Wanting is an efficient Bangla keyboard layout

by Abu Jar M Akkas | Published: 00:05, Feb 23,2018 | Updated: 21:30, Feb 22,2018


Munier-Optima Bangla keyboard layout that is said to be based on letter frequency analysis.

AKEYBOARD layout is not a mere interface between users and computers. It is quite a religion in some cases as keyboarding habit dies hard because of high training and touch-typing that is ingrained in muscle memory.
Conforming to any such standard makes typing easy for all, across places and across computers — typewriters have been long dead. When typewriters came, people felt the need for a keyboard layout, an interface to the machine that is, for Bangla.
There had been one, broached by Remington, a firearms manufacturer that produced typewriters. It was in 1940. There could be few others, too. But none gained grounds as whatever the Remington layout did.
A proper Bangla keyboard layout came into being in 1965, both in Bangladesh and West Bengal separately. In Bangladesh, Munier-Optima Typewriter came out with a Bangla keyboard layout, often referred to as scientifically analysed on letter frequency counts.
Godrej and Boyce, which started manufacturing typewriters in the 1950s, marketed Bangla typewriters, with the Munier-Optima layout after 1971 under the brand name of Prima.
And they, especially the Munier-Optima layout in Bangladesh, catered to the demand for quite a long time until the Bangla computing entered the scene. The Munier-Optima layout is said to have been designed based on letter frequency — the letters that are most used in Bangla text — assigned to the keys in accordance with the fingers that can access them easily and without being tired.
The most used letter would remain close to the centre of the keyboard, preferably without the need for the shift key, that index fingers, usually the most capable ones in human hands, could press them without having, for the finger, to travel much distance on the keyboard.
In the early 1990s, the floodgate opened with the Bijoy layout driving away the Shahid Lipi layout. The Shahid Lipi layout, which was designed for Mac, was cumbersome as users needed to press various combinations of keys — not just the shift key but also more than two at the same time — to access conjunct letters.
At the turn of the 1990s, some started developing programs to write Bangla on disk operating system, or DOS, which was a command-line affair as what programs such as Word Star or Word Perfect were. Yet, they provided for a means to write Bangla on computers at least for office automation purposes. A few of the programs started coming out both on DOS and Windows, with graphical interface, and better typeface that could be used in publication.
In the early 1990s, many add-ons, which were not fully-fledged programs but were sets of an input mechanism and font files, started burgeoning. By the middle of the 1990s, the were several programs and add-ons, with font files, offering an array of keyboard layouts for users. In the absence of nothing uniform or standard, developers started shipping layouts according to their taste.
While some of the layouts more or less matched that of the typewriter and the Bijoy layout, some were just QWERTY-mapped or -mnemonic layout, also known as ‘phonetic’. Similar sounding letters were mapped onto QWERTY layout so that the learning curve remains very low and users can get going after fiddling for a while.
The Unicode Standard came into use after a prolonged opposition and a lack of understanding about the system by many of the Bangla computing program proponents. Some of the oppositions lied even in business interest. The developer or the marketing company of a program, which earlier used the legacy layout of the font file, often called ASCII or ANSI mapping, with arbitrary orders of the letters, was for certain to lose its market share if Unicode would come with a force in use.
The old layouts, with minor modifications, were continued for Unicode fonts. It made the situation complex. In the legacy system, conjunct letters were composed of components and users need to access those components directly or by way of a circumventing mechanism. Besides, in the old system, certain vowel markers needed to be typed before the consonants as the written form demands. But Unicode stipulates all vowel markers after the consonants, as the computer stores the text, and all conjunct letters are more than two, three or even four consonants with invocalisers — virama or hasanta which kills the inherent first vowel associated with every consonants — in between.
As Unicode defines the font file and how the text, stored on the computer, is rendered on the screen keeping to the font file and does not define the text input mechanism, many of the keyboard layouts had to do some retrofitting. A letter here and another there were modified. While none of the keyboard layouts out there that time were designed keeping to the letter frequency to make them efficient, the retrofitting of the typewriter or legacy layouts to match the Unicode system made the layouts less inefficient.
