Copyright © 2001 by Piotr Gąsiorowski

The durative includes imperfective or progressive meanings (prolonged, continuous or incomplete actions) and may also refer to repeated or habitual activity. It has preterite forms (a.k.a. imperfects) and present-tense forms (known as presents, for the sake of brevity).

The aorist expresses perfective or non-progressive meanings (completed actions). It does its job mainly in the preterite tense, where it functions rather like the English past simple.

The perfect (or stative) aspect denotes a present state resulting from an action.

The forms of the perfect are very characteristic. They usually involve reduplication (with the vowel *e) and a special kind of ablaut: the stressed *o-grade of the root in the singular forms, and the unstressed nil or reduced grade in the plural:

The very common perfect stem *woid-/*wid- ‘know’ (from *weid- ‘see, perceive’) is never reduplicated.

The distinction between durative and aorist verbs is trickier, since these aspects need not be overtly signalled. Verbs are regarded as inherently durative or aorist, and the marked (non-default) aspect is expressed by derived stems.

For example, the root *b^her- ‘carry’ is inherently durative, so the preterite *b^hér-e-t (corresponding to the present *b^hér-e-ti ‘he carries’) means ‘he was carrying’. To express a completed action, one would use a specially marked aorist form, in this case *b^hé:r-s-t ‘he carried, lifted’ (a so-called sigmatic aorist).

The aorist may also have a distinctive stress pattern (with stress falling on the thematic vowel), e.g. *b^hug-é-t ‘he escaped’ as opposed to *b^héug-e-t ‘he was running away’, or employ reduplication, e.g. *we-uk^w-é-t ‘he said’ as opposed to *wék^w-t or *wék^w-e-t ‘he was saying’.

Conversely, *doh₃-t ‘he gave’ contains an aorist stem. To express the corresponding durative aspect (‘he was giving’), we can use a reduplicated stem with *i as the reduplication vowel, *di-doh₃-t.

{jeug-} ‘connect’

{g^when-} ‘strike’

{derk^{^}-} ‘look’

Stem *g^whén-/*g^whn- ‘strike’

A stem may consist of a root with a vocalic appendix (*-e-/*-o-) called the thematic vowel (hence the term “thematic conjugation”):

Stem *bHér-e- ‘carry’ (note the constant stress pattern and root vocalism)

Partial reduplication in which a copy of the root-initial consonant or consonantal cluster is followed by the vowel *i is employed to form thematic and athematic “intensive” verbs (a conveniently vague term, implying an intensified or vigorous activity) or durative counterparts of aorist stems:

Transitive “intensives” of a different kind involve the suffix *-áh₂-/*-h₂- added to the weak form of a root to produce athematic verbs:

The suffix *-je-/*-jo- forms thematic durative verbs:

A similar suffix is used to derive durative verbs from nouns:

When added to verb roots with the *o-vocalism, *-eje-/*-ejo- forms causatives (a very important formation, historically, as it remained productive in several branches):

The nasal suffix *-néu-/*-nu-, usually enforcing the weak vocalism of the root, produces (often transitive and vaguely causative) athematic verbs that refer to the beginning or termination of an action (the learned word for that is ‘inchoative’), or suggest that something is done once (rather than repeated). A rarer variant of this pattern involves *-nu- formations with stress alternating between the full-vowelled root and the inflection. A closely related formation involves verbs in *-náh₂-/*-nh₂-.

Similar functions can be attributed to the so-called nasal infix *-né-/*-n-, which is normally inserted after a liquid or semivowel (R = *w, *j, *r, *l) in *CeRC- roots, producing the characteristic alternation *CR-né-C-/*CR-n-C- preserved in Indo-Iranian.

The suffix *-sk^{^}e-/*-sk^{^}o-, usually added to nil-grade bases, forms iterative (or inchoative) stems. Its common variant is *-isk^{^}e-. Apparently, the same *-sk^{^}e- can also produce denonimal duratives like *med^hu-sk^{^}e- ‘get drunk’ (from *med^hu ‘mead, intoxicating drink’) or *wod-sk^{^}e- ‘wash’.

