> >  'Raising' and 'Equi' date from the 60s. 'Control' dates
> > from (I hazard) the late 70s.
> 'control' does appear to be related to GB theory, but it's got to a better
> term than 'equi'!

I prefer 'equi', because it has become purely descriptive, whereas
'control' is still used by many as a theoretical notion. Likewise
I prefer the more descriptive 'object raising' to 'ECM' (exceptional
case marking), because the latter derives from a widely popular but
egregiously silly theoretical analysis.

> Would the other alternative 'catenative' be more neutral?

But like most terms from traditionalist grammar, it is less
well-defined. Furthermore, it is loosely equivalent only to
subject equi and not object equi.

> > but if you hold, as most syntactic
> > theory holds, that it is syntax that builds up the semantic
> > structure of the sentence (particularly with regard to matching
> > predicates to their arguments, i.e. the syntagmatic dimension
> > of semantics), then the distinction is perforce syntactic.
> But there's the rub. The evidence for syntactic distinction is, as
> you say, "scanty at best" in English and seems to be almost non-
> existent in Classical Latin. Do natlangs in fact provide clear
> evidence from their syntagmatic relations for this distinction?

A valid & interesting question, to which I don't know the answer.