12.1234, Disc: Moss Review/Verbal Complexes

[The LINGUIST Network]

LINGUIST List:  Vol-12-1234. Thu May 3 2001. ISSN: 1068-4875.

Subject: 12.1234, Disc: Moss Review/Verbal Complexes

Moderators: Anthony Aristar, Wayne State U.<aristar at linguistlist.org>
            Helen Dry, Eastern Michigan U. <hdry at linguistlist.org>
            Andrew Carnie, U. of Arizona <carnie at linguistlist.org>

Reviews (reviews at linguistlist.org):
	Simin Karimi, U. of Arizona
	Terence Langendoen, U. of Arizona

Editors (linguist at linguistlist.org):
	Karen Milligan, WSU 		Naomi Ogasawara, EMU
	Lydia Grebenyova, EMU		Jody Huellmantel, WSU
	James Yuells, WSU		Michael Appleby, EMU
	Marie Klopfenstein, WSU		Ljuba Veselinova, Stockholm U.
		Heather Taylor-Loring, EMU		

Software: John Remmers, E. Michigan U. <remmers at emunix.emich.edu>
          Gayathri Sriram, E. Michigan U. <gayatri at linguistlist.org>

Home Page:  http://linguistlist.org/

The LINGUIST List is funded by Eastern Michigan University, Wayne
State University, and donations from subscribers and publishers.



Editor for this issue: Karen Milligan <karen at linguistlist.org>

=================================Directory=================================

1)
Date:  Thu, 03 May 2001 14:53:25 -0500
From:  "Peter Hallman" <phallman at unt.edu>
Subject:  Re: 12.1162, Disc: Moss Review/Verbal Complexes

-------------------------------- Message 1 -------------------------------

Date:  Thu, 03 May 2001 14:53:25 -0500
From:  "Peter Hallman" <phallman at unt.edu>
Subject:  Re: 12.1162, Disc: Moss Review/Verbal Complexes

In his recent posting, 12.1117, concerning Michael Moss's review,
12.1083, of the Koopman and Szabolcsi book 'Verbal Complexes', Nilsen
Oystein expresses concern that Moss's review is tangential in certain
ways, in that it does not address the most important aspects of the
analysis presented there.  I wish to reiterate Oystein's position and
remark on certain specific details of Moss's original posting.

Koopman and Szabolcsi (2000) present an analysis of verbal complex
formation in Hungarian and Dutch.  They begin in chapter 3 by showing
that no head movement analysis of verbal complex formation can
accommodate all of the patterns of verbal complex formation in
Hungarian.  This is an important step because the previous literature on
verbal complex formation in both languages has primarily treated the
phenomenon as a type of head movement.  Koopman and Szabolcsi show that
only a phrasal movement approach to the phenomenon is tenable in the
first place, an important contribution to our understanding of verbal
complex formation.

Next, Koopman and Szabolcsi present such an analysis, and the analysis
they present fulfills traditional criteria of correctness:  (1) it
generates the patterns of verbal complex formation that exist in
Hungarian and Dutch (it is complete), and (2) it does not generate
patterns that do not exist (it is sound).  Furthermore, it does so using
not only the same syntactic deep structure for both Dutch and Hungarian
(unrelated languages), but also the same set of movement operations in
both Dutch and Hungarian.  The parameter along which Dutch and Hungarian
differ is how much 'extra' material can or must pied pipe in a given
step of movement, which the authors formulate in terms of filters on the
size of what can appear in certain syntactic positions.  Because the
parameter is how much material moves and not whether material moves, the
analysis generates the differences between Dutch and Hungarian verbal
complexes without making use of covert movement.  The analysis is
therefore also an important contribution to the current debate about the
role of covert movement in the derivation of syntactic structures,
lending credence to the view that there is no covert movement.

This analysis treats a vast array of data comprehensively and it does
so with a single syntactic basis and a single set of transformations
common to both languages, and a single variable in which they vary
parametrically:  the amount of structure than undergoes movement in a
given step.  Koopman and Szabolcsi's book lends solid empirical
grounding to the notions of Universal Grammar and parametric variation
as theoretical constructs, and sets a standard of theoretical
significance, empirical comprehensiveness and formal precision that I
think few linguists can hope to one day duplicate.

Yet, in his review of this work, Michael Moss does not address the
significance of the argument against head movement and the rejection of
covert movement in this work, and his only remark on the empirical
coverage of the analysis is:  "It is ... worth pointing out the value of
an analysis that can explain word order in Hungarian and Germanic
sentences.  Such a generalization that shows similarities in seemingly
very different languages is surely a step in the right direction."  The
remainder of Moss's review details the following three criticisms:

a.  The argument is not developed in a linear form;
b.  The concepts used are often partially or poorly defined;
c.  The diagrams are incomplete and as such are difficult to analyze.

a. and c. are purely aesthetic judgments that I will not address.  b.
is a significant criticism and the one that Moss explains in the most
detail.  He levels the following four specific criticisms:

1)  "Much of the technology such as remnant movement, licensing
positions and stacking positions is poorly defined.  Furthermore, many
times the reader is simply referred to other works which reportedly
discuss this technology in detail."

