aiqbench analysis
This study benchmarks fine-tuned neural machine translation models against few-shot large language models for the challenging Arabic-Russian pair, revealing that specialized NMT architectures significantly outperform general LLMs in low-resource scenarios. The primary failure mode identified is low lexical overlap rather than morphological complexity, suggesting that current foundation models struggle with specific language pairs lacking shared vocabulary despite their broad capabilities. For researchers focusing on low-resource translation and model specialization, these findings highlight the continued necessity of task-specific fine-tuning over relying solely on zero-shot or few-shot prompting for high-stakes linguistic tasks.
Benchmarking Arabic--Russian Machine Translation: A Comparison of Fine-tuned NMT and Few-shot LLMs under Rich Morphology and Low Lexical Overlap
arXiv:2609.29559v1 Announce Type: new Abstract: Arabic-Russian machine translation (MT) remains under-explored due to the rich morphology of Arabic and low lexical overlap between the two languages. We benchmark seven fine-tuned neural machine translation (NMT) models against four few-shot large language models (LLMs) on a 20k/5k/5k split of a new 15.47M-pair corpus. Fine-tuned NLLB-1.3B achieves the highest BLEU (16.3) and COMET (0.738). Aya-Expanse 8B leads the few-shot LLMs (BLEU 1.7 on 500 sentences, chrF 25.7), but all LLM scores remain far below the fine-tuned NMT baselines. Error analysis identifies low lexical overlap as the dominant failure mode; among the worst translations, mT5-small produces 32% too-short outputs. Bootstrap tests confirm significant differences among most models. Our results demonstrate that fine-tuned NMT significantly outperforms few-shot LLMs for Arabic-Russian translation under low-resource conditions.
Abstract:Arabic-Russian machine translation (MT) remains under-explored due to the rich morphology of Arabic and low lexical overlap between the two languages. We benchmark seven fine-tuned neural machine translation (NMT) models against four few-shot large language models (LLMs) on a 20k/5k/5k split of a new 15.47M-pair corpus. Fine-tuned NLLB-1.3B achieves the highest BLEU (16.3) and COMET (0.738). Aya-Expanse 8B leads the few-shot LLMs (BLEU 1.7 on 500 sentences, chrF 25.7), but all LLM scores remain far below the fine-tuned NMT baselines. Error analysis identifies low lexical overlap as the dominant failure mode; among the worst translations, mT5-small produces 32% too-short outputs. Bootstrap tests confirm significant differences among most models. Our results demonstrate that fine-tuned NMT significantly outperforms few-shot LLMs for Arabic-Russian translation under low-resource conditions.
Preprint. 20 pages, 1 figure, 15 tables
Computation and Language (cs.CL)
arXiv:2609.29559 [cs.CL]
(or arXiv:2609.29559v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2609.29559 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 226 words
Related coverage
Links #3: 2026/06 Part 1
LessWrong · Jun 15, 2026
E-CONAN (Entailment, CONtradition And Neutral) Benchmarks: Arabic Textual Entailment and Natural Inference Datasets
arXiv · Sep 12, 2026
Fine-Tuning LLMs for Translation: General Forgetting Mitigation Does Not Preserve MT-Specific Instruction Following
arXiv · Sep 24, 2026
Bridging Scientific Heritage: An Arabic--Russian Parallel Corpus and LLM Benchmark for Sustainable Knowledge Transfer
arXiv · Sep 15, 2026