Also with reduplication:

The suffix *-ah2- added to adjectives produces “factitives”: *X-ah₂- means ‘make X’:

The suffix *-eh₁- and the combinations *-eh₁-s- (?) and *-eh₁-sk^{^}e- yield intransitive verbs denoting change of state (‘become X’)

A specialised preterite marker (the so-called “augment”, a stressed particle reconstructed as *h₁e, meaning ‘then’ and placed before the verb, e.g. *h₁é weuk^wet ‘[then] he said’) occurs only in a few branches (Greek and Phrygian, Indo-Iranian, Armenian) and may be a late local innovation restricted to the southeastern IE dialects.

Sets of endings contrasting the present with the preterite exist for all the three persons singular and the third person plural:

Athematic stems (*g^when-/*g^whn- ‘strike’)

Thematic stems (*b^her-e- ‘carry’)

The morphologically simpler preterite endings are called ‘secondary’, while the present endings (usually involving the present-tense marker *-i) are called ‘primary’. These traditional terms are rather confusing, and I’m in favour of avoiding them.

The remaining endings – the first and second persons plural and the dual endings of all three persons (PIE had a three-way number system) can be reconstructed only approximately. In their early history they were probably prone to dialectal variation and analogical restructuring:

(*N = *m or *n; the colour of the thematic vowel shown in square brackets)

Some of the inflections of the perfect are completely unlike those found in the other two aspects. This is how a perfect stem (here *wóid-/*wid- ‘know’) is conjugated:

The present-tense marker *-i could apparently be added to some of these endings, e.g. *woid-h₂a-i ‘I know’, but this extension seems to have been non-obligatory. What is really curious about the endings of the perfect is the regularity with which they match the pattern of the so-called hi-conjugation in Hittite (especially -hi, -ti, -ai in the singular). The formal correspondence compels us to regard these formations as different survivals of the same ancestral pattern despite the fact Hittite hi-verbs do not correspond to non-Anatolian perfects etymologically and their stem-internal structure is different (they are not reduplicated, while typical perfect stems are).

One conspicuous feature of the non-perfect paradigm is the first person singular ending *-o: of thematic presents (instead of expected *-o-m-i). Various analyses have been put forward to account for this curious asymmetry – so curious, in fact, that it seems to conceal something of historical importance, as morphological irregularities often do. Miguel Carrasquer Vidal has just contributed a brand-new hypothesis that attempts to analyse IE verb inflections as agglutinated personal pronouns; *-o: is explained there as the outcome of a regular phonological process (with what is reconstructed as pre-PIE *-a:-m^w-i becoming *-owi > *-o:u > *o: rather than *-omi). The explanation that I find most convincing, if not unproblematic, is that *-o: reflects PIE *-o-h₂, an ending related to that of the perfect. Many scholars have argued that the *-s-i, *-t-i and *-nt-i endings of thematic verbs are analogically modelled on those of the athematic conjugation, and that the original endings were like those found in the IE perfect and the hi-conjugation in Hittite, yielding patterns more or less like this:

Apparently a number of athematic verbs (e.g. those with characteristic *o-grade presents, like *k^{^}onk- ‘hang’) were conjugated in a similar way:

As, technically speaking, these stems are athematic, the general terms “thematic” vs. “athematic” should be replaced with something more appropriate. One could speak, very tentatively, of the “first conjugation” (singular forms in *-m-, *-s-, *-t-) and the “second conjugation” (singular forms in *-h₂(a), *-th₂(a), *-(e), with “thematic” and “athematic” subclasses).

It is hard to tell to what extent or by what means tense was signalled in the second conjugation. If *-i was employed as the present-tense tense marker, we should expects presents like 1sg. *k^{^}ónk-h₂a-i or 3sg. *k^{^}ónk-e-i, which could indeed be supported by some comparative evidence. However, it is also possible that preterite endings were levelled out very early, so that, for example, the preterite *b^hér-e-s contrasted with a present like *b^hér-e-th₂a no matter if the latter was additionally marked with *-i or not.

