You are here
Compound words
Compound words (समास :'thrown together i.e. sam:together + as:throw') in Sanskrit are formed by a semantic composition of constituent terms. Such compound words are very common in Sanskrit, and can be observed in almost every stanza of the Srimad Bhagavad Gita (some examples are provided below). Most of these compounds are treated as nominals (except the avyayIbhAva samAsa), undergoing the usual nominal declension processes (see [Kale1995][1] for exceptions).
The major types are as follows:1
| Type | Description | Example from Srimad Bhagavad Gita | Semantic equivalent (vigrahavAkya) |
| dvandva samAsa (द्वन्द्व) | There are broadly two sub-types:
|
|
|
| tatpuruSHa samAsa (तत्पुरुष) | There are two terms, of which the final term is the Head. The first term bears a Case relation with the second term, and can be any one of the Cases ACC, INS, DAT, ABL, GEN, and LOC. The Number and Gender of the final constituent determines the Number and Gender of the compound. | Stanza 1.2 pANNDavAnIkam ('army of the Pandavas') | pANNDavAnAm-(GEN-P) anIkam (ACC-S) |
| karmadhAraya samAsa (कर्मधारय) | There are two terms, of which the final term is the Head, and both terms share the same Case (samAnAdhikaraNNa: governed in common). The first term bears an appositional or adjectival relation that qualifies the second term in some sense. The Number and Gender of the final constituent determines the Number and Gender of the compound. In the common case, the first term will be an adjective or a pronoun that modifies the second term. There are also a few specialised forms of this compound that deserve special mention:
|
Stanza 2.65 sarvaduHakhAnAm ('[of] all sorrows') | |
| upapada samAsa (उपपद) | There are two terms, of which the final term is the Head. The second term is usually a term derived from a verbal root with an abstract suffix (for e.g., 'aNN') that requires the presence of a nominal prefix called the 'upapada' term (for the 'aNN' abstract suffix this may be the object of the verb) for the transformation. Hence, for example, in the compound word 'kumbhakAra: one who makes pots', 'kAra: one who makes' is a verbal noun derived from the verb 'kRu:to do' with the abstract affix 'aNN', while its 'upapada' term is 'kumbha:pot'. | Stanza 1.5 purujit ('the conqueror of many') | purUn jayati iti |
| naJ tatpurusha samAsa (नञ) | A prefix 'a' (or 'an') results in the negation of the nominal. | Stanza 2.11 ashochyAn ('not lamented') | na shochya |
| bahuvrIhi samAsa (बहुव्रीहि) | The compound word acts in an adjectival role for some external nominal entity, and adopts the key syntactic features of the external nominal (i.e. the Number, Gender, and Case of the nominal it modifies). In Stanza 1.47, the bahuvrIhi compound ACC-S 'sasharam' term modifies its Head Noun ACC-S 'chApam:bow' and adopts its features despite the fact that 'shara:arrow' has a Masculine Gender. By convention, the vigrahavAkya (the interpretation of a compound word) has the form of a relative clause with the relative pronoun expressing the Case relation with the external nominal entity. | Stanza 1.9 tyaktajIvitAHa ('the ones by whom lives were renounced') | tyaktAHa jIvitAHa yEHa-(INS-P) te |
| avyayIbhAva samAsa (अव्ययीभाव) | The first term is more important (usually a preposition or an adverb). The compound is treated as an Indeclinable, and does not undergo Case inflection. | Stanza 1.11 yathAbhAgam ('each in their respective places') | yathA bhAgam |
NOTES:
It is frequently very difficult to correctly identify the type and semantic equivalent of a 'samAsa' (only the 'dvandva' compound is easy to spot, and that too only if it has two constituent terms). This analytical exercise should only be carried out after the sentence is parsed successfully, and the syntactic and semantic relations between the various underlying terms of the sentence are fully understood. For example, a 'bahuvrIhi' compound is usually bound to an external Nominal and shares its Gender, Number and Case (in much the same manner as an Adjective), whereas a 'tatpuruSHa' or a 'karmadhAraya' compound may stand independently. However, note that the referent of a 'bahuvrIhi' compound will not usually be explicitly present in the sentence.
Take, for example, the compound word 'pItAmbara', which is a composition of the underlying terms 'pIta:yellow' and 'ambara:garment/sky'. The reader must figure out from a semantic analysis of the sentence whether the compound word refers to Lord Vishnu (in which case it would be a 'bahuvrIhi' compound), or whether it refers to a garment or the sky (in which case it would be a 'karmadhAraya' compound). In this specific case, a test of the gender of the term will inform us whether it refers to a garment or the sky (i.e. if it has a Neuter gender), and is therefore a 'karmadhAraya' compound. But note that it is far from trivial to determine the gender of a term, unless one is fortunate to find a Masculine Nominative or Vocative form in the sentence. This is because the Masculine and Neuter declensions of most nominals are identical except for the Nominative and Vocative declensions; a detailed syntactic and semantic analysis would thus be required to determine the gender of the term (note that a syntactic analysis may not be able to identify the correct gender in many cases).
