@inproceedings{lam-etal-2026-encoder,
title = "Encoder-only and decoder-only models exhibit distinct patterns of linguistic bias",
author = "Lam, Suet-Ying and
Zeng, Qingcheng and
Voigt, Rob",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.673/",
pages = "15508--15520",
ISBN = "979-8-89176-459-0",
abstract = "Large language models (LLMs) based on encoder and decoder architectures exhibit strong performance across a range of tasks, yet systematic comparisons of how training objectives shape their behavior remain limited. Building on prior work establishing systematic comparison between decoder-encoder differences (Weller et al., 2026), we ask whether this difference shapes different linguistic bias using two production-related tasks: semantic role assignment and syntactic structure choice (actives vs. passives). Across architectures, LLMs exhibit a robust animacy-first bias, preferring to place animate entities in subject position, diverging from human bias. However, when animacy cues are removed, encoders generalize better than decoders and show a stronger preference for passive constructions. We argue that these differences reflect training objectives: encoders' sensitivity to contextual compatibility supports flexible mapping from semantics to structure, whereas decoders' reliance on surface frequency favors dominant patterns and disfavors infrequent structures such as passives."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lam-etal-2026-encoder">
<titleInfo>
<title>Encoder-only and decoder-only models exhibit distinct patterns of linguistic bias</title>
</titleInfo>
<name type="personal">
<namePart type="given">Suet-Ying</namePart>
<namePart type="family">Lam</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qingcheng</namePart>
<namePart type="family">Zeng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rob</namePart>
<namePart type="family">Voigt</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Large language models (LLMs) based on encoder and decoder architectures exhibit strong performance across a range of tasks, yet systematic comparisons of how training objectives shape their behavior remain limited. Building on prior work establishing systematic comparison between decoder-encoder differences (Weller et al., 2026), we ask whether this difference shapes different linguistic bias using two production-related tasks: semantic role assignment and syntactic structure choice (actives vs. passives). Across architectures, LLMs exhibit a robust animacy-first bias, preferring to place animate entities in subject position, diverging from human bias. However, when animacy cues are removed, encoders generalize better than decoders and show a stronger preference for passive constructions. We argue that these differences reflect training objectives: encoders’ sensitivity to contextual compatibility supports flexible mapping from semantics to structure, whereas decoders’ reliance on surface frequency favors dominant patterns and disfavors infrequent structures such as passives.</abstract>
<identifier type="citekey">lam-etal-2026-encoder</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.673/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>15508</start>
<end>15520</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Encoder-only and decoder-only models exhibit distinct patterns of linguistic bias
%A Lam, Suet-Ying
%A Zeng, Qingcheng
%A Voigt, Rob
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F lam-etal-2026-encoder
%X Large language models (LLMs) based on encoder and decoder architectures exhibit strong performance across a range of tasks, yet systematic comparisons of how training objectives shape their behavior remain limited. Building on prior work establishing systematic comparison between decoder-encoder differences (Weller et al., 2026), we ask whether this difference shapes different linguistic bias using two production-related tasks: semantic role assignment and syntactic structure choice (actives vs. passives). Across architectures, LLMs exhibit a robust animacy-first bias, preferring to place animate entities in subject position, diverging from human bias. However, when animacy cues are removed, encoders generalize better than decoders and show a stronger preference for passive constructions. We argue that these differences reflect training objectives: encoders’ sensitivity to contextual compatibility supports flexible mapping from semantics to structure, whereas decoders’ reliance on surface frequency favors dominant patterns and disfavors infrequent structures such as passives.
%U https://aclanthology.org/2026.emnlp-main.673/
%P 15508-15520
Markdown (Informal)
[Encoder-only and decoder-only models exhibit distinct patterns of linguistic bias](https://aclanthology.org/2026.emnlp-main.673/) (Lam et al., EMNLP 2026)
ACL