The infinitive is often a problem. An ending -ië is mentioned in UT:317, stated to be infinitival or gerundial: enyalië "to recall". (When this means "[in order] to recall", the dative ending -n is added: enyalien.) This ending can probably be used on all the basic verbal stems: *quetië "to say", *matië "to eat", *tulië "to come". Perhaps it can be used on some of the verbal stems ending in vowels; in Old Sindarin (the "ON" of the Etymologies) we find ortië as the infinitive of orta- "rise, raise" (LR:379), and Old Sindarin is quite close to Common Eldarin, the common ancestor of Quenya and Sindarin. So we may assume that the Quenya verbs in -ta, and possibly most verbal stems ending in a vowel, have infinitives formed by dropping the final vowel and adding -ië: anta- "give", infinitive *antië, harna- "wound", infinitive *harnië etc. (no attested examples). But what about the many verbs in -ya, like metya "put and end to"? Forms like *metyië are hardly possible, for the combination yi does not occur in the language. Perhaps the whole ending ya would disappear, the infinitive of metya being *metië, or perhaps the stem metya itself can be used as infinitive. We don't know. To some extent the whole problem can be circumvented: It seems that rather than using this gerund in -ië, Quenya often uses simply a verbal stem where English would have an infinitive. In the phrase áva carë "don't do [it]" (WJ:371), carë "do" seems to be an "aorist stem" (cf. below) formed from the basic verbal stem car- "make, do" with the ending -ë. Other basic verbs would behave in the same way, e.g. *áva tulë "don't come". Derived verbs (with endings like -ya and -ta) would presumably be unchanged in such constructions (invented example: *Áva hilya Fëanáro Endorenna, "don't follow Fëanor to Middle-earth", the derived stem hilya "follow" being used as infinitive).
The present (or continuative) tense is formed with the ending -a. On the numerous verbal stems that end in a already, such as the -ya and -ta verbs, it is of course "invisible": lanta is the stem of the verb "fall", but also the present tense "falls, is falling". When -a is added to the stem of a basic, "strong" verb, the stem vowel seems to be lengthened: sil- "shine" > síla "is shining", mel- "love" > méla "is loving" (not simply **sila,**mela). The corresponding "aorist" forms (silë, *melë, see below) rather mean "shines", "loves", without the continuative aspect.
The basic verbs also have a distinct form called aorist by Christopher Gilson and Patrick Wynne, though its function cannot be exactly the same as the aorist of Classical Greek. Our best example of the aorist is the sentence i carir quettar ómainen "those who form words with voices" (WJ:391). This is a description of the Elves, so here the aorist verb carir "form" denotes a "timeless truth" beyond specific tenses. The aorist has the ending -ë, that changes to -i if any ending is added. RS:324 indicates that the aorist of sil- "shine" is silë, or silir with the plural ending -r (added when the subject is plural; see below). In the Etymologies in LR:347-400, many basic verbs are given in the 1. person aorist (with the ending -n "I"), e.g. lavin "I lick", tirin "I watch" (see LAB, TIR), some also in the 3. person aorist (e.g. tinë "it glints" under TIN). In VT39:11, macë (there spelt "make") is translated "hews with a sword". This is also an "aorist" form, assuming that this is the proper term. It seems that the relationship between macë and the corresponding present-tense (or continuative) form *máca closely corresponds to English "hews" vs. "is hewing".
The past tense of the regular derived verb seems to be formed with the suffix -në, e.g. orta- "rise, raise", pa.t. ortanë (cf. a line from Namárië: Varda...máryat ortanë *"Varda.... raised her hands", translated "Varda...has uplifted her hands" by Tolkien). The "strong", basic verbs often form their past tense with nasal infixion + the ending -ë, e.g. quet- "say" > quentë "said", top- "cover" > tompë "covered". (The infix has the form m before p and n otherwise.) But in the case of basic stems ending in -r or -m, the ending -në is suffixed instead, as in the case of the derived verbs: tir- "watch", pa.t. tirnë "watched", tam- "tap", pa.t. tamnë "tapped". Likely, basic verbal stems ending in N behave in the same way: cen- "see", pa.t. perhaps *cennë "saw" (not attested). When a stem ends in -l, the suffix -në seems to be assimilated to -lë, so the past tense of wil- "fly" is willë. (Compare ullë "poured", a past tense apparently formed directly from the stem UL [WJ:400 - but ULU in LR:396]. But "poured" in the transitive sense is ulyanë, formed from the derived verb ulya-.) There are also a few verbs that form their past tense by lengthening the stem vowel and adding -ë; for instance, the past tense of lav- "lick" is lávë. (For a while, Tolkien may have been toying with the idea of making this the universal system; in LR:46, 72 we find túlë as the past tense of tul- "come" and cárë as the past tense of car- "make", instead of the expected forms *tullë, carnë - the latter is actually given in LR:362, under KAR.)
