Skip to content
LING 582 (FA 2025)
GitHub

Paper Summary Assignment

Author: jakemns

— paper summary — 1 min read

Citation

He, W., Vieira, T. K., Garcia, M., Scarton, C., Idiart, M., & Villavicencio, A. (2025). Investigating idiomaticity in word representations. Computational Linguistics, 51(2), 505–555. https://doi.org/10.1162/coli_a_00546

Citation summary
PaperInvestigating Idiomaticity in Word Representations
AuthorsWei He, Tiago Kramer Vieira, Marcos Garcia, Carolina Scarton, Marco Idiart, Aline Villavicencio
Year published2025
VenueCL
Paper URLhttps://aclanthology.org/2025.cl-2.4/
Code URLhttps://github.com/risehnhew/Finding-Idiomaticity-in-Word-Representations

Description

This paper analyzes current models' ability to process idiomatic expressions, measuring Affinity and Scaled Similarity. The authors found that current models are lacking in this area, and they explore the limitations that currently exist.

Motivation

Why did you select this paper? I love the concept of idioms in general in linguistics and am interested in reading more about it. I think that this phenomenon is particularly interested in the lense of ML.

LLM

LLM model summary
LLM model???
LLM model version???
Model/service URL???
Why this model????

Description (LLM)

In the LLM's words, what is this paper about?

Prompt
1prompt here

What are the authors proposing?

Prompt
1prompt here

What is the motivation for the work?

Prompt
1prompt here

What is the approach or innovation?

Prompt
1prompt here

What are the results and how do they compare with competing approaches?

Prompt
1prompt here

Is the comparison fair?

Prompt
1prompt here

What are the takeaways according to the authors?

Prompt
1prompt here

What are the takeaways according to you?

Prompt
1prompt here

Would you use this? If so, how/where would you use this?

Prompt
1prompt here
Prompt
1prompt here

What problems remain and what are the next steps?

Prompt
1prompt here

Experience using the LLM

Describe your process for using the LLM. How did the LLM perform?

Errors and limitations of the LLM

Where did it fall short or make mistakes?