Thinking out loud

from Wikipedia, the free encyclopedia

Thinking aloud is the audible verbalization of thinking .

One application in computer science is the Thinking Aloud test . He is considered one of the most important methods for realistic evaluation of user interfaces ( English user interfaces ). In cognitive psychology and linguistics , protocols of thinking aloud are regularly used for research purposes; H. the test persons should not only work on a given task, but also report what they are currently thinking, e.g. B. Which sub-step you are currently editing in which way. It is important that all thoughts are mentioned, including those that appear irrelevant to the test subjects. The individual statements are recorded as objectively as possible, i.e. without any interpretation by the experimenter. Often the thinking aloud is also recorded (for example with a dictation machine) to facilitate a subsequent analysis. The investigator's role is to encourage the subject to continue speaking during pauses. Experienced subjects often find it easier to “think aloud”.

The aim of thinking aloud is to get a picture of which cognitive processes are taking place while working on a task. Not only the result, but above all the solution provides information about it. To interpret these protocols, K. Anders Ericsson and Herbert A. Simon presented their "Theory of Thinking aloud" in 1984. They distinguish three possible relationships between verbalization and task-guided cognitive processes:

  • Type 1 - Direct verbalization: the inner voice is made audible, i.e. thinking aloud in the narrower sense. The test person says everything that goes through his head while working on a task. The problem-solving behavior is not influenced by this. The processes guided by the task and their verbalization are separate and different.
  • Type 2 - If the task to be solved is in the form of symbols (e.g. in pictures), the time required for the solution is longer, since the test persons first have to put the task into words. The verbalization of the observed information is produced by the task-directed processes.
  • Type 3 - Have the test persons explain why they are doing what they are doing. The result: more correct solutions, longer solution times, a different order of the partial steps, etc. The requirements for thinking aloud change the task-guided processes. Since no control is possible, the information provided by the test subjects is unreliable.


Other protocol methods:

Motor protocol:

In the engine log, the obvious physical activity is recorded and detailed. This method is particularly suitable for subjects with limited language options such as B. animals and children.

Eye Movement Protocol:

With this method the eye movements of the test persons are recorded. The focus is on when and for how long the gaze is directed where while working on a task. There is the possibility to simply follow the gaze to get information about it or to record the exact points of view with an eye movement camera. This second option is much more accurate and provides a more detailed picture.

literature

  • Karl A. Ericsson, Herbert A. Simon: Verbal Reports as Data . In: Psychological Review , Vol. 87 (1980), pp. 215-251.
  • Karl A. Ericsson, Herbert A. Simon: Protocol analysis. Verbal reports as data . MIT Press, Cambridge, Mass. 1993, ISBN 0-262-05047-1 .
  • Alison Green: Verbal protocol analysis in language testing research . Cambridge University Press, Cambridge, Mass. 1998, ISBN 0-521-58413-2 .
  • AH Joergensen: Using the “thinking-aloud” method in system development . In: Gavriel Salvendy, Michael J. Smith (Eds.): Designing and using human-computer interfaces and knowledge-based systems Amsterdam . Elsevier Science Publishers, Amsterdam 1989, ISBN 0-444-88079-8 , pp. 743-750.
  • JR Hayes: The complete problem solver , Lawrence Erlbaum Associates (1989).