@inproceedings{gargova-etal-2026-multilingual,
title = "A Multilingual Linguistic Analysis of Human vs {LLM}-Generated News in a Disinformation Context",
author = "Gargova, Silvia and
Perez-Montero, Alba and
Lloret Pastor, Elena and
Moreda Pozo, Paloma",
editor = "Frenda, Simona and
Stranisci, Marco Antonio and
Ashraf, Shaina and
Ren, Ada and
Konstas, Ioannis and
Naseem, Usman",
booktitle = "Proceedings of the 1st Workshop on Information Disorder ({I}n{D}or) @ {LREC} 2026",
month = may,
year = "2026",
address = "Palma de Mallorca, Spain",
publisher = "ELRA Language Resources Association (ELRA)",
url = "https://aclanthology.org/2026.indor-1.9/",
doi = "10.63317/345wj5pp9nkh",
pages = "85--96",
ISBN = "978-2-493814-87-6",
abstract = "The rise of Large Language Models has shifted the Information Disorder landscape toward automated threats. This study investigates the linguistic construction of synthetic news by comparing GPT-5, Gemini 2.5, and Grok 4 across English, Spanish, and Bulgarian. Using multilingual human-authored verified news and disinformation as seeds, we analyze how prompt informativeness and model architecture influence deceptive content production. Our methodology employs five metrics: semantic similarity, factual consistency, readability, lexical richness, and persuasion technique frequency. Our analysis reveals that while prompt scarcity leads to informational loss, LLMs maintain a homogenized stylistic template regardless of input length. Unlike human authors, who intensify rhetorical and emotional markers to drive deceptive intent, LLMs adhere to a neutral register. This study identifies distinct statistical patterns in generated content characterized by hyper-standardized readability and high lexical density (p {\ensuremath{<}} 0.001). These features serve as robust ``LLM signatures'', enabling a classification accuracy of 96{\%} across English, Spanish, and Bulgarian. These findings suggest that generated disinformation relies on invariant syntactic structures rather than nuanced human rhetoric, providing a framework for detection tools centered on structural patterns rather than content veracity."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gargova-etal-2026-multilingual">
<titleInfo>
<title>A Multilingual Linguistic Analysis of Human vs LLM-Generated News in a Disinformation Context</title>
</titleInfo>
<name type="personal">
<namePart type="given">Silvia</namePart>
<namePart type="family">Gargova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alba</namePart>
<namePart type="family">Perez-Montero</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Elena</namePart>
<namePart type="family">Lloret Pastor</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Paloma</namePart>
<namePart type="family">Moreda Pozo</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 1st Workshop on Information Disorder (InDor) @ LREC 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Simona</namePart>
<namePart type="family">Frenda</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marco</namePart>
<namePart type="given">Antonio</namePart>
<namePart type="family">Stranisci</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shaina</namePart>
<namePart type="family">Ashraf</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ada</namePart>
<namePart type="family">Ren</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ioannis</namePart>
<namePart type="family">Konstas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Usman</namePart>
<namePart type="family">Naseem</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>ELRA Language Resources Association (ELRA)</publisher>
<place>
<placeTerm type="text">Palma de Mallorca, Spain</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">978-2-493814-87-6</identifier>
</relatedItem>
<abstract>The rise of Large Language Models has shifted the Information Disorder landscape toward automated threats. This study investigates the linguistic construction of synthetic news by comparing GPT-5, Gemini 2.5, and Grok 4 across English, Spanish, and Bulgarian. Using multilingual human-authored verified news and disinformation as seeds, we analyze how prompt informativeness and model architecture influence deceptive content production. Our methodology employs five metrics: semantic similarity, factual consistency, readability, lexical richness, and persuasion technique frequency. Our analysis reveals that while prompt scarcity leads to informational loss, LLMs maintain a homogenized stylistic template regardless of input length. Unlike human authors, who intensify rhetorical and emotional markers to drive deceptive intent, LLMs adhere to a neutral register. This study identifies distinct statistical patterns in generated content characterized by hyper-standardized readability and high lexical density (p \ensuremath< 0.001). These features serve as robust “LLM signatures”, enabling a classification accuracy of 96% across English, Spanish, and Bulgarian. These findings suggest that generated disinformation relies on invariant syntactic structures rather than nuanced human rhetoric, providing a framework for detection tools centered on structural patterns rather than content veracity.</abstract>
<identifier type="citekey">gargova-etal-2026-multilingual</identifier>
<identifier type="doi">10.63317/345wj5pp9nkh</identifier>
<location>
<url>https://aclanthology.org/2026.indor-1.9/</url>
</location>
<part>
<date>2026-05</date>
<extent unit="page">
<start>85</start>
<end>96</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T A Multilingual Linguistic Analysis of Human vs LLM-Generated News in a Disinformation Context
%A Gargova, Silvia
%A Perez-Montero, Alba
%A Lloret Pastor, Elena
%A Moreda Pozo, Paloma
%Y Frenda, Simona
%Y Stranisci, Marco Antonio
%Y Ashraf, Shaina
%Y Ren, Ada
%Y Konstas, Ioannis
%Y Naseem, Usman
%S Proceedings of the 1st Workshop on Information Disorder (InDor) @ LREC 2026
%D 2026
%8 May
%I ELRA Language Resources Association (ELRA)
%C Palma de Mallorca, Spain
%@ 978-2-493814-87-6
%F gargova-etal-2026-multilingual
%X The rise of Large Language Models has shifted the Information Disorder landscape toward automated threats. This study investigates the linguistic construction of synthetic news by comparing GPT-5, Gemini 2.5, and Grok 4 across English, Spanish, and Bulgarian. Using multilingual human-authored verified news and disinformation as seeds, we analyze how prompt informativeness and model architecture influence deceptive content production. Our methodology employs five metrics: semantic similarity, factual consistency, readability, lexical richness, and persuasion technique frequency. Our analysis reveals that while prompt scarcity leads to informational loss, LLMs maintain a homogenized stylistic template regardless of input length. Unlike human authors, who intensify rhetorical and emotional markers to drive deceptive intent, LLMs adhere to a neutral register. This study identifies distinct statistical patterns in generated content characterized by hyper-standardized readability and high lexical density (p \ensuremath< 0.001). These features serve as robust “LLM signatures”, enabling a classification accuracy of 96% across English, Spanish, and Bulgarian. These findings suggest that generated disinformation relies on invariant syntactic structures rather than nuanced human rhetoric, providing a framework for detection tools centered on structural patterns rather than content veracity.
%R 10.63317/345wj5pp9nkh
%U https://aclanthology.org/2026.indor-1.9/
%U https://doi.org/10.63317/345wj5pp9nkh
%P 85-96
Markdown (Informal)
[A Multilingual Linguistic Analysis of Human vs LLM-Generated News in a Disinformation Context](https://aclanthology.org/2026.indor-1.9/) (Gargova et al., InDor 2026)
ACL