CLAISep 22, 2025

Scaling, Simplification, and Adaptation: Lessons from Pretraining on Machine-Translated Text

arXiv:2509.17317v11 citationsh-index: 5
Originality Incremental advance
AI Analysis

This addresses the data scarcity issue for lower-resource languages in NLP, offering practical insights for model adaptation, though it is incremental in exploring translation-based pretraining.

The study tackled the problem of limited monolingual data for pretraining in lower-resource languages by using machine-translated text, finding that scaling model capacity improves performance, source-side simplification harms generalization, and adaptation on native text often outperforms native-only models, with specific gains like better accuracy in some tasks but challenges in cultural nuance tasks.

Most languages lack sufficient data for large-scale monolingual pretraining, creating a "data wall." Multilingual pretraining helps but is limited by language imbalance and the "curse of multilinguality." An alternative is to translate high-resource text with machine translation (MT), which raises three questions: (1) How does MT-derived data scale with model capacity? (2) Can source-side transformations (e.g., simplifying English with an LLM) improve generalization to native text? (3) How well do models pretrained on MT-derived data adapt when continually trained on limited native text? We investigate these questions by translating English into Indonesian and Tamil--two typologically distant, lower-resource languages--and pretraining GPT-2 models (124M-774M) on native or MT-derived corpora from raw and LLM-simplified English. We evaluate cross-entropy loss on native text, along with accuracy on syntactic probes and downstream tasks. Our results show that (1) MT-pretrained models benefit from scaling; (2) source-side simplification harms generalization to native text; and (3) adapting MT-pretrained models on native text often yields better performance than native-only models, even with less native data. However, tasks requiring cultural nuance (e.g., toxicity detection) demand more exposure to native data.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes