LINGUIST List 12.2230
Thu Sep 13 2001
Review: Bresnan, Lexical Functional 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.
Directory
Anna Feldman, Review of Bresnan, Lexical Functional Syntax
Message 1: Review of Bresnan, Lexical Functional Syntax
Date: Thu, 13 Sep 2001 00:48:49 -0400 (EDT)
From: Anna Feldman <afeldmanling.ohio-state.edu>
Subject: Review of Bresnan, Lexical Functional Syntax
Bresnan, Joan (2001) Lexical Functional Syntax. Blackwell Publishers,
paperback ISBN 0-631-20974-3, 446pp., Blackwell Textbooks in Linguistics,
$39.00.
Anna Feldman, The Ohio State University
Bresnan's book aims at providing both an introduction to
Lexical-Functional Grammar and a survey of the research done within this
framework during the last two decades. It is addressed to the readers
familiar with standard derivational theories and with elementary formal
concepts such as the definition of functions and relations. The book
provides an accessible, empirically motivated treatment of the
mathematical structure of LFG. It also covers the theoretical linguistic
ideas that LFG can model, and discusses the wide range of cross-linguistic
phenomena to which it has been applied.
The book is organized as follows.
Part 1 On the Architecture of Universal Grammar
1. Nonconfigurationality
2. Movement Paradoxes
3. Lexicality and Argument Structure
Part 2 Formally Modelling the Architecture
4. A Formal Model of Syntactic Structure
5. Monotonicity and Some of its Consequences
Part 3 Inflectional Morphology and Phrase Structure Variation
6. A Theory of Structure-Function Mappings
7. Endocentricity and Heads
8. Pronoun Incorporation and Agreement
9. Topicalization and Scrambling
Part 4 On Functional Structures: Binding, Predication, and
Control
10. Basic Binding Theory
11. Types of Bound Anaphors
12. Predication Relations
13. Anaphoric Control
14. From Argument Structure to Functional Structure
Problem Sets and Solutions
5 Problem Sets
Solutions to Selected Exercises
References
Index of Languages Referenced
Index of Concepts
Part 1 of the book empirically and informally motivates the LFG
architecture by looking at the core linguistic phenomena:
non-configurationality, movement paradoxes, and the lexicality of relation
changes such as passivization.
Warlpiri is one of the best examples of non-configurationality due to 1)
its free word order at the clause level. There are few syntactic rules
determining word order in Warlpiri apart from the position of the
auxiliary (sentence-initial position or after the first constituent) and
the position of interrogative words (almost always sentence-initial
position). The thematic choices of the speaker, rather than syntactic
rules, largely determine word order in Warlpiri with the sentence level
topic being positioned in sentence-initial position; 2) its use of
syntactically discontinuous expressions. Two nominal (or verbal) elements
forming a single expression may not necessarily occur in linearly adjacent
positions; 3) its extensive use of null anaphora;
Typological considerations such as noncongfigurationality motivate a
relational design of universal grammar. It is shown that
noncongfigurationality is possible because the same grammatical
information can be specified by word "shapes" as by word "groups". In
Lexical Functional Grammar lexical elements are as important as syntactic
elements in expressing grammatical information ("lexical") and the
grammatical information is not identified with particular structural forms
of expression but is viewed as a "system of abstract relators" of
expressions to eventualities ("functional").
The category mismatches in extraction configurations found in English are
expected in the relational architecture of LFG due to the fact that the
correspondence between structure and function is not perfect: there can be
mismatches between the functional structure attributes of an element and
the constituent structure positions it can appear in (these notions are
discussed below in more detail).
A third motivation for the relational architecture of LFG is based on the
lexicality of passivization and other changes in argument structure
realization.
Part 2 shows how the intuitive ideas of part 1 can be formally modelled as
flexible correspondence mappings between parallel structures
(c(ategorial)-structure and f(unctional) structure). Some of the ideas
are summarized below.
The LFG formalism postulates two levels of syntactic representation for a
sentence, a c-structure and an f-structure. These are related by a
piecewise correspondence that permits the properties of the abstract
functional structure to be defined in terms of configurations of
constituent structure phrases. The basic architecture crucially separates
the three notions of structure, structural description and structural
correspondence.
Mathematically, an f-structure is a finite set of pairs of attributes and
values. Every attribute has a unique feature (Uniqueness Condition), but
different attributes may have the same value. The formal model of LFG
embodies three general design principles: variability, universality, and
monotonicity.
The principle of *variability* states that external structures vary across
languages (the formal model of external structure in LFG is
c(onstituent)-structure). The constituent parts of sentences and phrases
can be ordered by precedence, dominance, and structural types. Fully
inflected words are the terminal elements of c-structures of sentences,
and every word belongs to exactly one node (this relation is often
referred to as 'lexical integrity').
