Why Computers Shouldn’t Grade Writing Assignments
MS Word’s grammar checker isn’t always right, so why trust a computer program to grade writing assignments? For understaffed and cash-strapped schools or schools that want to use the newest technology, computerized grading may seem like an idea worth considering. However, closer examination of such software shows that it undermines the purpose of writing and gives feedback inferior to that of teachers and professors.
About Computerized Grading
I learned about computerized grading by reading “Writing to a Machine is Not Writing At All” by Anne Herrington and Charles Moran. Programs such as ETS’s Criterion, EdX, and Vantage Learning’s My Access! allow students to submit their writing online to gain quick feedback with a score and comments on individual sentences based on writing standards of students’ grade level (2). The software assesses not only correctness of grammar and punctuation but also presence of structural elements, such as introductions and quality of evidence (6). On the surface, computerized grading may sound promising, but in reality this innovation is undesirable.
Undermining the Purpose of Writing
We write for an audience, even if that audience is just ourselves or an instructor. Learning to write to an audience is important. In the workplace, employees must write appropriately to coworkers, bosses, and customers. A machine is not an audience. A machine cannot react to, learn from, or take to heart ideas conveyed in writing. So, writing to a machine makes writing pointless.
Giving Inferior Feedback
In addition to being pointless, writing to a machine may also yield misleading feedback. When Herrington and Moran analyzed the feedback Criterion gave a sample essay, they found inaccuracies. For example, the program flagged e.g. as a misspelling (7). Because the essay’s introduction was inductive (going from specific to general), the program couldn’t detect the presence of any introduction. In all, 95 percent of the mistakes the program flagged in development/organization, grammar, spelling, style, and mechanics weren’t mistakes (8). This inferior feedback is likely to harm rather than help writers.
Even if the accuracy of the feedback from computers has improved since 2010, when Herrington and Moran wrote their article, another problem remains. It seems that grading programs discourage deviations from formulaic kinds of writing, such as five-paragraph essays. Formulas are easier for computers to identify. Although formulaic writing may be effective when responding to standardized writing test prompts and when in middle school, good writers write in more sophisticated ways, especially in high school and college. Encouraging formulaic writing could be detrimental to writers in the end.
A third problem is that, as far as I know, computers don’t praise specific aspects of writing. This is problematic because positive feedback shows what students do well, helping them continue to use their strengths and encouraging efforts to become stronger writers.
I’m not against using technology in writing, just in grading it. Let’s not make writing assignments pointless and allow inaccurate and negative feedback to hinder writers’ growth. Computers can’t replace human readers in school effectively.
Work Cited: Herrington, Anne and Charles Moran. “Writing to a Machine is Not Writing At All.” Digital Is. September 24, 2010. PDF file.
Photo: Valley Library (Oregon State University), Flickr Creative Commons
About Darla Word
I'm a writing tutor and editor from Michigan. My favorite subject to write about is writing because making better writers is my calling. I also enjoy reading, singing, swimming, and cardmaking.