tk You're right, Forth has not stabalized. It does change, it is the nature of the language and the philosophy of the language to change and adapt.
I wouldn't say that Chuck Moore is reinventing the language. The basics always stay, definitions and stacks. The operators have evolved from his chip design experience over the last 20 years. The use of color is a new innovation, a simplification of previous notation. Blocks have been around Forth for quite while. There are a lot of ideas and innovations in the implementation, but it could well be implemented much differently without change from the user's perpective.
We clearly differ on the philosophy of computer languages. To me, having so many computer languages kept alive just means a much greater maintenece cost. If there are better methods people should be using them. A legacy language, is just one which new software should not be written in.
C is known primarily as a systems language. Compared to Forth it is a very poor systems language. Forth can interface hardware much more directly, more compactly, and allows clean integration of machine language code. It also features interactive development, which is ideal for debuging hardware device code.
C as an application language is beaten by many languages including Forth. C provides you with one set of abstractions with a clumsy syntax. As a result, higher level operations are very clumsy compared to languages which support these operations directly or extensible languages which allow you to add these operations in an integrated way.
C as an Algol family language seems poor. I'm not up on Algol languages, but I would be very suprised if C has not been surpassed. The fact that it takes two languages for C for what should easily be accomplished in one, makes C an easy target for improvement.
There are some cases where writing new C code makes sense. The most usable free software systems and many commercial systems are C based. When writing code for these systems, C is almost always available, and many libraries are written in C.
New systems should avoid the use of C, as clearly better methods exist.
Of course this is only my perpective. I have no control over what people use. Language choice is mostly guided by popularity anyway, not techinical merit.