The perfect tense of basic verbs is formed by dropping the final vowel of the stem (if there is any), adding the suffix -ië, lengthening the stem vowel and prefixing an augment to the stem. The augment is identical to the stem vowel, so verbal stems like hat- "break", tec- "write", ric- "twist", top- "cover", tul- "come" have the augments a, e, i, o, u, respectively. Hence, the perfects of these verbs must be *ahátië "has broken", *etécië "has written", irícië "has twisted" (attested in VT39:9), *otópië "has covered", utúlië "has come" (the latter form is attested in several places: utúlie'n aurë "the day has come", Silm. ch. 20; Endorenna utúlien "to Middle-earth I am come", LotR3/VI ch. 5). Non-basic, derived verbs like harna- "wound" or horta- "urge" cannot lengthen the vowel because it is followed by a consonant cluster: perfects aharnië, ohortië?
We have no good example of the perfect of ya-verbs like harya- "possess"; a form like ?aharyië is probably not possible, for yi does not seem to occur in the language. My guess is that the ending -ya disappears with no trace, and the remaining part of the verbal stem can be treated as "basic": perfect ahárië. This is also the guess of Christopher Gilson and Patrick Wynne in a succinct Quenya grammar of theirs (if indeed they are guessing).
In the corpus, there are some examples of perfects that lack the augment, e.g. fírië "hath breathed forth" in MR:250 (for *ifírië, not attested).
The future tense is formed by dropping the final vowel of the stem (if any) and adding the suffix -uva; examples from Namárië include enquantuva "shall refill" and hiruva "shall find". The stem of the former (ignoring the prefix en- "re-") is quat- "fill" (WJ:392), suggesting that the verbs that show nasal infixion in the past tense do so also in the future.
The imperative takes the ending -a, just like the present tense, so laita = present tense "blesses, praises" or imperative "praise!" (In this case, the stem ends in -a already, so both endings are "invisible".) Basic verbal stems are distinct in form from the imperatives, since the latter do not lengthen their stem vowels in the imperative: SIL- "shine" > present tense síla "shines", imperative *sila! "shine!" An independent particle a or á is often used with imperatives: A laita te, laita te! "Bless them, bless them!" (the praise Frodo and Sam received on the Fields of Cormallen, translated in Letters:308); á vala Manwë "may Manwë order it" (WJ:404, lit. *"o rule Manwë!"). The imperative does not normally show number (whether the command is directed to one or several persons). However, imperatives may take the optional pronominal endings -t and -l, denoting sg. and pl,. respectively.
We don't know if Quenya has a distinct subjunctive or optative form. There is a "wishing formula" that involves nai "be it that" + a verb in the future tense, cf. nai hiruvalyë Valimar in the final lines of Namárië. In LotR the translation "maybe thou shalt find Valimar" is given, but Tolkien's notes in RGEO indicates that this is really a wish: "May thou find Valimar!" Cf. also UT:317: Nai tiruvantes = "be it that they will keep it, may they keep it".
Finite verbs agree in number with plural subjects by adding the ending
-r, as in the first line of Namárië: laurië
lassi, "like gold fall the leaves", with the verb agreeing with
its plural subject lassi "leaves". Singular "like gold falls
a leaf" would be *laurëa lanta lassë
with no plural
-r on the verb. (Note how the adjective translated "like gold",
literally "golden", also shifts from a plural to a singular form - see