Linguistic Nuances and Cross-Cultural Validity: A New Approach to Comparative Surveys

This project examines how linguistic and cultural translation issues distort survey responses and compromise comparability in cross-national surveys. Despite significant investments in global studies like the European Social Survey and the World Values Survey, little is known about how respondents perceive translated terms, particularly regarding emotional intensity. For instance, differences in reported happiness levels between Finland and Denmark may reflect translation choices in response scales rather than actual differences, underscoring the need for better translation practices.
Using the LES Online Distributional Semantic Lexicon, the project systematically analyzes linguistic nuances, including semantic similarity and word intensity, across languages. Through survey experiments in six countries, it examines how translation-driven variations in emotional intensity influence response behavior, particularly at threshold points along satisfaction and happiness scales. By focusing on bilingual populations, it isolates translation effects from broader cultural factors.
Beyond detecting translation issues, the project develops a scalable method to improve survey translations and retrospectively adjust existing data through computational reweighting. These innovations will enhance measurement equivalence, sharpen analytical precision, and strengthen the foundation for evidence-based policymaking across areas such as well-being, inequality, public trust, and other domains.