense sunc, belongs to this group of verbs):

groga- "feel terror" > 3 pers. sg. pa.t. *grunc (original stem RUK, WJ:415)
[l]oda- "float" (Noldorin lhoda-) > pa.t. *lunt (stem LUT)
tob[a]- "cover, roof over" > pa.t. *tump (stem *TUP)
In the Etymologies, the verb tob[a]- (listed in the infinitive: tobo) is derived from a stem TOP, and then the past tense would simply be *tomp. However, in the Quenya song Namárië in LotR, Tolkien seems to be presupposing *TUP as the Elvish root for "cover" (the literal meaning of the verb untúpa occurring in this song is given as "down-roofs" in RGEO:67). If the stem is to be *TUP instead, then the nasal-infixed past tense of *tob[a]- should probably be *tump.
          As we have demonstrated above, the verbs of the Mixed Conjugation seem to use -e- as their connecting vowel before pronominal endings, and then there would be no umlaut. From sunc (and *grunc, *lunt, *tump) we would have simply *sungen "I drank" (and *grungen "I felt terror", *lunnen "I floated", *tummen "I covered"): As always, -nc, -nt, -mp correspond to -ng-, -nn-, -mm- between vowels, but the vowel u would be unchanged. However, the forms *sungen, *grungen, *lunnen, *tummen would also be the past passive participles of these verbs, and in the plural forms of these participles (ending in -in, as in the attested example dengin "slain"), we would see the umlaut u > y: *Syngin, *gryngin, *lynnin, *tymmin. (Compare a being umlauted to e in the attested pair dangen pl. dengin.) Likewise, *tunc and *nunt as the past-tense forms "brought" and "tied" would correspond to the passive participles *tunnen, *nunnen (same English translation), with plural forms *tynnin, *nynnin.

Notice, however, that *mudannen as the past participle of the verb mud[a]- should probably have the plural form *mudennin rather than ?mydennin, for in this case the u is derived from an old long ô, mud[a]- being the cognate of Quenya móta- (root ). The (original) long vowels would be immune to I-umlaut. However, the intransitive verb muda- "labour, toil" may not normally have a past participle anyway.

          All of these elaborations are of course merely an attempt to work out what seems to be the logical and necessary consequence of Tolkien's general system. The reader will understand that the primary sources only provide hints, like Tolkien noting that *sogo has the past tense form sunc. It is up to ourselves to find out what kind of forms would arise if we apply the grammatical rules we have tried to make out, taking into account what can be inferred about the intended phonology of the language. Hopefully we are not too clever here.

