Notes: Xin (2012): A Critique of the Community of Inquiry Framework

Xin, C. (2012). A Critique of the Community of Inquiry Framework. The Journal of Distance Education, 26(1). Retrieved from

Thanks to Stephen Downes for pointing this paper out. I’m up to my eyeballs, processing data for my Community of Inquiry based MSc research, and could have missed this.

The Community of Inquiry model provides a framework for describing interactions within a community or classroom environment. It involves using textual analysis and coding of messages to interpret the type of interaction for each message – whether it involves social, teaching, or cognitive components. As I’ve been coding the data for my thesis, I’ve been adding as many types of “presences” as are appropriate – a message may include a number of things, indicating social, teaching and cognitive presences in a non-exclusive manner. I’m imagining each message having its own little Venn diagram for Social/Teaching/Cognitive component, as per the CoI model. It’s a simplification and abstraction, certainly, but looking at the coded output, I think it’s still got a fair bit of fidelity to describe the interactions at a high level. In my data, I’m also adding coding to describe the type of content (links, images, attachments, embedded media, etc…) as well as how involved the message is (is it a simple one-liner? a 2 paragraph response? a multi-page essay?) – and I’m thinking about how to include data on the timeline of the discussion (how rapid were the responses? staccato rapidfire conversation, or long drawn-out periods of silence?) I’m still thinking about how to represent that kind of data for an online discussion, but I think there’s something there, there.

In a total aside, I’m reading Foundation’s Fear (by Asimov & Benford), and it’s rekindled thoughts about psychohistory and the like – models at such a high level of abstraction that individual actions do not matter. That’s kind of how my thinking as shifted with respect to the use of CoI for my online discussion data – it’s not the individual messages that are interesting to me at this level, as much as the overall patterns and flow…

Now, on to the article!


This conceptual paper critiques the popular Community of Inquiry framework (CoI) that is widely used for studying text-based asynchronous online discussion (Garrison, Anderson, & Archer, 2000). It re-examines the three main aspects of CoI (cognitive presence, social presence, and teaching presence) and their relationship, and further highlights the specificity and complexity of online discussion forums. The paper explains that the multi-functionality of communicative acts which often combine instruction, knowledge construction, and social interaction in a single utterance. It further argues that all online expressions are inherently social. It clarifies the confused relation of cause and effect in CoI and specifies the leadership functions of teaching presence and how they are intertwined with social and cognitive presence. A game metaphor and Gadamer’s notion of “play” are employed to explain the dynamics of online discussion forums. The article concludes with a discussion of the practical and methodological implications of its main arguments.

on the coding itself:

Each of the three components of the CoI is an abstraction from the whole, i.e., the communication process. Often all three aspects are performed simultaneously in a single communicative act, e.g., a sentence or paragraph, their precise function depending on what said previously leading up to that point, the contexts, and the dynamics of the discussion.

Agreed. I think it’s essential to be coding each message in a way that documents each aspect. A message may include a couple instances of “social”, one of “teaching” and a few “cognitive”. The Venn diagram, rather than a single “this is a Teaching Presence Message”… In addition, other components of the message and discussion also convey meaning – the rapidity of the responses conveys a message about engagement or urgency. The depth of the responses conveys a message about importance or complexity, etc…

I think that by documenting all presences or aspects indicated in each message in a discussion, it should be possible to capture the more fluid communicative functions while still providing data at an abstract level that can be analysed as a whole.

on abilities vs. performances

Why does this distinction matter? It has to do with identifying what is operative or at work in the medium as opposed to capacities or desiderata that may or may not be active forces in the communication process. Abilities are not the same as performances. Knowing a rule is not the same as obeying the rule. The CoI encourages one to think about what a successful conference would entail, but it does not adequately account for how to get there or make it happen. The communicative functions provide the operative means for accomplishing that. They are important for clarifying what one can and must do online so that participants are able to play their part in a successful educational experience.

I may be misinterpreting the CoI model, but when I’m coding my data I’m not looking at abilities or roles – I’m looking at what actually happened in the message(s) – the performances. I don’t think CoI is incompatible with that view, and it shouldn’t force an emphasis on potential (and undocumented) abilities or the nature of institutional roles.

