Psychologist logo
Obama gives a speech on the election trail
Government and politics, Social and behavioural

Comparing Obama’s and Romney’s speech styles and the way their audiences react

Is it possible to estimate a political candidate's electoral performance by measuring how audiences react to their speeches?

11 November 2015

By Alex Fradera

In the run up to the 2012 US election, President Obama visited the undecided swing-states he needed to win in order to hold on to the Presidency. A new study published in the Journal of Language and Social Psychology features an analysis of the speeches he gave, together with those of his opponent Mitt Romney, and finds it's possible to estimate the candidates' subsequent electoral success by measuring how audiences reacted to their speeches. It also describes how speeches are intentionally designed to trigger applause and cheers through the use of rhetorical devices dubbed, appropriately enough, claptraps.

Public speaking researcher Max Atkinson coined the term claptraps in his 1984 book Our Master's Voices after analysing contemporary UK party conference speeches. He identified seven, all of which seek applause implicitly by creating a beat in the conversation to which the audience feels compelled to respond:

  • Puzzle-solution – pose a problem, give an answer
  • Headline-punchline – say you will make an announcement or declaration, then do it.
  • Take a position – describe something, then praise or condemn it
  • Three-part list – use the rule of threes to produce a unit of speech that the audience sees coming
  • Contrast – again, produce a unit by stating two things in contrast
  • Pursuit – resummarise a point to chase a reaction that wasn't yet given
  • A combination of the above 

For the new research, Peter Bull and Karolis Miskinis at the University of York pored over 11 Obama and Romney swing-state speeches, transcripts of the text of those speeches, together with video evidence of audience reactions. The researchers found that Obama and Romney made heavy use of these seven claptrap devices to generate responses.

There are also explicit ways to get an audience response, such as naming a person, expressing gratitude, or asking for support, but the implicit claptraps were used twice as often as these explicit requests. This is similar to findings from Atkinson's British research, but contrasts with work involving Japanese political speeches, where explicit claptraps appear much more important.

How did the candidates do? Over a similar number of minutes of speechmaking (around 145) Obama racked up more rhetorical devices in almost every category. Romney expressed gratitude twice as often, but Obama made 122 lists to Romney's 70; took 42 positions to his 12; 38 headline-punchlines to his 19; and used 61 incidents of naming to 28. And Obama racked up more positive audience responses too: 2.57 per minute, vs 2.23 per minute. Jokes are another explicit way to get an audience response, and Obama led in this field too: 47 jokes to 18 … and generated almost twice as many laughs.

Oh, and the booing. The majority of cases were "affiliative booing", where the boos weren't directed at the speaker, but at a target the speaker showed disapproval towards. Mitt Romney, however, was the recipient of boos in a few instances. Even in these cases, the booing came right after a rhetorical device – such as cries of 'for shame' after a punchline-headline about repealing Obamacare – suggesting the audience was employing the right of reply baked into the speech, only in this case to complain. As to whether these instances were truly unsolicited, we simply don't know: Romney later admitted to courting booing at times to show himself capable of being tough under fire.

Can audience responses be used to predict the subsequent success of the candidates? There was no link between the two in the previous Japanese research, but here Bull and Miskinis correlated the responses in each swing state with the vote received there, and found this to be statistically significant (an r of .67 meaning that audience responses accounted for 45 per cent of the variability in voting) – Obama provoked better reactions and more vote in several states including Wisconsin and Iowa, Romney more in North Carolina.

The researchers speculate that the US-Japan difference may just reflect the type of audience – the Japanese politicians were presenting to existing supporters, the US ones to an undecided audience. But it may also relate to national culture: the US is an individualistic one, as reflected in the more varied types of audience response, the chants and boos together with uninvited remarks and solo claps.

As such, taking the opportunity to applaud or cheer is more likely to reflect actual disposition towards the candidate, contrasted with collectivist Japan, where public addresses are punctuated by socially expected responses that may not reflect true feelings.

This study doesn't present evidence that usage of rhetorical devices causally influences elections in any way; most obviously, an audience's response may simply be a barometer of candidate popularity by the day of the speech. But it does suggest that audience response is a meaningful predictor of future action, as well as making a case for experimental research to see whether claptraps really can tug applause that would otherwise never arise.