Other special verbs would be the so-called impersonal verbs, verbs that by their meaning can have no logical subject. In the Noldorin of the Etymologies, we have eil "it is raining" (ULU) and bui, the latter explicitly said to be impersonal but not clearly glossed (MBAW). However, bui is quite clearly meant to be the cognate of Quenya mauya- "compel". We may assume that bui would mean "(it) compels", "(it) is necessary", used in such sentences as *bui anim teli, "(it) compels for me to come" = "I must come" (as we could theorize that this sentence might appear in Etym-style Noldorin).
          As for eil (older form "oeil" = öil) "it is raining", Tolkien derived this verb from ulyâ, which would also be the source of Quenya ulya- "pour". The final vowel of ulyâ has been lost, and the original y has also disappeared, but it has umlauted the stem-vowel to produce the dipthong öi, later ei.
          At this point, a question arises: if the old A-stem ulyâ comes out as eil, why does (say) *dulyâ "conceil" produce döli[a], deli[a] instead? (These A-stems are cited in infinitive form "doelio, delio" in the entry DUL.) Why is the original final -â here preserved as -a, whereas in ulyâ > eil it has been lost? Indeed, how can Sindarin have such present-tense forms as penna "slants" (observed in the hymn to Elbereth) given the fact that at one point, Noldorin/Sindarin is known to have lost all final vowels? The development suggested by ulyâ > eil would be expected to be normal, but actually it is highly exceptional.
          Where other N/S verb-forms end in a vowel, the explanation is simple: this vowel was not final at the point where the loss of final vowels occurred. For instance, the Noldorin infinitives in -i are descended from older forms in -ie (e.g. trenarie > treneri, NAR2), so the final vowel was indeed lost; what remains is the original second-to-last vowel. The imperative ending -o is descended from an originally independent imperative particle á (WJ:371-372); apparently it was suffixed to the verbal stem at a relatively late stage and so escaped the loss of final vowels. Also in the case of forms like penna "slants", it is tempting to assume that the surviving final vowel somehow was not final at the time of the loss of final vowels. The solution to this little mystery may be that verbs originally received 3rd person markers including the consonant -s. In the entry S- in the Etymologies, references are made to primitive endings -so or -se, apparently meaning "he" and "she" respectively; there may also have been an ending *-sa "it" (compare Noldorin ha "it" mentioned in the same entry); Quenya also preserves -s as an ending for "he, she, it". If, as it seems, a similar 3rd person singular marker -s (with or without a following vowel) also existed in early Sindarin, it would later become lenited to -h. Compare the entry BARÁS in Etym, where we have Old Noldorin barasa "hot, burning" later becoming baraha and finally yielding Noldorin bara, again with a surviving final vowel because it was not final when the original final vowels disappeared (baraha > *barah > bara). Likely, then, the immediate ancestor of penna "slants" was *pennah with a short-lived final -h which in turn derives from the ancient 3rd person sg. endings in -s-.
          Impersonal verbs like ulyâ > eil "it is raining" would indirectly confirm this scenario. Precisely because these verbs were entirely impersonal, having no logical subject, they did not receive the 3rd person sg. subject marker -s: It would seem that unlike speakers of English, the Sindar of pre-historic times did not slip in the same kind of dummy-subject as the one occurring in the phrase "it is raining" ("it" having no real meaning here). The original ulyâ- received no ending that could "shield" the final -â, and eventually it was lost like all final vowels, leaving eil as the comtemporary form. It may be assumed, though, that where various endings were added, the old -â would survive as -a- (shielded by the endings), and ulyâ- would regularly evolve into *elia- (compare, for instance, deli[a]- from *dulyâ, entry DUL in Etym). So as the past tense of this verb, our Suggested Conjugation lists *eliant, as the future tense, *eliatha, as the active participle, *eliol etc. However, as indicated in the Suggested Conjugation, it is possible that the past tense "it was raining" could also be *aul; compare daul as an archaic past tense of *deli[a]- "conceal" (the entry DUL seems to suggest that this can be the past tense of deli[a] and doltha- alike, these verbs having the same meaning). The monosyllabic past tense *aul would go well with eil as the present tense. Incidentally, eil would have become *ail in Third Age Sindarin, since Middle Sindarin ei turned into ai where this diphthong occurred in a final syllable; hence the reading *ail occurs in our Suggested Conjugation.
          The other Noldorin impersonal verb, bui *"it compels, it is necessary", is derived from "mauy-" in the entry MBAW in Etym. This cannot be right; *mauy- would produce Noldorin *mui instead. As indicated by the entry-head MBAW, the actual ancestral form must be *mbauy-, or in full form *mbauyâ. As in the case of ulyâ > öil > eil > *ail, *mbauyâ "[it] compels/is necessary" was apparently considered an impersonal verb which therefore did not receive the 3rd person marker -s-; therefore there is no trace of the final -â either, there being no following ending to shield it. The -y- before it is also gone, but its former presence caused I-umlaut and changed the original diphthong au to ui; hence *mbauyâ > bui. In Etym-style Noldorin, I-umlaut regularly turns au into ui. Consider, for instance, rhaw (= *rhau) "lion" and its plural form rhui, the latter being the cognate of Quenya rávi "lions", the Quenya form preserving the old plural ending which now only manifests as I-umlaut in Noldorin (see entry RAW).
          However, this is one feature of Noldorin which did not survive unmodified into Sindarin proper. In Sindarin, au + I-umlaut results in oe, not ui. For instance, the Sindarin plural form of naug "dwarf" is noeg rather than **nuig (cf. Nibin-noeg "Petty-dwarves", WJ:187). Thus Noldorin bui would become *boe if updated to Sindarin phonology, and this is the form listed in my Suggested Conjugation. One may of course question the value of such "updated" forms: Maybe Tolkien liked the word when it appeared as bui, and would have rejected it altogether when his own phonological revisions would turn it into *boe instead? But most likely, he never even considered this; he may have spontaneously created the form bui when writing the Etymologies, and we cannot know whether he ever returned to this verb. In the "Neo-Sindarin" lines composed by David Salo for the Jackson movies, the updated form *boe is used repeatedly. For instance, in The Fellowship of the Ring Aragorn tells Haldir: Boe ammen i dulu lín, "is-necessary for-us the support of-yours" = "we need your help". - Whether *boe would have any further tense-forms is unclear; none are listed in our Suggested Conjugation.

Discussing special verbs, we may finally mention a few verbs the exact behaviour of which is uncertain, because Tolkien's notes are obscure. In the entry MBAKH in the Etymologies, we read: "Q manka- trade; makar tradesman; mankale commerce. N banc, banga." There can be little doubt that banga is a verb, the cognate of Quenya manka- "trade", both from primitive *mbankâ-. But what does banc, the form listed immediately before banga, mean? It does not seem to mean either "tradesman" (a distinct word for "pedlar" is listed afterwards) or "commerce". May banc rather be a form of the verb banga? If so it would be the past tense, used instead of the longer form *bangant. Banc could be a past tense formed by nasal-infixing the original stem: primitive *mbank(h)ê. Our Suggested Conjugation presents banc as the past tense of banga-, but it should be understood that this is merely an attempt to make sense of Tolkien's brief notes.

Another somewhat obscure verb is listed in the entry DAT-, DANT-. Following Quenya lanta- "to fall", the corresponding Noldorin verb is simply cited as "dant-". Tolkien's annotation is highly elliptical. Quenya lanta- would come from *dantâ-, the Noldorin/Sindarin cognate of which should be *danna- (so in our Suggested Conjugation). Maybe "dant-" is simply meant to represent the primitive form underlying the actual later form. It is, however, entirely possible that the past tense of *danna- should be, not *dannant, but rather *dant, formed directly from the stem DAT-. If so, it would parallel banc as a possible past tense of banga- discussed above. It may be noted that after listing the strange verb "dant-", Tolkien immediately went on to cite the form dannen "fallen". According to the general principle that such a past participle would be formed by adding -en to the past tense form (and bearing in mind that -nt becomes -nn- between vowels), dannen "fallen" could be formed from a past tense *dant "fell". Yet we cannot be sure, for even if the past tense of *danna- were *dannant, the past participle *dannannen would probably still be shortened to dannen by haplology.

Appendix A: The Past Tense System Revised?

According to the rules reconstructed above, the past tense of the verb car- "do, make" should be *carn (corresponding to the Quenya pa.t. karne, listed in the Etymologies, entry KAR). However, in WJ:415, Tolkien makes reference to "a primitive past tense" which is "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."
          Agor as the past tense of car- is of course a rather surprising form. It seems that car- is indeed the strong or primary verb referred to; the root KAR can be discerned in the ancestral form akâra cited by Tolkien. Since Tolkien insists that these formations are "usual" in this class of verbs, we must try to generalize some rules. To quote my main Sindarin article: "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.)" By this system we would have forms like *ebid "spoke" (ped- "speak"), *ewidh "bound" (gwed[h]- "bind"), *idir "watched" (tir- "watch"), and *onur "ran" (nor- "run").
          Before pronominal endings, the connecting vowel -e- is slipped in, as in the Turin Wrapper form agorech (irrespecitive of what meaning the pronoun -ch had when Tolkien wrote this).

