Search papers, labs, and topics across Lattice.
This study evaluates the effectiveness of multilingual sentence embeddings as a replacement for translated English inputs in Linguistic-Integrated Reliability Auditing (LiRA) across 11 PIRLS constructed-response items. The findings indicate that native-language embeddings can closely replicate the reliability estimates derived from traditional translation methods while also recovering responses that were previously excluded due to translation failures. Importantly, the use of these embeddings does not significantly alter the reliability outcomes, suggesting a viable alternative to translation in multilingual assessment systems.
Native-language embeddings can match translation-based reliability estimates while recovering lost responses, challenging the necessity of translation in multilingual assessments.
Multilingual assessment systems commonly rely on translation for scoring and quality-control processes. We evaluate whether multilingual sentence embeddings can replace translated English input for Linguistic-Integrated Reliability Auditing (LiRA) across 11 PIRLS constructed-response items and three embedding models. Native-language embeddings reproduced translation-based reliability estimates closely while recovering responses excluded after translation failure, with no meaningful change in reliability.
