Intention and Face in Dialog

Adil Soubki,Owen Rambow

2024-06-06

Abstract:The notion of face described by Brown and Levinson (1987) has been studied in great detail, but a critical aspect of the framework, that which focuses on how intentions mediate the planning of turns which impose upon face, has received far less attention. We present an analysis of three computational systems trained for classifying both intention and politeness, focusing on how the former influences the latter. In politeness theory, agents attend to the desire to have their wants appreciated (positive face), and a complementary desire to act unimpeded and maintain freedom (negative face). Similar to speech acts, utterances can perform so-called face acts which can either raise or threaten the positive or negative face of the speaker or hearer. We begin by using an existing corpus to train a model which classifies face acts, achieving a new SoTA in the process. We then observe that every face act has an underlying intention that motivates it and perform additional experiments integrating dialog act annotations to provide these intentions by proxy. Our analysis finds that dialog acts improve performance on face act detection for minority classes and points to a close relationship between aspects of face and intent.

Computation and Language

What problem does this paper attempt to address?

### The Problem the Paper Attempts to Solve

This paper aims to explore and analyze the relationship between "face" behaviors and intentions in conversations. Specifically, the paper focuses on the following aspects:

1. **How intentions influence politeness behaviors in conversations**:
- The authors point out that although Brown and Levinson's (1987) face theory has been widely studied, a key aspect—how intentions influence the planning of conversational turns that affect face—has received less attention.
- The paper explores how intentions influence politeness behaviors by analyzing three computational systems to classify intentions and politeness.

2. **Improving the recognition of face behaviors**:
- The authors trained a model using an existing corpus to classify face behaviors, achieving a new State-of-the-Art (SoTA) level in the process.
- They further observed that each face behavior has an underlying intentional motivation and provided these intentions by integrating dialogue behavior annotations.

3. **The impact of dialogue behaviors on face behavior detection**:
- The authors found that dialogue behavior annotations can improve the detection performance of minority face behavior categories and highlighted the close relationship between face and intentions.

### Main Contributions

- **Experimental Design**: The authors conducted a series of experiments using dialogue behavior annotation information to train a face behavior classifier, demonstrating how this information helps classify specific face behaviors.
- **Performance Improvement**: Through these experiments, the authors not only improved the classifier's performance but also achieved a new state-of-the-art level, increasing the F-measure from 69% to 73%.
- **Theoretical Contribution**: The authors argue that Brown and Levinson's face theory is not just a marginal aspect of Natural Language Processing (NLP) but can play a key role in improving interactive NLP and understanding how humans use language.

### Conclusion

The paper demonstrates through experiments the importance of dialogue behavior annotations in improving face behavior detection and lays the foundation for further research. The authors emphasize that understanding and modeling intentions are crucial for enhancing the interactive capabilities and cultural adaptability of natural language processing systems.