Implications of Interpersonal Communication Competence Research on the Design of Artificial Behavioral Systems that Interact with Humans Goran Bubaš
Faculty of Organization and Informatics University of Zagreb 42000 Varaždin, Pavlinska 2 Croatia
Faculty of Organization and Informatics University of Zagreb 42000 Varaždin, Pavlinska 2 Croatia [email protected]
Abstract – Numerous communication skills and traits are related to competence in human communicative interaction. A model of interpersonal communication competence (ICC) has been recently developed that can be utilized to guide the design of artificial behavioral systems that interact with humans. A formalization of the generalized dimensions of ICC is outlined, and also of the skills and traits that contribute to those dimensions. Finally, suggestions are made concerning possible implementations of this formalized model in the design of an interface that could enhance the effectiveness and appropriateness in communication of an artificial system with humans.
I. INTRODUCTION Developments in semi-autonomous robotics are illustrated by the mobility of the Honda humanoid robot named ASIMO , as well as by the abundance of efforts to develop diverse android parts, prototypes, and products (see numerous examples at ). They indicate the growing importance of the design of human-humanoid interaction (HHI) systems . However, in order to effectively interact with humans, the robots should be able to communicate using human friendly modalities. Such HHI systems need not only to act acceptably in some real-world domain, but also to behave and interact semisentiently, e.g. (i) communicate with humans verbally and/or nonverbally, (ii) reason about actions and the environment at some level to have something to “discuss”, and (iii) learn and adapt on the basis of human feedback (see: ). Accordingly, the humanoid elements in robot design should incorporate not only physical features that correspond to human-like locomotion and manipulation, but also social features that facilitate the interaction of robots with people . Some recent efforts in the field of HHI were directed toward interaction systems that simulate emotional exchange , joint attention and rudiments of empathic identification , and the imitation-based interactive learning of a toy robot . However, perhaps the most successful of the latest achievements has been in the domain of entertainment robots: the latest Sony SDR-4X model of humanoid robot is capable of biped walking, speech and image recognition, speech synthesis and singing voice production, as well as face recognition and multi-human interaction . With expanding mobility, environment perception and potential for communicative exchange with humans, there will be growing demand for more effective and appropriate HHI in forthcoming robots. For this purpose, at least an outline of the factors of communicative competence in human communication should be conceived that could (a) guide the design efforts in HHI and (b) classify diverse communication skills and traits that, eventually, would be implemented not only in humanoid robots but also in other types of artificial behavioral systems that will interact with humans.
It must be noted that various empirical investigations of the variables that are relevant to interpersonal communication have reported diverse numbers of factors or dimensions, and that more than a hundred different labels have been assigned to those dimensions . The latest and elaborate theoretical classification of skills and traits that are related to human communication competence  has offered four basic dimensions of altercentrism, composure, expressiveness, and coordination as a means for the conceptual organization of roughly a hundred different labels for interpersonal variables. The six dimensions of interpersonal communication competence that are considered in this paper are derived from empirical analysis  and are theoretically elaborated . The results of these previous investigations are projected in this theoretical study to the field of HHI design and to the communication of artificial behavioral systems. II. DIMENSIONS OF HUMAN INTERPERSONAL COMMUNICATION COMPETENCE Six possible dimensions of human interpersonal communication competence were recently identified by empirical research : (i) encoding and decoding; (ii) intentionality, (iii) communication effectiveness, (iv) otherorientedness, (v) expressiveness, and (vi) social relaxation. The results of data analysis suggested that these dimensions are complexly associated to diverse specific communication skills and traits (Figure 1). Furthermore, they correspond to the following general objectives of actors in communicative interaction (adapted from: ): i) Effectively decode verbal and/or nonverbal messages; encode (produce) verbal and/or nonverbal messages in an efficient and appropriate way. ii) Manage social interactions and use communicative ability with a more conscious and intentional orientation; use tactics or strategies that maximize the chance for reaching the desired outcomes of communicative endeavor. iii) Achieve goals and satisfy personal (individualistic) needs by means of communication; engage others in the realization of the activities that are directed toward the fulfillment of one’s own goals and needs. iv) Integrate in the social environment, and contribute to the fulfillment of other actors’ (collectivistic) goals and needs; develop relationships and engage in the formation of interactive dyads or groups (communions) with other actors that share mutual interests and bonds. v) Produce messages that attract attention, generate the desired emotional response, and/or aesthetically stimulate those who receive them. vi) Overcome potential unreasonable communication handicaps and barriers that prevent the actor from unleashing the potential for communicative activity and participation in diverse interactive situations; continue communication incentives and efforts in unexpected or uncertain interactions or situations.
