Human Readers for Tests
As readers of my blogs know, I'm never opposed to using technology when it is an effective tool. I am opposed to the blind embrace of the latest trends without critical examination of the potential side effects. Computer-assisted grading, I can endorse to some extent because I use software to help me analyze student papers — and my own writings. But, I cannot and will not endorse any system that gives weight to the computer-based scoring.
If you're a teacher, consider this petition:
http://humanreaders.org/petition/index.php
Now, I also want to add a critical comment on human graders.
If the graders of standardized tests are using rigid scoring rubrics, they are little better than software algorithms. Bad grading is bad grading. Inflexible = bad.
Again, I am not opposed to using a computer for fact checking, some plagiarism verification, and as formatting aids. Computers can and do help many of us write more effectively. But, I don't use computers to grade papers. Sometimes that distinction escapes my students at first, but eventually they recognize that I'm using software to help me highlight potential problems.
Software is better than ever at grammar and spelling, but it isn't perfect. There are also new applications that can fact check documents, a project funded by the Knight Foundation to fact check political speeches and papers. But, computers cannot yet grade anything that cannot be flowcharted and categorized.
Can I grade a multiple choice test with a computer? Certainly. Can I grade a spelling test automatically? Sure. But, the computer might not recognize any patterns in the errors a student makes. In time, maybe a computer will be able to grade a test and offer suggestions for future study. But, computer technology is years, maybe decades, from being able to "grade" an essay, poem, or creative work.
Even when computers do reach that capability, I want a human to have the final say on grading students. We already are reducing too much education to an isolated process, focused on high-stakes test results. We need some humanity, for lack of a better word, in our classes.
If you're a teacher, consider this petition:
http://humanreaders.org/petition/index.php
Now, I also want to add a critical comment on human graders.
If the graders of standardized tests are using rigid scoring rubrics, they are little better than software algorithms. Bad grading is bad grading. Inflexible = bad.
Again, I am not opposed to using a computer for fact checking, some plagiarism verification, and as formatting aids. Computers can and do help many of us write more effectively. But, I don't use computers to grade papers. Sometimes that distinction escapes my students at first, but eventually they recognize that I'm using software to help me highlight potential problems.
Software is better than ever at grammar and spelling, but it isn't perfect. There are also new applications that can fact check documents, a project funded by the Knight Foundation to fact check political speeches and papers. But, computers cannot yet grade anything that cannot be flowcharted and categorized.
Can I grade a multiple choice test with a computer? Certainly. Can I grade a spelling test automatically? Sure. But, the computer might not recognize any patterns in the errors a student makes. In time, maybe a computer will be able to grade a test and offer suggestions for future study. But, computer technology is years, maybe decades, from being able to "grade" an essay, poem, or creative work.
Even when computers do reach that capability, I want a human to have the final say on grading students. We already are reducing too much education to an isolated process, focused on high-stakes test results. We need some humanity, for lack of a better word, in our classes.
Comments
Post a Comment