| Part of language | English |
|---|---|
| Meaning of the construction |
The central senses of argument structure construction have been argued to be in association with humanly relevant scenes like: transferring |
| comment | The main data includes 2000 sentences that are generated by GPT-4, which is a text file. The extension data includes a representation of these data in LLM and brain, Representation in LLMs are matrixes including hidden layer activity and attention factor. The representation in the brain includes EEG recordings from 10 participants while listening to 200 of these sentences. Both are matrixes in the format of NumPy arrays but can be converted to any supporting format. This is a similar publicly available dataset: https://github.com/SPOClab-ca/neural-reality-constructions |
| inherited by | Ditransitive construction | English |
| inherited by | Resultative construction | English |
| inherited by | Transitive construction | English |
| inherited by | Caused-Motion construction | English |
| Findings | So far we have results from LLM analysis that can significantly recognize and differentiate between argument structure constructions. |
|---|
| Date Created | 2025-02-14 |
|---|---|
| Annotator | Pegah Ramezani |
| References | Adele E. Goldberg (1995). Constructions: A Construction Grammar Approach to Argument Structure. |
| References | Achim Schilling, Patrick Krauss & Pegah Ramezani (2025). Analysis of Argument Structure Constructions in the Large Language Model BERT. https://doi.org/10.3389/frai.2025.1477246 |