Capacity Constraints and the Multilingual Penalty for Lexical Disambiguation
Sean Trott, Pamela D. Rivière · 2026
Read Capacity Constraints and the Multilingual Penalty for Lexical Disambiguation by Sean Trott, Pamela D. Rivière (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-05035.txt.
Description
Multilingual language models often trail their single-language cousins, and Sean Trott and Pamela D Rivière measure one slice of that gap in this 2026 computational linguistics paper. Their test is lexical disambiguation: judging which meaning of an ambiguous word fits, using human relatedness ratings in English and Spanish.
Multilingual models scored consistently worse than monolingual ones from the same families. Three possible bottlenecks were then checked: less orderly embeddings, weaker attention to the clues that fix a word's meaning, and more words split into extra tokens. The multilingual models showed signs of all three, and together the three factors statistically explained the penalty once credited to multilingual status alone. The findings suggest the gap is a capacity problem with identifiable parts.
Catalog record
| CREATOR | Sean Trott, Pamela D. Rivière |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-05035