One test, many tongues: Surveying language proficiency across the globe

Area of Science:

• Cognitive Science
• Psycho-linguistics
• Second Language Acquisition

Background:

• Objective characterization of linguistic background is crucial for cognitive and social science research.
• Traditional language proficiency tests are manually created, limiting their scalability for lab and online settings.
• Existing methods struggle to accommodate the vast diversity of global languages.

Purpose of the Study:

• To develop an automated pipeline for deriving language proficiency tests from text corpora.
• To create new language proficiency tests for 1,939 languages.
• To investigate global variations in language skills and reduce English language overreliance in research.

Main Methods:

• Developed an automated pipeline to generate language proficiency tests from text corpora.
• Applied the pipeline to create tests for 1,939 languages.
• Conducted a large-scale survey with 4,137 participants across 34 countries, testing proficiency in all 34 languages.

Main Results:

• The automated test effectively distinguishes native, second-language, and non-speakers within one minute.
• Linguistic and demographic backgrounds systematically influence language proficiency and self-reported skills.
• Vocabulary tests strongly correlate with other linguistic competences (listening, writing) across diverse languages.

Conclusions:

• The automated pipeline provides an efficient and scalable tool for assessing language proficiency.
• The study offers a valuable resource for global language skill variation research.
• This work can help mitigate the overreliance on English in cognitive and social sciences.