@inproceedings{kezar-etal-2026-surveying,
title = "Surveying Representation Collapse in Pretrained Models of {A}merican {S}ign {L}anguage",
author = "Kezar, Lee and
Aichler, Laurel and
Willis, Athena S. and
Imai, Saki and
Inan, Mert and
Wooten, Alicia K. and
Alikhani, Malihe and
Quandt, Lorna",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1976/",
pages = "42132--42146",
ISBN = "979-8-89176-462-0",
abstract = "The American Sign Language (ASL) lexicon contains heterogeneous sign types, including established signed words, neologisms, ad-hoc depictions of real-world phenomena, and fingerspelled words. Productions within each type exhibit unique phonotactics and substantial variability, potentially confounding models pretrained on downstream ASL tasks. To survey the capacity for representation collapse, we isolate and annotate 14 types of ASL productions as they appear in science, technology, engineering, and math (STEM) educational content. We then probe seven pretrained models' latent representations of the annotated clips, reporting collapse metrics with respect to domain, sign type, and meaning. Our findings show that out-of-distribution signs are more frequent in STEM content, and that pretrained models lose information about their visual and semantic contents in varing ways. Among the tested models, those which encoded the phonological properties of signs generalized to STEM data more effectively, while English- and gesture-based models can exhibit more brittleness. Future work can design interventions to mitigate the kinds of collapse documented here or extend this survey by evaluating additional models and datasets."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kezar-etal-2026-surveying">
<titleInfo>
<title>Surveying Representation Collapse in Pretrained Models of American Sign Language</title>
</titleInfo>
<name type="personal">
<namePart type="given">Lee</namePart>
<namePart type="family">Kezar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Laurel</namePart>
<namePart type="family">Aichler</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Athena</namePart>
<namePart type="given">S</namePart>
<namePart type="family">Willis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Saki</namePart>
<namePart type="family">Imai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mert</namePart>
<namePart type="family">Inan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alicia</namePart>
<namePart type="given">K</namePart>
<namePart type="family">Wooten</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Malihe</namePart>
<namePart type="family">Alikhani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lorna</namePart>
<namePart type="family">Quandt</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>The American Sign Language (ASL) lexicon contains heterogeneous sign types, including established signed words, neologisms, ad-hoc depictions of real-world phenomena, and fingerspelled words. Productions within each type exhibit unique phonotactics and substantial variability, potentially confounding models pretrained on downstream ASL tasks. To survey the capacity for representation collapse, we isolate and annotate 14 types of ASL productions as they appear in science, technology, engineering, and math (STEM) educational content. We then probe seven pretrained models’ latent representations of the annotated clips, reporting collapse metrics with respect to domain, sign type, and meaning. Our findings show that out-of-distribution signs are more frequent in STEM content, and that pretrained models lose information about their visual and semantic contents in varing ways. Among the tested models, those which encoded the phonological properties of signs generalized to STEM data more effectively, while English- and gesture-based models can exhibit more brittleness. Future work can design interventions to mitigate the kinds of collapse documented here or extend this survey by evaluating additional models and datasets.</abstract>
<identifier type="citekey">kezar-etal-2026-surveying</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1976/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>42132</start>
<end>42146</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Surveying Representation Collapse in Pretrained Models of American Sign Language
%A Kezar, Lee
%A Aichler, Laurel
%A Willis, Athena S.
%A Imai, Saki
%A Inan, Mert
%A Wooten, Alicia K.
%A Alikhani, Malihe
%A Quandt, Lorna
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F kezar-etal-2026-surveying
%X The American Sign Language (ASL) lexicon contains heterogeneous sign types, including established signed words, neologisms, ad-hoc depictions of real-world phenomena, and fingerspelled words. Productions within each type exhibit unique phonotactics and substantial variability, potentially confounding models pretrained on downstream ASL tasks. To survey the capacity for representation collapse, we isolate and annotate 14 types of ASL productions as they appear in science, technology, engineering, and math (STEM) educational content. We then probe seven pretrained models’ latent representations of the annotated clips, reporting collapse metrics with respect to domain, sign type, and meaning. Our findings show that out-of-distribution signs are more frequent in STEM content, and that pretrained models lose information about their visual and semantic contents in varing ways. Among the tested models, those which encoded the phonological properties of signs generalized to STEM data more effectively, while English- and gesture-based models can exhibit more brittleness. Future work can design interventions to mitigate the kinds of collapse documented here or extend this survey by evaluating additional models and datasets.
%U https://aclanthology.org/2026.findings-emnlp.1976/
%P 42132-42146
Markdown (Informal)
[Surveying Representation Collapse in Pretrained Models of American Sign Language](https://aclanthology.org/2026.findings-emnlp.1976/) (Kezar et al., Findings 2026)
ACL
- Lee Kezar, Laurel Aichler, Athena S. Willis, Saki Imai, Mert Inan, Alicia K. Wooten, Malihe Alikhani, and Lorna Quandt. 2026. Surveying Representation Collapse in Pretrained Models of American Sign Language. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 42132–42146, Budapest, Hungary. Association for Computational Linguistics.