NOTE: Carl F. Hostetter argues that before the extra syllable that is added when a connecting vowel and a pronominal ending is suffixed, the formerly long vowel in the preceding syllable should remain long (e.g. *idíren rather than *idiren for "I watched", because the older form would be *itîr-). (Incidentally, Hostetter also argues that the connecting vowel before the pronominal suffix may not always be -e-, but analogy would certainly work in this direction.) The attested example agorech (instead of **agórech) he dismisses because we would here see au at an intermediate stage between the original long â and the later o: akâra- > *agaur- > agor-. Apparently he feels that au was directly monophthongized to o, without considering the possibility of it first becoming a long ó (later shortened, as other long vowels would also be). Such a long vowel probably did exist at one stage; compare ónen for "I gave" in Gilraen's linnod according to many LotR editions. This ó is commonly understood to represent an older au, ultimately representing a long â in the primitive language. Anyway, the shortening of formerly long vowels in polysyllabic words is a common though not universal phenomenon in Noldorin/Sindarin. There are some cases where a vowel maintains its former length, e.g. in the name Tin(n)úviel "Nightingale", which Tolkien in the Etymologies (entry TIN) derived from Tindômiselde. The fact that the vowel of Tinnúviel receives the main stress in a long compound may have helped to preserve its original length. But the tendency is rather to shorten long vowels in polysyllabic words; compare Noldorin hiril "lady" with the Old Noldorin form khíril (from even older *khêrill-, compare the original root KHER). Another example would be milui "friendly" vs. the root MEL; to have its quality altered from E to I, the stem-vowel must have been a long ê at an older stage, initially yielding a long î the length of which does not survive in milui. If Hostetter will argue that milui is likely an adjective derived in more recent times from the related noun mîl "love, affection" (< *mêl-), then milui would still demonstrate the shortening of long vowels in polysyllabic words: Analogy with the noun would tend to preserve the long vowel, but this does not happen, indicating a strong tendency to shorten vowels in longer words. In the case of verbs, analogy with the unsuffixed forms (the 3rd person sg., or in Hostetterian terminology "personless" forms) would also be a strong influence: While Hostetter at one point seemed to argue that the formerly long vowel could remain long even in the final syllable of a word, he now insists that this was not what he really meant; he and I apparently agree that *itîr- by itself would yield *idir rather than *idír with the long vowel intact. So if *idir is indeed a possible translation of "watched" (and this is far from certain - see below), then I do think there is very good reason to expect *idiren rather than *idíren for "I watched", even though the vowel in the second-to-last syllable was long at an older stage. A survey of the Noldorin material indicates that in polysyllabic words, long vowels rarely occur outside compounds one element of which has a long vowel when it occurs by itself (as a monosyllable).

          As I also point out in my main Sindarin article, this way of forming preterites would seem to contradict other sources. The past tenses of ped- and gwed[h]- are attested as pent and gwend, respectively (see our discussion of the past tense above). Are such forms to be considered conceptually obsolete? Moreover, where would this leave the past passive participles that are apparently formed by adding -en to the past tense? Does not [t]irnen as the word for "watched, guarded" (lenited in Talath Dirnen "Guarded Plain") presuppose *tirn as the past tense of tir-? Surely there is no evidence that Tolkien ever considered altering this name to *Talath Idiren to go with *idir as the "new" past tense of tir-! Yet if car- "do, make" is to have the past tense agor, should "done, made" be *agoren or *carnen (the latter presupposing *carn as the pa.t. "did"?) Evidently following the pattern of [t]irnen vs. the verb tir-, David Salo used *carnen in the lines he constructed for the Jackson movies (in The Two Towers, the filmatic version of Aragorn at one point says mae carnen "well done" to a horse).
          In this case, Tolkien's intentions cannot be reconstructed with any confidence. May "fossilized" past tense forms underlie some participles, so that the old past tense *tirn "watched, guarded" (= Quenya tirne) survives in a participle like [t]irnen, whereas the living past tense of tir- is rather *idir? However, agor remains the sole attested example of this kind of past tense, though Tolkien did state that this formation is "usual" in the primary verbs. Yet what does "usual" mean? That it was the regular or at least the dominant system, or merely that it was relatively common? Until more material is available, I would accept agor as the past tense of car- but otherwise use the past tense system reconstructed above (e.g. *tirn rather than *idir as the past tense of tir-).

Appendix B: A Distinction Between Aorist and Continuative Tense?

