Adverb position and verb raising

[George Aaron Broadwell]

Dear Colleagues,

I'd like your opinions on the extent to which adverb placement is a valid
diagnostic for verb raising.  I believe the earliest form of this argument
may be due to Edmonds (1978) for French.  The basic idea is that adverbs
need to be adjoined to a XP.  Therefore, the contrast between English and
Hebrew  implies a difference of verb position

1.)  	Dani opened the door gently.
	*Dani opened gently the door.


2.)  	Hebrew  (Shlonsky 1997)

	Dani patax be-'adinut  	        'et ha-delet.
	Dani opened with-gentleness   acc the-door

	'Dani opened the door gently.'

Since 'gently' needs to be adjoined to XP, there is some phrasal boundary
between the V and its object.  In a movement-based framework, we would say
that the verb has moved out from V to some higher functional head (Agr,
Infl, etc.)  In recent LFG, we could say that the verb is realized in Infl.

My question is the extent to which data like (2) force us to such an
analysis.  In Bresnan (2001:102)'s principles of structure-function
mapping, the principle that deals with adverbs and other adjuncts says

"(d) Constituents adjoined to phrasal constituents are nonargument
functions AF' or not annotated.'

Compare

"(e) Complements of lexical categories are the nondiscourse functions CF."

[CF is the non-discourse argument-functions OBJ, OBJ-th, OBL-th, COMP]

My reading of these principles is that (so long as the language has an
endocentric phrasal organization) a Hebrew example like (2) cannot have a
flat VP like the following.

[V Adv NP]

That would follow because the Adv would be a complement of the lexical
category V.  Because it doesn't receive a complement function, this
structure violates the principles.  And because it should receive a
non-argument function ADJUNCT, it needs to be adjoined to a phrasal
constituent.

	Do readers of this list agree with this implication?  If it is correct, it
means that adverb position has a strong predictive power in telling us when
a verb is positioned outside VP.  In particular, according to this test,
the verb is outside VP in most Romance languages, since French, Spanish,
and Italian show the Hebrew pattern.  It also means that in general, we can
use adverbs to find phrase boundaries in less familiar languages.

	But I have seen little explicit LFG work on adverb positions, so I am not
sure of the extent to which people agree or disagree with this implication
of the structure-function principles.  I'd be interested to hear reactions.

Thanks,
Aaron Broadwell
==================================
George Aaron Broadwell
Dept. of Anthropology
University at Albany, SUNY
Albany, NY 12222
g.broadwell at albany.edu
http://www.albany.edu/anthro/fac/broadwell.htm