Do metaphors put us on the right track?

February 2017

Earnsy Liu, TechCommNZ member and GDID student, looks for evidence (not just opinions) to help you manage the daily conundrums we face in our profession. If you have a question for Earnsy to tackle, please email This month, Earnsy investigates whether metaphors matter in technical communication.

What do the crossroads of life, the greenhouse effect, and jump-starting the economy have in common?

They’re all metaphors: expressions describing ‘a person or [an] object by referring to something that is considered to have similar characteristics to that person or object’ (‘Metaphor’, n.d.).

Do metaphors matter in technical communication? Do they influence thinking – how, and how effectively? Which should we use?

Do metaphors matter?

They’re common, ‘often so common in our daily lives that we do not think of them as metaphors’ (Asplund, 2011, p. 2). It is estimated one out of every 25 words is a metaphor (cited in Gorlick, 2011). We even have visual metaphors – think of the folder icon on your computer (Praetorius, cited in McClure, 2009, p. 41). In fact, they’re extremely important in graphical design (E. Harding, personal communication, January 24, 2017).

It is argued that metaphors facilitate understanding by making concepts less abstract, and that they improve user experience by increasing appeal, engagement, and motivation (Idler, 2012; McClure, 2009).

They can provide new insights and facilitate discussion: ‘Novel metaphors can lead us to think about old problems in new ways and to discover new solutions. Further, by simplifying the problem space, and allowing reuse of knowledge from everyday experience, metaphors can allow more people to participate meaningfully in policy discussion’ (Thibodeau & Boroditsky, 2013, p. 1).

However, metaphors ‘always [highlight] certain perspectives while hiding others’ (Asplund, 2011, p. 5). A New York Times columnist preferred to talk about towing the 2011 US economy rather than about jump-starting it. He argued the jump-start metaphor would ‘[lead] people to downplay the problems that arise when temporary stimulus fades away’, whereas towing would suggest the extended support it needed (Krugman, 2010). Psychologists Thibodeau and Boroditsky contend metaphor selection ‘may lead to an illusion of simplicity and bad policy decisions, even among experts’ (2013, p. 1).

Do they influence thinking?

Recent evidence on how they affect thinking comes from a series of experiments by various researchers, starting with Thibodeau and Boroditsky.

Typically, half the participants read about crime as a wild beast preying on a city, while the other half read about crime as a virus infection. Here are the text and questions from the original experiment (Thibodeau & Boroditsky, 2011, p. 3):

Crime is a {wild beast preying on/virus infecting} the city of Addison. The crime rate in the once peaceful city has steadily increased over the past three years. In fact, these days it seems that crime is {lurking in/plaguing} every neighborhood. In 2004, 46,177 crimes were reported compared to more than 55,000 reported in 2007. The rise in violent crime is particularly alarming. In 2004, there were 330 murders in the city, in 2007, there were over 500.

1) In your opinion what does Addison need to do to reduce crime?

2) Please underline the part of the report that was most influential in your decision.

Subsequent experiments varied the text and tasks, but tasks were generally about crime mitigation solutions, and about which part of the text affected their response.

Although a few of the experiments were carried out on undergraduates, including the very first one, most experiments crowdsourced participants online. This made it possible for people of varying ages to take part, often between 18 and 70 years, but averaging just over 30 years. This also enabled huge samples (frequently a few hundred), usually native speakers. The ages and sample sizes make me think the findings probably apply to the general population better than experiments involving ‘just’ a few dozen undergraduates.


Responses were categorised as ‘reform’ or ‘enforce’:

  • reform: involving diagnosis, treatment, educational reform, economic welfare, etc.
  • enforce: involving capture, enforcement, punishment, etc.

Those who read about the virus preferred reform, whereas those who read about the beast preferred enforcement (table 1):

Proportions preferring enforcement after reading different texts

Beast text

Virus text

Thibodeau & Boroditsky, 2011, expt 1 74% 56%
Thibodeau & Boroditsky, 2011, expt 2 71% 54%
Thibodeau & Boroditsky, 2011, expt 4 40% 22%
Thibodeau & Boroditsky, 2013, expt 1 87% 40%
Thibodeau & Boroditsky, 2013, expt 2-4 42% 31%
Thibodeau & Boroditsky, 2015, expt 2 61% 53%
Christmann & Göhring, 2016 44% 26%

Table 1: More participants who read about a beast opted for enforcement than participants who read about a virus.

