<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="trigg-etal-2026-logic">
    <titleInfo>
        <title>Logic-Level Evaluation of Logical Table-to-Text Generation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Lena</namePart>
        <namePart type="family">Trigg</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Dean</namePart>
        <namePart type="given">F</namePart>
        <namePart type="family">Hougen</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ahsan</namePart>
        <namePart type="family">Bilal</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 30th Conference on Computational Natural Language Learning</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Claire</namePart>
            <namePart type="family">Bonial</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Yevgeni</namePart>
            <namePart type="family">Berzak</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">San Diego, California, USA</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-410-1</identifier>
    </relatedItem>
    <abstract>Logical Table-to-Text (LT2T) generation aims to produce natural-language sentences that are logically faithful to structured tabular data. While recent Large Language Models (LLMs) show high performance on aggregate fidelity metrics, these scores provide only a coarse view of performance, obscuring specific logic-type reasoning failures and models’ meta-logical awareness. We propose an operation-aware diagnostic framework that evaluates four core competencies: (1) Logical Form (LF) execution accuracy, (2) fidelity of LF-conditioned generation, (3) logic-type identification, and (4) LF-free generation.We apply this framework to a suite of frontier LLMs and perform fine-grained analysis across logic types such as aggregation, ordinal, and superlative reasoning. Our results show that LT2T fidelity assessment can be unstable; the choice of verifier and logic type can substantially alter conclusions and model rankings. Crucially, we identify a meta-logical gap: models often generate faithful statements while failing to identify the underlying operation.</abstract>
    <identifier type="citekey">trigg-etal-2026-logic</identifier>
    <location>
        <url>https://aclanthology.org/2026.conll-main.41/</url>
    </location>
    <part>
        <date>2026-07</date>
        <extent unit="page">
            <start>677</start>
            <end>691</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Logic-Level Evaluation of Logical Table-to-Text Generation
%A Trigg, Lena
%A Hougen, Dean F.
%A Bilal, Ahsan
%Y Bonial, Claire
%Y Berzak, Yevgeni
%S Proceedings of the 30th Conference on Computational Natural Language Learning
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, USA
%@ 979-8-89176-410-1
%F trigg-etal-2026-logic
%X Logical Table-to-Text (LT2T) generation aims to produce natural-language sentences that are logically faithful to structured tabular data. While recent Large Language Models (LLMs) show high performance on aggregate fidelity metrics, these scores provide only a coarse view of performance, obscuring specific logic-type reasoning failures and models’ meta-logical awareness. We propose an operation-aware diagnostic framework that evaluates four core competencies: (1) Logical Form (LF) execution accuracy, (2) fidelity of LF-conditioned generation, (3) logic-type identification, and (4) LF-free generation.We apply this framework to a suite of frontier LLMs and perform fine-grained analysis across logic types such as aggregation, ordinal, and superlative reasoning. Our results show that LT2T fidelity assessment can be unstable; the choice of verifier and logic type can substantially alter conclusions and model rankings. Crucially, we identify a meta-logical gap: models often generate faithful statements while failing to identify the underlying operation.
%U https://aclanthology.org/2026.conll-main.41/
%P 677-691

Markdown (Informal)

[Logic-Level Evaluation of Logical Table-to-Text Generation](https://aclanthology.org/2026.conll-main.41/) (Trigg et al., CoNLL 2026)

• Logic-Level Evaluation of Logical Table-to-Text Generation (Trigg et al., CoNLL 2026)

ACL

• Lena Trigg, Dean F. Hougen, and Ahsan Bilal. 2026. Logic-Level Evaluation of Logical Table-to-Text Generation. In Proceedings of the 30th Conference on Computational Natural Language Learning, pages 677–691, San Diego, California, USA. Association for Computational Linguistics.