InScript: how one layout covers eight Indian scripts

One keyboard, eight scripts

Most keyboard layouts are an arrangement for one language. InScript is a rule for arranging many, and it is the reason a person who types Marathi can sit down at a Kannada keyboard and already know where half of it is.

The rule is this: the phonetically equivalent letter sits at the same position in every script InScript covers. The key that gives the ka sound in Devanagari gives the ka sound in Bengali, in Gujarati, in Odia, in Telugu, in Kannada and in Malayalam. Not a similar sound in a roughly similar place — the same position, by design and by standard.

It is easier to see than to describe. Below are four ordinary consonant keys, read out of the key maps this site teaches from, with what each one produces in eight scripts.

Eight different alphabets, four keys, and one row of hands that does not have to move.

Why one layout can cover eight alphabets at all

This works because the scripts are relatives rather than strangers. All eight descend from Brahmi, and they inherited more than their letter shapes from it: they share an inventory of sounds and, crucially, the traditional order those sounds are listed in — vowels first, then consonants grouped by where in the mouth they are made, from the back of the throat forward to the lips.

That order is not a convenience somebody imposed later. It is how these alphabets have been recited and taught for well over two thousand years, and it is why a letter in one of them has an unambiguous counterpart in the others. A layout built on that order is a layout that transfers.

Unicode then encoded the scripts in exactly that order, one block each and laid out in parallel, which has an unintended and very useful consequence: the claim is checkable. Subtract a script's block start from the character a key produces and you get an offset, and two keyboards agree about a letter when their offsets match. This site does that arithmetic against Devanagari — the Tamil guide carries the table, because Tamil is the interesting exception to it.

The short version of that table is that the agreement is close to total. Where these layouts define the same key, they almost always put the same letter on it, and the handful of disagreements are letters one script has and another does not.

Vowels on the left, consonants on the right

The second thing worth knowing before you start is how the layout is organised, because it is organised, and that makes it far easier to hold in your head than thirty unrelated positions.

Every letter key under the right hand produces a consonant. There are fourteen of them and there are no exceptions. The left hand carries the vowel signs — the marks that attach to a consonant and change the vowel it carries — along with the virama that strips the vowel off entirely, and three consonants parked at the bottom.

So an ordinary syllable in these scripts, which is a consonant plus a vowel sign, alternates hands by construction. You are not chasing hand alternation the way a Latin layout has to; the writing system hands it to you.

Tamil 99 arrived at the same split independently, and so did Korean's Dubeolsik. It is not a coincidence so much as a consequence: when a writing system is built out of two alternating categories, separating those categories by hand is the obvious design, and it is only available to layouts whose language works that way.

There is one more piece of order in it. Where a consonant has an aspirated counterpart — the breathy version of the same sound — the two share a key, with the plain letter unshifted and the aspirated one under Shift. That halves the number of positions there are to learn, and it means a wrong guess usually lands on a letter that is at least related to the one you wanted.

What actually transfers, and what does not

If you already type one of these scripts, the honest accounting is that the keyboard transfers and the writing does not.

What transfers is the map: where your fingers go. That is most of what makes a new script slow, and getting it for free is a large head start — the second script in this family costs a fraction of what the first one did.

What does not transfer is everything downstream of the key. The letter shapes are different and have to be read. The conjunct forms — what happens when two consonants meet with no vowel between them — resolve differently from script to script, and Devanagari's are not Malayalam's. And the vocabulary is a different language's, which is the part that actually limits typing speed once the positions stop needing thought.

The practical advice follows from that. Do not relearn the layout; you have it. Spend the time on reading instead, because in this family the bottleneck moves from your hands to your eyes faster than people expect.

Where InScript stops

Two scripts on this site are written in India and Sri Lanka and are not part of this at all.

Tamil is typed on Tamil 99, designed for Tamil rather than adapted from the pan-Indic standard, and measured key by key it agrees with Devanagari almost nowhere. That is a design decision with a real argument behind it: Tamil has a much smaller consonant inventory than the Sanskrit-derived scripts, so a layout arranged around theirs is carrying keys Tamil does not need.

Sinhala uses Wijesekara, which predates InScript and is the standard in Sri Lanka. Same outcome, different history.

Both are worth knowing about before you assume the family covers everything south of the Himalayas. It covers a great deal, and it has two well-documented edges.