In Sanskrit, a nominal can form a compound with another nominal (Aṣṭādhyāyī 2.1.4 saha supā). Compound words can be quite large, but they are almost always analyzeable into binary constituents: that is, at any given level of analysis, a compound can be broken up into two parts, and these two parts must be related to each other in a specific way. This chapter will introduce the types of compounds and the relations that obtain between members of a compound.
An important concept when talking about compounds is that of the head and its dependent. The head of a compound is the word which determines the syntactic category of the compound as a whole. It can be defined more simply and straightforwardly in a negative way: the head is the word that is not subordinate to, or does not modify, any other word within the compound. The dependent word always modifies the head.
In the Indian grammatical tradition, the head is called pradhānam, and the dependent is called upasarjanam.
Generally compounding one word with another entails the loss (luk in Pāṇini’s terms) of the inflectional ending of the first word. The first word therefore appears in its stem form, although for certain types of nouns, the form used in compounds might be slightly different from the stem form (see note below). Since inflectional endings normally mark the relation between words, in their absence, word order becomes important for specifying the relationship. In Sanskrit compounds (as in English compounds), the head appears to the right of the dependent in the vast majority of cases. Sometimes, however, the first word retains its inflectional ending even when compounded with another word; these compounds are called aluk-samāsaḥ.
The following exceptions should be noted to the general principle that the form of a word in compound is identical to its stem form:
- stems in -n lose the final nasal and simply use the preceding vowel:
- balin- “strong” → bali-
- karman- “action” → karma-
- rājan- “king” → rāja-
- words that have multiple stem forms generally use their weakest form:
- vidvāṁs-, viduṣ- “scholar” → viduṣ-
- bhavant-, bhavat- “you” → bhavat-
- pronouns (sarvanāmāni) often use a special form for the stem, generally (but not always) identical to the neuter nominative-accusative singular:
- tatparaḥ ← tat param yasya saḥ “intent upon that”
- yuṣmadasmatpratyayaḥ ← yuṣmākam asmākam ca pratyayaḥ “the idea of ‘you’ and ‘me’”
- kiṁnāmaḥ ← kim nāma yasya saḥ “having what as his name?”
- yatputraḥ ← yasya putraḥ “the son of whom”
An exception to the above is mahant-, mahat-, which appears as mahā- when it is coreferential with the following word in a compound (either a karmadhārayaḥ or a samānādhikaraṇabahuvrīhiḥ); see Aṣṭādhyāyī 6.3.46.
- mahājanaḥ ← mahāṁś ca sa janaś ca “great people, the majority”
- mahārājaḥ ← mahāṁś ca sa rājā ca “great king”
When a word appears in its stem form in a compound, you will have to use word order and context in order to determine the categories that are usually expressed by the inflectional endings: the relationship with the following word, usually expressed as case; the number; and (in the case of adjectives) the gender.
Reference will be made in this section to the “analytic paraphrase” (vigrahavākyam) that are the semantic equivalent of compound expressions. The analysis of compounds in Sanskrit typically proceeds by identifying what kind of compound it is, and then offering an equivalent expression in analytic terms. This strategy is used very often, for example, in Sanskrit commentaries. For details and plentiful examples, see Tubb and Boose, Scholastic Sanskrit.
Endocentric compounds are those wherein the head is actually a constituent of the compound. The general test for endocentricity is whether the referent of the compound as a whole can be described by one of its constituent terms: since a bluejay is a jay, a checkerboard is a board, and a guest lecturer is a lecturer, all of these expressions are endocentric; contrast these with the exocentric expressions Blackbeard (who is not a beard) or skinhead (who is not a head).
Under the general category of endocentric compounds we may distinguish the following subtypes:
- compounds wherein the second member is the head of the first member;
- compounds wherein the first member is the head of the second member; and
- compounds where the first and second member both serve as the head.
The second subtype is a passable approximation of what Indian grammarians call avyayībhāvāḥ, or “adpositional phrase compounds” (see below). The third describes what Indian grammarians call dvandvāḥ, or “coordinative compounds” (see below). The first corresponds to the large category of tatpuruṣaḥ compounds.
