Tuesday, 29 May 2007
I use the spell checker and grammar checker in Word all the time. These thing are tools for me, ways to help somewhat with the gruntwork of examining every letter of every word of every sentence in all the documents I edit. The spell checker finds words all the time that have been fumbled (often by me as I edit), although it finds many, many more that it thinks are errors but are just fine in context (e.g. lots of technical names). The grammar checker doesn't have as much opportunity to be helpful, but it's good at finding problems like subject-verb agreement when the subject of sentence has been edited but the verb has not.
But these tools are often looked at askance. As I've noted before (I think), professional editors can be snotty about the grammar checker in particular
ly, focusing on errors that the checker doesn't find, or constructions that confuse the grammar checker and make it believe it's found an error when there is none. Similarly, virtually everyone has examples where the spell checker has missed words (the spell checker is helpless in the face of their-they're-there confusion, for example).
You occasionally hear about situations where someone has relied entirely on the grammar and spell checkers to find errors, and if you hear such a story, it's because things didn't turn out well. For some reason this is seen as a reason to bash the tools rather than bashing the people who rely on them too much.
I got another perspective the other day. I had the good fortune to exchange email with Geoff Pullum of the Language Log; I had commented on one of his entries, and in the exchange the subject of the Word grammar checker came up. One of Pullum's repeated theses is that we are saddled with a bunch of rules in English that shouldn't be rules at all, because they're just made up -- they have no basis in the linguistic history of English. (that/which, split infinitives, that sort of thing.) A point that Pullum made was that the grammar checker in Word dutifully marked violations of a number of these non-rules, which, of course, he doesn't approve of.
I noted that the grammar checker is configurable and that one can include and exclude things to be checked. His response:
It's good that Word can be ratcheted up or down in its prissiness level. But it's not exactly an easy on-the-dashboard adjustment. Most people who use Word never customize it at all, so they think it is the program (or even the computer) that permanently criticizes their writing and they just have to hang their head and put up with it.Ah. Assuming this is true, most people don't share my view that these tools are like having an extremely thorough but not particularly bright underling double-check your work; instead, people view the tools as authorities pronouncing on the correctness of one's writing. If Word out of the box tells you that a split infinitive is bad (I don't know that it does, but let's assume), who are you to question Word, or, as Pullum notes, the computer?
He did concede, however, that the tools in Word have a tough job, expressed as perhaps only a linguist might:
I do realize, by the way, that it's difficult to deal with a general public who might as well be asked to learn quantum mechanics as asked to appreciate that at least five different things are involved in what Word is trying to do for them: (i) intradialectal fixed spelling conventions, (ii) dialect-invariant syntactic constraints ("the the" is ungrammatical for absolutely everyone), (iii) features differentiating between Standard English and its non-standard dialects ("ain't"), (iv) mythical prescriptive bugaboo rules that are not relevant for any dialect ("which"/"that"), and (v) stylistic recommendations of an optional nature (try to avoid very long sentences with many subordinate clauses). Some will want all five to be enforced with insane stringency; others may only want help in spotting gross violations of (i) and (ii).
Considering all this, the tools are pretty good at what they do, imo. But no matter how good they are, people need to understand the tools' limitations, or perhaps more fundamentally, the tools are just tools, and they should never have the last say. Don't let that computer boss you around.