When your Cousin has the Right Connections: Unsupervised Bilingual Lexicon Induction for Related Data-Imbalanced Languages
This addresses the challenge of building bilingual resources for low-resource languages against related high-resource ones, which is crucial for language preservation and NLP applications but often neglected due to data scarcity.
The paper tackled the problem of unsupervised bilingual lexicon induction for low-resource languages paired with related high-resource languages, where data imbalance causes near-zero performance in existing methods, and introduced a new method using only a masked language model of the high-resource language, achieving effective results on languages like Bhojpuri and Magahi with less than 5M tokens each.
Most existing approaches for unsupervised bilingual lexicon induction (BLI) depend on good quality static or contextual embeddings requiring large monolingual corpora for both languages. However, unsupervised BLI is most likely to be useful for low-resource languages (LRLs), where large datasets are not available. Often we are interested in building bilingual resources for LRLs against related high-resource languages (HRLs), resulting in severely imbalanced data settings for BLI. We first show that state-of-the-art BLI methods in the literature exhibit near-zero performance for severely data-imbalanced language pairs, indicating that these settings require more robust techniques. We then present a new method for unsupervised BLI between a related LRL and HRL that only requires inference on a masked language model of the HRL, and demonstrate its effectiveness on truly low-resource languages Bhojpuri and Magahi (with <5M monolingual tokens each), against Hindi. We further present experiments on (mid-resource) Marathi and Nepali to compare approach performances by resource range, and release our resulting lexicons for five low-resource Indic languages: Bhojpuri, Magahi, Awadhi, Braj, and Maithili, against Hindi.