At least in the case of basic verbs, Quenya is able to make a distinction between the aorist tense and the present or present continuative tense. Thus the verb sil- "shine" has the aorist form silë "shines", whereas in the present or continative form, the same verb appears as síla instead; this may be translated "is shining", emphasizing the on-going nature of the verbal action. The aorist simply describes the verbal action without qualifying its duration, and may often refer to momentary or habitual actions or describe events in "proverbial time". Is it possible to maintain a similar distinction in Sindarin, at least in the case of primary verbs, or does Sindarin merely possess a single "present" tense? We have no definite evidence either way. We can only consider the primary verbs. In the diachronic perspective, what we have called the present tense appears to be the cognate of the Quenya aorist. [P]êd (VT41:11) as what we have called the present tense of ped- "say, speak" would seem to descend from *kweti, which would also be the origin of the Quenya aorist quete. Likewise, tôg as the 3rd person singular (present tense) form of tegi "to bring" is probably meant to be the descendant of *tuki, which would yield the Quenya aorist *tuke "draws" (1st person tukin *"I draw" listed in the entry TUK in Etym). In Quenya, the "continuative" form of primary verbs is derived by lengthening the stem-vowel and adding -a, e.g. síla "is shining" from sil-. Is there any evidence for such formations in Sindarin? It may be noted that the verb síla is attested not only in Quenya (as in elen síla lúmenn' omentielvo), but also in a Sindarin context in Lúthien's Song (menel-vîr síla, evidently *"[the] sky-jewel is shining", referring to the moon). If the basic verb sil- is common to Quenya and Sindarin, its Sindarin present tense would be *sîl according to the rules reconstructed above. May *sîl and síla actually coexist in the language, as the aorist and the continuative tense, respectively? If so, these forms would correspond to Quenya silë and síla (again respectively). Otherwise, the only possible evidence for continuative forms is found in the King's Letter. The form aníra "desires" there occurring could conceivably be the continuative form of a verb *anir-, hence more literally *"is desiring", but the exact derivation and etymology of aníra are uncertain. The King's Letter also provides the gerund/infinitive tírad "to see". This is obviously related to the verb tir- "watch, guard". According to the rules reconstructed above, tir- should have the gerund *tired (to go with such attested forms as cabed, cened, [g]onoded). May tírad actually be formed from *tíra as a continuative form of tir-? If so, would e aníra tírad i Cherdir Perhael "he desires to see the Master Samwise" actually imply *"he desires to be seeing the Master Samwise", emphasizing the continuative nature of the action? (Our Suggested Conjugation lists tíra and tir- as separate entries, without discussing whether these are two forms of the same verb or distinct formations from the primitive root TIR.) It is somewhat unfortunate that our possible attestations of continative formations (síla and the gerund tírad) involve verbs formed from roots with the stem-vowel i (SIL, TIR); this may also be the case with aníra, if indeed it belongs here. The lengthening of the stem-vowel in continuative forms would have to be ancient if this grammatical feature is to be shared by Quenya and Sindarin, and whereas old long î could survive into Sindarin with its original quality unchanged, older â, ê, ô would turn into ô, î, û at the stage corresponding to the "Old Noldorin" of the Etymologies (what Tolkien might later have called pre-historic Sindarin). Original û would also survive with its original quality intact, whereas short u would turn into o at an early stage. In short, verbal stems with the stem-vowel a would have continuative forms in ó, stems with the vowel e would show continuative í, and stems with the vowel o (whether original or derived from primitive u) would have ú in the continuative forms. Examples of possible aorist/continuative distinctions:
mad- "eat": aorist *mâd "eats" (< *mati), continuative *móda "is eating" (< *mât-)
ped- "say": aorist pêd "says" (< *kweti), continuative *pída "is saying" (< *kwêt-)
nor- ?"run": aorist *nôr "runs" (< *nori, if the root is *NOR), continuative *núra "is running" (< *nôr-)
tol- "come": aorist tôl "comes" (< *tuli), continuative *túla "is coming" (< *tûl-)
Only verbal stems with the vowel i would simply lengthen it in the continuative tense, like sil- "shine" having the aorist form *sîl and continuative form síla *"is shining".
          If the form tírad occurring in the King's Letter is indeed the gerund of a continuative stem *tíra- (as contrasted with the simpler stem tir- "watch, see, guard"), we must assume that all the "continuative" forms listed above could also produce distinct gerunds. We would have distinctions like *maded "eating, to eat" (< *matita) vs. * módad "(continuous) eating, to be eating" (< *mâtata). Conceivably the continuous stems could also be used as the basis of distinct forms in the past tense and future tense, e.g. *módant "was eating" (vs. *mant "ate"), future *módatha "will be eating" (vs. *meditha "will eat"), and maybe even distinct active participles (e.g. *módol < *mâtala vs. *madel < *matila). It would be difficult to maintain the distinction in the Perfect Participle, though.
          However, all of this is very hypothetical. Currently, there is little or no hard evidence for an aorist/continuative distinction in Sindarin, and while *tíra- is given a separate entry in our Suggested Conjugation, no explicit continuative forms are listed - since they would be even more hypothetical than many of the forms that do appear there.

