The current study examined relationships between expert human judgments of text quality and grammar and mechanical errors in student writing. A corpus of essays (N = 100) written by high school students in the W-Pal system was collected, coded for grammar and mechanical errors, and scored by expert human raters. Results revealed weak relations between grammar errors and holistic essay scores and stronger relations between mechanics and holistic essay scores. Implications for essay scoring algorithms and providing feedback to writers are discussed. [This paper was published in "Proceedings of the 7th International Conference on Educational Data Mining" p300-303.]