The Sindarin Verb System

Three Rings
Thorsten Renk — June 2004
needs some updating
Analytical ArticleAnalytical Articles: These articles provide a detailed overview of the theme they cover. However, they require some prior knowledge of the main works of J.R.R. Tolkien.


The following article is an attempt to present some ideas about the systematics of the Sindarin verb system. It owes much to Carl F. Hostetter’s article The Past-Tense Verb in the Noldorin of the Etymologies. This article is intened as a guideline for people who want to use Sindarin to write texts and therefore often contains plausible (but unproven) speculations - these may or may not be wrong, there’s no easy way to tell. The article uses the Noldorin of the Etymologies and the later Sindarin examples as sources, this may to some degree questionable, but there’s little choice if one is interested in the practical question of what form to use. Therefore, in the following verbs will be quoted as Noldorin forms when examples from the Etymologies are given, but (rather well-known) sound shifts will be carried out to give the best estimate of how the corresponding Sindarin forms might look like when actual suggestions for verb conjugations are given. Some basic phenomena of Elvish phonology we will frequently use in the following are:

  • Sindarin words undergo i-affection, i.e. if the ending of a verb involves the vowel -i-, the other vowels inside the verb adjust following the pattern a → e, o → e and u → y. An example would be the infinitive ending -i leading from echad- to echedi (to fashion)1).
  • single consonants following a vowel undergo lenition in the development from CE roots to Sindarin words, hence a primitive root NOT yields the (prefixed) verb gonod- and not **gonot-
  • primitive root vowels are also shifted - most notably, primitive -u- appears as -o- in Sindarin, but this shift is not carried out under some conditions, one of them being if the cluster -nc follows the -u-. The verb *sog- (to drink)2) exemplifies this - it is derived from a primitive root SUK, its present tense is sôg (showing lenition in addition) and if endings are present, most likely i-affection turns this into *segin (I drink), but the past tense involves the cluster -nc and so the original root vowel is restored in sunc (drank).
  • vowels which are long in OS/ON undergo characteristic shifts é → í, ó → ú and á → au/o. In the last example, -au- is preserved in monosyllabic words whereas this turns into -o- or -ó- in words with more than one syllable. An example is the past tense of *anna- (to give) which is *aun (gave) but becomes ónen and not **aunen (I gave) as soon as endings are present.
  • a nasal appended to a root ending in a stop consonant freuqently swaps position with the stop. The past tense sunc (drank) mentioned above is presumably derived from primitive PQ: *suk-nê which subsequently changed into the less awkward CE: *sunkê → N: sunc. This phenomenon is called ‘nasal infixion’

It is beyond the scope of this article to give the detailed reasoning leading to these rules. Sometimes, several of these rules conspire to produce results which look at first glance rather surprising. We do not know for sure if those are really what Tolkien had in mind or if there are exceptions to the rules not known to us, so following the rules is only our best guess. In the following, we will ocasionally refer to ‘weak’ or ‘strong’ past tense formation. What is meant by these is that past tense can be formed by appending an ending to the verbal stem (this is called ‘weak’) or by a modification of the verbal stem as such (this is called ‘strong’). For example, the nasal infixion mentioned above alters the verbal stem and is therefore a strong past tense formation.

General systematics

We can distinguish two main classes of verbs, termed by Tolkien ‘weak’ and ‘strong’ or ‘primary’ and ‘derived’ verbs (see e.g. WJ:415). Primary verbs show a base form which is basically the underlying CE root modified by sound shifts in the development from CE to Noldorin/Sindarin, cf. RIP yielding the verb rhib- (to flow like a torrent)3). In contrast, derived verbs are formed from a root by means of a derivational ending appended on the CE root, the most common of these endings are -jâ and -tâ. An example is the root , the primitive CE: môta yielding Q: móta- and N: *muda- (attested is the infinitivemudo) (to labour)4).

However, the scenario of verb conjugation emerging from the attested examples in the Etymologies and elsewhere shows a rich pattern of subdivisions of these main classes; there are several different ways of forming the various tenses and obviously some older forms have been replaced by analogies with others. These classes are by no means unique, in the course of the (fictional) evolution of the language, Tolkien apparently had drifts of verbs from one into a different class in mind; for example for a given verb there may be more than one possible way of forming the past tense.

Primary Verbs ending with CE roots ending in stop consonants (p, t, k, d) (Class P1)

This class of verbs seems to be close to the underlying CE forms since many elements of the formation occur both in Sindarin and in Quenya. P1-verbs show infinitives in -i (cf. echedi from et-kat (to fashion)5)) and past tense formation with nasal infixion (cf. echant). We see similar past tense formations in Quenya; cf. Q: quentë (spoke)6) to the infinitive Q: quetë (to speak)7) and the Sindarin past tense S: pent (said), presumably corresponding to a verb *ped- with infinitive *pedi. As the example pêd (from guren bêd enni (my heart tells me)8)) shows, the basic (personless/3rd person) form of a verb of this class is apparently formed by lengthening of the stem vowel.

We have no direct clean example of how the present tense of verbs of these class is formed when endings are appended (there’s ú-chebin (I do not keep) in Gilraen’s linnod, but we don’t know the infinitive of this verb, it might fall into the next class), but general principles make it very likely that endings in present tense should be appended using a connecting vowel -i-. The example *had- (to hurl)9) with infinitive hedi and past tense hant (*hurled) hennin (*I hurled) suggests that the past tense of P1 verbs likewise uses -i- as the vowel to connect the ending with the basic form of the past tense. No example is attested in future tense, so trying to instert -i- again as connecting vowel to append the ending -tha (which in turn is only attested as ending for derived verbs) is a mere guess. Using this ending does not require and connecting vowel to be inserted in the future tense.

