@inproceedings{de-seyssel-etal-2025-discriminating,
title = "Discriminating Form and Meaning in Multilingual Models with Minimal-Pair {ABX} Tasks",
author = "de Seyssel, Maureen and
Chi, Jie and
Seto, Skyler and
Ter Hoeve, Maartje and
Fedzechkina, Masha and
Schluter, Natalie",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.emnlp-main.1210/",
doi = "10.18653/v1/2025.emnlp-main.1210",
pages = "23704--23725",
ISBN = "979-8-89176-332-6",
abstract = "We introduce a set of training-free ABX-style discrimination tasks to evaluate how multilingual language models represent language identity (form) and semantic content (meaning). Inspired from speech processing, these zero-shot tasks measure whether minimal differences in representation can be reliably detected. This offers a flexible and interpretable alternative to probing. Applied to XLM-R (Conneau et al, 2020) across pretraining checkpoints and layers, we find that language discrimination declines over training and becomes concentrated in lower layers, while meaning discrimination strengthens over time and stabilizes in deeper layers. We then explore probing tasks, showing some alignment between our metrics and linguistic learning performance.Our results position ABX tasks as a lightweight framework for analyzing the structure of multilingual representations."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="de-seyssel-etal-2025-discriminating">
<titleInfo>
<title>Discriminating Form and Meaning in Multilingual Models with Minimal-Pair ABX Tasks</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maureen</namePart>
<namePart type="family">de Seyssel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">Chi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Skyler</namePart>
<namePart type="family">Seto</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maartje</namePart>
<namePart type="family">Ter Hoeve</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Masha</namePart>
<namePart type="family">Fedzechkina</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Natalie</namePart>
<namePart type="family">Schluter</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-332-6</identifier>
</relatedItem>
<abstract>We introduce a set of training-free ABX-style discrimination tasks to evaluate how multilingual language models represent language identity (form) and semantic content (meaning). Inspired from speech processing, these zero-shot tasks measure whether minimal differences in representation can be reliably detected. This offers a flexible and interpretable alternative to probing. Applied to XLM-R (Conneau et al, 2020) across pretraining checkpoints and layers, we find that language discrimination declines over training and becomes concentrated in lower layers, while meaning discrimination strengthens over time and stabilizes in deeper layers. We then explore probing tasks, showing some alignment between our metrics and linguistic learning performance.Our results position ABX tasks as a lightweight framework for analyzing the structure of multilingual representations.</abstract>
<identifier type="citekey">de-seyssel-etal-2025-discriminating</identifier>
<identifier type="doi">10.18653/v1/2025.emnlp-main.1210</identifier>
<location>
<url>https://aclanthology.org/2025.emnlp-main.1210/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>23704</start>
<end>23725</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Discriminating Form and Meaning in Multilingual Models with Minimal-Pair ABX Tasks
%A de Seyssel, Maureen
%A Chi, Jie
%A Seto, Skyler
%A Ter Hoeve, Maartje
%A Fedzechkina, Masha
%A Schluter, Natalie
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-332-6
%F de-seyssel-etal-2025-discriminating
%X We introduce a set of training-free ABX-style discrimination tasks to evaluate how multilingual language models represent language identity (form) and semantic content (meaning). Inspired from speech processing, these zero-shot tasks measure whether minimal differences in representation can be reliably detected. This offers a flexible and interpretable alternative to probing. Applied to XLM-R (Conneau et al, 2020) across pretraining checkpoints and layers, we find that language discrimination declines over training and becomes concentrated in lower layers, while meaning discrimination strengthens over time and stabilizes in deeper layers. We then explore probing tasks, showing some alignment between our metrics and linguistic learning performance.Our results position ABX tasks as a lightweight framework for analyzing the structure of multilingual representations.
%R 10.18653/v1/2025.emnlp-main.1210
%U https://aclanthology.org/2025.emnlp-main.1210/
%U https://doi.org/10.18653/v1/2025.emnlp-main.1210
%P 23704-23725
Markdown (Informal)
[Discriminating Form and Meaning in Multilingual Models with Minimal-Pair ABX Tasks](https://aclanthology.org/2025.emnlp-main.1210/) (de Seyssel et al., EMNLP 2025)
ACL