If you’ve ever done almost any sort of incident response at all, you know that coordination and communication are important. But did you ever wonder or stop to think about how it works?
That’s exactly what I’m going to be talking about today, how common ground is repaired and the techniques used to do so. This is what Herbert Clark and Susan Brennan studied in this paper.
Specifically I’ll be talking about “grounding,” which is the way that common ground is repaired. We’ll be talking about how grounding works across a variety of media, including video calls and text chat.
As far as coordination goes, there brought the two different types: coordination of content and coordination of process. To demonstrate this the authors of grounding in communication use the example of two people playing a duet on the piano. The need to coordinate content of course because they’re going to be playing the same piece but they’ll also need to adjust to how loud or fast the other person is playing which is coordination of process. This is where grounding comes in to play.
Given that we are seeking the grounding criterion then we must keep looking for evidence that we’ve been understood. We could take an approach where we only look for some sign that the been misunderstood, we should be looking for “negative evidence.” While this might sound reasonable on the surface it was causing a lot of problems for us, there’d be a lot of cases where we thought everything was fine, but of course we would have been heard and understood at all. This is why people tend to look for “positive evidence” instead.
There are three different major or common types of positive evidence that people look for. That isn’t to say that there are only three, these are the main ones.
The first kind and the most obvious kind are acknowledgments. These are things like “yeah” or “uh-huh.”
The second is called the “relevant next turn.” This is where I ask a question and then you answer it, creating a pairing. If you reply to my question with something that doesn’t make sense to me as an answer, then I take this to mean you haven’t understood and attempt a repair.
The authors use the silly example of a Marx brothers skit:
Miss Dimple: Where can I get a hold of you?
Chico: I don’t know, lady. You see, I’m very ticklish.
Miss Dimple: I mean, where do you live?
Chico: I live with my brother.
Depending on the content that I’m trying to convey, I’ll use different strategies for grounding. These largely fall into two categories: references, when I’m talk about someone or something and verbatim, when I want to make sure you get the information exactly right, like in a phone number or an address.
What makes a medium?
We’re all familiar with a number of different communication media and probably also have a sense of which ones are better for different tasks, but what makes that so? There are a number of traits that can be present or absent that makes a medium better or worse in different cases.
Different media have different properties that help shape what costs they incur and in what situations that matters. Before I talk about costs though, I’ll go over the properties.
- Copresence – We’re in the same physical space.
- Visibility – We can see each other
There are varying degrees to this. In a video call I can still see you, but I can’t see what else you might be looking at or what else you’re doing. It might also be harder for me to see some gestures.
- Audibility – We can here each other.
This means we can (and need to) time our communications so we don’t speak over each other.
- Cotemporality – I get the message at about the same time that you produce it.
- Simultaniety – We can send and receive at the same time.
While some forms of text chat might come to mind, smiling while you say something fits here. Most forms of text chat don’t have this. One that does is
talk. This is because I can compose a message, but you don’t see it until I send it, you can’t see it as I write it.
- “Sequentiality” – “turns cannot get out of sequence” or rather they need to stay in sequence to be sensical
Email and similar don’t have this, but in those cases that may or may not be a bad thing since “interruptions are not as impactful.”
- “Reviewability” – I can review your message.
Any recorded medium or even ephemeral chat channels have this as long as they’re active.
- “Revisability” – I can revise a message that I’m developing.
For example in email and similar mediums, I can edit my writing before I send it. You never see those edits. Whereas in speech, I can revise, but all those revisions are public.
Slack is like email in that I can revise before I send and you wouldn’t see those edits, but also I can revise after I send (and you may or may not have seen the original). This is not addressed the paper as none of the mediums they discuss directly have this. But as we can see these properties still apply so we can still think in these terms for new mediums.
The costs of grounding
Grounding is needed to repair common ground and allow coordination, but it doesn’t come for free. It takes time and effort on the parts of both the speaker (or message sender) and the receiver. Some of the costs are paid by both, some just by one. Also, more than one cost can apply in a situation. It’s also possible to make trade-offs between costs.
I’ll list them here and give a brief description of each of them. Other than knowing that grounding exists at all, I think understanding these costs is one of the most important things to understand from this paper. When you understand these costs it can become more clear how to effectly communicate and coordinate whether in incident response or just in the normal course of your day. It can also help as you develop tools. How does the tool or automation communicate with the user? How do they coordinate?
