Almost everyone I speak to understands what it means to ‘click’ with someone” explains Beth Richardson, as we sit down to talk about her PhD research. “But until recently we had little idea of how this manifest in dialogue, or what impact it really has on interaction.”

Richardson is one of a new wave of researchers examining the process and consequences of verbal ‘clicking’ in interaction. Overwhelmingly she and others have found that clicking, or mimicking another’s use of language, is a positive thing. It breeds cooperation, liking, trust, and even better interaction outcomes.

Take, for example, Rick van Baaren’s study of waitress tipping. He compared the amount of tips waitresses normally received to the amount they received when they deliberately repeated what patrons said when ordering. Tips almost doubled when waitresses mimicked their patrons’ language. A single, simple repetition delivered cold, hard cash.

For her part, Richardson spent four years with me at Lancaster University examining how verbal mimicry works within interviews.  Using a range of different scenarios, she’s explored the role of mimicry in encouraging information exchange and unprompted self-disclosure. She’s even explored whether or not it is possible to elicit cooperation by purposefully mimicking, an idea that presumably owes its genesis to van Baaren’s waitresses.

“Often the problem for interviewers is not that people are trying to conceal information, but that they don’t recognize the importance of providing the information in the first place” explains Richardson. “Most interview training suggests that the trick is to ask the right question. That advice is spot on. But it can be tricky when what you’re interested in is unknown to you.” Richardson’s findings suggest that verbal mimicry might be one of the things that can help elicit such unknown information.

Her first study in this area examined the far more austere interactions of police interrogations. With help from a Canadian colleague, Richardson got hold of a set of Canadian police interrogations and started to examine the language used by interrogator and suspect. “I knew that I wanted to study the process by which people come to confess, but doing so wasn’t straightforward. By examining verbal mimicry it was possible to cut through the complexity of words and exchanges to a higher level of coordination between the parties.”

At this point Richardson produces a graph Mimicry Graphthat might be attributed to the floor of the London Stock Exchange. “The path toward confession is never smooth,” jokes Richardson, as she gestures at the ebb-and-flow of lines across the horizontal axis. “What you see here is two gradually dividing forms of mimicry. The positive slope is the change in verbal mimicry that occurs with confessions. It’s the verbal mimicry side of clicking. The other, level slope is associated with no confession.”

Critical to the differences in these slopes is who matches whose language. Confessions occur when suspects increasingly match the language of their interviewer. They tend to match in the way that they describe issues—typically achieved through auxiliary verbs and prepositions—and they match in their point of reference—as suggested by similar use of personal pronouns. In non-confessions, the balance of who matches who oscillates back and forth with no clear direction.

Interestingly, a similar finding was reported a few years ago by another member of my lab, Sally Thomas. She and I examined the dialogue of hostage negotiations and how this related to incident outcome. Negotiators who were successful at convincing  the perpetrator to surrender typically managed to maintain high degrees of verbal mimicry across the incident. By contrast, when negotiations were unsuccessful, the periods of high mimicry were punctuated by periods of cross-talk and low mimicry. In these incidents, the negotiator and perpetrator never managed to find the same wavelength on which to build a resolution.

In June 2010, encouraged by the findings of her first study, Richardson set out to determine what causes and inhibits mimicry. In one study she had pairs complete a task in which a ‘director’ guided a ‘follower’ to a location on a map. An omission on the follower’s map created a difficulty for the pair that could only be resolved by sharing information. Did verbal mimicry facilitate journey completion? In a second set of studies Richardson had pairs negotiate a scenario where the best outcome could only be achieved when each trusted their counterpart to not take advantage of their cooperation. Did verbal mimicry facilitate that exchange? In a third study, Richardson had pairs discuss personal stories for a set period of time; a scenario that loosely resembled a vetting interview. Was mimicry effective at generating self-disclosure?

The answer to each of these questions is a qualified ‘yes.’ The occurrence of verbal mimicry was associated with more information sharing, cooperation, liking and self-disclosure in many of Richardson’s interactions, and this wasn’t affected by gender or physical appearance. This was true even in competitive situations where one person held all the cards. As verbal mimicry increased so the other party was willing to cooperate on those terms.

There were, however, two important exceptions. When the nature of the interaction was cooperative, in the sense that those involved had no competing agendas, verbal mimicry backfired when one person took a dominant role. The power asymmetry created by the dominance tended to get mimicked, which only served to eschew what should have been the natural, informal cooperation of an equal-basis relationship. Sometimes it is more effective to appear as though you are not leading the interview.

The second exception mirrors the findings of the police interrogations. When the interaction was hostile, with competing agendas and distinct roles, the use of mutual verbal mimicry caused a stalemate. Competitive interactions tend to resolve only when one person concedes to the other. In the absence of one mimicking the other the interaction went back-and-forth, just like the no confession interrogations.

Richardson’s explanation for both of these exceptions is that they are ‘schema-inconsistent’ scenarios. By this she means that the person’s behaviour is inconsistent with what is expected in such an interaction. This appears to trigger—perhaps even subconsciously trigger—suspicion in the observer. A feeling of unease that stalls cooperative encounters and provokes an unwillingness to concede in competitive encounters.

If mimicry is linked to cooperation—the two exceptions notwithstanding—it represents a potentially useful way of gauging the nature of relationships. It may, for example, be used to identify periods of low cooperation in an interview. A post-interview review of mimicry may flag sticking points that need to be revisited. It may also help interviewers identify and refine their approach to difficult periods of interaction. Similarly, mimicry may be used by investigators who want to learn more about the relationships among gang members. Here the most useful insights will come from monitoring the changing patterns of mimicry over time.

Arguably the most interesting possibility, however, is that people can be trained to mimic in a way that generates cooperation. This is the question Richardson was drawn to answer during the final stages of her project. Her first study of this possibility borrowed an experimental set-up used extensively in cognitive science. Pairs sat opposite one another with a pile of picture cards by their side, and a second set of cards laid out in front of them.  Each member of the pair then took turns describing what was on the uppermost card, while the listening member searched for that card on the table. By manipulating what one member said (they were conscious of the experiment’s true purpose), Richardson showed that it is possible to educe mimicry by changing language use.

“So,” recalls Richardson, “I had data that would convince the scientists, but not those from investigative practice. I needed to demonstrate that it was possible, through mimicry, to gain cooperation and information from somebody in a context where doing so may have negative implications for that person.” She finally showed this in a series of mock interactions where outcome was tied intimately to the extent to which people shared information and acted cooperatively. “I trained my interviewer, Steven, to mimic key aspects of a persons behaviour whenever he could. When utilising this strategy, he educed more cooperation from his opposite number, and he gained a better outcome as a result.”

This is obviously exciting. If Richardson can train people to mimic, then she can, in effect, train people to act in a way that encourages cooperation. However, before this goes mainstream there are some unanswered questions. Chief among these will be whether or not people can deliberately mimic when under pressure, or while having to think about tasks more complex than a simulated negotiation. “I would expect experience to help things considerably” suggests Richardson, “and we have developed software that helps people practice their mimicry. But, ultimately, it’s an unanswered question.”

In the lab where Richardson works, another application of verbal mimicry has been implemented in software. As Taylor explained, “while face-to-face matching appears a difficult task, matching the language somebody uses online, in a forum or through email, is comparatively easy because an individual has time to formulate his or her response. It becomes possible to tweak an email response to increase the degree it mimics the sender’s message. We’ve developed software that integrates with Outlook to help people achieve this.”

There is a second, less obvious implication of this work. If people mimic different types of behaviour, it is possible that inadvertent use of particular language by an interviewer may shape the response given by an interviewee. This becomes quite important when a statement is being assessed for its veracity. Those seeking to authenticate an account will often look for references to perceptual and sensory detail. The inclusion of such detail usually suggests that an account originates from a true memory. But what if such ‘sensory’ language was also part of the interviewer’s question. Does that prime a suspect to include sensory details in their answer? If true, this would make it difficult to know whether sensory details stem from a true memory or from the type of question asked. Worryingly, early evidence suggests that such priming does occur.

Overall, Richardson is cautious about the value of verbal mimicry. “The strategic use of verbal mimicry is something for the future. We’ve only demonstrated its possible in experimental conditions, and it needs testing in the wild before it can be given true credence. Its use as a measure of cooperation may be more valuable, especially in terms of unexpected changes in cooperation.” Whatever the applicability of mimicry, there appears to be some truth to Martin Heidegger’s quip: “Man acts as though he were the shaper and master of language, while in fact language remains the master of man.”

PDF version

Download here

Read more

Richardson, B., Taylor, P. J., Snook, B., Conchie, S. M., & Bennell, C. (2014). Language style matching and confessions in police interrogations. Law and Human Behavior, 38, 357-366. doi:10.1037/lhb0000077