Periodic Reporting for period 3 - SMIC (Formal Models of Social Meaning and Identity Construction through Language)
Período documentado: 2023-02-01 hasta 2024-07-31
The main objective of the SMIC project is to construct an empirically well-founded, mathematically explicit, and computationally implemented theory of social meaning and the role that socially meaningful linguistic expressions play in the construction of personal identities. The term "social meaning" refers to the information about the properties of the speaker that are communicated to the listener through the speaker's use of a particular pronunciation, word or syntactic structure drawn from a set of linguistic alternatives (called "sociolinguistic variables" in the literature). In other words, "social meaning" corresponds to the subtle informational differences that exist between, for example, pronouncing an English word like working as `work[ın]' (-in’) versus `work[ıŋ]' (-ing), using or omitting the negative particle "ne" in a French sentence like "Je ne l’aime pas" vs "Je l’aime pas" ‘I don’t like it’, or using the French pronoun "tu" vs "vous" when addressing a colleague. The term "identity construction (through language)" refers to the process through which speakers use socially meaningful linguistic expressions to communicate information about themselves to their interlocutors and, in doing so, establish and reinforce their place in their communities.
Social meaning and identity construction are theoretical notions that are central to many disciplines of the humanities and social sciences, playing a key role in our understanding of linguistic and non-linguistic phenomena studied in anthropology, sociology, philosophy, psychology and gender studies. Moreover, methods based on identity construction through language are widely employed outside academia, for example in education, management, social work, digital communication, and even social justice. Given how important and widespread these notions are, it is crucial that our understanding of their concepts and the relationship between them be as detailed, as explicit and as well-founded as possible. Nevertheless, we are still a long way away from such an understanding. In particular, linguistic communication and identity construction through language are extremely complex cognitive and social processes, and a lot of open issues in the study of language, variation and identity are very subtle. The SMIC project proposes that formalization can be a powerful tool for carefully distinguishing different aspects of theoretical proposals in this area and for identifying empirical predictions made by competing analyses. Additionally, although there is general agreement that identity plays an important role in human behavior, because of their lack of formalization, insights from identity oriented theories of language have remained isolated from work in cognitive science, computer science and artificial intelligence, which limits the scope of work in these areas.
SMIC proposes that "game-theoretic pragmatics" provides a promising framework for formalizing current sociolinguistic theories and integrating them into cognitive science. There was initial interest in marrying mathematical and ethnographic studies of linguistic interaction in the 1970s, when game-theoretic methods started to become widely used outside economics. However, this interest never flourished into a full-fledged research program, primarily because the models accessible to scholars in the humanities at the time were not adapted for modeling language use and interpretation. Fortunately, in the past decade, much progress has been made in the application of epistemic game theory to linguistic communication, particularly the use of signaling games and a Bayesian approach to reasoning under uncertainty. The goal of the project is to build on these recent advances and create a fully formalized game-theoretic framework modeling social meaning, and to situate the study of identity construction at the intersection of the humanities, cognitive and information sciences.
Social meaning and identity construction are theoretical notions that are central to many disciplines of the humanities and social sciences, playing a key role in our understanding of linguistic and non-linguistic phenomena studied in anthropology, sociology, philosophy, psychology and gender studies. Moreover, methods based on identity construction through language are widely employed outside academia, for example in education, management, social work, digital communication, and even social justice. Given how important and widespread these notions are, it is crucial that our understanding of their concepts and the relationship between them be as detailed, as explicit and as well-founded as possible. Nevertheless, we are still a long way away from such an understanding. In particular, linguistic communication and identity construction through language are extremely complex cognitive and social processes, and a lot of open issues in the study of language, variation and identity are very subtle. The SMIC project proposes that formalization can be a powerful tool for carefully distinguishing different aspects of theoretical proposals in this area and for identifying empirical predictions made by competing analyses. Additionally, although there is general agreement that identity plays an important role in human behavior, because of their lack of formalization, insights from identity oriented theories of language have remained isolated from work in cognitive science, computer science and artificial intelligence, which limits the scope of work in these areas.
SMIC proposes that "game-theoretic pragmatics" provides a promising framework for formalizing current sociolinguistic theories and integrating them into cognitive science. There was initial interest in marrying mathematical and ethnographic studies of linguistic interaction in the 1970s, when game-theoretic methods started to become widely used outside economics. However, this interest never flourished into a full-fledged research program, primarily because the models accessible to scholars in the humanities at the time were not adapted for modeling language use and interpretation. Fortunately, in the past decade, much progress has been made in the application of epistemic game theory to linguistic communication, particularly the use of signaling games and a Bayesian approach to reasoning under uncertainty. The goal of the project is to build on these recent advances and create a fully formalized game-theoretic framework modeling social meaning, and to situate the study of identity construction at the intersection of the humanities, cognitive and information sciences.
