Moro, Andrea (2000) Dynamic Antisymmetry. MIT Press, paperback ISBN:
0-262-63201-2, xii+142pp, $20.00, Linguistic Inquiry Monograph 38
Publisher's announcement at
http://linguistlist.org/issues/12/12-785.html#1
Previous review at http://linguistlist.org/issues/12/12-1030.html#1
Reviewed by Kerstin Hoge, University of Oxford
Introduction
In this book Moro formulates a theory of syntactic movement
which links movement to the geometry of phrase structure and
dissociates it from the morphological properties of lexical
items. The central idea, from which the theory derives its
name, is that "movement is driven by the search for
antisymmetry" (p. 28, (23)). Moro thus argues that movement is
a by-product of linearisation, i.e. it applies to rescue those
phrase markers which are not compatible with Kayne's (1994)
Linear Correspondence Axiom (LCA).
Although Dynamic Asymmetry advances a radical alternative to
the feature-driven syntax of the Minimalist Programme, it
remains Minimalist in at least two respects: (i) contra Kayne
(1994:49), it takes the LCA to be a principle of the
phonological component rather than a principle which holds at
all levels of syntactic representation (p. 2, cf. Chomsky
1995:340); and (ii) it views the syntactic operation Move as a
reflex of the legibility conditions at the interface, even
though it differs from standard Minimalist assumptions
(according to which movement facilitates the formation of
legitimate L(ogical)F(orm)-objects) in that it takes the
relevant interface level to be Phonetic Form (PF), not LF
(p. 14, cf. Chomsky 1995:277).
Synopsis
The monograph is divided into four chapters and an appendix,
which, together with the book's central thesis, are briefly
outlined in the Introduction.
Chapter 1 ("Theories of Movement") considers the Minimalist
background for Dynamic Asymmetry. Moro discusses Chomsky's
approach to the question how and why syntactic movement is
realised, in particular his suggestion that apparent
'imperfections' of natural language (dislocation,
uninterpretable features) constitute an optimal solution to
bare output conditions (Chomsky 1995:317). On Minimalist
assumptions, movement takes place to ensure interpretability at
LF, given that uninterpretable features (i) must be eliminated
for LF convergence and (ii) can be eliminated only in a proper
local relation with a matching feature. Dynamic Asymmetry
takes up the idea that "movement is a way grammar avoids
presenting the interfaces with uninterpretable [structures]"
but proposes that the relevant interface level is PF, where the
set of terminals must be associated with a linear ordering
through the LCA.
Chapter 2 ("Movement as a Symmetry-Breaking Phenomenon")
presents the core aspects of Kayne's (1994) theory of
Antisymmetry and thereby prepares the ground for Moro's
discussion of the empirical content of Dynamic Asymmetry.
Crucially, the chapter introduces the notion of 'point of
symmetry' (p. 22). A point of symmetry in a syntactic tree is
defined by two properties:
"first, it involves two elements belonging to
the same category (i.e., both heads or both nonheads)
that c-command each other; second, the categories are
both overt" (p. 100).
A tree structure which contains a point of symmetry is not LCA
compatible. Since the hierarchical structure is not ordered
conclusively by means of asymmetric c-command, the terminal
nodes dominated by the two symmetric nodes cannot be associated
with a linear ordering. Given that words must be arranged into
sequences, structures involving a point of symmetry will fail
to converge at PF. The question is now whether such structures
can be generated in the course of the derivation. Answering
this question in the affirmative, Moro (departing from Kayne's
original conception) takes the LCA to hold only at PF and
proposes that points of symmetry constitute triggers for
movement. Movement 'neutralises' a point of symmetry by
turning one of its elements into a trace, which will be
invisible to the PF component and hence exempt from the LCA.
Chapter 3 ("Sources of Symmetry") constitutes the core of the
book, in which Moro identifies three basic types of points of
symmetry and their empirical correlates:
(i) structures that pair two maximal projections
(associated with the syntax of small clauses);
(ii) structures that involve multiple adjunction to
a maximal projection
(associated with the syntax of wh-questions);
(iii) structures that pair two heads
(associated with the syntax of clitics).
