“Verbs” ākhyātam are those forms which have verbal endings (also called ākhyātāni), that is to say, finite verbs, which are inflected for the grammatical categories of person, number, diathesis (or “voice”), as well as tense and mood. The process of inflecting a verbal form for these categories is called conjugation.
English has a fairly minimal system of conjugation (I do, you do, he does, etc. for the present tense, I did, you did, he did etc. for the past), and expresses most of the distinctions of tense, mood, and diathesis through compound verbal expressions that make use of auxiliary verbs. In Sanskrit, by contrast, all of these categories are typically expressed in a single verbal form.
Finite verbs are so called because they are limited (finitum) by these grammatical categories; non-finite verbs do not express all of these categories. Usually only a finite form can function as a verbal predicate. Non-finite forms are used as the complements of certain types of verbal and adjectival phrases (as in the case of infinitives), or as the head of a subordinate clause (as in the case of converbs).
As noted above, verbal forms are understood by Indian grammarians refer to “processes” bhāvaḥ, in contrast to “existing things” sattvam, which are designated by nominal forms. However, because of Sanskrit’s rich processes of derivation, it is possible to express processes with nominal forms as well.
In Pāṇini’s grammar, the starting point for the formation of any finite verbal form, and several non-finite verbal forms, is a verbal root dhātuḥ. The next step is to select one of the lakāraḥ, abstract signs that combine the notion of tense and mood. One a tense and mood has been assigned with a lakāraḥ, a conjugational stem is formed through morphological processes such as reduplication (in the case of the perfect tense or liṭ) and suffixation (in the case of most other tenses and moods). In the latter case, the stem-forming suffix is selected lexically: certain verbal roots take certain stem-forming suffixes and not others. Finally, once a conjugational stem aṅgam has been formed, the conjugational endings tiṄ are added, which differ according to the grammatical categories that are to be expressed by them (such as person, number, and diathesis) and according to the tense–mood of the stem to which they are added. Each step of this process will be discussed in the following sections.
A verbal root is an abstract entity from which the process of verbal inflection begins. It expresses a basic meaning (e.g., “going”) that is specified by the addition of suffixes, specifically, the tense and mood suffixes (called lakārāḥ) to which the conjugational endings of verbs are added.
Most primary roots in Sanskrit are monosyllabic, as was evidently the case in Indo-European as well. The qualification “primary” is necessary because, in Sanskrit, a verbal root, in the strictly morphological sense of a unit to which tense and mood suffixes can be added, can be formed secondarily from either a nominal stem (a prātipadikam) or from another verbal root by means of certain suffixes (called sanādipratyayāḥ). These denominal and deverbal roots are almost always polysyllabic.
Learning the verbal roots and their meanings has long been one of the “first steps” of learning Sanskrit. Pāṇini’s grammar refers to a traditional “recitation of verbal roots” dhātupāṭhaḥ, and many such lists were edited and compiled.
Verbal roots in Proto-Indo-European could end in a consonant called a laryngeal, a sound which does not survive in Sanskrit but which has left traces here and there. One of these traces is that roots which historically ended in a laryngeal take the augment i before certain suffixes. (This i is called iṬ, and it is the regular outcome of a Proto-Indo-European laryngeal between consonants.) This augment was gradually extended to roots which did not end with a laryngeal in Proto-Indo-European. Thus one will have to know whether a given root is sēṬ (literally “with the augment i”) or aniṬ (“without the augment i”). Thus:
|root:||vr̥t sēṬ||hu aniṬ|
|root:||lū sēṬ||yā aniṬ|
Sēṭ roots will be noted in transliteration with a superscript i, as in lū.
As noted above, the class of sēṬ roots largely corresponds to roots that ended in a laryngeal in Proto-Indo-European. Because of the laryngeal’s effects on the preceding sounds, it’s often possible to guess whether a root belongs to this class or not on the basis of its form. Roots that end in voiceless aspirates (such as grath) and in long vowels (such as lū) are often sēṭ.
Every finite verb can be said to have both a tense, which expresses the time in which the action referred to by the verb occurred relative to the time in which the verb form is used, as well as a mood, which expresses the “way” (modus) in which the action is referred to, that is to say, whether it is something the speaker refers to as actually occurring (either in the past, present, or future), or as something that either might happen or ought to happen.
The five tenses of Sanskrit are:
- present: For referring to an action that takes place more or less more or less at the same time that the verbal form is used Aṣṭādhyāyī 3.1.123. The Sanskrit present tense corresponds to the English habitual present (“I go to the store”) and progressive present (“I am going to the store”).
- aorist: For referring to an action that takes place in the past, without further specification of time Aṣṭādhyāyī 3.2.110.
- imperfect: For an action that takes place in the relatively recent past. (“Before the present day,” according to Pāṇini’s definition in Aṣṭādhyāyī 3.2.111.)
- perfect: For referring to an action that takes place in the distant past. (“Beyond the speaker’s personal experience,” according to Pāṇini’s definition in Aṣṭādhyāyī 3.2.115.)
- future: For referring to an action that takes place in the future, relative to the speaker’s time.
Sanskrit therefore has three past tenses, and more can be added if we consider the common use of the past participle to be a past tense. For practical purposes, there is no difference in meaning between the imperfect, aorist, and perfect, although fastiduous authors will observe Pāṇini’s rules and they will refrain from using, for instance, a perfect-tense verb to describe an action which the speaker has personally witnessed.
The following moods may also be distinguished:
- indicative: For a referring to an action that actually did, does, or will take place. This is sometimes called the realis mood.
- optative: For referring to an action that ought to or ought not to take place, as well as an action that might or might not take place. This is the principal irrealis mood.
- imperative: For referring to an action that, from the speaker’s perspective, must or must not take place, especially when giving commands or orders.
Lists of Sanskrit moods often include the following additional moods, which are only used in very specific circumstances, and which in some cases are clearly derivative of the three moods above:
- subjunctive: So called because of its formal identity with the Indo-European subjunctive, but used in Vedic texts as a future tense.
- injunctive: In post-Vedic Sanskrit, this mood is used exclusively in prohibitions, in complementary distribution with the imperative; in Vedic Sanskrit, it is a tenseless verb used for events that occur “outside of time.”
- precative: A mood used for wishing blessings, formally similar to the optative.
- conditional: A mood used to express counterfactual conditions, of comparatively rare occurrence.
It is important to note, however, that the system of tenses and moods has been restructured in Sanskrit vis-à-vis what we can reconstruct for Indo-European. Most importantly, Indo-European distinguished a category of aspect that has been more or less folded into the category of tense in post-Vedic Sanskrit. There were three such aspects:
- imperfective: For referring to an action as occurring over an extended period of time, including continuous, progressive, or habitual action.
- aorist: For referring to an action as a punctual event.
- stative: For referring to an action as a state consequent upon a previous action.
Each of these three aspects was expressed with a particular form of the stem. Each aspectual stem “had” the indicative, subjunctive, imperative and optative moods, in the sense that endings that characterized these moods could be added onto a given aspectual stem. Moreover, in the case of the imperfective aspect, there were two tenses of the indicative mood, one indicating present time, and another indicating past time. Hence the verbal system that Sanskrit inherited from Indo-European looked something like the following:
|aspect||mood and tense|
|imperfective||indicative present (laṭ)|
|indicative past (laṅ)|
There are therefore three different “systems” of tenses and moods, corresponding to the three different aspects of Indo-European: the present (the inherited imperfective), the aorist (the inherited aorist), and the perfect (the inherited stative).
These systems have been modified in Sanskrit in a number of ways:
- First, the distinction between tense and aspect has been largely erased, so that the past indicative of the imperfective, the indicative of the aorist, and the indicative of the stative all function as past tenses, namely, as the imperfect, aorist, and perfect.
- Second, the non-indicative moods, besides those of the present system, are only used in the Vedic language. In the post-Vedic language, the imperative and optative moods can only be formed from the present stem.
- The subjunctive is no longer used in the post-Vedic language, having been functionally replaced with two future tenses.
In the above table, we have noted in parentheses those aspect-tense-mood combinations that are represented in Pāṇini’s system of lakārāḥ. As noted above, these lakārāḥ stand for fixed tense-mood combinations. They are listed below, together with the stem of the verb that they select for:
Pāṇini’s names reflect the patterns of these different tense-mood combinations. For example, what the present indicative and imperfect indicative share, in contrast to all of the other tense-mood combinations, is the medial element -a-, which signifies that these forms are built by adding a set of personal endings to the present stem without any additional suffixation. By contrast, the present indicative is characterized by a final -ṭ, whereas the imperfect indicative is characterized by a final -ṅ, and therefore they take two different sets of endings.
