<p>Risk literacy is increasingly important in today’s society. It involves the adequate understanding of risks for informed decision making. Probability is central for defining risks and thus, an adequate understanding of probabilities (in the mathematical world) is crucial for risk literacy. Further, risk literacy is characterised by a&#xa0;switch between the mathematical world and the risks’ context. As language is used for communicating in both (mathematical world and risks’ context), language reception and production may be particularly relevant for risk literacy. Therefore, we study receptive and productive tasks in risk literacy. Thereby, we focus on typical Bayesian reasoning tasks including conditional probabilities. We vary the verbal expression of conditional probabilities by manipulating the syntactic structure (conditional clause vs. noun phrase) and the sequence of condition and conditional (condition in the beginning vs. end). We measure the ability to determine conditional probabilities based on these varying given verbal expressions (i.e., receptive tasks) and study students’ verbal expression of conditional probabilities (i.e., productive tasks). Moreover, we study the influence of language proficiency on receptive and productive tasks. According to our study with <i>N</i> = 124 secondary-school students, differences in the sequence of condition and conditional are more influential than the syntactic structure in receptive tasks, implying no inherent difficulty in specific linguistic expressions. Further, performance in receptive tasks only partly predicts performance in productive tasks, suggesting a&#xa0;need for explicit training of verbalising probabilities. Finally, language proficiency predicts performance in both tasks.</p>