For years, psychologists have argued that since the speed of vocabulary learning increases dramatically at a certain age (somewhere around 18 months), it must mean that there is a fundamental change in the learning process, a shift in strategy perhaps.
I haven't read the Science article, but on his web page and on the various media accounts, this change in learning is said to be accounted for by differences in the input rather than differences in the processing. McMurray shows that word difficulty/frequency can account for the change in learning speed. Basically, there are only a few very easy words and once you get past them, there are more and more words at that level of difficulty/frequency. This is sort of the upside of what I described here.
It's certainly an interesting result. But there's one assumption that remains unquestioned. Do children really have a learning spurt at this age? Paul Bloom thinks not. Citing various studies in his book How Children Learn the Meanings of Words, he produces the following table on p. 44 (you can search inside on Amazon.com):
12 months to 16 months: 0.3 words per day
16 months to 23 months: 0.8 words per day
23 months to 30 months: 1.6 words per day
30 months to 6 years: 3.6 words per day
6 years to 8 years: 6.6 words per day
8 years to 10 years: 12.1 words per day
In other words, children are gradually increasing their word-learning rate at least until the age of 10. It seems likely that the early changes are quite visible to us because we can keep track of which words they know and we readily notice new words. As the stock of words grows, it becomes much harder to do this.
That's not to say that McMurray's results are wrong. Not having read the paper, I can't really say. But it is another reason to believe that there is no particular change in the learning process that happens early on.