How Can Teachers and Machines Grade the Same?
A recent study comparing essay scoring by computers and by humans demonstrated that the machines were capable of producing pretty much the same scores as humans.
Walt Gardener adroitly notes that as a former English teacher, he was excited at first that computers might be able to relieve the heavy burden of essay grading borne by English teachers. Further study, however revealed that machines were incapable of judging content, focusing only on the mechanics of an essay.
So what does the study say about the way teachers grade essays if the results were the same for teachers and machines?
As a former English teacher myself, I have graded my share of essays. It is arduous work, but so is writing. Good writing demands careful and logical thought, supportive evidence for an argument, careful word choices, sound conclusions. Mechanics are important only insofar as they enhance or obfuscate the argument.
So if the machine scored as well as the teachers, I can only assume that the teachers too were grading on mechanics, not on thought. And that, my friends, is why many kids can learn to punctuate without learning how to write. (I’m talking basic punctuation here, not the use of apostrophes. Hardly anyone these days knows how to use apostrophes correctly.)
Think, for a moment, about how English teachers correct/grade essays in your own school. For years I’ve been on a one-woman crusade against assigning number grades to essays. Why? Because it makes it easy to focus only on the mechanics of an essay -- misspellings, fragments, lack of subject-verb agreement, misuse of punctuation. The essay can be docked a point for every error. It’s cut and dried. The problem, of course, is that any drivel that is correctly punctuated and spelled can get a perfect score. On the other hand, the kid who can think but can’t punctuate gets a 78. This scoring method makes it easy to explain the difference between a 76 and a 78 or between a 69 and a 70.
Some teachers give two grades, one for mechanics and one for content. To me, it’s like trying to grade the paint and brushes separately from the painting. Spelling, punctuation, word choice – all are simply tools to be used to enhance the meaning of the work. In addition, kids already have access to Microsoft editors when they write. So what they hand in should, for the most part, be error-free. Unless, of course, your students don’t have access to computers or you haven’t shown them how to use computers to enhance their writing. I suspect there are still teachers who make their students write out their essays on paper before they type them into the computer, still treating the computer like basically an upgraded typewriter.
Teachers who score written work using rubrics have a better chance of developing their students’ writing skills, especially if the rubric is shared before the work is begun, letting kids know beforehand what’s expected and what it takes to get a top grade. When rubrics are carefully written, they address content and mechanics, with the emphasis on content.
Writing is thought; oftentimes you don’t know what you really think about a subject until you sit down to write. Machines cannot evaluate a thoughtfully reasoned argument replete with supportive evidence like teachers can. So if machines and teachers came up with the same grades, teachers must be scoring like machines.