Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models

Boxuan Wang; Mariët Theune; Sumit Srivastava

doi:10.1007/978-3-031-54975-5_6

Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models

Boxuan Wang, Mariët Theune, Sumit Srivastava^*

^*Corresponding author for this work

Human Media Interaction

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic › peer-review

20 Downloads (Pure)

Abstract

This study employs a quantitative approach to investigate lexical alignment in human-agent interactions involving GPT-3.5 and GPT-4 language models. The research examines alignment performances across different conversational contexts and compares the performance of the two models. The findings highlight the significant improvements in GPT-4’s ability to foster lexical alignment, and the influence of conversation topics on alignment patterns.

Original language	English
Title of host publication	Chatbot Research and Design
Subtitle of host publication	7th International Workshop, CONVERSATIONS 2023 Oslo, Norway, November 22–23, 2023 Revised Selected Papers
Editors	Asbjørn Følstad, Theo Araujo, Symeon Papadopoulos, Effie Law, Ewa Luger, Morten Goodwin, Sebastian Hobert, Petter Bae Brandtzaeg
Publisher	Springer
Pages	94-114
Number of pages	21
ISBN (Electronic)	978-3-031-54975-5
ISBN (Print)	978-3-031-54974-8
DOIs	https://doi.org/10.1007/978-3-031-54975-5_6
Publication status	Published - 2024
Event	7th International Workshop on Chatbot Research and Design, CONVERSATIONS 2023 - Oslo, Norway Duration: 22 Nov 2023 → 23 Nov 2023 Conference number: 7

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer Nature
Volume	14524
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	7th International Workshop on Chatbot Research and Design, CONVERSATIONS 2023
Abbreviated title	CONVERSATIONS 2023
Country/Territory	Norway
City	Oslo
Period	22/11/23 → 23/11/23

Keywords

2024 OA procedure
linguistic accomodation
linguistic style matching
linguistic alignment
lexical alignment
human-agent interaction
GPT3.5
GPT4

Access to Document

10.1007/978-3-031-54975-5_6

ArticleFinal published version, 0.98 MBLicence: Taverne

Cite this

Wang, B., Theune, M., & Srivastava, S. (2024). Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models. In A. Følstad, T. Araujo, S. Papadopoulos, E. Law, E. Luger, M. Goodwin, S. Hobert, & P. B. Brandtzaeg (Eds.), Chatbot Research and Design: 7th International Workshop, CONVERSATIONS 2023 Oslo, Norway, November 22–23, 2023 Revised Selected Papers (pp. 94-114). (Lecture Notes in Computer Science; Vol. 14524). Springer. https://doi.org/10.1007/978-3-031-54975-5_6

Wang, Boxuan ; Theune, Mariët ; Srivastava, Sumit. / Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models. Chatbot Research and Design: 7th International Workshop, CONVERSATIONS 2023 Oslo, Norway, November 22–23, 2023 Revised Selected Papers. editor / Asbjørn Følstad ; Theo Araujo ; Symeon Papadopoulos ; Effie Law ; Ewa Luger ; Morten Goodwin ; Sebastian Hobert ; Petter Bae Brandtzaeg . Springer, 2024. pp. 94-114 (Lecture Notes in Computer Science).

@inbook{cba6dc00f2184ff9ac4a19fc5161c48d,

title = "Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models",

abstract = "This study employs a quantitative approach to investigate lexical alignment in human-agent interactions involving GPT-3.5 and GPT-4 language models. The research examines alignment performances across different conversational contexts and compares the performance of the two models. The findings highlight the significant improvements in GPT-4{\textquoteright}s ability to foster lexical alignment, and the influence of conversation topics on alignment patterns.",

keywords = "2024 OA procedure, linguistic accomodation, linguistic style matching, linguistic alignment, lexical alignment, human-agent interaction, GPT3.5, GPT4",

author = "Boxuan Wang and Mari{\"e}t Theune and Sumit Srivastava",

year = "2024",

doi = "10.1007/978-3-031-54975-5_6",

language = "English",

isbn = "978-3-031-54974-8",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "94--114",

editor = "F{\o}lstad, {Asbj{\o}rn } and Theo Araujo and Papadopoulos, {Symeon } and Effie Law and Ewa Luger and Goodwin, {Morten } and Hobert, {Sebastian } and {Brandtzaeg }, {Petter Bae }",

booktitle = "Chatbot Research and Design",

address = "Germany",

note = "7th International Workshop on Chatbot Research and Design, CONVERSATIONS 2023, CONVERSATIONS 2023 ; Conference date: 22-11-2023 Through 23-11-2023",