The principle of *universality* states that internal structure are largely
invariant across languages. The internal structure is reflected in
phenomena such as case government, pronominal binding, and agreement
relations among the predicators and arguments of a sentence.
The correspondence between the nodes of the c-structure and f-structure is
many-to-one, i.e. different c-structure nodes may correspond to the same
f-structure. The grammar does not prescribe a particular computational
process of deriving f-structures from c-structures (or the reverse).
However, there is an algorithm for deriving f-structure from a
c-structure. Given a lexical-functional grammar and lexicon for a language
L, there is an algorithm for deriving the c-structure and f-structure of
any sentence of L. The method of solution proceeds on three steps: 1)
annotate the c-structure tree with the appropriate functional schemata; 2)
instantiate the schemata to generate a functional description; 3) solve
the simultaneous equations of the functional description by constructing
the minimal f-structure that satisfies them.
In LFG the correspondence mapping between internal and external structures
does not preserve sameness of form. Instead it is designed to preserve
inclusion relations between the information expressed by the external
structure and the content of the internal structure (*monotonicity*). This
notion is discussed in chapter 5 in detail. In a nutshell, in mapping from
a c-structure to an f-structure, we accumulate local functional
descriptions from parts of the c-structure to get a functional structure
for the whole. As new functional equations are added to the functional
descriptions, the resulting f-structures that satisfy these descriptions
become increasingly specific; they have more attributes. The mapping from
c-structure to f-structure increases monotonically. This formal property
is significant. The monotonicity implies that the grammatical relations of
parts are preserved in the whole. The mapping does not destroy or change
the grammatical relations.
Completeness and Coherence are general well-formedness conditions on
f-structures. Completeness requires that every function designated by a
PRED be present in the f-structure of that PRED. It also requires a
further matching between PREDs and their f-structure functions: if a
designator is associated with a semantic role by the PRED, the f-structure
element satisfying the designator must itself contain semantic feature
[PRED v]. Coherence requires that every argument function in an
f-structure be designated by a PRED. Furthermore, any function that has a
semantic feature must match with a designator associated with a semantic
role by its PRED. The extended coherence condition applies not just to
argument functions, but also to all syntactic functions, requiring that
they be integrated appropriately into the f-structure.
After the examining the basic formal architecture of LFG, the reader is
offered problem sets where s/he can try to apply the formalism learned in
the chapter to real linguistic data.
Part 3 presents a theory and typology of structure-function
correspondences, and several case studies of languages in which syntactic
functions are created morphologically rather than by constituent
structures.
This part mainly discusses the problem of invariability. The idea is that
principles of conflict resolution can play an explanatory role in
accounting for variation in forms of expression.
The book adopts the specific economy principle: All syntactic phrase
structure nodes are optional and are not used unless required by
independent principles (completeness, coherence, semantic
expressivity);("syntactic phrase structure nodes" refers to those
nonterminal nodes which do not immediately dominate a lexical element).
The principle of economy of expression implies that if a syntactic phrase
structure node provides only redundant information, it is not allowed.
Economy of expression creates potential competition between different
forms of expression that carry overlapping information within the same
sentence or phrase. Economy of expression creates potential competition
between different forms of expression that carry overlapping information
within the same sentence or phrase.
The principle of lexical integrity - Morphologically complete words are
leaves of the c-structure tree and each leaf corresponds to one and only
one c-structure node- implies that while morphemic words and syntactic
phrases are different types of forms of expression in c-structure, they
may carry the same types of information in f-structure. In other words,
these different forms of expression - words and phrases- may be
functionally equivalent (in terms of f-structure content). In such cases
of equivalence, economy of expression privileges words over phrase
structure nodes: it is only the syntactic nodes whose presence must be
justified by economy of expression. Words are considered to be more
economical than phrases. If the syntactic phrase structure nodes do not
bear additional functions that distinguish them from the morphological
structures, they must be omitted. Such a setup provides a theoretical
explanation for the existence of phrase structure variation within the
formal model of UG.
The chapter outlines a particular theory of c-structure to f-structure
mappings based on these principles. In the subsequent chapters, the author
applies this theory to several types of phrase structure variation seen
crosslinguistically: head movement, pronominal incorporation and
scrambling.
Part 4 motivates functional structure by showing how invariances of
language are captured on functional structures and outlines a theory of
how functional structures are projected from argument structures.
In Part 3 it has been shown that hierarchical phrase structure is
diminished in varying ways by the presence of rich morphological
specifications. Part 4 addresses the f-structure issues. F-structure
models the abstract predication relations that grammars systematize.
Section 10 is concerned with Binding Theory. In LFG, binding theory is
defined at the level of functional structure (and a-structure), because
anaphors, pronominals, and referring expressions are not uniformly
represented in c-structure positions across languages or even within the
same language.
After outlining the basic concepts and examining the grammatical binding
constraints, the formalization of the Binding Constraints is proposed.
According to the theory outlined, constraints on anaphoric binding are not
expressed in terms of general principles holding invariably for all
anaphoric elements but are directly associated with the lexical properties
of the anaphors themselves.
The constraints associated with the anaphoric elements specify (a)
coreference requirements (positive constraints) or disjointness
requirements (negative constraints), (b) the syntactic domain in which the
anaphor may be bound or free (domain constraints) and (c) the required
grammatical function (e.g., SUBJ, OBJ, OBL [$_{\theta}$]) of the
antecedent (antecedent constraints). Binding constraints are defined at
the level of feature structure and expressed in terms of the grammatical
concepts of predicate (PRED), subject (SUBJ) and tense (TENSE). They are
stated as binding equations that define the permissible relations between
the f-structure of an anaphoric or pronominal element and the elements
with which it may or may not corefer. These constraints are formally
expressed by ``inside-out'' functional uncertainty equations which define
an infinite disjunction over the possible f-structures which may contain
the anaphor or the pronoun. An expression lexically associated with the
anaphor picks out a set of less embedded f-structures which must be the
antecedent of the anaphoric element, or f-structures with which the
antecedent may not corefer.
Three dimensions play a central role in anaphoric binding systems: 1) the
subjective dimension, which indicates whether or not the antecedent must
be a grammatical subject; 2) the nuclear dimension, which indicates
whether or not the antecedent must lie in the same nucleus as the anaphor;
3) the logophoric dimension, in which the pronoun refers to one whose
speech, thoughts, or feelings are represented in indirect discourse, from
that person's own point of view.
Different types of bound anaphora are discussed in section 11 which reveal
more dimensions of anaphoric binding. The languages discussed are
Icelandic, Norwegian, Ewe (a West African language) and some others.
Section 12 examines predicative complements (XCOMPS), using participial VP
complements as the paradigms case. The characteristic binding patterns of
predicative VP complements extend across categories and can be explained
by identifying in f-structure a complement subject with the subject or
object of the matrix verb. The relation between this implicit subject and
the matrix argument is called functional predication or functional
control.
Section 13 examines a contrasting verbal construction that exhibits a
different type of control, called anaphoric control, known in English as
gerundive VPs (true for some infinitival constructions too). The last
section is devoted to the lexical mapping theory, which projects skeletal
f-structures from argument structures by general principles. The basic
syntactic principles for mapping a-structure to surface grammatical
functions are as follows. The underspecified roles are freely mapped onto
all compatible grammatical functions subject to a few general constraints:
if it is the initial argument of the predicator, a most prominent role
classified [-o] has to be mapped onto the subject function; if such a role
is unavailable, a non-agentive unrestricted role is mapped onto the
subject position. All other roles are mapped onto the lowest compatible
function on the partial ordering.
All in all, the book provides a very clear introduction to LFG, a
framework that differs from both transformational and relational grammar
in assuming a single level of syntactic structure. LFG rejects syntactic
movement of constituents as the mechanism by which the surface syntactic
realization of arguments is determined and it disallows changes of
grammatical relations within the syntax. A unique constituent structure
that corresponds to the superficial phrase structure tree is postulated.
The dissociation of syntactic structure from predicate argument structure
is crucial to the LFG framework. The single level of syntactic
representation, constituent structure, exists simultaneously with a
functional structure representation that integrates the information from
c-structure and from the lexicon. While c-structure varies across
languages, the f-structure representation, which contains all necessary
information for the semantic interpretation of an utterance, is claimed to
be universal. Phenomena that had been accounted for by the interaction of
transformations are explained by the regular interaction of lexical
processes.
This book should be of interest to anyone concerned with theories of
grammar. It is addressed not only to formally and computationally inclined
linguists but also to linguists interested in typology and language,
thanks to the ample cross-linguistic evidence that is used to illustrate
the scope of the theory. The book demonstrates the applicability of the
LFG approach to a wide range of empirical problems. The author makes clear
how an architecture of imperfect correspondences between parallel
information structures compares with and improves upon approaches
undertaken in transformational frameworks. The book can be used both as a
textbook for students and as a reference text for researchers. The problem
sets (with the solutions) included in the book support the text and
provide an essential practice in using the formalism and analytical
concepts of LFG. Almost every section is supplemented with discussions and
suggestions for further readings.
Anna Feldman is a student at the Ohio State University, Department of
Linguistics.