I just read one of the three journal articles that Dr. Weekes gave me to read before I work in his lab this semester. It looked at the progression of false memories using the DRM paradigm for semantic relations as well as phonological relations (For those who don't know, the DRM gives a list of words that are all related to a target word, but the target isn't in the list. However, very often subjects recall the target word as part of the list). They found a dissociation between the two, in that as children got older, semantic errors became more prevalent and phonological errors less. So as children develop, they stop focusing on the surface structure of the words and tune in to the meaning.
But the part that really fascinated me is what they interpreted this to mean. They claim that it shows the child's ability to learn and remember the "gist" of the list and refute the founding work of Rodeiger and McDermott (the R and M in DRM) who attributed it to semantic priming. I think I'll read it over again when I have time, but that struck me as a really powerful assessment of the results. I think I'm going to like this a lot!