@inproceedings{cotterell-2026-surprisal,
title = "Surprisal Theory is Tautological (without Rational Grounding)",
author = "Cotterell, Ryan",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.2096/",
pages = "47731--47747",
ISBN = "979-8-89176-459-0",
abstract = "Surprisal theory holds that the human processing difficulty of a linguistic unit is an affine function of its surprisal in context under some language model. I argue this claim is tautological, i.e., for any non-negative difficulty measure over units in context, there exists a language model whose surprisal is an affine function of it under mild technical conditions. Therefore, because any pattern of difficulty is consistent with some language model, without an additional constraint on the language model itself, surprisal theory makes no falsifiable predictions. I further claim that the tautology was obscured by an assumption implicit in two decades of psycholinguistic work {--} that the language model relevant for studying human sentence processing is the distribution that generated the training corpus. Recent empirical work, however, has undermined this assumption, demonstrating that better corpus models can be worse predictors of processing difficulty. I conclude that breaking the tautology requires a rationalist intervention, i.e., the language models used for the study of human sentence processing must be derived from an independently motivated model of the comprehender, which could be based on, for instance, memory constraints or processing goals, and that, crucially, does not depend on the behavioral data surprisal theory is meant to explain."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="cotterell-2026-surprisal">
<titleInfo>
<title>Surprisal Theory is Tautological (without Rational Grounding)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ryan</namePart>
<namePart type="family">Cotterell</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Surprisal theory holds that the human processing difficulty of a linguistic unit is an affine function of its surprisal in context under some language model. I argue this claim is tautological, i.e., for any non-negative difficulty measure over units in context, there exists a language model whose surprisal is an affine function of it under mild technical conditions. Therefore, because any pattern of difficulty is consistent with some language model, without an additional constraint on the language model itself, surprisal theory makes no falsifiable predictions. I further claim that the tautology was obscured by an assumption implicit in two decades of psycholinguistic work – that the language model relevant for studying human sentence processing is the distribution that generated the training corpus. Recent empirical work, however, has undermined this assumption, demonstrating that better corpus models can be worse predictors of processing difficulty. I conclude that breaking the tautology requires a rationalist intervention, i.e., the language models used for the study of human sentence processing must be derived from an independently motivated model of the comprehender, which could be based on, for instance, memory constraints or processing goals, and that, crucially, does not depend on the behavioral data surprisal theory is meant to explain.</abstract>
<identifier type="citekey">cotterell-2026-surprisal</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2096/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>47731</start>
<end>47747</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Surprisal Theory is Tautological (without Rational Grounding)
%A Cotterell, Ryan
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F cotterell-2026-surprisal
%X Surprisal theory holds that the human processing difficulty of a linguistic unit is an affine function of its surprisal in context under some language model. I argue this claim is tautological, i.e., for any non-negative difficulty measure over units in context, there exists a language model whose surprisal is an affine function of it under mild technical conditions. Therefore, because any pattern of difficulty is consistent with some language model, without an additional constraint on the language model itself, surprisal theory makes no falsifiable predictions. I further claim that the tautology was obscured by an assumption implicit in two decades of psycholinguistic work – that the language model relevant for studying human sentence processing is the distribution that generated the training corpus. Recent empirical work, however, has undermined this assumption, demonstrating that better corpus models can be worse predictors of processing difficulty. I conclude that breaking the tautology requires a rationalist intervention, i.e., the language models used for the study of human sentence processing must be derived from an independently motivated model of the comprehender, which could be based on, for instance, memory constraints or processing goals, and that, crucially, does not depend on the behavioral data surprisal theory is meant to explain.
%U https://aclanthology.org/2026.emnlp-main.2096/
%P 47731-47747
Markdown (Informal)
[Surprisal Theory is Tautological (without Rational Grounding)](https://aclanthology.org/2026.emnlp-main.2096/) (Cotterell, EMNLP 2026)
ACL