By Gordon Rugg and Sue Gerrard
This is the second in a series of posts about explicit, semi-tacit and tacit knowledge.
It’s structured around a four way model of whether people do, don’t, can’t or won’t state the knowledge. If they do state it, it is explicit knowledge, and can be accessed via any method. If people don’t, can’t or won’t state the knowledge, then it is some form of semi-tacit or strictly tacit knowledge, which can only be accessed via a limited set of methods such as observation, laddering or think-aloud.
This is summed up in the image below.
The previous article in this series gave an overview. In the present article, we focus on do and don’t knowledge, i.e. explicit and semi-tacit knowledge.
This category includes various types of memory, communication and reasoning identified in psychology and related fields; for instance semantic memory (memory for facts, such as Paris is the capital of France) and episodic memory (memory for events, such as The time we went to Paris). These types of knowledge can be stated explicitly in words, and are amenable to being represented in classical formal logic and in software.
A first key point is that not all knowledge falls into the do category. A lot of knowledge cannot be stated explicitly in words, and cannot be easily represented in classical formal logic or in software. Handling these other types of knowledge can be done rigorously and systematically, but requires different approaches.
This category includes various types of semi-tacit memory, communication and reasoning. A key feature of this category is that the knowledge involved is accessible, but is only accessible via a limited set of methods, such as think-aloud, observation, laddering, and card sorts, for reasons outlined below.
Short term memory: Knowledge in short term memory can be stated explicitly in words, but because its duration is very brief, an interview or questionnaire or focus group after the task would almost certainly fail to access it. The only realistic chance of accessing that knowledge would be to ask for it during the task; via think-aloud technique for instance.
Taken For Granted (TFG) knowledge is knowledge that the speaker/writer assumes to be so familiar to their audience that there is no need to mention it. This is one of Grice’s communication principles; it helps keep communications down to a manageable length. For instance, when you say something about your aunt, you don’t usually need to mention that she is female, because you can take it for granted that the person you’re talking to will know that this is part of the definition of “aunt”. However, if you’re talking about a second cousin once removed, you may need to say explicitly what that means, since not everyone knows the meaning of this term, so it can’t be taken for granted.
Taken For Granted knowledge is a common cause of misunderstandings and communication failures. These are often serious precisely because the concepts involved are likely to be fundamental ones; the fact that they are fundamental is what leads to people taking them for granted and not thinking that they need to be said explicitly. This situation often arises when experts in a field are interacting with non-experts, such as medical experts dealing with patients, or engineers in safety-critical fields dealing with members of the public. A typical example involves the expert telling the non-expert that if X happens, it will lead to Y, and taking it for granted that the non-expert knows how bad Y would be as an outcome.
Not Worth Mentioning (NWM) knowledge is similar to Taken For Granted knowledge, in that it involves knowledge being filtered out of a communication. The first key difference is that usually NWM knowledge is consciously, rather than unconsciously, filtered out. The second is that NWM knowledge is by definition considered by the speaker/writer to be trivial, whereas TFG knowledge is often considered by the speaker/writer to be extremely important.
NWM knowledge can cause problems when the speaker/writer is mistaken in their judgment that the knowledge is trivial. Often, something is trivial in one way but extremely important in another way. For instance, airborne dust may not be a problem from a respiratory health viewpoint in a workplace where all staff wear breathing gear, but may be a huge risk in terms of its potential for exploding, if the dust is from flammable materials.
Recognition and recall are known in different fields by various names, including passive memory and active memory respectively. Typically, people are better at recognising something when it is in front of them than they are at actively recalling something to memory. A classic example is asking someone to name states in the USA, or countries in the EU; people typically perform better at recognising the correct names from a list than at recalling them if asked to produce a list from memory.
Matching knowledge types onto elicitation methods
Each of the knowledge types above has significant implications for choice of appropriate elicitation method. These are summarised in the table below.
|Type of semi-tacit knowledge||Appropriate elicitation technique(s)|
|Short term memory||Think-aloud, to catch someone’s thoughts in their own words before they’re lost from memory; Observation, to see what they do|
|Taken For Granted||Observation, to see what they do; Laddering, to unpack their knowledge systematically|
|Not Worth Mentioning||Observation, to see what they do; Laddering, to unpack their knowledge systematically|
|Recognition and recall||Showing examples, to jog their memory, combined with think-aloud to catch their short term memory responses|
Other methods can also help when dealing with these types of knowledge. The table below gives examples; there’s a more detailed description here of different forms of report, including the methods in the table.
|Scenarios (“what would you do if…?”||Rare or dangerous situations that can’t be studied directly||People’s predictions of their own future behaviour are often unreliable|
|Critical incident technique/hard case technique||Teasing out rare but important knowledge||People’s memories of past events are often systematically distorted and unreliable|
|Real-time commentary on someone else||Accessing short term memory knowledge about tasks where think-aloud isn’t possible||People’s judgments of the reasons for other people’s behaviour are often systematically distorted and unreliable|
These methods can give valuable insights that might otherwise be missed, but the insights need to be treated with caution. There’s an extensive literature on biases in human judgment, memory, and beliefs about other people; a recurrent finding is that these biases are widespread, even among experts in the relevant field. An important related finding is that there’s little or no correlation between how vivid a memory or belief is, and how accurate that memory or belief is. This is why the table about knowledge types and elicitation techniques focuses on real-time techniques involving the person talking about their own performance, where the scope for biases and memory distortions is less.
It’s also possible to improve the chances of capturing TFG and NWM knowledge by using a representation that shows all the possible options, such as a table. When you populate the table with the knowledge you know about, you see at a glance where knowledge is missing.
Unfortunately, a lot of TFG and NWM knowledge involves isolated pieces of knowledge that don’t fit into a neatly tabular structure, such as “of course it’s more important to keep drivers alive than to protect the paintwork of the car”. Laddering uses a more flexible representation than a table, and offers a better chance of catching TFG and NWM knowledge by unpacking goals and explanations systematically, but laddering still doesn’t guarantee completeness. This is why we often advise using two or several appropriate methods in combination, to improve the chances of catching something with one method that the other missed.
The types of knowledge described above have serious implications for anyone trying to elicit valid knowledge from human beings. By their very nature, they require specific elicitation methods that correspond to the issues of memory, communication, etc that are involved; methods such as card sorts, laddering, think-aloud and observation.
In the next post we’ll look at can’t (i.e. strictly tacit) and won’t knowledge, their implications for choice of elicitation methods, and their implications for other areas such as human error, and education theory and practice.
Links and references
Card sorts: A short tutorial, and an in-depth tutorial.
Laddering: A short tutorial, and a longer tutorial (PDF download).
- An overview
- The STROBE method for making inferences about a working environment from observation.
- Identifying issues with the built environment, part 1
- Identifying issues with the built environment, part 2
Reports: An overview that includes think-aloud, critical incident technique, hard case technique, and scenarios.
Aesthetics and design:
Reference to original article:
The Maiden & Rugg ACRE reference is: Maiden, N.A.M. & Rugg, G. (1996). ACRE: a framework for acquisition of requirements. Software Engineering Journal, 11(3) pp. 183-192.
Copyleft: You’re welcome to use the image in this article for any non-commercial purpose, including lectures, as long as you retain the copyleft statement within the image.
Pingback: Tacit knowledge: Can’t and won’t | hyde and rugg