February 01, 2005

More dubious grammar testing

In response to my post on the SAT's badly-designed grammar questions, Ray Girvan points out "a recent MetaFilter thread about the online test for the Blue Book of Grammar and Punctuation (http://www.metafilter.com/mefi/39080), whose author similarly makes particular choices on disputable constructions".

Genuine grammatical or semantic errors exist even in well-edited texts, and finding such errors can be a real test of grammatical and stylistic acuity. It's even easier to find uncontroversial typos, grammatical errors, malapropisms and other infelicities in unedited texts written by less skilled writers, or by writers in a hurry. Examples could be selected to test any desired level of verbal ability and familiarity with linguistic norms. And you could even test the ability to identify features that are inappropriate for a given genre or register, from easy things like contractions and slang, to more subtle things such as the use of anarthrous noun phrases in fiction as opposed to journalism.

Given the wealth of indubitable solecisms littering every textual streetcorner, it's strange that the people who make up grammar tests insist so often on picking examples where the norms are at best subject to dispute, and the "rule" that selects one of the alternatives is doubtful if not totally bogus.


Posted by Mark Liberman at February 1, 2005 12:31 AM