In 2006, former US president George Bush supported his embattled defense secretary Donald Rumsfeld with the words: “But I’m the decider, and I decide what is best.” This quotation quickly entered the folklore of political humour. But to psychology researchers, it revealed something fundamental about human language.
At that time, most Americans had not encountered the word decider. While this is a common word in some parts of the world, it refers to the part of a game that determines the winner. So how did people understand what it meant? They understood it because across all of the words that people know, the suffix –er often transforms a verb into a person (as in teacher, builder, dancer). Thus, a decider must be someone who decides.
The ability to extract general principles from a small number of examples is fundamental to language and literacy. In teaching children how to read, teachers introduce sets of words like chin, church, chest, chess, chop, to convey information about how to pronounce particular letters. This general knowledge might then be applied to new words like chick. In later years of primary school, children develop general knowledge about the functions of affixes. Through exposure to relevant sets of words like uncertain, unknown, unhappy, children become able to use affixes like -un in new contexts.
What’s a dunklomb?
In new research published in the journal Cognitive Psychology, we investigated the brain processes responsible for acquiring this type of general knowledge. We trained adults on a fictitious language, in which groups of individual words were bound together by a rule that was not disclosed to participants. For example, participants learned:
• a clinglomb is a small device used by cat burglars to cling to skyscrapers
• a dunklomb is the gadget used by royalty to dunk biscuits in tea politely
• a skimlomb is a professional tool which is used to skim cream off milk
• a weighlomb is the official scale used to weigh boxers before an important match
Our interest was not whether people could learn the individual words, but whether they could uncover the rule – in this case, the function of –lomb. We tested this by examining people’s understanding of untrained words like teachlomb when they were presented in sentences.
The remarkable power of sleep
Our key finding was that participants could apply their understanding of the rule (that –lomb refers to some kind of tool) to untrained words such as teachlomb. But this was the case only if participants were tested some days after training. Participants showed no such ability immediately after training.
In later experiments, we mixed words that conformed to the rule with examples of words that violated it – for example, that a mournlomb is “the cost of organising a wake to mourn a loved one”. This word is an exception because -lomb is supposed to refer to a tool. Crucially, the introduction of exceptions abolished people’s learning of the rule being taught. However, people did learn the rule if we inserted a period of overnight sleep between training on the rule-based examples and the exceptions.
Our findings fit neatly into dual-mechanism theories of memory. These theories argue that rapid learning of individual episodes is followed by a slower process of integrating that knowledge into long-term memory. The claim is that these processes rely on different brain structures optimised for fast and slow learning. Critically, these theories suggest that sleep may be a necessary component of the second, slower process.
Research in adults and children has shown that the brain continues to process new memories during sleep, allowing them to become stronger, more resistant to interference – and better integrated with existing knowledge. Our findings advance these theories by showing that sleep may also be necessary for discovering regular patterns across individual episodes and encoding these in the brain.
Helping children understand patterns
This research has clear messages for the teaching of language and literacy. Our work suggests that if teachers want to convey some general linguistic principle, then they must structure the information in a way that promotes learning. If a teacher is trying to illustrate use of the suffix –ing, for example, then presenting a child with a spelling list including the words standing, jumping, swimming, kicking, dancing, talking, nothing would be unlikely to facilitate learning.
For one, the letters –ing do not function as a suffix in the word nothing (noth is not a verb). And although the spelling alterations present in swimming and dancing are highly systematic, these items appear to be exceptions in the list presented. Our research suggests that lists that include such exceptions may disrupt a child’s learning of the pattern.
Most generally, our work adds to a growing body of research implicating overnight sleep in aspects of language learning. It suggests that key aspects of learning arise after classroom instruction – and thus reinforces the importance of proper sleep behaviour in children.