The human brain is not a system with unlimited memory, but a narrow processing unit with a limited working memory, only able to process a small number of items for a very short time. The number often quoted, is 7 plus or minus 2 items, lasting only 20 seconds. If we want to retain this limited number of items longer, then we have to internally rehearse them. This is the process of schema formation, retrieving and storing items in our long term memory. So obviously this system can easily be overwhelmed by useless information. In order to mitigate that, it is essential to consider the different categories of cognitive load, and free up any mental resources that are being wasted or misused.
Intrinsic load is the element of interactivity of material. It is the inherent difficult of the material which can only be managed by segmenting materials, isolating sub-skills into smaller steps. Extraneous load is considered the really bad load, since it is generally made up of instructional junk like poorly designed presentations, or conflicting elements that split the learner’s attention. This has to be rigorously avoided if any serious learning can happen. The mental effort involve in learning, comparing new schema to old ones is known as germane load. This can only be increased if extraneous and intrinsic load is reduced.
Teachers often fall into the “more is better” trap, and materials become more dense. For example, if they provide an audio, a full transcript, an L1 translation, and a detailed grammar reference, all on the same presentation slide. The learner’s brain isn’t processing the language, it is processing the relationships between the text, the audio, and the translation. It is splitting the learner’s attention between too many elements.
If the learner understands the audio, the transcript becomes redundant noise, information that they actively filter out. This is wasting energy that could be used for deep, germane processing. The redundancy problem is when the learners are not challenged by material that they already know. The learner has already established these mental schema, and now they are being forced to read something entirely redundant, increasing that awful extraneous cognitive load. The problem is that this doesn’t just waste valuable mental space, it actively steals precious processing power from their working memory. It halts real schema formation.
One of the worst examples of the redundancy problem is listening to a podcast in your target language while reading the subtitles below. This might be interesting for a beginner learner who is unable to understand the text. What’s happening is that the auditory cortex and the visual cortex are processing identical data. This actually slows down listening fluency because the brain relies on the crutch of text, rather than training the ear to parse phonemes. If you can read it, the audio becomes redundant background static.
Another example is providing L1 translation as a standard practice. If the learner knows 80% of the words in a text, then these translations are totally redundant. The eye is forced to scan them, creating visual clutter that distracts from the 20% of target vocabulary that is truly new.
We also provide a lot of support in communication activities like sentence stems and substitution tables. This is probably good for lower levels, but it can become patronizing, as if we are trying to get the learners to behave like parrots. The higher levels learners may have already automated that structure, and they will see it as irrelevant and distracting. The same could be said of helpful vocabulary lists that are appended to speaking tasks. For a stronger learner they are irrelevant, as they will perform the tasks in a more intuitive way without referirng to such lists.
Of course, Subtitles should not automatically appear, and should only be used if the learners actually require them. Learning should be more adaptive. For example, if a learner clearly understands a grammar point, they should not need to read long grammar definitions. These should only be activated if a learner is not able to to complete a task. In other words, if you make an error, then grammar explanation escalates: first a one-sentence rule, and finally, only after repeated failures, the full detailed explanation. If you answer correctly, the rule is deemed redundant and withheld. The same approach should be taken with translations, as they should only be shown if they are absolutely necessary
Some of the warning signs could be, for example, when learners complete a gap fill task too quickly. This is an indication that this schema is already automated, so the trainer will have to revert to a less scaffolded activity, matching the abilities of the learners. Another example, could be a learner has a lower level and struggles with a free speaking task. The trainer will have to revert to a more scaffolded activity.
The bottom line is that every redundant piece of information you put in front of a learner clutters their very limited working memory. They don’t need more data, they need liberated working memory.
So kill the redundancy and free the cognitive load.
Comments are closed