One-Shot Neural Cross-Lingual Transfer for Paradigm Completion
This addresses the challenge of morphological inflection for low-resource languages, though it is incremental as it builds on existing neural encoder-decoder models.
The paper tackles the problem of paradigm completion for low-resource languages by using cross-lingual transfer from high-resource languages, achieving up to 58% higher accuracy and enabling zero-shot and one-shot learning in experiments across 21 language pairs.
We present a novel cross-lingual transfer method for paradigm completion, the task of mapping a lemma to its inflected forms, using a neural encoder-decoder model, the state of the art for the monolingual task. We use labeled data from a high-resource language to increase performance on a low-resource language. In experiments on 21 language pairs from four different language families, we obtain up to 58% higher accuracy than without transfer and show that even zero-shot and one-shot learning are possible. We further find that the degree of language relatedness strongly influences the ability to transfer morphological knowledge.