Reading politics in a second language: how it shapes what voters remember

A study by the MultiPoD team shows that when voters read political arguments in a second language, their own views have much less influence on what they remember.

MultiPoD is building a multilingual and multicultural European space for political deliberation, where citizens from different language communities can debate the issues that affect them. In such a space, many participants will engage with arguments written in a language that is not their own. This raises an important question for the project: does the language in which we read political arguments change what we take away from them? A new study by our team, currently under peer review, offers a first answer.

Reading in a second language narrows the bias

We asked 765 bilingual English and French speakers in Canada to read two short texts on immigration, one supportive and one critical, either in their main language or in their second language. We then tested what they remembered, distinguishing between what the texts actually stated and claims that could only be inferred from them.

When reading in their main language, participants were considerably more likely to mistakenly recall unstated claims that matched their own views than claims that challenged them. This is the kind of error that turns a report of increased police patrols into a “memory” of rising crime. Reading in a second language reduced this gap by more than half, as participants became more willing to accept implied claims from the opposing side as well. These results are consistent with the idea that a second language slows readers down and encourages a more careful evaluation of what they read.

Beyond access: language as part of the message

Multilingual deliberation is usually approached as a question of access, ensuring that everyone can read information in their own language. That remains essential. Our findings, however, point to a second dimension. Language also shapes how arguments are understood, and prior views influence memory mainly through what readers infer rather than through what they are told.

This connects directly to MultiPoD’s work on reconstructing the implicit premises behind arguments. Making unstated assumptions visible could help participants, whatever their language, distinguish what was argued from what they read into it. The study also shows that language groups do not always respond in the same way, a reminder that tools designed for one community may work differently in another.

What comes next

These insights will inform the design of our pilots. We will record the language in which participants take part and how comfortable they feel using it, develop materials and summaries that state central claims clearly, and assess not only whether participants understand the content but also what they remember from it.

In multilingual democracies, the way information reaches people matters as much as what it says. For MultiPoD, this means treating language as an integral part of the message and building deliberation spaces where it is easier to tell what was said from what was assumed.

About the study: “Reading Between the Lines: How Language Shapes What Voters Infer” is currently under peer review. We will share the published version once it becomes available.