In all tatpuruṣaḥ compounds, the first member is the dependent of the second member, which is the head. The nature of that dependency, however, may differ across compounds, and accordingly three further subtypes of tatpuruṣaḥ compounds are distinguished:
- vibhaktitatpuruṣaḥ, wherein the relationship can be expressed by a case suffix;
- karmadhārayaḥ, wherein the two members are coreferential and the second qualifies the first; and
- upapadatatpuruṣaḥ, wherein the first member is a dependent argument of the second member, which is a verbal noun or adjective.
Each of these three subtypes admits of further subtypes, as will be discussed below.
Case compounds vibhaktitatpuruṣāḥ are those in which the relationship between the head and the dependent can be expressed analytically as a relationship between two case-forms. What distinguishes these case compounds from coreferential compounds is that the constituent words in a case compound do not refer to the same thing, but rather express a relationship between two different things.
In theory, the relationship between the two constituents can be expressed by any of the cases, since all of the cases can, in theory, express a relationship between two nouns. In practice, however, the genitive is by far the most common case for expressing a relationship between nouns — indeed, that is its primary grammatical function — and therefore the ṣaṣṭhītatpuruṣaḥ is the most common of these case compounds. Among the other cases, the nominative and the accusative are not generally used adnominally, but rather to mark core arguments of the sentence (the subject and the object, respectively). Adnominal usages of these cases, however, do occur, and therefore case compounds can be formed.
Nominative case compounds prathamātatpuruṣāḥ are, as noted above, quite uncommon, since the nominative case does not generally express an adnominal relationship. (As noted above, if the two words actually refer to the same thing, they are coreferential and hence the compound will not be considered a case compound, or vibhaktitatpuruṣaḥ, but a coreferential compound, or karmadhārayaḥ.) Nominative case compounds are allowed by Aṣṭādhyāyī 2.2.1 and Aṣṭādhyāyī 2.2.2, when compounding a word with another word expressing a part thereof. In the analytic expressions, the word expressing a part is in the nominative, and the word expressing that of which it is the part is in the genitive.
- ardhapippalī ← ardhaṁ pippalyāḥ “half of a black pepper”
- pūrvakāyaḥ ← pūrvaṁ kāyasya “the forepart of the body”
- uttarakāyaḥ ← uttaraṁ kāyasya “the top part of the body”
This type of formation is necessary to account for the position of words within the compound. Other compounds involving a word for a part, such as grāmārdhaḥ, can be more straightforwardly analyzed as genitive case compounds.
Accusative case compounds dvitīyātatpuruṣāḥ are usually formed when a verbal adjective enters into a compound with an accusative object.
- grāmagataḥ ← grāmaṁ gataḥ “having gone to the village”
- kr̥ṣṇaśritaḥ ← kr̥ṣṇaṁ śritaḥ “having taken refuge in Kr̥ṣṇa”
As can be seen from these examples, the verbal adjective is usually a past participle in -ta that expresses the agent of the action, which is often the case in verbs of motion, and the noun with which it is compounded is usually the patient of the action, which in the case of verbs of motion is generally the destination.
Instrumental case compounds tr̥tīyātatpuruṣāḥ are formed whenever a nominal form can take a modifier in the instrumental case, which might express accompaniment, instrumentality, or a characteristic, or, in the case that the head is a verbal adjective, the agent or instrument of the verbal action.
- guḍamiśrāḥ ← guḍena miśrāḥ “mixed with jaggery”
- asikalahaḥ ← asibhiḥ kalahaḥ “a fight with knives, a knife-fight”
- māṣōnaḥ ← māṣēṇa ūnaḥ “short by a gram”
- yatnakr̥taḥ ← yatnēna kr̥taḥ “produced with effort”
- lakṣmaṇasahitaḥ ← lakṣmaṇēna sahitaḥ “together with Lakṣmaṇa”
Dative case compounds caturthītatpuruṣāḥ are permitted by Aṣṭādhyāyī 2.1.36 when one noun is said to be “for the purpose of” another, and particularly when used with the words arthaḥ “purpose,” baliḥ “offering,” hitaḥ “beneficial,” sukham “pleasure,” and rakṣitam “protection.”
