Grounding in computer-supported collaborative problem solving

Grounding

Let us say that it is common knowledge in a population P that X if and only if some state of affairs A holds such that:

• Everyone in P has reason to believe that A holds.

• A indicates to everyone in P that everyone in P has reason to believe that A holds.

• A indicates to everyone in P that X.

Clark and Marshall (1981) pointed out that using such a schema requires a number of assumptions in addition to the mere accessibility or presentation of information. Clark and Schaefer (1989) went beyond this, claiming that feedback of some sort was needed to actually ground material in conversation, and that this grounding process was collaborative, requiring effort by both partners to achieve common ground. They point out that it is not necessary to fully ground every aspect of the interaction, merely that they reach the grounding criterion: “The contributor and the partners mutually believe that the partners have understood what the contributor meant to a criterion sufficient for the current purpose.” What this criterion may be, of course, depends on the reasons for needing this information in common ground, and can vary with the type of information and the collaborator’s local and overall goals. They also point out that the conversants have different ways of providing evidence which vary in strength. These include display of what has been understood, acknowledgments, and continuing with the next expected step, as well as continued attention.

This study addresses grounding when two subjects (a) solve a problem together and (b) communicate via a groupware. Grounding in collaborative problem solving is probably more tightly constraint than in simple conversation. The specific features of the task, which affect the grounding process, are presented in section 5.1. We focus here on how the use of groupware may impact on grounding mechanisms. The term 'groupware' refers to a large variety of synchronous and asynchronous tools for communication and action including written communication (electronic mail, news groups, bulletin boards, MOOs, ...), oral communication (audio link, voice messages, ...), visual communication (video link, video messages) and shared workspaces (shared editors, whiteboards, task-specific shared interfaces, ...). These tools are generally not used alone but organized into different configurations to support decision processes in groups (McLeod, 1992), collaborative design (Fischer et al, 1992), meetings (Shrage, 1990), and so forth. This study is concerned by virtual collaborative environments (VCEs), a category of groupware aiming to empower collaborative work. There exists a large variety of VCEs. We do not pretend that the grounding mechanisms observed in one VCE will be identical with another VCE. The VCE system we have chosen is a MOO environment plus with a whiteboard. These tools are described in section 5.

1. Grounding in a MOO

A static document such as this report can hardly give an idea of interactivity in the MOO. The reader who is not familiar with the MOO should read our description of basic MOO interactions (Appendix 1) and even connect to our MOO (tecfamoo.unige.ch).

As a research tool, MOO environments paradoxically constitute both ecologically valid environments and laboratory devices: on one hand, our experiments are run with a standard MOO, used in various communities, but on the other hand, since the MOOs includes a programming language, we can tailor a sub-area of the MOO to experimental purposes and create the objects necessary to do the task. Moreover, because of these programming features, we were able to add facilities for recording automatically the trace of all actions and interactions. This reduces the traditional cost of studies on collaborative learning.

Clark and Brennan (1991) established that the cost of grounding varies according to the medium. We review now several parameters and will provide examples and quantitative data when presenting the results of our experiments:

• 
  Production costs (articulating or typing the message). MOO interactions have to be typed on keyboard. In addition to the message itself, the user must type the communication command -either 'say' either 'page'- followed by the name of the message receiver³. The cost of production is high. Note, that we ran two experiments with voice conversations to have an appraisal of these costs.
  
• 
  Formulation costs (how easy is it to decide exactly what to say) depend on the task. In the MOO, in addition to choose the content of his message, the user must reason on the position of his partner(s) to choose the communication command. The communication commands are different according to two parameters, space and privacy. A 'say bla bla' message is local and public: it is received by any character in the same room. A 'page Spiridon bla bla' message is global but private: it will be received only by Spiridon, but wherever he is located⁴.
  
