Word 2004’s French grammar checker: Beyond useless

Posted by Pierre Igot in: Macintosh
September 17th, 2004 • 5:14 am

Today, I just finished proof-reading a very long French translation (80,000 words) on which I had been working for quite a while. I had re-read the document several times, so I was pretty confident that there were very few errors left.

Still, just for fun, I thought I’d give Word 2004’s French grammar checker a chance to scan through the text and flag what it thought were grammatical errors. I am human, after all, and I’ve probably missed a handful of mistakes such as verbal or adjectival agreements, etc.

I knew that Word’s grammar checker was bad, but I hadn’t really tried it in years. Well, I must said that its degree of failure exceeded all my expectations.

I only did about 10 pages before I gave up. In these 10 pages, Word’s grammar checker found a myriad of errors. (I didn’t count, but every paragraph had at least 4 or 5 of them.) The trouble is that not a single one of these “errors” were actually errors. All the things that Word flagged as errors were actually correct, and all the suggestions that Word made were errors — and blatant ones at that: wrong agreements, wrong structures, etc. Not a single suggestion was even only slightly incorrect. They were all bad, very bad mistakes themselves.

As I said, I knew it was useless, but not to such a degree. I thought that it would at least find one real error here and there, and that it would display a lesser degree of self-confidence in its suggested “corrections”. But no — this thing is such pure crap that it should actually be illegal.

And then there was one particular “error” flagged by Word’s French grammar checker that was actually a epiphany for me. See, I used to be a French teacher in university here in Nova Scotia, and we had a lot of English-speaking students whose command of French was far from adequate. I would spend my time correcting all kinds of basic grammatical errors in papers. And in many of these papers, there was one particular error whose origin I couldn’t quite fathom. Quite often, when the grammatical structure of the French sentence required a relative pronoun such as qui or que, my students would use dont, which is more or less the French equivalent of whose.

I never figured out where this was coming from. I couldn’t imagine anyone teaching French, even with very limited qualifications (and God knows there are many French teachers in this part of the world who are not really qualified to teach French), teaching students that they had to use dont in such a context. I mean, it’s not exactly easy to confuse which with whose, is it?

And then today I found out why. Replacing qui or que with dont is something that Word’s French grammar checker suggested to me several times while I tried to use it with my French translation — and every time it was wrong, of course.

I am quite convinced now that what happened is that these poor university students actually used Microsoft Word to proof-read their papers, and followed some of its recommendations.

It’s a scary thought — and a good example of how evil Microsoft’s software actually is. Because it’s software, and it’s so self-assured in its recommendations, there are some people who actually take them seriously. Ouch.

As I said, I seriously believe such software should be banned. It’s more than useless. It’s actually dangerous. Some people obviously think that what a software grammar checker says is as reliable as what a grammar book says. Dear oh dear.


Comments are closed.

Leave a Reply

Comments are closed.