For the structures in (i)-(iii) to function as a trigger for
movement in their respective empirical domains, they must all
be syntactic objects which can be constructed by Merge. To
this end, Moro proposes that Merge allows generation of both
labelled and unlabelled constituents. More explicitly, Merge
is viewed as an operation on two syntactic objects, a and b,
which creates a new syntactic object K of the form {a,{a,b}}
(substitution, instantiated by (iii)), {<a,a>,{a,b}}
(adjunction, instantiated by (ii)), or {<0>,{a,b}} (predicative
linking, instantiated by (i)). Unlabelled constituents are
claimed to be "fully compatible with Merge's essential property
of not adding extra information with respect to a and b and
with the idea that there are no mixed labels g" (p. 33).
Turning to specific instances of movement which can be related
to the three distinct types of points of symmetry, Moro first
examines canonical and inverse copular constructions. Building
on his theory of copular sentences, articulated in earlier work
(Moro 1988, 1997) and summarised in the appendix, Moro argues
that copular constructions involve small clause (SC)
complementation and are derived by raising either the subject
or the predicative noun phrase to preverbal position, cf. (1).
(1) a. John is [t the cause of the riot]. (canonical)
b. The cause of the riot is [John t]. (inverse)
Under a Dynamic Antisymmetry approach the complement of the
copula must be a point of symmetry and thus a bare SC, i.e. an
unlabelled constituent dominating two maximal projections.
According to Moro, bare SCs have a wider distribution than
standardly assumed. Thus, he argues for wh-phrases to be SC
complements of a phonologically null D(eterminer) (or, in the
case of split wh-constructions, an overt P(reposition)). The
wh-element is analysed as the predicate of a bare SC, which
must adjoin to DP (or PP) in order to neutralise the point of
symmetry constituted by the SC, cf. (2)
(2) a. [which D [book t]]
b. [wat voor [romans t]] (Dutch)
[what for [novels t]]
'what novels'
In the case of wh-objects, movement of the wh-element to the
adjoined position will create a second point of symmetry
between the V(erb) and the wh-element, given Moro's assumption
that wh-elements have both XP and X properties. Wh-elements
are similar to clitics in that they are nonterminals projected
by heads (and thus XPs) which do not dominate another
nonterminal (and thus Xs). Accordingly, V and wh are two
elements belonging to the same category, i.e. they are both
(phonologically overt) heads. Moreover, they c-command each
other, provided that we adopt Kayne's (1994:16) definition of
c-command, which distinguishes between segments and categories
It follows that a wh-object cannot stay in situ but is required
to move to the left periphery of the clause so that the tree
structure can be rendered LCA compatible. In non-wh-splitting
constructions the entire DP complement of V must raise, which
Moro derives from the supposition that a wh-element must be
locally present for the empty functional head D to be able to
license the wh-trace contained in the bare SC.
In his discussion of the second possible type of points of
symmetry, viz. structures that involve multiple adjunction to a
maximal projection, Moro further explores wh-movement. He
claims that movement of a wh-object from its postverbal
position (motivated by the need to neutralise the point of
symmetry between V and wh) results in adjunction of the wh-
object to I(nflection) P(hrase) and thereby creates a point of
symmetry. The search for antisymmetry induces movement of the
wh-object from its IP-adjoined position to a position within
the domain of the C(omplementiser). In contrast, a wh-subject
in a root interrogative will not have to move to CP because the
structure does not contain a point of symmetry which requires
neutralisation.
The asymmetry between wh-subjects and wh-objects concerning
their phrase-structural position is reflected in the subject-
object asymmetry as regards so-called 'do'-support. While 'do'
must appear in wh-object (root) interrogatives, it is absent
with wh-subject interrogatives. Moro suggests that these facts
follow from the independent requirement that C be overt in all
root clauses with an operator in SpecCP. In a Dynamic
Asymmetry account, 'do'-support is linked to wh-movement to the
C-domain, i.e. the appearance of 'do' "signals the exploiting
of a further layer of clause structure, syncretically the CP
periphery" (p. 65).
