Karpāsa, Karpos, Karpas: How Sanskrit Borrowed the Word for Cotton from Dravidian Trade

Open the Ṛgveda (Rigveda), India's oldest surviving text, composed orally sometime between 1500 and 1200 BCE, and a handful of its words simply refuse to be Sanskrit. Linguists estimate roughly 300 "substrate words" (words absorbed from an earlier, underlying language into a dominant incoming one) sit inside its hymns. A good number of them look Dravidian. Michael Witzel, along with F.B.J. Kuiper and Franklin Southworth, has spent decades tracing these words by their phonology (sound patterns) and semantics (meanings). They are not scribal slips. They are fossils—evidence that speakers of Sanskrit and speakers of Dravidian languages lived in the same stretch of northwestern India, trading words along with grain and cattle.

Yet, to understand the full material history of this contact, we must look at what the oldest Sanskrit text leaves out. The Ṛgveda has no word for cotton. The poets of the early hymns wore wool (ūrṇā, wool) and skins. It is only in later Vedic texts, such as the Gautama Dharmasūtra (a Sanskrit text on law and custom) and the Manu Smṛti (Laws of Manu), that a specific word for cotton suddenly appears: karpāsa (कर्पास, karpāsa, cotton). This word went on to travel the global trade winds, becoming karpos in Greek, karpas in Hebrew, and carbasus in Latin. But its origins do not lie in the Indo-European family tree. It is a loanword from Dravidian, reflecting a deep history of trade and agricultural transfer.

The Etymology of Cotton: From South India to the Mediterranean

To trace the word karpāsa, one must look to the reconstructed roots of Proto-Dravidian (the ancestral language from which Tamil, Kannada, Telugu, and Malayalam descended). Philologists link the word to Proto-Dravidian roots associated with spinning, bending, or the cotton plant itself, such as the Tamil karppai (cotton or grass used for sacrificial purposes) and the Kannada karpa or kaṟpa (cotton).

Unlike primary Indo-European nouns, which can be broken down into verbal roots and suffixes native to Sanskrit, karpāsa has no internal Indo-Aryan etymology. It appears in the Sanskrit lexicon as a fully formed borrowing. Linguists argue that the word entered Sanskrit as nomadic Indo-Aryan pastoralists encountered established agricultural communities who were already cultivating, spinning, and trading the fiber. The term traveled along trade routes, carrying the subcontinental product to the Near East and the Mediterranean, where it became Hebrew karpas (fine cotton or linen, as seen in the Book of Esther) and Greek karpos (fine fabric).

This linguistic transition mirrors a major shift in trade and agricultural technology, as explored in our companion study on the material history of early Indian textiles.

The Archaeological Gap: When Cotton Met Sanskrit

To understand why the Ṛgveda is silent on cotton while later texts are not, we must look to physical archaeological evidence.

Archaeological evidence shows that cotton (Gossypium arboreum) was cultivated in the Indus Valley as early as the Harappan period. Archaeologists have recovered preserved cotton threads from silver vessels at Mohenjo-daro, dating back to roughly 2500 BCE.

However, tradition holds that the Ṛgveda was composed in the post-Harappan northwest, a period of transition. Scholars debate the exact timeline, but the text's composition is generally placed between 1500 and 1200 BCE. During this period, the center of cotton cultivation had shifted. As the urban centers of the Indus Valley collapsed, cotton agriculture persisted in Gujarat and southern India.

When nomadic Indo-Aryan speakers entered the northwest, they entered an environment where cotton was no longer the dominant regional crop. It was only as these communities migrated further east into the Ganga-Yamuna Doab (the fertile land between the Ganga and Yamuna rivers) and established trade links with southern and western merchants that they re-encountered the textile. When they did, they did not invent a new Sanskrit word; they borrowed karpāsa from the Dravidian-speaking merchants who controlled the trade.

Seven Dravidian Words inside the Rigveda

While cotton appears in later Sanskrit, several other Dravidian loanwords made their way directly into the earliest layers of the Ṛgveda. Sanskrit is an Indo-European language, built from roots you can usually trace back to a reconstructed ancestor shared with Greek, Latin, and Persian. When a Rigvedic word has no such root, carries sounds foreign to Indo-European, and lines up with reconstructed Proto-Dravidian, philologists take notice. Here are seven:

• Mayūra (मयूर, mayūra, peacock), in Ṛgveda 1.191.14. The peacock is native to South India and the lowlands, not the Central Asian steppe. Witzel notes mayūra has no Indo-European cognates but matches Proto-Dravidian *mayil (peacock), preserved in Tamil mayil and Kannada mayil.
• Khála (खल, khála, threshing floor), Ṛgveda 10.48.7. Matches Dravidian *kaḷ (open ground, threshing place), surviving in Tamil kaḷam (place, threshing floor) and Kannada kaḷa. Indo-Aryan speakers adopted this local farming term along with local agricultural techniques.
• Ulūkhala (उलूखल, ulūkhala, mortar), Ṛgveda 1.28.6, used for pressing the ritual plant soma. Its second element, khala, traces to the same Dravidian "grinding place" root, fused to an Indo-Aryan prefix.
• Kuṇḍa (कुण्ड, kuṇḍa, pit or round vessel), Ṛgveda 8.17.13. This matches Proto-Dravidian *kuṇṭ-V (well, hollow), seen in Tamil kuṇṭu (hollow, pool) and Telugu kuṇḍamu (pot).
• Muñja (मुञ्ज, muñja, sedge grass, Saccharum munja), Ṛgveda 1.191.3. This reed, used for weaving baskets, ropes, and priestly girdles, matches Tamil muñci (sedge grass).
• Taṇḍula (तण्डुल, taṇḍula, rice grain), appearing in later Vedic texts. Scholars trace this to a Dravidian root for husked grain rather than any Indo-European source. Rice was a subcontinental wet-crop that incoming pastoralists had no native word for.
• Kāṇa (काण, kāṇa, one-eyed), Ṛgveda 10.155.1. Scholars link this to the Dravidian *kaṇ (eye), a rare case of basic anatomical vocabulary crossing over.

Sounds, Grammar, and a Contested Outlier

This contact went past vocabulary. Retroflex consonants (ट, ठ, ड, ण—made by curling the tongue back against the hard palate) are rare in Indo-European languages generally but common in Sanskrit from its earliest hymns onward. Scholars attribute this structural shift to Dravidian influence.

Furthermore, Vedic Sanskrit's gerund (a verbal form used to chain consecutive actions) functions in a manner identical to Dravidian gerunds, unlike its sister Indo-Iranian language, Avestan. Syntax and phonology do not cross between languages casually; they require deep, bilingual domestic contact.

Then there's Brahui, a Dravidian language still spoken in Balochistan, Pakistan, far from every other Dravidian-speaking region. Some scholars read Brahui as a leftover population, proof Dravidian speech once stretched across the northwest. Others argue Brahui speakers arrived there later, in the medieval period. That argument is unresolved. But the loanwords sitting inside the Rigveda's own hymns do not depend on Brahui to make their case—they are a separate, earlier line of evidence for the same conclusion.

Read together, these words are not trivia. They are the residue of two peoples sharing fields, tools, and language on the same soil, three thousand years before either side wrote any of it down as history.