Wurmbrand then proceeds to argue that the account for this three-way distinction--which is that RIs are VPs, RNRIs vPs/TPs, NRIs CPs--allows us (i) to derive the distinctive syntactic properties, and (ii) to explain the syntax-semantics correlation showing up in (25).
328), Wurmbrand applies two test criteria for CP status of the infinitival complement: first relative clause pied piping (which, predictably, does not co-occur with the operations indicative for RIs and RNRIs), then extraposition vs.
Wurmbrand concedes herself that (c) does not correlate 1:1 with LRIs, since it also occurs with RNRIs and NRIs; hence, it is not a defining feature.
Turning now to the relationship between Wurmbrand's RIs and RNRIs, Wurmbrand herself mentions just two properties they share vis-a-vis NRIs, namely: (31) a.
With (31b), due to Wurmbrand's fuzzy conception of scrambling phenomena (see Section 4.4 below), things are not so clear, but in any case, there are many more features uniting RIs and RNRIs against NRIs than Wurmbrand mentions: RIs as well as RNRIs, but not NRIs, allow for (31) c.
Of course, this by itself doesn't exclude the possibility that ALL the phenomena listed in (31) require a vP/TP structure rather than a CP- or VP-structure of the infinitival complement, hence that RNRIs turned RIs vs.