Proceedings of the 6th International Conference on Intelligent Engineering Systems - INES 2002, Opatija, Croatia, pp. 507-512.
Figure 1. The six empirically derived dimensions of human interpersonal communication competence with related communication skills and traits
When compared to the dimensions of human interpersonal communication competence, the latest developments in humanoid robotics and HHI design are still at the general level of the (i) decoding and encoding dimension and rudimentary (v) expressivity dimension. Perhaps, with more elaborate HHI designs, the other dimensions of communicative competence could also be implemented. To facilitate the implementation of more human-like attributes of communication competence in HHI systems the selected components of the three important dimensions of the communication competence of artificial systems (CCAS) will be outlined in more detail. Those dimensions are (A) decoding and encoding, (B) communication effectiveness, and (C) other-orientedness. Each of those dimensions is illustrated by four characteristic communication skills/traits that are further elaborated through a list of means and predispositions that enable their enactment. Even though more than four skills or traits are associated with most of the interpersonal communication competence dimensions of humans (see Figure 1), for reasons of simplification only a reduced set of four skills/traits will be analyzed in more detail for each of the three presented CCAS dimensions. A. Decoding and encoding The dimension of decoding and encoding is basic for any type of communicative interaction since it is comprised of skills and knowledge that contribute to the receiving and sending of affective and nonaffective messages by means of verbal and nonverbal behavior. If observed at the level that is superordinate to elementary perceptive and motoric functions, among other skills, it consists of (1) nonverbal sensitivity, (2) verbal understanding, (3) verbal and nonverbal encoding, and (4) self-monitoring. It must be emphasized that the skills of verbal understanding and verbal and nonverbal encoding were not included as variables in previous empirical research because of measurement problems (they are,
therefore, not represented in Figure 1), while more aggregated conversational skills were included. Some of the components of nonverbal sensitivity (NVerSens) that interact in message interpretation are: Interpretation of facial expressions (i.e. emotions). Comprehension of gestures and body movements. Understanding of paralinguistic communication. Knowledge of other forms of nonverbal interaction. Motivated observation and evaluation of situation. Knowledge of sociocultural and subcultural models. Familiarity with an actor’s specific traits/experiences. Clarification of message interpretation problems due to uncertainty, ambiguity, incongruence, etc. Internal representation or modeling of co-actor's mental and emotional state. The components that interact in the process of verbal understanding (VerbUndr) are: Attentive listening/perception of verbal expressions. Processing of denotative and connotative meaning. Domain knowledge related to the message content. Creating meaning at different levels of abstraction. Discerning the mental and emotional component of a message. Use of complementary nonverbal cues to improve comprehension of verbal content. Management of verbal ambiguity, uncertainty, partiality, incompleteness etc. The elements of verbal and nonverbal encoding (VeNveEnc) are: Verbal and nonverbal behavioral repertoires (“motoric acts”) that can be used for message encoding. Inherited, instinctive, learned, deliberate, or in some other way created associations between motoric acts as messages and their potential meaning for other actors. Assessment of situational/environmental capacity and limitation for message encoding and transmission by means of different communication channels. Consideration for message encoding that is acceptable (understandable) by the co-actor(s).
