Limits of the foreign language effect: intertemporal choice

Abstract

Intertemporal choice requires one to decide between smaller sooner and larger later payoffs and is captured by discount rates. Across two preregistered experiments testing three language pairs (Polish vs. English, Spanish, and German; Experiment 1) and with incentivized participants (Experiment 2), we found no evidence that using a foreign language decreased the strength or increased the consistency of intertemporal choices. On the contrary, there was some evidence of stronger discounting when a foreign language was used. We confirmed prior findings that more reflective individuals discount less strongly but observed that they were just as (un)affected by using foreign language as less reflective individuals. Thus, we provide preliminary evidence that the foreign language effect is robust to individual differences in cognitive reflection.

Keywords:

• Foreign language effect
• delay discounting
• intertemporal choice
• cognitive reflection test

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 In all our sample of participants who declared enough understanding of the materials to qualify, there was just one person that failed the consistency requirement in the native language condition, and four in the foreign language condition. Remaining participants did not differ in their consistency of responding to the MCQ, t(291) = −0.61, p = .543. Including participants who provided unsatisfactory consistent responses in discounting problems resulted in similar results with identical conclusions as reported here.

2 A probability of 60% was used to assign participants to the foreign language condition because greater drop out was expected in this condition. Participants tend to quit the more demanding tasks (i.e., the foreign language condition) more often. This was not needed, as the drop-out rates were identical across experimental conditions, χ2 = 0.85, p = .652

3 We mistakenly preregistered linear mixed effects analysis, but since there was just one measure of the DV for each individual, this was infeasible. We substituted this analysis with ANCOVA because it is a closest analogy, and standard in psychological research.

4 This is likely to be the case because linguistic comprehension should have been almost perfect in the native language condition. Additional evidence comes from the observation that understanding and proficiency were weakly correlated in participants in the foreign language condition, τ_B(141) = .16, p = .008. If low self-reported understanding were an indicator of poor comprehension, understanding and foreign language proficiency should have been strongly correlated.

5 To put this in context, 100 PLN is equivalent to the Polish 8-hour minimal wage job salary, or $25.

6 The difference between areas under the curve obtained with reference points in the present vs. the future has opposite signs for gains and losses. By considering the RPDI as an absolute value of this difference, we can account for these reversed signs and boost statistical power without losing information about reference point dependency.

Additional information

Funding

The current research was supported by grant 2017/26/D/HS6/01159 from the National Science Centre (NCN, Poland) to Michał Białek. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.