- yūpadāruḥ ← yūpāya dāruḥ “wood for a sacrificial post”
- brāhmaṇārtham ← brāhmaṇēbhyō ’rtham “for the sake of Brāhmaṇas”
- kubērabaliḥ ← kubērāya baliḥ “an offering to Kubēra”
- gōhitam ← gōbhyō hitam “beneficial to cows”
- gōrakṣitam ← gōbhyō rakṣitam “protection for cows”
Ablative case compounds pañcamītatpuruṣāḥ are permitted by Aṣṭādhyāyī 2.1.37, to express fear of something, and Aṣṭādhyāyī 2.1.38, when the latter word is a verbal adjective implying separation (although Pāṇini says that these are only rarely compounded). Furthermore, Aṣṭādhyāyī 2.1.39 allows certain ablative forms with an adverbial meaning to be compounded with a following verbal adjective.
- vr̥kabhayaḥ ← vr̥kēbhyō bhayaḥ “fear of wolves”
- apētadōṣaḥ ← dōṣēbhyō ’pētaḥ “free of faults”
- svargapatitaḥ ← svargāt patitaḥ “fallen from heaven”
- kr̥cchralabdhaḥ ← kr̥cchrāl labdhaḥ “obtained with difficulty”
Genitive case compounds ṣaṣṭītatpuruṣāḥ are the most common type of case compound. Most often they have a possessive meaning, although nearly any other type of relationship between the two nominal forms is possible. When the final word expresses a verbal action, the genitive is permitted in the sense of the patient.
- rājapuruṣaḥ ← rājñaḥ puruṣaḥ “the king’s man”
- daśāhētuḥ ← daśāyā hētuḥ “the cause of the condition”
- ōdanabhōjanam ← ōdanasya bhōjanam “the eating of rice”
- dēvaguruḥ ← dēvānāṁ guruḥ “the teacher of the gods”
Locative case compounds saptamītatpuruṣāḥ, like their corresponding analytic expressions, are used to express location and reference, as well as some expressions of time.
- girigrāmaḥ ← giriṣu grāmaḥ “a village in the mountains”
- lalāṭalōcanam ← lalāṭē lōcanam “eye in the forehead”
- saṁvatsaradēyam ← saṁvatsarē dēyam “to be given within [one] year”
- sthālīpakvaḥ ← sthālyāṁ pakvaḥ “cooked in a plate”
- akṣapravīṇaḥ ← akṣēṣu pravīṇaḥ “skilled at dice”
In the analytic paraphrase vigrahavākyam of such compounds, the first element is simply expressed in the appropriate case-form, followed by the second member, which is usually expressed in the nominative singular, as in the above example.
Coreferential compounds karmadhārayáḥ are those in which the two constituents refer to the same thing (Aṣṭādhyāyī 1.2.42 tatpuruṣáḥ samānā́dhikaraṇaḥ karmadhārayáḥ). This relationship is called “coreferentiality” (sāmānādhikaraṇyam). In these compounds, as endocentric compounds more generally, the final member is the head of the compound, which is modified by the first member. Hence the compound as a whole is inflected exactly in the same way as its final member, regardless of whether it is a noun or an adjective.
Generally, in the analytic paraphrase of such a compound, an anaphoric pronoun (such as tád, idám, or adáḥ) is employed in order to show that the two constituents describe the same thing, as shown in the examples below demonstrate.
One very common category of coreferential endocentric compounds is that in which the first constituent is an adjective. Although this is not made explicit in the analytic expression, the first constituent will almost always qualify the second—even if the second is also an adjective. Usually the most accurate translation is “y that is x.” In such cases the adjective takes its basic stem form, that is, it uses the masculine/neuter stem (e.g., sundara-) rather than a stem with a feminine suffix (e.g., sundarī-).
