Politeness and Etiquette Modeling: Beyond Perception to Behavior

Politeness and Etiquette Modeling: Beyond Perception to Behavior

Christopher A. Miller (Smart Information Flow Technologies, USA), Tammy Ott (Smart Information Flow Technologies, USA), Peggy Wu (Smart Information Flow Technologies, USA) and Vanessa Vakili (Smart Information Flow Technologies, USA)
DOI: 10.4018/978-1-61520-883-8.ch017
OnDemand PDF Download:
No Current Special Offers


If culture is expressed in the patterns of behavior, values and expectations of a group, then a central element in the practical modeling and understanding of culture is the expression of politeness and its roles in governing and influencing behavior. The authors have been developing computational models of “politeness” and its role in power and familiarity relationships, urgency, indebtedness, etc. Such a model, insofar as it extends to human-machine interactions, will enable better and more effective decision aids. This model, based on a universal theory of human politeness, links aspects of social context (power and familiarity relationships, imposition, character), which have culture-specific values, to produce expectations about the use of polite, redressive behaviors (also culturally defined). The authors have linked this “politeness perception” model to a coarse model of decision making and behavior in order to predict influences of politeness on behavior and attitudes. This chapter describes the algorithm along with results from multiple validation experiments: two addressing the model’s ability to predict perceived politeness and two predicting the impact of perceived politeness on compliance behaviors in response to directives. The authors conclude that their model tracks well with subjective perceptions of American cultural politeness and that its predictions broadly anticipate and explain situations in which perceived politeness in a directive yields improved affect, trust, perceived competence, subjective workload, and compliance, though somewhat decreased reaction time. The model proves better at accounting for the effects of social distance than for power differences.
Chapter Preview


Much significant and useful work (e.g., Hofstede, 2001; Nisbett, 2003; Klein, 2004; House, et al., 2004) has been done on characterizing cultures according to deep-seated and, presumably, durable attitudes, values and, perhaps, psychological tendencies. This work can and has been used to train individuals, predict behaviors and even customize devices and human-machine interactions to the culture as a whole. While helpful, such models are extremely coarse-grained and provide little guidance at the level of specific interactions between individuals in context.

While these broad and deep descriptions of cultural differences and similarities may be of interest in their own right, that interest becomes practical when we need to get a job done, to communicate information and/or obtain resulting actions from members of another culture. It may be helpful to know, for example, that an individual I’m working with is from a culture with a lower Power Distance Index (PDI-- cf. Hofstede, 2001) than my own and that, therefore, s/he is more likely to deviate from or critique an instruction from a superior, but how do I practically convey, with any reasonable hope of success, that this instruction is critical and it’s very important that it be followed exactly?

This makes apparent the role of social interaction “etiquette”—that is, interactions based on the social characteristics and assumptions of each agent as an intentional entity and drawing from culturally familiar patterns of expectations about and interpretations of appropriate behaviors among such agents. While the study of culturally-correlated attitudes, cognitive styles, and sense-making mechanisms, not to mention specific attributes and histories of alternate cultures, are all important, action and work across diverse cultures almost inevitably involves communication. Communication means verbal and non-verbal exchange between intelligent agents by means of language and gesture, prosody and posture. Hence these topics and their subtle role in social interaction expectations, interpretations and behaviors, should be a topic of study in multi-cultural interactions, whether they be human-human or human-machine.

It may well be objected that attention to these “superficial” aspects of cultural expression—overt politeness behaviors—will hardly convey a rich or deep understanding of cultural thought, behavior and nuance. This is almost certainly true, but such overt behaviors not only have predictive power and are capable of producing desired (or avoiding undesired) effects as we will illustrate, but they are also perhaps a necessary entry point toward interacting with those of other cultures with sufficient goodwill to enable such deeper learning.

Over the past several years, our work has focused on developing a concrete model of observable human behaviors—specifically communication behaviors having to do with etiquette and politeness—which in turn have relevance for human performance, attitudes and broader decision making. In this paper, we will describe our model and then, briefly, the results of several experiments illustrating its ability to both predict perceptions of politeness and to account for variations in very specific behaviors associated with those perceptions.

Key Terms in this Chapter

Redressive Act: A redressive act (which we sometimes also label a “politeness behavior”) is a verbal, non-verbal (e.g., gestural, postural, etc.) or extra-verbal (e.g., prosody, tone or volume of voice, etc.) behavior which, in a given culture, has the potential to mitigate or “redress” some degree of face threat. Redressive acts are what are typically thought of as “polite” behaviors: e.g., saying “please” and “thank you”, taking off a hat before speaking, etc. Specific redressive acts (e.g., taking off one’s hat vs. bowing vs. kissing a ring, etc.), as well as their redressive values, are culturally defined and are not universal, though Brown and Levinson (1987) have identified and reported broad classes of redressive “strategies” (e.g., apologies, honorifics, etc.) which they have found recurring across multiple cultures.

Politeness: Politeness is one the means by which we convey, interpret, maintain and alter social relationships. In keeping with Brown and Levinson’s (1987) model, politeness operates by enhancing or mitigating face threat. Also in keeping with our interpretation of the Brown and Levinson model, it is strictly incorrect to speak of the amount of “politeness” in an interaction or utterance. Instead, “perceived politeness” is the result of the balance between perceived face threat in the interaction and perceived value of the “redressive actions” used in the interaction.

Etiquette: Etiquette is the code, or protocol, by which we signal and interpret politeness. It makes use of verbal, physical, gestural and even more primitive modes of interaction. The set of expectations and interpretations that give contextual elements their social and cultural face threat values and behaviors their redressive values is an etiquette. We suspect that each culture has its own, unique etiquette, though similar cultures will have similar etiquettes. In fact, etiquette is one of the most readily observable, overt signals of a culture.

Behavioral Effects of Politeness: In this paper we use the phrase “behavioral effects of politeness” to refer to the impact of perceived politeness on the reported attitudes and decision making processes and outcomes. In our experiments, we directly sampled multiple behavioral and/or attitudinal effects of perceived politeness including: directive compliance, reaction time to comply, memory for the context of the directive, affect toward the directive giver, trust in the directive giver, and perceived workload in interaction with the directive and directive giver.

Culture: We have not provided a deep definition of culture and are, instead, largely concerned with its surface manifestations: particularly, in the ways that politeness is reasoned about and its effects on behavior. “Culture”, then, as far as we are concerned, is a characteristic pattern of situations in which more or less politeness is required or expected by the people who share the culture, combined with a characteristic pattern of behaviors (e.g., verbal, non-verbal and extra-verbal patterns) which can be used to provide or satisfy those expectations. We largely begin with the politeness model of Brown and Levinson (1987), which claims to be culturally universal, but are extending it to provide a more cognitive grounding combined with behavioral expectations. We suspect that some aspects of culture may be characterized by common, deep-seated attitudes (e.g., Hofstede, 1987) and decision making styles or methods (e.g., Nisbett, 2003), which combine to make various behaviors more or less likely in context, but our primary interest has been in how such “deep” cultural factors may characteristically interact with politeness to yield predictable behaviors in response to directives.

Cultural Factors: We use the term “cultural factors” to refer to “deeper” descriptors of common attitudes or thought patterns characteristic of cultures than are captured by the overt, “shallow” and potentially rote nature of our politeness behaviors. Several models of such deeper “cultural factors” exist such as Hofstede’s (1987) descriptive taxonomy or Nisbett’s (2003) largely cognitive explanation of differences in Eastern and Western thought.

Face Threat: Face was defined by Goffman (1967, p. 5) as the “positive social value a person effectively claims for himself”. Face can be saved or lost, and it can be threatened or conserved in interactions. The desire to preserve and enhance one’s face is presumed to be a universal social goal in the Brown and Levinson (1987) model, and commonly manifests itself as the twin desires to autonomously direct one’s actions and attentions where one wills and to have one’s goals and actions seen as valuable. Yet virtually all interactions between social agents are Face threatening acts (FTAs) since, at a minimum, they require the hearer to shift attention and acknowledge the speaker, thereby impinging upon one or both of the face goals. Brown and Levinson’s model claims that the degree of face threat in an interaction is a function of three terms: the power that the Hearer has over the Speaker, the Social distance between Speaker and Hearer, and the raw imposition of the topic or request inherent in the interaction.

Complete Chapter List

Search this Book: