PPIs and Movement in Hindi-Urdu

Typically, Positive Polarity Items (PPIs), e.g. `would rather', cannot
be interpreted in the scope of a clausemate negation (barring rescuing
or shielding) (Baker 1970, van der Wouden 1997, Szabolcsi 2004 a.o.):

1a. John would rather leave.
1b. *John wouldn't rather leave.

The scope of most of them is uniquely determined by their surface
position. But PPI indefinites are special: they can surface under
negation and yet yield a grammatical sentence under a wide scope

2. John didn't understand something. ok: SOME > NEG;*NEG > SOME

Here we address the question of the mechanism through which a   PPI of
the `some' type  takes wide scope out of an anti-licensing
configuration. One possibility is (covert) movement, another is
mechanisms that allow indefinites to take (island-violating)
ultra-wide scope such as choice functions (Reinhart 1997). The
relevant configurations that have motivated choice functions for other
languages can be set up for Hindi-Urdu too.

We can therefore assume that a device that generates wide-scope for
indefinites without movement is available in Hindi-Urdu too. We show
that in Hindi-Urdu at least,  this device is unable to salvage PPIs in
the relevant configuration. Only good old fashioned overt movement
does the needful. If we think of overt movement in Hindi-Urdu as being
the analogue of covert movement elsewhere, then the Hindi-Urdu facts
are an argument that it is movement, albeit covert, that salvages PPIs
in English too, not alternative scope-shifting devices. We explore
whether the conclusion from Hindi-Urdu does in fact extend to English.