Among the categories that finite verbs are inflected for is person. Sanskrit distinguishes between the prathamapuruṣaḥ or “first person,” the madhyamapuruṣaḥ or “middle person,” and the uttamapuruṣaḥ or “last person.” Beware that these correspond to what we call the third person, the second person, and the first person, respectively. Sanskrit does not distinguish between an inclusive and exclusive first person.
Like nouns, finite verbs in Sanskrit are inflected for singular ēkavacanam, dual dvivacanam, and plural bahuvacanam numbers.
The diathesis (also called voice) of a verb, in linguistics, generally refers to the connection between the arguments of a verb, which are features of its syntax, to one or another of the participants in the verbal action (called thematic roles below), which are features of its semantics. In English, we distinguish between “active voice,” where the agent is the principal argument, or subject, of the verb, and “passive voice,” where the patient is the subject of the verb.
In Sanskrit, there is a clear distinction between verbal constructions wherein the subject is an agent, and verbal constructions where in the subject is a patient. The latter are called “passive” or “P-oriented” (for patient) constructions. These will be discussed below.
Sanskrit makes a further diathetic distinction within the category of “active” or “A-oriented” voice between parasmaipadám, “a word for another,” and ātmanēpadám, “a word for oneself.” This distinction maps onto what grammars in the Greek and Latin tradition call “active” (ἐνέργεια) and “middle” (μεσότης), respectively. The expression “middle,” in the Greek and Latin tradition, is meant to capture the fact that the subject of the verb in question is represented as neither the agent of a particular action, nor the patient of the verbal action, but something in between. Here is an analogous case from English:
- I am baking the bread in the oven. :: active voice, because the subject of the verb “bake” is represented as the agent of the verbal action.
- The bread is being baked in the oven. :: passive voice, because the subject of the verb “bake” is represented as the patient of the verbal action.
- The bread is baking in the oven. :: middle voice, because the subject of the verb “bake” is the bread, which is neither represented as the agent of an action of baking, nor the patient of the same action.
The “middle” voice, so construed, is used to represent the subject of the verb as a “non-agentive agent,” that is, someone who undergoes or experiences the action of the verb, but who is nevertheless not represented as the patient of an action performed by someone else. It is therefore often, cross-linguistically, with the following types of verbs:
- verbs of perceiving and experiencing;
- verbs of motion;
- other verbs referring to changes in state.
Note that these types of verbs are typically intransitive, that is, they do not take a direct object. They may, however, take dependents that agree with the subject: in “I feel tired,” “tired” is not a direct object, but an adjective that agrees with the subject of the verb.
Whereas the distinction between “active” and “passive” voice in Sanskrit is made by the verbal stem—one form of the stem being used for “A-oriented” constructions, and another for “P-oriented” constructions—the distinction between parasmaipadám and ātmanēpadám is made by the verbal endings. There are thus, for any verb, two sets of endings, and the choice of whether to use parasmaipadám or ātmanēpadám endings is mostly conditioned lexically, along the lines sketched above:
- Most transitive verbs have an agent subject, that is, the subject of the verb is an “agentive agent.” These verbs take parasmaipadám endings.
- indrō vr̥traṁ hanti “Indra slays Vr̥tra.”
- A few transitive verbs have an experiencer subject, that is, the subject of the verb is one who experiences something. These verbs take ātmanēpadám endings.
- indrō sōmaṁ bhuṅktē “Indra consumes the Sōma.”
- Among intransitive verbs, some will treat the subject as an “agentive agent,” and will hence take parasmaipadám endings (these verbs are usually called unergative in the linguistics literature):
- brāhmaṇāś calanti “The Brahmans are walking.”
- Some intransitive verbs, however, treat the subject as a “non-agentive agent,” and will take ātmanēpadám endings (these verbs are called unaccusative in the linguistics literature):
- sa manyatē “He is thinking.”
The verbal endings—or tiṄ, to use Pāṇini’s abbreviation for them—are the final element in a finite verbal form. The endings are what express the categories of person, number, and diathesis; additionally, each tense-mood combination lakārāḥ takes a specific set of endings.
As with the nominal endings, Pāṇini teaches the verbal endings in “triplets,” consisting of the singular, dual, and plural ending in the third person, second person, and first person, in that order. There are two such 3x3 matrices taught in Aṣṭādhyāyī 3.4.78, the first being the endings of the parasmaipadám and the second being the endings of the ātmanēpadám.
These two sets of endings are actually the starting-point for the derivation of additional sets of endings, because both the parasmaipadám and ātmanēpadám include special endings for each lakāraḥ. We will see, however, that Pāṇini has classified the lakārāḥ on the basis of the kinds of endings that they take, and therefore there is a broad division—subsequently discovered by philologists, who gave it a new name—between primary endings, used in lakārāḥ with the marker Ṭ, and secondary endings, used in lakārāḥ with the marker Ṅ.
In some verbal inflections, just as in some nominal inflections, there is a distinction between the strong and weak form of the stem, which historically arose from the interplay of the accent of the root, or stem-forming suffix, and the accent of the inflectional ending. The endings that require the strong form of the stem will be noted below.
This section will explain the parasmaipadám endings for the various lakārāḥ. We begin with the parasmaipadám endings that Pāṇini teaches as basic:
Note that the endings of the singular all have the marker P, which indicates that the accent is on the preceding morpheme; all of the other endings are accented. These singular endings also require the strong form of the stem, if the paradigm distinguishes between strong and weak forms. The endings listed above are the primary endings, which are used in the present indicative laṭ. If we “translate” Pāṇini’s endings into the forms that actually appear, we obtain the following 3x3 matrix:
|prathamapuruṣaḥ||-ti||-táḥ||-ánti / -áti|
Important points to notice about these endings are:
- The endings of the singular are unaccented, and additionally, they and the plural of the prathamapuruṣaḥ end in an element -i, which Indo-Europeanists have considered a “hic-et-nunc particle,” indicating that the verb refers to the present. This element disappears in the secondary endings.
- The remaining endings of the uttamapuruṣaḥ end in an element -ḥ, which also disappears in the secondary endings.
- The plural ending of the prathamapuruṣaḥ has two variants, -ánti and -áti, which derive from two different grades of an earlier ending -ént-i and -n̥t-i.
- Most of these endings are well-attested in the other Indo-European languages:
- -ti एक॰ प्र॰: Lat. -t, Hittite -zi, Greek -τι in ἐστί.
- -si एक॰ मध्य॰: Lat. -s, Hittite -si.
- -mi एक॰ उत्त॰: Hittite -mi, Greek -μι in verbs like δείκνῡμι.
- -ánti बहु॰ प्र॰: Hittite -anzi, Greek (Doric) -οντι, Latin -unt.
|prathamapuruṣaḥ||-t||-tā́m||-án / -úḥ|
Notes on these endings:
- The “hic-et-nunc particle” does not appear on the endings of the singular, or of the plural prathamapuruṣaḥ.
- The singular ending of the uttamapuruṣaḥ is -am, which derives from -m (the primary ending without -i); it is taught as -am because it takes this form after consonants, whereas after vowels, it is -m.
- Similarly, the element -ḥ of the dual and plural uttamapuruṣaḥ endings does not appear, as it does in the primary endings.
- The plural ending of the prathamapuruṣaḥ has two variants. The first, -án, is used in the imperfect laṅ, and it is simply -ánt, after the application of the rule that no Sanskrit word can end in more than one consonant.
- The second plural ending of the prathamapuruṣaḥ, -úḥ, is used in the aorist luṅ and optative liṅ. Historically, it derives from -r̥s.
|prathamapuruṣaḥ||-tu||-tā́m||-ántu / -átu|
|madhyamapuruṣaḥ||-dhí / -hí / ∅||-tám||-tá|
The imperative endings are most similar to the secondary endings, with the following differences:
- The primary endings that terminate in an element -i—the three singular endings as well as the plural of the prathamapuruṣaḥ—take a similar element in the imperative, namely, -u.
- The singular of the madhyamapuruṣaḥ, however, always takes a special ending in the imperative:
- after the stem-final vowel -a- (the so-called “thematic vowel”), the ending is ∅.
- otherwise, the ending is -dhí (after consonants) and -hí after vowels (compare Greek -θί)
- There is no imperative of the uttamapuruṣaḥ. The subjunctive forms are used instead.
