On Difficulties of Cross-Lingual Transfer with Order Differences: A Case Study on Dependency Parsing
This addresses the problem of improving NLP tools for low-resource languages by enhancing cross-lingual transfer, though it is incremental as it builds on existing architectures.
The paper tackled cross-lingual transfer challenges in dependency parsing due to word order differences, finding that self-attentive models outperform RNN-based ones, especially on distant languages, with experiments on 30 languages.
Different languages might have different word orders. In this paper, we investigate cross-lingual transfer and posit that an order-agnostic model will perform better when transferring to distant foreign languages. To test our hypothesis, we train dependency parsers on an English corpus and evaluate their transfer performance on 30 other languages. Specifically, we compare encoders and decoders based on Recurrent Neural Networks (RNNs) and modified self-attentive architectures. The former relies on sequential information while the latter is more flexible at modeling word order. Rigorous experiments and detailed analysis shows that RNN-based architectures transfer well to languages that are close to English, while self-attentive models have better overall cross-lingual transferability and perform especially well on distant languages.