Tuesday, December 21, 2010

Fetal Testosterone and Autistic Traits - Part IV: Verbal Abilities

ResearchBlogging.orgApart from eye contact and looking behavior, one of the earliest signs that a child might be autistic is a delay in the acquisition of speech*. Many of us continue to have difficulties with language as older children, adolescents and adults --- difficulties both with producing speech, and with understanding other people's speech.

Thus, if autism really results from hypermasculinization of the developing brain in utero by testosterone, you would expect children exposed to higher levels of testosterone during gestation to 1) develop speech somewhat later than children exposed to less testosterone, 2) have somewhat poorer verbal skills in general, perhaps reflected by an imbalance between verbal and performance IQ, and 3) do relatively poorly at comprehending spoken language, compared to their less-androgenic peers.

To help answer those questions, Svetlana Lutchmaya, Simon Baron-Cohen and Peter Raggatt (full text here) measured testosterone levels in the amniotic fluid of 87 women and then tested their children twice, at the ages of eighteen and twenty-four months, on the size of their vocabularies using the Communicative Development Inventory (CDI).

They found a significant sex difference in vocabulary size --- on average, the 40 girls in this study knew about 87 words at eighteen months old, while the 47 boys had an average vocabulary size of around 42 words. There was a huge amount of variation within each group, though: the boys' scores ranged from zero (which I'm assuming to mean they didn't speak at all --- three boys had this result) to 222 words; while the girls' scores ranged from 2 to 318. So, the least verbal girl(s) could all say at least a few words, while three boys couldn't say any words at all; while the most verbal girl(s) knew almost a hundred (or ~50%) more words than the most verbal boy(s).

This pattern persisted when the children (most of them, at least --- at this stage, the sample consisted of 43 boys and 38 girls) were followed up at twenty-four months old: then, the boys' average vocabulary consisted of 197 words (ranging from a low score of zero to a high score of 414); the girls knew about 275 words on average (ranging from 15 to 415). So, while the most talkative boys had caught up with the most talkative girls by this time, the sexes' average and low-end scores still diverged.

In their statistical analysis of these data, and on the seven other variables they tracked, the researchers found the strongest relationship between sex and vocabulary size, followed by level of parental education. Fetal testosterone came in third for both sexes together, and was not a significant predictor of vocabulary size at all within either sex. So I think it's probably fair to conclude that the reason testosterone showed up as a significant factor at all in the analysis of the pooled data is because it's strongly correlated to sex.

A much older study (Finegan, Niccols, and Sitarenios, 1992) also looked for a potential relationship between prenatal testosterone (measured in amniotic fluid) and young children's performance on a broad assortment of cognitive tests. Some of these tests measured verbal abilities; the Comprehension scale of the Reynell Developmental Language Scales tested a child's understanding of individual words, concepts, and different meanings conveyed by different sentence structures; and several subdomains within the McCarthy Scales of Children's Abilities --- Oral Vocabulary, Verbal Fluency and Opposite Analogies --- tested the children's expressive language abilities. Of those latter three, the first is a rough measure of vocabulary size (though, unlike the measure used by Lutchmaya and her colleagues, it doesn't ask the parents how many words their child knows; instead, it asks children to define a series of 10 words, increasing in difficulty from familiar objects and animals to more abstract concepts, like "loyal"), while the second two measure how quickly a child can come up with words of a specific type.

Of all those measures, only one --- Language Comprehension --- was found to correlate with prenatal testosterone, and that relationship was only seen in girls.

(Figure 1, in Finegan et al., 1992 --- graphs showing the relation of prenatal testosterone levels, plotted on the x-axes, with scores on the Reynell Language Comprehension subscale. The graph on the left, with the black dots, shows the results for girls; the graph on the right shows boys' results. You can see that there's only a correlation of any kind for the girls).

In girls, there was a negative quadratic relationship between prenatal testosterone and scores on the Language Comprehension test, meaning that girls with middling levels of prenatal testosterone exposure --- not the highest or the lowest --- tended to get the highest scores on this task. No relationship whatsoever was found between prenatal testosterone and either comprehension or expression of language for boys.

That study tracked other variables as well as prenatal testosterone, and while none of those variables (including testosterone) predicted how well the children did on any of the expressive-language tasks, family background (a measure including parental socioeconomic status, and the mother's IQ and education) was also a significant predictor of language comprehension. This is somewhat in line with Lutchmaya et al.'s results, which found parental education to be an important predictor of a child's vocabulary size, but also a little different, since Lutchmaya and colleagues were measuring how much the children in their sample spoke, not how much they understood --- expressive language, not receptive language, which Finegan and colleagues did not find any predictors for. Finegan et al.'s children were older than Lutchmaya et al.'s, though, and the measures the two groups of researchers used were very different. The one thing they seem clearly to agree on is that there probably isn't any correlation between prenatal testosterone exposure and language abilities as manifested in early childhood, or if there is it's a very weak one.

*This impairment in verbal communication is actually one of the diagnostic criteria for various ASDs --- with (classical autism) or without (Asperger syndrome) delayed acquisition of speech.

Lutchmaya, S., Baron-Cohen, S., & Raggatt, P. (2001). Foetal testosterone and vocabulary size in 18- and 24-month-old infants Infant Behavior and Development, 24 (4), 418-424 DOI: 10.1016/S0163-6383(02)00087-5

Lutchmaya, S., Baron-Cohen, S., & Raggatt, P. (2001). Foetal testosterone and vocabulary size in 18- and 24-month-old infants Infant Behavior and Development, 24 (4), 418-424 DOI: 10.1016/S0163-6383(02)00087-5

No comments: