This is an interesting observation. However, there already is an established way to deal with such apparent conflicts of (possibly) meanings: the correct meaning is defined by the authoritative dictionary.
Tanru are coined on the fly. Lujvo mean what they mean. Cf. CLL 4.5, CLL 4.10.
So maybe the perception that there is a problem with lujvo stems from a slight confusion of their function in the language?
One solution I thought of might be to ditch the school of thought that all lujvo that have the same ve lujvo are equivalent and completely interchangeable
Thus {zangasnu} could be {zabna fa lo nu X1 gasnu X2},
While {zabnygau} could be {X1 gasnu lo nu X2 zabna X3 X4}
Kontraŭ-Fundamenta, thus bad. (I believe it to be one of the unnaturalistic traits of the language that word-forms as different as {xamgyzmadu}, {xaurmau}, and {xagzma} may never acquire different nuances of meaning. But that's how it is.)
{ta'ocu'i pe'i} that Lojban as-is relies way too much on the short rafsi anyway, which may inhibit learning it naturally. For example, {snura} "to be safe/secure", often gets combined with {gasnu} to make {nurgau} "to secure/protect". But that relies on the student remembering the rafsi of a not-quite-so-promiscuous gismu. It would probably be much easier to see the connection if the "standard form" remains {snurygau}.
You need to *know* the lujvo's meaning anyway (see above); the rafsi are just kind of memory hooks. And would it be wise to tweak the language around learners' difficulties in remembering parts of the language?