- nīlōtpalam ← nīlaṁ ca tad utpalaṁ ca “a lotus that is blue,” “blue lotus”
- lōhitakr̥ṣṇaḥ ← lōhitaś ca sa kr̥ṣṇaś ca “black that is red,” “reddish-black”
- sajjanaḥ ← sañ ca sa janaś ca “a person that is good,” “good person”
- vicakṣaṇastrī ← vicakṣaṇāś ca sā strī ca “a woman that is clever,” “clever woman”
Another type involves two substantives that are combined. Once again, the first constituent qualifies the second.
- brahmarākṣasaḥ ← brahmā ca sa rākṣasaś ca “a Rākṣasa that is a Brahman”
- rājarṣiḥ ← rājā ca sa r̥ṣiś ca “a sage that is a king,” “a royal sage”
A large number of coreferential compounds cannot be analyzed “in their own words” (hence they are said to be asvapadavigrahaḥ, “not having an analytic paraphase that uses the same words”), because their first member is a “bound form” which can only occur when followed by another form. This category includes the following:
- nañtatpuruṣaḥ, negative endocentric compounds;
- prāditatpuruṣaḥ, coreferential compounds beginning with a prefix, among which are counted the standard prefixes that are also affixed to verbal forms (prādi) as well as a few special prefixes that are only used for nouns, such as ku;
A nañtatpuruṣaḥ is simply formed by prefixing the negative suffix nañ, which generally takes the form a- before consonants and an- before vowels, to a nominal form. The resulting compound has the opposite meaning of the negated term, which may be either a noun or an adjective:
- abrāhmaṇaḥ ← na brāhmaṇaḥ “a non-Brāhmaṇa”
- apitā ← na pitā “a non-father”
- anāraktaḥ ← nāraktaḥ “not attached”
- aprasaktaḥ ← na prasaktaḥ “not relevant
They are considered karmadhārayáḥ because the negative particle, although not inflected, directly modifies the head of the compound. As can be seen from these examples, the analytic expression generally employs the word nañ in its independent form, as the prepositive particle na.
The prāditatpuruṣāḥ are formed by prefixing one of the indeclinable prefixes (prādi, see verbal prefixes above) to the head. In some cases, they function exactly like adjectives, and are generally rendered as adjectives in an analytic paraphrase. These adjectives generally incorporate the indeclinable prefix.
- supuruṣaḥ ← śōbhanaḥ puruṣaḥ “a good person”
- atipuruṣaḥ ← atiśāyitaḥ puruṣaḥ “a first-rate person”
- durjanaḥ ← duṣṭō janaḥ “a bad person”
Note that most of these prefixes are also used as the first element of “governing compounds” upapadatatpuruṣāḥ, that is, as adverbial modifiers of the verbal action expressed the second member of the compound, when that is a verbal noun or adjective (see below). I discuss those compounds below. Here, we are concerned only with compounds wherein the two members are coreferential.
The prefix ku and its substitutes are used with a perjorative sense that is often paraphrased as kutsitaḥ “blamed, found fault with.”
- kukaviḥ ← kutsitaḥ kaviḥ “a bad poet”
- kāpuruṣaḥ ← kutsitaḥ puruṣaḥ “a bad person”
- kaduṣṭraḥ ← kutsita uṣṭraḥ “a bad camel”
- kāmadhuraḥ ← kiṃcin madhuraḥ “slightly sweet”
While most compounds—except for dvandvas and, arguably, karmadhārayas—involve the “governing” of one word by another, the term “governing compound” (upapadasamāsaḥ, lit. “a compound involving a subordinate word”) is reserved for a special class of tatpuruṣa compounds, wherein the relation of the head and its dependent cannot be expressed through a case-relationship between two independently-occurring nominal forms, because the head is a bound form which cannot occur outside of a compound. A parallel in English is chimney sweep, since the word sweep is not used as an agent noun in its own right.
