Periodic Reporting for period 3 - SPAGAD (Speech Acts in Grammar and Discourse)
Okres sprawozdawczy: 2022-01-01 do 2023-06-30
Speech acts are “the basic or minimal units of linguistic communication” (John Searle). They connect two aspects of language: its ability to refer to entities and state of affairs, and its use in communication. This includes informing others about state of affairs, but also asking questions, giving commands, making promises, expressing attitudes, and even enacting changes of the world. In linguistics, successful formal theories have been devel-oped to address the referential aspects of language – how speakers can identify the virtu-ally unlimited possible state of affairs with limited linguistic means. But speech acts proved much more difficult to model with the same level of formal explicitness.
The SPAGAD project is addressing this task by elaborating the notion of dynamic inter-pretation, according to which the participants in a conversation enrich their shared in-formation, the Common Ground. SPAGAD develops a model of shared information with the novel notion of Commitment Spaces, which contains the information accrued at a par-ticular point in conversation, but also the proposed future developments. This forward-looking perspective is crucial for the modelling of questions, commands and promises. Furthermore, this model does not only contain the shared factual information, but also the commitments by the participants – e.g. information about who vouches for the truth of which proposition, and who has which attitude towards an entity or proposition.
SPAGAD undertakes research in the way how aspects of speech acts are expressed in grammar, by particular words, and by prosody. For example, adverbs like possibly and truly have a particular effect on assertions; questions like Is it not raining? and Isn’t it raining? express different pre-assumptions of the speaker. SPAGAD investigates how speech acts work in discourse: How are speech acts combined, what does the selection of one act over another tell us about the communicative goals of the interlocutors? SPAGAD also analyzes the role of speech acts in communication, in particular the social norms, like truthfulness, that come with the enactment of speech acts and that are crucial for com-munication. SPAGAD is based on experimental and corpus-linguistic research, with an aim at precise formal modelling.
SPAGAD should lead to a better framework for the description of the grammar, prosody, and communicative devices of languages, and to a more coherent notion of how linguistic communication works.
The SPAGAD project is addressing this task by elaborating the notion of dynamic inter-pretation, according to which the participants in a conversation enrich their shared in-formation, the Common Ground. SPAGAD develops a model of shared information with the novel notion of Commitment Spaces, which contains the information accrued at a par-ticular point in conversation, but also the proposed future developments. This forward-looking perspective is crucial for the modelling of questions, commands and promises. Furthermore, this model does not only contain the shared factual information, but also the commitments by the participants – e.g. information about who vouches for the truth of which proposition, and who has which attitude towards an entity or proposition.
SPAGAD undertakes research in the way how aspects of speech acts are expressed in grammar, by particular words, and by prosody. For example, adverbs like possibly and truly have a particular effect on assertions; questions like Is it not raining? and Isn’t it raining? express different pre-assumptions of the speaker. SPAGAD investigates how speech acts work in discourse: How are speech acts combined, what does the selection of one act over another tell us about the communicative goals of the interlocutors? SPAGAD also analyzes the role of speech acts in communication, in particular the social norms, like truthfulness, that come with the enactment of speech acts and that are crucial for com-munication. SPAGAD is based on experimental and corpus-linguistic research, with an aim at precise formal modelling.
SPAGAD should lead to a better framework for the description of the grammar, prosody, and communicative devices of languages, and to a more coherent notion of how linguistic communication works.
The project has carried out three workshops on different aspects of speech acts (on the formal modelling of grammatical phenomena related to speech acts, on discourse phe-nomena, and on questions).
The model of Commitment Spaces was used to explain different types of questions, for example questions that express a bias, such as negated questions, declarative questions (It is raining?) and question tags (It is raining, isn’t it?). We also gave an account of questions in Turkish, which has grammatical means to highlight certain parts of questions. Current-ly we are working on a volume on biased questions.
A model of assertions was developed that distinguishes between three levels that affect certain aspects of assertive speech acts. These levels can be addressed by different speech-act modifiers, such as epistemic and evidential adverbials (probably, apparently), commitment modifiers (truly, seriously) and other specifications (frankly, as you are in-terested in this).
We investigated in an experimental study the uses of epistemic adjectives such as it is cer-tain / possible and the corresponding adverbials certainly / possibly that are used in quite different ways. Elaborating on previous work we found evidence that the adverbials are used to down-tone the commitment to propositions that the speaker wants to introduce into the communication.
We investigated the effect of prosodically distinct contours used for calling a person (as to attract attention) and for cheering a person (as in sport events). There are interesting variations between speakers, languages, and purposes for the way how such speech acts are enacted.
We also developed a theory of pronoun use for languages like Vietnamese that, compared to English, tend to avoid the use of first and second person pronouns.
We have experimental results showing that speech acts can be referred to by demonstra-tives, just as other events in the situation of utterance (e.g. if A says: Bill is an idiot. then B can react with: That is not nice of you., not with It is not nice of you). We also have experi-mental results about the different ways how discourse particles like okay are used, which strongly depend on prosody.
Due to the Covid19 crisis, several planned experiments could not be carried out. We had to resort to online experiments, which excluded, for example, work on the acquisition of speech acts by children.
The model of Commitment Spaces was used to explain different types of questions, for example questions that express a bias, such as negated questions, declarative questions (It is raining?) and question tags (It is raining, isn’t it?). We also gave an account of questions in Turkish, which has grammatical means to highlight certain parts of questions. Current-ly we are working on a volume on biased questions.
A model of assertions was developed that distinguishes between three levels that affect certain aspects of assertive speech acts. These levels can be addressed by different speech-act modifiers, such as epistemic and evidential adverbials (probably, apparently), commitment modifiers (truly, seriously) and other specifications (frankly, as you are in-terested in this).
We investigated in an experimental study the uses of epistemic adjectives such as it is cer-tain / possible and the corresponding adverbials certainly / possibly that are used in quite different ways. Elaborating on previous work we found evidence that the adverbials are used to down-tone the commitment to propositions that the speaker wants to introduce into the communication.
We investigated the effect of prosodically distinct contours used for calling a person (as to attract attention) and for cheering a person (as in sport events). There are interesting variations between speakers, languages, and purposes for the way how such speech acts are enacted.
We also developed a theory of pronoun use for languages like Vietnamese that, compared to English, tend to avoid the use of first and second person pronouns.
We have experimental results showing that speech acts can be referred to by demonstra-tives, just as other events in the situation of utterance (e.g. if A says: Bill is an idiot. then B can react with: That is not nice of you., not with It is not nice of you). We also have experi-mental results about the different ways how discourse particles like okay are used, which strongly depend on prosody.
Due to the Covid19 crisis, several planned experiments could not be carried out. We had to resort to online experiments, which excluded, for example, work on the acquisition of speech acts by children.
The Commitment Space model is turning out to have several advantages over alternative proposals, for the representation of biased questions, for commands, and for speech acts expressed under a condition (like if we are hungry, what should we eat?).
The assumption that the grammatical structure of an assertion provides for different lay-ers goes beyond previous proposals, as it allows for a compositional interpretation of these layers.
We expect that at the end of the project, the Commitment Space model can be shown to provide for a natural and enlightening account for a wide range of speech act phenomena.
The assumption that the grammatical structure of an assertion provides for different lay-ers goes beyond previous proposals, as it allows for a compositional interpretation of these layers.
We expect that at the end of the project, the Commitment Space model can be shown to provide for a natural and enlightening account for a wide range of speech act phenomena.