Karnataka Samskrit University
Computational Linguistics Research

Sandhi Splitting
< सं-भाषा
This tool is derived from the Sandhi Splitter developed by the Samskrit Studies Department of the University of Hyderabad. Initially a simple port from Perl to Python, it is now a work-in-progress research project pending future improvement.
Please type in a Samskrit form to be split (in Devanagari or WX) and click "Submit":

Introduction to Sandhi

Every language has a set of sounds that is used to make words and sentences. Usually, the sounds are quite easy to pronounce, especially for native speakers. But even though the sounds might be easy to say when they're separate, it can be quite difficult to say some of them when they're put together, especially when speaking quickly! Sanskrit speakers faced these exact problems, and they did what people everywhere did with their own languages: in one place or another, they started to simplify their pronunciation. These simplifications did not happen everywhere, but they certainly did in ordinary speech. The early Sanskrit grammarians, trying to study their own language and preserve it for the future, gave a single name to the set of all of these changes: sandhi.

The word "sandhi," more properly written as saṃdhi, means "junction" or "combination." It refers to the "combination" of two sounds that sit next to each other. The word "sandhi" was borrowed into English, where it refers to the same sorts of changes in any language. So, we can talk about English sandhi, Chinese sandhi, Latin sandhi, and so on.

The concept of "sandhi" might seem strange to you. Fortunately, English shows some signs of sandhi rules. For example, consider three English words that are borrowed from Latin: "indirect," "impossible," and "illuminate." Each of these words starts with "in," but the "n" of this "in" has changed to more closely match the letter that follows it. Note that the pronunciation and spelling of these words have both changed. The same thing happens in Sanskrit, and if sandhi is applied in a text, we must write out all of the changes.

In the oldest parts of the Vedas, sandhi changes do not uniformly occur. That is, a change can appear once in one sentence and not at all in the next. But in the later form of the language — the form that we're studying today — the grammar of the language is more consistent. With the rare exception, all Sanskrit texts apply the sandhi rules.

Sandhi Examples

भव अर्जुन → भवार्जुन (bhava arjuna → bhavārjuna)
दग्ध अरण्य → दग्धारण्य (dagdha araṇya → dagdhāraṇya)
सु-उक्त → सूक्त (su-ukta → sūkta)
राज इन्द्रः → राजेन्द्रः (rāja indraḥ → rājendraḥ)
महा ऋषिः → महर्षिः (mahā ṛṣiḥ → maharṣiḥ)
हित उपदेशः → हितोपदेशः (hita upadeśaḥ → hitopadeśaḥ)
तस्य औदार्यम् → तस्यौदार्यम् (tasya audāryam → tasyaudāryam)
इति आह → इत्याह (iti āha → ityāha)
अपि अस्य → अप्यस्य (api asya → apyasya)
लेह् + सि → लेक् + सि → लेक्षि (leh + si → lek + si → lekṣi)
दह् + त → दग्ध (dah + ta → dagdha)
दिह् + ध्वे → दिग्ध्वे (dih + dhve → digdhve)
बुध् + त → बुद्ध (budh + ta → buddha)
रुन्ध् + थः → रुन्द्धः (rundh + thaḥ → runddhaḥ)
दुह् + त → दुग्ध (duh + ta → dugdha)


Karnataka Samskrit University
Pampa Mahakavi Road,
Chamarajpet, Bangalore – 560 018.