Key ideas

Relationship between grounding and problem-solving

Learning effects seem to be related to the effort that group members devote to building/maintaining a shared understanding of the domain (Schwarz, 1995). This process in psycholinguistics is called “grounding” (Clark & Brennan, 1991).

The conversational participants 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.” (Clark and Schaefer 1989)

Ways of grounding

• display of what has been understood
• explicit acknowledgments with words such as “ok” and “right”
• continuing with the next expected step
• continued attention

Grounding implies

• anticipating
• preventing
• detecting
• repairing misunderstanding

Misunderstanding has different epistemic value in research on efficient communication and in research on collaborative learning:

communication breakdown to be avoided VS a learning opportunity

To repair misunderstandings, partners engage in constructive activities

• build explanations
• justify themselves
• make explicit some knowledge that would otherwise remain tacit and therefore reflect on their own knowledge

although this might slow down communication, this extra effort for grounding may lead to better understanding of task

Whereas Clark and Wilkes-Gibb’s (1986) notion of least collaborative effort emphasizes the economy of grounding, Schwartz (1995) pointed out that some effort is necessary to produce learning. Hence, we focus on optimal collaborative effort (Dillenbourg, Traum, & Schneider, 1996): Up to a certain level where communication becomes too difficult, misunderstandings are opportunities that, under some conditions, may produce learning.

(Might not need to be full-on disagreement, slight misunderstanding might be enough for learning - Blaye 1988).

(ref Suthers, 1995Suthers, D. D. (1995). {Designing for Internal vs . External Discourse in Groupware for Developing Critical Discussion Skills}. In Cognitive Science. about how an ideal CSCW environment is not necessarily an ideal CSCL environment - some difficulties are positive).

Building a common visual representation (textual or graphical) of the problem at hand contributes to the construction of shared understanding.

shared info spaces NOT shared knowledge spaces because two people look at the same thing, doesn't mean that they understand it in the same way

(based on Allwood et al 1991, Clark 1994) (see table 2)

• access (B has access to something)
• perception (B has seen sth)
• understanding
• agreement

• visibility
• audibility
• cotemporality
• simultaneity
• sequentiality
• reviewability (→ persistence)
• revisability (→ mutual revisability)

Media

• shared whiteboard
• MOO environment with text chat

How does acknowledgment rate (how often partners give feedback of having perceived, understood, and accepted partner’s contributions) vary according to media and content of interaction?

Efficiency of interaction and low production cost, but no persistence (and higher analysis costs). In typed interactions, peers acknowledge 90% of utterances, in typed, only 41% (see also Traum & Heeman 1997). Reception cost also higher - have to read messages, pay attention to message area (split visual attention between whiteboard and chat).

High grounding costs means that grounding is only performed when it is really necessary.

The whiteboard would serve to draw schemata that disambiguate chat utterances.

The whiteboard is primarily used to represent the state of problem solving and the chat is used for grounding information created on the whiteboard.

Rate of acknowledgment not related to verbosity. Also not related to success in problem solving.

Fewer redundant actions for frequent grounders, but given cost of grounding, redundant action might be an efficient strategy in some cases.

Acknowledgment for inferences almost twice as high (46% vs 26%) as for facts - due to probability of disagreement.

• MOO utterances acknowledged through whiteboard action
• Whiteboard action acknowledged through MOO utterances
• MOO utterances acknowledged through MOO actions

(interesting that because it is a MOO, acknowledgment can be not only typing, but also action - going into a room, asking a question to a person, etc)

• Napkin-model: two people discuss in a restaurant and draw sketches on napkin to disambiguate utterances
• Mockup-model: two architects draw a sketch of a new building and utterances aim to disambiguate what is meant by drawings

(ie. does the whiteboard support the chat, or the chat support the whiteboard - which is primary?)

They started from napkin hypothesis, but participants behaved according to mockup-model.

These results are interpreted in terms of persistence: More persistent information is exchanged through the more persistent medium. The whiteboard was used as a shared memory rather than a grounding tool.

Persistency of

• display (scrolling etc)
• information (how long info is valid, for example facts vs meta-discussion)

Non-persistent categories 44% in chat, 10% in whiteboard.

Persistent in display means higher probability that information piece is grounded at level 2 “perception”. A large part of chat messages simply grounded “I've seen what you wrote”.

Facts and inferences are very different in needed grounding of level 3 (understanding) and 4 (agreement).

The persistency of display also plays the role of individual and group memory. The tool maintains for the group a representation of the state of the problem: which facts have been collected, how these facts related to each other, which suspects have been discarded, and so on. Moreover, as the key inferences seem to be grounded before or just after they appear on the whiteboard, this tool not only provides the users with a representation of what they know but also a representation of what they—roughly—agreed on. A group memory is more than what the group knows; it is what the group considers as being mutually known.

A sentence can contain a large number of references (bindings) “the guy that went to the room that night”. Is the number of references limited by the scale of working memory?

Representations can be referred to as working memory. No need for information rehearsal (displayed with no effort) and has spatial organization. Can off-load individual cognition, but does not completely inhibit individual working memory - each individual still maintains a representation of the problem state.

This state is both close to the white board (constant visual access) and different (results from personal interpretation of external representation).

Collaborative environments should consider need of persistency

• chat tool augmented with more persistent area (FAQ in forums)
• whiteboard with less persistent area where non-persistent information could be automatically removed rather than requiring “clean-up”

Interesting to compare with Suther's work on representational guidance, persistent and non-persistent VS discourse and artefact, grounding VS deixis