Com-proofer?

Don’t believe the hype, as man battles machine

bookediting.co.uk editor Lee Dickinson

By Lee Dickinson

Man versus machine. It’s the plot for some of Hollywood’s best blockbusters, and now online language checker Grammarly threatens to be a proofreader terminator.

If you believe the hype.

Millions do, according to figures from the program’s marketers. They trust it to do everything from “Increase Team Productivity” to “Write better sentences” and “Save Time”.

Wow. Who doesn’t want to save time? And if it helps write better sentences, I’m in.

I’m suspicious though, not least because all those claims are in a three-line web blurb. The one from a Google search. This one:

“Increase Your Team’s Productivity. Eliminate Errors, Add Clarity, And Save Time. Eliminate grammar errors. Easily improve any text. Fix punctuation errors. Write better sentences. Improve word choice. Services: Advanced Grammar Rules, Proofreader, Corrects Writing Mistakes.”

So in arguably their most important marketing material, the first words you’ll see in a search engine, they have mistakes? Lots of them.

See how all the first letters of words up to the second “Eliminate” are capped up? After that they’re lower case except at the start of a sentence, before reverting to caps again for the parting shot of “Corrects Writing Mistakes”. Clearly, one mistake it doesn’t correct is inconsistency.

Repetition, neither, judging by the three uses of “Errors” (or “errors”) and two for “Eliminate”. The latter word also makes me doubtful it does “Improve word choice”, because that’s an awful, four-syllable mouthful. Good language simplifies. Using ‘removes’ or ‘amends’ would’ve done that without repeating “Fix” or “Corrects”. A good proofer changes that.

Anyone can make a mistake. Maybe the software is better than the flawed words promoting it? So I delved deeper.

Cards on the table here, because I’m a proofreader and editor. You’d no more expect me to think a computer can do my job than you’d expect a Paxo-carrying turkey to don a paper hat and cavort around your Christmas dinner table. But I’m a fan of automated language programs. Check the last point in my 10 Better Writing Tips. Spellcheckers are useful.

When you use one, though, know its limits. It’s a tool, not a cure, spotting some mistakes and missing many others. Sometimes it’ll even flag up perfect writing as wrong.

One thing it will never do is spot the worst type of howler. It will never go the ‘extra mile’ with the sort of fact and word checking done by top-level proofers.

To prove the point, I ran my last job through it. The client is a PhD-level student whose work reflects the fact English is his second language. For example, the vital, bolded-up second word of his dissertation was wrong, because he’d written ‘The Conception of (his subject)’. Reading the piece and cross-checking it the way Grammarly never can, it was clear he meant ‘concept’, because no-one was pregnant and none of his work covered the ‘forming or devising’ of anything.

The document continued like this, with sources’ names misspelt, ‘employment’ used instead of ‘unemployment’, missing words altering meanings, the wrong years applied to sources, and the International Monetary Fund labelled the ‘International Monterey Fund’. No problem; that’s why people use proofers.

How many of these major mistakes did I find in his 12,500-word submission? Twenty-one. Grammarly? Three. So that’s at least 18 howlers staying in a vital piece of coursework.

Others have checked Grammarly. A 2017 probe by language site Grammarist gauged its ability to spot everything from plagiarism to sentence structure. Of the 43 errors, Grammarly found 31.

Writing skills firm Emphasis concluded it’s “far from a substitute for the human eye”, with basic mistakes approved. The phrase “plain English” was challenged as “too generic” and “give your writing colour” rejected for “make your hand richness”.

Hundreds of customers have road-tested it. Trustpilot’s rating? Two stars out of five, or 3.2 out of 10. If you prefer sitejabber, that’s also two out of five. A total of 136 of the 206 reviews give it one star, and the comments are damning. Yes, angry customers are more likely to review than happy ones, but the level of spleen-venting is shocking. Check for yourself here or here.

The headline on one of those reviews concludes: “don’t believe the hype”.

I won’t. Nor will I entrust words to a computer which can never fully check them or appreciate the beauty of their subtleties.

This article’s writer, Lee Dickinson, is an advanced professional member of the Chartered Institute of Editing and Proofreading and chief editor at bookediting.co.uk. You can read his blog here.