This section will explain the ātmanēpadám endings for the various lakārāḥ, starting, once again, from the endings that Pāṇini teaches as basic:
These are the basic ātmanēpadám endings, which can also be referred to using Pāṇini’s abbreviation taṄ. In contrast to the parasmaipadám endings, Pāṇini teaches the secondary endings as basic, and has rules (e.g., Aṣṭādhyāyī 3.4.79) which change the secondary endings to primary endings in the lakārāḥ that require them. For the sake of convenience, here are the primary endings of the ātmanēpadám, that is, those which are used in the present indicative laṭ:
|prathamapuruṣaḥ||-tḗ||-ā́tē||-ántē / -átē|
Notes on these endings:
- The element -i, which characterizes a certain set of primary endings in the parasmaipadám, also characterizes a set of primary endings in the ātmanēpadám. In the ātmanēpadám, however, this element almost always combines with a final vowel -a to yield -ē. The endings with this element can thus be analyzed as follows:
- -tá + i → -tḗ एक॰ प्र॰, cp. Greek -εται
- -sá + i → -sḗ एक॰ मध्य॰
- -á + i → -ḗ एक॰ उत्त॰
- -ánta + i → -ántē बहु॰ प्र॰, cp. Greek -ονται, Hittite -anta
- The terminal element -ē appears throughout the primary endings of the ātmanēpadám, even when it cannot be analyzed as the result of combining a final -a with a particle -i. Several of the corresponding secondary endings do not end in a -a but in -am or -ām.
- Several elements are recognizably the same as the parasmaipadám endings:
- -t- as a marker of the एक॰ प्र॰ (here followed by a vowel -a)
- -s- as a marker of the एक॰ मध्य॰ (here followed by a vowel -a)
- -nt- as a marker of the बहु॰ प्र॰ (here followed by a vowel -a)
- -va- as a marker of the द्वि॰ उत्त॰ (here followed by an element -hi / -hē)
- -ma- as a marker of the बहु॰ उत्त॰ (here followed by an element -hi / -hē)
|prathamapuruṣaḥ||-tá||-ā́tām||-ánta / -áta/ -rán|
|uttamapuruṣaḥ||-í / -á||-váhi||-máhi|
The similarities of these endings to those of the parasmaipadám have been discussed above. When some relationship is evident between the forms, the principal difference is that the secondary endings of the ātmanēpadám, in contrast to those of the parasmaipadám, are followed by an element -a (extended to -ē in the primary endings) or -hi (extended, probably analogically, to -hē in the primary endings). A number of other points:
As noted above, the optative liṅ always receives the “secondary” endings, like the aorist and imperfect. The optative, however, always includes an element between the verbal stem and the secondary endings which serves as the marker of this mood.
There are two related forms of this marker, yā and ī. The former is used only in athematic verbs, that is, verbs belonging to classes 2, 3, 5, 7, 8 and 9, and then only in the so-called strong forms, i.e., the singular forms of the parasmaipadám. The marker ī is used in all other contexts, including in all of the thematic conjugations, where it combines with the stem-final vowel to form the ending ē.
The marker of the optative in Indo-European was -ih₁- in the zero grade and -yeh₁- in the full grade.
The past tenses—that is, the imperfect laṅ and the aorist luṅ—employ an augment that appears prior to the verbal form, although after any verbal prefixes. This augment, which Pāṇini teaches as aṬ, is generally a short vowel a:
- ákarōt “he did”
- ákārṣīt “id.”
However, in the case of verbal roots beginning with a vowel, the combination of augment and initial vowel always results in vŕ̥ddhiḥ of the initial vowel:
- aícchat “he desired” (iṣ)
- aít “he went” (i)
- ā́rcchat “he went” (r̥ch)
In a small number of other cases, the augment is lengthened before an initial consonant of a verb:
- ā́var “he blocked”
The lengthening of the augment before a consonant is an effect of a root-initial laryngeal (in the above example, the proto-form is é-h₂u̯er-t.
The present system refers to a family of verbal forms that have in common the fact that they are formed from the same verbal stem, which is conventionally called the present stem. (Not all of these forms have a reference to the present time, however.) The verbal forms that belong to the present system are:
- the present indicative, or laṭ;
- the imperfect indicative, or laṅ;
- the imperative, or lōṭ; and
- the optative, or liṭ.
The present stem is formed in different ways from different verbal roots, and hence we talk about ten classes of verbs. A verb belongs to a class solely by virtue of how its present stem is formed, which is to say, which present-stem-forming suffix vikaraṇaḥ is added to the verbal root before the endings of the mood-tense complexes lakārāḥ that belong to the present system. In part, a verb’s belonging to one or the other class is determined by phonological considerations; in part, it is also determined by semantic considerations. The reason is that some of the present-stem-forming suffixes vikaraṇāḥ had particular meanings—generally, nasal-infix presents are more “transitive,” and -ya- presents are more “stative” or “reflexive”—but most of those semantic nuances have been lost. A verb’s membership in one (or more) of the ten classes must be memorized.
The ten classes are distinguished, as noted above, by the present-stem-forming suffix, or vikaraṇaḥ, that is used to form the present stem. The following suffixes are used:
- ŚaP (Aṣṭādhyāyī 3.1.68): Unaccented -a- which causes the root to be accented and full-grade (guṇáḥ). The list of verbs of this first class is called bhvādigaṇaḥ.
- No suffix (technically: luk-elision of the formant ŚaP, cf. Aṣṭādhyāyī 2.4.72). The root takes the full grade (guṇa) in the strong form of the stem, and the zero grade in the weak form of the stem. The list of verbs of this second class is called adādigaṇaḥ.
- No suffix (technically: ślu-elision of the formant ŚaP, cf. Aṣṭādhyāyī 2.4.75). Reduplication abhyāsaḥ of the verbal root, which takes the full-grade (or guṇa) form in the strong forms of the stem, and the zero-grade form in the weak forms of the stem. The list of verbs of this third class is called hvādigaṇaḥ.
- ŚyaN: An unaccented suffix -ya-. The root takes its weakest (zero) grade before this suffix, but it is accented. The list of verbs of this fourth class is called divādigaṇaḥ.
- Śnu (Aṣṭādhyāyī 3.1.73): The suffix is -nō- in the strong form of the stem, and -nu- in the weak form of the stem. The list of verbs of this fifth class is called svādigaṇaḥ.
- Śa (Aṣṭādhyāyī 3.1.77): The suffix is an accented -á-, and the root stays in the zero grade. The list of verbs of this sixth class is called tudādigaṇaḥ.
- Śnam (Aṣṭādhyāyī 3.1.78): The vikaraṇaḥ is actually an infixed -na- before the final consonant in the strong forms, and -n- in the weak forms. The list of verbs of this seventh class is called rudhādigaṇaḥ.
- u (Aṣṭādhyāyī 3.1.79): The suffix is -ō- in the strong form of the stem, and -u- in the weak form. The list of verbs of this eighth class is called tanādigaṇaḥ.
- Śnā (Aṣṭādhyāyī 3.1.81): A suffix -nā- in the strong form of the stem, and -nī- in the weak form of the stem. The list of verbs of this ninth class is called kryādigaṇaḥ.
- ṆiC (Aṣṭādhyāyī 3.1.25): The root is first suffixed with ṆiC, which induces Ṇ-vŕ̥ddhiḥ in the root syllable; then the suffixed root takes the vikaraṇaḥ ŚaP, which causes the preceding suffix to undergo guṇáḥ. Hence the entire suffix appears as -áya-. The list of verbs of this tenth class is called curādigaṇaḥ.
Full paradigms of all of the mood-tense complexes lakārāḥ belonging to the present system will follow for verbs belonging to each of the ten classes.
Formed with a vikaraṇaḥ -a- that induces guṇa on the root, if the root is capable of taking guṇáḥ. (Recall that “superheavy” roots, which contain either a long vowel followed by a consonant, like jīv, or any vowel followed by two consonants, like nind, cannot take guṇáḥ.) The root is accented.
The first class is the largest. Here are some of the most common verbs:
- bhū “become” (parasmaipadi)
- smr̥ “remember” (parasmaipadi)
- nad “hum” (parasmaipadi)
- nind “blame” (parasmaipadi)
- vraj “wander” (parasmaipadi)
- tap “burn” (parasmaipadi)
- cam “sip” (parasmaipadi)
- nam “bow” (parasmaipadi)
- ji “win” (parasmaipadi)
- dah “burn” (parasmaipadi)
- dhāv “run” (parasmaipadi)
- sr̥ “spread” (parasmaipadi)
- tr̥ “cross” (parasmaipadi)
- vad “speak” (parasmaipadi)
- vas “dwell” (parasmaipadi)
- pat “fall” (parasmaipadi)
- īkṣ “fall” (ātmanēpadi)
- īh “desire” (ātmanēpadi)
- dyut “shine” (ātmanēpadi)
- syand “sprinkle” (ātmanēpadi)
- ram “enjoy” (ātmanēpadi)
- tvar “hurry” (ātmanēpadi)
- nī “lead” (ubhayapadi)
- sah “bear” (ātmanēpadi)
- khan “dig” (ubhayapadi)
- hr̥ “take” (ubhayapadi)
- yaj “worship” (ubhayapadi)
- vap “sow” (ubhayapadi)
- vah “carry” (ubhayapadi)
- śri “betake oneself” (ubhayapadi)
A number of roots form their present stem with a suffix cha. They are considered to belong to the first class, if the root syllable is accented, or the sixth class, if the suffix is accented. The root always takes the zero grade before this suffix. See Aṣṭādhyāyī 7.3.77.
