MYTE: Morphology-Driven Byte Encoding for Better and Fairer Multilingual Language Modeling
This addresses fairness and performance disparities in multilingual NLP for underrepresented languages, representing a new paradigm rather than an incremental improvement.
The paper tackled the bias in multilingual language modeling towards high-resource languages by introducing a morphology-driven byte encoding (MYTE) that encodes information with consistent segment sizes across languages, resulting in shorter encodings for all 99 analyzed languages and improved LM performance with reduced perplexity gaps.
A major consideration in multilingual language modeling is how to best represent languages with diverse vocabularies and scripts. Although contemporary text encoding methods cover most of the world's writing systems, they exhibit bias towards the high-resource languages of the Global West. As a result, texts of underrepresented languages tend to be segmented into long sequences of linguistically meaningless units. To address the disparities, we introduce a new paradigm that encodes the same information with segments of consistent size across diverse languages. Our encoding convention (MYTE) is based on morphemes, as their inventories are more balanced across languages than characters, which are used in previous methods. We show that MYTE produces shorter encodings for all 99 analyzed languages, with the most notable improvements for non-European languages and non-Latin scripts. This, in turn, improves multilingual LM performance and diminishes the perplexity gap throughout diverse languages.