I have not previously encountered the notion that there is something
unusual about referring to other works that spell out the empirical
and/or theoretical justifications of 'borrowed' technology.  All work in
linguistics and other sciences is built on the work that preceded it,
and familiarity with the relevant literature on a given topic is usually
assumed.  Moss continues:

"This is the case with VMs or verbal modifiers.  VMs are introduced on
page 17 but are defined on page 20, and there the definition in only
vaguely referent:  'All of these expressions have been dubbed 'verbal
modifiers' (VM) in recent Hungarian descriptive literature.  For ease of
pre-theoretical reference, we adopt this cover term.'  This is
unfortunate.  I feel that a work should be as self-contained as
possible.  It is very difficult to really evaluate an analysis if the
definitions and concepts are not included in the book itself."

The reference to verbal modifiers on page 17 follows a discussion of
the behavior of verb prefixes, and states:  "Shortly, in section 3.4, we
introduce another set of expressions whose behavior is significantly
similar to that of the prefixes; the collective label to be used is VM."
 Section 3.4 begins on page 19 and gives examples of the behavior of
expressions which either optionally or obligatorily invert with the
immediately preceding verb, specifically:  separable prefixes,
directional or locative PPs containing a full DP, predicative APs and
NPs, determinerless arguments, and VM-less infinitival verbs.  On page
20, the authors state they will adopt the cover term 'verbal modifier'
for these elements.  This definition is pre-theoretical but formally
explicit.  There is no lack of precision in defining a class by listing
the elements in it.  Further, Moss's general criticism that the book is
not self-contained is incongruous with the fact that the book contains a
20 page appendix that offers the reader background information on the
relevant aspects of Hungarian syntax, including verbal modifiers (this
in addition to a 'General Reference' appendix which lists all the
assumptions on which the analysis is based including references to where
these assumptions are discussed in the main text--a first for a
linguistics book, as far as I know).

2)  "A new type of '+' phrase (+P) such as VP+, InfP+ are introduced as
elements which immediately dominate VP and InfP and attract elements
which would normally have landed in the [spec,VP] or [spec,InfP]
positions for spec-head agreement. . .  This really just seems to be a
way of producing Chomsky's (or Gazdar's) multiple-specifier
configurations without referring to Chomsky (1995) (or Gazdar (1982)).
As such, the +Ps create a position into which an element may move, but
justification of this type of phrase is far from obvious."

The notion of a '+' projection is introduced at the beginning of
chapter four, on the heels of the demonstration that head movement is an
inadequate approach to verbal complex formation.  Verb-verb inversion is
necessarily phrasal movement, and this is itself justification for
inferring a landing site for the inverted phrase, which the authors
label VP+.  There may be a sense in which VP+ is like an 'extra'
specifier of VP, however VP and VP+ are separate categories in that they
can be separated, i.e., VP can move out from under VP+, stranding VP+.
Hence they are labeled differently.  This is unlike double specifiers.

3)  "LPs or licensing positions are similar in this respect.  They are
introduced on page 39, where we find:  'We are led to assume that both
arguments and adjuncts have their own licensing positions (to be notated
as LP(xp)) and move into them as soon as possible.'  Apparently, LPs are
'motivated by Case and other feature checking positions' (pg. 43).  It
is not clear what these LPs do, or how they come into the derivation.
No clear definition is given, and no defense of their existence is
offered.  Yet they are a central part of the book's analysis."

The authors introduce licensing projections in connection with the
following datum:

Mutogatni fogja   akarni       a jatekot a gyerekeknek
show-inf  will-3sg want-inf the toy   the children-to
'(He/she) will want to show the toy to the children'

