Predicting naming scores from language history: A little immersion goes a long way, and self-rated proficiency matters more than percent use.

Language proficiency is a critically important factor in research on bilingualism, but researchers disagree on its measurement. Validated objective measures exist, but investigators often rely exclusively on subjective measures. We investigated if combining multiple self-report measures improves prediction of objective naming test scores in 36 English-dominant versus 32 Spanish-dominant older bilinguals (Experiment 1), and in 41 older Spanish-English bilinguals versus 41 proficiency-matched young bilinguals (Experiment 2). Self-rated proficiency was a powerful but sometimes inaccurate predictor and better predicted naming accuracy when combined with years of immersion, while percent use explained little or no unique variance. Spanish-dominant bilinguals rated themselves more strictly than English-dominant bilinguals at the same objectively measured proficiency level. Immersion affected young more than older bilinguals, and non-immersed (English-dominant) more than immersed (Spanish-dominant) bilinguals. Self-reported proficiency ratings can produce spurious results, but predictive power improves when combined with self-report questions that might be less affected by subjective judgements.