Quenya has five simple verb tenses: aorist (timeless), past, future, present/imperfect and perfect. Aside from the “default” aorist tense, two of these tenses have to do with time (past and future) and two have to with the verbal aspect of the action: imperfect (an ongoing action not yet complete) and perfect (a completed action). Unlike Quenya, the English language addresses the imperfect and perfect aspects by using various auxiliary verbs: “I am making” (imperfect/ongoing) and “I have made” (perfect/completed). In English these aspects can be combined with the past tense to form a past imperfect (“I was making”) and past perfect (“I had made”). Likewise you can form an English future imperfect (“I will be making”) and future perfect (“I will have made”).

Quenya has similar compound tenses combining time and verbal aspects. Strictly speaking, these Quenya tenses are not “compound tenses” as in English, since they did not involve an auxiliary verb, but they serve a similar function. Tolkien referred to them as either “derivative tenses” (PE22/105) or less accurately as “compound tenses” (PE22/122), and I use the latter term for convenience so I don’t have switch terminology between Quenya and English. The most detailed descriptions of the compound tenses appears in the Quenya Verbal System (QVS) of the late 1940s, but there are hints of it in Tolkien’s later writing as well, notably:

Make a Q. past continuous. “was eating”. mātante. istante, oryante, ortante.
More like aorist in syntax. and also helps to explain wide use of -ant as past in S.
better mātānē > mātane (Late Notes on Verb Structure, 1969, PE22/157).

Setting aside the apparently rejected mātante, the form mātane seems to be a past continuous (aka. past imperfect) tense, and it is essentially the same as the past imperfect appearing in QVS, at least for basic verbs:

The past imperfect is used less than the English form “I was making”. This is formed in Quenya by the addition of past suffix of “weak” verbs -nē to the bare “present” stem. Thus kárane “was making”; túlane “was coming” (PE22/100-101).

In QVS the imperfect (aka. present continuous) form of derived verbs was often based on the imperfect participle orta- “rise” → ortalya “is rising”, and the past imperfect was thus ortalyane “was rising” (PE22/100, second half of footnote #5). In the 1960s, Tolkien introduced a new present continuous/imperfect form for derived verbs: ortea (or ortia). Perhaps the past imperfect of derived verbs in the 1960s would likewise have become orteáne (or perhaps ortiáne) with prosodic lengthening.

Similarly, in QVS the past perfect, also known as the pluperfect, was a combination of the perfect tense with the past suffix -ne, but in QVS this form was based on the (active) perfect participle, or perhaps the “long perfect” (itself based on the perfect participle):

A weak “pluperfect” was made in Quenya, by adding to the perfect participle the past-suffix -nḗ. So karnelyane “I was having made = I had made”; túlielyane “I had come”, lasselyane “I had heard”, etc. (PE22/104).

Tolkien’s explanation for the origin of the pluperfect in QVS is quite convoluted:

See further under weak verbs. Since these could not normally distinguish between past and perfect [in footnote: sc. when they had no strong pa.t.] their perfect was periphrastic: thus ni nahtanelya “I (am) having slain” = “I have slain”; but also where no pronoun was concerned Orome nahtanelya “O. has slain”. From the latter developed the form nahtanelye or by analogy with the strong verbs anahtalye “has slain”. The final vowel -e is on analogy of the normal perfect ending. These are called “long perfects” (PE22/104).

This discussion of the long perfect immediately follows that of the pluperfect, and it was presumably this long perfect that was the basis for how the pluperfect was formed. Tolkien’s description of the origin of the long perfect nahtanelya (later nahtanelye) is itself a bit mysterious. It is based on the assumption that weak verbs “could not normally distinguish between past and perfect”. However, in the finished version of QVS weak verbs generally had distinct past and perfect forms, such as:

• tă-formative orta- “rise” → past oronte “rose” and perfect orórie or (rare) orontie “has arisen” (PE22/115).
• tā-causative orta- “raise” → past ortane “raised” and perfect ortanie “has raised” (PE22/117).

However in the initial drafts of QVS, the pasts and perfects were more confused:

[From a rejected draft] In Q. there is no distinction at all at any period between the significance or uses of the two different formations [past and perfect]: they are used primarily as true “past tenses”, the past of the “aorist” ... In Quenya the forms mantē-, karnē, tūlie became recognized all with “past” meaning (PE22/102 note #13).

