@inproceedings{singh-etal-2019-bert,
    title = "{BERT} is Not an Interlingua and the Bias of Tokenization",
    author = "Singh, Jasdeep  and
      McCann, Bryan  and
      Socher, Richard  and
      Xiong, Caiming",
    editor = "Cherry, Colin  and
      Durrett, Greg  and
      Foster, George  and
      Haffari, Reza  and
      Khadivi, Shahram  and
      Peng, Nanyun  and
      Ren, Xiang  and
      Swayamdipta, Swabha",
    booktitle = "Proceedings of the 2nd Workshop on Deep Learning Approaches for Low-Resource NLP (DeepLo 2019)",
    month = nov,
    year = "2019",
    address = "Hong Kong, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/iwcs-25-ingestion/D19-6106/",
    doi = "10.18653/v1/D19-6106",
    pages = "47--55",
    abstract = "Multilingual transfer learning can benefit both high- and low-resource languages, but the source of these improvements is not well understood. Cananical Correlation Analysis (CCA) of the internal representations of a pre- trained, multilingual BERT model reveals that the model partitions representations for each language rather than using a common, shared, interlingual space. This effect is magnified at deeper layers, suggesting that the model does not progressively abstract semantic con- tent while disregarding languages. Hierarchical clustering based on the CCA similarity scores between languages reveals a tree structure that mirrors the phylogenetic trees hand- designed by linguists. The subword tokenization employed by BERT provides a stronger bias towards such structure than character- and word-level tokenizations. We release a subset of the XNLI dataset translated into an additional 14 languages at \url{https://www.github.com/salesforce/xnli_extension} to assist further research into multilingual representations."
}