Conclusions
The field of visual programming languages is starting to mature as evidenced by the flow of current research toward formalizing the theoretical foundations of VPLs and developing standardized classifications for VPLs.
It is not worthwhile to try to eschew text entirely. (e.g. Atomic operations, like addition, can be represented graphically in VIPR, but programs are easier to read and write when such operations are presented as text.)
VPL research provides an interesting mix of problems in computer graphics, programming languages, and human-computer interaction.
So far, the most successful visual systems have been those which target a well-defined application area. (e.g. Explorer, BALSA, ARK, LabVIEW)