- “Formulation Costs”
This is essentially the cost I pay when I take time the think before I speak (or otherwise compose my message).Complicated communications cost more than simpler ones. Regardless of complexity, communications that need the be perfect cost more. (Another way of thinking of this is that communications that are critical or mediums where revision is difficult costs more in formulation time and effort). This is our first hint that we can trade off different costs across media.
- “Production Costs”
This is the actual saying or composing of the thing. An example of how this differs in different mediums is maybe the difference between typing on a computer and T9 input? This cost (or effort) is highly variable across mediums. Speaking is low for most people, but typing and handwriting are harder. This leads us to our next insight, that people may change the message depending on the medium. We’ve all likely experienced this ourselves and we can see this occur in other research as well. This might look like using more words when speaking, but fewer when typing depending on how fast you type or how difficult you find typing.
- “Reception Costs”
Listening is generally easy, and reading harder, although it may be easier to read than to listen to complicated instructions or abstract arguments. It also softs to have to wait while the speaker produces a Turner. This way takes its toll in keyboard conversations when addressees must suffer as they watch an utterance appear letter by letter with painstaking backspacing to repair misspellings.
- Understanding Costs
Some words or concepts take more effort than others to understand. Medium can make this worse when context or other clues (such as tone or timing) are missing.We can see this in email, because both sides have “to imagine appropriate contexts for both the sender and message and to remember what the message is in response to.”
- Setup Costs
This is literally the cast of starting a new conversation. Face-to-face this can he very low: I only need to catch your attention then start talking. It’s higher in phone calls and may be even higher still in emails (and letters).
- Delay Costs
“The costs of delaying… in order to plan, revise, and execute” what you’re going to say. Costs of this type are very high face-to-face because of how interruptions are interpreted. I may accidentally signal that I’ve dropped out of conversation entirely. This creates a pressure on me, as a speaker, to say things that to hurry and say things, which means I may need to reexplain them later, which itself will be a higher cost. Even when listeners know they’re waiting for “production” time, the listeners incurring a cost in that waiting. This cost is very low or even non-existent in asynchronous communications like email.
- Asynchrony Costs
When people are talking, face to face or over the phone, they can use timing as part of their communication. Any technique that relies on timing, when you’re not face to face has higher costs. We can see this when even slight delays are introduced, like on video calls where it can be difficult to start your turn or know when others are done. The more asynchronous you are the higher this cost is.
- Speaker Change Costs
Face-to-face or at least in person this is easier. When all those cues are unavailable those costs are higher. This is another cost that we can see in video calls with even slight delays. In email and similar media this cost is so high that changing speakers basically the same as starting a new communication. When speaker change costs are high, people tend to try to do more during their turn.
- Display Costs
In person, I can point at things, gesture, and show you things. Where as in other media, some gestures are lost, though we do have the ability to do things like share documents, such as in a video call.
- Fault Costs
The cost of making a mistake in your communication. How big of a mistake determines the cost. There are tradeoffs here that you can make, you can pay higher formulation cost to try and have a communication that doesn’t need fixed. You could spend less on formulating, but then risk spending more when there is a fault. This also is a trade off with the next cost.
- Repair Costs
This one is also highly variable across medium. In speech, this cost tends to be low and expected. Whereas in email the cost to repair some fault after its detected can be very high and unexpected. People also tend to prefer to initiate report for their own messages and tend to find this easier, whereas repairs initiated by others tend to be harder.
High costs of repair in one medium can drive a switch to some other medium. Most of have experienced this, perhaps even in an incident, where we may switch from Slack to Zoom.
For example, are there situations where users pay high revision costs of their input isn’t perfect?
- Grounding is the process to repair common ground.
- This is what allows people to coordinate and communicate.
- People look of positive evidence that they are being understood.
- Knowing how grounding works can help communication and coordination in incident response.
- Exactly what about grounding consists of varies on the communication medium. Email is different from text chat which is different from being in person, etc.
- When you know what constraints that media can place on grounding, you can plan and choose which wants you want to use for different situations.
- By understanding how people communicate and coordinate effectively, you can help support those patterns in your own tools and processes.
- All communication methods have some cost associated with them, some paid by on the speaker, some by the listener, and some by both.
- There are trade-offs that can be made between many of the costs.
Subscribe to Resilience Roundup
Subscribe to the newsletter.