@inproceedings{jin-etal-2026-stage,
title = "{STAGE}: Diagnosing Semantic Transfer at Grounded Execution in Embodied Agents",
author = "Jin, Baosheng and
Liang, Yushen and
Shen, Hua",
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.1148/",
doi = "10.18653/2026.findings-emnlp.1148",
pages = "24309--24331",
ISBN = "979-8-89176-462-0",
abstract = "Embodied language grounding requires more than identifying the referent of an instruction: recovered semantics must also control the action an agent exposes. We study this missing link as a semantic{--}action gap, where instruction semantics are recoverable but weakly expressed in native continuous actions. We introduce SAT-Bench, a fixed-observation counterfactual benchmark that holds the visual scene and agent state fixed while changing only instruction semantics. On LIBERO target-name and pixel-grounded relation swaps, target recovery reaches 100.0{\%} and 95.8{\%}, whereas OpenVLA action sensitivity remains only 6.8{\%} and 7.7{\%}. The gap persists across 1,000 additional compositional and temporal/procedural counterfactuals, with overall action sensitivity of 6.1{\%}. Hidden-state, threshold-free, cross-policy, and rollout diagnostics further support this semantic{--}action transfer failure. We introduce VISA, a lightweight execution-time interface that converts recovered semantics into ALLOW, DEFER, target-consistency, and verified-selection decisions. VISA reduces invalid-instruction blind execution from 92.7{\%} to 2.8{\%} while preserving 94.0{\%} of normal commands, and verified selection further improves target-consistent action exposure without updating the underlying policy. Overall, embodied language evaluation should measure semantic{--}action transfer, not semantic parsing alone."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jin-etal-2026-stage">
<titleInfo>
<title>STAGE: Diagnosing Semantic Transfer at Grounded Execution in Embodied Agents</title>
</titleInfo>
<name type="personal">
<namePart type="given">Baosheng</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yushen</namePart>
<namePart type="family">Liang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hua</namePart>
<namePart type="family">Shen</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>Embodied language grounding requires more than identifying the referent of an instruction: recovered semantics must also control the action an agent exposes. We study this missing link as a semantic–action gap, where instruction semantics are recoverable but weakly expressed in native continuous actions. We introduce SAT-Bench, a fixed-observation counterfactual benchmark that holds the visual scene and agent state fixed while changing only instruction semantics. On LIBERO target-name and pixel-grounded relation swaps, target recovery reaches 100.0% and 95.8%, whereas OpenVLA action sensitivity remains only 6.8% and 7.7%. The gap persists across 1,000 additional compositional and temporal/procedural counterfactuals, with overall action sensitivity of 6.1%. Hidden-state, threshold-free, cross-policy, and rollout diagnostics further support this semantic–action transfer failure. We introduce VISA, a lightweight execution-time interface that converts recovered semantics into ALLOW, DEFER, target-consistency, and verified-selection decisions. VISA reduces invalid-instruction blind execution from 92.7% to 2.8% while preserving 94.0% of normal commands, and verified selection further improves target-consistent action exposure without updating the underlying policy. Overall, embodied language evaluation should measure semantic–action transfer, not semantic parsing alone.</abstract>
<identifier type="citekey">jin-etal-2026-stage</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1148</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1148/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>24309</start>
<end>24331</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T STAGE: Diagnosing Semantic Transfer at Grounded Execution in Embodied Agents
%A Jin, Baosheng
%A Liang, Yushen
%A Shen, Hua
%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 jin-etal-2026-stage
%X Embodied language grounding requires more than identifying the referent of an instruction: recovered semantics must also control the action an agent exposes. We study this missing link as a semantic–action gap, where instruction semantics are recoverable but weakly expressed in native continuous actions. We introduce SAT-Bench, a fixed-observation counterfactual benchmark that holds the visual scene and agent state fixed while changing only instruction semantics. On LIBERO target-name and pixel-grounded relation swaps, target recovery reaches 100.0% and 95.8%, whereas OpenVLA action sensitivity remains only 6.8% and 7.7%. The gap persists across 1,000 additional compositional and temporal/procedural counterfactuals, with overall action sensitivity of 6.1%. Hidden-state, threshold-free, cross-policy, and rollout diagnostics further support this semantic–action transfer failure. We introduce VISA, a lightweight execution-time interface that converts recovered semantics into ALLOW, DEFER, target-consistency, and verified-selection decisions. VISA reduces invalid-instruction blind execution from 92.7% to 2.8% while preserving 94.0% of normal commands, and verified selection further improves target-consistent action exposure without updating the underlying policy. Overall, embodied language evaluation should measure semantic–action transfer, not semantic parsing alone.
%R 10.18653/2026.findings-emnlp.1148
%U https://aclanthology.org/2026.findings-emnlp.1148/
%U https://doi.org/10.18653/2026.findings-emnlp.1148
%P 24309-24331
Markdown (Informal)
[STAGE: Diagnosing Semantic Transfer at Grounded Execution in Embodied Agents](https://aclanthology.org/2026.findings-emnlp.1148/) (Jin et al., Findings 2026)
ACL