Thus Tolkien’s introduction of the long perfect may have been part of those earlier drafts where the ancient past and perfect forms were more confused, necessitating the introduction of a more distinct long perfect. Hat tip to Raccoon for pointing out this possibility to me.

Turning to the future tense, the future imperfect was similarly formed by adding the QVS future suffix -(u)va to the (present) active participle which also functioned as the long imperfect (analogous to the long perfect mentioned above and likewise introduced due to ambiguities in the imperfects of derived verbs). Similarly, the future perfect was formed by adding the future suffix to the (normal) perfect form:

Imperf. future: karalyuva, etc. (PE22/109).

A future perfect was similarly made by adding vā to the perfect stem: thus akariéva “shall have made”; similarly utúliéva, aláviéva, etc. The augment is often omitted (PE22/109).

There was also a past future form, that is “the future from the perspective of the past”:

A past-future = “was about to, was going to” is frequently made by suffixing -nē to the future stem: thus karuváne “was going to make”; later usually accented karúvane (PE22/105).

The variation between karuváne and karúvane was due to the vagaries of Quenya stress patterns. There was not a corresponding “past from the perspective of the future”; presumably the future perfect akariéva served this role. There was, however, the rather arcane “past future perfect” formed by adding the past suffix to the future perfect. This obscure tense was only used for conditionals:

A rare form, a past future perfect, “I was going to have made”, (a)kariévane is sometimes found, chiefly = “would have gone” in conditional clauses. See syntax of verbal forms.

Rounding out the QVS compound tenses, there was also a “consuetudinal past” used to describe habitual actions in the past. This was formed by adding the past suffix to the imperfect participle. This participle was distinct from the normal present/imperfect form of basic verbs, but indistinguishable from the present/imperfect for many derived verbs:

A “consuetudinal past” only used in the sense “I used to make” can be formed by suffixing -ne to the imperfect participle (q.v.). Thus karalyane; tuluryane “I used to come” (PE22/101).

Tolkien also described a “consuetudinal future” formed by combining the imperfect participle with the future suffix; this would in all cases be indistinguishable from the future imperfect described above:

Also occasionally used is a consuetudinal or imperfect future = “I shall be making, I shall go on making, it will be my habit to make”, expressed by adding the future (q.v.) uva to the participle, karalyuva (PE22/101).

To summarize the QVS compound tenses:

• Past imperfect = imperfect + ne: kárane = kára + ne (the same form reappears in the 1960s).
• Past perfect/pluperfect = past active participle/long perfect + ne: karnelyane = karnelya + ne.
• Future imperfect = present active participle/long imperfect + uva: karalyuva = karaly(a) + uva.
• Future perfect = perfect + (u)va: akáriéva = akárie + (u)va.
• Past future = future + ne: karuváne = karuva + ne (later karúvane).
• Past future perfect = future perfect + ne: akáriévane = akáriéva + ne.
• Consuetudinal past = imperfect participle + ne: karalyane = karalya + ne.
• Consuetudinal future = imperfect participle + uva: karalyuva = karaly(a) + uva.

The QVS consuetudinal future and future imperfect forms were identical, and how they were distinguished contextually isn’t clear.

Conceptual Development: In the Early Quenya Grammar (EQG) of the 1920s there was no distinct imperfect or perfect verb forms. The imperfect and perfect aspects were expressed, like in English, by combining an auxiliary verb (various tenses of ᴱQ. e- “to be”) with present, past and future (active) participles [tulin(d-), túlien(d-), tuluvan(d-)], thereby forming various compound tenses (PE14/57). This could happen in two ways, first by having the auxiliary verb precede the participle:

• ᴱQ. e tulien (= “is having come”) “has come” (perfect) [“is” + past participle].
• ᴱQ. va tulien (= “will be having come”) “will have come” (fut. perfect) [“will be” + past participle].

More commonly the form of “to be” was suffixed to the participle:

ᴱQ. tulinde	“is coming”	pres. imperf.	[present participle + e “is”]
ᴱQ. tulindie	“was coming”	past. imperf.	[present participle + ie “was”]
ᴱQ. tulinwa	“will be coming”	fut. imperf.	[present participle + va “will be”]
ᴱQ. tuliende	“has come”	perf.	[past participle + e “is”]
ᴱQ. tuliendie	“had come”	past perf. (pluperfect)	[past participle + ie “was”]
ᴱQ. tulienwa	“will have come”	fut. perf.	[past participle + va “will be”]
ᴱQ. tuluvande	“is going to come”	fut. imperf.	[future participle + e “is”]
ᴱQ. tuluvandie	“was going to come”	past fut.	[future participle + ie “was”]
ᴱQ. tuluvanwa	“will be going to come”	fut. fut.	[future participle + va “will be”]

There do not appear to be any examples of these particular compound tenses outside EQG.

