Weak, needs revision

People have long been fascinated with the idea of time travel and often speculate on the marvelous things that it would enable them to do. Here’s a reason we can be thankful that we cannot readily receive visitors from the past: some of English’s greatest writers, if they should drop in for a visit to the early 21st century, might be tempted in a moment of curiosity to submit their creations to one of the new-fangled online writing assistants and learn, to their deep chagrin, that they’d got it all wrong.

I tried this recently with some writers whose works are great and enduring monuments of English expression. It was surprising to see how poorly they fared with the modern, algorithmically-driven programs that chew your text up and spit it out with advice about how it might be improved. Take, for example, Abraham Lincoln’s Gettysburg Address. It is regarded as the greatest speech in American history and is one of the most studied and celebrated texts in the world. Not so, however, by the lights of Grammarly.com. Abe, if he’d been able to hop online while riding the train to Gettysburg in 1863 for a quick writing check-up, would have been confronted with many more “issues” than he would have had time to address; eleven, to be exact. The software makes a mad dash through text, looking for faulty parallelism, comma splices, squinty modifiers, split infinitives, and all sorts of lurking monsters that might have been unknown to the great statesman and orator. Both his writing style and his vocabulary use are called into question by Grammarly, though Mr. Lincoln might take comfort in the fact that the software generated four word choice corrections for his short text.

Emily Dickinson, a contemporary of Lincoln whose hundreds of poems are among the jewels of English, fares no better at Grammarly. If she had submitted her short and grammatically straightforward poem Safe in their Alabaster Chambers to Grammarly, she would learn that it had 13 critical writing issues, and that like Mr. Lincoln’s writing, hers was “weak, needs revision.” The good news for Emily about the poem is that Grammarly regards it as original. Though the website touts itself as a plagiarism checker in addition to its many other offerings, Dickinson’s oeuvre does not seem to be included in the eight billion documents that Grammarly checks your submissions against.

We may be thankful that Jane Austen was not able to take advantage of the services at Paperrater.com when she was writing Mansfield Park. We tried the first paragraph of Chapter 19 there to learn that Ms. Austen would be congratulated on her sophisticated word choice, but her overall “grade” was only a C. Did she use too many nominalizations? If you ask a computer program, the answer might be “yes.” At Polishmywriting.com, E. M. Forster would have been warned four times in the first paragraph of Chapter 14 alone of A Passage to India about his use of the passive voice. But this suggests that the esteemed author of this novel and many other masterpieces is middle-schooler who it is hoped can be scolded out of his bad writing habits.

Computer and web technology have made great contributions to language learning and study. The greatest successes at present tend to be more at the micro level than the macro level: learning about words online can be a rewarding and enriching experience, but learning about writing does not seem to be ready for its close-up when it is mediated by programs rather than by people. All of the automated websites that purport to help you with your writing seem to be based on the simplistic notion that good writing can be reduced to a set of rules; or in computer-speak, algorithms. But the idea that algorithms are now capable of evaluating the quality and accuracy of writing—especially when they are driven by old-fashioned ideas about what makes writing “good” or “bad”—does not hold up very well in practice. It’s weak; needs revision.