}

Wang, B, Theune, M & Srivastava, S 2024, Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models. in A Følstad, T Araujo, S Papadopoulos, E Law, E Luger, M Goodwin, S Hobert & PB Brandtzaeg (eds), Chatbot Research and Design: 7th International Workshop, CONVERSATIONS 2023 Oslo, Norway, November 22–23, 2023 Revised Selected Papers. Lecture Notes in Computer Science, vol. 14524, Springer, pp. 94-114, 7th International Workshop on Chatbot Research and Design, CONVERSATIONS 2023, Oslo, Norway, 22/11/23. https://doi.org/10.1007/978-3-031-54975-5_6

Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models. / Wang, Boxuan; Theune, Mariët ; Srivastava, Sumit.
Chatbot Research and Design: 7th International Workshop, CONVERSATIONS 2023 Oslo, Norway, November 22–23, 2023 Revised Selected Papers. ed. / Asbjørn Følstad; Theo Araujo; Symeon Papadopoulos; Effie Law; Ewa Luger; Morten Goodwin; Sebastian Hobert; Petter Bae Brandtzaeg . Springer, 2024. p. 94-114 (Lecture Notes in Computer Science; Vol. 14524).

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic › peer-review

TY - CHAP

T1 - Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models

AU - Wang, Boxuan

AU - Theune, Mariët

AU - Srivastava, Sumit

N1 - Conference code: 7

PY - 2024

Y1 - 2024

N2 - This study employs a quantitative approach to investigate lexical alignment in human-agent interactions involving GPT-3.5 and GPT-4 language models. The research examines alignment performances across different conversational contexts and compares the performance of the two models. The findings highlight the significant improvements in GPT-4’s ability to foster lexical alignment, and the influence of conversation topics on alignment patterns.

AB - This study employs a quantitative approach to investigate lexical alignment in human-agent interactions involving GPT-3.5 and GPT-4 language models. The research examines alignment performances across different conversational contexts and compares the performance of the two models. The findings highlight the significant improvements in GPT-4’s ability to foster lexical alignment, and the influence of conversation topics on alignment patterns.

KW - 2024 OA procedure

KW - linguistic accomodation

KW - linguistic style matching

KW - linguistic alignment

KW - lexical alignment

KW - human-agent interaction

KW - GPT3.5

KW - GPT4

U2 - 10.1007/978-3-031-54975-5_6

DO - 10.1007/978-3-031-54975-5_6

M3 - Chapter

SN - 978-3-031-54974-8

T3 - Lecture Notes in Computer Science

SP - 94

EP - 114

BT - Chatbot Research and Design

A2 - Følstad, Asbjørn

A2 - Araujo, Theo

A2 - Papadopoulos, Symeon

A2 - Law, Effie

A2 - Luger, Ewa

A2 - Goodwin, Morten

A2 - Hobert, Sebastian

A2 - Brandtzaeg , Petter Bae

PB - Springer

T2 - 7th International Workshop on Chatbot Research and Design, CONVERSATIONS 2023

Y2 - 22 November 2023 through 23 November 2023

ER -

Wang B, Theune M , Srivastava S. Examining Lexical Alignment in Human-Agent Conversations with GPT-3.5 and GPT-4 Models. In Følstad A, Araujo T, Papadopoulos S, Law E, Luger E, Goodwin M, Hobert S, Brandtzaeg PB, editors, Chatbot Research and Design: 7th International Workshop, CONVERSATIONS 2023 Oslo, Norway, November 22–23, 2023 Revised Selected Papers. Springer. 2024. p. 94-114. (Lecture Notes in Computer Science). Epub 2024 Mar 13. doi: 10.1007/978-3-031-54975-5_6