One John Elliott of Leeds Metropolitan University believes that it is doable, at least to some extent:

Dr Elliott's programme would compare an alien language to a database of 60 different languages in the world to search see if it has a similar structure.

He believes that even an alien language far removed from any on Earth is likely to have recognisable patterns that could help reveal how intelligent the life forms are.

"Language has to be structured in a certain way otherwise it will be inefficient and unwieldy," he told New Scientist magazine.

So this approach doesn't give us the content of an alien signal, but it might enable us to distinguish language from music or images. Elliott suggests that his software might give us some indication of how intelligent the aliens are, but I'm not sure that I agree that how many words get packed into a sentence is a particularly good measure of this. Intelligence might ultimately be reflected by economy of language.

Anyhow, the whole thing strikes me as a waste of time. I've been watching aliens on TV for years, and I can assure you that the vast majority of them are English-speaking white people. There is a little ethnic diversity, but not much.

Oh, and many of them have interesting facial ridges. But that doesn't mean we can't communicate with them. Experience shows that we can even reproduce with them if we wish.

I mean -- if they wish, too, obviously.

But seriously, Elliott's ideas provide a good reminder that any aliens we eventually encounter are likely to be far more alien than we can begin to imagine. Even the idea that we can recognize alien language as such might be optimistic.