New Benchmark Reveals LLM Weakness in Distinguishing Similar Arabic-Hebrew Words
Researchers introduced SemCog Bench, a benchmark of 1,858 Arabic-Hebrew word pairs, to test how well large language models distinguish between true cognates, false friends, and loanwords in these related Semitic languages. Current LLMs perform well on true cognates but fail significantly on false friends and loanwords, showing they rely too heavily on surface-level similarity rather than semantic understanding. The findings highlight a fundamental limitation in cross-lingual reasoning that could affect multilingual AI applications.
Researchers have created SemCog Bench, a curated benchmark containing 1,858 Arabic-Hebrew word pairs with sentence-level annotations, to evaluate how large language models handle cross-lingual semantic understanding between two closely related Semitic languages. Testing both open-source and commercial LLMs across multiple input formats (raw text, diacritized, Romanized, and phonetic), the study found that models achieve high accuracy on true cognates but performance drops sharply when distinguishing false friends and loanwords. The research reveals that LLMs rely heavily on surface-form similarity and that providing sentence-level context yields only modest improvements, suggesting that contextual information alone cannot overcome misleading formal similarities. These findings establish a critical gap in current LLM capabilities for resolving cross-lingual form-meaning conflicts and provide a rigorous benchmark for future work in multilingual semantic reasoning.
What's missing
The study does not discuss potential implications for real-world applications such as machine translation, cross-lingual information retrieval, or multilingual chatbots, nor does it propose specific solutions to address the identified limitations.
What different sources said
- arXiv cs.CLCenter
When Similar Means Different: Evaluating LLMs on Arabic--Hebrew Cognates
Related
Topology-Aware Thermodynamics Improves DNA Probe Specificity Design
Researchers developed a new framework for designing DNA probes that accounts for the spatial organization of matched sequences, not just overall thermodynamic stability. Traditional methods rely on scalar measures like melting temperature and free energy, which miss how mismatches are distributed along the probe. The approach could improve diagnostic accuracy in applications like HPV detection and gene expression profiling.
Study Identifies Optimal Thermal Dose for Combining Focused Ultrasound with Immunotherapy in Tumors
Researchers used multimodal PET imaging to identify an optimal thermal dose range for focused ultrasound ablation that destroys tumor tissue while preserving conditions for immunotherapy delivery. The study found that excessive heating collapses blood vessels needed for antibody access, while insufficient heating fails to adequately reduce tumor burden. The findings could guide clinical design of combination treatments pairing thermal ablation with immunotherapies.
Plant MSH1 Protein Functions as Mismatch-Directed Nuclease for Organelle Genome Maintenance
Researchers have identified the precise mechanism by which the AtMSH1 protein in Arabidopsis plants recognizes and cleaves DNA mismatches and lesions, preventing mutations in organellar genomes. The protein combines a DNA mismatch recognition module with a nuclease domain that makes staggered cuts at specific positions relative to DNA damage. This discovery explains how plants maintain unusually low mutation rates in their mitochondrial and chloroplast DNA compared to other eukaryotes.