Could We Have Had Better Multilingual LLMs If English Was Not the Central Language?

TL;DR

This study investigates whether alternative central languages to English could enhance multilingual LLM performance, revealing that linguistic similarities and model scaling influence translation quality more than training data alone.

Contribution

It introduces a linear model linking linguistic features to translation scores, challenging the English-centric paradigm in multilingual LLM development.

Findings

Scaling up model size improves translation into unseen languages.

Syntactic similarity correlates with translation performance beyond language data size.

Languages like Swedish and Catalan perform well despite less training data.

Abstract

Large Language Models (LLMs) demonstrate strong machine translation capabilities on languages they are trained on. However, the impact of factors beyond training data size on translation performance remains a topic of debate, especially concerning languages not directly encountered during training. Our study delves into Llama2's translation capabilities. By modeling a linear relationship between linguistic feature distances and machine translation scores, we ask ourselves if there are potentially better central languages for LLMs other than English. Our experiments show that the 7B Llama2 model yields above 10 BLEU when translating into all languages it has seen, which rarely happens for languages it has not seen. Most translation improvements into unseen languages come from scaling up the model size rather than instruction tuning or increasing shot count. Furthermore, our correlation analysis reveals that syntactic similarity is not the only linguistic factor that strongly correlates with machine translation scores. Interestingly, we discovered that under specific circumstances, some languages (e.g. Swedish, Catalan), despite having significantly less training data, exhibit comparable correlation levels to English. These insights challenge the prevailing landscape of LLMs, suggesting that models centered around languages other than English could provide a more efficient foundation for multilingual applications.