@inproceedings{rodriguez-etal-2024-neural,
title = "Do Neural Language Models Inferentially Compose Concepts the Way Humans Can?",
author = {Rodriguez, Amilleah and
Wang, Shaonan and
Pylkk{\"a}nen, Liina},
editor = "Calzolari, Nicoletta and
Kan, Min-Yen and
Hoste, Veronique and
Lenci, Alessandro and
Sakti, Sakriani and
Xue, Nianwen",
booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)",
month = may,
year = "2024",
address = "Torino, Italia",
publisher = "ELRA and ICCL",
url = "https://aclanthology.org/2024.lrec-main.472/",
pages = "5309--5314",
abstract = "While compositional interpretation is the core of language understanding, humans also derive meaning via inference. For example, while the phrase {\textquotedblleft}the blue hat{\textquotedblright} introduces a blue hat into the discourse via the direct composition of {\textquotedblleft}blue{\textquotedblright} and {\textquotedblleft}hat,{\textquotedblright} the same discourse entity is introduced by the phrase {\textquotedblleft}the blue color of this hat{\textquotedblright} despite the absence of any local composition between {\textquotedblleft}blue{\textquotedblright} and {\textquotedblleft}hat.{\textquotedblright} Instead, we infer that if the color is blue and it belongs to the hat, the hat must be blue. We tested the performance of neural language models and humans on such inferentially driven conceptual compositions, eliciting probability estimates for a noun in a minimally composed phrase, {\textquotedblleft}This blue hat{\textquotedblright}, following contexts that had introduced the conceptual combinations of those nouns and adjectives either syntactically or inferentially. Surprisingly, our findings reveal significant disparities between the performance of neural language models and human judgments. Among the eight models evaluated, RoBERTa, BERT-large, and GPT-2 exhibited the closest resemblance to human responses, while other models faced challenges in accurately identifying compositions in the provided contexts. Our study reveals that language models and humans may rely on different approaches to represent and compose lexical items across sentence structure. All data and code are accessible at https://github.com/wangshaonan/BlueHat."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="rodriguez-etal-2024-neural">
<titleInfo>
<title>Do Neural Language Models Inferentially Compose Concepts the Way Humans Can?</title>
</titleInfo>
<name type="personal">
<namePart type="given">Amilleah</namePart>
<namePart type="family">Rodriguez</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shaonan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liina</namePart>
<namePart type="family">Pylkkänen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nicoletta</namePart>
<namePart type="family">Calzolari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Min-Yen</namePart>
<namePart type="family">Kan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Veronique</namePart>
<namePart type="family">Hoste</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alessandro</namePart>
<namePart type="family">Lenci</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sakriani</namePart>
<namePart type="family">Sakti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nianwen</namePart>
<namePart type="family">Xue</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>ELRA and ICCL</publisher>
<place>
<placeTerm type="text">Torino, Italia</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>While compositional interpretation is the core of language understanding, humans also derive meaning via inference. For example, while the phrase “the blue hat” introduces a blue hat into the discourse via the direct composition of “blue” and “hat,” the same discourse entity is introduced by the phrase “the blue color of this hat” despite the absence of any local composition between “blue” and “hat.” Instead, we infer that if the color is blue and it belongs to the hat, the hat must be blue. We tested the performance of neural language models and humans on such inferentially driven conceptual compositions, eliciting probability estimates for a noun in a minimally composed phrase, “This blue hat”, following contexts that had introduced the conceptual combinations of those nouns and adjectives either syntactically or inferentially. Surprisingly, our findings reveal significant disparities between the performance of neural language models and human judgments. Among the eight models evaluated, RoBERTa, BERT-large, and GPT-2 exhibited the closest resemblance to human responses, while other models faced challenges in accurately identifying compositions in the provided contexts. Our study reveals that language models and humans may rely on different approaches to represent and compose lexical items across sentence structure. All data and code are accessible at https://github.com/wangshaonan/BlueHat.</abstract>
<identifier type="citekey">rodriguez-etal-2024-neural</identifier>
<location>
<url>https://aclanthology.org/2024.lrec-main.472/</url>
</location>
<part>
<date>2024-05</date>
<extent unit="page">
<start>5309</start>
<end>5314</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Do Neural Language Models Inferentially Compose Concepts the Way Humans Can?
%A Rodriguez, Amilleah
%A Wang, Shaonan
%A Pylkkänen, Liina
%Y Calzolari, Nicoletta
%Y Kan, Min-Yen
%Y Hoste, Veronique
%Y Lenci, Alessandro
%Y Sakti, Sakriani
%Y Xue, Nianwen
%S Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)
%D 2024
%8 May
%I ELRA and ICCL
%C Torino, Italia
%F rodriguez-etal-2024-neural
%X While compositional interpretation is the core of language understanding, humans also derive meaning via inference. For example, while the phrase “the blue hat” introduces a blue hat into the discourse via the direct composition of “blue” and “hat,” the same discourse entity is introduced by the phrase “the blue color of this hat” despite the absence of any local composition between “blue” and “hat.” Instead, we infer that if the color is blue and it belongs to the hat, the hat must be blue. We tested the performance of neural language models and humans on such inferentially driven conceptual compositions, eliciting probability estimates for a noun in a minimally composed phrase, “This blue hat”, following contexts that had introduced the conceptual combinations of those nouns and adjectives either syntactically or inferentially. Surprisingly, our findings reveal significant disparities between the performance of neural language models and human judgments. Among the eight models evaluated, RoBERTa, BERT-large, and GPT-2 exhibited the closest resemblance to human responses, while other models faced challenges in accurately identifying compositions in the provided contexts. Our study reveals that language models and humans may rely on different approaches to represent and compose lexical items across sentence structure. All data and code are accessible at https://github.com/wangshaonan/BlueHat.
%U https://aclanthology.org/2024.lrec-main.472/
%P 5309-5314
Markdown (Informal)
[Do Neural Language Models Inferentially Compose Concepts the Way Humans Can?](https://aclanthology.org/2024.lrec-main.472/) (Rodriguez et al., LREC-COLING 2024)
ACL