Sleeping on banara and the fate of banana: consolidation of new words in the lexicon

How do we recognize the written word? While this seems trivially easy to us, we need to remember that words that are quite close perceptually can be drastically different in meaning. Fin and fine, crew and crow, and deck and desk may look nearly the same but their meanings differ considerably.

In the sentence “I went to the stone to pick up bread and milk”, you might imagine a very useful rock. The relatively small perceptual differences between ‘n’ and ‘r’ require the visual system to quickly and effectively decide which words are on the page, and activate their meanings. When two words – like stone and store – are visually close, how does the visual system solve this task?

One way psychologists think this process might occur is through a process called lexical competition. When a written word is seen, a set of similar-looking candidates is activated. So, when you see the word “stone,” your visual system may consider the word “stone”, but also “store,” and maybe “scone,” “shone,” or “stove.” One candidate becomes dominant as the material is processed further, and its meaning is then retrieved. In the sentence above, the visual word “stone” contradicts the implicit meaning in the context of the sentence, suggesting the word “store.”

The lexical competition hypothesis provides a way to test whether a new word has been integrated into the lexicon. In English, many words have visual neighbors – words that are off from each other by just one letter. According to a lexical competition account, when “stone” is seen, “store” also becomes activated because it is a candidate word, but “stane” or “slone” probably do not because these aren’t words.

Some English words are visual “hermits” – none of their neighbors are actual words. Consider anchor, banana, or syzygy. When the word “banana” is seen, “banara” is probably not activated because it’s not an actual word (nor is barana or bonana, but these are fun to come up with). These visual hermits provide the opportunity to explore how new words are integrated into the lexicon.

A recent study by Wang and colleagues, published in the Psychonomic Bulletin & Review, investigated whether participants who learned new neighbors of these visual word hermits (like “banara”) would show more evidence for lexical competition after a night of sleep. Sleep is thought to consolidate information accrued during the day into long-term memory, making it more easily accessible and searchable in the future. A new word such as “banara” should therefore begin to exhibit lexical competition the day after a good night’s sleep.

To test this, the researchers developed a list of 40 new words, each of which was paired with one visual hermit. Participants learned 20 of these new words, leaving the other 20 visual hermits with no new neighbor. After learning the words, participants were tested in two ways. They were asked to recall and write down as many of the new words as they could. They also categorized the familiar words as being man-made or natural.

After the first testing session, the participants left the lab. One group of the participants had completed this first session at 8 AM; a second group had completed it at 8 PM. Both groups then came back 12 hours later to complete the two tasks a second time. The morning crew had a day’s worth of activities, while the night crew spent the majority of that time asleep. (Because these are undergraduates, the researchers asked how much sleep the participant got that night, ensuring it was a mostly restful evening). Finally, both groups came back twelve hours later for a third session, which allowed the AM group a chance to sleep before being tested again.

Critically, the two tasks tap different aspects of word learning. The recall task simply measures the number of words learned, and is a test of the participant’s memory for the new words. On the other hand, the categorization task should expose differences between the 20 words for which participants learned a new word neighbor and the 20 words that remained hermits. If “banara” was one of the new words a participant learned, then he or she should be slower to categorize “banana,” because extra time is needed to recognize that the word was not “banara.” Furthermore, this should only be true if the new word has been integrated into the lexicon.

The figure below displays the results of the recall task. The researchers found an interaction – the PM group, who slept between Sessions 1 and 2, saw less of a drop in memory for the new words than the AM group, who remained awake. After a good night’s sleep, however, this group rebounded to match the AM group by Session 3. As has been previously found, and discussed on this blog, sleep boosts a number of cognitive faculties, so this finding isn’t all that surprising.


But how did performance on the categorization task change as a function of sleep? Remember that this task, unlike the recall task, would show evidence for whether the new words had entered the participants’ lexicons. To measure the lexical competition effect, the researchers calculated the reaction time for words that the participant did not learn a new neighbor for. Because these words remained visual hermits, that reaction time only includes time to process the meaning of the word. The researchers subtracted the reaction time for those words from the reaction time for words that the participant did learn a neighbor for. These words had to be discriminated (the screen says “banana” not “banara”) in order to be categorized.

The figure below displays the results of the categorization task. In the first session, neither group showed a competition effect, suggesting that the new words had not been lexicalized yet. In Session 2, however, the PM group showed a large competition effect, whereas the AM group did not. The AM group does, however, show an effect after they too had a night’s sleep, namely in Session 3.

As I discussed last month with the Stroop effect, word reading occurs seemingly automatically. Yet, it requires a sophisticated visual process to disambiguate one character from another smoothly, and efficiently. Whether we are distinguishing familiar words from each other, learning a new word, or learning a new set of characters entirely, integrating words into our lexicon to be processed seamlessly and effectively is a critical skill.
Reference for the article discussed in this post:

Wang, H., Savage, G., Gaskett, M.G., Paulin, T., Robidoux, S., & Castles, A. (2016). Bedding down new words: Sleep promotes the emergence of lexical competition in visual word recognition. Psychonomic Bulletin & Review. DOI: 10.3758/s13423-016-1182-7.

You may also like