Can you tell them apart? Using machine learning to classify bilinguals’ and multilinguals’ cognitive and linguistic performance.

The debate of whether bilingualism provides a cognitive and or
linguistic advantage is a lasting one. Underlying this debate is the
idea that an additional language shapes cognition and linguistic
processing. The current research analyzes a behavioral dataset
containing individuals’ performance in different general
cognitive and linguistic tests using a machine learning approach
to classify individuals as bilinguals or multilinguals based on their
performance. Using an extreme gradient boosting model, we were
able to achieve a balanced accuracy of 77%. High scores on a
prescriptive grammar test, a verbal fluency test, and a picture
naming test were predictive for multilingualism. The implications
of the reported results for the field and future research are
discussed.