(Hart pp. 59-61, 64-65, 68-74)
Sanskrit has four types of compounds: tatpuruṣa (तत्पुरुष), bahuvrīhi (बहुव्रीहि), dvandva (द्वंद्व/द्वन्द्व), and avyayībhāva (अव्ययीभाव). All compounds are formed in the same general way. Stem forms of nouns and adjectives are put together using saṃdhi rules, and only the last member of the combination is declined. The stem forms are usually the form cited as entries in dictionaries, except for nouns ending in -अम् and -अन् whose stem form is simply -अ without the nasal. All declensions of compounds follow the form of their last member, except for compounds ending in -अन् (for example, compounds ending in राजन्), such compounds are declined like देव or फलम्.
Pronouns have their own unique stem forms as follows:
| Pronoun | Stem form |
| अहाम् | मत् |
| वयम् | अस्मत् |
| त्वम् | त्वत् |
| यूयम् | युष्मत् |
| स | तत् |
Compounds have at least two components but can have more. In cases with three or more components, each juncture could be a different kind of compound. The best way to parse such multiple compounds is to start at the end and consider the last word’s relationship to the rest of the multi-compound and work backwards.
Sanskrit grammarians analyze Sanskrit compounds according to which part has syntactical predominance or emphasis, or prādhānyam (प्रधान्यम्). In tatpuruṣa compounds, the emphasis is on the second part of the compound as in the compound राजकोप, “the anger of the king”. In dvandva compounds the emphasis is on each element equally, as in देवगन्धर्वमनुष्याः, “gods, gandharvas, and men”. In bahuvrīhi compounds, the emphasis lies outside the compound on the word it modifies, though this external noun can be an implied pronoun so that धनुर्हस्त standing alone means “the one in whose hand there is a bow“ or more simply “the one who holds a bow”, but धनुर्हस्तो राजा means “the king with bow in hand”. The last type of compound, avyayībhāva, is rarer. They are undeclined adverbs, where the emphasis lies on the second word.
Tatpuruṣa Compounds
Tatpuruṣa compounds are those in which the first member of the compound has a case-relationship to the second member. The first member may be considered either singular or plural depending on the context. Thus, there are the following varieties of tatpuruṣa compounds:
| Case | Example |
| Nominative | राजृषि = राजन् + ऋषि, “a sage who is a king” नीलोत्पल = नील + उत्पल, “blue lotus” |
| Accusative | धनदातृ = धनम् + दातृ, “giver of money” |
| Instrumental | अग्निपाक = अग्नि + पाक, “cooking by fire” |
| Dative | प्रजाहित = प्रजा + हित, “good for (one’s) subjects” |
| Genitive | नदीतिरम् = नदी + तीरम्, “the shore of the river” |
| Locative | गिरिनदी = गिरि + नादी, “the river on the mountain” |
Because of the common use, Tatpuruṣa compounds that utilize a nominative relationship are called karmadhārayas (कर्मधार्य), “that hold/bear the action/role”, which is the preferred way to refer to them. Pronouns can be the first member of tatpuruṣa compounds, such as मद्धनम् (“my money”). Only the last member of the compound is declined when used in a sentence. So, नदीतिरे means “on the bank of the river”. The term “तत्पुरुष” is itself an example of a tatpuruṣa meaning “the servant of him”.
Dvandva Compounds
Dvandva compounds are essentially lists of items, compounded together so that only the last item is declined. (Dvandva means “pair or couple”.) There are two types of dvandva compounds: Itaretara dvandva (इतरेतर द्वन्द्व) and Samāhāra dvandva (समाहार). Itaretara dvandva, or “list pairs”, are simply a string of different items compounded together, where only the last one is declined. Samāhāra dvandva, or “composite pairs”, are two or more items that are considered as a single unit, such as “hands and feet” (पाणिपादम) or “food, sleep, and fear” (आहारनिद्राभयम्), which as a compound is used to mean “animal life”.
Bahuvrīhī Compounds
Bahuvrīhī is one of the most common types of compounds. It is unique in that its syntactical emphasis lies outside the compound itself. They act like adjectives agreeing in gender, case, and number with their antecedent or the word they modify. (Hart uses the term “antecedent”. However, the governing word is antecedent only in a grammatical sense. Because of the looseness of Sanskrit word order, the governing word for a Bahuvrīhī compound can be placed after the compound.) Gender is determined by the antecedent and not by the last word in the compound. So, in the above example, we had:
धनुर्हस्तो राजा
“The king with bow in hand”
That has a masculine antecedent, king. If however the antecedent was feminine, say the pronoun “she” (सा), we could have:
तस्यै धनुर्हस्तायै
“... to her with a bow in her hand.”
Because the bahuvrīhī compound matches its antecedent in gender, case, and number, if the antecedent is plural, then the compound is plural. This means that the grammatical number for the words within the compound is indeterminate. Thus,
वीरमित्रो राजा
Could mean either “the king whose friend is a hero” or “the king whose friends are heroes”, because वीरमित्रो must agree with राजा in gender, case, and number.
