Virginia's new governor, Gerald L. Baliles, worried about the recidivism of the state's prison inmates, has come up with a policy that is intriguing, well- motivated and doomed.

He has ordered a "no read, no release" rule for granting parole. There might be exceptions (details are still being worked out), but the general rule would be that no illiterate inmate would be eligible for parole.

Everything about Baliles' reasoning makes sense, except his conclusion. Surely some significant number of convicts turn to crime in the first place because, being unable to read, they have no job and no prospects of gainful employment. The same handicap also makes them likely candidates for recidivism. It follows that if an inmate learned to read, it would improve his chances of staying out of trouble in the future.

It follows, that is, if the inmate learns to read because he wants to learn to read. The fatal error (or so it strikes me) of the Baliles plan is in supposing that inmates -- or anyone else, for that matter -- can be coerced into learning in prison what they failed to learn as children in school.

Scratch an illiterate inmate, and you will find a kid who wanted to learn to read, who spent some years in classes where most of his peers learned to read, who had a teacher who wanted him to learn to read, and whose failure to acquire that basic skill was a source of daily disgrace and humiliation -- and who still did not acquire it.

What makes Baliles believe that it can work to tell such a person that he will be locked up until he learns to read?

And yet that is what he clearly believes. "Far too many prisoners are also prisoners of illiteracy," he correctly observed. "When released, many can't function in a complex, fast-moving society. Many cannot read and write well enough to fill out job application forms or balance a checkbook. They return to what they know too well: crime and prisons. The recidivism rate reflects it. We must stop it."

I understand the governor's noble intention; I understand that his plan would provide tutors, reading clubs, volunteer teachers, new techniques and a million- dollar educational budget; I understand, too, that what he proposes is for the inmate's own good, as well as the good of the larger society. My doubts have to do with my idea of how people learn.

It is a part of the American lore that "reading and 'riting and 'rithmetic" can be "taught to the tune of a hick'ry stick." But I'm not sure it ever worked that way in reality. I remember a handful of slow-learners who were enticed into learning by teachers who had the patience and insight to give them special help and also to work at enhancing their self-confidence. But I am unable to recall any nonreaders among my elementary school classmates who were successfully caned out of their illiteracy. They stayed in school (with growing resentment) as long as they had to, but as far as I can recall, they mostly left with their ignorance (if not their egos) intact.

Can the coercion and humiliation that turned out to be a pedagogical failure with children work with locked-up adults? I doubt that it would work even if basic literacy guaranteed a job, which, of course, it doesn't. Most of the millions of jobless Americans already can read at the level likely to be achieved by the Baliles plan, and many of them, having had jobs at some time in their lives, may possess rudimentary work skills as well. If thousands of laid- off (and literate) laborers and factory workers remain jobless long after their unemployment benefits have run out, is there any reason to suppose that marginally literate stick-up artists will see basic literacy as their salvation?

I suspect that Baliles sees his idea not as a stick but as an enticing carrot: Learn to read and you can get out of prison early. But the fact is, most inmates view parole not as a bonus but as a routine right, and they view the denial of parole as a punishment. Leaving aside the questionable constitutionality of imprisoning people for illiteracy (which is what "no read, no release" would amount to), there is just one thing wrong with what Baliles has in mind: It won't work.