The 2010 UK Election: The Second Debate – A new heartthrob?
Dr Raj Persaud and Professor James W. Pennebaker

Previously on The S Word – The New Scientist Blog, we filed a report of research based on a computer analysis of the words each candidate used on the first of the Thursday night debates.  Using the same program for last night’s debate, we are finding some new trends.

The LIWC (Linguistic Inquiry and Word Count) program Professor Pennebaker has developed examines each word in text supplied to it, and calculates the percentages of words used allocated to particular categories, found from previous research to be psychologically significant.
Using this technique Pennebaker and colleagues analysed in depth what candidates said during the 2004 US Presidential Election Campaign, and was able to determine that John Kerry, the ultimately failed Democratic candidate, used more depressive language compared to the other candidates.
The study, published in the prestigious academic journal Journal of Research in Personality, and entitled ‘Winning words: Individual differences in linguistic style among US Presidential and Vice Presidential Candidates’ argued that, as previous psychological research has established, voters much prefer particular language patterns (eg more optimism), it was key linguistic differences exemplified in speeches made during the campaign, that ultimately determined the electoral outcome.
Given both Democratic candidates used more depressive words than their Republican rivals, this could have been an essential electoral factor, perhaps operating at an unconscious level in the voter’s mind.
In another study with colleagues published in the journal American Psychologist, Martin Seligman, a US Professor of Psychology who advocates optimism as the fundamental variable that determines outcomes in competitive predicaments, was able to predict US election results by merely comparing levels of optimistic vocabulary in candidates’ speeches.
So if Pennebaker deployed his computerized psychological analysis of language would his results stand up in the UK as they did for the US? Would essential linguistic contrasts correlate with voter appeal?
In the first debate on 15 April, Liberal Democrat Nick Clegg wowed the country with his warmth, humility, and charm.  Using Pennebaker’s computerized text analysis methods, Clegg’s language during the debate was found to be distinctively personal, positive and honest compared to Gordon Brown’s academic distance.  David Cameron was linguistically the least distinctive, as the key striking differences emerging were between Brown and Clegg, with Cameron  falling somewhere in-between.
One vital psychological and strategic decision in any election is how different you might want to appear, as opposed to not standing out as too unusual, which could be interpreted as a stratagem of caution. Given before the last televised debate Cameron was widely perceived as the heir apparent to the throne, his key task, and he may have been so briefed, may have been to simply not slip up, and so let the crown slip from his grasp.
However, given the surge in Liberal Democrat support following Nick Clegg’s distinctive performance last week, the demands on Cameron may have altered. Now, with the crown indeed apparently slipping from his grasp (polling showed Liberal Democrats had pushed the Conservatives into second place), not because he made any gaff but because of the distinctiveness of a new opponent, a change in approach may have been forced upon Cameron.
Oh what a difference a week makes in politics… and in linguistic analysis. 
Text analyses of the 22 April debate in Bristol suggest a flipping of linguistic roles between Cameron and Clegg.  David Cameron, by using the most I-words, is now giving the impression of being more personal than his two competitors.  He also practically bubbled with happiness and upbeat language – the role that Nick Clegg grabbed last week. He has also adopted Clegg’s strategy of using high rates of present tense verbs and not referring much to the past.
Compared to the other two adversaries, last week, Pennebaker found Nick Clegg used more personal language (more I – words for example), deployed the most positive emotion words, and tended to talk in the present tense at the highest rates.  These are strong indicators of psychological immediacy, in other words, he was speaking more of the here-and-now.
Last week, Clegg’s overall linguistic style on the night was characterised by verbal markers of honesty, consistent with previous research on differences between truthful and deceptive language. Linguistic honesty is associated with, among other features, higher use of I – words. Now this week again it’s Cameron who has not just closed the gap between him and Clegg, but significantly, has overtaken Clegg here.
Last week Camerion scored highest on negative emotion and Clegg the lowest, this week Cameron has managed to drop his score down so dramatically that he now bottoms out as lowest on negative emotion. The reason he has been able to pull off this spectacular switch is not just that he himself has dropped his use of negative emotion (last week negative emotion words were 1.85% of his total output compared to 1.52% this week), but also Clegg allowed his use of negative emotion to climb (last week it was 1.35% of his total output compared with 1.61% now). 
Cameron didn’t just adjust his game in what would appear to be an adept grasp of where he went wrong last time, Clegg doesn’t appear to have either understood where he went right, or if he did, he has for some reason struggled to keep his eye on the linguistic ball, ensuring he maintained the gap between himself and the others verbally and emotionally.
This could be worrying to Liberal Democrat strategists, as it may suggest that their candidate is going to struggle to maintain the distinctive persona over the next few weeks.