Understanding the meaning of bahuvrīhī compounds is best done by considering the two words that make up the compound as A (the first word) and B (the second word). The most common meaning is then “whose B is A”, but there can be different implied grammatical relationships within the compound that are not explicitly indicated. Below is a table of some of the possibilities:
| Meaning | Example | Translation |
| Whose B is A | वीरमित्र | “whose friends are heroes” |
| By whom B is A | हतगज | “by whom an elephant is killed” |
| To whom B is A | दत्तधन | “to whom money was given” |
| For whom B is A | पुत्रधन | “for whom money is sons” |
| In whose B is A | धनुर्हस्त | “in whose hands is a bow” |
| Whose B is for A | पुत्रकाम | “whose desire is for sons” |
| Whose B is of A | पुष्पतिर | “whose shore is of flowers” |
Thus, both B and A can have implied case meanings without the actual endings. Only the roots of the words are used for the compound, and, as mentioned above, the gender, case, and number of the last word matches the external antecedent.
The exception to this interpretation of a bahuvrīhī’s meaning is when the first member of the compound is a prefix. In such cases, the prefixes have special meanings as follows:
| Prefix | Meaning | Example | Notes |
| अ- / अन्- | “-less” “non-” |
अपुत्र – sonless |
अन्- is used before words beginning with vowels |
| दुस्- | “having bad B” “having difficult B” |
दुष्कर्मन् – having bad karma | Before क, ख, प, and फ, दुस् becomes दुष्. This is the antonym of सु-. |
| निस्- | “without B” “devoid of B” “lacking in B” |
निष्पुष्पम् उद्यानम् – garden without flowers | Before क, ख, प, and फ, निस् becomes निष् |
| स- / सह- | “accompanied by B” “in the company of B” “possessing B” “with B” |
सविरो राजा – “king with his heroes” सहविरो राजा – “king with his heroes” |
स- and सह- can be used interchangeably. |
| सु- |
“having lovely B” |
सुकर्मन् – “having good karma” | This is the antonym of दुस्- |
Because the bahuvrīhī takes the gender of its antecedent but the case ending is applied to the last word, which in its stand-alone form may be of a different gender, it is necessary to know how these last words are declined in each situation. The following rules apply:
- When the last word ends in -अ, -अम्, and -आ, the masculine, neuter, and feminine respectively end in -अ, -अम्, and -आ, following the declensions of देव, फलम्, and सेना.
- When the last word ends in -इ, they are declined like the masculine, neuter, and feminine in -इ with a few exceptions in long -ई.
- When the last word ends in -उ, they are declined like the masculine, neuter, and feminine in -उ with a few exceptions in long -ऊ.
- When the last word is a consonant stem, use the masculine, neuter, and feminine of that consonant stem.
- When the last word ends in -ई, -ऋ, or -ऊ, then add -क, कम्, or का for the masculine, neuter, and feminine respectively. These are declined like देव, फलम्, and सेना. Thus, सपत्निक from पत्नि; सदातृक from दातृ; सवधूक from वधू.
- When the last word ends in -अन्, either add -क, -कम्, or -का or else they are declined like राजन् (m.), नामन् (n.), or नदी (f.). Thus, स- plus राजन् can be सराजक (m.), सराजकम् (n.), सराजका (f.), सराजन् (m./n.), or सराज्ञी (f.).
Some examples of bahuvrīhi compounds follow:
- दीर्घकेश – “he whose hair is long”
- दीरघकेशा – “she whose hair is long”
- महाबाहु – “he whose arm is great/large”
- अन्यरूप – “having another form”
- सुमनस् – “good-minded, well disposed”
- अप्रज – “childless”
- चारचक्षुस् – “having spies for eyes”
- रामनामन् – “he whose name is Rāma”
Note that most of the bahuvrīhis above can also be interpreted as tatpuruṣas in which case they become nouns instead of adjectives. In that case, their gender and declension follows that of the last word. So दीर्घकेश can also mean “long hair”; महाबाहु , “great arm”; and रामनामन्, “the name of Rāma”. If there is a matching antecedent, then the compound is definitely a bahuvrīhī. If not, it could be either a bahuvrīhī with an implied pronoun antecedent or a tatpuruṣa, depending on context.
Avyayībhāva Compounds
These are rarer and not as important. Their syntactical emphasis lies on the first member, and they are not declined. (“Avyayībhāva”, the word itself, is a bahuvrīhī compound meaning, “whose nature is unchanging”.) Instead, they act as adverbs. Examples are:
- उपकुम्भम् – “in the pot”, from उप- a prefix meaning “near” and कुम्भ or “pot”
- अन्तर्गिरि – “inside the mountain”, from अन्त- meaning “inside” and गिरि meaning “mountain”