- gam “go” (parasmaipadi) → gáccha-
- yam “stop” (parasmaipadi) → yáccha-
- r̥ “go” (parasmaipadi) → ŕ̥ccha- (cf. ἔρχομαι)
- prach “go” (parasmaipadi) → pr̥cchá- (cf. poscō)
- iṣ “want” (parasmaipadi) → icchá-
This suffix was originally a present stem forming suffix (-sḱe-/-sḱo-), and is attested in several other Indo-European languages. It probably had an “inchoative” sense, indicating that the agent is just beginning the action: hence gʷm-sḱé-ti “he sets out.”
A very small number of verbal roots in this class form their present stem with an irregular kind of reduplication:
- sthā “stand” (parasmaipadi) → tíṣṭha- (cf. ἵστημι)
- pā “drink” (parasmaipadi) → píbati-
- ghrā “sniff” (parasmaipadi) → jíghra-
The following roots, historically ending in a laryngeal, form the present stem by lengthening the root vowel:
- kram “stride” (parasmaipadi) → krā́mati
The first class is exemplified with bhū “become” in the parasmaipadám and ruc “shine” in the ātmanēpadám.
This is the root present. There is no present-stem forming suffix: the default suffix, ŚaP (Aṣṭādhyāyī 3.1.68), is deleted by Aṣṭādhyāyī 2.4.72. There is a distinction in vowel gradation for roots of this class: a strong stem, which generally features a full-grade root (or guṇáḥ), and a weak stem, which generally geatures a zero-grade root. The strong stem is used before the singular endings of the parasmaipadám only, i.e., those endings that Pāṇini teaches with the anubandha p (tiP, siP, and miP), and that, too, only in the present indicative laṭ and imperfect indicative laṅ, as well as certain forms of the imperative lōṭ, namely, all forms of the first person (which are taken from the subjunctive) as well as the third person singular of the parasmaipadám.
If the verb is accented (see above), then the accent is on the root when the strong stem is used, i.e., in the singular forms of the parasmaipadám; the accent is on the ending in the other forms.
Here is a list of relatively common second class verbs, which I provide with their third persons singular and plural::
- ad “eat” (parasmaipadi), átti, adánti
- yā “go” (parasmaipadi), yā́ti, yā́nti
- han “kill” (parasmaipadi), hánti, ghnánti
- vaś “desire” (parasmaipadi), váṣṭi, uśánti
- ās “sit” (ātmanēpadi), ā́stē, ā́satē
- vid “know” (parasmaipadi), vḗtti, vidánti
- i “go” (parasmaipadi), ḗti yánti
- as “be” (parasmaipadi), ásti, sánti
- vac “speak” (parasmaipadi), vákti (no third person plural)
- śās “command” (parasmaipadi), śā́sti, śāsáti
- īś “rule” (ātmanēpadi), īśḗ, īśátē
- brū “speak” (ubhayapadi), brávīti or brūtḗ, bruvánti or bruvátē
To exemplify the paradigm, we use i “go” in the parasmaipadám and ās “sit” in the ātmanēpadám.
All of the roots beginning with a long vowel in this class, including ās, maintain their accent on the root throughout.
The verb i has the strong stem ē and the weak stem i. The strong stem only occurs before endings that begin with a consonant. The weak stem, however, occurs before both consonant-initial and vowel-initial endings. In the parasmaipadám, the only vowel-initial ending is -anti, and there is a special rule (Aṣṭādhyāyī 6.4.81) that tells us that the vowel of the stem, i, is replaced by the corresponding semivowel, y, before this ending, and hence we have third-person plural form yánti.
In the ātmanēpadám, however, there are many endings that begin with vowels. The ātmanēpadám forms of this verb only occur after a verbal prefix, and in this context, by a special rule, the stem i is not replaced by the corresponding semivowel y, but by the sequence iy (which keeps the root as a distinct syllable), according to Aṣṭādhyāyī 6.4.77. Hence the forms in the ātmanēpadám would be iyē, iyāthē, iyātē, iyatē.
This is the “reduplicating” present, so called because it is formed with a reduplicated syllable, or abhyāsaḥ, which is always a modified version of the root. The rules for the formation of the reduplicated syllable are almost exactly the same as for the reduplication of the perfect. Specifically:
- No aspiration is permitted in the reduplicated syllable. Aspirate stops become their corresponding unaspirated forms, and the voiced aspirate h becomes j.
- The reduplicated syllable itself is always light. This means that vowel of the root, if it is long, is shortened; if it is a diphthong (ē, ō, ai and au), it becomes the corresponding simple vowel (i or u). Generally the vowel ā in the root syllable becomes a in the reduplicated syllable (e.g., dadā́ti from dā), but in a few roots, it becomes i (e.g., mimā́ti from mā). This rule also means that any root-final consonants are lost, since they would make the reduplicated syllable heavy.
- No complex consonants are permitted in the reduplicated syllable. Generally, if the onset of the root is C₁C₂, the onset of the reduplicated syllable is C₁ (e.g., jihrḗti from hrī). If, however, root begins with a sibilant-stop combination, then the stop, rather than the sibilant, appears in the reduplicated syllable.
- Velar kaṇṭhyaḥ consonants are replaced by their palatal tālavyaḥ counterparts. This is because the vowel of the reduplicated syllable in Indo-European was high (either e or i), which caused the palatalization of a preceding velar stop in Indo-Iranian.
The examples are hu “pour out” for parasmaipadám and mā “measure” for ātmanēpadám.
The fourth class (“[the list of verbal roots] which begins with div”) forms its present stem with the present-stem-forming suffix ŚyaN (Aṣṭādhyāyī 3.1.69). This has the following properties:
- The form of the present-stem-forming suffix is an unaccented -ya-.
- The suffix requires that the root be in the zero grade. Hence:
- No guṇáḥ or vŕ̥ddhiḥ applies to the verbal root.
- If the verbal root is taught in a full-grade form (e.g., tam) then it will go into the corresponding zero-grade form (e.g., tām).
Here is a list of relatively common fourth class verbs:
- naś “be lost” (parasmaipadi)
- kṣudh “be hungry” (parasmaipadi)
- tr̥p “be satisfied” (parasmaipadi)
- nr̥t “dance” (parasmaipadi)
- siv “sew” (parasmaipadi)
- div “gamble” (parasmaipadi)
- man “think” (ātmanēpadi)
- nah “tie up, bind” (parasmaipadi)
- puṣ “nourish” (parasmaipadi) or “be nourished” (ātmanēpadi)
The root śram “become tired” lengthens its vowel when the present stem forming suffix is added, hence śrāmyáti.
The forms are exemplified with puṣ “nourish” in the parasmaipadám and man “think” in the ātmanēpadám.
This class adds the stem-forming suffix vikaraṇaḥ -nṓ-/-nu- to the verbal root. It is essentially identical to the eighth class, which adds the suffix -ṓ-/-u-. The strong forms are built with the full-grade suffix -nṓ-, and the weak forms, with the zero-grade suffix -nu-.
The accent is on the present-stem forming suffix in the strong forms and on the ending in the weak forms.
Roots of this class behave somewhat differently in their internal sandhi depending on whether they end in a vowel or a consonant. For roots that end in a vowel, the final u of the stem becomes v before endings that begin with a vowel (e.g., cinvánti). This final u also be dropped before the endings of the first person dual and plural (e.g., cinmáḥ or cinumáḥ).
For roots that end in a consonant, however, the consonant cluster -C-n- prevents the final u of the stem from becoming v (because that would result in an awkward consonant cluster -Cnv-). Instead, we have uvaṄ-sandhi, where the final u is replaced by uv before endings that begin with a vowel (e.g., āpnuvánti). The u of the stem is retained before the endings of the first person dual and plural (e.g., āpnumáḥ).
Here is a list of relatively common fifth class verbs:
- śak “be capable” (parasmaipadi)
- śru “hear” (parasmaipadi); though this forms its stem as if from śr̥
- āp “obtain” (parasmaipadi)
- aś “pervade” (ātmanēpadi)
- su “press” (ubhayapadi)
- ci “collect” (ubhayapadi)
- str̥ “cover” (ubhayapadi)
- vr̥ “choose” (ubhayapadi)
The paradigms are given using āp “obtain” for the parasmaipadám and ci “gather” for the ātmanēpadám.
The sixth class (“[the list of verbal roots] which begins with tud”) forms its present stem with the present-stem-forming suffix Śa. This has the following properties:
- The form of the present-stem-forming suffix is an unaccented -á-.
- The suffix requires that the root be in the zero grade. Hence:
- No guṇáḥ or vŕ̥ddhiḥ applies to the verbal root.
- If the verbal root is taught in a full-grade form (e.g., prach) then it will go into the corresponding zero-grade form (e.g., pr̥ch).
- Typically these roots end in a consonant, but for roots that end in the vowel r̥̄, the stem ends in -irá-.
Here is a list of relatively common verbs that belong to the sixth class:
- viś “enter” (parasmaipadi)
- diś “point out” (parasmaipadi)
- likh “write” (parasmaipadi)
- spr̥ś “touch” (parasmaipadi)
- kr̥̄ “scatter” (parasmaipadi)
- kṣip “throw” (parasmaipadi)
- vidh “worship” (parasmaipadi)
- lajj “be embarrassed” (ātmanēpadi)
- sphur “tremble, flash, throb, pulse” (parasmaipadi)
- viś “enter” (parasmaipadi)
- sr̥j “emit” (parasmaipadi)
- kr̥ṣ “plow” (ubhayapadi)
The following roots form their present stem by inserting a nasal between the vowel and the final consonant Aṣṭādhyāyī 7.1.59:
- muc “release” (ubhayapadi) → muñcá-
- lup “cut” (ubhayapadi) → lumpá-
- vid “find” (ubhayapadi) → vindá-
- lip “smear” (ubhayapadi) → limpá-
- sic “sprinkle” (ubhayapadi) → siñcá-
- kr̥t “cut” (parasmaipadi) → kr̥ntá-
- khid “oppress” (parasmaipadi) → khindá-
- piś “form” (parasmaipadi) → piṁśá-
The verbs listed as prach and iṣ form their present stems as pr̥ccháti and iccháti respectively. See the note on the verbs yácchati and gácchati under first-class verbs above.
The forms are exemplified with sr̥j “emit” in the parasmaipadám and lajj “be embarrassed” in the ātmanēpadám.
This class is sometimes called the ‘nasal infix’ class, because its stem-forming affix vikaraṇaḥ is a nasal, which is not added onto the verbal root, but infixed between the root’s vowel and its final consonant.
In the strong forms, the infix full-grade form -ná-. In the weak forms, the infix takes the zero-grade form -n-. Note that, because the root ends in a consonant, there will be internal sandhi between the root and endings beginning with a consonant. Note, also, that the -n- of the infix is subject to retroflexion by nati in its strong forms.
The accent is on the infix in the strong forms and on the ending in the weak forms.
Some of the most common roots in this class are:
- rudh “block” (ubhayapadí) → ruṇádh-, rundh-
- yuj “join” (ubhayapadí) → yunáj-, yuñj-
- bhuj “enjoy” (ātmanēpadí) → bhunáj-, bhuñj-
- his “strike” (parasmaipadí) → hiṁs-, hinás-
- śiś “distinguish” (parasmaipadí) → śináṣ-, śiṁṣ-
- bhañj “break” (parasmaipadí) → bhanáj-, bhañj-
- añj “annoint” (parasmaipadí) → anáj-, añj-
The following paradigms use the verb rudh “block” for parasmaipadám and bhuj “enjoy” for ātmanēpadám.
The stem-forming affix vikaraṇaḥ in his class is the vowel u, which takes the full-grade or guṇáḥ form -ṓ- in the strong forms, and the zero-grade form -u- in the weak forms.
The accent is on the suffix in the strong forms and on the verbal ending in the weak forms.
This class of verbs is exactly parallel to the fifth class, which builds its strong forms with the affix -nṓ- and its weak forms with the affix -nu-.
The primary verb of the eighth class, tan “stretch,” historically was a nu-present. The root takes the zero-grade form before the suffix, so the inherited form would have been tn-nu- or tn-néu-. But because an inherited n between consonants develops into a in Sanskrit, the zero-grade form of the root ends up looking like ta, and for convenience, the Sanskrit grammarians considered the n of the suffix to belong to the root.
The forms are examplified with tan “stretch” in both the parasmaipadám and the ātmanēpadám.
In addition, the full paradigm of the verb kr̥ “do” is given here, which presents a number of particularities: in the strong forms, the stem takes the form karṓ-, and in the weak forms, it appears as kuru- or kur-. The latter is the reflex of the zero-grade form kr̥ before the stem-forming suffix -u-. In addition, the verb kr̥ drops the stem-final u before the endings of the first person dual and plural (like fifth-class verbs such as ci, su, etc.).
The present-stem forming suffix (vikaraṇaḥ) for this class is -nā- in the strong forms and -nī- (before consonants) or -n- (before vowels) in the weak forms. The accent is on the suffix in the strong forms and on the verbal endings in the weak forms.
Historically, this class is a subset of the seventh class, which forms the stem with an infix -na- or -n-. The roots of the ninth class happened to end in a laryngeal in Indo-European, and the combinations of the infix and the final consonant, -ne-H- and -n-H-, resulted in -nā- and -nī- in Sanskrit.
The parasmaipadám forms are exemplified with bandh “bind,” and the ātmanēpadám forms with jñā “know.”
The tenth class mostly includes deverbal verbs, such as causative verbs. But it includes a number of roots that have no corresponding simple verb. These roots are listed in the curādigaṇaḥ.
They are formed with the present stem forming suffix -áya-, which induces Ṇ-vŕ̥ddhiḥ on the root syllable. (In Pāṇini’s system, these stems are formed by adding the suffix ṆiC to the root, and then adding the suffix ŚaP, just as in the first class of verbs: ṆiC strengthens the root syllable, and ŚaP strengthens the root-extension ṆiC to guṇáḥ, hence the composite form -áy-a-.)
Some roots in this class are:
- gaṇ “count” → gaṇáya-
- kath “say” → katháya-
- cint “think” → cintáya-
The parasmaipadám forms are exemplified with cur “steal.” There are no non-causative ātmanēpadám verbs.
All of the finite verb forms listed above are used “in reference to the agent of the verbal action” kartári prayōgáḥ. Sanskrit, however, allows for any verb of any of the ten classes to be used either with reference to the patient of the verbal action karmáṇi prayōgáḥ, if the verb is transitive, or with reference to the verbal action itself bhāvḗ prayōgáḥ. (For more on these three constructions, see the agentive, patientive, and impersonal constructions below.)
In the present system, a passive/impersonal construction is available with a special present stem forming suffix vikaraṇaḥ that Pāṇini calls yaK. The effect of the final diacritical letter anubandhaḥ K is to put the root into the zero grade (see vowel gradation above). Once the passive stem (i.e., the stem used for both passive and impersonal expressions) has been formed by adding the suffix yaK to the verbal root in the zero grade, the ātmanēpadám endings are added, since these endings are obligatory whenever the verbal form expresses the patient or the verbal action Aṣṭādhyāyī 1.3.13.
The formation of the passive stem is generally straightforward: add the suffix -ya- to the unstrengthened verbal root:
- gam + yaK [ + tē ] → gamyátē “goes” [impersonal]
- nī + yaK [ + tē ] → nīyátē “is led”
- pac + yaK [ + tē ] → pacyátē “is cooked”
- paṭh + yaK [ + tē ] → paṭhyátē “is recited”
However the root sometimes undergoes changes. One type of change relates to the combination of root-final vowel with the -ya- of the present stem forming suffix. Before the suffix, the vowels i and u are lengthened, and the vowel r̥ is changed to ri; a long r̥̄ is changed to īr generally, but to ūr after labial consonants:
- śru + yaK [ + tē ] → śrūyátē “is heard”
- stu + yaK [ + tē ] → stūyátē “is praised”
- ci + yaK [ + tē ] → cīyátē “is piled”
- kr̥ + yaK [ + tē ] → kriyátē “is done”
- hr̥ + yaK [ + tē ] → hriyátē “is taken”
- tr̥̄ + yaK [ + tē ] → tīryátē “is crossed”
- pr̥̄ + yaK [ + tē ] → pūryátē “is filled”
An exception to this rule is śī “lie,” which forms the impersonal stem śayya- Aṣṭādhyāyī 7.4.22.
