A few thoughts:
1. I agree that it would be interesting to see if his other WISC verbal subtests were more similar to the higher score or the lower score, but I think you already have information in that area, in the form of the CELF-5 results, which are generally consistent with the VCI, with a core language index identical to the VCI, and individual test scores across the same range as the VCI subtests (from Average up to the beginning of the Extremely High range--though on the CELF, these typically would be described only as Average and Above Average). Your observation on the relationship between Similarities and Word Classes is generally accurate, though the big difference is that there is no expressive language requirement for WC. Students only have to identify the words that go together, not describe how two provided words are related. The two tasks he did particularly well on on the CELF both derive significant benefit from his exceptional working memory.
2. I think Coding may have benefited from his memory as well. Typically, students of this age look up and down between the key and the response rows. Memorizing them quickly saved him a lot of time, and reduced any possible effects from getting "lost" while scanning up and down. I wonder a bit if that (scanning) was a factor in his lower SS score.
3. On another note, SaturnFan, I imagine that you may have considered that you yourself might be 2e? And -because- of that, much more capable of teaching the child that you know better than anyone else does than you might think.
