Cross-lingual Alignment Methods for Multilingual BERT: A Comparative Study
This work addresses the challenge of enhancing cross-lingual performance in NLP for multilingual applications, but it is incremental as it builds on existing alignment methods.
The paper tackled the problem of improving zero-shot cross-lingual transfer in multilingual BERT by comparing alignment methods and supervision types, finding that parallel corpus supervision is generally superior and proposing a normalization method that yields a 3% F1 improvement for distant languages.
Multilingual BERT (mBERT) has shown reasonable capability for zero-shot cross-lingual transfer when fine-tuned on downstream tasks. Since mBERT is not pre-trained with explicit cross-lingual supervision, transfer performance can further be improved by aligning mBERT with cross-lingual signal. Prior work proposes several approaches to align contextualised embeddings. In this paper we analyse how different forms of cross-lingual supervision and various alignment methods influence the transfer capability of mBERT in zero-shot setting. Specifically, we compare parallel corpora vs. dictionary-based supervision and rotational vs. fine-tuning based alignment methods. We evaluate the performance of different alignment methodologies across eight languages on two tasks: Name Entity Recognition and Semantic Slot Filling. In addition, we propose a novel normalisation method which consistently improves the performance of rotation-based alignment including a notable 3% F1 improvement for distant and typologically dissimilar languages. Importantly we identify the biases of the alignment methods to the type of task and proximity to the transfer language. We also find that supervision from parallel corpus is generally superior to dictionary alignments.