Verbal roots ending in r̥ that begin with conjunct consonants, however, do not turn the r̥ into ri, but into ar, probably to avoid an awkward combination of consonants such as smr-, Aṣṭādhyāyī 7.4.29
- smr̥ + yaK [ + tē ] → smaryatē “it is remembered”
The following verbs take saṁprasā́raṇam in the passive stem Aṣṭādhyāyī 6.1.15:
- vac + yaK [ + tē ] → ucyátē “is said”
- svap + yaK [ + tē ] → supyátē “sleeps” [impersonal]
- vaś + yaK [ + tē ] → uśyátē “is wished”
- yaj + yaK [ + tē ] → ijyátē “is sacrificed”
- vap + yaK [ + tē ] → upyátē “is sown”
- vah + yaK [ + tē ] → uhyátē “is carried”
- vas + yaK [ + tē ] → uṣyátē “stays” [impersonal
- vē + yaK [ + tē ] → ūyátē “is weaved”
- vyē + yaK [ + tē ] → vīyátē “is covered”
- hvē + yaK [ + tē ] → hūyátē “is called”
- vad + yaK [ + tē ] → udyátē “is spoken”
- śvi + yaK [ + tē ] → śūyátē “expands” [impersonal]
- grah + yaK [ + tē ] → gr̥hyatē “is taken”
- jyā + yaK [ + tē ] → jīyátē “fails” [impersonal]
- vyadh + yaK [ + tē ] → vidhyátē “is pierced”
- vyac + yaK [ + tē ] → vicyátē “is surrounded”
- vraśc + yaK [ + tē ] → vr̥ścyátē “is cut”
- prach + yaK [ + tē ] → pr̥cchyátē “is asked”
- bhrajj + yaK [ + tē ] → bhr̥jjyátē “is fried”
- syam + yaK [ + tē ] → simyátē “cries” [impersonal]
The following verbal roots ending a long vowel (and historically in a laryngeal) form their zero-grade forms with ī Aṣṭādhyāyī 6.4.66:
- dā + yaK [ + tē ] → dīyátē “is given”
- dhā + yaK [ + tē ] → dhīyátē “is placed”
- mā + yaK [ + tē ] → mīyátē “is measured”
- sthā + yaK [ + tē ] → sthīyátē “stands” [impersonal]
- hā + yaK [ + tē ] → hīyátē “is abandoned”
- gai + yaK [ + tē ] → gīyátē “is sung”
- pā + yaK [ + tē ] → pīyátē “is drunk”
- sō + yaK [ + tē ] → sīyátē “is finished”
The verbal root śās forms the passive stem śiṣyá- Aṣṭādhyāyī 6.4.34.
Roots that have a penultimate nasal typically lose it in the passive:
- rañj + yaK [ + tē ] → rajyátē “is attached” [impersonal]
- sraṁs + yaK [ + tē ] → srasyátē “falls” [impersonal]
- bandh + yaK [ + tē ] → badhyátē “is bound”
Among the roots ending in a nasal consonant, most do not change, but a few can optionally lose the final nasal and lengthen the preceding vowel Aṣṭādhyāyī 6.4.43. All of these historically ended in a laryngeal consonant.
- jan + yaK [ + tē ] → janyátē or jāyátē “is born”
- tan + yaK [ + tē ] → tanyátē or tāyátē “is stretched”
- san + yaK [ + tē ] → sanyátē or sāyátē “is attained”
- khan + yaK [ + tē ] → khanyátē or khāyátē “is dug”
The verbal adjective of the present stem is formed by the addition of two suffixes: one corresponding to parasmaipadám forms, which Pāṇini calls ŚatR̥, and one corresponding to ātmanēpadám forms, which Pāṇini calls ŚānaC. The form that these suffixes take depends on the class of the verb, according to the following rule:
- verbal roots belonging to the thematic classes—that is, those whose present-stem forming suffix ends in the thematic vowel a, including the first bhvādigaṇaḥ, the fourth divādigaṇaḥ, the sixth tudādigaṇaḥ, and the tenth curādigaṇaḥ—affix the suffix -at in the parasmaipadám and the suffix -māna- in the ātmanēpadám;
- verbal roots belonging to all other classes, or athematic roots, affix the suffix -at- in the parasmaipadám and the suffix -āna- in the ātmanēpadám.
The reason for the different suffixes of the ātmanēpadám participle in the thematic and non-thematic verb classes is the different development of the inherited suffix, -mh₁no-, after a vowel (as was always the case in the thematic verb classes) and after a consonant (as was often the case in the athematic verb classes). After a vowel, the suffix probably developed into -mīna-, for which there is some inscriptional evidence. After a consonant, it developed into -āna-. Analogy from the postconsonantal version of the suffix probably reshaped the postvocalic version.
The declension of the present participle of the parasmaipadám is covered above: in the masculine and nominative, it is inflected as a stem ending in -ant in the strong cases and -at in the weak cases. The feminine is formed with the suffix ṄīP. As noted in the section on the participle’s declension, there is an important exception to general principle that the strong stem is formed with -ant- and the weak stem with -at: in the participle of verbs of the third class, which have a reduplicated stem, the form of the stem is always -at-, even in the strong cases (although it is optionally -ant- in the nominative-accusative-vocative of the neuter).
Whether the feminine suffix is added onto the stem in -ant- or -at-, and also whether the neuter nominative-accusative dual is formed by adding the suffix -ī onto the stem in -ant- or -at-, depends on the verb class. The general principle is as follows:
- verbs of the first, fourth, and tenth classes use the stem ending in -ant-;
- verbs of the second, third, fifth, seventh, eighth, and ninth classes use the stem ending in -at-;
- (except verbs of the second class ending in -ā, which may use either the stem ending in -at- or the stem ending in -ant-)
- verbs of the sixth class may use either stem.
|2. adādiḥ||yā́n||yā́ntī / yātī́|
|6. tudādiḥ||tudán||tudántī / tudatī́|
The perfect system refers to a set of verbal forms that are derived with the tense-mood marker liṭ. Unlike the present system, there is no distinction between classes in the perfect: all verbs are formed according to the same rules. Not all verbs, however, can be inflected in the perfect: verbs beginning with a vowel (besides a or ā) that is “superheavy” (i.e., a long vowel followed by a consonant, or a short vowel followed by two consonants), such as īḍ, und, ēdh, and indh, cannot be inflected in this tense.
The forms of the perfect are generally used with reference to past time; Pāṇini prescribes the perfect tense for events in the past that one has not personally witnessed (parōkṣē liṭ in Aṣṭādhyāyī 3.2.115), and it is commonly used in discussing the remote past, or as a general narrative tense..
As with all other verbal forms, the perfect can be thought of as a combination of a stem with a given ending. As with athematic verbs in the present system, the verbs in the perfect generally distinguish between a strong and weak form of the stem, the former displaying a full-grade form of the root, and the latter a zero-grade form.
The characteristic of the perfect stem is not a suffix, but rather a reduplication abhyāsaḥ of the root. In addition, the perfect takes a distinctive set of endings in both the parasmaipadám and the ātmanēpadám.
The basic principles of reduplication in the perfect are similar to that of third-class verbs in the present tense (see above): the reduplicated syllable is the verbal root, only subject to a greater number of phonological constraints, and subject to certain modifications:
- No coda consonants. If the root ends in a consonant, it is omitted from the reduplicated syllable. (This is due to a broader constraint on heavy syllables.)
- car “go” → ca-
- pac “go” → pa-
- labh “touch” → la-
- No long vowels. If the vowel of the root is long, it is shortened in the reduplicated syllable. The short equivalents of ē and ō are i and respectively Aṣṭādhyāyī 1.1.48. (This, too, is due to the aforementioned constraint on heavy syllables.) Note that the vowel of the reduplicated syllable of bhū is ba- rather than bu- Aṣṭādhyāyī 7.4.73.
- nī “lead” → ni-
- sēv “serve” → si-
- dā “give” → da-
- bhū “become” → ba-
- No vocalic r. r̥ becomes a.
- vr̥ “open” → va-
- mr̥ “die” → ma-
- Saṁprasā́raṇam if possible. Roots that contain a semivowel followed by the vowel a and which are subject to saṁprasā́raṇam form the reduplicated syllable with the corresponding vowel (Aṣṭādhyāyī 6.1.17). Roots that would form a saṁprasā́raṇam stem with r̥ use a in the reduplicated syllable because of the above constraint on r̥.
- vac “speak” → u-
- vad “speak” → u-
- vap “sow” → u-
- vaś “wish” → u-
- vas “dwell” → u-
- svap “sleep” → su-
- yaj “sacrifice” → i-
- vyadh “strike” → vi-
- No velars. Velars turn into the corresponding palatal sounds. See the above examples, as well as:
- kāś “shine” → ca-
- kr̥ “play” → ca-
- gāh “dive into” → ja-
- No aspirates. Aspirates are replaced by the corresponding non-aspirate sounds, and the phoneme h is replaced by j.