Appendix C: Sindarin Phonology Revised?

The system here outlined depends on certain features of phonology. We have repeatedly referred to the rule that when final -nc, -nt, -mp become intervocalic because some ending is added, these groups turn into -ng- (technically -ññ-), -nn-, -mm-, respectively. Thus we have, say, dangen "slain" from the past-tense form *danc, prestannen "affected" from the past-tense form *prestant, or drammen *"I hewed" from the ending-less past-tense form dramp (Etym, entries NDAK, PERES, DARÁM). As we will demonstrate below, there cannot be much doubt that this accurately represents Tolkien's ideas about Noldorin/Sindarin phonology as he held them in mind when he actually wrote LotR (and earlier, the Etymologies).
          Yet in VT42:27, reproducing a late text, Tolkien insists that in the kind of Sindarin that was used as a language of lore in Gondor, final -nc, -nt, -mp became -nch-, -nth-, -mf- (sc. the spirants ch, th, f following a nasal); later these groups were often pronounced as long voiceless ñ, n, m (which may be represented as , hn, hm). Furthermore, it is said that "in true Sindarin of the Elves or Elf-friends the final form [presumably meaning -nc, -nt, -mp] was often introduced medially". This would thoroughly mess up the system outlined above. The (attested!) forms dangen, prestannen, drammen would then rather be either *danchen, *prestanthen, *dramfen (later pronounced *dahñen, *prestahnen, *drahmen with voiceless nasals), or - "in true Sindarin of the Elves" - *dancen, *prestanten, *drampen.
          Is this a lasting or "authoritative" idea in Tolkien's evolution of his languages? Intervocalic nth does occur in the verb "glintha-" in WJ:337 (reproducing a post-LotR source), but in our Suggested Conjugation we prefer the reading *glinna-, though this could well descend from glintha- in prehistoric Sindarin (the stage corresponding to the Old Noldorin of the Etymologies). The idea that older -nt- (via -nth-) becomes -nn- at the classical stage of Noldorin/Sindarin is a long-standing idea. One may consider such an example as primitive winta- yielding Old Noldorin wintha- and (classical) Noldorin gwinna- (see entry WIN-, WIND- in Etym). At the - in external time - very late stage of Noldorin exemplified in PM:135, reproducing Tolkien's drafts for the LotR appendices, this idea was still valid: Considering a word for "fall" (autumn), Tolkien first wrote Dant, but then decided to add the ending -as, and the resulting word is Dannas (not "Danthas" or "Dantas"). Some may point out that in the same source, Tolkien also experimented with the form Dantilais and cite this as evidence that intervocalic nt is possible, but Dantilais is transparently a so-called "improper compound" (not a value judgement): This is really the phrase *Dant i lais "Fall of the leaves" written without spaces rather than a genuine unitary word.
          Most importantly, one seeming example of intervocalic nt becoming nn appears in the published text of the LotR itself, which (from a certain perspective) would elevate this feature of Sindarin phonology to canonical status, even if Tolkien presents contradicting ideas in later sources. The word in question is ann-thennath, which Aragorn mentions as the name of an Elvish verse-mode in the chapter A Knife in the Dark in Book One of The Fellowship of the Ring. Ann-thennath is probably to be interpreted as "long-shorts", referring to a certain pattern of long and short rhythmic units. In any case the element thent (probably meaning "short" as in the Noldorin of Etym, see entry STINTÂ) also occurs in the name of another Elvish verse-mode, the Minlamad thent/estent (WJ:311). If, as seems overwhelmingly probable, the thennath of ann-thennath is thent with the plural ending -ath attached, then the rule that intervocalic nt becomes nn must be considered canonical and fixed, and the surprising system described in VT42:27 can be dismissed as a (far from unprecedented) case of Tolkien forgetting what he had already said or implied in published sources.
