Kids as young as two can judge the reliability of their own word knowledge

Ask a child under four whether a word is familiar to them, or whether they know the name of an object, and typically they will find it hard to tell you. Researchers have taken this to suggest that young children aren't too sure about the reliability of their knowledge of words and their meanings. In other words, they lack 'metacognitive awareness' in this area.

As the authors of a new paper in Psychological Science point out, metacognitive awareness of the reliability of your memories and knowledge is crucial for learning. If, for example, you fail to realise that your understanding of the meaning of certain words or a topic is shaky, you're unlikely to seek out the correct answers. Now Isabelle Dautriche at CNRS in France and colleagues provide evidence that by the age of two, children do in fact have metacognitive awareness of their knowledge of word meanings (even if they can't explicitly state this).

For the first of two experiments, the team recruited 50 children aged between 18 and 29 months, with a mean age of 23 months. The study involved five pairs of pictures of objects whose names were familiar to the children (such as a banana and a dog), and also five pairs of unfamiliar objects, such as a 3D virus shape and a double helix, for which the team came up with novel names, such as 'blicket' and 'toma'.

Each time, the child was shown a pair of pictures and prompted to look at one of them. For instance, if shown a banana and dog, they might be asked 'Where is the dog?"; or, for a pair of unfamiliar pictures, "Where is the blicket?". Virtual curtains were then drawn over both pictures. The child was then asked where the picture they had previously been prompted to look at had been (e.g. 'Where was the dog?') After a few seconds, the correct picture was revealed, accompanied by a cheering sound and a star-burst animation.

The researchers used eye-tracking to monitor the child's gaze throughout the trials. They found that in the initial phase, when both pictures were visible, the children were more likely to look at the correct picture in the known-words trials, indicating that they recognised the known words. As expected, in the novel word trials, they didn't show a preference for either of the pictures. Once the virtual curtains were closed, and the children were asked 'Where was the…' they were more likely to glance first at where the correct picture had been, but again only for trials where they knew the words.

The children were pretty good, then, at identifying known words. But they weren't always right on every trial. To explore their metacognitive awareness of their understanding of each word, the team looked at what happened right after that first glance. They found that, for familiar words, when a child's first glance had been correct, they persisted in looking in this direction for longer. Their further analysis of the data suggested that their 'post decision preference' reflected their confidence in having made the right choice.

A second study then explored whether (as might be expected), this confidence is shaped by their experiences of listening to adults. A fresh group of 60 young children watched videos of an adult either correctly or incorrectly naming five familiar objects (such as a ball) and then teaching the child the names of two novel objects (identifying an odd-looking fluffy toy as a 'danu', for example). Using a similar set-up to the first experiment, the children were then tested on pairs of other known words and the two newly learned words.

The team found that whether the speaker had shown themselves to be reliable or unreliable (in that they'd earlier got the names of familiar objects wrong), the children were more likely to look at the correct picture for both the familiar words and the novel words that they had just been taught. However, when the curtains were closed and the children were asked where the object had been, if the speaker had been unreliable the children shifted their gaze between the two possible locations more often  (this was only the case for trials involving familiar words). The team interprets this as meaning that the speaker's earlier unreliability had led them to be less confident about their decision.

"These findings provide novel evidence that the capacity to internally monitor one's own decisions beliefs and memories…exists before children are old enough to talk about what they know," they write.

So — while a toddler might not be able to tell you whether they know the name of an object, they probably do have an implicit awareness of how sure or unsure they are about this. Still, the work does also suggest that a caregiver who makes mistakes will lead children to lose confidence in their knowledge, even when it's right.