E02¶


Research	R0027 — Multilingual prompt engineering challenges
Run	2026-03-26
Query	Q001
Source	SRC02
Evidence	SRC02-E02
Type	Statistical

Low-resource languages show dramatic improvement with selective pre-translation

URL: https://arxiv.org/html/2502.09331v1

Extract¶

"Low-resource languages showed relative improvements exceeding 200% compared to full pre-translation (e.g., Malayalam, Telugu)." This demonstrates that the choice of prompting strategy has an outsized impact on low-resource languages, and that naive full translation to English can be counterproductive.

Relevance to Hypotheses¶

Hypothesis	Relationship	Strength
H1	Supports	Confirms major performance gaps exist, especially for low-resource languages
H2	Contradicts	200%+ improvement demonstrates significant, measurable variation
H3	Supports	The effect magnitude varies dramatically by language resource level

Context¶

The 200% figure is a relative improvement over full pre-translation, not absolute accuracy. It shows that prompting strategy choice matters more for languages where the baseline performance is poor.