The perfect passive participle (PPP) can be inferred from the root NDAK, yielding a verb *dag- (to slay) with infinitive degi and past tense danc and PPP dangen (slain) - this verb however is already halfway into the next class (A1) to be discussed. The gerund may be inferred from gonod- (to count up)10) and the dervied from aronoded (countless) - evidently it is formed by appending -ed to the basic verb form. No examples for the present active and perfect active participle can be found - the best guess is that they are formed by appending the endings -el and -iel respectively, for the last form, the stem vowel is probably lengthened and changes in the transition from ON to N (OS to S) according to to rules described above.

Hence we find the prototype conjugation: Class P1

  • Stem: *ped- (to speak) Infinitive: *pedi (to speak)
  • Present tense: pêd (speaks) with endings *pedin (I speak)
  • Past tense: pent (spoke) with endings *pennin (I spoke)
  • Future tense: *peditha (will speak) with endings *pedithon (I will speak)
  • Active participles: *pedel (speaking) pídiel (having spoken)
  • Passive participle: *pennen (spoken) Gerund: peded ([the] speaking)

Analogical developments from class P1 verbs (Class A1)

There seems to be a class of verbs which can be understood if one assumes that some forms of the (more numerous) derived verbs have started to replace the ‘genuine’ older forms of the P1 conjugation. The first hint is the verb dag- (to slay) which has both past tenses danc (corresponding to P1) and degant. The last form shows an ending -ant which looks like a common past tense of derived verbs; cf. CE: tirjâ N: *tiria- (to watch)11) with past tense tiriant (*watched). However, this ending is appended to a form which shows internal i-affection, indicating that the ending may really be appended to the infinitive of the verb (with the final -i lost in the process), hence not **dagant but degant. The verb treneri (to be discussed below) shows a similar analoguous formation of the past tense as trenerant.

But there are verbs for which the analogy is even one step further - those show infinitives in -o (as typical for derived verbs) and at least one possible past tense in -ant (but since the infinitive in -o doesn’t cause i-affection, the past tense is likewise not i-affected). The primary example is sog- (to drink)12) with infinitive sogo, present tense sôg and past tenses sunc, sogant. Here, in the primary past tense the stem vowel turned to -u- reflects the fact that the underlying root SUK has the -u-, and this can survive if ‘screened’ behind the cluster -nc created by the past tense nasal infixion. Theorizing that the endings are really appended by analogy, we can assume that the connecting vowels for the endings would be -i- in present tense, -e- for the past tense (see the example dravo from the class A2 below), hence *segin (I drink) but *sungen, *sogannen (I drank). Since these verbs are progressively close to derived verbs, the best guess for the future tense is to insert a connecting vowel -a- for any primary verb that shows a past tense in -ant, but there are no attested examples. Hence *sogatha (will drink) or *degatha (will slay), the latter assuming the same internal i-affection as in the past tense formation.

The form sogannen13) may be 1st person sg. of the past tense or the PPP - this is hard to tell. If it is a PPP, the PPP could be derived by means of the ending -en appended to the analogical past tense in -ant. On the other hand, the form dangen (slain)14) is evidently derived from the original past tense danc and not the analogical form degant. Evidence from the roots NAR and TIR suggest however that the PPP tends to be fossilized and should probably rather be derived from the original and not the analogical past tense, hence *sungen (drunk).

For the gerund, no attested examples are available. The general line of reasoning suggests that gerunds would also be rather fossilized forms and reflect the original rather than the analogical form, even more since the present tense of A1 verbs is identical with the present tense of P1 verbs. Probably the gerund is therefore formed using -ed as ending, hence *soged (drinking) *daged (slaying). No examples for the present active and perfect active participle can be found - the best guess is that they are formed by appending the endings -el and -iel respectively, for the last form, the stem vowel is probably lengthened and changes in the transition from ON to N (OS to S) according to to rules described above. As prototype conjugations, we may present:

Class A1a (only analoguous past tense)

  • Stem: dag- (to slay) Infinitive: *degi (to slay)
  • Present tense: dâg (slays) with endings *degin (I slay)
  • Original Past tense: danc (slew) with endings *dengin (I slew)
  • Analoguous past tense: degant (slew) with endings *degannen (I slew)
  • Future tense: *degatha (will slay) with endings *degathon (I will slay)
  • Active participles: *dagel (slaying) *dógiel (having slain)
  • Passive participle: dangen (slain) Gerund: *daged ([the] slaying)

Class A1 (infinitive, past tense and presumably future analoguous)

  • Stem: sog- (to drink) Infinitive: sogo (to drink)
  • Present tense: sôg (drinks) with endings *segin (I drink)
  • Original Past tense: sunc (drank) with endings *sungen (I drank)
  • Analoguous past tense: sogant (drank) with endings *sogannen (I drank)
  • Future tense: *sogatha (will drink) with endings *sogathon (I will drink)
  • Active participles: *sogel (drinking) *súgiel (having drunk)
  • Passive participle: sogannen, maybe also *sungen (drunken) Gerund: *soged ([the] drinking)

Primary Verbs ending with CE roots ending in nasal consonants (m, n) (Class P2)

Verbs of this class do not seem to be different from the class P1 except for a peculiar quirk in the formation of the past tense. We have the attested haf- (to sit)15) from the root KHAM with past tense hamp(sat) and hemmin (I sat). This almost looks like a P1 verb, except that the past tense hamp looks like it would have been derived from a root KHAP by nasal infixion - but of course this can hardly be the case, therefore the appearance of the -p is surprising. Probably this form has developed in analogy with the more numerous endings -mp coming from nasal infixion where the root ends with a stop. The fact that the connecting vowel in the past tense is -i- and not -e- may give some confidence that the infinitive of this verb would be *hevi and that it is not influenced by more analogies with derived verbs. Unfortunately, this interpretation is not without a problem, because the verb has an alternative past tense hafant which would point to an infinitive *havo since the analogical past tense is not subject to i-affection, i.e. we don’t see **hevant in analogy with degant. Bearing this in mind, we will nevertheless quote the verb here as the best example of what we can infer about the class P2. Its conjugation might look like

Class P2 (primary verbs ending in nasal)

  • Stem: haf- (to sit) Infinitive: *hevi (to sit) or *havo (to sit)
  • Present tense: hâf (sits) with endings *hevin (I sit)
  • Past tense: hamp (sat) with endings hemmin (I sat)
  • Analoguous past tense: hafant with endings *hafannen (I sat) (may not be present for all P2 verbs)
  • Future tense: *hevitha (will sit) with endings *hevithon (I will sit)
  • Active participles: *havel (sitting) *hóviel (having sat)
  • Passive participle: *hammen (sat) Gerund: *haved ([the] sitting)

We don’t really know for sure if other verbs with roots ending in nasals would also show the equivalent of the curious -mp in the past tense - it seem possible, though.

Analogical developments from class P2 verbs (Class A2)

Not surprisingly, there is also evidence that verbs ending in nasals can be subject to the influence of analogies with derived verbs. The best example is a verb dravo (to hew) derived from the root DARAM which shows the archaic past tense dramp (hewed) with drammen (I hewed) and the additional past tense dravant (hewed). This is the best evidence that the A verbal classes would have their past tense with the connecting vowel -e- instead of -i- even for the original past tense (since the analoguous past tense would be *dravannen). We may assume that besides the slightly irregular -p in the past tense, this class in general follows the pattern of A1:

Class A2 (infinitive, past tense and presumably future analoguous)

  • Stem: draf- (to hew) Infinitive: dravo (to hew)
  • Present tense: *drâf (hews) with endings *drevin (I hew)
  • Original Past tense: +dramp (hewed) with endings drammen (I hewed)
  • Analoguous past tense: dravant (hewed) with endings *dravannen (I hewed)
  • Future tense: *dravatha (will hew) with endings *dravathon (I will hew)
  • Active participles: *dravel (hewing) *dróviel (having hewed)
  • Passive participle: *drammen, maybe also *dravannen (hewed) Gerund: *draved ([the] hewing)

Primary Verbs with CE roots ending in liquid consonants (r, l) or voiced stops (b, d, g) (Class P3)

There is a class of primary verbs for which we do not see nasal infixion but some strengthening of the stem vowel. This may either be a lengthening or an a-infixion. A good example is heli (to lift) from the root KHAL (hence the stem is apparently hal-) with past tense haul. For some verbs, this formation may already have been present in CE since Quenya has a similar way of forming the past tense, cf. unduláve (downlicked) in Namárië from the verb lav-16).

For others, however, this past tense may have developed later on. The entry NAR2 of the Etymologies17) and the information from the ‘Addenda and Corrigenda’ can help shed some light on this: We find the ON past tense narne, but a poetic Noldorin past tense is given as narante and the compound verb treneri (obviously derived from tre-nar- (to tell through) has the past tense trenor which supposedly represents an intermediate *trenaur where we again see strengthening of the root vowel. Thus, it may well be that for some verbs strengthening of the root vowel as a means of past tense formation is a rather recent development. The entry TIR18) in combination with the participle tirnen (guarded) seen in Talath Dirnen19) can add some more information: the participle again reflects a ON past tense ending -ne added to the bare stem - the Noldorin past tense however seems to be tiriant, analoguous to the case of NAR, this form of inflexion seems to have been lost somewhere in the transition from ON to N. Finally, the ‘Addenda and Corrigenda’ show a verb from the root *nidh- (hurt, bruise) from the root NID2 with (supposedly) past tense nîdh and corresponding ON past tense níde. Tolkien discusses this formation of the past tense in his Early Qenya grammar (unfortunately for a different language):

“The past stem is obtained by the suffix -ye, (…) -ie, or -ne, but -ie (the commonest) is normally accompanied by stem strengthening consisting of (1) a-infixion (2) n-infixion, (3) vowel lengthening (this last perhaps largely an analogical extension from the â in many stems.” (from PE14)

Nevertheless, apart from *nidh-, all other attested verbs using this way of past tense formation have a stem vowel -a- or -u- in their CE root and can be explained by a-infixion, so it is not impossible that the Early Qenya scenario holds in Noldorin as well and that lengthening of the stem vowel is only a later development. This suggests the following scenario: Verbs with roots ending in voiced stops b, d and g and stem vowel a or u could well have a preferred original past tense by strengthening of the root vowel, cf. Q: lávë (licked) which makes N: *lhauf (licked) a likely guess. On the other hand, for verbs with stems ending in liquids r, l or stems in b, d, g with stem vowels other than a or u this formation would not be the original one an older past tense in -ne could still be reflected in some forms like the participles, so a guess for ‘told’ would be *narnen. Indeed, we find for the root WED a verb gwedi (to bind) with past tense gwend - exhibiting nasal infixion because the stem vowel is e and not a or u.