Matters get even more complicated where the nominal Head and the compound term share the same gender. What if the Head of the compound word were a noun that had a Masculine gender form (remember that it is common for nominals in Sanskrit to have both Masculine and Neuter forms, such as 'pApa:Neuter:sin' vs. 'pApa:Masculine:wicked man') ? And what if the external referent of a 'bahuvrIhi' compound had Masculine as well as Neuter forms (for e.g., the compound word 'duHashAsana:bad governance' is composed from the derived nominal 'shAsana:governance' which has Masculine, Neuter, and Feminine forms, but it could also refer to a son of Dhritarashtra named 'duHashAsana')? Clearly, the analysis of compound words requires far more than a mere mechanical splitting of compound words into their components; it requires a deeper semantic analysis that is best left to human experts.
Compound words can be nested within other compound words, requiring great care when such nested structures are being analysed. Note that it is vital to know the order in which the compounds were nested, in the first place, by the original author. This unstated ordering of terms within the compound leads to the possibility of the compounds being misread by the reader in complex cases.
- Simple left-to-right nesting:For example, a tatpuruSHa compound ('lobhena upahata': [that is] damaged by greed) is nested inside a bahuvrIhi compound ('lobhopahatam chetas yasya saHa': [one whose] mind is damaged by greed) in Stanza 1.38 of the Srimad Bhagavad Gita resulting in the compound word 'lobhopahatachetasAHa' ('those whose minds are damaged by greed'). In this example, the tatpuruSHa compound was created first (the first two terms on the left a='lobha:greed', b='upahata:[that is] damaged'), and then subsequently compounded with the remaining term on its right (c='chetas:mind') to form the bahuvrIhi compound ((a+b)+c).
- Complex nesting: However, it is also possible to construct compound words that are not nested from left to right. For example, in Stanza 2.43 of the Srimad Bhagavad Gita, we can see the compound term 'janmakarmaphalapradAm: a bestowment of rebirth as the result of actions'. Here, a tatpuruSHa compound word (a+b) is nested inside a karmadhAraya compound word (c+(a+b)), which is in turn nested inside an upapada compound word ((c+(a+b))+d) (given that a='karman:action', b='phala:fruit/result', c='janman:birth', d='pradA:bestowment'). Note that, the tatpuruSHa compound (a+b) is the
second term of the karmadhAraya compound (c+(a+b)).
Not all compound words are compositional in respect of their meanings. Some compound words ('nitya samAsa') are prohibited from being broken down into their constituent terms because the meaning of the overall compound differs from the meaning derived from its constituent terms.
While most terms in a compound word (except the last term), lose their declension suffixes during the composition process, there are some exceptions (known as 'aluk: unelided') where an initial term is permitted to retain its declension suffix. Examples of 'aluk samAsa' include:
- 'yudhiSHThira' [i.e. yudhi (GEN-S-Masc.:in battle) + sthira:steady] in Stanza 1.16 of the Srimad Bhagavad Gita.
- 'parasmEpada' [i.e. parasmE (DAT-S-Neut.:for another) + pada:term]: a grammatical term denoting a Verb that acts to benefit someone other than the Agent of the Verb.
- 'Atmanepada' [i.e. Atmane (DAT-S-Neut.:for the self) + pada:term]: a grammatical term denoting a Verb that acts to benefit the Agent of the Verb.
Note that a participle embedded inside a compound word retains its syntactic relations with other terms that may be either inside the compound word or outside it. If the participle has external relations, the syntactic parser should be made aware of the embedded participle inside the nominal.
- External relation: Thus, for example, we have the compound word 'anAryajuSHTam' in Stanza 2.2 of the Srimad Bhagavad Gita, in which the passive past participle 'juSHTa:[that is] befitting' is embedded. In this stanza, the 'Link_subj_idam' comment links the compound word 'anAryajuSHTam' with its Subject (i.e. 'idam:this' and 'kashmalam:despair') of the participle. Hence we have the reading '... this despair [that is] not befitting for a respected person ...'. In this case, the syntactic parser must be made aware of the embedded participle inside the compound word that has a syntactic relation with another term elsewhere in the sentence.
- Internal relations: It can also be the case that the arguments of the participle are fully contained within the compound word, as seen in the compound word 'lobhopahatachetasAHa:[those who have] minds [that are] damaged by greed' in Stanza 1.38 of the Srimad Bhagavad Gita. Here the passive past participle 'upahata:[that is] damaged' has both its arguments (its Subject 'chetas:mind' and its Complement 'lobha:greed') contained within the compound word. In this case, the syntactic parser is not made aware of the participle inside the compound word, and the term appears to be similar to any other nominal in the parser output. This is a 'bahuvrIhi' compound word that bears an Adjectival relation with an external entity i.e. 'those ones [who have]' that is not explicitly present in the sentence.
The interpretation of compound words is a non-trivial semantic exercise, primarily because the constituent terms are usually in their stem forms (with no Case suffixes). It is beyond the capabilities of a syntactic parser to figure out the internal relations in a compound word, as this requires a deep 'understanding of the world'. There are several Paninian sutras that govern the construction of compound words, but despite this, human experts may not always agree on the correct analysis of compound words (as can be seen, for example, in the numerous differences between [KAL2015] [3] and [MM2015] [2] in their respective analyses of various stanzas), as it may involve a subjective interpretation of the intent of the original author of the sentence.
- 1. Note that the karmadhAraya, upapada, and naJ compounds words are conventionally regarded as tatpuruSHa compounds, but have their own unique characteristics, and the first term does not bear a Case relation with the second term.