Coordination of verbal and nonverbal message(s) for congruence and other intuitive or purposeful effect(s). Social knowledge and context (situational) awareness. Skills and traits related to rhetoric and expressive style. Creation of an intentional or subconscious/impulsive reflection and its expression as a message via communicative motoric act(s). The elements that contribute to self-monitoring (SelfMntr) are: Awareness of one’s own behavior and how it could be observed by other actors. Concern for the appropriateness of one’s own behavior before the co-actors. Sensitivity for situational and interpersonal cues that indicate how to model one’s behavior before others. Assessing opportunity and developing goals for selfpresentation. Monitoring and controlling one’s verbal and nonverbal behavior for the purpose of self-presentation. B. Communication effectiveness The communication effectiveness dimension is related to the ability to realize individualistic goals by means of communication with other actors, e.g. it denotes the capacity to influence or persuade. The skills and traits that are essential for this dimension are (1) initiation of interaction, (2) assertiveness, (3) interaction management, and (4) adaptability. More specific components of initiation of interaction (InitIntr) are: Motivation (goal, need) for a novel acquaintance. Behavioral repertoire of approaches to other actors. Assessment of the potential attributes of other actors. Choice of approach according to co-actor attributes. Immediacy / naturalness vs. formality / etiquette. Continuation of preliminary talk (i.e. by chatting). Finding shared motive(s) for additional encounters. Arranging additional contacts or encounters. Some of the important elements of assertiveness (Assertiv) are: Awareness of goal / purpose / interest for interaction. Attentiveness to the interaction process with co-actor. Closeness and amicability (or dominance) in conduct. Estimate of potential interests / needs of the co-actor. Use of non-coercive tactics/manipulation to influence. Conformance with co-actor interests, rights and dignity. Clear and convincing argument(s) and/or demand(s). Relationship compensation or reward(s) for co-actor. The common components of interaction management (IntManag) are: Defined interaction goals for the specific situation and co-actor(s). Awareness of the sequencing of interaction processes. Impression management and/or self-presentation. Assessment of the situation and attributes of co-actor. Choice of manipulation tactics and their mental simulation; Means for interpersonal control and/or dominance. Scripting, personal role-play, structuring of interplay. Outcomes appraisal, relational repair, impression maintenance. The possible elements of the adaptability (Adaptabt) skill/trait are: Development of behavioral repertoires for interaction.
Avoidance of stereotypical / rigid / dogmatic conduct. Exploration of other actor's traits and dispositions. Assessment of the interaction situation/environment. Insight into preferred interaction goals / impressions. Choice of optimal pattern of behavior to display. Attentiveness to co-actor’s reactions and impressions. Reassessment of previously set personal interaction goals, adaptation of goal plans and behavior patterns.
C. Other-orientedness The dimension of other-orientedness reflects the ability to achieve relational goals and to contribute to the wellbeing of others. The skills that are important for the manifestation of this dimension are (1) empathy, (2) support, (3) self-disclosure, and (4) collaboration. Some of the essential components of empathy (Empathy) are: Motivation or desire to identify with other actors emotionally or conceptually. Ability to recognize specific behaviors of other actors that could relate to their emotional or mental state. Observation of co-actor's situation and behavior. Perspective taking in relation to situation of co-actor. Determining possible emotions, thoughts and roles of co-actor. Internal emotional and mental reflection or representation of what the co-actor is feeling / experiencing. External expression of similar or corresponding feelings and of identification with the co-actor. Disclosure of concern and care for the co-actor. The components that contribute to the expression of support (Support) are: Feeling of desire/obligation to provide aid/assistance. Adequate behavioral repertoire for successful support. Insight into the actual needs/interests of the co-actor. Assessment of the situation and choice of favorable type/form of support for the co-actor. Display of readiness to provide support in a manner that is agreeable. Concern, understanding and respect for the co-actor. Intervention by means of supportive message(s) and/or action(s). Advancement of affinity, affective bonding and reciprocation among the co-actors. Some of the potentially important elements of selfdisclosure (SelfDisc) are: Reason(s) for self-disclosure and trust or confidence in the co-actor. Filtering of produced verbal and nonverbal messages for impression management. Adaptation to the co-actor in topic, extent, depth, and manner of self-disclosure. Receiving of interpersonal feedback from the co-actor, relational self-appraisal. Encouraging closeness, immediacy and reciprocal selfdisclosure of the co-actor. Collecting personal information about the co-actor. Assessment of potential relationship goals / prospects. Search for similarity with co-actor and reduction of uncertainty in the relationship. Regard for the other actor's conventions, as well as strategic (relationship-oriented) self-disclosure. The components that contribute to collaboration (Collabor) are: Need/affinity for collective/group activities and work.