Likewise, we find the Quenya past tense carnë (*made)20) but the Sindarin form is said to be agor21). This form shows in addition to the strengthening of the stem vowel an ‘augment’, a repetition of the stem vowel as prefix (this seems very similar to the Quenya formation of the perfect *acárië by the way) - but it is attested in ‘Quendi and Eldar’, a much later source than the Etymologies - it certainly is a long shot, but maybe in the late scenario, there should be an augment in addition to the strengthening of the stem vowel? We don’t know for sure, but we will assume this scenario in the following. Neither of these verbs shows clearn indication of what the connecting vowels in the presence of endings may be - but since they form infinitives in -i, it seems reasonable that they would follow the regular pattern and insert -i- in present tense. For the past tense, there is one form known where a derived verb shows a past tense by stem vowel strengthening: ónen (I gave) and another one where the augmented form agor receives an ending: agorech (whatever -ch may be here, the connecting vowel is -e-). From this, we may infer that -e- is the most likely vowel for this class of past tense formations. For the future, we may again turn to the class P1 and try an -i- as the best guess. Hence we may try

Class P3 (primary verbs ending in liquids or voiced stops)

  • Stem: *hal- (to lift) Infinitive: heli (to lift)
  • Present tense: *hâl (lifts) with endings *helin (I lift)
  • Past tense: haul (lifted) with endings *holen (I lifted)
  • Future tense: *helitha (will lift) with endings *helithon (I will lift)
  • Active participles: *halel (lifting) *hóliel (having lifted)
  • Passive participle: *holen (lifted) Gerund: *haled ([the] lifting)
  • Stem: *nidh- (to hurt) Infinitive: *nidhi (to hurt)
  • Present tense: *nîdh (hurts) with endings *nidhin (I hurt)
  • Past tense: nîdh (hurt) with endings *nídhen (I hurt)
  • Future tense: *nidhitha (will hurt) with endings *nidhithon (I will hurt)
  • Active participles: *nidhel (hurting) *nídhiel (having hurt)
  • Passive participle: *nídhen (hurt) Gerund: *nidhed ([the] hurting)

Incidentally, the personless present tense and past tense seem to be both nîdh and cannot be distinguished. However, whoever thinks that this would be a major problem is invited to figure out what tense ‘You hurt me.’ is supposed to be - English has the same problem.

Analogical developments from class P3 verbs (Class A3)

Not surprisingly, analogical formations are also present for this class of verbs. We find rather weak traces of them in the alternative past tense trenerant zu treneri where we again see that this form is subject to i-affection and presumably derived directly from the infinitive removing the final -i. The formation seems to be largely analoguous to what we have termed above class A1a. For other verbs, we see again more pronounced deviations from what is presumably the original development (P3) in both infinitive and past tense, examples include naro (to tell) with past tense narante, garo with present tense gerin (I hold) and past tense garant22) and melo (to love) with past tense melant23). Again, since the infinitive of these formations is in -o, they do not show any internal i-affection. Since for these class of verbs the past tense by stem vowel strengthening is really replaced by the analoguous past tense, they presumably do not show any difference to the class A1 verbs.

We do have an example of the perfect active participle for those verbs - tíriel (having watched) from LR/RGEO:72, corresponding to the verb *tir-. This is the main basis for the suggested scenario of stem vowel lengthening and an ending -iel. The past tense of the verb tiri may be interpreted to be tiriant, so the verb as such is probably to some degree analogical. However, the perfect passive participle tirnen (watched) seems to indicate that in some forms, an original CE past tense formation in -nê survived. Since the ON past tense form of the verb naro is given as narne, it is probably not unreasonable to assume that this verb would still retain the participle*narnen (told) in spite of the fact that its past tense would now be narante and formed by a different mechanism. Hence:

Class A3a (only analoguous past tense)

  • Stem: *trenar- (to recount) Infinitive: treneri (to recount)
  • Present tense: *trenar (recounts) with endings *trenerin (I recount)
  • Past tense: trenor (recounted) with endings *trenoren (I recounted)
  • Analoguous past tense: trenerant (recounted) with endings *trenerannen (I recounted)
  • Future tense: *treneritha (will recount) with endings *trenerithon (I will recount)
  • Active participles: *trenorel (recounting) *trenóriel (having recounted)
  • Passive participle: *trenoren or maybe *trenarnen (recounted) Gerund: *trenared ([the] recounting)

Class A3 (infinitive, past tense and presumably future analoguous)

  • Stem: *gar- (to hold) Infinitive: garo (to hold)
  • Present tense: *gâr (holds) with endings *gerin (I hold)
  • Past tense: garant (held) with endings *garannen (I held)
  • Future tense: *garatha (will hold) with endings *garathon (I will hold)
  • Active participles: *garel (holding) *góriel (having held)
  • Passive participle: *garannen (held) Gerund: *gared ([the] holding)

Transitive derived verbs (Class D1)

As mentioned above, derived verbs are not directly formed from the verbal stem but with the help of an ending. Common endings include -jâ, -tâ, -nâ, -bâ and possibly . Given their large number, surprisingly few forms of derived verbs are directly attested. The verbal stems in Noldorin/Sindarin of these verbs end in -a - this is not so for basic verbs which usually end with a consonant (hence this class of derived verbs is often referred to as A-verbs). An example of such a bare stem can be found in LRW:391 as trasta- (to harass, to trouble), derived from a root TARAS.

