Last year, I began using Slack in my classes, and in the early days of the Spring 2016 semester, I wrote a post about how I thought it was going and how it seemed to be filling in many of the communicative modalities otherwise occupied by an LMS like Canvas. After completing that semester and having now reflected a bit on how the online environment in Slack supported a sense of community among the four sections (two of mine plus one each by Lee Skallerup Bessette and Jesse Stommel, I want to do some analysis of what that Slack community looked like in terms of participation. My comments here extend some thoughts I originally shared in a presentation back in May.
Basically, I want to look further into an assumption I made in my original post, where I talked about the sense of community developing in my class’s Slack conversations. But what is a community? And what does the content shared in that space tell us about that community? Does the fact that a slack group of 120 people (students + faculty) produced around 5,000 messages over the course of a semester mean that we were “successful” in some way? And finally, how does my role within this community impact my impression of its cohesiveness and sense of shared purpose?
Numbers and graphs can’t tell the whole story, of course, but they start to illuminate a few things about our Slack group. Slack only provides detailed statistics to its paying users, so to produce the data for all of the charts and visualizations that follow, I simply exported the public message archive and cleaned that up in a spreadsheet.
First, I was curious about who participated in this conversation and who didn’t. So here’s a pie chart of the messages by source:
Here and elsewhere, a “message” is something someone typed into Slack, including interacting with integrations and bots, e.g.
/catfacts. This chart shows that, while students did do a lot of messaging (and there’s an additional 2.4K private messages that I’m not including), most of the communicating in this Slack group was done by the three instructors, of which I personally made up the vast majority, and the next largest chunk overall was the output or responses from various bots and integrations.
Clearly, a lot of the “participation” in this community is actually just me. It’s even more starkly visible in a network graph. For the visualization that appears below (or full screen here), a node represents a user (student’s names have been removed), and the size of the node indicates how many posts that user contributed. A connection between nodes records at least one instance of a user
@-mentioning another user, which usually indicates a conversation. The thicker the line, the greater the number of mentions between the two.
I’m pleased by all of the connections in this graph, especially that it’s still relatively well-connected if you take me out of it (Right-click on my face and select “Hide”), but it’s hard not to worry about all the unconnected users floating out to the right. These are the same users that occupy the right side of this bar-graph:
There are plenty of legitimate reasons these users may have had for not participating as much as the rest of us. Anecdotal information can fill that data-gap somewhat: some students had difficulty installing the platform or getting used to it, and others used it frequently but tended to lurk more than they posted. Moreover, I suspect that if I were to make a similar set of graphs for how and when people talk irl in my classes, I’d probably find a similar curve with me doing most of the talking, a few outliers doing almost none of the talking, and most others finding themselves somewhere in the middle.
So what about the content? What kinds of things were we talking about in all these messages? Here’s a word cloud:
I know word clouds aren’t great at providing insights into a corpus, but I like that giphy stands out because it reveals how often we used the
/giphy integration, and since pretty the only reason to use
/giphy is to be silly, its popularity helps characterize the playful tone most often employed in our community.
To further explore tone, I looked into how we used emoji. Slack has a rich, easy-to-use emoji keyboard that also supports adding custom emoji, and, importantly, there are two ways emoji show up: 1) inside a regular message and 2) as a “reaction” to another message. Analyzing these two uses separately was essential to get a good read, because we used “in-message” emoji differently. For example, some groups tried to re-tell the novel we read using only emoji, so the emoji they used for recurring characters are inevitably going to show up quite a lot.In the first cloud emoji used inside of messages, one finds that the cat face is the largest and most popular image, due to the popularity of `/catfacts`. The following cloud actually gives a better sense of our community, because it is less subject to ironic, sarcastic, or diegetic uses. Taken together, these pictures of our community only provide a cursory glimpse of what, to me, really felt like an energetic, interactive community of learners engaged in the task of figuring things out and supporting each other along the way. I’m an optimist, though, and I look on the bright side of these data. Anecdotal evidence can’t avoid the trap of my own perspective as I look at this community through the lens of my experience with it. Clearly, I’m a biased witness, but my anecdotal, educated-guess feeling is that Slack affords community building in a way that is no worse than in-person classes. And if it’s more playful integrations help ease its adoption and encourage students to use those integrations to renegotiate the power relationships that otherwise could restrict conversation, then more the better.
This past Summer, Lee and I worked on a proposal to teach DGST 101 fully online, with Slack playing a major role in how we conceive of the class going forward. It’s my hope that, through the ways in which Slack’s adoption and continued use supported our sense of community across sections, it will also help provide a common ground for a community to develop via primarily asynchronous communication.