Concern for communality and fellowship in goal selection. Awareness of the importance of cooperation in goal attainment. Ability to adapt to collaborators and collective work. Respect for the individuality, rights and interest of coactor(s). Readiness to provide (receive) assistance or support. Capacity for sharing the results of collective efforts (i.e. rewards). Acceptance of disproportional rewards when work effort or credit was greater than that of the co-actor. Relationship maintenance for the sake of future collaboration with the co-actor. C. Intentionality, expressivity, and social-relaxation The analyzed structural elements of the dimensions of decoding and encoding, communication effectiveness, and other-orientedness illustrate to what extent the higher-level communication capacities of humans are elaborate and imply an extraordinarily high level of social knowledge and skill. Since the detailed elaboration of the further dimensions of expressivity, social-relaxation, and intentionality would be too extensive for this paper, only the components of those dimensions will be listed. The expressivity dimension is predominantly related to verbal expressivity (VerbExpr) and nonverbal expressivity (NVerExpr). The skills needed for expressivity enable the production of messages that are especially illustrative, lively, inspiring, moving, emotionally contagious, and may create the impression of having “personality” or “style”. The social-relaxation dimension is associated with composure (Composr), interaction involvement (IntInvol), and communication motivation (CommMotv). The related skills and traits regulate the approach-withdrawal behavior in communicative interaction and facilitate the enactment/effectuation of diverse other communication dimensions and skills. The intentionality dimension is related to awareness and consciousness in interpersonal interaction regarding the communication process, actors, situation, roles/rules, goals, and means for achieving specific outcomes. The goal-driven and purposeful communication activity may better utilize the available skills and knowledge, as well as the characteristics of the situation and of other co-actors. This dimension is related to assertivity, self-monitoring, conversational skills (Converst), and knowledge of the communication process (Knowledg). III. CONCEPTUAL MODEL OF CCAS Implementation problems exist if the structural model of communication competence that was previously presented is not observed in relation to humans, but at the level of CCAS, since the elaborate design of most of the higher level communication skills and traits is still technologically infeasible. Most of the behavioral logic that would guide the communication of such systems has to be programmed or is developed by machine learning at a rather low level of complexity. Therefore, the intentionality dimension of CCAS is perhaps the primary structure that could organize the activity and interaction of other dimensions and skills/traits in the modeling of CCAS at the current stage of technological development (see Figure 2 for a conceptual model of CCAS structure).
INTENTIONALITY (organizing function)
DECODING AND ENCODING (message exchange function)
(message enhancement function)
OTHERORIENTEDNESS (community function)
SOCIAL RELAXATION (activation and intensity moderation)
Figure 2. A conceptual model of the structure of CCAS
The social relaxation dimension in CCAS can be conceptualized as a semi-autonomous mechanism that functions as a regulator of approach-withdrawal behaviors, as well as of activation-deactivation and intensity aspects of communicative engagement (in fact “involvement regulation” would probably be a more appropriate name for this dimension in CCAS structure). The decoding and encoding dimension is fundamental since it makes possible the perception of the communication-related stimuli in the external environment and produces “motoric” acts that create response messages. The expressivity dimension enhances produced messages to make them more attractive or influencing. Finally, the dimensions of communication effectiveness and otherorientedness regulate the communicative engagement with respect to self-directed or community (relationship) oriented goals. IV. FORMALIZATION OF CCAS COMPONENTS Communication-related behavior in humans can be observed at the macroscopic, mezzoscopic and microscopic level . This kind of abstraction is used for a formalization of the CCAS structure with an additional atomic level. Let be the set of dimensions of communication competence (macroscopic level). Let be the set of unlisted or still undefined elements of the sets , , , and that denote dimensions, skills or traits, primitives, and atomic behaviors at various levels of communication behavior analysis. = Decoding and encoding, Intentionality, Communication effectiveness, Other-orientedness, Expressivity, Social Relaxation, Let be the set of skills and traits that are related to the dimensions of communication competence (mezzoscopic level). The denotations for the skills and traits at this level were introduced in the preceding parts of this text. Decoding and encoding = NVerSens, VerbUndr, VeNveEnc, SelfMntr, Knowledg, Composr, Empathy, Adaptabt, IntrInvo, NVerExpr, Converst,
Intentionality = SelfMntr, Assertiv, Knowledg, IntrMngm, Collabor, Adaptabt, IntrInvo, IntrpCtr, Converst,
Ai , Bi i
Communication effectiveness = Motivatn, InitIntr, Assertiv, IntrpCtr, VerbExpr, Adaptabt, Intmngm, IntInvol, Collabor ,
Other-orientedness = Empathy, Support, NVerSens, NVerExpr, IntInvol, SelfDisc, Collabor, Converst, Expressivity = NVerExpr, VerbExpr, Social relaxation = Composr, Motivatn, InitIntr, Adaptabt, IntInvol, IntrpCtr, Let be the set of intermediate communication “primitives” (at the microscopic level) or moderately complex verbal/nonverbal behaviors, behavioral scripts, skill-related knowledge or experience, communication traits, interaction tactics, etc., that contribute to the enactment and manifestation of a specific skill or trait of the set. It must be noted that the optimal components of skills and traits (as well as the configuration of their enactment) at this level may vary depending on the specific situation or goal. An example is given of the components of nonverbal sensitivity as a subset of . NVerSens = Facial expression decoding; Gesture and body movement decoding; Paralinguistic decoding; Other forms of nonverbal messages decoding; Observation and situation evaluation; Sociocultural and subcultural knowledge; Knowledge of co-actor traits and experiences; Message uncertainty, ambiguity, and incongruence management, Representation or modeling of co-actor's mental or emotional states, Let be the set of simple perceptual or motoric behaviors (at the atomic level) that are engaged in the realization of the communication “primitives” of the set. This level will not be elaborated, but one example is provided. Facial expression decoding = Gaze in the direction of co-actor's face; Focusing attention; Facial feature extraction in temporal sequence; Processing for basic emotion identification; Comparison with data collected from other sources/channels - verbal, paralinguistic - on the actual emotional/mental state of the co-actor; Incongruence management; Let be a set of semantic interpretations of the perceived features by a behavioral system of all possible and detectable situations in the external environment. Let i be a communication-relevant semantic interpretation of an external situation at temporal point i. Elements f of i are semantic features that are communication-relevant. i
Let Ai be a semantic representation of the perceived features by a behavioral system of its own external communication-relevant behavior at temporal point i, and let Bi be a semantic representation of the perceived features by a behavioral system of the consequences of Ai at temporal point i.
Let be a function defined by (3)
Let i be a representation of the intentional computation of all acceptable elements of regarding i
In this phase only an outline of the formalization is provided, although it can be advanced in more detail for the analysis or implementation of a specific dimension of CCAS or for more specific communication "primitives". Further elaboration of this model should also solve the problems related to conflict situations where i = (e.g. when there is no action which is satisfactory for all elements of i) and problems where | i | > 1 (e.g. when there is more than one acceptable action for a given i). Also, the interaction between components of the set (that originate from different levels of CCAS) should be treated in more detail. Of course, such problems do not always apply to human communicators since they are resolved by the activation of complex intentionality, social relaxation, communication effectiveness and otherorientation dimensions in the structure of their communication competence. V. CONCLUSIONS One of the reasons for creating humanoid robots is that they present a natural interface for HHI that enables humans to utilize their instinctive and culturally developed (“subconscious”) communication patterns, techniques, and channels for interacting with other humans when they are faced with the problem of HHI (see: ). Since robots with “communication skills” and “knowledge” similar to those of humans would enable more effective HHI, it is important to investigate and create models of communication competence of both human and artificial systems. For example, the other-orientedness dimension of CCAS could be important in robotic systems for helping the disabled. A robot that is designed for such purpose could be programmed to learn from its own experiences . With the development of robotics, more and more artificial behavioral systems that interact with humans will be engaged, not only in homes, but also in public places with functions of attracting and influencing people to participate in some activity, and maintaining their interest and attention (see: ). This will imply the need for an implementation of at least rudimentary communication behaviors that correspond to the CCAS dimensions of communication effectiveness and expressivity. Recent advances in speech recognition and dialogue management , as well as the potential for the application of dialogue systems in autonomous entertainment robots , resolve some of the problems associated with the microscopic () and atomic () level of communication behavior implementation. Some other solutions for those levels of CCAS have already been developed or are in
development . Other advances are related to humanoid robots that can learn and interact by imitating simple motoric acts , which can be applied to learning nonverbal communication by gesture and body movement, and in multimodal human-robot interfaces that use human natural language and gestures . From there it should not be too difficult to incorporate more expressivity (as in: ) or even “personality”  in embodied humanoid robots. Even though the practical approach in the development of CCAS or in HHI design has to start with the dimensions of intentionality and decoding and encoding, it must be emphasized that in human-computer interaction the human agent is likely to judge or evaluate the CCAS on the basis of criteria that he/she has developed beforehand in humanhuman communication, and such criteria would probably include not only judgments of global communicative (social) skill or competence, but also the competence dimensions of (a) other-orientedness, (b) communication effectiveness, and (c) expressivity. Therefore, it would be favorable to include at least some related communication “skills” or “traits” in the HHI interface or CCAS design. In other words, to feature these characteristics in the humancomputer interface of even the simplest artificial behavioral system that has to interact (“communicate”) with humans would probably be profitable because of the created impression of other-orientedness, communication effectiveness, and expressivity during the interactions of a human actor/user with such a system. In this paper an outline of the potential components of CCAS is provided, as well as an initial formalization of the description of such systems.