In the present tense 3rd person/personless form, these verbs apparently show the bare stem, see the examples thia (it appears)24) to the verb thio (to seem), to the root THĒ - apparently, this is not a stem verb since those cannot acquire an -a in the 3rd person. We might see true personless use in e aníra from the King’s Letter - if this corresponds to a verb aníra-. If inflected for endings, in the first person the final -a of the stem is changed into an -o-, cf. le nallon (to thee I cry)25), le linnon (to thee I sing)26) where the verb is clearly derived from the root LIN227). This doesn’t seem to be the case for other endings, the Ae Adar28) has gohenam (we forgive) and SIL has Dor Firn i Guinar (Land of the dead that live), showing that the endings are appended to the stem if they are different from the first person. The infinitive of derived verbs can be read off from the example thio29) quoted above. This in fact agrees well with the overall scenario outlined in this acticle that the infinitives in -o seen for some derived verbs represent analogical formations inspired by the forms of the more numerous A-verbs. There are quite a lot of infinitives in -o found in the Etymologies where we cen deduce (e.g. from their Quenya counterpart) that the verb has to be derived. To give some examples, tirio (to watch) from the root TIR as CE: *tirjâ30), ortho (to rise) from the root ORO31) as CE: *orotâ or harno (to wound) from the root SKAR32) where the Quenya equivalent Q: harna- is given.

The past tense of these forms (at least for transitive verbs) can be inferred from the examples tiriant (*watched)33) and orthant (*raised)34) or the form teithant (wrote) seen in the Moria gate inscription corresponding to a verb teitho (to write) derived from a root TEK35) - obviously, it is formed by appending -nt to the verbal root. Again, this agrees nicely with the scenario that basic verbs would have ‘inherited’ a past tense -ant by analogy with the more numerous derived verbs. Since a past tense formation in -ant is not seen in Quenya, this is presumably in itself a development genuine to Sindarin and not present in the CE inflection of derived verbs. The question of the connecting vowel for endings in the past tense is a bit tricky - we don’t have a clear example, but the fact that the A2 verb dravo (which should be closer to derived verbs than a P2 verb) shows drammen (hewed) and not **dremmin gives some indication that this vowel ought to be -e-. Likewise, the form sogannen might be a past tense corresponding to sogant (drank) - if so, this would again point to a connecting vowel -e-.

Imladris (© John Howe)

The future can be read of from anglennatha (will approach) from the King’s Letter - apparently this is a word derived from the root LED by means of a primitive ending -nâ, and the future tense is simply formed by appending -tha to the verbal stem. If endings are present, this behaves apparently just like the verb in present tense, e.g. the -a is shifted into -o- in the first person but remains else. See for example le linnathon (to thee I will sing)36) and estathar in the King’s Letter (refraining from interpreting the latter form, it seems to be the Sindarin equivalent of the Quenya verb Q: esta-37) with the future ending -tha).

For the present active participle, we find some evidence in LRW:358 where a verb glavro (to babble) is given from the root GLAM in addition to a form glavrol (babbling). Likewise, LRW:388 has a verb chwinio (to whirl) derived from the root SWIN and a form chwiniol (whirling). Thus, the best guess for the formation of participles for derived verbs is that the ending -a is replaced by -ol. This is in fact fully consistent with the assumption that present active participles in Sindarin and Quenya are both derived from a primitive CE ending -lâ. We can only guess what the perfect active participle may be, there’s no attested form. It is commonly assumed to be formed by strippind any vocalic endings from a derived verb, appending -iel and carrying out a lengthening of the stem vowel (if possible) along with the necesary sound shifts (this is based on the analogy with the basic verbs and the perfect tense formation in Quenya). For the perfect passive participle however, we do have attested examples - for the verb presto (to affect) we find the form prestannen (affected)38) - indicating that the perfect passive participle is apparently formed from the past tense with the help of the ending -en, just as in the case of stem verbs. Finally, we may infer the gerund from the example eithad (*insult) corresponding to a verb eitha- (to insult)39) - apparently it is formed by appending -d to the verbal stem. Hence, the conjugation of such a verb my look like

Class D1 (transitive derived verbs)

  • Stem: *teitha- (to write) Infinitive: teitho (to write)
  • Present tense: *teitha (writes) with endings *teithon (I write)
  • Past tense: teithant (wrote) with endings *teithannen (I wrote)
  • Future tense: *teithatha (will write) with endings *teithathon (I will write)
  • Active participles: *teithol (writing) *teithiel (having written)
  • Passive participle: *teithannen (written) Gerund: *teithad ([the] writing)

Intransitive derived verbs (Class D2)

There are several examples of derived verbs showing a past tense different from -(a)nt. The first class of verbs shows a past tense in -(a)s, see e.g. mudas (*laboured) from mudo (to labour, to toil)40), erias(*arose) from erio (to rise)41) and istas (*had knowledge) from isto (to have knowledge)42). Especially the difference between erio → erias (rose) and ortha → orthant (raised) in combination with the fact that all attested verbs with past tenses in -s are intransitive seems to indicate that this (rather than) -ntis the preferred ending for intransitive derived verbs. Note that there is no reason to believe these verbs would somehow an exception - past tenses in -s are about attested as often as past tenses in -nt for derived verbs.