Moro's conception of wh-movement raises the question how wh-
extraction from an embedded clause can be derived. Limiting
the discussion to wh-object interrogatives for reasons of
space, we can formulate the problem as follows: under Dynamic
Asymmetry, wh-extraction from an embedded clause is predicted
to be possible, only if (i) movement of the embedded wh-object
from its IP-adjoined position to the embedded CP creates a
further point of symmetry, e.g. with matrix V; or (ii) movement
of the embedded wh-object from its IP-adjoined position to the
embedded CP is not available as a strategy to neutralise the
point of symmetry created at the IP-level. Since Moro adopts
Rizzi's (1997) split CP hypothesis, the CP layer may contain
any number of abstract heads which would "protect the wh-phrase
from constituting a point of symmetry with the matrix verb,
hence from further movement" (p. 70). This leaves only (ii) as
a viable option. Moro suggests that wh-movement from an
embedded clause results from the incompatibility of a [-wh] C
and a [+wh] wh-element in the same local domain. In other
words, the point of symmetry created at the IP-level cannot be
neutralised by movement of the wh-object to the embedded CP,
and the structure can be rescued only by movement to the matrix
CP.
Proceeding from the assumption that verbs uniformly select
[-wh] complements, Moro then sets out to develop an account of
wh-movement within embedded clauses. His analysis crucially
relies on the existence of two distinct structures for clausal
complementation. While 'believe'-type verbs take a CP-
complement, verbs which tolerate an embedded wh-phrase (e.g.
'wonder') take a functional projection (FP) as their
complement, to which CP is adjoined, cf. (3).
(3) a. VP b. VP (p. 74, (71))
/ \ / \
V CP V FP
believe wonder / \
CP FP
In (3b), C may be [+wh], since V does not select and govern CP.
Consequently, adjunction of a wh-object to CP will not result
in two opposite values for the wh-feature in the same local
domain. Movement to a position within the embedded CP suffices
to neutralise the point of symmetry created at the IP-level.
The third basic type of points of symmetry, constituted by
structures that pair two heads, is argued to serve as trigger
for clitic movement. Within a Dynamic Asymmetry approach to
phrase structure, which distinguishes clitics from pronouns by
the fact that the former do not dominate another nonterminal
whereas the latter are unambiguously XPs, the different
syntactic behaviour of clitics and pronominals follows without
further stipulation. Clitics obligatorily move to preverbal
position to neutralise the point of symmetry which they
constitute with V. Pronouns, on the other hand, are not found
in a symmetrical structure with V and remain in postverbal
position.
Chapter 4 ("Some Consequences and Speculations") reflects upon
the proposed theory's range of application and its implications
for the design of grammar. The discussion is guided by the
following two questions:
(4) a. Are all movements explained by Dynamic Antisymmetry?
b. Does Dynamic Antisymmetry allow parametric variation?
(p. 96, (1))
In regard to (4a), Moro considers whether covert movement has
any place in a Dynamic Asymmetry framework. He suggests that
even a theory in which movement is derivative from the geometry
of phrase structure may allow for covert movement as long as it
is purely optional. Obligatory covert movement cannot exist
because, by definition, movement is required only to neutralise
points of symmetry. Concerning (4b), Moro observes that
crosslinguistic variation may arise from crosslinguistic
differences that bear on either of the properties defining a
point of symmetry. For example, in a language in which the
subject is expressed by a clitic, or stays in a lower position
within the IP system, or can be expressed by an empty category,
movement of the wh-object will not create a point of symmetry
at the IP-level, which has the result that the wh-object need
not move to CP.
The book concludes with an appendix in which Moro gives a brief
summary of his unified theory of copular constructions.