The Communicative Functions are interesting, and might make it more explicit that the actions are what are important, but they should be compatible with the CoI coding of presences.

on desired states vs. modeling behaviour:

Desired states of affairs such as “open communication” are descriptors that may or may not apply in any given situation no matter how much they are referenced as goals. This is why modeling behavior is important. A teacher could say, “Feel free to speak up” but until someone does so (e.g., by saying something daring) and is accepted as normal (e.g., acknowledged by the teacher), the norm is only an unrealized ideal. Again, this is an example of how communicative functions (CFs) operationalize a rule or a desired state of affairs. “Presence” must be understood as just such an operative component of the discourse. In this sense, the cognitive, social, and teaching presences are best specified in terms of the functions that make them “present.” “Presence” is an effect of what I call “functions.” In other words, CFs are actions that cause the effect of someone or something being present.

Bingo. This is something I struggled with as well – how do you know if a message is something that indicates “open communication?” In my coding, I have been looking at the actual messages as indicators of this – it’s not open communication because someone says open communication is encouraged, it’s open communication when a message shows that someone is communicating openly. This could be as simple as asking a question, or challenging something that was posted earlier.

on social presence, and confusing cause and effect:

The three categories of social presence also confuse action with outcome. Affective expressions are communicative actions that can cause certain effects. Open communication and group cohesion are not actions but positive effects or outcomes of successful forum management. Putting the three together confuses cause (what one does) and effect (what the action leads to) and obscures what a participant should (or should not) do online in order to achieve the desired state of affairs.

Now this is interesting. The indicators given for defining “social presence” are somewhat muddy or confused. Some look at actions directly, some look at indicators of desired outcomes. Is this just semantic nitpicking, or an indicator of something deeper?

I’m wondering if it might be interesting to analyse my data using Communicative Functions in addition to CoI…

on teaching presence:

Earlier I raised the concern that from a strictly communicative perspective, “instructional management” and the “design and organization” of content and activities such as “setting curriculum and methods” (Garrison, et al, 2000, p. 89) contribute to the learning experience but do not constitute the teacher’s presence online since the only way to be present online is to express oneself in words. This is important because no matter how thoughtfully a conference is designed and structured, without the active involvement of the teacher in the process of dialogue, its success is left to chance.

This is consistent with what I’ve seen – it’s not the initial design of an activity or discussion, as much as the continued authentic interaction by both teacher and student.

Xin maps the CoI teaching presences directly to Feenberg’s Communicative Functions. If they are able to map directly, doesn’t that suggest that they are roughly equivalent (but with different vocabularies)? If my goal is to gather a high level abstract analysis of the discussions, the exact vocabulary used to describe the interactions doesn’t matter all that much.

on multiple contexts in messages:

The highly context-dependent nature of online discussion also explains why facilitation is not just instructional, or social, or intellectual; it is often all of them simultaneously. Process and content are intertwined at the communicative level. Teaching presence both sustains the social relations of communication and advances understanding of the subject matter at hand. Because the messages that perform communicative functions often encapsulate cognitive contributions, the effective use of those functions fulfills an intellectual as well as a communicative role

Agreed. This is why I’m coding my messages using CoI in a way that can represent multiple contexts for each message. I may have to write up something to describe how I’m doing this – I haven’t found any solid resource on the mechanics of CoI coding at that level.

discussion and conclusion

Xin’s strongest point involves the confusion of cause and effect in indicators for CoI presences. Some of the coding is directly triggered by actions – someone uses an emoticon or shares an autobiographical narrative, and that indicates a social presence – and some are indirectly triggered by actions that are interpreted to imply desired goals or intents – someone appears to speak freely, so we infer “open communication” and a Social Presence.

This may just be confusion about how CoI can work – I’m using it as a hierarchy. Each of the 3 “presences” is represented by 3 categories, and each category is represented by a number of indicators. The wording used to describe the indicators may be a bit inconsistent (the cause-and-effect confusion Xin describes), but they can be interpreted consistently by a researcher to mitigate that.

Furthermore, the analysis of the communicative functions of online talks should be considered together with other aspects of interest — who said what, how, why and when.

Exactly. CoI is just one component of analysis – I’ll also be looking at other components to build the full picture.