sanskrit transliteration at a glance

This chart shows the letters of the Devanagari script used to write Sanskrit, their unicode values, and their equivalents in three of the most commonly used transliteration schemes.

vowels devanagari unicode (dependent) unicode iast harvard kyoto itrans	अ U+0905 a a a	आ U+0906 ा U+093E ā A A,aa	इ U+0907 ि U+093F i i i	ई U+0908 ी U+0940 ī I I,ii	उ U+0909 ु U+0941 u u u	ऊ U+090A ू U+0942 ū U U,uu
vowels devanagari unicode (dependent) unicode iast harvard kyoto itrans	ऋ U+090B ृ U+0943 ṛ R RRi,R^i	ऌ U+090C ॄ U+0944 ḷ lR LLi ,L^i	ए U+090F े U+0947 e e e	ऐ U+0910 ै U+0948 ai ai ai	ओ U+0913 ो U+094B o o o	औ U+0914 ौ U+094C au au au
gutturals devanagari unicode iast harvard kyoto itrans	क U+0915 ka ka ka	ख U+0916 kha kha kha	ग U+0917 ga ga ga	घ U+0918 gha gha gha	ङ U+0919 ṅa Ga ~Na
palatals devanagari unicode iast harvard kyoto itrans	च U+091A ca ca cha	छ U+091B cha cha Cha	ज U+091C ja ja ja	झ U+091D jha jha jha	ञ U+091E ña Ja ~na, JNa
retroflexes devanagari unicode iast harvard kyoto itrans	ट U+091F ṭa Ta Ta	ठ U+0920 ṭha Tha Tha	ड U+0921 ḍa Da Da	ढ U+0922 ḍha Dha Dha	ण U+0923 ṇa Na Na
dentals devanagari unicode iast harvard kyoto itrans	त U+0924 ta ta ta	थ U+0925 tha tha tha	द U+0926 da da da	ध U+0927 dha dha dha	न U+0928 na na na
labials devanagari unicode iast harvard kyoto itrans	प U+092A pa pa pa	फ U+092B pha pha pha	ब U+092C ba ba ba	भ U+092D bha bha bha	म U+092E ma ma ma
semi-vowels devanagari unicode iast harvard kyoto itrans	य U+092F ya ya ya	र U+0930 ra ra ra	ल U+0932 la la la	व U+0935 va va va,wa
sibilants devanagari unicode iast harvard kyoto itrans	श U+0936 śa za sha	ष U+0937 ṣa Sa Sha,shha	स U+0938 sa sa sa
(miscellaneous) devanagari unicode iast harvard kyoto itrans	ह U+0939 ha ha ha	ं U+0902 ṃ M M,.n,.m	ः U+0903 ḥ H H,.h	ॐ U+0950 Oṃ oM OM, AUM
itrans irregular combinations regular irregular	क्ष kSha xa	ज्ञ j~na GYa,dnya

notes

Transliteration is just a simple matter of looking up a character in one alphabet and substituting a character from another, right? If only.

The ideal transliteration scheme would …

be reversible. You could convert a text in and out (in either direction) and end up with the same text you started with.
be be easy for non-specialists to read.
map one-to-one between characters in the two scripts.
be easy to type on a normal keyboard.

Let’s see how the commonly used schemes for Sanskrit stack up:

	IAST	ITRANS	HK	latin
Reversible	10	7	10	0
Legible	7	6	5	10
One-to-one	8	6	7	6
Normal keyboard	0	9	10	10
Total	25	28	32	26

IAST

yogaścittavṛttinirodhaḥ

The IAST (International Alphabet for Sanskrit Transliteration) is the script with all the dots commonly seen in western academic writings. The trouble is, some of those dotted characters exist only in IAST and aren’t used in any normal writing system, so there’s no standard keyboard in the world that can type them. IAST was invented in 1912, and was relevant in the days of limited typesetting capabilities, but it’s hard to see what it has to offer in the all-unicode 21st century. Specialists should be able to read and print the real thing just as easily as some invented specialist alphabet.

plain latin letters

yogash chitta-vritti-nirodhah

For the non-specialist western reader, all the dots and accents in IAST are just offputting and confusing. For something that's intended for non-specialist western readers, the best option is to just use the nearest equivalent latin letters. Most people who aren’t native speakers of Indian languages can’t hear or pronounce the difference between dental and retroflex Ts and Ds anyway.

You can forget converting back from pure latin texts.

“ASANA” is the commonly used latin spelling for the word for “yoga posture” (it literally means “seat” or “seated position”). The corect sanskrit spelling is आसन. Five letters in latin, three in devanagari – how hard can that be? Well. Short “A” and long “A” are different devanagari characters. (So is “even shorter A” in some modern Indian languages, but if we confine ourselves to Sanskrit we don’t have to worry about that just yet.) There are three devanagari characters that roughly correspond to “S”, and four “N”-like sounds.

That, if I calculate correctly, gives us 96 possible Sanskrit renderings of “ASANA”. OK, we can look up the correct one in a dictionary. There are ways of coding that that aren’t as slow as it might sound, but you won’t like your bandwidth bills when people start downloading your “little” transliterator program that now includes a comprehensive Sanskrit dictionary. Get used to it, because it’s only going to get worse when you decide to start supporting Hindi, Marathi, Nepali …

And your troubles have only just begun. Supposing five of those 96 variant spellings turn out to be valid, different, Sanskrit words? In order to work out which one you meant, the program is going to have to start understanding what it probably means in the context of the words before it; and backtracking to review that in the light of the words that come after it. And when it can do that, congratulations, your simple little lookup table is about ready to sit the Turing Test.

ITRANS

yogashchittavR^ittinirodhaH

You can convert Devanagari to ITRANS and back no problem. You can go the other way, but because of the many character variations ITRANS allows there’s no guarantee that the ITRANS text you get out will be the same as the ITRANS text you put in.

You can type ITRANS on a normal western keyboard as long as you know how to get “~” and “^”.

Harvard Kyoto

yogazcittavRttinirodhaH

Harvard Kyoto uses only normal keyboard characters and is reversible both ways, but the heavy use of uppercase letters makes it look ugly and some of the phonetic mappings aren’t obvious.

References

sanskrit-sanscrito.com: comparing transliteration systems

ITRANS 5.3 official reference

ITRANS supplemented with older/irregular characters used in texts from the Sanskrit Documents List