US cinema is perenially talked about in terms of images, seldom in the context of sound. Which is a pity, because despite the trailers for Dolby and the THX system ('The audience is listening') that preface an evening's entertainment, I can't remember dialogue recording ever being worse, not even in those Thirties flicks when actors spoke directly into hidden microphones or during the Seventies rage for fast, furious overlapping dialogue.

Dialogue is different from sound per se. Sound per se is fine: technicans spend weeks blending background noises (cars, chatter, the hum of a fridge) to soundtracks to simulate aural reality. And they can just as easily fabricate a dinosaur's roar: remember how the cinema shook when the T Rex bellowed in Jurassic Park? And remember how the actors kept on eating their words? Prehistoric noise doesn't throw the sound boys - just human speech.

Try sitting through a preview of The Flintstones (right). Technological magic allows sabre-toothed tigers to roam the screen, yet you can only catch about half of what the humans say. True, when you catch the other half, this seems like an act of mercy, but that's hardly the point. The picture is big on volume but unintelligible. The same is true of The Crow, which can perfectly capture a dead superhero's guitar solo yet develops cauliflower ears once Brandon Lee tries old-fashioned verbal communication.

This is either mass coincidental carelessness or there's a presumption at work, the presumption being that audiences no longer care about what's being said. Which may be true: Arnie and Sly have managed maximum careers on minimum talk. In fact, it can only be a matter of time before some whizz-kid jumps up at a studio meeting and shrieks, 'I've got it] The future is. . . silent movies]'

(Photograph omitted)