Why don't people use character-level machine translation?

TL;DR

This paper critically examines the use of character-level models in machine translation, revealing they are less effective than subword models in competitive settings despite some robustness advantages.

Contribution

It provides a comprehensive literature review and empirical analysis showing the limitations of character-level MT systems compared to subword systems.

Findings

Character-level MT systems are less competitive than subword systems in WMT.

They do not exhibit superior domain robustness or morphological generalization.

Character-level systems are robust to source noise and unaffected by larger beam sizes.

Abstract

We present a literature and empirical survey that critically assesses the state of the art in character-level modeling for machine translation (MT). Despite evidence in the literature that character-level systems are comparable with subword systems, they are virtually never used in competitive setups in WMT competitions. We empirically show that even with recent modeling innovations in character-level natural language processing, character-level MT systems still struggle to match their subword-based counterparts. Character-level MT systems show neither better domain robustness, nor better morphological generalization, despite being often so motivated. However, we are able to show robustness towards source side noise and that translation quality does not degrade with increasing beam size at decoding time.