Clay Spinuzzi, University of Texas at Austin
At the meso level, actions are
These actions involve communicating with other actors and with community stakeholders using tools. So when we examine them, we may ask questions such as:
Communicative Event Models (CEMs) are a way to answer those questions. CEMs let you map out the chains of communicative events that people use at your site, allowing you to detect common patterns (or scripts) and divergences.
Communicative events are events in which one actor "hands off" a communication to another; emails, phone calls, document handoffs, and "do not disturb" signs on doors could all be considered communicative events. Usually these "handoffs" are instantiated in texts or speech.
Communicative events tend to follow patterns, and these patterns tend to cycle within the larger activity at the research site. For instance,
Since these communicative events involve identifiable transactions, we can detect them (especially if we ask the participants to help interpret them). And in most cases, these events follow regular and recurrent patterns: people tend to develop or learn a sequence for communicating, and they tend to stick to that sequence unless other circumstances intervene. So examining communicative events tells us a lot about how people understand their work and how they handle disruptions..
One way to represent the communicative "handoffs" is through Communication Event Models (CEMs) (Hart-Davidson, "Modeling," "Seeing"). CEMs provide a simplified, easily comparable description of event sequences, a description that can help us detect patterns in people’s work, compare patterns, and see sequential divergences. Any given action is a contingent choice made in response to situational constraints. In the CEM, these contingent choices are essentially portrayed as strings of verbs and nouns. If we were to apply CEMs to longer segments of work, we should be able to formally detect consistent patterns, identify larger units of interaction, and consistently explore places where sequences diverge across workers or conditions.
In a CEM, you record communicative events (events in which actors exchange information by exchanging texts, speech, or other signs). To find these communicative events, start with the genres you identified in the GEM. Which ones are handed off? Based on the genres you identify, define symbols for frequently occurring events (such as face-to-face meetings, email, and phone calls). See Figure 3.
As you analyze how people accomplish things at your site, you'll look for their chains of communicative events. Think in terms of
To identify communicative events, examine your data for any sort of direct exchange that participants had with other people.
Now we have a good idea of the communicative events involved. Let's go to the next step: figuring out sequences.
As you look over your field notes and interviews, you'll notice that these communicative events often form patterns. For instance,
If you observe people who do fairly structured or repetitive work over a short amount of time - such as calling customers whose accounts are past due - it's easy to spot sequences in communicative events. On the other hand, when people do work over longer cycles, such as proposal writing, sequences might be harder to spot. But in most work, you'll find sequences of communicative events. That's because people tend to create structure in their work. That structure helps them to make better estimates, to make better guesses about what their collaborators are doing, to optimize their work, and to recover from disruptions.
So, to get a better understanding of why people do what they do, we need to determine how people chain their communicative events into sequences.
At this point, you have enough information that you can develop a Communicative Event Model (CEM).
Now that you have documented the communicative events in their sequences, you can depict these in a Communicative Event Model (CEM). In a CEM, you map these events, showing regular patterns and divergences. Once you have this map of texts, analytical possibilities open up:
To construct a CEM, start with a single set of field notes, artifacts, and interview.
At the end of the process, you should have a CEM that looks something like Figure 3.
You can repeat this process across a participant's observations, across the group, or across the organization. Consider generating separate CEMs whenever you see significant differences. These might include
At this point, you should be able to show how people follow event sequences and how those sequences do or do not differ. Now let's look at the discoordinations you may have detected.
Often, work suffers from action-level disturbances called discoordinations: points at which two or more texts just don't "fit." At these points, the sequence is interrupted, and the participant has to find a way to repair the coordination before she can continue the sequence.
When you look for discoordinations, look for these characteristics in the data:
When you detect a discoordination, mark it in a different color of ink - I suggest red - on your CEM.