Modeling creative and social uses of language

Placeholder Show Content


This thesis examines the computational basis of how people understand and appreciate creative uses of language. In the first half of the thesis, I describe a computational model that interprets diverse types of creative and figurative language use, such as hyperbole, irony, and metaphor. In the second half, I focus on modeling the social consequences of using language in creative ways, such as strengthening social bonds and evoking humor. Across the chapters, I describe computational models that formalize theories of language comprehension and show that these formalizations predict people's interpretations of a wide range of creative language use. Part of the goal of this thesis is to show that the interpretation of creative and social uses of language arises through many of the same rational principles that govern standard language understanding. In Chapter 2, I formalize basic principles of communication by extending the Rational Speech-Acts (RSA) framework (Frank & Goodman, 2012) with an explicit representation of speakers' communicative goals. I show that by incorporating inferences regarding which dimensions of the world the speaker wishes to address, the extended model produces appropriate interpretations for three types of figurative uses: hyperbole, verbal irony, and metaphor, suggesting that diverse figurative meanings may arise from the same principles of communication. In Chapter 3, I focus specifically on nonliteral interpretations of number words such as hyperbole and pragmatic halo. I provide a detailed account of how utterance cost, informativity, background knowledge, and communicative goals interact to produce hyperbolic and loose interpretations as well as associated affective subtexts. In Chapter 4, I explore the social consequences of figurative language use by formalizing the idea that listeners reason about common ground during communication. Through a series of simulations, I show that the extended RSA model infers a higher probability of common ground given figurative utterances than given literal ones, suggesting that figurative language may uniquely license inferences that strengthen social bonds. Finally, in Chapter 5, I use a simple model of sentence processing to derive quantitative measures of humor that are motivated by both humor theory and general theories of sentence comprehension. Taken together, this work sheds light on creative and social uses of language and advances formal approaches to language understanding, such that computational models can explain a broader range of phenomena that enrich our linguistic and social lives.


Type of resource text
Form electronic; electronic resource; remote
Extent 1 online resource.
Publication date 2016
Issuance monographic
Language English


Associated with Kao, Justine T
Associated with Stanford University, Department of Psychology.
Primary advisor Goodman, Noah
Thesis advisor Goodman, Noah
Thesis advisor Clark, Herbert H
Thesis advisor Frank, Michael C
Thesis advisor Potts, Christopher, 1977-
Advisor Clark, Herbert H
Advisor Frank, Michael C
Advisor Potts, Christopher, 1977-


Genre Theses

Bibliographic information

Statement of responsibility Justine T. Kao.
Note Submitted to the Department of Psychology.
Thesis Thesis (Ph.D.)--Stanford University, 2016.
Location electronic resource

Access conditions

© 2016 by Justine T Kao
This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).

Also listed in

Loading usage metrics...