There is also a mention of an alternate “past future” form in some notes from the early 1940s:

√UB “ponder, have in mind” ... The verb only survives in form -uva used as future ending in Q. The old str. pa.t. is seen in OQ umne future-past, matumne “I was going to eat” (VT48/32).

It is not clear how matumne relates to the past future form in QVS: matuváne or matúvane “was going to eat”. It could be a distinct paradigm or it could be an archaic (OQ.) form but still part of the same conceptual paradigm. However, a similar construction was mentioned in Common Eldarin: Verb Structure from the 1950s:

In AQ an old basic past tense (of intransitive form [of the verb uv-] with suffix nē: see above) was still in use to form analytically the “future in the past”: {umneste >>} umbeste mate “he was going to eat” (PE22/132 and note #16).

In this 1950s paradigm, {umne >>} umbe was an auxiliary verb rather than a suffix. Both uses of umne/umbe from the 1940s and 50s seem to be archaic and no longer part of “modern” Quenya.

Neo-Quenya: There is a 1969 note that mentions a past imperfect form mātane (PE22/157), which indicates that Tolkien still imagined compound tenses to be part of Quenya into the late 1960s. This form happens to agree with the past imperfect form of QVS. However, many of QVS compound tenses are based on the 1940s various active participles (with the suffix -lya) that had intruded into the tense system as the long perfect and long imperfect. In Tolkien’s later writings (a) the 1940s participle suffix -lya was abandoned and replaced by -(i)la in the 1960s and (b) there is no sign that the participles played a role in the tense system. Thus it is not obvious how (or if) QVS compound tenses can be updated to fit Tolkien’s later conception of the language.

We do have one late example of a compound tense: a past imperfect mátane that is simply a combination of the imperfect verb form with the past tense suffix -ne. We could extrapolate from this pattern to get compound tenses by combining various verb forms with past/future suffixes, with prosodic lengthening as appropriate:

• Past imperfect: mátane “was eating” = máta (present/imperfect) + ne.
• Past perfect/pluperfect: *(a)mátiéne “had eaten” = amátié (perfect) + ne.
• Future imperfect: *mátauva “will be eating” = máta (present/imperfect) + uva.
• Future perfect: *amátiéva “will have eaten” = amátie (perfect) + (u)va.
• Past future: *matúvane “was going to eat” = matuva (future) + ne.
• Past future perfect: *amátiévane “would have eaten” = amátiéva (future perfect) + ne.

The above is basically the same as the QVS system, except that it uses the base perfect and imperfect stems instead of participles to form the pluperfect and future imperfect. Alternately, you could generalize the QVS system by updating to the 1960s participle forms: pluperfect *mátiélane “had eaten” = mátiéla (past active participle) + ne or future imperfect *mátaluva “will be eating = mátala (imperfect active participle) + uva.

It’s not entirely clear how the future suffix -uva would combine with other tenses in the paradigm of the 1960s. In QVS, this suffix was -(u)va from primitive vā, with u inserted only where necessary if there was a preceding consonant. Thus in QVS the future perfect was amátiéva with no u. But in the 1960s the future was derived from √UB, and the u was a more intrinsic feature of the future inflection. However, it is possible that the ancient long ē in the perfect would “overwhelm” the (short) future ŭ, so that *amātiyē + ŭbā > amātiyēbā > amátiéva. As for the patterns of prosodic lengthening, it is possible that -(u)vane was perceived as a single suffix (like -lion) and so the stress preceded it.

I suspect the compound tenses of derived verbs would follow the same patterns, such as: ortea + ne = past imperfect orteáne “was rising” (attested óreáne from orya- seems to be an example of this) or avánie + uva = future perfect avániéva “will have departed”. This would produce past future perfect avániévane “would have departed”, a form whose use is limited to subjunctives. Regardless, with the exception of the past imperfect for basic verbs (mátane), none of these forms are attested in Tolkien’s later writing and they remain very speculative. Of these, only the past imperfect (mátane) and the pluperfect (mátiéne or amátiéne) are likely to see much use in Neo-Quenya.