Cross-lingual Word Embeddings beyond Zero-shot Machine Translation
This addresses the problem of extending machine translation to low-resource languages for researchers and practitioners, but it is incremental as it builds on existing multilingual models.
The study investigated whether a multilingual neural machine translation model can transfer translation knowledge to unseen languages using only cross-lingual word embeddings, finding weak transfer that depends on language relatedness.
We explore the transferability of a multilingual neural machine translation model to unseen languages when the transfer is grounded solely on the cross-lingual word embeddings. Our experimental results show that the translation knowledge can transfer weakly to other languages and that the degree of transferability depends on the languages' relatedness. We also discuss the limiting aspects of the multilingual architectures that cause weak translation transfer and suggest how to mitigate the limitations.