In the Aṣṭādhyāyī, such compounds are generally taught by specifying
- the first member, or dependent (in the locative case), sometimes with additional specifications regarding the thematic role that the dependent has in relation to the action designated by the verbal root;
- the second member, or head, which will always be a bound form (in the sense that it does not occur outside of such compounds) and will also be a primary nominal derivative of the verb, and hence is taught by Pāṇini as consisting of:
- the verbal root (in the ablative case);
- the primary derivational suffix (i.e., kr̥t).
Because the head is a bound form, the analytic expression vigrahavākyam) corresponding to an upapadasamāsaḥ will usually take the form a sentence involving a conjugated verb. Examples will be given below.
- vr̥traghnaḥ ← vr̥tram hanti “Vr̥tra-slayer, slayer of Vr̥tra”
- sōmasut ← sōmaṁ sunōti “Sōma-presser, priest who presses the Sōma”
- jaladaḥ ← jalaṁ dadāti “water-giver, cloud”
Since such compounds are the only context in which certain types of primary derivatives occur, I provide here a list of the most common primary derivational suffixes that are used to form upapadasamāsāḥ, along with examples.
- aṆ Aṣṭādhyāyī 3.2.1. A very general suffix used to form compounds with Ṇ-gradation in the root.
- kumbha- + kr̥- + aṆ- ( + sU) → kumbhakāraḥ “pot-maker” (= कुम्भं करोति)
- aṬ Aṣṭādhyāyī 3.2.16. Another suffix that is simply -a-, but this one requires guṇa of the root. The effect is that the suffix is -a- with zero grade of the root. The feminines formed with this suffix use the suffix ṄīP Aṣṭādhyāyī 4.1.14.
- puraḥ- + sr̥- + aṬ- ( + sU) → puraḥsaraḥ “one who goes in front” (= पुरः सरति)
- śōka- + kr̥- + aṬ- + ṄīP ( + sU) → śōkakarī “one who causes sorrow” (= शोकं करोति)
- KhaL Aṣṭādhyāyī 3.3.126. The suffix -a, with guṇa, attached to roots when preceded by the indeclinable prefixes īṣad, su, and duḥ. This suffix has a patientive meaning by Aṣṭādhyāyī 3.4.70
- duḥ + tr̥- + KhaL ( + sU) → duṣtaraḥ “difficult to cross” (= तरितुमशक्यः)
- duḥ + labh- + KhaL ( + sU) → durlabhaḥ “difficult to obtain” (= लब्धुमशक्यः)
- ṬaK Aṣṭādhyāyī 3.2.8. Actually identical to aṬ, but taught after roots that historically ended in laryngeals. The effect is that the suffix is -a- with zero grade of the root. The feminines formed with this suffix use the suffix ṄīP Aṣṭādhyāyī 4.1.14.
- sāman- + gai- + ṬaK- ( + sU) → sāmagaḥ “sāman-singer” (= सामानि गायति)
- surā- + pā- + ṬaK- + ṄīP ( + sU) → surāpī “drinker of wine” (= सुरां पिबति)
- Ḍa Aṣṭādhyāyī 3.2.48. The suffix a, replacing the vowel and any consonants that follows; taught for the roots ending in nasals, like gam and han.
- pāra- + gam- + Ḍa- ( + sU) → pāragaḥ “one who goes to the furthest shore”
- KhaC Aṣṭādhyāyī 3.2.38. Thus suffix induces the augment mUM (Aṣṭādhyāyī 6.3.37).
- priya + vad- + KhaC ( + sU) → priyaṁvadaḥ “one who speaks kindly”
- KhaŚ Aṣṭādhyāyī 3.2.28. This suffix also causes the affixation of the augment mUM, but the root takes the form that it takes in the present system, i.e., reduplication and present-stem forming suffixes are
- jana- + ēj- + KhaŚ (( + sU) → janamējayaḥ “making the people tremble”
- paṇḍita- + man- + KhaŚ (( + sU) → paṇḍitaṁmanyaḥ “one who thinks he is learned”
- KviN Aṣṭādhyāyī 3.2.58. A zero affix, which causes the root to take its zero-grade form. Taught for a small class of words.