          Another possible example of intervocalic -nt- becoming -nn- in the LotR itself may be the word govannen in Glorfindel's greeting to Aragorn: Mae govannen, "well met" (translated in Letters:308). The view presented in the Suggested Conjugation is that there exists a verb *govad- "meet" (go-vad- *"go together", compare Etym-Noldorin trevad- "traverse" or more literally "through-go" = *tre-vad-; see entry BAT and compare TER). Its past tense would be *govant. Adding the participial ending -en and taking into account the hopefully regular change of -nt- to -nn- between vowels would indeed land us on govannen as the past participle "met".
          It may also be noted that according to the system set out in VT42:27, the last word of the name Haudh-en-Ndengin occurring in the Silmarillion would have to be altered to either *(n)denchin or *(n)dencin. I would tend to ascribe a semi-canonical status to the names occurring in the published Silmarillion, even though Tolkien himself never produced or published a definite version of this work in his lifetime. As we have just seen, the change of intervocalic nt to nn does occur in a cannonical source, and then phonological consistency would require that intervocalic -nc- is to become -ng- (i.e. -ññ-) in the same position. Likewise, mp should become mm when intervocalic.
          Others have also observed that the ideas manifesting in VT42:27 are in some respects difficult to reconcile with information Tolkien had already published in LotR. For instance, we suddenly have Tolkien insisting that the medial ll of words like mallorn actually represents voiceless lh; this spelling is supposedly inspired by "the manner of modern Welsh". Already the editor of the document in question, Carl F. Hostetter, pointed out that this is "in stark contrast to Tolkien's earlier comment in Appendix E to The Lord of the Rings" (VT42:31) - since in that Appendix, Tolkien implied that ll is simply a long or double l.
          Furthermore, Tolkien's new idea about intervocalic nc, nt, mp evolving into voiceless nasals , hn, hm would make these sounds relatively common in Sindarin, whereas in Appendix E he had stated that voiceless nasals "were of very rare occurrence in the languages concerned" (evidently including Sindarin). It may seem that the system set out in VT42:27 represents wild experimenting rather than a particularly authoritative or well-considered set of ideas, and it probably contradicts what Tolkien had implied in canonical sources already published by himself - unless we adopt very special readings of the passages in question, or arbitarily assume that certain words are to be assigned new etymologies.
          The academic may try to diligently map every conceptual development in Tolkien's evolving scenario, with no considerations of internal consistency (which it is not the academic/descriptive linguist's task to achieve). However, at least as things appear now, I think the practical linguist accepting the (often ungrateful) task of suggesting how Sindarin can be normalized should rather maintain the system observed in numerous sources: If the groups -nc, -nt, -mp become intervocalic, they turn into -ng-, -nn-, -mm-. In the classical language they do not come out as nasal + spirant, nor as voiceless nasals, and much less are they preserved unchanged. Our Suggested Conjugation is based on this assumption.