This echoes a previous finding that metaphors encourage some solutions over others (Robins & Mayer, cited in Reijnierse, Burgers, Krennmayr, & Steen, 2015, p. 246).

Experiments in German by Christmann & Göhring replicated the effect – and found an interesting difference: overall, irrespective of the text read, responses in the English language studies favoured enforcement, while responses in the German study favoured reform (figure 1). They attribute the difference to the infrequency of the word ‘beast’ in German (relative to ‘virus’), and its unconventional use to describe situations or issues (2016).

Figure 1: English speakers preferred enforcement measures, but German speakers preferred reform. (Thibodeau and Boroditsky’s studies were conducted in the US; Christmann and Göhring’s study was conducted in Germany.)

Challenging the findings

Steen, Reijnierse, & Burgers (2014) have challenged Thibodeau and Boroditsky’s findings for:

  • using ‘supporting metaphors’ that act as prompts and ‘maintain and develop the initial metaphorical frame’ (p. 4). They refer to these words in bold (p. 4):

Crime is a beast/virus ravaging the city of Addison. Five years ago Addison was in good shape, with no obvious vulnerabilities. Unfortunately, in the past five years the city’s defense systems have weakened, and the city has succumbed to crime. Today, there are more than 55,000 criminal incidents a year - up by more than 10,000 per year. There is a worry that if the city does not regain its strength soon, even more serious problems may start to develop.

  • not comparing metaphor text with non-metaphor text
  • not having a baseline (pre-experiment) measure of opinions.

Steen et al.’s four experiments ‘found no effects of [metaphors] on reasoning’ (2014, p. 21).

After Thibodeau and Boroditsky (2015) analysed their critics’ work and carried out fresh research, they maintained their findings were correct. They concluded that Steen et al.’s experiments:

  • were underpowered (samples were too small)
  • analysed results incorrectly (responses should not have been aggregated)
  • relied on outdated opinion norms (perceptions of crime had changed)
  • assumed non-metaphor text was neutral (it implied a different severity).

How do metaphors work?

Metaphors work ‘covertly’ – as Thibodeau and Boroditsky put it – and efficiently when they are integrated upfront into text:

  • Covertly. For one thing, 10 per cent or less of participants acknowledged the metaphor affected their responses (Thibodeau & Boroditsky, 2011 and 2013; Christmann & Göhring, 2016). For another, whether they remembered the metaphor didn’t affect their responses (Thibodeau & Boroditsky, 2013 and 2015).
  • Efficiently. Metaphors do not have to be extended (repeated). When the only reference to the metaphor was ‘Crime is a {beast/virus} ravaging the city of Addison’, without mentioning it {lurking in/plaguing} the city, the effect remained (Thibodeau & Boroditsky, 2011). Likewise, using a metaphor in one or in four sentences did not clearly influence opinion (Reijnierse et al., 2015).
  • Integrated. Indirect prompts do not work. Asking participants to think of synonyms for ‘beast’ or ‘virus’ beforehand, without using the words in the text, did not affect their responses (Thibodeau and Boroditsky, 2011).
  • Upfront. Leaving ‘Crime is a {beast/virus} ravaging the city of Addison’ till the last sentence was ineffective. Metaphors work when presented early and can ‘help organize and coerce further incoming information’ (Thibodeau & Boroditsky, 2011, p. 9).
  • Without discrimination. Metaphors affect both genders similarly (Thibodeau and Boroditsky, 2011; Christmann & Göhring, 2016).

How effective are they?

Effectiveness seems to depend on perspective and prior knowledge.

  • Perspective. Republicans preferred enforcement to reform, and were less influenced by metaphors than Democrats and Independents (Thibodeau and Boroditsky, 2011, 2013, 2015; Reijnierse, 2015). Such flexibility may be topic dependent, as Republicans have been found to be more flexible than Democrats on other matters (Hardisty, Johnson & Weber, cited in Thibodeau & Boroditsky, 2013; Plumm, Borhart, & Weatherly, 2012).
  • Prior knowledge. Other research has found metaphors work only when explaining unfamiliar subjects (Göhring and Christmann, cited in Christmann & Göhring, 2016; Robins & Mayer, cited in Steen et al., 2014).

This could partly explain why The Economist used 55 per cent more metaphors than The International Economist (Cardini, 2014); perhaps The Economist caters to its less expert audience by using metaphors to explain concepts.

The researcher, a linguist, put this down to different writing styles or a need for The Economist to make things interesting. That’s possible too: participants find text more vivid and sometimes less complex when a metaphor is used in more sentences (Reijnierse et al., 2015).

So which metaphors should we use?

Participants do not always prefer metaphors. Presented with a choice of ‘beast’, ‘virus’, ‘problem’, and ‘horrific problem’ to describe severe crime, most preferred ‘horrific problem’, as shown in figure 2. We should be careful before generalising from this though, since metaphors vary in different dimensions, including novelty, vividness, and conventionality (Steen, 2014; Thibodeau and Boroditsky, 2015).

Figure 2: Nearly half the participants preferred a plain expression – horrific problem – to a metaphor (adapted from Thibodeau and Boroditsky, 2015).

How do we know when to use metaphors, and which ones? I don’t know, to be honest. I tried asking Thibodeau and Boroditsky but one of them was on leave and the other did not reply. Perhaps future research could test metaphors on users, to check for familiarity and unintended associations. For now, it’d pay to pick metaphors that highlight the appropriate perspective, and to consider using them when explaining something technical or dry to beginners.


Thanks a million to Emma Harding for her guidance and valuable comments.


Asplund, T. (2011). Metaphors in climate discourse: An analysis of Swedish farm magazines. Journal of Science Communication, 10(4). Retrieved from

Cardini, F. E. (2014). Analysing English metaphors of the economic crisis. Lingue e Linguaggi, 11, 59-76. Retrieved from

Christmann, U., & Göhring, A. L. (2016). A German-language replication study analysing the role of figurative speech in reasoning. Scientific Data, 3. Retrieved from

Gorlick, A. (2011). Is crime a virus or a beast? When describing crime, Stanford study shows the word you pick can frame the debate on how to fight it. Retrieved from Stanford University website:

Idler, S. (2012). 5 reasons why metaphors can improve the user experience. Retrieved from Six Revisions website:

Krugman, P. (2011). Block that metaphor. Retrieved from New York Times website:

Plumm, K. M., Borhart, H., & Weatherly, J. N. (2012). Choose your words wisely: Delay discounting of differently titled social policy issues. Behavior and Social Issues, 21, 26-48. Retrieved from

Metaphor (n.d.). In Cambridge Dictionary. Retrieved from

McClure, A. (2009). Humanizing technical communication with metaphor (Master’s dissertation, University of Central Florida). Retrieved from

Reijnierse, W. G., Burgers, C., Krennmayr, T., & Steen, G. J. (2015). How viruses and beasts affect our opinions (or not): The role of extendedness in metaphorical framing. Metaphor and the Social World, 5(2), 245-263. Retrieved from

Steen, G. J., Reijnierse, W. G., & Burgers, C. (2014). When do natural language metaphors influence reasoning? A follow-up study to Thibodeau and Boroditsky (2013). PLoS ONE, 9(12), e113536. Retrieved from EBSCO database.

Thibodeau, P. H., & Boroditsky, L. (2011). Metaphors we think with: The role of metaphor in reasoning. PLoS ONE, 6(2), e16782. Retrieved from EBSCO database.

Thibodeau, P. H., & Boroditsky, L. (2013). Natural language metaphors covertly influence reasoning. PLoS ONE, 8(1), e52961. Retrieved from

Thibodeau, P. H., & Boroditsky, L. (2015). Measuring effects of metaphor in a dynamic opinion landscape. PLoS ONE, 10(7), e0133939. Retrieved from