Arguments for VP

[LFG List]

Referring to George Broadwell's message of 2/27:

> If you'll indulge me, here are four arguments in GB for a VP in Choctaw.
> I'd like to see which translate to valid arguments in LFG.
> 
> a.) subject -object asymmetries in binding.  These suggest asymmetric
> c-command in GB, but this is not a valid argument in LFG, since binding
> relationships are defined over f-structures.

--And also a-structures (see recent CSLI books by Manning and Alsina,
and Bresnan 1995 "Morphology Competes with Syntax"--reference below).

> b.) a variety of evidence that SOV is more basic than OSV or SVO.  Objects
> in these non-canonical positions show morphological and intonational
> evidence of being displaced.
> I am unsure about the validity of these arguments in LFG -- it seems that a
> flat structure for S and a structure with a VP are both compatible with the
> evidence. 

There are several types of LFG analyses of "displacements": 
(i) nonconfigurationality (Kroeger 1993 CSLI book, Austin and Bresnan
1996 NLLT article, Nordlinger 1997 Stanford Ph.D. dissertation
available by anonymous ftp from
ftp-csli.stanford.edu/linguistics/Papers/nordlinger-diss.ps), (ii)
scrambling (e.g. Choi 1996 dissertation, also Bresnan 1995 paper
"Morphology Competes with Syntax" available on my web site
http://www-csli.stanford.edu/users/bresnan/download.html), (iii)
various types of topicalization and dislocation (e.g. King 1995 CSLI
book, Bresnan and Mchombo 1987 Lg article, Bresnan 1994 Lg article,
and Bresnan 1995; (iv) Object Shift (see Sells 1998; sells at csli.stanford.edu).

> c.)  Some adverbials appear only before the V or before the OBJ but nowhere
> else.  It's possible, then, to say that they must be daughters of VP, but
> not daughters of S.  This argument seems valid to me.

Sounds good.  Alsina's dissertation and (I believe) his 1996 CSLI book
use a similar argument bearing on the c-structure of Romance
causatives.  In general word-order evidence is very strong evidence
for constituency in LFG, because c-structure constrains the ordering
and grouping of overt constituents.

> d.) Some proforms (approximately equal to "do so" in English) seem to refer
> to the V + OBJ as a constituent.  I am unsure whether LFG wants to assume
> that proforms refer to constituents.

This is not a theory-internal matter.  Both this hypothesis and its
opposite can and have been made in various frameworks, including LFG.