PhD defence: The future is a linguistic thing of possibility

Date of news: 18 November 2021

Would you rather get 100 euros directly or have a 50 percent chance of 200 euros later? The choices people make are influenced by the degree of uncertainty that is expressed in a language. That is the outcome of Cole Robertson's research among speakers of Dutch and English. He will defend his thesis on 23 November.


Dividing time into past, present and future is a human trait. The future has not yet happened and therefore, anything is possible. All languages allow speakers to refer to times in the future. But the grammatical marking of future time differs between languages. Linguist Cole Robertson looked at how these differences might impact the way people make decisions. ‘Specifically, decisions that involve weighing present versus future outcomes’, says Robertson. ‘For instance, not spending money now to save it for the future, making an effort to exercise now to become fitter and healthier later, or investing in climate change mitigation efforts to protect the global climate in the future.’

Devalue outcomes

People tend to perceive outcomes as less valuable if they are risky or far in the future. Most people would rather have 20 euros immediately than receive it a month from now because they feel that the delayed 20 euros are subjectively less valuable. Similarly, many people prefer a guaranteed 100 euros over a 50 percent chance of 200 euros. The phenomenon that the subjective value of things decreases with distance in time or risk is known as 'psychological discounting'.

In his research, Robertson looked at future marking in two languages—English and Dutch—to find out how this impacted the way people make decisions. Robertson: ‘In Dutch, it is possible to say Het regent morgen (‘It rains tomorrow’), without using a modal or future tense, but rather using the present tense. In English you have to say: It will rain tomorrow. English obliges speakers to use will, be going to, or shall when talking about the future.’ Previous research by economist Keith Chen suggested that if a language mandates that speakers use the future tense—like English does—speakers will think of future events as further away in time, and would therefore devalue future outcomes more. In Dutch that wouldn’t be the case.

Robertson tried to replicate Chen’s results by having Dutch and English participants complete online intertemporal-choice and risky-intertemporal-choice tasks. ‘We asked our participants questions like ‘Would you rather have 10 euros now or 20 in a month? Would you rather have 10 now or 20 in a year? Then we looked at the rate of their devaluation over time. Surprisingly, in our data we found no support for Chen’s claim that future tense causes increased discounting’, says Robertson. ‘In fact, we found it caused decreased discounting, likely because future tenses like will encode modal notions of high certainty.’

Expressing uncertainty

Instead, Robertson and his team started to realise that the outcomes were influenced by another factor: modality. Since the future cannot be known, a speaker has to guess, estimate and predict. Language about the future, therefore, tends to express that uncertainty. The somewhat broad term for the linguistic expression of such notions is modality. In English, for example, speakers are required to use a modal verb when making predictions, such as may, might, or could. Robertson: It could rain, or it might rain or it may rain. What is obliged in English isn’t the future tense when one makes predictions about the future, but the use of these modal verbs, which tend to encode uncertainty: it is possible, but not guaranteed that things will happen in the future.

In Robertson’s experiments, English speakers used more low-certainty terms when talking about the future. This suggests English obliges speakers to pay more attention to the risk and/or uncertainty which often surrounds future events. Dutch speakers compensated to some extent by using other (non modal verb) constructions, for instance, denken ‘think’, mogelijk ‘possibly’, or wel approximately ‘well’. However, the encoding of low certainty was higher in English. It turned out that it is precisely this difference that caused English participants to devalue risky intertemporal rewards to a greater extent than Dutch participants. Robertson: ‘Not as a function of using tense structures, but because of the grammatical obligation to use low-certainty modals.’


Robertson, therefore, concludes that the grammatical obligation to use low-certainty modal verbs, and not the future tense, drives his results. ‘There might be real-world implications of how language impacts people’s intertemporal decisions. At the same time, the amounts involved in our experiments were small, like 35 euros, so it is difficult to extract bigger claims from these small findings. I don’t think we can say: Okay English speakers, you just have to speak a little bit more like the Dutch and then we can solve climate change’, Robertson says laughing. ‘But my research does suggest that language influences important decisions in the real world.’

Cole Robertson (1984, Vancouver) completed his Bachelor of Arts in Western Society and Culture (Concordia University, Montreal), followed by the Master of Science in Cognitive and Evolutionary Anthropology (University of Oxford). In 2016, he began his PhD research at the Centre for Language Studies at Radboud University. He lives in Oxford and is employed as the Chief Technology Officer for Texture AI.

Cole Robertson will defend his thesis on 23 November.