- r̥tu- + yaj- + KviN ( + sU) → r̥tvik “one who sacrifices at the proper moment”
- KviP Aṣṭādhyāyī 3.2.61. A zero affix, which causes the root to take its zero-grade form; in addition, if the root is a light syllable, the augment
t is affixed to it in order to make the root-syllable heavy.
- śatru- + ji- + KviP ( + sU) → śatrujit “victorious over enemies” (śatrūñ jayati)
- Ṇvi Aṣṭādhyāyī 3.2.62. A zero affix, which causes the root to take its “Brugmann” form (induced by Ṇ).
- ardha- + bhaj- + Ṇvi ( + sU) → ardhabhāk “getting half a share” (ardhaṁ bhajatē)
Adpositional compounds are what I call avyayībhāvaḥ, compounds that “become indeclinable” and function as adverbs. Like many other adverbs, they take the form of a neuter singular nominal. In these compounds, the first member is usually an indeclinable word—often an adposition karmapravacanī́yaḥ—and the second word is a nominal form that functions as its dependent. Thus, in this type of compound, the first member is the syntactic head.
Sometimes such compounds can be analytically paraphrased as two independent words, that is, as a karmapravacanī́yaḥ governing a case-form of the dependent:
- ābrahma ← ā brahmaṇaḥ “starting from Brahma”
Most often, however, these compounds cannot be analytically paraphrased in their own words asvapadavigrahaḥ. This is because the indeclinable first member does not generally govern a case-form, or does not govern one in the given sense. Avyayībhāvaḥ compounds with práti, for example, generally have a distributive sense, whereas as an adposition, práti most often conveys the sense of a goal of motion.
- pratidinam ← dinē dinē “every day”
In many other cases, other paraphrases must be found:
- yathāśakti ← śaktim anatikrāmya ”in proportion to one’s power”
- upanadi ← nadyāḥ samīpē “near the river”
- anujyēṣṭham ← jyēṣṭhānupūrvēṇa ”in order of age”
Compounds that have multiple heads are called “coordinative” compounds or dvandvāḥ. There are two types of such compounds.
A countable or additive coordinative compound itarētaradvandvaḥ is one where the grammatical number of the resulting compound is a function of the grammatical number of each of its constituents. The grammatical gender of the compound as a whole is usually that of the final member of the compound.
- rāma- + lakṣmaṇa- → rāmalakṣmaṇau “Rāma and Lakṣmaṇa”
- aśva- + gaja- + ratha- → aśvagajarathāḥ “horses, elephants, and chariots”
The standard analysis vigrahaḥ of such compounds involves inserting the coordinating particle ca after each member, hence rāmaś ca lakṣmaṇaś ca, or aśvāś ca gajāś ca rathāś ca.
A collective coordinative compound samāhāradvandvaḥ refers to its constituents as forming a collective, and hence the compound as a whole has singular morphology. Usually these compounds are neuter.
- jaṅgama- + sthāvara- → jaṅgamasthāvaram “mobile and immobile”
The analysis of these kinds of compounds is identical to that of itarētaradvandvāḥ, except with the addition of samāhāraḥ or samāhr̥taḥ, hence jaṅgamāni ca sthāvarāṇi ca samāhr̥tāni, or jaṅgamāni ca sthāvarāṇi ca tēṣāṁ samāhāraḥ.
Exocentric compounds, in contrast to endocentric compounds, are those wherein the head is not a member of the compound. The compound as a whole gets its syntactic category from its head, and in many cases, it qualifies the head as an adjective, hence it also receives its gender and number from its head. These compounds are thus “headless.” They will always bold fail the above-mentioned test of endocentricity: if x-y represents a compound, the answer to the question “is x-y either an x or a y?” will be no. Thus, to use English examples, the following are exocentric: skinhead (not a head, but a person); barefoot (not a foot, but a person); pale-faced (not a face).
