Why in the world did we make a Malay grammar checker?

Years ago, back in 2008, we were asked the question: “Why isn’t there a grammar checker for Bahasa Melayu?”

At that time, English grammar checker had existed for quite a while, initially pioneered by Microsoft. The tool has been integrated in MS Office, constantly monitoring what users type and highlight possible grammatical mistakes with a green (later, blue) wavy line.

It’s not a catch-all solution, nor was it meant to be one. It is a safety net to help users avoid common grammatical mistakes that a spell checker cannot catch. For example:

Michael did not chose to go home.

“Chose” is correctly spelled, but it ought to be in present tense. A spell checker would completely miss that. That’s when a grammar checker can help.

“So, why don’t we have a grammar checker for Bahasa Melayu?” So we asked, too.

Nobody believed it was commercially viable. But the same was said about a Malay spell checker. We made that work. We didn’t wait for an answer. We wanted to be the answer ourselves. Our intention was quite simple, maybe even naïve — it’s pride.

Malaysians deserve to have their own proofing tools. And we don’t need to wait for foreigners to build it. If nobody else is willing to risk investing, we would do it. But we as a small company did not have the vast resources Microsoft had. How should we approach the problem?

It wasn’t entirely true that nobody ever worked on a Malay grammar checker. We had heard that something was being worked on in academia for years., but we didn’t have any access to details. Perhaps the researchers were aiming for a perfect system, but a perfect system in the lab isn’t going to benefit the mass consumers.

The first step towards developing our Malay grammar checker was to clearly define goals and constraints. We wanted a “useful” grammar checker that people could use there and then, not something perfect that is confined to the lab.

Then, we had to define what “useful” meant. To be useful, it should suffice that the grammar checker could detect most commonly-made mistakes, not all mistakes. Most importantly, the grammar checker must function in a word processor that everybody already uses, namely MS Word.

And then that was our mission for more than two years. We identified and researched about 60 types of common mistakes that we can reasonably work with. With no existing algorithm for reference, we started from scratch. We bulldozed our way through the massive work of tagging every Malay word. And we adapted our software to the MS Office API.

The only outside help we received was from MDeC (was MDC). They provided us with a grant to fund part of our development effort. We are forever grateful for the help. The result was Dewan Eja Pro, the first commercially available grammar checker for Bahasa Melayu. Because the approach was unique, we also patented the technology.

So, we did it. Till this day, it is still the only solution available for Bahasa Melayu. If you haven’t already, download a free trial.