ファブリス Wrote:I'm not sure where the "novel approach" is in "LV Approach". Seems to me an implementation of Heisig, where users can choose how to breakdown the primitives.Yeah I wasn't convinced by that stuff either in terms of 'personal styles' and LV being somehow 'novel' despite referencing Heisig right off the bat, but I do feel the paper lent support to the Heisig method by showing how learners, when given the opportunity to combine visualization in a radical-based approach, group the kanji into progressively larger chunks until they see the kanji as a whole. The variations could be seen more on the level of how people come up with their own stories i.e. mentally frame the primitives differently.
Now if I was to build an app to do that, I would question the benefits of users using different primitives seeing as there is so much common ground there; versus the benefits of having a common set of learning material. Heisig did the right thing.
I'm not convinced about the effects of "personal visual cognition" either. If someone was already exposed to kanji, then it follows logically that they might recognize a part where another learner won't. Then it can be expected that this person could break down the character in a different way, because they already "see" something in there. Even so, I doubt there is much advantage for learners to create their own variations in primitives. I would think a computer program could compute the most effective primitive sets purely on graphical analysis of the characters, using character/word frequency of use to determine what the best primitives would be.
I don't have anything against the "LV Approach" per se, just I feel that practically asking learners to choose how to breakdown each character is adding unnecessary complexity for the learner. The perceived benefit may instead be due simply to having exposed the learners to the character deconstruction in an interactive way.
PS: Also it is interesting that the paper presents the approach as superseding Heisig's " Component Approach" but it doesn't address the effects of using mnemonics, and especially the naming of those components or component groups (ie. "primitives").
I feel a customized (ie. per user) naming of primitives would be more interesting to study than a customized break down of characters, which can only deviate so far given the limited possibilities of breaking down the characters.
I seem to be noticing a pattern with these papers where they use a rhetoric that plays up the novelty of their own methods even when they're clearly similar to and even inspired by stuff that's already out there. Maybe they want to avoid branding or something? Except a lot of them seem to want to develop their own branded tools. ;p
But yeah, it would be interesting to see if a basic optimum 'story' structure could be found for each kanji based on the limited range of primitive customization.