Reputable linguistic "lie detection"?

« previous post | next post »

Several readers have noted the article by Anne Eisenberg in Saturday's New York Times, "Software that listens for lies":

SHE looks as innocuous as Miss Marple, Agatha Christie's famous detective.

But also like Miss Marple, Julia Hirschberg, a professor of computer science at Columbia University, may spell trouble for a lot of liars.

That's because Dr. Hirschberg is teaching computers how to spot deception — programming them to parse people's speech for patterns that gauge whether they are being honest.

For this sort of lie detection, there's no need to strap anyone into a machine. The person's speech provides all the cues — loudness, changes in pitch, pauses between words, ums and ahs, nervous laughs and dozens of other tiny signs that can suggest a lie.

Dr. Hirschberg is not the only researcher using algorithms to trawl our utterances for evidence of our inner lives. A small band of linguists, engineers and computer scientists, among others, are busy training computers to recognize hallmarks of what they call emotional speech — talk that reflects deception, anger, friendliness and even flirtation.

Much of the article is not about "lie detection" at all, but rather about things that go variously under names like "emotion detection" and "sentiment analysis", among other things: sometime Language Logger Dan Jurafsky, whose stern visage is prominently displayed at the top of the onine version of the article, has worked on "Detecting Friendly, Flirtatious, Awkward, and Assertive Speech in Speed-Dates", among other relevant things, but has never worked on deception.

Anyhow, given my previously-expressed skepticism about current commercial systems (e.g. "Speech-based 'lie detection'? I don't think so", 11/10/2011), several people have asked what I think of Saturday's article and the work referenced in it.

A full answer requires going through the research in some detail, which I won't have time to do until this evening or perhaps tomorrow.

My background perspective — which is not worth much without the detailed discussion of specific research — is that "deception" is very diverse category. It includes acting; bluffing in poker; flattery; exaggeration in a fish story or a sales pitch; covering up guilt in a false denial; making a false accusation; and so on. Each specific instance of deception might be rehearsed or improvised, routine or unusual, going down easy or running into skepticism and resistance, inconsequential or life-changing, and so on. And each would-be deceiver has relevant individual histories, and relevant individual psychological traits.

The effects of these diverse combinations of purpose, context, and character are likely to be so diverse, whether in terms of physiology or in terms of discourse strategy and tactics, that it would be surprising if  the question "what is the effect of deception on speech or writing?" had a coherent answer.

On this view, it may be rather like asking "what is the effect of not having a job on blood chemistry?", or "what is the effect of travel on brain waves?"

I'll add a detailed discussion of some of the relevant research — none of which really claims to address the effects of deception in general — when I have a little more time.



Comments are closed.