Good question. I *think* it's because the very high directional stability of the highly swept delta wing, which makes it effectively impossible to spin, allows the wing, once a turn has developed, to continuously weathercock into the sideslip component of the unbalanced turn, and then the fact of the tilted lift vector, causes the accelerating rate of descent. Probably.
Recovery is easy, reacting against the trike, just place the control bar parallel with the visual horizon (well, easy to understand, but can require quite a lot of physical effort). But without that, the aircraft just maintains an accelerating and tightening turn, which presumably would result in either structural failure or ground impact.
I have played some years ago with significant rolling inputs at the point of stall in a flexwing to see what would happen. The rolling departure feels for maybe a quarter of a turn rather like a 3-axis incipient spin, but rapidly speed builds and it has clearly become a spiral dive, which then requires the standard recovery.
G