• 
  Reception costs (listening to or reading the message, including attention and waiting time). In the MOO, reception costs are threefold. First, there is the time necessary to read incoming messages. Second, when a lot of information is suddenly displayed on the screen, finding one's partner messages requires a real effort⁵. Third, when working with several windows (as in our experiments) the cost is increased by the necessity to maintain visual attention on the bottom of MOO window⁶. Conversely, when the receiver is away for some time, he changes one of the displayed features of her character ('mood') to inform potential senders that she is not looking at MOO window for a while.
  
• 
  Repair costs vary according to the type of repair. If the subject repairs by sending the same message but changing one or two words, he can generally uses some command which redisplay in the entry zone his last message or a previous one. At the opposite, if repair involves complete rephrasing, then the cost is high since formulation costs are high.
  

• 
  Co-presence (can see the same things). MOO environments rely on a spatial metaphor. The characters move from room to room, see and interact with objects in their room and use different communication commands whether their partner are in the same room or not. The accessibility of knowledge is bound to the MOO topology: A can infer that B has access or not to information-X if A knows both where B and where information-X are located. Knowing that the partner can access to some information is the first level of mutuality of knowledge. Hence, the process of grounding in itself acquires a spatial dimension which can be traced by the observer. The drawback is that the subject ability to infer what her partner can access depends on her understanding of how MOO functions. This introduces heterogeneity in our sample. Note that MOO environments provide users with information about the respective position of their partner.
  
• 
  Visibility (can see each other). Visibility is important in grounding. For instance, in video communication, agents report to be more aware of their partner's attentional state ). (e.g. "I could readily tell when my partner was concentrating on what I was saying") when the CMC setting includes video-conference than when it is only audio (Watts, Monk and Daly-Jones; 1996). Since MOOs are text-based, 'visibility' is obtained by a verbal description of a character, its 'mood' and where it is located. For instance, every time agent-A arrives in or leaves a room where agent-B is, both agent-A and agent-B are informed of this arrival/departure. Moreover, agents can type the 'who' command to see where are the other agents (and how long they have been active.)
  
• 
  Audibility (can hear each other). Using 'say' instead of 'page' is a common source of miscommunication in our experiments, when A believes that B has received his message while it is not the case⁷.
  
• 
  Cotemporality (messages received at the same time as sent). MOO rely on synchronous communication. Actually, synchronicity is more encompassed into the practices socially established around that tool than due to technical constraints. Technically speaking, a set of bits will take more or less the same time to cross the net whether it is generated by a MOO client or by an electronic mail client. Is it synchronous when it takes 2 seconds and asynchronous when it takes 10? Synchronicity rather describes the sender's expectation that the receiver is waiting for his message. In collaborative problem solving, synchronicity involves that the partner is carrying a more or less similar reasoning.
  
• 
  Simultaneity (can both parties send messages at the same time or do they have to take turns) Grounding mechanisms are sensitive to the quality of turn taking, a delayed acknowledgment may be perceived as a request for explanation. Turn taking rules, as acquired through voice conversations, are modified in the MOO. The main constraint in voice conversations, the fact that people cannot talk simultaneously, is relaxed in MOO conversation: A and B can type simultaneously. In section 6.6, we report several cases of parallel or interwoven turns.
  
• 
  Sequentiality (can the turns get out of sequence). The effects of parallel interactions is increased when more than two people talk together.
  
• 
  Reviewability (can they review messages, after they have been first received). In general, audio message are not reviewable: once they are pronounced, they disappear. The MOO provides users with the interaction trace. When a new interaction is typed, the window scrolls one or a few lines up. According to the size of this window and the nature of commands, the user can generally see between 10 and 20 interactions. Reviewability compensate the non-sequentiality of turns: If the user receives an answer which does not match with the last question, she can look upwards which questions is actually concerned by this answer.
  
• 
  Reviseability (can the producer edit the message privately before sending). With most MOO clients, the user can edit his message. Given simultaneity and sequentiality problems, it occurs, that one types a long message but do not send it because, in the meanwhile, one of interlocutors said or did something which makes this new message irrelevant.