We do not have any special indication that tenses other than the past tense would be different for the intransitive verbs - it is possible, but for the time being it seems best to assume that the only difference occurs in the past tense only. A past tense ending -s has to be derived from a longer -ssV (with V a vowel), otherwise it would be lenited to -h following a vowel and eventually be lost. This longer ending would be restored whenever the verb receives pronomial endings. We can only make an educated guess what the vowel should be - the most likely solution is -e- in analogy with the transitive derived verbs. The vowel -i- in the past tense seems to be relevant for basic verbs only. On the other hand, the passive participles of stem and derived verbs alike usually seem to be formed directly from the past tense - if they exist (which is not obvious for intransitive verbs because they would need to take a different meaning), they should probably derived from the ending -sse- as well, hence ?istassen (known) (note that there is no real passive participle for a verb ‘to have knowledge’). Hence, the complete conjugation might look like:

Class D2 (intransitive derived verbs)

  • Stem: *muda- (to labour) Infinitive: mudo (to labour)
  • Present tense: *muda (labours) with endings *mudon (I labour)
  • Past tense: mudas (laboured) with endings *mudassen (I laboured)
  • Future tense: *mudatha (will labour) with endings *mudathon (I will labour)
  • Active participles: *mudol (labouring) *múdiel (having laboured)
  • Passive participle: ?mudassen (?done) Gerund: *mudad ([the] labouring)

Intransitive and causative derived verbs with strong past tenses (Class D2a)

For the verb isto (to have knowledge), there is an past tense given as +sint (had knowledge) in addition to istas43). This strong past tense formation is a bit surprising for a derived verb. There are, however, more examples of this phenomenon: The verbs *lhimmid- (to moisten)44) and *nimmid- (to whiten)45) have past tenses lhimmint (moistened) and nimmint whitened (We quote these verbs here as *lhimmid-, nimmid- assuming Tolkien’s entries lhimmid, nimmid are supposed to represent verb stems. The verb aphad- (to follow)46) derived from *ap-pata (the star written by Tolkien) shows that indeed the final -a of a verb can be shed to give a shortened stem and lends some support to this interpretation). And yet these verbs are derived from the roots LINKWI and NIK-W and are apparently the result of a causative verbal ending -tâ appended to adjectives, hence CE: *linkwi-tâ ‘to cause to be wet’ and CE: *ninkwi-tâ ‘to cause to be white’. It is interesting that these verbs seem to have lost the final -a of this ending, though, since the stems given in the Etymologies clearly don’t show this. It is possible that the thing these examples have in common is that a causative meaning ‘to whiten, to make white’ can easily blend with an intransitive meaning ‘to whiten, to become white’. If so, the example +sintwould suggest that intransitive verbs can form strong past tenses even if they are derived verbs, and the blending of the forms would then produce lhimmint, nimmint by analogy. We have too few examples to be sure, but it seems to be a likely explanation.

We have one more clearly causative verb in the corpus; tangado (to make firm)47), and some for which it is not obvious if they were derived using a causative ending, but which are nevertheless longer forms where the ending is applied to a verbal stem, e.g. lhathrado (to listen in)48) (cf. lathro (to eavesdrop)), gannado (to play a harp)49) (cf. ganno (to play a harp)). It seems plausible that the past tense of tangado should be *tangant to agree with the examples lhimmint, nimmint - but what about the present tense of these verbs? Do they fall into the same class, i.e. should the infinitive of *lhimmid- be *lhimmido, or is the loss of the final vowel somehow peculiar to these two examples and the present tense should rather be *tangada-? What connecting vowel needs to be insterted?

There is no easy answer to these questions, but the following scenario seems most likely: Both *nimmid- and *lhimmid- seem to have had an ending -a at a former stage. Therefore, their infinitive might well be preserved as *lhimmido and *nimmido in order to agree with tangado. Their shortening to *lhimmid- instead of **lhimmida- may be a rather recent phenomenon caused by an analogy with their past tense (which in the case of +sint is explicitly denoted as archaic/poetic). If so, they would presumably be conjugated like basic verbs in present tense, i.e. *lhimmidin (I moisten), *nimmid (he whitens). It is mere speculation if the same thing would happen to tangado - I would think *tangadon (I make fast) more likely since Tolkien doesn’t remark anything more about this form, but ?tengedin may just be possible. What would the connecting vowel in the past tense be? Since we theorized that these verbs did originally end in -a, they may be similar to the A class verbs and have -e- as connecting vowel, cf. drammen, but we don’t know this for sure. As for the other verbs mentioned, there is no reason to believe that they would represent particularly causative meanings blending with intransitive verbs - but they are of course intransitive themselves, so they would presumably show past tenses in -a(s) or even the strong past tense formation. As for other tenses, there is no particular reason to assume that they would be different from verbs of the classes D1/D2. Hence:

Class D2 (intransitive derived verb with optional strong past tense)

  • Stem: *ista- (to have knowledge) Infinitive: isto (to have knowledge)
  • Present tense: *ista (has knowledge) with endings *iston (I have knowledge)
  • Past tense: istas (had knowledge) with endings *istassen (I had knowledge)
  • Alternative past tense: sint (had knowledge) with endings *sinten (I had knowledge)
  • Future tense: *istatha (will have knowledge) with endings *istathon (I will have knowledge)
  • Active participles: *istol (having knowledge) *istiel (having had knowledge)
  • Passive participle: ?istassen (?known) Gerund: *istad ([the] having knowledge)

Class D2a (causative derived verb with strong past tense and altered present tense)

  • Stem: *nimmid- (to whiten) Infinitive: *nimmido (to whiten)
  • Present tense: *nimmid (whitens) with endings *nimmidin (I whiten)
  • Past tense: nimmint (whitened) with endings *nimminnen (I whitened)
  • Future tense: *nimmidatha (will whiten) with endings *nimmidathon (I will whiten)
  • Active participles: *nimmidol (whitening) *nimmidiel (having whitened)
  • Passive participle: *nimminnen (whitened) Gerund: *nimmidad ([the] whitening)

Derived verbs with past tense formation by strengthening of the root vowel (Class D3)

We have one clear example of a derived verb showing past tense formation by strengthening of the root vowel - this example is ónen (I gave)50), presumably the past tense of anno (to give)51). It is a bit surprising to find such a past tense for a derived verb, especially since the root ends neither with a voiced stop nor with a nasal. However, considering that the usual formation **annant (gave) and **annannen (I gave) would be somewhat awkward, one can perhaps understand how this form came to pass. On the other hand, it may be that óne- is actually the old past tense form - we have no way to be sure. The form ónen is presumably derived from *aun (gave) by means of a connecting vowel -e-. We have no reason to assume that the other tenses should somehow be irregular, the perfect passive participle however probably ought to be ónen as well. The verb delia (to conceal) shows the poetic/archaic past tense +daul next to the more usual deliant52).

There is one more verb which may belong to this class: for thoro (to fence)53) we have the perfect passive participle thoren (fenced) which would point to a past tense *thaur (fenced). The trouble with this interpretation is that we cannot be sure if this is really a derived verb ?thora-, since it may well be that this is actually a verb ?thor- with an analoguous infinitive in -o. However, if so we could expect an analoguous alternative past tense ?thorant, and since this is not mentioned explicitly, the odds are slightly in favour of the interpretation that this is in fact a derived verb (and its past tense is irregular). We don’t know for sure. In any case, the conjugation of our known example would be

Class D3 (derived verb with past tense by strengthening of the root vowel)

  • Stem: *anna- (to give) Infinitive: anno (to give)
  • Present tense: *anna (gives) with endings *annon (I give)
  • Past tense: *aun (gave) with endings ónen (I gave)
  • Future tense: *annatha (will give) with endings *annathon (I will give)
  • Active participles: *annol (giving) *anniel (having given)
  • Passive participle: *ónen (given) Gerund: *annad ([the] giving)

Irregular verbs (Class I)

There are still verbs which do not seem to fit in any of the classes outlined above. The past tense of egledhi (to go into exile)54) is given as egledhas or eglant (*went into exile). Both forms appear to be irregular shortenings of the expected forms egledhias and egledhant (the first one the D2 past tense for intransitive verbs to *egledhia-, the second one the analoguous A1 past tense to *egledh-. Lacking any additional information under what circumstances these irregularities may occur, we do not know if this formation pattern would affect other verbs as well. LRW:354 has a verb dant- (to fall), which would be very irregular in the sense that it has a two consonant cluster as ending of the root - it is unclear how such a verb would form its past tense, the present tense would probably be dant (falls) with endings *dennin (I fall). There are several more irregular verbs listed below - their detailed discussion is somewhat beyond the scope of this article and will be postponed.

Some (popular) examples

Since the systematics of Sindarin verbs outlined here has quite drastic consequences for the conjugation of some verbs compared with the ‘Suggested Conjugation’ of the Ardalambion, here are some examples of how often used verbs might look like:

aphad- (to follow)55) is very likely a D2a with the final -a stripped off in present tense, so this ought to be:

  • Stem: aphad- (to follow) Infinitive: *aphado (to follow)
  • Present tense: *aphad (follows) with endings *ephedin, maybe *aphedin (I follow)
  • Past tense: *aphant (followed) with endings *aphannen (I followed)
  • Future tense: *aphadatha (will follow) with endings *aphadathon (I will follow)
  • Active participles: *aphadol (following) *aphadiel (having followed)
  • Passive participle: *aphannen (followed) Gerund: *aphadad ([the] following)

*dar- (to stop)56) would be of class P3, since the root ends with a liquid and the stem vowel is a, but like *nar- it might retain an original past tense formation in -ne in the participle:

  • Stem: dar- (to stop) Infinitive: deri (to stop)
  • Present tense: *dâr (stops) with endings *derin (I stop)
  • Past tense: *daur (stopped) with endings *doren (I stopped)
  • Future tense: *deritha (will stop) with endings *derithon (I will stop)
  • Active participles: *darel (stopping) *dóriel (having stopped)
  • Passive participle: *darnen (stopped) Gerund: *dared ([the] stopping)

*gir- (to shudder)57) might be in class P3 or A3a, so its past tense would either be *gîr (lengthening of the stem vowel in analogy with *nidh-) or *giriant (formation analoguous to *tir-. Maybe both are permissible. Again, the participle (if it exists at all, after all this is not a transitive verb) should reflect an older form in analogy with tirnen:

  • Stem: *gir- (to shudder) Infinitive: giri (to shudder)
  • Present tense: *gîr (shudders) with endings *girin (I shudder)
  • Past tense: *gîr (shuddered) with endings *gíren (I shuddered)
  • Analoguous tense: *giriant (shuddered) with endings *giriannen (I shuddered)
  • Future tense: *giritha (will shudder) with endings *girithon (I will shudder)
  • Active participles: *girel (shuddering) *gíriel (having shuddered)
  • Passive participle: ?girnen (?shuddered) Gerund: *gired ([the] shuddering)

thinna- (to fade, to grow towards evening) would be an intransitive verb (D2) and hence probably show past tense in -(a)s:

