Last February Francisco Lacerda, of Stockholm University, made international headlines as one of a number of scientists claiming their academic freedom has been stifled by British libel law.
A peer-reviewed journal had removed his published paper (Eriksson, A. & Lacerda, F. Intl J. Speech Lang. Law 14, 169–173 (2007)) from its website in 2008, following the threat of legal action from Nemesysco Limited. The company, based in Natanya, Israel, makes voice analysis technology which the research criticized (see Nature, Science for coverage at the time).
Lacerda came to London’s Houses of Parliament today to tell a panel of UK politicians about the case, and how he thinks the libel law should be reformed to protect scientific debate. “If you publish in English, as scientists must, you are at risk,” he says (The Times).
A coming deadline may push the action forward: the UK government’s Department for Work and Pensions (DWP) has spent some £2.4 million on pilot studies of Nemesysco’s system and is supposed to decide this spring whether to roll out the technology across the nation. They hope to use the technology to identify potential benefit fraudsters.
Nemesysco says its “Layered Voice Analysis” technology detects emotions such as stress and deception in speech, and can help “truth-detection investigation activities”. Lacerda and his co-investigator, Anders Eriksson of the University of Gothenburg, having analysed the software, say there is no scientific basis for the claims.
“Study after study shows low validity and chance level for reliability. But people won’t listen. They don’t try them in controlled trials,” another (anonymous) specialist tells The Guardian.
At the seminar today, politicians such as the Liberal Democrat science spokesman, Evan Harris, said they would ask the DWP whether its trials to test the technology were scientific. Lacerda does not know: when he asked, he explained today, the DWP sent him a chart of numbers without any methodology behind it.
The problem, as Charles Arthur in The Guardian explains, is that “telling people they are being monitored by a lie detector (real or not) makes them more likely to be truthful” (an effect that has itself been documented in peer-reviewed papers). If this deterrent bluff is the only gain from the technology, then it is either unethical, or DWP could employ a placebo machine at much lower cost.
As for Lacerda’s research, circumstances have not changed much since last year. Though the Swedish Research Council, a government funding agency, signed a statement expressing concern over the removal of his paper, it remains pulled from the journal’s website . Equinox Publishing, who publish the International Journal of Speech, Language, and the Law, where the paper first appeared, say they can’t afford the potential costs of legal action. However, the paper, entitled “Charlatanry in forensic speech science”, is still widely available.
A spokesman for Nemesysco told Nature he believed Lacerda was “piggybacking on Nemesysco to build a reputation for himself. But he is barking up the wrong tree”. He points to references on the company’s website.