Appendix D: "3rd Person" or "Personless" Forms: a Question of Terminology

Carl F. Hostetter, this ever-diligent critic of this article, has questioned the terminology here used. He presented the following comments on the Lambengolmor list:
Fauskanger repeatedly refers to past-tense verbs in -nt as "3rd pers. sg." (in some cases "transparently" so). In light of this, it is noteworthy that none of the Noldorin verbs of this form found in Etymologies is translated with specifically 3rd-person sg. glosses; short of an explicit statement by Tolkien that all Noldorin past-tense verbs in -nt are specifically 3rd pers. sg., there is in fact no way to prove what Fauskanger silently asserts. In fact, given what we know about personless verb forms in the Eldarin tongues [...] and based on the evidence we actually have, it is far more likely that these forms are singular personless forms, and therefore would be used in any case where an explicit, singular subject precedes the verb. Indeed, the same holds true of the Sindarin past-tense verbs in -nt, for although teithant is indeed used by Tolkien with a 3rd pers. sg. subject (Celebrimbor ... teithant 'Celebrimbor ... drew'), this does not in itself necessarily preclude its possible usage as a personless verb with 1st and 2nd person subjects, any more than the use of endingless present-tense verbs in Quenya with 3rd pers. sg. subjects precludes their use with other singular persons, such as in elye hiruva 'thou shalt find" (where the subject is 2nd sg.). Indeed, another such verb, echant, is used by Tolkien with a 1st pers. sg. subject, and translated as such: Im, Narvi, ... echant 'I, Narvi, ... made'.
First of all, it should be noted that this "controversy" only has to do with what terminology it is best to use when describing the Sindarin verb system; there is not (in this case) any disagreement about how the verb system actually works.
         Hostetter is undoubtedly right that an Eldarin verb, where it occurs without pronominal endings, is strictly speaking "personless" rather than belonging specifically to the 3rd person. We are dealing with the "basic" finite form of the verb. Where the subject is a pronoun and this pronoun is not added directly to the verb as a suffix, but appears as an independent word, then the "personless" form is used - no matter what "person" the pronoun belongs to. If the pronoun is singular, the verb receives no ending at all (as in the example Hostetter refers to: Im ... echant "I...made"). If the pronoun is plural, then the verb would probably receive the plural marker -r; this system is attested in Quenya (as in emme avatyarir "we forgive", VT43:8, 20).
          Yet such verbs may conveniently be referred to as "3rd person" verbs, since they typically would occur with a noun as their subject, and this noun is thought of as "3rd person" because it could be replaced by a 3rd person pronoun. This is Tolkien's own terminology: A verb-form like tôg (Etym, entry TUK) would typically translate into English as "leads", an English 3rd person sg. form. One could say, for instance, *Narvi tôg (and/or maybe *tôg Narvi) for "Narvi leads". Yet this verb is by Hostetter's terminology surely a "personless" form since it has no pronominal markers, and indeed it could probably be combined with any independent (sg.) pronoun regardless of what "person" the pronoun belongs to (e.g. the first person if we rewrite the Moria Gate inscription as *im, Narvi, tôg "I, Narvi, lead"). I do not disagree with any of this - there is just one detail which Hostetter should notice: Tolkien himself called tôg a "3 sg." form in the Etymologies, entry TUK.
          So what I have done, really only amounts to applying Tolkien's own terminology to the past tense as well as the present (or aorist?) tense. I, like Tolkien, find this a convenient way of referring to the forms in question, though as pointed out by Hostetter, these forms as such do not belong specifically to the 3rd person: When the subject is an independent pronoun, this pronoun shows what "person" we are dealing with, whereas the verb is simply unmarked in this regard. In general, we can still call these verb forms "3rd person" forms. Calling them "personless forms" because in certain contexts they can do service for any "person" is like calling the Quenya nominative case an "indefinite case" because the "nominative" forms can sometimes do service for any case: In a genitive phrase like Elendil Vorondo "of Elendil the Faithful" (UT:305) it is only the last word that receives the genitive ending -o, whereas the word Elendil appears in a form that looks like the nominative. The point is that the explicit case marker is added to another word (vorondo as the genitive of voronda "faithful"), but applies to the entire phrase. Similarly, in im...echant "I...made" it is another word - the independent pronoun - which shows what "person" we are dealing with, though the verb echant looks like a 3rd person sg. form. This does not mean that terms like "nominative" and "3rd person verb" must be scrapped as useless and inadequate, to be replaced by something like "indefinite/unmarked case" or Hostetter's "personless form". Tolkien's terms describe the typical function of the relevant forms well enough.


Other studies of Sindarin verbs:

The Sindarin verb system by Thorsten Renk

The Past-Tense Verb in the Noldorin of the Etymologies by Carl F. Hostetter (an important source for Renk's treatment)

Ardalambion Index