What isn't mentioned in the summary is from what subset the sample of just 7 brains was taken. This sort of research is usually drawn from postmortem dissection of LFA brains.
Not so long ago, a lot of excitement was created by discoveries concerning the autistic brain's neural minicolumns being more tightly packed. This sounds not only plausible, but explanatory, right?
Until you read it, and find out, that on the very Highest end, they are only 10% closer than the general population's minicolumn distribution.
The "normal" as well as, most of our autistic brains, grow fast in infancy, and then early on, "prune" away a lot of the infant connections.
It's hypothetically possible that at this time an original language system is removed (pruned), and the complex, permanent language system begins to take over, (or doesn't).
It's also possible, that the structure of sensory differences and memory differences, are habituated in the brain at this time.
In 2007 Sarah White and Uta Frith had a conference poster entitled "Big Heads, Small Details".
It was meant to be funny.