 G. Bubaš, “The Interpersonal Communication Competence Inventory and the structure of trait communication competence”, 51st annual conference of the International Communication Association, Washington, DC, USA, 2001.  G. Bubaš, “Toward competence in interpersonal communication: Constitutive traits, skills, and dimensions”, World Futures: A Journal of the General Evolution, 57, 2001, pp. 557-581.  K. Kawamura, R.A. Peters II, D.M. Wilkes, W. A. Alford and T.A. Roger, “ISAC: Foundations in human-humanoid interaction”, IEEE Intelligent Systems, vol. 15, no. 4, July/Aug. 2000, pp. 38-45.  H. Kozima and H. Yano, “In search of ontogenetic prerequisites for embodied social intelligence”, International Conference on Cognitive Science, Workshop on Emergence and Development of Embodied Cognition, Beijing, China, 2001.  L.S. Lopes and J.H. Connell, “Guest editors' introduction: Semisentient robots-- routes to integrated intelligence”, IEEE Intelligent Systems, vol. 16, no. 5, Sept./Oct. 2001, pp. 10-14.  M.J. Mataric, “Getting humanoids to move and imitate”, IEEE Intelligent Systems, vol. 15, no. 4, July/Aug. 2000, pp. 18-24.
 D. Perzanowski, A.C. Schultz, W. Adams, E. Marsh and M. Bugajska, “Building a multimodal humanrobot interface”, IEEE Intelligent Systems, vol. 16, no. 1, Jan./Febr. 2001, pp. 16-21.
 B. Adams, C. Breazeal, R.A. Brooks and B. Scassellati, “Humanoid robots: A new kind of tool”, IEEE Intelligent Systems, vol. 15, no. 4, July/Aug. 2000, pp. 25-31.
 W.S.N. Reilly, “A methodology for building believable social agents”, Proceedings of the First International Conference on Autonomous Agents, Marina del Rey, CA, USA, 1997, pp. 114-121.
 H. Asoh, Y. Motomura, F. Asano, I. Hara, S. Hayamizu, K. Itou, T. Kurita and T. Matsui, “Jijo-2: An office robot that communicates and learns”, IEEE Intelligent Systems, vol. 16, no. 5, Sept./Oct. 2001, pp. 38-45.
 B. Salem and N. Earle, “Designing a non-verbal language for expressive avatars”, Proceedings of the Third International Conference on Collaborative Virtual Environments, San Francisco, USA, 2000, pp. 93–101.
 A. Billard, “Play, Dreams and Imitation in Robota”, Workshop on Interactive Robotics and Entertainment (WIRE-2000), Pittsburgh, USA, 2000.
 “Sony Develops Small Biped Entertainment Robot”, Press Release, Sony Corporation, Tokyo, Japan, March 19, 2002. http://www.sony.co.jp/en/SonyInfo/News/Press/2002 03/02-0319E/
 C. Breazeal and B. Scassellati, “Infant-like social interactions between a robot and a human caretaker”, to appear in Special issue of Adaptive Behavior on Simulation Models of Social Agents, guest editor Kerstin Dautenhahn, 2000.  R.A. Brooks, C. Breazeal, B. Scassellati, U.-M. O'Reilly, “Technologies for human/humanoid natural interactions”, HURO-99: Proceedings of the Second International Symposium on Humanoid Robots, Tokyo, Japan, 1999, pp. 135–147.  G. Bubaš, D. Bratko and I. Marušić, “The dimensions of interpersonal communication competence: An analysis of data collected by multiple self-rating scales”, 85th Annual Convention of the National Communication Association, Chicago, IL, USA, 1999.
 B.H. Spitzberg and W.R. Cupach, “Interpersonal Skills”, in M. L. Knapp and J. Daly (Eds.), Handbook of interpersonal communication, Sage, Thousand Oaks, CA, in press.  B.H. Spitzberg and W.R. Cupach, “Handbook of interpersonal competence research”, SpringerVerlag, New York, 1989.  L. Steels, “Language games for autonomous robots”, IEEE Intelligent Systems, vol. 16, no. 5, Sept./Oct. 2001, pp. 16-22.  S. Thrun, J. Schulte and C. Rosenberg, “Robots with humanoid features in public places: a case study”, IEEE Intelligent Systems, vol. 15, no. 4, July/Aug. 2000, pp. 7-11.