An entropy-based discrimination between translation and second language production - Frontiersin.org
This study investigates to what extent second language production (L2) and translational production are influenced by cognitive constraints due to their bilingual processing.
Compared with monolingual production in first language (L1), the two bilingual productions are subject to the same cognitive constraints due to the co-activation of two linguistic systems and the language contact interference, though they involve two distinctive conceptualization stages, which may cause variations to their cognitive constraint.
Entropy is utilized as an indicator of cognitive constraints in the study to illuminate how bilingual processing causes cognitive constraints to the two bilingual productions. Specifically, word and part-of-speech n-gram features are computed in the machine learning models to compare the three productions.
The results show that L1 production could be effectively distinguished with the two bilingual productions, with L1 production exhibiting greater diversity and a more even distribution in most features than L2 production and translational production. This is clear evidence that both L2 production and translational production endure more cognitive load in the bilingual processing than L1 production.
The results also reveal that L2 production and translational production could be discriminated against each other, with the former one exhibiting greater diversity and a more even distribution in most features than the latter one. These findings support the idea that both translators and L2 users belong to bilinguals affected by the cognitive load in bilingual processing although the two bilingual productions are constrained differently.