  • Stem: thinna- (to fade) Infinitive: *thinno (to fade)
  • Present tense: *thinna (fades) with endings *thinnon (I fade)
  • Past tense: *thinnas (faded) with endings *thinnassen (I faded)
  • Future tense: *thinnatha (will fade) with endings *thinnathon (I will fade)
  • Active participles: *thinnol (fading) *thinniel (having faded)
  • Passive participle: ?thinnassen (?faded) Gerund: *thinnad ([the] fading)

Verbs and their classes

It remains the labourious task of sorting the attested and semi-attested verbs into the different classes. Note that for many verbs, this is only a suggestion, not much can be deduced e.g. from an infinitive only. Please note that this list includes what is commonly assumed as being a more or less valid Sindarin verb without giving the reference where it is attested. Please refer to one of the compiled Sindarin dictionaries for this information. Some verbs which have been previously included in the ‘mixed conjugation’ are assigned into different classes here and consequently appear without a final -a, hence the list contains e.g. sog- instead ofsoga-. For all we know, Noldorin verbs (and probably their Sindarin counterparts as well) seem to have more than one way of e.g. forming their past tense. Several verbs listed below may well belong to more than one class, listing them only in what I think is most likely class is therefore to some degree artificial.

  • P1: bad-, blab-, cab-, echad-, gad-, gonod-, govad-, hab-, had-, heb-, mab-, mad-, nag-, nestag-, nod-, ped-, rib-, trevad-
  • P2: can-, cen-, draf-
  • P3: car- (however with augment in past tense), dar-, gal-, gir-, glir-, lav-, orthor-, osgar-, tol-
  • A1: dreg-, rad-, sog-, tob-
  • A1a: dag-, gwedh-, redh-, tog-
  • A2: drav-
  • A3: gar-, mel-, orthel-, pel-, tir-, thel-
  • A3a: fir-, gir-, glir-, nor- (?), sol-, trenar-, tir-
  • D1: adertha-, anglenna-, aníra- (possibly a distinct present tense), awartha-, bartha-, baugla-, beria-, bertha-, boda-, breitha-, brona-, bronia-, buia-, critha-, dagra-, damma-, delia-, díhena-, dilia-, doltha-, dringa-, edra-, egleria-, eitha-, ercha-, ertha-, esta-, fara-, fuia-, ganna-, genedia-, glavra-, gleina-, gohena-, gonathra-, gosta-, gruitha-, gwatha-, gweria-, gwesta-, haltha-, harna-, hartha-, hasta-, heltha-, henia-, heria-, hortha-, iuitha-, lasta-, lathra-, leitha-, linna-, luitha-, luithia-, maetha-, matha-, minna-, naegra-, nalla-, nara- (past tense narante, PPP *narnen), narcha-, nasta-, nautha-, neitha-, nella-, nesta-, nuitha-, oltha-, ortha-, pada-, panna-, pathra-, pelia-, penia-, presta-, puia-, rada-, rista-, ritha-, rosta-, sautha-, seidia-, thuia-, taetha-, tamma-, teitha-, telia-, tiria-, tíra- (possibly a distinct present tense), toltha-, tortha-, trasta-
  • D2: adlanna-, batha-, cuia-, cuina-, dartha-, dortha-, eria-, faltha-, gwanna-, hwinia-, ista- (alt. past tense sint), laba-, lacha-, mista-, muda-, ovra-, penna-, renia-, revia-, síla- (possibly a distinct present tense), siria-, thia-, thilia-, thinna-, tinna-, tuia-
  • D2a: athrad- (?), aphad-, gannad- (?), lathrad- (?),limmid-, nimmid-, tangad-
  • D3: abonna-, anna-, delia-, edonna-, onna-, suilanna-, thora-
  • I: ava-, banga-, boe, dant- (?danna-), edledh-, elia-, gaw-, neledh-, *yr-

The verbs aníra-, tíra- and síla- might represent the Sindarin analogues to the present tense (what has been glossed as present tense above would then be the Aorist). If so, these verbs should rather read anir-, tir- and sil-.


In the preparation of this article, I have made heavy use of both The Past-Tense Verb in the Noldorin of the Etymologies by Carl F. Hostetter and Reconstructing the Sindarin Verb System by Helge Fauskanger, often as a reference to what forms are attested where. However, much of what I have written about the past tense rests on the foundations laid in Carl Hostetter’s article. I would like to thank Lothenon and Carl Hostetter for helpful comments during the preparation of this article.

See also

On Tolkiendil

On the net

1) , 5) , 9) LRW:363
2) , 12) , 13) LRW:388
3) LRW:384
4) , 40) LRW:373
6) PM:401
7) VT41:6
8) VT41:11
10) , 45) LRW:378
11) , 18) , 30) , 33) LRW:394
14) LRW:375
15) VT:45
16) LRW:367
17) LRW:374
19) UT:465
20) LRW:362
21) WJ:415
22) , 23) , 42) , 43) , 52) ‘Addenda and Corrigenda’
24) , 29) LRW:392
25) , 50) LR
26) LB:354
27) , 44) LRW:369
28) VT44:21
31) , 34) LRW:379
32) LRW:386
35) LRW:391
36) LR; RGEO:72
37) LRW:356
38) LRW:380
39) WJ:365
41) LRW:379; ‘Addenda and Corrigenda’
46) , 55) WJ:387
47) LRW:389
48) , 54) LRW:368
49) LRW:377
51) LRW:348
53) LRW:393
56) LRW:353
57) LRW:358
langues/english/i-lam_arth/sindarin_verb_system.txt · Dernière modification: 27/11/2010 09:41 par elendil
Partager sur
Nous suivre sur
Tolkiendil - - Tous droits réservés © 1996-2017