If there is any substance to these highly speculative reconstructions, the IE “perfect” may turn out to be just one of several archaic formations belonging to a distinct conjugation, the other well-preserved example being the Hittite hi-conjugation. There were apparently more formations of that class in PIE (see section 5 on the PIE middle voice), but analogical changes obliterated them almost everywhere in the family, leaving only isolated traces ignored by earlier scholarship. What can be said at present is that IE studies are at the crossroads in this respect: the classic reconstruction no longer appears satisfactory, but there is little consensus as to what should replace it.

The imperative mood was used in commands, requests and prohibitions. As in many other languages, it lacked first-person forms. It is clear that this category included formations of different origin.

The “injunctive”, attested chiefly in Indo-Iranian but supported also by archaic Greek and Hittite forms, is formally indistinguishable from the preterite indicative (except for the fact that it was never accompanied by tense-markers such as the preverbal augment in Indo-Iranian and Greek). It can be regarded as an “underspecified” form of a verb, inflected for person and number (and voice as well, though I’ll ignore it here) but having no overt tense or mood markers. For the stem *weg^{^h}-e- ‘carry, transport’ we have the following forms:

For example, *wég^{^h}-e-t, when used as an injunctive, can express meanings such as ‘let it carry’, ‘it must carry’, or ‘may it carry’. In post-Vedic Old Indic the use of the injunctive was by and large restricted to prohibitions with the prohibitive particle ma: (< *meh₁). The 2pl. form *wég^{^h}e-te occurs very widely across the IE family as the exclusive type of imperative for that person/number combination.

One widespread and apparently very archaic variety of the 2sg. imperative consists of the bare stem without any inflections (this is analogous to the formation of PIE vocatives):

In the case of root verbs, the root was more frequently in the nil grade, followed by the particle *-d^hi, as in the following examples:

The third-person imperative includes evidently old forms in *-tu (3sg.) and *-ntu (3pl.), which look like injunctives with an obscure clitic extension (*-u). They are found in Indo-Iranian and Hittite, and so can be plausibly projected back to PIE:

The so-called future imperative in *-to:t is attested in Indo-Iranian, Greek and Latin, and probably represents an innovation of ‘non-Anatolian IE’ date; *to:t looks like the ablative of the demonstrative pronoun *to- (contracted *to-et), so the ending may be interpreted as ‘hence’, i.e. ‘from now on; in the future’:

Here, either the 3pl. form has analogical *-(o)nt- corresponding to 3sg. *-(e)t- (assuming that the original imperative was *b^hére-to:t across the board), or both third-person forms have been simplified (from conjectural *b^héret-to:t, *b^héront-to:t) by dropping the inflection-final *-t before an enclitic. Similar forms occur for athematic verbs:

The optative mood expresses wishes, choices or preferences. It is rather well attested, though absent from Anatolian. In athematic verbs the optative stem-extension is *-jéh₁- alternating with *-ih₁-. The suffix is followed by the ordinary personal endings (without the present-tense marker *-i):

There is no alternation (and no paradigmatic stress-shift) in thematic verbs, where the suffix occurs invariably in the nil grade. The thematic vowel is also invariably realised as *-o-, e.g.

The subjunctive mood is used in clauses expressing doubts, fears, predictions, hypothetical or guarded statements (as opposed to assertions). Hence its use in conditional constructions, various kinds of subordinate clauses (speaking of things inferred, assumed or reported second-hand), and as a surrogate future tense -- the future being inherently uncertain.

The subjunctive is moderately well attested. It is probably a rather late category; at any rate, there is no trace of it in Anatolian.

The subjunctive can be based on durative or aorist stems and seems to have taken the ordinary present-tense endings of the thematic conjugation. Interestingly, subjunctives corresponding to athematic indicatives look exactly like their thematic counterparts:

The subjunctive of thematic verbs shows a lengthened thematic vowel, as if resulting from contraction:

The contrast between the two voices must have been, at least partly, a matter of stress placement at some point in the internal history of PIE, middle forms having once sported final stress falling on inflectional endings, whose accented variants ended in *-o. Athematic verbs often preserve the original stress variation and the corresponding vowel-grade contrasts. In some branches the middle preterite has the following endings (the reconstruction is problematic for persons other than those given):

For example, *jug-s-tó [juk.'sto] ‘he joined (= became allied with, met, etc.)’ as opposed to *jé:ug-s-t [je:ukst] ‘he united (something with something else)’.

