I picked up a book on the human voice, memorably titled The Human Voice, by Anne Karpf, mostly because it’s about what I’ve always regretted as a lack in linguistics… essentially, everything that voice can do that’s not reducible to writing– intonation, expression, accent, prosody. 

It’s astonishing how much information content there is in the voice.  If I hear you speak even briefly, I know your sex, region, class, rough age, your emotional state, your closeness to me, a good deal of your personality.  It encodes general information and also much of your irreducible individuality.  It’s also a powerful social force: it can soothe or inflame, insult or amuse, completely independent of the words used; it’s a major means of bonding between mother and child.  (Fathers too, to some extent; one of Karpf’s curious facts is that babies aren’t terribly moved by their father’s voice.)

Linguistics generally ignores these things, or issues a vague promissory note– we’ll deal with it after we nail down syntax, maybe.  It’s true that science proceeds as much by excluding as by including domains: Galileo and Newton wouldn’t have got so far with the laws of motion if they hadn’t ignored friction.  On the other hand, if we haven’t explored the territory, the one thing we should know is that we have no idea how big it is or what it contains.  One reason I distrust many current models of language is because they’re adapted to what we do know– the relatively discreet worlds of phonemes and words, which seem amenable to largely computational methods.  But our methods just don’t seem so useful for voice, which is full of analog effects, individual variation, and phenomena we can barely name or talk about.

Unfortunately the book really only underlines how little we know.  It’s full of interesting facts and tantalizing studies, but it’s entirely theory-free.  It’ll probably take another thirty years before we understand, not what the voice can do, but how the brain handles it.  I suspect it’ll be a revolution.