In the process, a few programs or add-ons appeared on the computing scene which used transliteration keyboards of a sort, following heuristic algorithm. In the system, users need to spell the word using Roman letters and the program tries to bring up all possible Bangla words for users to choose from. While heuristic algorithm works well, it does not guarantee that the best solutions will always be found. It finds a choice close to the best one but it works fast and easily. This makes typing slow, slower than what fixed layout, in which users know that certain keys will call certain letters, could so far offer. It slows down typing in that in fixed layout, which were largely inefficient as they were not designed based on letter frequency, the assignment of each of the keys, yet in programs that use heuristic algorithm or run a dictionary in the background to find the choices, users often need to press more than one keys to call the letter intended.
In the history of Bangla computing in Bangladesh, users came to receive a number of keyboard layout down the journey. It began with Shahid Lipi and, almost, ended with fixed layout such as Bijoy (Unicode), which has still been prominent in the publishing industry because of its huge number of users because of high training in the arena, Probhat and Inscript, which is the national standard in India and was later endorsed by the Microsoft Corporation, and heuristic programs such as Avro which gained popularity with people who think that there is no reason for them to climb a high learning curve. People had a national standard layout called National Keyboard, designed by the Bangladesh Computer Council in 1993. But it hardly could gain currency among ordinary people.
All the fixed keyboard layouts that the Bengalis so far have had both in Bangladesh and West Bengal are hardly designed to be efficient. They are focused more on the ease of use in terms of knowing how to type, with low learning curves. This leaves the Bangla computing scene still wanting an efficient keyboard layout, fixed, tailored to the way Unicode input mechanism demands. And in a situation like this where a good number of layouts share the market, more or less, people are baffled about what to use.
This has also caused great inconveniences to people in not being able to type Bangla text using the same layout everywhere, a fallout of the lack of a standard. Although a handful of people, probably not in Bangladesh, may use other variant layouts such as Dvorak to type English or Roman text, and most people can interchange their devices to type text the QWERTY layout which is almost universal, in the case of Bangla computing, this has so far remained a problem.
It is neither that private entities should come up with a good, efficient fixed layout nor government agencies responsible for such jobs should do this. But what people need is an efficient fixed layout, arrived at scientifically through the research needed. And they need to have it.
In an ideal case, keyboards should be efficient and optimised for speed. This can only be done by considering letter frequency, as is required in Unicode input mechanism, bigram, which is the frequency of two consecutive letters so that the same finger does not need to make two presses consecutively, and even trigram, which is the frequency of three consecutive letters.
Letter frequency makes the movement of fingers easy and brings in the speed. An analysis of bigrams and trigrams helps not to make fingers tired. These are all meant to stop the fingers from travelling too much distance while they make key-presses. And users must know which key-press calls up which letter so that they do not need to select one from many choices.
Software developers have always tried to grab the market, with whatever they could roll out in the initial days of Bangla computing. Later they may have tried to keep whatever of the market share they could grab. Speed, an ease of use in terms of finger movement and efficiency are what people should need.
There is no royal road to learning, not even in learning how to type, or touch-type, which is the proper way of typing. People would learn how to type with an efficient keyboard layout as they do in learning how to type on a QWERTY layout. There could be easy ways to type with QWERTY-mnemonic layouts or transliteration layouts for hobbyists and enthusiasts, but there must be efficient, fixed layout for professionals or people who do the typing job professionally so that they can type at speed and ease. With the electronic text corpora for the Bangla language, it is not a very difficult task. The government can do it. Or any private entity, any of the universities or institutions can do it. But do they it must.

Abu Jar M Akkas is deputy editor at New Age.

More about:

Want stories like this in your inbox?

Sign up to exclusive daily email