Skip to Content

Ask the editor 12: The limitations of editing software


The limitations of editing software

A new writer, setting out on that curious and occasionally perilous journey that, sometimes, ends in publication, needs help getting there; very few writers get it all right without advice or intervention. The internet has made the task of finding appropriate help and advice much easier; if your ivory tower is connected, that is. Recently, however, the limpid waters of literary support have been muddied somewhat by the advent of a plethora of AI tools. In this article, I'll consider the advantages - and limitations - of editing software.

Digital editing tools have, in fact, been around since Microsoft first added a spell and grammar checker to its Word suite. They can be pretty useful too; they certainly take some of the effort out of checking for spelling errors and grammatical howlers. Note that little qualification on my part; some, but not all, proofing and editing can be carried out by machine. I would argue that it's the bit that's left after ‘some' that's key to understanding the limitations of machine editing.

Word's checker may be the mother of machine editing tools but it's relatively crude. Complex grammar defeats it, and some fairly basic grammatical dilemmas (that or which, for instance) are as likely to produce howlers as correct them. One might assume that, with the new AI editing tools, such problems have been solved; one would, unfortunately, be wrong. While the latest editing apps are a step up from Word, they are not by any means infallible.

Part of the problem with digital editing pertains to its design. There are two basic resources: an ‘in-house' grammar containing rules and alternatives; and a broader approach, using statistical probability. In theory, that broader approach can use the entire internet as its resource. And that's a nice theory. In practice, using everything ever written is a case of wielding a double-edged sword.

The problem with this approach is simple to state: if you use statistical analysis to predict the correct word, phrase or grammatical point, you are as likely to get the most popular answer as the correct one. And there is no reason to suppose that the most popular answer is always correct. It may be, and in very simple cases, it probably is; but as soon as you move from simple to more complex statements, gaps, and elephant traps, can appear.

Think of it this way: you are using a grammar check because you are not totally confident in your own knowledge and expertise. You ask the app to check your grammar and it rushes off to look at everything ever written. It discovers that lots of people say ‘that' precisely where you are saying ‘which'. If all of those people are grammar mavens, all is well and your problem is solved.

But how likely is it that everyone who wrote the words ‘that' or ‘which' knew exactly what they were doing? The chances are that most of them are in the same position as you; they are a little uncertain about the intricacies involved in using the words correctly. The resulting suggestion, then, may be a victim, simultaneously, of popularity and uncertainty. To quote the late, great H L Mencken, ‘For every problem, there is a solution that is simple, popular and wrong.'

And if this is the case for fairly basic grammatical elements, what happens when you look for answers to more complex questions? Here, I think, the waters are very muddy indeed. The market is currently saturated with new apps that claim to be able to offer the right word, the perfect metaphor, the telling phrase; in effect, to do some of the writing for you.

These apps (often called ‘paraphrasing tools') are like auto-correct on steroids. I imagine that some, or maybe most, of you have encountered comically dubious suggestions from auto-correction apps. You make a mistake while typing the word ‘stretch' and, before you get to the ‘h', the auto-correct app has offered ‘scratch'. If you're not paying attention, your prose will soon be littered with literals: words that are perfectly correct in themselves but completely wrong in the context. One might suspect that auto-correct has replaced ‘grammar' with ‘grimoire'.

This phenomenon - occasionally hilarious and always irritating - is especially noticeable when people use an online thesaurus. If you have ever used a traditional thesaurus (and lots of wonderful writers have done so; Sylvia Plath kept one on her writing desk) you will know that it's not just a question of entering a term and waiting for an alternative to pop up. You need to check the suggested word against a dictionary definition to make sure it fits the context. If you are writing literary fiction or poetry, you also need to check that the word fits the music of the line or the structure of the sentence; you must, ultimately, employ human intelligence and wisdom; and, hopefully, a little creativity.

If you have read any AI-generated text, you may have noticed that there is something homogeneous, even bland, about the tone and style of it. This is due to the statistical probability problem too; the software inevitably tends towards the lowest common denominator. That's to say, the text is composed of the words and phrases used most often in the particular context; it is, effectively, a smorgasbord of clichés. If you are knocking out an instruction manual for a kettle, that's just fine; if you are attempting to encapsulate a delicate emotional situation, not so much.

There is, and I am quite confident there always will be, a significant gap between what a machine can offer to the writer, in terms of resources and support, and what a capable writer can do for themselves; or what a writer and editor, working together, can achieve. And that gap, dear reader, is the place where real writing happens.


You can see our wide range of editorial services here

If you have any queries or suggestions for our new series, Ask the Editor, please email us.

When he isn't editing, Noel Rooney writes a regular column for Fortean Times magazine, and wilfully obscure poetry. He lives in South London with his family and rather too many animals.

Ask the Editor 1: What genre is my book?

Ask the Editor 2: the submission letter

Ask the Editor 3: Writing a synopsis

Ask the Editor 4: Why do I need you?

Ask the Editor 5: Non-fiction submissions

Ask the Editor 6: Writing non-fiction

Ask the Editor 7: Researching for a book

Ask the Editor 8: How I assess a manuscript

Ask the Editor 9: Why do I need a report?

Ask the Editor 10: Writing your blurb or cover copy

Ask the Editor 11: English Language Editing

Which report?

Copy editing services

The Pedant series

An Editor's Advice series