The infinitive 'mutogatni' ('show') has moved to clause initial
position, and this movement is unquestionably phrasal movement.  But the
two arguments of the head of the fronted phrase ('the toy' and 'to the
children') are stranded clause finally, which means that they must have
moved out of the fronted phrase before the fronted phrase moved.  This
leads the authors to assume that there are landing sites for syntactic
dependents of heads of phrases that undergo movement that these
dependents extract to, and hence they are not carried with the moved
phrase (all orders in which 'the toy' or 'the children' move with the
infinitive 'show' to the clause initial position are ungrammatical).
These positions are termed 'licensing positions' and overtly and
obligatorily attract elements with certain features--case features in
the example above, but also including, crucially, categorial features.
So LP(cp) attracts the closest CP and LP(vp+) attracts the closest VP+,
etc.  The argument for licensing positions is a distributional one.  In
the position where an element of a certain type occurs, a landing site
is postulated that attracts the relevant element to that position.  This
argument is a classical one, as Nilsen Oystein points out in his
previous posting concerning Moss's review, stating:  "LP (licensing
phrase) is a new name, but LP is nothing but a generalized version of
the familiar AgrPs; generalized in the sense that, according to Koopman
and Szabolcsi, other constituents than noun phrases need licensing and
move to the specifier of such projections."  Moss dismisses this remark,
saying "AgrP (for either subjects or objects) is restricted in its use
in early Minimalist Theory both by place and function.  LP does not seem
to be constrained in the same way, especially in its use as a 'stacking
position'".  In doing so, Moss misses Oystein's point that the function
of agreement phrases is more important than ever in contemporary syntax.
 And of course it is not constrained in the same way, because it is
generalized to features other than Case.  Structures are built by adding
elements to a tree which 'attract' other elements.  The surface word
order of the attracted phrases is determined by the position of the
attractors in the underlying syntactic hierarchy.  A projection like
LP(vp+) attracts the closest constituent of type VP+, just like AgrSP,
(or TP) attracts the closest nominative marked element.  This is the
motivation for movement in the Minimalist Program.  There is nothing new
about Koopman and Szabolcsi's analysis in this respect, and there is
nothing about the inferences they make about what licensing positions
exist and where they occur that is poorly defended.

4)  "Next, we have 'stacking positions' which are also labeled (LP)...
Apparently stacking positions occur in the derivation to preserve word
order:  'movement into them is constrained by the convention that it
must replicate the already existing linear order of the pertinent XPs'
(pg. 44).  I thought that the derivation was supposed to explain linear
order phenomena.  How is it that the linear order is now supposed to
determine structural projections?  While the authors themselves are
hesitant about stacking positions saying that they do not seem to be in
the spirit of the 'minimal [-ist (sic)] analysis', they offer the
following defense for using them in the derivation:  'since it is
possible to employ them in a completely mindless, mechanical fashion, we
choose to live with them as a provisional solution that we hope will
give way to a more insightful one' (pg. 44).  This does not seem to be
an adequate defense for an element which is central to a derivation."

Stacking positions are introduced in order to make the analysis the
authors present compatible with cyclicity (aka 'Extension').  As the
authors point out, stacking positions are not necessary if cyclicity
violations are tolerated, as they are often said to be (cf. the
Minimalist notion 'Procrastinate'), and for this reason stacking
positions are not in fact central to the analysis.  Further, the
function of stacking positions is not to preserve word order.  Movement
to stacking positions is constrained by the convention that when several
things in a row get 'stacked' (move to stacking positions), the order
they get stacked in must follow the order they appear in before they are
stacked.  But this is the same as saying that movement to stacking
positions strictly observes superiority.  The 'superior' element prior
to movement remains the superior element after movement.  Stacking
positions are not there to explain word order.  Their function is to
deal with certain cases where a specifier seems to be extracted to a
position higher than the position to which the category it is a
specifier of moves (an Extension violation).  The motivation for them is
theory internal:  they are postulated because an observed transformation
appears to be an Extension violation, and the authors note the problem
here, choosing to 'live with it' in the form of stacking positions and
the associated restrictions on their use.  It is not uncommon nor
considered bad practice for linguists to introduce an ad hoc device as
long its role is clear and it is secondary to an otherwise principled
analysis (as it is here, contra Moss's assertion).  Certainly, no work
in linguists can portend to present 'the whole picture', and there are
always points on which an analysis can be expanded upon, clarified and
improved.

But Moss is not willing to evaluate an analysis which could be improved
upon in one (rather insignificant) respect.  His criticisms do not
address the main claims of the book, and in his response to Oystein's
remarks regarding his review, we find out why:

"Nilsen writes that I did not address the issue of the head versus
phrasal movement analysis presented in the book and that the review did
not mention the question brought up in chapter 6 concerning the issue of
whether or not infinitival verbs represent full clauses.  These issues
were not mentioned due to the fact that the mechanisms on which they are
founded ('+' phrases and LP/Stacking Positions) are not justified or
defended.  As such, it is difficult to estimate the strength of the new
approach, no matter how innovative or interesting it may be."

But it is not the task of a reviewer to read a book until he or she
finds a premise with which he or she disagrees, and then abandon the
task of reviewing the book because the mechanisms on which it is based
are in his or her mind unjustified.  There are few linguists who have
had the pleasure of reading a work in their field with whose every
premise he or she wholeheartedly agrees (the pleasure has not been
mine), but it is nonetheless possible to consider, given the premises
the authors adopt, whether the argument is valid or not.  This means, in
an empirical discipline like linguistics, whether the analysis generates
the phenomenology it is supposed to, and does not overgenerate.  Koopman
and Szabolcsi's analysis accomplishes this goal, and makes important
contributions to a number of current theoretical debates in addition.
The arguments the authors bring to bear are classical methods in
syntactic argumentation, and the technology they introduce is formally
defined and articulately defended throughout.

Sincerely,

Peter Hallman

---------------------------------------------------------------------------
LINGUIST List: Vol-12-1234