As the last example shows, exocentric compounds in English often use the suffix -ed. This is not a past passive participle (there is no verb “pale-face”) but a compound-final suffix, which is called samāsāntapratyayaḥ in Sanskrit. Exocentric compounds in Sanskrit often use such suffixes as well, as noted below.
The Sanskrit term for these kinds of compound is bahuvrīhiḥ, which is, like tatpuruṣaḥ, an instance of the grammatical phenomenon it names: a bahuvrīhiḥ is not “a lot of rice,” but a man who has “a lot of rice.” In contrast to the preceding categories of compounds, the formation of exocentric compounds can be thought of as the formation of an adjectival stem.
Typically the constituents of an exocentric compound are coreferential, that is, they refer to the same thing. But there are exocentric compounds whose constituents are not coreferential (called vyadhikaraṇabahuvrīhayaḥ), which will be exemplified below. They are also commonly analyzed using relative clauses, a strategy that will be followed in the analyses below; the relative pronoun supplies the gender and number of the head noun, which in these examples will always be cited in the nominative prathamā case.
- bahur vrīhiḥ yasya saḥ → bahuvrīhiḥ “one who has a lot of rice” (samānādhikaraṇabahuvrīhiḥ)
- nīlaḥ kaṇṭhaḥ yasya saḥ → nīlakaṇṭhaḥ “one whose neck is dark blue” (samānādhikaraṇabahuvrīhiḥ)
- nīlam ambaram yasya saḥ → nīlāmbaraḥ “one who wears a dark cloak” (samānādhikaraṇabahuvrīhiḥ)
- vyastaṁ viśvaṁ yābhiḥ tāḥ → vyastaviśvāḥ “those that have put the universe asunder” (samānādhikaraṇabahuvrīhiḥ)
- mahān anubhāvaḥ yasya saḥ → mahānubhāvaḥ “one whose dignity is great” (samānādhikaraṇabahuvrīhiḥ)
- dattaṁ sarvasvaṁ yēna saḥ → dattasarvasvaḥ “one who has given everything he owns” (samānādhikaraṇabahuvrīhiḥ)
- cakraḥ pāṇau yasya saḥ → cakrapāṇiḥ “one in whose hand there is a discus” (vyadhikaraṇabahuvrīhiḥ)
The gender of the final word will sometimes change in a bahuvrīhiḥ compound. That is, a word that standardly appears in one gender might have to appear in another gender if it is final within a bahuvrīhiḥ compound that describes something or someone of that gender. This is a very useful way of recognizing bahuvrīhiḥ compounds, provided that you have actually learned the gender of the final noun!
- dattāni ratnāni yēbhyaḥ tē → dattaratnēḥ “[men] to whom jewels have been given” (ratnam is neuter)
- pītaṁ jalaṁ yēna saḥ → pītajalaḥ “[a man] by whom water has been drunk” (jalam is neuter)
- mr̥tāḥ narāḥ yasmin tat → mr̥tanaram “[a family] in which the men have died” (naraḥ is masculine)
How do we know what form a word will take at the end of a bahuvrīhiḥ compound if it is used in a gender different from the one with which it is usually associated? Masculine and neuter words usually use the same stem, so the only challenge is remembering the corresponding masculine or neuter endings, which may indeed be a challenge for classes of nouns that are typically associated with one gender (for instance nominal stems ending in -s are almost always neuter, and they have different forms in the nominative-accusative of the masculine). Feminine words, however, usually use a different stem, formed with a feminine stem forming suffix (strīpratyayaḥ; see above). Going from a feminine to a masculine-neuter stem or vice versa is sometimes, but not always, as simple as shortening (Aṣṭādhyāyī 1.2.48) or lengthening the stem vowel:
- citrā gauḥ yasya saḥ → citraguḥ “[a man] who has a brindled cow” (gauḥ is feminine)
- kauśāmbyā niṣkrāntaḥ → niṣkauśāmbiḥ “[a man] who has left Kauśāmbī” (kauśāmbī is feminine)
- dattā mālā yasmai saḥ → dattamālaḥ “[a man] to whom a garland has been given” (mālā is feminine)
- dattā hastaḥ yasyai sā → dattahastā “[a woman] to whom a hand has been given” (hastaḥ is masculine)
No shortening takes place, however, if a feminine word is not formed with a feminine stem forming suffix:
- atikrāntā śrīḥ yasya saḥ → atiśrī “[a man] who has great prosperity” (śrīḥ is feminine)
In general, however, one must know which of the feminine stem forming suffixes to use in order to convert a masculine/neuter stem into a feminine stem. There are only two common ones: -ā- and -ī-.
