Supervision versus Demonstration-Based In-Context Learning for Multiword Expression Classification
Can AI learn to spot hidden idioms by example instead of training data?
When large language models are shown just one or two examples of Turkish idioms in prompts, they dramatically improve at recognizing them—but only if the examples are chosen carefully. A traditional supervised model performed roughly as well overall, suggesting that examples matter more than scale for this particular language task.
Turkish and many other languages rely heavily on idioms that look identical to literal phrases, making them genuinely hard to classify. This research shows that current AI systems struggle with this distinction unless they receive well-designed guidance, and that bigger models aren't automatically better at it. For anyone building translation tools or search systems for Turkish, the findings suggest investing in smarter example selection might work better than simply scaling up.