Comments
Dynamic Antisymmetry presents an important and truly original
contribution, as it challenges two fundamental assumptions of
current Minimalist theory: (i) the conception of movement as
morphology-driven, and (ii) the view that all constructions are
endocentric. In place of (i), Dynamic Antisymmetry offers the
view that movement is a function of the geometry of phrase
structure; with respect to (ii), it suggests that small clauses
play a larger role in syntactic organisation than has generally
been thought. As with any theory which presents a new
theoretical architecture, the burden of proof rests with the
innovator, i.e. the proposals made need to be justified both
theoretically and empirically. Theoretical justification is
provided if it can be shown that the new theory moves more
fully towards explanatory adequacy. Empirical justification
involves demonstrating that the new theory is at least as
descriptively adequate as its predecessors (if not more so).
Moro himself formulates the task at hand when he asks whether
Dynamic Asymmetry can account for all movement (cf. (4a)
above).
In this context, one cannot avoid commenting on the limited
empirical scope of Moro's book. While Moro is right to point
out that Dynamic Antisymmetry is "too powerful [a hypothesis]
for all its consequences to be checked in a single work"
(p. 29), his labelling of the examined phenomena (raising in
copular constructions, wh-movement in interrogatives, clitic
movement in Italian) as "paradigmatic cases of movement" may be
questionable. As he readily admits (p. 126, n.53, n.54), the
book does not examine any of the core cases of A-movement
(i.e., passive, raising, unaccusatives) or head movement (i.e.,
V-movement). Therefore, the success of Dynamic Antisymmetry
crucially rests on its being descriptively more adequate than
the morphological theory of movement within the empirical
domain on which he focuses. The following comments consider
some of the issues raised by Moro's analysis of inverse copular
constructions, wh-movement and cliticisation.
Moro argues that inverse copular constructions provide a
genuine counterexample to the morphological theory of movement.
His line of argument goes as follows. First, he observes that
the postverbal noun phrase in an Italian inverse copular
sentence is assigned nominative Case. Case is an
uninterpretable feature and must be deleted for LF convergence.
Irrespective of whether movement is directly induced by Case-
checking requirements, Case-checking always entails movement
(p. 41). It follows that the postverbal noun phrase in inverse
copular constructions "should move covertly to delete the
uninterpretable nominative Case features associated with it"
(p. 42). Moro then demonstrates that the postverbal noun
phrase is opaque to movement, i.e. neither extraction of, nor
from, the postverbal noun phrase is grammatical. If the
postverbal noun phrase cannot move, the uninterpretable Case
feature cannot be deleted, and the derivation should not
converge, contrary to fact.
Note that Moro does not comment here on the mechanism of Case
assignment for the postverbal noun phrase. Given his stated
goal of severing the link between movement and Case checking,
one would have liked to see at least some sort of suggestion
how Case assignment is assumed to proceed. The problem might
have a possible solution in viewing Case as a reflex of Agree
rather than Move (cf. Chomsky 1999:13). Moro explicitly
rejects this idea, arguing that Agree preempts Move and thus
that we will not be able to derive canonical copular
constructions (in which the SC subject raises), once we assume
that Agree can apply to delete the uninterpretable features of
the SC subject in an inverse copular sentence. However, this
objection might not hold if movement to the preverbal position
in inverse copular constructions will result only in the
valuation of an EPP-feature, i.e. if T(ense) continues to probe
because the predicative noun phrase does not have a full set of
phi-features. If so, the morphological theory of movement
might not fare any worse than Dynamic Antisymmetry with respect
to inverse copular constructions.
As for wh-movement, Dynamic Antisymmetry is able to offer a
natural account of the subject-object antisymmetry observed
with 'do'-support. Note, however, that English multiple wh-
questions fall outside the proposed analysis. If wh-objects
undergo movement to the clausal periphery in order to
neutralise not only the point of symmetry constituted by the
small clause but also, in a second step, the point of symmetry
between V and the wh-element, it is predicted that wh-objects
cannot remain in postverbal position. This is clearly false,
as seen in (5).
(5) Who gave what to Zoe?
Within Dynamic Antisymmetry, the problem posed by multiple
interrogatives may be approached from two different angles.
