madojapan

Analysis

The foreign language effect

Four claims about why you are braver in Japanese, sorted by how much evidence actually stands behind each one.

Figure 01

The bias this research was built on

Keysar, Hayakawa and An (2012) gave people the Asian disease problem: identical odds, described once as lives saved and once as lives lost. In their native language, the wording moved them thirty points.

Chose the guaranteed option · native language

Framing effect in a native languageIn the native-language condition, 77% chose the sure option when the problem was framed as lives saved, but only 47% when the same problem was framed as lives lost, a 30 point gap.77%framed aslives saved47%framed aslives lost30 pointsof purewording

Asked in a foreign language, that gap did not hold up. The published paper reports the difference as no longer reliable rather than as a second pair of percentages, so there is nothing honest to draw beside these bars, which is itself worth noticing.

Figure 02

What 13,886 people actually show

Circi, Gatti, Russo and Vecchi (2021) pooled 91 experiments across 15 papers. The effect is real. It is also small, and the spread between studies is wide enough that the next one could find nothing.

Pooled odds ratio and prediction intervalThe pooled odds ratio is 1.34 with a 95% confidence interval from 1.10 to 1.60, sitting to the right of the no-effect line at 1.0. The prediction interval runs from 0.29 to 6.11 and crosses the no-effect line, meaning a future study could plausibly find no effect at all.0.250.51246no effect1.3495% CI 1.10 to 1.60

The yellow band is the prediction interval, 0.29 to 6.11, where the next study could land.

Odds ratio, log scale · k = 91 experiments · N = 13,886 · I² = 72.8%

Read the yellow band

The black bar is where the average sits. The yellow band is where a new study would plausibly land, and it crosses the no-effect line. The authors call it a "small but significant" effect. It is not nothing, and it is not the personality transplant the popular write-ups imply.

Figure 03

Four explanations, sorted by evidence

Three come from the literature. The fourth does not, and it is the only one that is actually about speaking.

  1. 01

    心理的距離

    Psychological distance

    measured

    Second-language words carry less emotional charge than the ones learned in childhood.

    Skin-conductance studies show weaker physical response to taboo words in a second language.

  2. 02

    意識的処理

    Deliberate processing

    measured

    A foreign language will not run on autopilot, which widens the gap between impulse and speech.

    Consistent with dual-process accounts, though the mechanism is inferred rather than isolated.

  3. 03

    社会的自己

    Distance from the social self

    proposed

    Japanese was not present when your social roles were formed, so it sits outside them.

    Well argued in the qualitative literature; not directly tested by the decision studies.

  4. 04

    失敗の許可

    Permission to fail

    untested

    A learner is held to a lower standard and knows it. Errors read as learning, not as carelessness.

    Not part of the foreign-language-effect literature at all. The only one of the four that is about speaking.

Figure 04

The leap nobody has tested

The gap between what was measured and what is claimedThe studies measured framing bias and moral choices. The popular claim is about confidence in conversation. No study measured the second, so the link between them is an inference, not a finding.What was measuredframing biasmoral dilemmasinference· not a findingWhat people claimconfidence speakingsocial courage

Every study here measured a decision on paper. None measured whether you felt brave enough to speak. The bridge is plausible, and it is the part of the argument doing the most work while carrying the least support.

Sources

  • Keysar, Hayakawa & An (2012). The Foreign-Language Effect: Thinking in a Foreign Tongue Reduces Decision Biases. Psychological Science 23(6), 661-668. Figure 01 uses the Experiment 1a native-language percentages.
  • Circi, Gatti, Russo & Vecchi (2021). The foreign language effect on decision-making: a meta-analysis. Psychonomic Bulletin & Review. All figures in 02 are from this paper.
  • Costa et al. (2014). Your Morals Depend on Language. PLOS ONE.
  • Caldwell-Harris & Ayçiçeği-Dinn (2009), on emotional reactivity in a first versus second language.

All lessons