@inproceedings{tuck-verma-2026-orthographic,
title = "Orthographic Constraint Satisfaction and Human Difficulty Alignment in Large Language Models",
author = "Tuck, Bryan E. and
Verma, Rakesh",
editor = "Piperidis, Stelios and
Bel, N{\'u}ria and
van den Heuvel, Henk and
Ide, Nancy and
Krek, Simon and
Toral, Antonio",
booktitle = "Proceedings of the Fifteenth Language Resources and Evaluation Conference",
month = may,
year = "2026",
address = "Palma de Mallorca, Spain",
publisher = "ELRA Language Resource Association",
url = "https://aclanthology.org/2026.lrec-1.426/",
doi = "10.63317/3erhoom72odv",
pages = "5466--5481",
abstract = "Large language models must satisfy hard orthographic constraints during controlled text generation, yet systematic cross-family evaluation remains limited. We evaluate 39 configurations spanning three model families (Qwen3, Claude Haiku 4.5, GPT-5-mini) on 58 word puzzles requiring character-level constraint satisfaction. Cross-family differences produce substantially larger performance gaps (2.0{--}2.2{\texttimes}, F1 = 0.761 vs. 0.343) than parameter scaling within families (83{\%} gain from 4B to 32B scaling), and a partial-correlation analysis rules out tokenizer design as a confound for within-family scaling. Thinking budget sensitivity proves heterogeneous: high-capacity models show strong returns (+0.102 to +0.136 F1), while mid-sized variants saturate or degrade, showing inconsistent compute benefits. Using difficulty ratings from 10,000 human solvers per puzzle, we establish modest but consistent calibration ({\ensuremath{\rho}} = 0.28{--}0.42) across all families, yet identify systematic failures on common words with unusual orthography ({``}data'', ``loll'', ``acai'': 83{--}91{\%} human success, 94{--}98{\%} model miss rate). These failures point to over-reliance on distributional plausibility that penalizes orthographically atypical but constraint-valid patterns."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tuck-verma-2026-orthographic">
<titleInfo>
<title>Orthographic Constraint Satisfaction and Human Difficulty Alignment in Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Bryan</namePart>
<namePart type="given">E</namePart>
<namePart type="family">Tuck</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rakesh</namePart>
<namePart type="family">Verma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Fifteenth Language Resources and Evaluation Conference</title>
</titleInfo>
<name type="personal">
<namePart type="given">Stelios</namePart>
<namePart type="family">Piperidis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Núria</namePart>
<namePart type="family">Bel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Henk</namePart>
<namePart type="family">van den Heuvel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nancy</namePart>
<namePart type="family">Ide</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Simon</namePart>
<namePart type="family">Krek</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Antonio</namePart>
<namePart type="family">Toral</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>ELRA Language Resource Association</publisher>
<place>
<placeTerm type="text">Palma de Mallorca, Spain</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Large language models must satisfy hard orthographic constraints during controlled text generation, yet systematic cross-family evaluation remains limited. We evaluate 39 configurations spanning three model families (Qwen3, Claude Haiku 4.5, GPT-5-mini) on 58 word puzzles requiring character-level constraint satisfaction. Cross-family differences produce substantially larger performance gaps (2.0–2.2×, F1 = 0.761 vs. 0.343) than parameter scaling within families (83% gain from 4B to 32B scaling), and a partial-correlation analysis rules out tokenizer design as a confound for within-family scaling. Thinking budget sensitivity proves heterogeneous: high-capacity models show strong returns (+0.102 to +0.136 F1), while mid-sized variants saturate or degrade, showing inconsistent compute benefits. Using difficulty ratings from 10,000 human solvers per puzzle, we establish modest but consistent calibration (\ensuremathρ = 0.28–0.42) across all families, yet identify systematic failures on common words with unusual orthography (“data”, “loll”, “acai”: 83–91% human success, 94–98% model miss rate). These failures point to over-reliance on distributional plausibility that penalizes orthographically atypical but constraint-valid patterns.</abstract>
<identifier type="citekey">tuck-verma-2026-orthographic</identifier>
<identifier type="doi">10.63317/3erhoom72odv</identifier>
<location>
<url>https://aclanthology.org/2026.lrec-1.426/</url>
</location>
<part>
<date>2026-05</date>
<extent unit="page">
<start>5466</start>
<end>5481</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Orthographic Constraint Satisfaction and Human Difficulty Alignment in Large Language Models
%A Tuck, Bryan E.
%A Verma, Rakesh
%Y Piperidis, Stelios
%Y Bel, Núria
%Y van den Heuvel, Henk
%Y Ide, Nancy
%Y Krek, Simon
%Y Toral, Antonio
%S Proceedings of the Fifteenth Language Resources and Evaluation Conference
%D 2026
%8 May
%I ELRA Language Resource Association
%C Palma de Mallorca, Spain
%F tuck-verma-2026-orthographic
%X Large language models must satisfy hard orthographic constraints during controlled text generation, yet systematic cross-family evaluation remains limited. We evaluate 39 configurations spanning three model families (Qwen3, Claude Haiku 4.5, GPT-5-mini) on 58 word puzzles requiring character-level constraint satisfaction. Cross-family differences produce substantially larger performance gaps (2.0–2.2×, F1 = 0.761 vs. 0.343) than parameter scaling within families (83% gain from 4B to 32B scaling), and a partial-correlation analysis rules out tokenizer design as a confound for within-family scaling. Thinking budget sensitivity proves heterogeneous: high-capacity models show strong returns (+0.102 to +0.136 F1), while mid-sized variants saturate or degrade, showing inconsistent compute benefits. Using difficulty ratings from 10,000 human solvers per puzzle, we establish modest but consistent calibration (\ensuremathρ = 0.28–0.42) across all families, yet identify systematic failures on common words with unusual orthography (“data”, “loll”, “acai”: 83–91% human success, 94–98% model miss rate). These failures point to over-reliance on distributional plausibility that penalizes orthographically atypical but constraint-valid patterns.
%R 10.63317/3erhoom72odv
%U https://aclanthology.org/2026.lrec-1.426/
%U https://doi.org/10.63317/3erhoom72odv
%P 5466-5481
Markdown (Informal)
[Orthographic Constraint Satisfaction and Human Difficulty Alignment in Large Language Models](https://aclanthology.org/2026.lrec-1.426/) (Tuck & Verma, LREC 2026)
ACL