Hindi/Devanagari Vowel Marks & Signs Cheatsheet

Devanagari is an abugida, which means each consonant carries an inherent a vowel by default. क is not just k, it is ka. To write a different vowel you attach a matra — a dependent vowel sign — to the consonant, and to write no vowel at all you add the virama. This is the fundamental difference from an alphabet, and it explains why this table is mostly vowel signs rather than vowels.

The matras attach in different positions around the consonant, and that position is purely visual convention rather than storage order. The aa matra ा and the o and au matras go to the right, the u and uu matras ु and ू go below, the e and ai matras े and ै go above, and the i matra ि goes to the left — but it is typed and stored after the consonant even though it displays before it. This is why cursor movement in Devanagari can feel unpredictable and why a character count rarely matches what you see. It is correct behaviour, not a bug.

The virama, called halant in Hindi, removes the inherent vowel and is what makes consonant clusters possible. When two consonants meet with a halant between them, most fonts render them as a single conjunct ligature rather than two letters with a visible mark, so क plus halant plus ष becomes क्ष. Hindi has hundreds of these conjuncts. If you see stray halant marks where you expect conjuncts, the font lacks the ligatures rather than the text being wrong.

The two nasal marks are frequently confused. Anusvara ं is a dot on the line that nasalises by adding a nasal consonant matching the following sound, as in हिंदी. Candrabindu ँ, the moon with a dot, marks pure nasalisation of the vowel itself, as in हँसना. In practice modern Hindi has drifted towards using anusvara for both, and many writers no longer distinguish them, but careful writing and most textbooks keep the difference. Visarga ः, a colon-like mark, appears mainly in Sanskrit loanwords.

Nukta ़ is a dot placed below a consonant to make sounds borrowed from Persian and Arabic — क़, ख़, ग़, ज़, फ़. Like the voiced kana in Japanese, these exist both as precomposed characters and as a base letter plus a combining nukta, so the same word can have two byte representations that look identical. NFC normalisation resolves it, and it is worth doing on any Hindi or Urdu-influenced dataset. The danda । and double danda ॥ are the Devanagari full stop and its verse-ending form; a Latin period is common in modern informal Hindi but the danda is standard. For typing, InScript is the government standard layout and places matras logically; phonetic layouts mapping Hindi sounds to QWERTY positions are more popular with casual typists.