New article: Eye-tracking research on toddlers with suspected developmental language disorder
In the Netherlands, five to seven percent of children have a developmental language disorder (DLD), which makes it difficult for them to understand and produce language. Within the group of children with DLD, there is also a lot of variation in the severity and nature of language problems. Can testing provide an indication of how toddlers with suspected DLD will develop later on in life?
In the case of a developmental language disorder (DLD) there is something wrong with the innate ability to learn language because the brain does not process language properly. Children with DLD have serious difficulty understanding words and verbs. The severity and nature of these language difficulties can vary considerably. It is very important to gain a better understanding of which children are likely to catch up, and which children are less likely to do so.
Measuring with eye tracking
To this end, linguists from Radboud University examined a group of young children (2 to 4 years old) with suspected DLD together with Kentalis*. ‘DLD is usually examined with standardised language tests’, explains Susanne Brouwer, one of the researchers. ‘For example, a child is shown four pictures, hears a word and has to indicate which of the four pictures matches the spoken word. This allows you to discover whether a child knows a word or not. But you don't see how the process of recognition works.’ To get a more precise picture of what happens in a child's mind when it performs such a task, Brouwer and colleagues investigated the processing of individual words and sentences using eye tracking. This measures eye movements and calculates to the millisecond how long it takes a child to recognise words and sentences. Brouwer: ‘This way we can see whether children with suspected DLD are just as quick as the normally developing children, and whether there are differences among children with suspected DLD.’
The study examined 51 toddlers with suspected developmental language disorder and 31 typically developing (TD) control toddlers in two experiments. In the first experiment, the children heard short sentences, such as ‘Look, a hat’. On the screen they saw a hat and a shoe. Eye tracking measured how quickly children looked at the correct picture. ‘At first, we saw little difference between the DLD children and the TD children’, says Brouwer. ‘But when we zoomed in, it turned out that the children with the developmental language disorder needed a little more time to look at the correct picture. We did not expect that.’
In the second eye movement experiment, children heard more complex sentences, such as ‘Hey, he's just reading a book’. Brouwer: ‘Based on the verb read, you can expect something readable. When a child sees a picture of a book and a shoe on the screen, you expect that with the verb reads they will already look at the book and not at the shoe, because you cannot read a shoe. This is how it works with TD children. We expected that DLD children would have more problems with this because they are less able to process verb structures.’
That expectation came true: toddlers with suspected DLD were less likely to look at the picture of the book when hearing the verb read and were slower compared to the TD children. In addition, the prediction behaviour within the group of DLD children appeared to vary considerably: some children performed the same as the TD children and others did not. This indicates that not all children with a suspicion of DLD have difficulty predicting nouns based on a verb.
Brouwer and colleagues also examined how the results of the standardised` language tests and the eye movement tasks are related. It showed that if a child is good at producing words in a standardised language test, they can also recognise words more quickly in the first eye movement experiment. And a child in the DLD group with low scores on the standard language tests (producing and recognising) had more difficulty predicting a word based on a verb compared to a child in the DLD group who scored higher on the language tests.
The results of the two eye movement experiments were also found to correlate. The children who are slower at recognising a word in the first experiment are slower at predicting a word in the second experiment. Brouwer: ‘That suggests that these children's problems may lie with the speed of processing.’ Treatment plans could be adjusted accordingly. Because in the eye movement tasks it may be about milliseconds per sentence, but that can cause communication to go wrong in the long term. Brouwer: ‘It is not so bad if a child takes just a little longer to process one sentence. But if there is another sentence and another, the delay will increase and children with DLD may miss parts of the conversation.’
Can these results predict which children will catch up as they get older and which children will not? No, is Brouwer's sobering answer. ‘For that we have to keep an eye on them in the coming years, to see how they develop. What we do show is that standardised language tests do not provide the full picture. Most children with a suspicion of DLD are able to predict successfully, while a small group is (still) lagging behind. It could be that the children who are able to predict successfully now will have fewer language problems when they are a bit older than the children who less successful at predicting.’
The article by Petra van Alphen, Susanne Brouwer, Nina Davids, Emma Dijkstra and Paula Fikkert - Word recognition and word prediction in preschoolers with (a suspicion of) a developmental language disorder: evidence form eye tracking - has been online in the Journal since 21 May of Speech, Language and Hearing Research.
* Koninklijke Kentalis is a Dutch organization that provides research, care and education to people who have a developmental language disorder (DLD) or are hard of hearing, deaf or deafblind. The children enrolled in this study were recruited from special toddler treatment groups.