Two researchers in Japan have invented a “SpeechJammer” device that can stop a person talking in mid-sentence, by just projecting back to them “their own utterances at a delay of a few hundred milliseconds”. The device does not stop them talking permanently, it is just that they become so confused, they can’t finish their sentence and begin to stutter or just shut up.

The two researchers are Kazutaka Kurihara, a media interaction research scientist at the National Institute of Advanced Industrial Science and Technology, and Koji Tsukada, an assistant professor at Ochanomizu University, and a researcher at JST PRESTO, a program that aims to “cultivate the seeds of precursory science and technology”.

They describe their prototype SpeechJammer, and the results of some experiments, in a paper published on 28 February on arVix, an e-print service owned, operated and quality controlled by Cornell University.

The researchers say the device causes no physical discomfort to the interrupted speaker, and the effect stops as soon as they stop speaking.

The prototype SpeechJammer looks like a black cube about the size of a shoebox mounted on a shaft which acts as a handle. The box contains a direction-sensitive speaker, and on top of it is a direction-sensitive microphone.

On Kazutaka Kurihara’s personal website there is a short video demonstrating the use of the device in two scenarios.

The first scenario shows a small group of people in an office, working at their computers, when one of them receives a call on her cellphone. The conversation begins to irritate the others, and then one of them decides to take action. He points the SpeechJammer at the irritating talker, interrupting her mid-sentence in her cellphone conversation, whereupon she appears confused, and then stops.

In the other scenario, a lecturer is talking and his lecture has run over time. Many of his students are looking quite bored and fed up and one of them takes the SpeechJammer, points it at the lecturer, and he trips over his own words and stutters, interrupting his flow.

The SpeechJammer works on the principle of Delayed Audio Feedback or DAF. There is a theory that when we speak, we use the sound of our own voice uttering the words to help us. But, if that “playback” is artificially delayed, it interrupts the cognitive processing that helps us maintain our flow. In fact, there is a theory that something akin to DAF is what happens to people who stutter, and it is known that artificially induced DAF can help reduce stuttering.

In their paper the researchers describe how they experimented with two speech contexts: one where the speaker was reading news out loud and another that was a “spontaneous monologue”.

It appears that speech jamming is more successful, with this prototype, in the news out loud than in the monologue context, and also, it became obvious that it never works when meaningless sound is uttered, like when someone says “Ahhh” over a long period of time.

With reference to research on communication and decision making, Kurihara and Tsukada point out that applying rules and constraints on verbal contributions can change the properties of the discussion, and they also mention how “negative features” of speech can be “barriers toward peaceful communication”.

They propose that using the SpeechJammer to place a constraint on communication, by simply making “speech difficult for some people”, it might “bring meaningful changes to communication patterns in discussions”.

Such a system “points the way to promising future research relating to discussion dynamics,” they write.

In their paper, the researchers focus very much on the science: the physics of the device and how it might be improved to deal with various parameters, plus the science of communication, and make no mention of the ethical and legal aspects of developing a machine that makes people stop talking.

Written by Catharine Paddock PhD