I think we sometimes ASSUME that nonverbal tasks are nonverbal in the same way. And you know what happens when we assume right?? This is true for IQ tests that test nonverbal abilities. We have to ask what kinds of abilities? How are these tested? How are they elicited? And, how are they observed?
There are different kinds of nonverbal tasks. Sometimes the instructions are given verbally but the response is pointing, manipulating, constructing, or gesturing. Sometimes both instructions and responses are nonverbal. Some IQ tests are fully nonverbal, others have nonverbal subtests. In a paper published a couple of years ago, we were interested in how bilingual children with and without developmental language disorder (DLD) performed on nonverbal tests.
Why does it matter how bilinguals perform on nonverbal measures?
There are a couple of reasons. First, nonverbal measures are said to be non-biased for children who are bilingual or who speak different languages. Second, children with DLD (or SLI) generally score within the normal range on nonverbal IQ tests. The idea is that their DLD is not because of a cognitive impairment.
But, is it possible that nonverbal IQ tests are STILL culturally biased even if they are nonverbal? Maybe. The tasks could be unfamiliar, or gestures could be used differently in different cultures. For children with DLD, it could be that some tasks still involve verbal skills. This is what we set out to test in our study.
We tested 169 (including 23 with DLD) Spanish-English bilingual kindergarten age children on the Universal Nonverbal Intelligence Test (UNIT). The two groups of children were comparable in age, exposure to the two languages and SES. On average, they started learning English when they were 2, they heard and used Spanish and English about 50% of the time, with a range from 20-80%.
We used two subtests appropriate for kindergarten age children. One was a cube design task where children look at a two-color geometric design and reconstruct it using two-color cubes. The other was symbolic memory where children see a sequence of symbols (varying by color and size) and then have to reproduce it from memory using a set of cards.
Our results indicated that children with typical development and DLD performed similarly on the cube design task. Further, they performed comparably to the normative sample which is based on a US population of monolingual English speakers. But, children with DLD performed lower than the typically developing children on the symbolic memory task (and lower than average compared to the normative sample).
We believe that this pattern of results indicates that these two tasks are not biased for bilinguals. And we also think that symbolic memory is not as nonverbal as we might hope. In order to complete this task, especially as it gets harder, children may be using their verbal skills to support their performance. Children with DLD by definition, have difficulty with language learning and they had a harder time with this task even though it was nonverbal. Anecdotally, we observed that some children would talk themselves through the task. They used language to hold the sequence of pictures in memory to support their completion of the task.
We need to question whether or not nonverbal IQ really is unbiased. In our study, we found that cube design and symbolic memory did not seem to be biased for bilingual children. But, symbolic memory required more language skills and was thus harder for children with DLD. When we use these measures, we need to understand their limits and whether or not they are truly nonverbal and if it matters. The symbolic memory task may underestimate children’s abilities if they a language related disability. This could lead to denied services, lack of access to general education, and lowered expectations. The educational and special education decisions we make are not trivial. They matter to the children and families we serve.