- bhū “become” → ba-
- bhuj “enjoy” → bu-
- dhā “place” → da-
- hā “leave” → ja-
- No conjunct consonants. If the root begins with a conjunct consonant, it must be reduced to a single consonant. Generally the least sonorous consonant remains, where stops are the least sonorous of all, followed by sibilants, followed by nasals, and finally followed by semivowels, the least sonorous consonants. Here are examples of the possible combinations:
- stop + sibilant:
- spr̥ś “touch” → pa-
- skhal “trip” → ca-
- kṣip “throw” → ci-
- sibilant + nasal:
- smr̥ “remember” → sa-
- stop + semivowel:
- kruś “shout” → cu-
- tras “be afraid” → ta-
Roots beginning with a vowel follow the same rules, although some sandhi changes occurs: in the weak form of the stem, the vowel of the reduplicated syllable with combine with the vowel of the root, and in the strong form, if the root takes guṇáḥ or vŕ̥ddhiḥ, the glide y or v will be inserted between the reduplicated syllable and the root.
The endings of the perfect are different from those of the present system. In the tables below, they have been presented with the augment i, which very often appears between a stem ending in a consonant, and an ending beginning with a consonant.
In the following, the parasmaipadám endings are given with Pāṇini’s diacritic letters (anubandhaḥ):
The ātmanēpadám endings are generally the same as the primary secondary endings that are used, for example, in the present tense:
Note, however, the distinct endings ē in the third person singular (in contrast to tē) and irē in the third person plural.
The augment iṬ very commonly occurs between the perfect stem and endings that begin with a consonant. The presence or absence of iṬ depends very much on the root in question, but the endings can be put into three groups:
|always with iṬ||rē||3pl.ātmanē.|
|mostly with iṬ||va||1du.parasmai.|
|less often with iṬ||tha||2sg.parasmai.|
Note that the augment is generally optional before the second person singular parasmaipadám ending tha, hence uváktha or uvácitha.
The verb kr̥ never takes the augment iṬ in the perfect, except before the third person plural ātmanēpadám ending rē (which is, for all intents and purposes, irē).
Roots ending in a vowel like dā “give” or dhā “place” take a special ending in the first person and third person singular parasmaipadam. In those cases the vowel of the root combines with the ending to produce au (see the paradigms below).
In the perfect, as in all of the other ‘athematic’ verb conjugations, there is a pattern of accent mobility between the stem and the endings, which accompanies a pattern of vowel gradation. According to this pattern, the stem exhibits a full grade of the root syllable when it is accented, and a zero grade of the root syllable when it is unaccented. Hence the perfect distinguishes between a strong and weak form of the stem.
As usual, the singular endings of the parasmaipadám are unaccented and therefore accompany the strong form of the stem, whereas the stem is weak before all of the other endings (the dual and plural of the parasmaipadám and all of the ātmanēpadám endings).
The weak form of the root is simply the zero grade, which involves all of the transformations associated with the diacritic marker K (which is implicit in all of the endings of the perfect other than the singular parasmaipadám). The most important such transformation is saṁprasā́raṇam: for many verbs that contain a semivowel followed by the vowel a, the a is lost and the semivowel becomes a vowel. In many of those cases, internal sandhi will unite the reduplicating syllable and the root syllable, since both are formed with saṁprasā́raṇam:
For roots ending in vowels, whether the final vowel of the weak stem becomes a semivowel before endings beginning with a vowel (yaṆ) or whether it is substituted with a combination of a short vowel plus a semivowel (iyaṄ/uvaṄ) depends on the number of consonants that precede it. If the final vowel is preceded by one consonant, it is regularly converted into a semivowel; if it is preceded by more than one, it is replaced with iyaṄ or uvaṄ:
In the case of the verb bhū, however, the augment vuK is always inserted between the root and endings beginning with a vowel:
One important class of verbal roots does not employ reduplication in the weak form of the stem, but instead uses a collapsed stem. These are roots that (1) contain the vowel a followed by a single consonant, and (2) form their reduplicating syllable with the exact same consonant that is used in the root, i.e., they do not begin with a conjunct consonant, a velar, or an aspirate. Such roots form the weak stem by replacing the medial vowel a with ē, as follows:
An even smaller class of verbs with a medial a (i.e., the vowel a followed by one and only one consonant) forms the weak stem of the perfect with reduplication, but accompanied by the deletion of the vowel of the root syllable. These roots are as follows Aṣṭādhyāyī 6.4.98:
Note that han substitutes its original velar, gh, for h in the perfect forms.
Finally, roots that end in the vowel ā form the strong stem with ā (which takes special endings in the first and third person parasmaipadam) and their weak stem with i before consonants and zero before vowels:
For verbs of a certain phonological shape, the strong form of the stem itself takes different forms depending on which of the endings follows. The vowel gradation of the strong them is:
|ṆaL (1sg.parasmai.)||optionally guṇáḥ or Ṇ-vŕ̥ddhiḥ|
“Ṇ-vŕ̥ddhiḥ” means that the root will take either guṇáḥ or vŕ̥ddhiḥ depending on its phonological shape. Roots endings in a vowel, or which have the vowel a followed by one and only one consonant, take vŕ̥ddhiḥ in these forms; all other roots take guṇáḥ.
A root like bhid “split” will never take vŕ̥ddhiḥ before endings marked with Ṇ, because it has neither a final vowel nor a medial a. Hence its paradigm will be as follows:
For a verb like nī, however, vŕ̥ddhiḥ is required in the third person singular, and guṇáḥ is required in the second person singular. In the first person singular, either guṇáḥ or vŕ̥ddhiḥ may be used. Hence:
The reason for the difference in vowel gradation in the strong form of the stem has to do with Brugmann’s Law. The Proto-Indo-European vowel o generally became a in Sanskrit. However, when it occurred in an open syllable—that is, when it was not followed by a consonant in the same syllable—it developed into the long vowel ā. The full grade of the perfect stem was formed with an accented ó in Proto-Indo-European. Most often, Indo-European roots end in a single consonant. Hence this ó stood in an open syllable before endings beginning with a vowel. The endings reconstructed for the perfect in Proto-Indo-European are e in the third person singular, th₂e in the second person singular, and h₂e in the first person singular. Hence Brugmann’s Law generally converted a medial ó into a in the first and second person singular, and ā in the third person singular. Possibly the option for guṇáḥ or Ṇ-vŕ̥ddhiḥ in the first person singular arose from the fact that the laryngeal consonant with which the first person singular ending began, h₂, was already unstable at the time that Brugmann’s Law operated. Below are the reconstructions for the perfect of śru “hear”, with a dash marking the boundary between the stem and the ending, and a dot marking the boundary between syllables:
The verb bhū “become” is peculiar first of all because it reduplicates with ba-, rather than the expected bu-, and second because it does not undergo any kind of vowel gradation in the perfect, and third because a glide (vUK) always appears between the root and endings beginning with a vowel.
The verb man “think” is conjugated in the ātmanēpadám, and it forms its weak stem through a “collapse” of the reduplicating syllable and the root.
The verb dā “give” exhibits the special ending of the first and third person singular parasmaipadam for verbs ending in a long ā:
The aorist refers to a set of verbal formations that constitute a system in alternation with the present and perfect system. Originally, the aorist system was associated with perfective aspect, that is, to speak of actions as complete in themselves, and usually thus as punctual (e.g., ‘he dropped the glass’ vs. ‘he kept dropping the glass’). The significance of aspect, however, has largely disappeared from the Sanskrit verbal system, and aorist forms are used as the ‘default’ past tense Aṣṭādhyāyī 3.2.110. They contrast with the imperfect, which is used for actions which have taken place within the lifetime of the speaker but not generally in the timeframe of discourse (e.g., ‘Obama was elected in 2008’), and with the perfect, which is generally reserved for the remote past (e.g., ‘Harṣa defeated Śaśāṅka’). The association of aorist forms with perfective aspect, however, lives on in the common use of augmentless aorist forms to express prohibitions (‘don’t do that’ = mā tat kārṣīḥ; see the injunctive below).
The aorist system, like the present system, comprises a number of different stem formations. But whereas each verbal root is generally associated with one and only one present stem, a given verbal root may be associated with several different aorist stems. Moreover the phonological shape of a root is of greater importance in determining the stems that it can form in the aorist system than in the present system. The major modes of aorist stem formation are:
- the sigmatic aorists:
- the simple sigmatic aorist or s-aorist SiC;
- the iṣ-aorist;
- the siṣ-aorist;
- the sa-aorist Ksa;
- the root aorist;
- the thematic aorist aṄ;
- the reduplicated aorist CaṄ; and
- the passive aorist CiṆ.
Some of these formations are reserved for specific syntactic functions. Thus, just as the suffix yaK forms an impersonal or passive present stem, so too the suffix CiṆ forms an impersonal or passive aorist. Similarly, just as the suffix ṆiC forms a causative root that can be conjugated in the present system, so too the suffix CaṄ forms a causative aorist. Whether a particular root takes a particular aorist formation, however, is otherwise largely determined by its phonological shape.
Pāṇini groups all of the aorist forms together firstly under the abstract tense-aspect-mood marker luṅ Aṣṭādhyāyī 3.2.110, which is always replaced with the suffix Cli Aṣṭādhyāyī 3.1.43. This general suffix of the aorist is then substituted as needed to form the aorist stems discussed below. The default form is the s-aorist Aṣṭādhyāyī 3.1.44.
The endings of the aorist are the secondary endings of both the parasmaipadám and ātmanēpadám. The only major point to note in connection with the aorist endings is the third person plural. The thematic aorist conjugations (those formed with Ksa and aṄ) take -an in the parasmaipadám and -anta in the ātmanēpadám. (The root aorist of bhū, which is athematic, also takes the ending -an.) The athematic aorist conjugations, however, take the ending -uḥ in the parasmaipadám and -ata in the ātmanēpadám.
The aorist, as an indicative past tense, always features the past tense augment aṬ (= á-). The augment is added before the root, but after all of the verbal prefixes. In case the root begins with a vowel, the augmented form of the root is represented by vŕ̥ddhiḥ of that vowel.
The most common forms of the aorist feature the aorist marker s, which appears between the root and the endings. There are, however, a number of different ways to form this aorist—which is called the sigmatic or sibilant aorist because of this marker—depending partly on the phonological shape of the verbal root:
- the s-aorist,
- the iṣ-aorist,
- the siṣ-aorist, and
- the sa-aorist.
The s-aorist is the default aorist formation, and is formed by the addition of the suffix s (Pāṇini’s siC) directly to the verbal root. The default pattern of vowel gradation for this form is lengthened grade (vŕ̥ddhiḥ) in the parasmaipadám and zero grade in the ātmanēpadám. However, verbal roots that end in the vowels i, ī, u and ū take full grade (guṇáḥ) in the ātmanēpadám. Verbs that end in a long ā turn this into i or ī in the ātmanēpadám (in the parasmaipadám such verbs usually form another aorist).
The endings of the s-aorist are the regular secondary endings for athematic verbs, including -uḥ in the third person plural parasmaipadám and -ata in the third person plural ātmanēpadám. However, because these endings are added directly onto a stem characterized by s, a number of sandhi changes, and pseudo-sandhi changes, occur:
- between the verbal root and endings that consist of a single phoneme (i.e., -s of the 2sg.parasmai. and -t of the 3sg.parasmai.) an augment -ī- is inserted, so that the endings of the 2sg. and 3sg. parasmai. are effectively -īḥ and -īt respectively;
- before the ending -dhvam of the 2pl.ātmanē., the aorist marker s is lost, and in case the s would have been retroflexed, the ending becomes -ḍhvam;
- before endings that begin with t or th stop, the aorist marker -s- is lost in two circumstances:
- if the sound preceding s is a consonant; and
- if the sound preceding s is a short vowel (as in the ātmanēpadam of certain vowel-final roots.
Regarding the insertion of the augment ī: this is a relatively new feature of the s-aorist, which begins to appear in the Atharvavēda and, with some regularity, in the Brāhmaṇa texts. The older form of the s-aorist was simply (e.g.) ánaiḥ, ájaiḥ, ákāḥ, ápāk, etc. The augment ī was introduced from the iṣ-aorist (see below).
Regarding the loss of s/ṣ before -dhvam, this is a normal (although rare) sandhi phenomenon, according to which the s/ṣ is voiced due to regressive assimilation of voicing. The voiced dental sibilant (call it z if you like) disappears without any visible effect, but the voiced retroflex sibilant (call it ẓ) almost always produces retroflexion in the following dental stop, and hence the outcome -ḍhvam.
Regarding the loss of s after the r̥ of a verbal root, this is likely not a sandhi phenomenon at all, but the appearance of root aorist forms in the paradigm of the s-aorist.
The iṣ-aorist is, in origin, just the form that the s-aorist took when the verbal root takes the augment iṬ (i.e., if it is a sēṬ root). It uses the same endings as the s-aorist, but the vowel gradation is different: the augment i generally causes the root to enter the full grade, or guṇáḥ, although roots ending in a vowel (and some roots with a penultimate a) take vŕ̥ddhiḥ in the parasmaipadám, just as in the s-aorist.
Note that the second and third person singular of the parasmaipadám are not *-iṣīḥ and *-iṣīt, but -īḥ and -īt (as noted above, these forms actually originated in the iṣ-aorist, and secondarily spread to the s-aorist).
The roots that took this augment, as explained above, historically ended in a laryngeal consonant, which became i in interconsonantal position.
In a few verbs (ram ‘besport,’ yam ‘control,’ nam ‘bow,’ and those ending in -ā), the augment -s- (sak) is added between the root and the augment -i-. This form is therefore often called the “siṣ-aorist.” The root stays in its full grade form (guṇáḥ), and only the parasmaipadám is used. In the ātmanēpadám of these verbs, the s-aorist is used instead.
Another aorist paradigm, using the suffix -sa, is used for the following types of verbs Aṣṭādhyāyī 3.1.45:
- those ending in ś, ṣ, s or h;
- those that have a medial i, ī, u, ū, r̥, or r̥̄;
- those that do not take the augment iṬ.
Hence roots that take this form include diś “point out,” viś “enter,” lih “lick,” duh “milk,” and so on. (Not dr̥ś “see,” which takes the s-aorist instead.)
In these forms, the root always stands in the zero grade. The endings -an and -anta of the third person plural are used instead of the endings -uḥ and -ata.
Four verbs that end in h (duh “milk,” dih “smear,” lih “lick,” and guh “hide”) optionally take the regular s-aorist endings before ātmanēpadám endings that begin with a dental (that is to say, the entire suffix of the sa-aorist is optionally deleted in such environments, see Aṣṭādhyāyī 7.3.73):
- duh 3rd person singular ātmanēpadám: ádhukṣata or ádugdha
- duh 2nd person singular ātmanēpadám: ádhukṣathāḥ or ádugdhāḥ
- duh 2nd person plural ātmanēpadám: ádhukṣadhvam or ádugdhvam
The root aorist refers to an aorist form in which the endings are added directly to the verbal root. In Classical Sanskrit, this form is only available for a small class of verbs — five that end in -ā (dā, dhā, sthā, pā, and gā, the last of which is a substitution of gam “go” in certain contexts), and bhū — and only in the parasmaipadám Aṣṭādhyāyī 2.4.77. (These same roots use the s-aorist in the ātmanēpadám.) But in the Vedas forms from other roots, and ātmanēpadám forms, are attested.
The third-person plural ending in -uḥ for roots endings in -ā, before which this vowel is lost; for bhū, the ending is -an, before which bhū becomes bhuv-.
The thematic aorist is formed by a suffix -a- (Pāṇini’s aṆ) added to root in the zero grade. The root dr̥ś, and roots ending in the vowel -r̥, take guṇáḥ instead Aṣṭādhyāyī 7.4.14. It has both parasmaipadám and ātmanēpadám forms, but not all roots can take both. Following Pāṇini we can think of three classes of roots that take this form:
- those that take the a-aorist in both the parasmaipadám and ātmanēpadám, like sr̥ “go,” śās “order,” and r̥ “go”;
- those that take it in the parasmaipadám and only optionally in the ātmanēpadám, like lip “smear,” sic “sprinkle,” and hvē “call”;
- those that take it in the parasmaipadám and not in the ātmanēpadám, like puṣ “nourish,” śuṣ “dry out,” and dyut “shine.”
The thematic aorist is very similar to the imperfect of thematic verbs, especially those in class 6, which also appear in the zero grade. One difference, however, is that the imperfect uses the present stem, which in some cases is formed by nasal infixation, whereas the aorist stem never has a nasal infix. Contrast:
- sic “sprinkle” → á-siñc-a-t “he sprinkled” (imperfect)
- sic “sprinkle” → á-sic-a-t “he sprinkled” (aorist)
There are two verbs that form their thematic aorist stems by reduplication: vac “speak” and pat “fly.” In both cases the root syllable appears in the zero grade:
- vac + aṆ → á-va-vc-a- → ávōca-
- pat + aṆ → á-pa-pt-a- → ápapta-
A few other verbs have idiosyncratic a-aorist formations:
- as “throw” + aṆ → ā́stha-
- śās “teach” + aṆ → áśiṣa-
- naś “perish” + aṆ → ánēśa-