The middle present could have the same endings as the middle preterite plus the present-tense marker *-r (unique to the middle) or *-i (shared with, and presumably borrowed from, the active), e.g.:

However, there is evidence suggesting that the spread of these formations is a late (post-PIE) phenomenon, and that there was also an old middle with endings like those of the perfect (and of the Hittite hi-conjugation) attached to durative or aorist stems. Those endings are especially well preserved in Anatolian, but their PIE status is supported (to a varying degree) by Italic, Celtic, Tocharian and Indo-Iranian evidence. The reconstruction is still extremely uncertain and the examples below are only meant to give you an approximate idea of what such forms may have looked like:

It is thinkable, in fact, that both formations are old. The detailed analysis of the middle, however, is not an autonomous problem; it must be discussed together with the question of the origin and early form of the PIE active conjugations – and here some points are so difficult and speculative that their discussion would get us well beyond the scope of an introductory presentation.

As if matters were not complicated enough, the first two persons have a unique set of middle endings in the plural and dual numbers:

The *-d^hh₂ of the first person is obviously a clitic enlargement of the normal plural and dual endings. I will refrain here from speculation about its origin (or about the structure of the 2pl. ending).

The suffix of the present participle is *-(o)nt-, added to verb stems. One peculiarity of this formation is that the thematic vowel is deleted before the suffix: *b^hér-ont- ‘carrying’, etc. In the declension of participles derived from athematic stems stress falls on *-ont- in some cases and on the inflectional ending in others; the stem typically occurs with weak vocalism (except, perhaps, in the inanimate class, see below). A similar vocalic alternation, but without stress variation, is analogically mimicked by thematic verbs: *b^hér-ont-/*b^hér-nt-.

The declension of a typical root participle (*h₁s-ont- ‘being, existent’) looks like this:

In post-PIE times, when the animate class split into the feminine and the masculine genders, most non-Anatolian branches developed additionally the feminine form *-nt-ih₂-.

A number of original present participles were lexicalised as nouns already in PIE times; they include, amongst others, *h₁dont- ‘tooth’ from *h₁ed- ‘eat’ and *h₂uh₁ont- ‘wind’ from *h₂weh₁- ‘blow’.

The suffix of the preterite active participle is *-wós-/*-us- (with the variant *-wot-/*-ut-, suggesting hard-to-reconstruct heteroclitic complications in its early declension). This formation yielded, among other things, participles derived from perfect stems. For example, the perfect *le-loik^w- ‘be rid of’ could form the following participles (only animate forms are given):

The feminine variant, of late origin, ends in *-us-íh₂.

Before it became associated with perfect stems, the suffix *-wós-/*-us- had a more general function, forming athematic verb participles with the meaning ‘having X-en’. The ablaut pattern was perhaps something like this:

Root: *weid- ‘see, perceive’ (perfect *woid- ‘know’)

The present middle participle (corresponding, as to its meaning, to the English “past” participle in its processual rather than stative sense), is reconstructed with the thematic suffix *-m(e)n-ó-. Its oldest form was probably *-mh₁n-, with *-h₁- easily dropped in the nil grade *-mh₁n-ó- > *-mn-ó-. Post-PIE gender assignment followed the normal pattern of thematic adjectives:

Several types of deverbal adjectives, semantically akin to middle participles, were incorporated into the conjugational system in various branches of the family, functioning as preterite middle participles, expressing stative meanings. The most important of them involve the stressed suffixes *-tó- and *-nó- (sometimes also *-mó-), originally added directly to weak-grade verb roots:

In their capacity as participles, they tended in some branches to be derived from stems rather than roots, e.g. *b^her-e-tó- ‘brought’, instead of *b^hr-tó-. Germanic strong-verb past participles in *-ana- < *-onó- possibly reflect simplified *-o-m(h₁)n-ó-.