The e-rater® Engine in Criterion® Online Writing Evaluation Service
The Criterion® Online Writing Evaluation Service uses the e-rater® engine. When writers read a topic and write an essay response for the Criterion service, they’ll receive a score that corresponds to a faculty test reader's evaluation based on the scoring criteria for that topic, as well as detailed feedback on possible errors made.
The e-rater score and trait-level feedback provided through the Criterion interface can help students improve their writing skills in the context of an instructor-led writing course.
The e-rater technology is an embedded real-time application that can’t read or respond to an essay as a teacher would. The computer learns to score essays on each topic by extracting linguistic features of writing based on regularities observed in a large group of "training" essays at each score point on the score scale.
The e-rater engine is designed to be used in conjunction with a writing instructor or teacher. The role of the teacher as manager and guide of a larger program of writing instruction is significant and has been repeatedly mentioned in research.
In some cases, the e-rater engine generates both a score and an advisory message. These advisory messages are an indication that the essay isn’t typical of other essays submitted on the topic. Writers should consider these messages along with the overall score and other feedback when they analyze their essays. Advisory messages serve as alerts to writers that the e-rater engine can intercept many essays that aren’t authentic responses to the essay topic.
Generating advisory messages and doing the ongoing work to improve the capability to detect nonsense or off-topic responses is part of our commitment to enhancing the value of the e-rater engine score.