First, we may pursue an analysis in which multiple
interrogatives are derived by IP-adjunction of the wh-object,
followed by a second movement operation which obscures the
effect of the object wh-movement. In other words, it is
conceivable that the syntax of multiple questions can receive
an analysis akin to the one proposed by Kayne (1994) for
language variation in linear order. However, it is difficult
to see how such an analysis might work in the particular case
of (5), given that the wh-subject and the verb do not form a
single constituent.
The second option is to argue that wh-objects in single and
multiple wh-questions differ in their phrase structure. Since
a wh-object can remain in situ only if it does not constitute a
point of symmetry but, conversely, can undergo wh-movement only
if it is one of the elements in a point of symmetry, wh-objects
must be able to occur in two different syntactic
configurations. Indeed, this is the strategy adopted by Moro
to account for the existence of wh-movement both from and
within embedded clauses. Movement from an embedded clause
takes place from a CP which is the complement of the matrix
verb; movement within an embedded clause occurs within a CP
which is the specifier of the FP-complement of the matrix verb.
While perhaps technically possible, such an approach would
ultimately lead us to posit different underlying structures for
nearly all phrase markers which differ in phonetic form
(perhaps with the exception only of canonical and inverse
copular constructions) - a result which is clearly not
desirable.
The last case of movement considered in this book is
cliticisation, which Dynamic Asymmetry assumes to be triggered
by a symmetrical head-head structure. In addition to
pronominal clitic movement, cliticisation also occurs in
copular sentences, e.g. (6).
(6) Queste foto lo sono t
these pictures LO are
(p. 46, (26b))
As a canonical copular construction, the sentence is derived by
movement of the SC subject 'queste foto' ('these pictures').
It also contains the propredicative element 'lo', which is an
uninflected clitic and thus in preverbal position. The
question now arises why movement of both the SC subject and the
SC predicate should be necessary. If noun phrase and
propredicative clitic are merged as sisters in a bare SC,
movement of one or the other should be sufficient to neutralise
the SC point of symmetry. In a chapter which illustrates the
applications of Dynamic Antisymmetry to SC and clitic
constructions, this reader would have appreciated clarification
of this matter.
A final, theoretical point may be raised concerning the place
allocated to uninterpretable features in Dynamic Antisymmetry.
Moro leaves open the issue as to the existence and role of
uninterpretable features in syntax, but appears to imply that
uninterpretable features do exist when he writes in a footnote
that "uninterpretable features can only be affected by ...
Agree operations" (p. 116, n.12). It should be emphasised that
in the continued presence of uninterpretable features, the move
to dissociate movement and morphology is not without
consequence but would elevate uninterpretable features from
apparent to true imperfections.
These comments notwithstanding, Dynamic Antisymmetry proves to
be a thought-provoking and lucidly written work, which has the
potential of advancing our understanding of syntactic theory.
Since it provides all necessary background, it should be
accessible to readers whose area of expertise does not lie
within the theory of Antisymmetry and/or current theories of
movement.
References
Chomsky, N. (1995) The Minimalist Program, MIT Press.
Kayne, R. (1994) The Antisymmetry of Syntax, MIT Press.
Moro, A. (1988) 'Per una teoria unificata delle frasi
copulari', Rivista di Grammatica Generativa 13, 81-110.
Moro, A. (1997) The Raising of Predicates, Cambridge University
Press.
Rizzi, L. (1997) 'The fine structure of the left periphery', in
L. Haegeman (ed.), Elements of Grammar, Kluwer Academic
Publishers, 281-337.
Kerstin Hoge holds the posts of Junior Lecturer in German
Linguistics at the University of Oxford, and of Yiddish Lector
at the Oxford Centre for Hebrew and Jewish Studies. Her
D.Phil. thesis (University of Oxford, 2000) investigates
superiority phenomena in English, German and Yiddish - a topic,
which continues to be a research interest, alongside issues in
theoretical syntax and Yiddish lingustics.
|
