Rethinking the teaching of grammar from the perspective of corpus linguistics

Despite calls from many composition and rhetoric scholars for instructors of writing to stop teaching prescriptive grammar, a vast number of handbooks intended for college writing classes encourage this tradition. For example, Hacker's Pocket Style Manual has a section on grammar with instructions for students on how to write appropriately. While Hacker may not intend for her instructions to be taken as dictums, they often are, and much time is spent in many classrooms teaching students these rules of grammar. This article uses the Corpus of Contemporary American English (COCA) to support the calls from composition and rhetoric scholars that prescriptive instruction in grammar is more a hindrance to writing instruction than a benefit. Focusing on a few specifics from frequently used grammar handbooks and illustrating how big data shows the "rules" are incorrect at best, this article argues that, just as scholars of English have begun using big data to better understand literary history, scholars of rhetoric and composition might better understand how to help students to write by understanding patterns within big data. Certainly, this argument recognizes that "common usage" may not necessarily be the most eloquent usage. In making this argument, this article focuses on the [neither or either of X] + Verb construction, where the Verb may have either a plural or a singular form. Our findings illustrate that the "real world" writing is different from what textbooks dictate, and we suggest the data-driven observations need to be appropriately incorporated in writing classes.