AI summaryⓘ
The authors study how to detect special Turkish verb phrases called idiomatic light verb constructions (LVCs) that look the same as regular verb phrases but have a partly idiomatic meaning. They treat the problem like a yes-or-no question: is the phrase idiomatic or literal? They compare a supervised model based on BERTurk to different instruction-tuned large language models (LLMs) using no examples, one example, or a few examples. They find that LLMs struggle to identify idiomatic cases without examples but improve a lot with few-shot prompting, though results depend strongly on how prompts are designed. Overall, the supervised model is still strong, but with the right prompts, some LLMs can do as well or better on this task.
Turkish light verb constructions (LVCs)Multiword expressionsIdiomatic expressionsBinary classificationBERTurkLarge language models (LLMs)Zero-shot learningFew-shot promptingMetalinguistic classificationPrompt sensitivity
Authors
Sercan Karakaş, Yusuf Şimşek
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, partially idiomatic predicate. We frame Turkish LVC detection as a binary classification task (literal meaning vs. idiomatic meaning) and evaluate on a manually created controlled set (N=147) with matched negatives: out-of-domain random sentences and in-domain literal controls (NLVC), alongside LVC positives. We compare a supervised Turkish encoder baseline (BERTurk with a classifier head) to three instruction-tuned LLMs from different families under zero-shot, one-shot, and few-shot prompting, and analyze how demonstrations shift error profiles. In zero-shot, LLMs perform well on negatives but show very low LVC recall. One-shot prompting sharply improves LVC detection but can induce strong, model-specific biases, leading models to overpredict or underpredict LVCs. A richer few-shot prompt improves calibration and yields robust overall performance for GPT-OSS-20B and Qwen 2.5-14B. Overall, the results highlight substantial prompt sensitivity in Turkish metalinguistic classification: the supervised baseline remains competitive, while prompted LLMs can match or exceed it on LVCs with carefully constructed demonstrations.