- candra iva mukham yasyāḥ sā → candramukhī “[a woman] whose face is like the moon” (mukham is neuter)
Some exocentric compounds have a word as their first member which is a “bound form,” in that it cannot typically be used as an independent word. There are several subvarieties of these compounds. One, called prādibahuvrīhayaḥ, have a preverb as their first member. Since they cannot be expressed as independent words in the analysis of the compound, they are usually “expanded” into a form that can be used as an independent word.
- utkaṇthaḥ ← unnataḥ kaṇṭhaḥ yasya saḥ “one whose neck is uplifted” (almost always in anticipation)
Another variety of exocentric compounds with bound forms as first members are negative compounds nañbahuvrīhayaḥ, which, as their Sanskrit name suggests, are formed with the negative particle nañ. This particle is usually expanded in the analytic phrase to a nañ-tatpuruṣaḥ, namely, avidyamānaḥ “non-existing.”
- agr̥haḥ ← avidyamānaḥ gr̥haḥ yasya saḥ “one who has no home, homeless”
Finally, there are sociative compounds sahabahuvrīhayaḥ, whose first member can be sa-, which is expanded to saha with an instrumental case-form in the analytic expression, and often expressed with the verb vartatē.
- samātāpitr̥kaḥ ← saha mātāpitarau vartatē “together with his mother and father” (for the form mātāpitr̥- see Aṣṭādhyāyī 6.3.25).
Under certain conditions, a suffix is added to the final member of a compound. Thus the stem of the word on its own may differ from its stem when it occurs at the end of a compound. These suffixes are called “compound-final” samāsāntāḥ Aṣṭādhyāyī 5.4.68.
When the word rā́jan- “king,” áhan- “day” or sákhi- “friend” is final within a tatpuruṣáḥ, they become rāja-, aha-, and sakha- Aṣṭādhyāyī 5.4.91, with a change to the masculine gender in the case of áhan- (normally neuter).
- mahārājaḥ ← mahāṁś ca sa rājā ca “great king”
- brāhmaṇasakhaḥ ← brāhmaṇasya sakhā “friend of a Brahmin”
- uttamāhaḥ ← uttamaṁ ca tad ahaś ca “the final day”
When rātri- “night” is preceded by the words áhan- “day,” sarvá- “entire,” a word for a part (e.g., pūrva- “earlier”), a numeral, or the word puṇya-, then it becomes rātra- Aṣṭādhyāyī 5.4.87 with a change to the masculine gender; similarly the word áhan- becomes ahna- under the same circumstances Aṣṭādhyāyī 5.4.88:
- pūrvarātraḥ ← pūrvā rātriḥ “the early part of the night”
- pañcarātraḥ ← pañca rātrayaḥ “five nights”
- aparāhṇaḥ ← aparam ahaḥ “the latter part of the day, evening”
In general, bahuvrīhiḥ compounds can take the suffix -ka- (feminine -ikā) which will very often make the noun easier to inflect.
- tat ātmā yasya tat → tadātmakam “[a thing] of which the essence is that”
- bahvī śrīḥ yasya saḥ → bahuśrīkaḥ “[a man] who has great prosperity”
- avidyamānaṁ manaḥ yasya saḥ → amanaskaḥ “without the mind”
- jīvantau pitarau yasya saḥ → jīvatpitr̥kaḥ “[a man] whose parents are still alive”