| Cuts costs of assessing a large volume of essays written for placement tests |
| Software can assess as many as six documents a second, saving time and providing results more quickly |
| According to some studies, can be as good as, even better than, human raters in assessing student writing |
| Can be fed an "ideal" essay and programmed to assess other essays based on this ideal |
| Not affected by exhaustion or biases, as some human raters may be |
| Able to flag essays deemed "unusual" in some way, indicating the need for human asssessment |
|
| Assesses writing based on standards of placement tests; doesn’t distinguish differences in students’ primary discourses |
| Looks for keywords to interpret meaning and content; unable to assess content holistically |
Given to misinterpretation of content, based on keyword searches |
| More likely to focus on conventions, such as syntactical errors, misspellings, sentence complexity, sentence length and punctuation |
| Removes consideration of differences of student background, culture, and language from the equation |
| Does not provide positive feedback; focuses on product, not process |
|
