<< >> TOC
Under current minimalist assumptions (cp.
Chomsky 1995b), base generation by operation Merge is costfree, whereas overt
applications of Move-a
are not. Structure building processes are assumed to be subject to economy
requirements, such that overt movements will have to be specifically motivated.
Whether scrambling is characterized as A- or A-bar movement, the question of
motivation thus becomes immediately relevant.
The common strategy of assuming case checking
as motivating factor seems to be ruled out: if case needs to be checked off at
all, the necessary mechanism is obviously available in unscrambled structures as
well (cp. Ueyama 1994 arguing against scrambling by case checking in Japanese).
A feature like [+ FOCUS] may be another likely candidate for triggering
movement, but it is quite obvious that in many cases intonation can override any
positional focussing. The question would then be: why undergo the cost of
positional focussing if the effect is subsequently voided by intonation?
Since there are also instances of scrambling
that seem to be truly optional, Fukui (1993) tackles the question from the
opposite direction by proposing a Parameter Value Preservation
(PVP) principle which states that Move-a
involves a cost only if it disturbs the base generated ordering relation of head
(i.e. verb) and complement (i.e. object). Since German is head-final (like
Japanese), scrambling is licensed, because no argument would need to cross the
(original) verb position to reach the landing site for scrambled elements
(presumably a VP-adjoined position). This proposal may well be true, but it does
not provide an answer to the puzzles that lie waiting in the details. For
example, why does English allow ‘scrambling’ as in (14c), but not in (14b)?
I gave JohnDAT the bookACC.
*I gave the book John.
I gave the book to John.
Thus, the employment of Move-a
as a mechanism for scrambling leads into a dilemma: if the application of Move-a
is cost-free, why is scrambling not always possible? If Move-a
does involve a cost, why does scrambling only seem to be optional and not
necessary in many instances?
Obviously, the fact that English does not have
an extensive overt case system plays a role in (14). German (and, for that
matter, Japanese), on the other hand, has overt markings available, which may
allow deviations from the default argument order without leading to a crash at
LF (because of unsaturated features?). If it is acknowledged, however, that
overt markings are necessary for licensing deviant word orders (see also 5.1),
the PVP may be superfluous. The easiest way out of these problems is to assume
that scrambling is not subject to Move-a.
<< >> TOC
© Philipp Strazny 1997
 I take Move-a as referring to any chain forming process in syntax without making any implications about the specific details (i.e. it does not concern me here whether Move-a should really be Attract-a or Attract-F etc.).
 It is not clear whether (14c) is to be analysed as a scrambled variant of (14a), but if it is, the analysis developed here will be able to account for it: (14c) is licensed because of overt case marking, whereas (14b) solely relies on positional evidence linking the overt arguments to A-structure.