Alignment in Communication:
Towards a New Theory of Communication
July 19-21, 2012 at Center for Interdisciplinary Research (ZiF), Bielefeld, Germany
Alignment in Communication is the theme of a novel direction in communication research under investigation in the Bielefeld Collaborative Research Center SFB 673 since July 2006. Our effort focuses on special modes of coordination in communication referring to interactive adaptation processes among agents that are assumed to be more or less automatic in humans. Its ultimate purpose is to extend our knowledge about cognitive processes underlying language production and comprehension in human communication, as well as in natural language use in human-machine interaction.
Since the seminal work on communication as signal transmission by Shannon, it has become clear that the communication between rational agents is too complex, flexible, and unpredictable to be based on signal processing alone. Theorists from many different frameworks have addressed the issue of agent-agent communication, with varying assumptions, methodologies, and empirical findings. Examples are the discourse-analytical approaches, the elegant but highly reflexive/recursive approach by Grice and later by neo-Gricean theorists, a related approach advocated by Herbert Clark and colleagues which is based on the central notion of “common ground”, and more recent approaches such as that by Pickering and Garrod, in which communication is seen as representational alignment arising from automatic processes. However, there is at present no theory of communication that is comprehensive enough to cover the multitude of observed communicational phenomena, and concrete enough to provide guidelines and blueprints for implementing communicative systems in artificial agents.
The Bielefeld research initiative has two overall goals: First, to investigate the role of alignment as a pioneering explanation of natural language use in conversation. Second, to explore the notion of alignment as a general principle of information interchange by testing the interactive alignment approach in situations that go beyond verbal conversation between humans. There are many phenomena demonstrating that communication is, to a large extent, a matter of joint activity based on fine-tuned ‘mechanistic’ coordination. Moreover, in communication and human information processing, the roles of automaticity, tacit conventions, and alignment (rather than explicit negotiation) have been underestimated. Consequently, we focus our research efforts on these less obvious aspects of communication and aim to develop a theory of communication around the notion of alignment taking place between interlocutors.
As our research program has progressed, our knowledge about alignment processes has become more fine-grained and more sophisticated, as has our awareness of limitations of the approach. There is ample evidence from several of SFB 673’s projects that alignment phenomena can be detected in experiments, reconstructed in small-scale theories and simulated using various methods such as real or virtual robots or programs mimicking speakers’ interaction. A case in point, illustrating a shift to a perspective complementary to the alignment approach, is the discovery that synchronization of timing in human verbal actions is an important prerequisite for a process model of communication. Dialogue participants’ timing mechanisms do not seem to be controlled and planned to the same degree as, for example, the selection of dialogue acts or the production of explicit feedback. Another limitation pertains to surface vs. structural alignment: It is well known in linguistics that a wide range of variation in surface forms is possible without obstructing content alignment. We are taking this into account by paying more attention to semantic and pragmatic levels and by specifying what “classical” processes are needed to supplement the alignment model.
Midway-through our research agenda, the ZiF workshop will expose our research progress to an international, interdisciplinary round of experts for discussion and critique. One motive derives from the observation that many communication researchers tend to have implicit assumptions when talking about communication and its modeling and that these should be made explicit. A second motive is to set up a discussion between the different streams prevalent in communication research and dialogue theory. In particular, the workshop will address the following questions:
Synchronization of timing: how important a prerequisite?
Surface vs. structural alignment: which is more important?
Which role does common ground of the participants in dialogue play?
Interactive alignment vs. (possibly implicit) negotiation: automatic or strategic?
What are the requirements for a process model of communication?