LINGUIST List 13.372
Mon Feb 11 2002
Review: Legendre, Grimshaw & Vikner, OT Syntax
Editor for this issue: Terence Langendoen <terrylinguistlist.org>
What follows is another discussion note contributed to our Book
Discussion Forum. We expect these discussions to be informal and
interactive; and the author of the book discussed is cordially invited
to join in.
If you are interested in leading a book discussion, look for books
announced on LINGUIST as "available for discussion." (This means that
the publisher has sent us a review copy.) Then contact
Simin Karimi at simin
linguistlist.org
or Terry Langendoen at terry
linguistlist.org.
Subscribe to Blackwell's LL+ at
http://www.linguistlistplus.com/ and
donate 20% of your subscription to LINGUIST! You get 30% off on
Blackwells books, and free shipping and postage!
Directory
Alex Alsina, review of Legendre et al., Optimality-Theoretic Syntax
Message 1: review of Legendre et al., Optimality-Theoretic Syntax
Date: Mon, 11 Feb 2002 09:36:34 +0100
From: Alex Alsina <alex.alsinatrad.upf.es>
Subject: review of Legendre et al., Optimality-Theoretic Syntax
Legendre, G�raldine, Jane Grimshaw, and Sten Vikner, ed.
(2001) Optimality-Theoretic Syntax. MIT Press, xviii+548pp,
paperback ISBN 0-262-62138-X, $42.00, A Bradford Book.
Alex Alsina, Universitat Pompeu Fabra.
Optimality-Theoretic Syntax is an edited collection of
papers that share the feature of applying the framework and
ideas of Optimality Theory (OT) to the domain of syntax. An
overview by the editors highlighting the main contributions
of each paper precedes the body of the book, consisting of
sixteen chapters. In what follows I first give a brief
summary of each paper, in the order in which they appear in
the book, and then make general remarks about the versions
of OT that are presented in this volume.
SUMMARY
G�raldine Legendre, An Introduction to Optimality Theory in
Syntax, presents an overview of OT as it applies to syntax
with the purpose of making the rest of the volume accessible
to a reader who is not familiar with OT. This chapter
illustrates the main ideas of this formal theory of
constraint interaction by means of analyses of various
syntactic phenomena. And points out the similarity between
syntax and phonology, since, within OT, "the two modules
operate on the basis of the same formal and markedness
principles."
Peter Ackema and Ad Neeleman, Competition between Syntax and
Phonology, account for the observation that certain complex
lexical items (such as verb-particle complexes in Dutch) can
alternate between a morphological realization and a
syntactic realization depending on the environment. Assuming
that such complex expressions are generated either in the
morphology or in the syntax, the choice between these two
modules will be the result of the application of constraints
(one favoring syntactic structures over morphological
structures and one requiring affixes to select morphological
hosts).
Judith Aissen, Markedness and Subject Choice in Optimality
Theory, accounts for the presence and use of morphology
signaling marked subject choice (ergative case, inverse,
passive, etc.) crosslinguistically. The analysis makes
crucial use of the formal devices of harmonic alignment of
prominence scales and local conjunction and of the claim
that there are scales of person and animacy, semantic role,
and grammatical relations underlying the association between
these types of information.
Eric Bakovic and Edward Keer, Optionality and Ineffability,
propose an analysis of the distribution of complementizers
in English and Norwegian, accounting for when the
complementizer is obligatorily present or absent and when
optional. Optionality in this case results from different
inputs and consequently different candidate sets with the
relevant faithfulness constraints outranking markedness,
whereas the obligatory situations result from a ranking of
the relevant markedness constraint above faithfulness, which
has the effect of neutralizing differences in the inputs.
Joan Bresnan, The Emergence of the Unmarked Pronoun,
proposes constraints relating the form of personal pronouns
(zero, bound, clitic, weak, or pronoun) with the properties
of pronouns (anaphoricity, topic-anaphoricity, and
classification by agreement features) in order to explain
certain crosslinguistic generalizations. These include the
claim that languages may lack bound or zero pronouns, but no
language lacks free forms, and that, in languages where
bound and free pronominals coexist, the free pronoun is used
for focus where a bound form is available, but has a
nonfocus use only when the bound form is lacking.
Hye-Won Choi, Binding and Discourse Prominence:
Reconstruction in "Focus" Scrambling, analyses the binding
and focality facts associated with scrambling in German as a
result of constraint competition between two components of
grammar, syntax and discourse. The scrambling facts follow
from the interaction of a syntactic constraint, requiring
subjects to precede nonsubjects, with two discourse
constraints, requiring old and prominent information to
precede new and nonprominent information respectively, and
the binding facts are explained by a local conjunction of
constraints in operator binding requiring the binder to
linearly precede its bindee and to outrank it in the
functional hierarchy.
Jo�o Costa, The Emergence of Unmarked Word Order, explains
the unmarked word orders in different languages through the
reranking of a small set of constraints. These constraints
are a combination of purely syntactic constraints (such as
the requirement that the subject be in SpecIP or a
constraint penalizing movement) and of constraints that take
into account discourse information such as focus and topic.
The languages that illustrate the various rankings of
constraints include Portuguese, Spanish (two dialects),
Italian, Greek, Arabic, Berber, Chamorro, Malagasy, and
Celtic.
Jane Grimshaw, Optimal Clitic Positions and the Lexicon in
Romance Clitic Systems, accounts for the order of clitics in
combination, some restrictions on clitic combinations, and
the actual inventory of clitic specifications in several
Romance languages. The order of clitics follows from a set
of positional constraints that are satisfied when an element
with the relevant specification is placed at the designated
edge of the cluster. The selection of unexpected clitics in
combination such as Spanish "spurious se" follows from
ranking certain positional constraints above faithfulness
constraints. Finally, the ranking of constraints predicts
the set of morphosyntactic combinations that can be found in
a clitic system.
G�raldine Legendre, Masked Second-position Effects and the
Linearization of Functional Features, proposes a unified
analysis of verb-second phenomena (such as is found the
Germanic languages) and second position clitics (as in many
South Slavic languages). The claim is that finite verbs and
clitics are linearized at Phonetic Form (PF) by two
conflicting constraints, one requiring the relevant feature
not to be realized in intonational phrase-initial position
and another one requiring this feature to be left-aligned
with the edge of the nearest projection of the head that
this feature is associated with. If the former constraint
outranks the latter, the relevant element appears in second
position; if the latter outranks the former, the relevant
element appears in initial position.
Gereon M�ller, Order Preservation, Parallel Movement, and
the Emergence of the Unmarked, proposes a constraint that
requires c-command relations to be preserved across levels
of syntactic representation, which interacts with
constraints favoring movement as well as with a constraint
penalizing movement. The newly proposed constraint accounts
for situations in which the D-structure order of
constituents is preserved at S-structure, including
superiority effects in English, wh-movement in Bulgarian,
pronominal object shift in Danish, object shift of lexical
NPs in Icelandic, Case-driven NP raising, pronoun fronting
in German, and relative scope and QR in German.
Vieri Samek-Lodovici, Crosslinguistic Typologies in
Optimality Theory, investigates the structure of the
typology entailed by any language-specific analysis within
OT. The formal properties of OT typologies are illustrated
through the analysis of the syntax of new-information focus.
The basic word order of a language and other facts are
claimed to emerge as epiphenomenal effects of constraints
governing phrase-structure and movement and constraints
governing the syntactic realization of focused phrases.
Peter Sells, Form and Function in the Typology of
Grammatical Voice Systems, proposes to derive the inventory
of voice forms in any language from the language-particular
ranking of constraints. Different inventories of voice forms
are evaluated relative to each other and the winning
candidate is the actual voice system of the language. The
analysis involves the assumption that arguments have three
levels of prominence, that these levels of prominence
constrain the linking to grammatical functions, and that
thematic roles also constrain the linking to grammatical
functions. The unmarked linking is the best candidate in an
OT evaluation and is claimed to be the morphologically
unmarked form in the language; the next most optimal
candidate with contrasting prominence relations will be a
morphologically marked voice form, if the language has one
contrast; and so on if the language has more than one
contrast.
Margaret Speas, Constraints on Null Pronouns, proposes a set
of violable constraints, which, variously ranked, give the
facts of English, where null pronouns occur only in the
subject position of nonfinite clauses, the facts of Thai,
where null pronouns occur as subjects and objects regardless
of finiteness, and the facts of Spanish, where null pronouns
occur only as subjects regardless of finiteness. She also
brings out the importance of the role of point of view in
the licensing and interpretation of null pronouns.
Sten Vikner, V-to-I movement and do-Insertion in Optimality
Theory, deals with the position of the finite verb relative
to a sentential adverb and to negation, the occurrence of
auxiliary do, and the optionality or obligatoriness of
complementizers in English, Danish, French, and Icelandic.
These facts, both within a language and across languages,
are derived from an interacting set of constraints that are
ranked differently in each language. An interesting
empirical claim is the idea that the obligatoriness or
optionality of complementizers in a given language
correlates with apparently unrelated properties of the
language such as the position of the finite verb.
Colin Wilson, Bidirectional Optimization and the Theory of
Anaphora, motivates two types of competition on the basis of
binding phenomena: interpretive competition, where competing
sentences differ in their semantic interpretation but not in
their syntactic structure, and expressive competition, where
candidates have the same intended meaning but differ in
their syntactic structure. Data revealing a variable binding
domain for certain anaphors motivate the first type of
competition, and the partial complementarity of pronouns and
anaphors in some languages motivates the latter. The paper
proposes to integrate both types of competition in a
bidirectional model of the syntax-semantics interface.
Ellen Woolford, Case Patterns, accounts for situations in
which there is a potential competition among Cases by having
markedness constraints that require avoidance of marked
Cases interacting with faithfulness constraints that require
the presence of lexically determined Case features. The data
accounted for include structures in which the presence of a
particular Case depends on the presence of another Case,
such as structures in which an accusative argument is only
possible if there is another nominative argument, structures
in which a particular Case is only possible with certain
aspectual features, as with ergative Case and perfective
aspect in Hindi, among others.
GENERAL REMARKS
The application of Optimality Theory to syntactic analysis
has a shorter history than the application of Optimality
Theory to theories of phonology, but, by now, OT has
established a solid foothold in syntactic analysis and this
volume is definite proof of this. OT represents an important
change in the way we think about linguistic theory and
analysis. Among other differences with non-OT analyses,
checking the predictions of a theory that involves OT is no
longer a matter of determining whether a given structure
violates the principles (or, rather, constraints) of a
theory, since a given structure may violate some constraints
and yet be well-formed. Such a structure is grammatical if
there is no other structure among those it competes with
that has fewer violations of the most highly ranked
constraint on which the two structures differ. Thus, the
grammaticality of a structure does not involve only the
structure in question and the set of principles or
constraints of the theory, but crucially the whole
(infinite) set of competing structures. This makes checking
the predictions of the theory potentially very complicated
(which may be a problem from the point of view of a reader
who wants to evaluate a theory, but is not necessarily a
defect of the theory or a drawback from the point of view of
the linguist proposing the theory). In any case, it becomes
very important to establish the set of competitors, or
candidate set, out of which a grammatical structure has to
emerge. The set of competitors is characterized by a common
input. And what the input is in syntax emerges from the
collection of papers in Optimality-Theoretic Syntax as one
of the unresolved problems facing the application of OT to
syntax: there is no consensus among the OT syntacticians
represented in this volume as to how the input is defined
and, consequently, as to how the candidate set is defined.
What follows concerns this important issue.
In OT applied to phonology, the input is taken to consist of
lexical contrasts (underlying representations). The set of
principles called GEN applies to any input producing a large
number of structures that constitute the candidate set. The
ranked constraints in EVAL pick out the optimal candidate as
the grammatical structure. This model of grammar is
perfectly consistent with the standard conception that there
is a division in grammar between general statements and
item-specific statements. The former kind of statement is
expressed as rules, principles, constraints, etc. that apply
to large classes of structures, whereas the latter kind is
information about a particular linguistic item, be it a
word, a morpheme, a syntactic structure, etc. This suggests
that there are two broad components in grammar reflecting
this distinction: the component of regularities, containing
all the general properties, and the lexical component,
containing all the item-specific properties. OT phonology
adopts this view: the lexical component provides the input
and the component of regularities is divided into two
subcomponents: the subcomponent that generates structures
for the inputs-GEN, which consists of inviolable
constraints-and the subcomponent that contains statements
that are sufficiently general and recurrent
crosslinguistically to merit being considered regularities,
but may fail to be satisfied in particular linguistic
expressions-EVAL, which consists of violable constraints.
One could imagine that this overall model might also be
applied to syntax. The input would consist of lexical items,
or, more precisely, the syntactic information of lexical
items, if we are only concerned with syntax, which would
include categorial information, argument structure,
agreement features, case features, etc. GEN would take this
input, build structures for it, manipulate it in various
ways by adding or deleting features, etc. and generate an
infinite number of candidates for EVAL. This simple (or,
perhaps, simple-minded) extension of OT to syntax is seldom
what we find in actual syntactic analyses that use OT. Most,
if not all, of the papers in this volume adopt a more
elaborate view in which the input does not consist only of
lexical items, that is, item-specific or idiosyncratic
information. Furthermore, there is considerable discrepancy
among the authors as to what else, in addition to lexical
information, is part of the input. Some discussion about the
nature of the input in syntax can be found in some of the
chapters, such as Speas's, pp. 399-400.
Whatever the reasons may be for rejecting the idea that the
input contains nothing but lexical information, the fact is
that it has some important consequences. In the first place,
it breaks the parallel between phonology and syntax that is
often presented as a positive feature of the extension of OT
to syntax (as in Legendre's introductory chapter in this
volume). Whereas the input in phonology consists only of
lexical information, that would not be the case with the
input in syntax. In the second place, allowing more than
just lexical information in the input implicitly introduces
a third subcomponent within the system of regularities, in
addition to GEN and EVAL. If the input contains structure,
features, specifications, etc., beyond what is provided by
the lexical component, it means that some set of rules has
added this non-lexical information. The existence of this
"pre-input" set of rules is not explicitly acknowledged by
any of the authors in this volume. In the third place, since
this "pre-input" set of rules is not even mentioned, there
is no discussion about what it is, what it does, or how it
does what it does, which means that there is a portion of
the theory that needs to be clarified. In the fourth place,
and closely related to the latter two points, there is a
considerable amount of variation among authors about what
the input is. For some authors the input has very little
more than the bare lexical information (this would seem to
be the case with M�ller's proposal in Optimality-Theoretic
Syntax); for other authors the input consists of
full-fledged syntactic structures (which seems to be the
proposal in Wilson's chapter). It is perfectly legitimate
for different authors to have different theoretical
conceptions. But, as long as there is no unified view about
what the input is in OT syntax, the term "OT" will simply
denote a formal framework of constraint interaction. One
might expect "OT" also to denote a model of grammar
specifying what the various components or subcomponents of
the grammar are (e.g., a lexical component and one or more
components of regularities) and how they interact. As it is,
there is no general consensus about OT as a model of grammar
in the sense intended here.
To try to make this last point clearer, if OT is no more
than a formal framework of constraint interaction, as
described in Legendre's chapter, page 3, there is no
restriction on what an input can be. Within this view, a
linguist interested in explaining a certain phenomenon has
to define an input so that a candidate set can be generated
by GEN and subsequently evaluated by EVAL, but the input can
be anything to which linguistic principles and constraints
can apply, ranging from sets of morphosyntactic features to
full-fledged syntactic structures. Depending on what
phenomenon the linguist is interested in explaining, the
nature of the input will be defined in one way or another.
For phenomenon A, the input would be defined using one set
of criteria, whereas, for phenomenon B, it would be defined
using a different set of criteria. This situation seems
undesirable because there is no substance attached to the
term "input": it is just a concept in the formal framework
of constraint interaction known as OT. I expect that
linguists working within OT will eventually converge on a
model of grammar for OT, in which concepts such as "input"
will have a commonly accepted definition in substantive
terms.
To conclude, Optimality-Theoretic Syntax is an excellently
edited volume, containing sixteen papers that present
interesting analyses of a variety of syntactic phenomena.
The fact that the papers taken as a whole reveal what seems
to me to be a problem for a unified conception of OT as a
model of grammar in the sense indicated earlier does not
detract from the high quality of the volume or of the
individual papers. I believe this volume is essential for
anyone interested in the application of OT to the syntactic
domain or, more generally, in syntactic theory.
BIOGRAPHICAL SKETCH
I received my PhD from Stanford University in 1993. My
research interests are argument structure, complex
predicates, object asymmetries and, more generally,
linguistic theory.