Hindi Computing: Unicode, InScript Keyboard and Fonts

A computer stores every letter as a number. English needed only a few dozen numbers, so the early codes were small; Devanagari, with its vowel signs, conjuncts and nukta, needed a far larger and better organised code. Hindi computing is the story of how those codes, keyboards and fonts were standardised, and it is a favourite area for computer awareness papers in SSC, Railway and banking exams, as well as for the Hindi typing tests that many state recruitments hold. Three ideas must be kept apart: the encoding, which is the number given to a character; the keyboard layout, which decides what key produces that character; and the font, which only draws the shape on screen.

From ASCII to ISCII

ASCII, the American Standard Code for Information Interchange, is a 7-bit code and therefore has 128 code positions, enough for the English letters, digits, punctuation and control characters. Extended ASCII used the eighth bit to reach 256 positions, which various makers filled with their own symbols. For Indian scripts the Bureau of Indian Standards adopted ISCII, the Indian Script Code for Information Interchange, as IS 13194 in 1991. ISCII is an 8-bit code that keeps the lower 128 positions identical to ASCII and puts an Indian script in the upper 128. Its clever feature is that a single code table serves ten Indian scripts, because those scripts share a common phonetic order descended from Brahmi; the same code point is read as a Devanagari letter or a Bengali or Tamil letter depending on the script in use. ISCII is still the base of much older government software but has been overtaken by Unicode.

Unicode and the Devanagari block

Unicode is the universal character encoding standard maintained by the Unicode Consortium. It gives every character of every script a unique number called a code point, written as U+ followed by hexadecimal digits. The code space runs from U+0000 to U+10FFFF, that is 17 planes of 65,536 code points each; plane zero is the Basic Multilingual Plane, where almost all living scripts sit. The Devanagari block occupies U+0900 to U+097F, and later additions sit in a Devanagari Extended block. Devanagari digits, the danda and double danda, the virama and the nukta all have their own code points in this range.

Unicode numbers must be stored as bytes, and that is done by an encoding form. UTF-8 is variable width, using one to four bytes; it keeps plain English text byte-identical to ASCII, which is why it took over the web, and it uses three bytes for a typical Devanagari character. UTF-16 uses two bytes for the Basic Multilingual Plane and four for the rest, while UTF-32 uses four bytes for every character.

Keyboard layouts

The InScript layout, short for Indian Script, is the standard keyboard layout for Indian languages notified by the Government of India, and it is built into Windows, Linux, macOS and Android. Its logic is phonetic and uniform: the vowels and their matras are placed on the left hand and the consonants on the right, in the dictionary order of the script, so one layout serves Devanagari, Bengali, Tamil, Telugu and the rest, and a typist who learns Hindi InScript can type Marathi or Nepali without relearning the keys.

Against it stands the Remington layout, which copies the arrangement of the old mechanical Hindi typewriter and is what most professional Hindi typists in offices and courts were trained on. Its variants, Remington GAIL and Remington CBI, are still offered in government typing tests, where candidates usually choose between Remington with a legacy font and InScript with Unicode Mangal. A third way to type is phonetic or transliteration input, in which the typist writes bharat in Roman letters and the software converts it to the Devanagari spelling; this is the easiest for a beginner and the slowest for a professional.

Fonts: Unicode against legacy

A font does not store text, it only draws it. Unicode fonts such as Mangal, Nirmala UI, Aparajita and Kokila on Windows, and Lohit Devanagari and Noto Sans Devanagari in the open source world, map the real Devanagari code points to glyphs. Text typed in them can be searched, sorted, copied into e-mail and read on any device. Legacy or non-Unicode fonts such as Kruti Dev, DevLys and Chanakya take a different route: they paste Devanagari glyphs on top of ASCII code positions, so the computer thinks the text is English while the eye sees Hindi. Open such a file without the matching font and it becomes unreadable symbols, and no search engine can index it. That is why government work has moved to Unicode, with converters available to turn old Kruti Dev material into Unicode. Because Devanagari joins letters into conjuncts, Indian language fonts must be OpenType fonts, which carry the substitution and positioning rules that a plain TrueType font cannot.

Government support

The Technology Development for Indian Languages (TDIL) programme of the Ministry of Electronics and Information Technology develops and distributes fonts, dictionaries, speech and translation tools for Indian languages. The Centre for Development of Advanced Computing (C-DAC), Pune built the GIST range of Indian language technology and the LEAP Office suite, and remains a key centre for script software. Article 343 of the Constitution makes Hindi in the Devanagari script the official language of the Union, which is the legal reason government computing must support Devanagari properly.