PulseAugur
EN
LIVE 10:17:17

LLMs show prompt sensitivity in Turkish idiomatic classification

Researchers investigated the effectiveness of in-context learning for classifying Turkish idiomatic light verb constructions (LVCs). They compared a supervised BERTurk baseline against instruction-tuned large language models (LLMs) using zero-shot, one-shot, and few-shot prompting. While LLMs struggled with LVC recall in zero-shot, few-shot prompting with carefully constructed demonstrations improved performance, with GPT-OSS-20B and Qwen 2.5-14B showing robust results that matched or exceeded the supervised baseline. AI

IMPACT Demonstrates how prompt engineering significantly impacts LLM performance on nuanced linguistic tasks, influencing how models are deployed for specialized NLP applications.

RANK_REASON Academic paper detailing a new evaluation of LLM performance on a specific linguistic task.

Read on arXiv cs.CL →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

COVERAGE [2]

  1. arXiv cs.AI TIER_1 English(EN) · Sercan Karaka\c{s}, Yusuf \c{S}im\c{s}ek ·

    Supervision versus Demonstration-Based In-Context Learning for Multiword Expression Classification

    arXiv:2606.07479v1 Announce Type: cross Abstract: Turkish idiomatic light verb constructions (LVCs) are challenging for multiword expression processing because they often share the same surface form as fully literal verb-object combinations while functioning as a single, partiall…

  2. arXiv cs.CL TIER_1 English(EN) · Yusuf Şimşek ·

    Supervision versus Demonstration-Based In-Context Learning for Multiword Expression Classification

    Turkish idiomatic light verb constructions (LVCs) are challenging for multiword expression processing because they often share the same surface form as fully literal verb-object combinations while functioning as a single, partially idiomatic predicate. We frame Turkish LVC detect…