> is the unnatural nature of Lojban something that blocks it from being learned in that way?
I doubt it. For example, I am fairly fluent in the programming language C++, such that I create syntactically correct and sensible code on the fly, and I would say C++ is even more unnatural than Lojban.
I actually think it will be possible to learn Lojban even more correctly than natural languages, because software will be able to automatically and quickly point out mistakes, given the language is unambiguous. This computer-checking definitely speeds up
the learning of programming languages. There are Lojban parsers, but I am not sure if there is yet anything quite at the same calibre of a static code analyser for programming languages.