Remember, the psychological and linguistic analysis as conducted by Pennebaker, suggested it was this distinguishing use of language and emotional appeal, which appears to have underpinned the astounding surge in Clegg’s popularity, following his enhanced exposure to the public via the televised debates, rather than a clear sense he was expounding a definitive policy difference, which ultimately appealed.
The fluidity of the two challengers is striking in comparison with rock steady Brown whose figures (1.83% last week; 1.84% this week on negative emotion) indicate that whatever is going to happen until the end of this race, Brown appears to be sticking doggedly to a particular linquistic and emotional strategy through thick and thin, and we may not be able to expect much alteration from him. This could be worrying to Labour strategists who, if Labour continues to bump along at the nadir of third place in many opinion polls, might be hoping for a transformational response from Brown. 
It also might suggest that the outcome of this election is going to turn most on how Cameron and Clegg continue to evolve.
Ominously for the Liberal Democrats given what a euphoric week they have just had, Cameron seems to have sidled his way into Nick Clegg’s verbal territory since last week, and annexed it almost completely. This could explain the immediate reaction opinion polls which generally have not put Nick Clegg as the overwhelming winner, compared to last time, and instead have either put Cameron marginally ahead, but more generally have put all three leaders at much more level pegging compared to last week.
From a language perspective, just as Cameron has shown a more personable side, the linguistic analysis suggests that Nick Clegg appears to be taking himself a bit more seriously.  This week, he is meaningfully lower than Cameron in personal pronouns in general and I – words in particular.  He also seems to be censoring his own thoughts and feelings compared to last week with a large jump in his use of negations (words such as no, not, never).
Gordon Brown’s language continues to be the most consistent.  He continues to be the least personal of the three in his use of pronouns and I-words.  He displays a thinking style that reflects a natural strategy of organizing complex ideas into highly specific concrete categories more than his opponents (as can be seen with his use of articles and prepositions). The previous S Word blog explained the potential psychological significance of this point.
As the debates continue it’s not only the differences between the candidates that lends itself to linguistic and psychological analysis, but also how they adapt to polling reaction and each other. It’s not just where you are that counts in a political campaign, but also your ability to adapt and change as circumstances evolve.
Perhaps the major but thus far unexplicated lurking influence on British politics at the moment, given the advent of political televised debating follows from a longer standing American tradition, are lessons learnt from their impact on results over there. The obvious teacher is going to be Obama, and politicians in the UK are likely to have been keen to learn whatever lessons his electrifying and unforeseen campaign had to teach.
Obama’s speechifying has been characterized as distinctively ‘flowery’ and one way of scientifically operationalising that vague concept, is the use of more adverbs – like ‘very’ and ‘really’. It is notable that in the second of their televised debates, one key contrast between Obama and McCain that emerged from Pennebaker’s linguistic analysis was Obama deployed significantly more adverbs. However, by the third televised debate McCain had learnt his lesson and narrowed the gap in adverb usage (though Obama was still a nose ahead on this point).
It is remarkable in this regard that in the first UK televised debate, Cameron and Clegg who are most likely to be influenced by Obama for a host of reasons, used many more adverbs than Brown, with Clegg ahead of Cameron. This paralleled polling reaction to who was the more successful debater. But by the second debate Cameron had significantly closed the adverb gap between himself and Clegg, while both remain linguistically distinctive compared to Brown who appears to be labouring in a pre-Obama low adverb electoral universe. 
Bottom line: Gordon Brown continues to be Gordon Brown.  Both David Cameron and Nick Clegg voices are evolving and they may be becoming key influences on each other, with Cameron so far learning the right lessons, while Clegg has at least temporarily apparently lost his grip on the distinctive strategy that previously put him ahead.
Word count 586358465963
Big words 17.7716.3417.02
Personal pronouns 9.8910.749.22
   I-wordsI, me, my2.103.032.52
  We-wordsWe, us4.324.643.40
ArticlesA, an, the7.666.796.78
VerbsIs, ran18.4419.3617.34
   Past tenseWas, ran2.952.502.80
   Present tenseAm, feel12.7613.8412.66
   Future tenseWill, shall1.42F0.89
AdverbsVery, so3.585.255.32
PrepositionsOn, to15.0313.0713.85
ConjunctionsAnd, but6.285.445.03
NegationsNo, not, never1.381.542.21
Positive EmotionsLove, nice2.813.713.00
Negative emotionsCry, hate1.841.521.61
Cognitive wordsRealize18.7322.0218.87
   CauseWould. Should2.522.332.30
   DiscrepLove, nice2.663.542.80
To learn more about the text analysis work, go to the website, an click on the “Explorations into Language” link.