Writing for Computers, part 2

My previous post reflected on the perils of using computer-generated response to student writing too early in the writing process. In this post I’ll argue for the proposition that such programs encourage teachers to ignore good instructional practices while deluding them into believing they are adopting good instructionl practices.

For years teachers have been using The Writing Process: a step-by-step sequence of pre-writing, drafting, revising, editing, and publishing. Despite the nearly universal acceptance of that formula by teachers – visit any language arts classroom and the chances are good that you’ll see those stages displayed somewhere in the room–there are still a number of teachers who don’t follow it in their writing instruction. They assign writing, but they don’t teach writing. There are far more who do embrace The Writing Process but who do so in a formulaic way. “Today we will brainstorm; tomorrow we will write drafts. I’ll read our drafts overnight and hand them back to you tomorrow, and then you will revise them based on my comments and corrections. After that you will have someone check your paper for errors, and then we will tape then to the classrrom wall.” The problem with both approaches is, that’s not the way writing gets done.

Writers write lots of ways; there is no single process, and there is no single formula that works for everyone. But unless you’re a writer yourself, you wouldn’t know that. And far too many teachers aren’t writers. That’s sad, because once teachers have had the opportunity to do genuine, authentic writing, they get it. All of them have written, or they wouldn’t have gotten through college. But for most of them, that was a long time ago, and that writing was more of a chore than a challenge. (The inauthenticity of most academic writing is clear to anyone who looks–but that’s a topic for another day.) You wouldn’t send your child to a piano teacher who didn’t play the piano, but that’s precisely what we are doing when we send children to schools where the writing teachers don’t write. You can’t teach the piano–or writing–from a book.

Teachers know that–at least, they know the results of trying to teach writing from a book or a program or a formula. You hear it at every level: college professors complain that kids are coming to them lacking writing skills; high school teachers blame the middle school; middle school teachers think that elementary school teachers haven’t done their job. The failure of teaching writing-by-the-numbers is apparent to everyone. But because so many teachers don’t do much writing themselves, they don’t know any other way. So they try, every year, to assign business letters and friendly letters, expository essays and persuasive essays, literary analysis essays and, of course, the research paper. Maybe, they hope, this time it will take.

Then, along comes the silver bullet of our times, technolgy. It seems that there are computer programs that will improve students’ writing. The claims made by the education-market companies promise astounding results, and there’s a certain logic to it. Kids grow up with tablets, video games and smart phones. It makes sense that they would embrace a writing medium that seems to be a part of their world.

Perhaps the teacher responds to an advertisement from a company. Before long, an aggressive salesman calls and says, “Let me come in and show you.” He visits the school, gives a quick demonstration of how his particular program works, and sets-up the teacher with trial licenses for her students. A week later the teacher has her class in the computer lab, responding to the pre-loaded prompts and sending their answers off into the cloud. With the speed we’ve become accustomed to, the writing comes back to them with all kinds of suggestions for improvement and correction. And with a score. That’s important, because once the student makes those changes and fixes those errors and once again sends the writing out into the void, instantly it comes back with a higher score. Wow! It’s like getting to the next level of a video game. Let’s keep going!

What the teacher sees is students crouched over keyboards for extended periods of time, engaged with what is on the screen. They are revising, the teacher concludes. This has never happened before. I’ve found the answer. I’ve got to tell my colleagues. And so the fever spreads across the school and throughout the district. The salesman gets additional ammunition: such-and-such school district has adopted our program, he can now boast. Sometimes the media picks-up on the frenzy and, in their quest to capture your dime, runs enticing stories about excited students, ecstatic teachers, and delighted parents. The juggernaut is rolling.

The problem is, none of these programs substantially improve students’ writing. Numerous studies have shown that. (See the NCTE Position Paper on Machine Scoring.) Because they can attend only to the surface features of writing, that’s all that improves. There is no real revising going on, no discussion with another person about what the writing says, no development or elaboration of ideas by the writer, and no one reading it except a machine. Surface features are all that improve. Yet all involved–the student, the teacher, the parents, even the company sales rep– think that the writing is getting better. The emperor has no clothes.

Entrepreneurs have, in recent years, begun marketing programs that will actually do the writing. For a fee, the author of a technical paper in science or engineering can send his work in and get back a computer-spawned revision of it written for the layman. I have an idea. Suppose all of those those computer generated papers were submitted to a computer reader? We’d have computers doing the writing and computers reading the writing; no human involvement would be necessary. Maybe then, in the words of Richard Brautigan, we’d finally be “All watched over by machines of loving grace.”

Writing for Computers, part 1

by David Arbogast

Lately, English teachers in my school district have been pestered by phone calls and emails from sales reps hawking their latest versions of computerized systems that will read, respond to, and rate student writing in a variety of domains. They offer the promise of giving students immediate and detailed feedback on their writing and suggestions for revising it. They encourage students to make the suggested changes and resubmit the writing for a higher score. Most attractively, they entice the teacher with the promise of simultaneously reducing the quantity of papers she has to read while motivating students to write more and better. It looks like a win-win situation.

Until you look deeper, when you see that all of these programs have some very serious flaws. Despite their reference to complex algorithms, even the best of them can read only surface features of writing. They encourage teachers to abandon best practices and turn the work over to machines. I would argue that they are inherently inimitable to the writing process itself. They encourage formulaic writing, such as the 5-paragraph essay. And now some states, like my own beloved Virginia, are going to use them for scoring high stakes tests.

My initial forays into the academic blogosphere will address these shortcomings one at a time. I begin with my contention that, as they are generally used, these programs intervene too early in the writing process and thereby reinforce the confusion that many of our students have in distinguishing editing from revision. The result is that they come to regard the surface features of their writing as being more important than its meaning.

A student, having written a draft of a piece of writing, submits it to the e-reader for scoring and response. Because the computer cannot comprehend anything but surface features, that is the level of the writing the student considers when making her revisions. She makes the spelling and punctuation changes prompted by the computer. If the e-reader tells her that a particular paragraph is lacking in development, she may insert a few sentences. If the feedback given suggests that her vocabulary is weak, she may search for synonyms. When she re-submits it to the e-reader, she probably will find that her score has risen, at which point she may decide that her paper is complete and ready to give to her teacher. But she has received no feedback on what her writing says and what it means.

If a computer program is used before an actual human reads the paper, revision is unlikely to occur. For revision is much more than fixing spelling errors, obeying static rules of punctuation, and increasing the number and length of the words used. Revision, as its etymology suggests, is about seeing the writing again, in a different way, and through different eyes. Before she can revise, someone has to tell the writer what it means, how it comes across, what happens in the mind of the person who reads it. Unless the writer is very experienced, that looking must be done by someone else, a peer or teacher. But when an e-reader tells the writer that her writing has improved because of the changes she has made, even though the reader has absolutely no idea what the writing says, then the student’s concept of revision, and her ability to do significant revision, is stifled. The lesson she learns is that the surface features of writing are what count and that improving her writing is accomplished by changing those features.

In my next post I will address my fear that computerized response to student writing has the effect of encouraging teachers to abandon best instructional practices, while simultaneously deluding them into thinking that they are actually improving their practice. Strong language, I know, and I know that there are many dedicated teachers who will defend these tools and endorse them enthusiastically. I’m traveling to one or our high schools today for a demonstration by one such teacher. Maybe I’ll be convinced; I’ll try to remain open to the possibilities. I’ll let you know my thoughts next time. Meanwhile, I’d love to hear your thoughts about computerized writing response programs. Please post your comments.