14 Jun 2023

Christopher lucks in as the real Mr Chippy

6:20 am on 14 June 2023

The prime minister may have the Chippy moniker but his opponent has emerged as the more chipper of the two in an RNZ analysis of political speech.

National leader Christopher Luxon used a greater percentage of positive words in his media interviews than either Prime Minister Chris Hipkins or his predecessor Jacinda Ardern.

The extent to which political leaders hammer certain words and phrases - and how they use language to spin their messages - is revealed in the RNZ series The Interview.

For the data project RNZ recorded and analysed 28 hours of weekly interviews with the Labour and National leaders on RNZ, TVNZ, Newshub and Newstalk ZB between July 2022 and May 2023.

According to a language sentiment data dictionary, which rates words as either positive, neutral or negative, Luxon was the more sunny communicator.

On RNZ's Morning Report the proportion of positive words spoken by Luxon was 3.5 percent, the negative words were 2 percent and the rest were neutral. That compared with Hipkins, whose proportion of positive words was 2.7 percent and Ardern 2.6 percent.

On Newshub's AM Luxon's proportion of positive words was 3.4 percent compared with 2.9 percent for Hipkins and 2.7 percent for Ardern.

On Newstalk ZB's Mike Hosking Breakfast Luxon's positive word count was 3 percent compared with 2.6 percent for Hipkins. (Ardern did not do interviews with Hosking in the period monitored by RNZ).

On TVNZ's Breakfast Luxon's proportion of positive words was 3.5 percent. Hipkins also scored 3.5 percent and Ardern 3 percent.

Despite Ardern's reputation for kindness and empathy, Hipkins was the more positive communicator of the two Labour leaders.

Andreea Calude, a data linguist who works on various aspects of language analysis at Waikato University, was surprised the data showed Luxon came out on top for use of positive language.

"Those in power want to tell the world how good it is and how it's wonderful because they've been in power and those coming in want to tell you why it's actually awful and you should be voting for them," she said.

"So generally speaking, you'd expect some differences there, which you don't find in this data, which is interesting."

Collective language

Ardern's reputation for inclusivity was more evident on another measure, however.

The analysis shows that Ardern makes more use of collective language than the other two leaders - invoking a sense of togetherness, as she did with her 'team of five million' phrase, used during the pandemic.

In Ardern's interviews, 4.4 percent of the words were "we/us" words and just 2.5 percent were "I/me" words.

Ardern's tendency to use language emphasising the collective rather than the individual was more marked than in Hipkins' interviews, where 3.1 percent of the words were "we/us" terms and 2.6 percent were "I/me" words.

Luxon came in between the two Labour leaders with 3.6 percent of his speech "we/us" words and 2.3 percent "I/me" words.

Calude said pronouns were little words with immense power.

"If I tell my kids 'it's winter now and we should all be washing our hands thoroughly' this is going to be very differently received to me saying, 'it's winter now and you should be washing your hands thoroughly'."

She said that when the speaker used 'we' they positioned themselves inside the group rather than as an outsider telling someone what to do.

"It can be a powerful way of implicitly saying that as a leader, 'I'm part of you, I'm in your group,'" she said.

"The framing is very different in those pronouns and even though they're little words, they work quite hard."

Ardern's phrase 'they are us' in the wake of the Christchurch mosque attacks was a perfect example.

"It's a really strong, simple, punchy way of essentially saying that we care and we feel your pain. They're not a minority in the corner. They're part of us."

Calude said politicians were very strategic in how they used language to endear themselves to voters.

"We use language, not just to convey information and pure content from one brain to another, we also use its packaging to send subtle messages about who we are, what our alliances are and what kind of persona we want to put forward."

RNZ also measured the lexical diversity of the leaders to see how broad their vocabularies were. There was little difference between Ardern, Hipkins and Luxon and most of the language was simple, standard and avoided complex words.

"I wonder to what extent we actually see ourselves as a kind of down to earth nation and so we don't like that flamboyant over the top kind of pompous and adorned language," Calude said.

"We want it straight and we trust people that speak like us - people who speak like everyday New Zealanders."

How we analysed the data:Weekly leader interviews on the following shows were included. Newshub's AM Show, TVNZ Breakfast, RNZ's Morning Report, and Newstalk ZB's Mike Hosking Breakfast Show. In total more than 28 hours of interviews were analysed including 107 interviews with Christopher Luxon, 49 with Jacinda Ardern and 50 with Chris Hipkins. Recordings were transcribed by Otter.ai and then manually cleaned to improve accuracy, however some minor errors may remain. Excessively long preambles to interviews by hosts were shortened to focus on the interviewer and leader exchange. Text was analysed using R and the Quanteda package.

The positive and negative sentiment analysis used the Lexicoder Sentiment Dictionary 2015. This was used outside of its academic license with permission from Stuart Soroka. The 49 most recent interviews for each leader were analysed for sentiment. To improve the accuracy of results the word 'well' was removed at the beginning of leader answers, and the word 'good' was removed where it was followed by 'morning'. No other changes were made. For emotion words the same approach was taken using the NRC Word-Emotion Association Lexicon to categorise words of different emotions. This dictionary allows words to represent more than one emotion, for example, 'money' is classed as a joy and anger word. For this analysis words were lemmatised, for example, 'happily' was changed to 'happy'.