When an undergrad asks me how to tell if a scientific article is primary or not, I always ask them, "Can you understand the title?" If they answer "no," then I say, "Bingo!" It's simplistic, probably overly so, but my follow-up explanation goes into how the audience for primary articles is other specialists in the same field, so it uses highly specialized language that the non-PhD will probably not understand yet. This seems to work. In the LIS field, I rarely get that feeling, of seeing such highly specialized terminology that I'm not sure what the darn thing is about. This is one of the closest so far: "Using MPEG-21 DIDL to Represent Complex Digital Objects in the Los Alamos National Laboratory Digital Library" from the lastest D-Lib. Every so often, D-Lib publishes something that probably doesn't belong there, more likely in the ACM/IEEE Conference on DLs -- this is a case in point.