Wine is Not v i n. On the Compatibility of Tokenizations Across Languages

Maronikolakis, Antonis; Dufter, Philipp and Schütze, Hinrich (November 2021): Wine is Not v i n. On the Compatibility of Tokenizations Across Languages. EMNLP 2021, Punta Cana, Dominican Republic, November 7–11, 2021. Moens, Marie-Francine; Huang, Xuanjing; Specia, Lucia and Yih, Scott Wen-tau (eds.) : In: Findings of the Association for Computational Linguistics: EMNLP 2021, Stroudsburg, PA: Association for Computational Linguistics. pp. 2382-2399

Abstract

The size of the vocabulary is a central design choice in large pretrained language models, with respect to both performance and memory requirements. Typically, subword tokenization algorithms such as byte pair encoding and WordPiece are used. In this work, we investigate the compatibility of tokenizations for multilingual static and contextualized embedding spaces and propose a measure that reflects the compatibility of tokenizations across languages. Our goal is to prevent incompatible tokenizations, e.g., “wine” (word-level) in English vs. “v i n” (character-level) in French, which make it hard to learn good multilingual semantic representations. We show that our compatibility measure allows the system designer to create vocabularies across languages that are compatible – a desideratum that so far has been neglected in multilingual models.

• Item Type: Conference or Workshop Item (Paper)
• EU Funded Grant Agreement Number: 740516
• EU Projects: Horizon 2020 > ERC Grants > ERC Advanced Grant > ERC Grant 740516: NonSequeToR - Non-sequence models for tokenization replacement
• Research Centers: Center for Information and Language Processing (CIS)
• Subjects: 000 Computer science, information and general works > 000 Computer science, knowledge, and systems; 400 Language > 400 Language; 400 Language > 410 Linguistics
• Place of Publication: Stroudsburg, PA
• Language: English