@inproceedings{afli-frank-2026-benchmark,
title = "The Benchmark Is Not the Construct: Toward Distribution-Based Evaluation in {NLP}",
author = "Afli, Haithem and
Frank, Manuel",
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.1992/",
pages = "42441--42465",
ISBN = "979-8-89176-462-0",
abstract = "Current NLP evaluation often treats a scalar score on a fixed benchmark dataset as if it directly measured the construct the benchmark is intended to represent. This Measurement-as-Instance paradigm is appropriate for a fixed evaluation configuration, but can suppress uncertainty, variance, and robustness when a single result is used to support broader capability or generalisation claims. In this position paper, we argue for extending static evaluation with Measurement-as-Distribution: a structured evaluation procedure involving principled data generation, validated perturbations, repeated evaluation, and distributional reporting. The framework is configurable: the amount of distributional evidence required should scale with the strength and breadth of the claims being made. We synthesise existing robustness, dynamic-evaluation, prompt-sensitivity, and uncertainty-reporting practices into a general measurement-theoretic framework for distribution-based NLP evaluation and provide a call to action for more rigorous reporting. Using the protocol $\mathcal{P} = (\mathcal{C}, G, \Phi, E, R)$, we identify three risks of relying exclusively on fixed scalar evaluation and derive implications for peer review, benchmark publication, leaderboard reporting, and computational cost. Two case studies demonstrate feasibility: a SQuAD evaluation of 14 models reveals modest but axis-specific performance changes and stable overall rankings with uncertainty-aware comparisons, while a verifier-backed Knights-and-Knaves study illustrates how perturbations of rigid logical tasks can be generated and validated formally."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="afli-frank-2026-benchmark">
<titleInfo>
<title>The Benchmark Is Not the Construct: Toward Distribution-Based Evaluation in NLP</title>
</titleInfo>
<name type="personal">
<namePart type="given">Haithem</namePart>
<namePart type="family">Afli</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Manuel</namePart>
<namePart type="family">Frank</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>Current NLP evaluation often treats a scalar score on a fixed benchmark dataset as if it directly measured the construct the benchmark is intended to represent. This Measurement-as-Instance paradigm is appropriate for a fixed evaluation configuration, but can suppress uncertainty, variance, and robustness when a single result is used to support broader capability or generalisation claims. In this position paper, we argue for extending static evaluation with Measurement-as-Distribution: a structured evaluation procedure involving principled data generation, validated perturbations, repeated evaluation, and distributional reporting. The framework is configurable: the amount of distributional evidence required should scale with the strength and breadth of the claims being made. We synthesise existing robustness, dynamic-evaluation, prompt-sensitivity, and uncertainty-reporting practices into a general measurement-theoretic framework for distribution-based NLP evaluation and provide a call to action for more rigorous reporting. Using the protocol \mathcalP = (\mathcalC, G, Φ, E, R), we identify three risks of relying exclusively on fixed scalar evaluation and derive implications for peer review, benchmark publication, leaderboard reporting, and computational cost. Two case studies demonstrate feasibility: a SQuAD evaluation of 14 models reveals modest but axis-specific performance changes and stable overall rankings with uncertainty-aware comparisons, while a verifier-backed Knights-and-Knaves study illustrates how perturbations of rigid logical tasks can be generated and validated formally.</abstract>
<identifier type="citekey">afli-frank-2026-benchmark</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1992/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>42441</start>
<end>42465</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T The Benchmark Is Not the Construct: Toward Distribution-Based Evaluation in NLP
%A Afli, Haithem
%A Frank, Manuel
%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 afli-frank-2026-benchmark
%X Current NLP evaluation often treats a scalar score on a fixed benchmark dataset as if it directly measured the construct the benchmark is intended to represent. This Measurement-as-Instance paradigm is appropriate for a fixed evaluation configuration, but can suppress uncertainty, variance, and robustness when a single result is used to support broader capability or generalisation claims. In this position paper, we argue for extending static evaluation with Measurement-as-Distribution: a structured evaluation procedure involving principled data generation, validated perturbations, repeated evaluation, and distributional reporting. The framework is configurable: the amount of distributional evidence required should scale with the strength and breadth of the claims being made. We synthesise existing robustness, dynamic-evaluation, prompt-sensitivity, and uncertainty-reporting practices into a general measurement-theoretic framework for distribution-based NLP evaluation and provide a call to action for more rigorous reporting. Using the protocol \mathcalP = (\mathcalC, G, Φ, E, R), we identify three risks of relying exclusively on fixed scalar evaluation and derive implications for peer review, benchmark publication, leaderboard reporting, and computational cost. Two case studies demonstrate feasibility: a SQuAD evaluation of 14 models reveals modest but axis-specific performance changes and stable overall rankings with uncertainty-aware comparisons, while a verifier-backed Knights-and-Knaves study illustrates how perturbations of rigid logical tasks can be generated and validated formally.
%U https://aclanthology.org/2026.findings-emnlp.1992/
%P 42441-42465
Markdown (Informal)
[The Benchmark Is Not the Construct: Toward Distribution-Based Evaluation in NLP](https://aclanthology.org/2026.findings-emnlp.1992/) (Afli & Frank, Findings 2026)
ACL