The SMIC project combines novel empirical studies of social meaning, sociolinguistic variation and identity construction through language with formal game-theoretic modeling of the data observed in these studies. The empirical studies take two shapes: psycholinguistic experiments, and quantitative corpus studies. The work in the first half of SMIC is divided into three phases.
The first phase aims at developing a picture of social meaning and sociolinguistic variation in Paris, a large francophone city in Europe. In collaboration with Postdoc Martin, we conducted matched-guise technique experiments investigating Parisian listeners' interpretations of a series of sociolinguistic variables: future temporal reference ("Je vais manger" vs "Je mangerai" "I will eat"), subject doubling ("Jean est là" vs "Jean il est là" "Jean is there") and liaison (ex. pronouncing (or not) the "P" in "trop important" "too important"). In a matched-guise experiment, participants listen to samples of recorded speech (called guises) that are designed to match as much as possible, differing only in the grammatical alternation studied. Each participant is exposed to only one of the guises, and after hearing it, their beliefs and attitudes towards the speaker are assessed, usually via questionnaire. Additionally, in collaboration with PhD student Liang, we conducted quantitative corpus studies investigating Parisian speakers' use of future temporal reference, subject doubling, and complementizer omission ("Je pense que Jean est là" vs "Je pense Jean est là" "I think (that) Jean is there"), as well as grammatical gender and inclusive language (eg. "les étudiants" vs "les étudiant.e.s" "the (gender neutral) students"). We found, as a result of these empirical investigations, that the social meanings associated with these variables, particularly liaison, are highly dependent on the social context of the utterance. This context-sensitivity could not be modeled by previously developed game-theoretic models (such as Burnett 2017, 2019). Therefore, as per the main objectives of the project, we worked to develop a new one. In collaboration with Postdocs Martin and Abbou, we developed (what we call) the "Pragmatic Sociolinguistics" game-theoretic framework. This was obtained by formalizing the "Pragmatic Sociology" non-formalized sociological framework of Boltanski & Thévenot (1991), applying it to language, and integrating it with Burnett's previous models. The development of this framework constitutes the second main result of the project so far.
The second phase of SMIC aims at comparing the results found in the first phase with the results of similar corpus and experimental studies on speakers in Montréal, a large francophone city in North America. These comparative studies are still in progress, in collaboration with PhD student Liang; however, we expect them to yield more information about how the social and cultural context of utterance affects sociolinguistic use and interpretation. These studies will also allow us to test and refine our "Pragmatic Sociolinguistics" game-theoretic framework.
The third phase of SMIC goes deeper into how the social and cultural context affects social meaning through quantitative corpus and experimental studies of feminist activists in Paris, Montréal and Marseille. Since there were no existing corpora of feminist activists in these cities, in collaboration with Postdoc Abbou, we constructed them by doing sociolinguistic interviews in the three cities. The resulting corpus, called CaFé "Cartographies linguistiques du féminisme", is the third main result of the project. Quantitative studies of sociolinguistic variables such as grammatical gender and inclusive language, and feminist social group labels ("queer" vs "materialist" vs "intersectional" etc.) are currently in progress. In order further refine the notion of "social and cultural context" within feminist activism, we conducted psycholinguistic categorization experiments aimed at empirically inferring feminists social and political ideologies, which, in principle, could be integrated into the "Pragmatic Sociolinguistics" framework. These experiments involved applying the "Conceptual Spaces" methodology, commonly used in cognitive science to model perceptual concepts, to the highly complex social domain of French feminism. The fact that we were successful in adapting this methodology in a sociolinguistic context constitutes the fourth main result of the project so far. The integration of conceptual spaces with "Pragmatic Sociolinguistics", and the application of the resulting framework to modeling quantitative corpus studies, is in progress.
The first phase aims at developing a picture of social meaning and sociolinguistic variation in Paris, a large francophone city in Europe. In collaboration with Postdoc Martin, we conducted matched-guise technique experiments investigating Parisian listeners' interpretations of a series of sociolinguistic variables: future temporal reference ("Je vais manger" vs "Je mangerai" "I will eat"), subject doubling ("Jean est là" vs "Jean il est là" "Jean is there") and liaison (ex. pronouncing (or not) the "P" in "trop important" "too important"). In a matched-guise experiment, participants listen to samples of recorded speech (called guises) that are designed to match as much as possible, differing only in the grammatical alternation studied. Each participant is exposed to only one of the guises, and after hearing it, their beliefs and attitudes towards the speaker are assessed, usually via questionnaire. Additionally, in collaboration with PhD student Liang, we conducted quantitative corpus studies investigating Parisian speakers' use of future temporal reference, subject doubling, and complementizer omission ("Je pense que Jean est là" vs "Je pense Jean est là" "I think (that) Jean is there"), as well as grammatical gender and inclusive language (eg. "les étudiants" vs "les étudiant.e.s" "the (gender neutral) students"). We found, as a result of these empirical investigations, that the social meanings associated with these variables, particularly liaison, are highly dependent on the social context of the utterance. This context-sensitivity could not be modeled by previously developed game-theoretic models (such as Burnett 2017, 2019). Therefore, as per the main objectives of the project, we worked to develop a new one. In collaboration with Postdocs Martin and Abbou, we developed (what we call) the "Pragmatic Sociolinguistics" game-theoretic framework. This was obtained by formalizing the "Pragmatic Sociology" non-formalized sociological framework of Boltanski & Thévenot (1991), applying it to language, and integrating it with Burnett's previous models. The development of this framework constitutes the second main result of the project so far.
The second phase of SMIC aims at comparing the results found in the first phase with the results of similar corpus and experimental studies on speakers in Montréal, a large francophone city in North America. These comparative studies are still in progress, in collaboration with PhD student Liang; however, we expect them to yield more information about how the social and cultural context of utterance affects sociolinguistic use and interpretation. These studies will also allow us to test and refine our "Pragmatic Sociolinguistics" game-theoretic framework.
The third phase of SMIC goes deeper into how the social and cultural context affects social meaning through quantitative corpus and experimental studies of feminist activists in Paris, Montréal and Marseille. Since there were no existing corpora of feminist activists in these cities, in collaboration with Postdoc Abbou, we constructed them by doing sociolinguistic interviews in the three cities. The resulting corpus, called CaFé "Cartographies linguistiques du féminisme", is the third main result of the project. Quantitative studies of sociolinguistic variables such as grammatical gender and inclusive language, and feminist social group labels ("queer" vs "materialist" vs "intersectional" etc.) are currently in progress. In order further refine the notion of "social and cultural context" within feminist activism, we conducted psycholinguistic categorization experiments aimed at empirically inferring feminists social and political ideologies, which, in principle, could be integrated into the "Pragmatic Sociolinguistics" framework. These experiments involved applying the "Conceptual Spaces" methodology, commonly used in cognitive science to model perceptual concepts, to the highly complex social domain of French feminism. The fact that we were successful in adapting this methodology in a sociolinguistic context constitutes the fourth main result of the project so far. The integration of conceptual spaces with "Pragmatic Sociolinguistics", and the application of the resulting framework to modeling quantitative corpus studies, is in progress.
So far, the SMIC project has progressed beyond the state of the art in two ways: firstly, we have observed new ways in which social meaning is sensitive to the social context experimentally, and we have developed a new game-theoretic framework (Pragmatic Sociolinguistics) that can capture these results. Pragmatic Sociolinguistics will be the framework that we will use to analyze the results of the empirical studies in the second half of the project. In particular, as specified in the Description of Action, we will enrich our experimental methodologies to include one based on video games, which we will develop. The video games will allow us to develop an even more complex definition of “social context of utterance”, one which takes into account the strategic aspects of the interaction situation. We expect that aspects of the context will be even more important in the video game experimental paradigm than they were in the Matched Guise paradigm. Given this, we expect that the video game experiments will provide even more support for the Pragmatic Sociolinguistics framework over previous proposals in the literature.
The second way that SMIC has gone beyond the state of the art is in its application of the Conceptual Spaces theory and experimental methodology to social and ideological domains, such as French feminism. Conceptual Spaces had been previously thought to be a framework that was only useful to study basic perceptual concepts; however, we have shown how to apply it to ideological representations of feminist political orientation in France. Going forward in the second half of the project, we will use these conceptual space representations to analyze the socio-political identities of speakers in the CaFé corpus, and we expect them to be useful in analyzing the sociolinguistic variables that we will study in our corpus. Thus, we expect to have to integrate Conceptual Spaces with the Pragmatic Sociolinguistic game-theoretic framework, which will result in another class of game-theoretic models which advances the state of the art in game-theoretic pragmatics.
The second way that SMIC has gone beyond the state of the art is in its application of the Conceptual Spaces theory and experimental methodology to social and ideological domains, such as French feminism. Conceptual Spaces had been previously thought to be a framework that was only useful to study basic perceptual concepts; however, we have shown how to apply it to ideological representations of feminist political orientation in France. Going forward in the second half of the project, we will use these conceptual space representations to analyze the socio-political identities of speakers in the CaFé corpus, and we expect them to be useful in analyzing the sociolinguistic variables that we will study in our corpus. Thus, we expect to have to integrate Conceptual Spaces with the Pragmatic Sociolinguistic game-theoretic framework, which will result in another class of game-theoretic models which advances the state of the art in game-theoretic pragmatics.