LINGUIST List 14.1691
Mon Jun 16 2003
Review: Linguistic Theories: Alexiadou, ed. (2002)
Editor for this issue: Naomi Ogasawara <naomilinguistlist.org>
What follows is a review or discussion note contributed to our Book
Discussion Forum. We expect 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 review." Then contact
Simin Karimi at simin
linguistlist.org.
Directory
Sergei Sey, Theoretical Approaches to Universals
Message 1: Theoretical Approaches to Universals
Date: Sat, 14 Jun 2003 16:45:54 +0000
From: Sergei Sey <rusakovAR2015.spb.edu>
Subject: Theoretical Approaches to Universals
Alexiadou, Artemis, ed. (2002) Theoretical Approaches to Universals,
John Benjamins Publishing Company, Linguistik Aktuell/Linguistics
Today 49.
Announced at http://linguistlist.org/issues/13/13-2678.html
Sergey Say, ILI RAN (Institute for linguistic research of the Russian
Academy of Sciences), Saint-Petersburg, Russia
This volume presents a selection of papers from the GLOW conference on
universals organized by the Research Center for General Linguistics
(Berlin), the Linguistic Department of the University of Potsdam and
the Dutch Graduate School in Linguistics (Berlin, March 1999). The
authors of the contributions to the volume all share a generativist
(in most cases, a minimalist) background; the goal of the volume is to
enrich our knowledge of what is universal and what is language-
specific in linguistic structures, and in particular to answer some of
the following questions: ''how are the primitive notions of the
structure building apparatus, Merge, Move, Agree or Attract defined?
(...) What does a typology of features look like? (...) [H]ow are we
to understand variation in morpho- lexical features exactly? Finally,
is it true that morphological variation dispenses with the need for
structural variation?'' (p. 5).
SUMMARY
The book opens with an introduction by the volume's editor, Artemis
Alexiadou. This introduction is itself divided into two parts, the
first containing a brief overview of the current state of affairs in
the universals research (with special emphasis on convergences and
discrepancies between generative and functional-typological approach),
the second giving a summary of the contributions to the volume. The
rest of the volume consists of 9 papers and a subject index (with an
in-built language index). References are given after each paper.
The first paper of the book (''Universal features and language-
particular morphemes'', by Maya Arad) elaborates on the well-known
view that assigns variation among languages to morphology (Chomsky
1995). A specific formulation of this view is offered, namely, that
''there are three sources for language variation: the inventory of
roots a language has, the features it has selected out of a universal
inventory, and the way these features are bundled together''
(p. 15). In particular, Arad investigates those features that are
bundled together in the upper head in the VP- shell; this head
(commonly referred to as ''little v'') introduces an external argument
in its specifier and enters into a relation with the object. Arad
observes semantic and syntactic properties of verbs denoting mental
states (psych verbs) in a number of languages (Italian and other
Romance first of all); she distinguishes between stative and agentive
readings of Object Experiencer verbs and ascribes this distinction to
the type of verbal morpheme with which the root is combined: it is
either ''(standard) little v'' or a morpheme labeled as ''stative
little v''. These ''little v'' morphemes are responsible for the
syntactic effects such as reflexivization, causativization and
extraction from the object (thus, in Italian, reflexivization is
ungrammatical with the stative reading of the ObjExp verbs). Finally
Arad proposes a preliminary typology of ''little v'' morphemes, basing
on the assumption that these morphemes have three types of properties:
verbalizing property (making a verb out of a category neutral root),
semantic content (agentive, stative, inchoative, causative etc.) and
transitivity. It is shown that languages may put together different
sets of features into their ''little v'' morphemes. It is further
generalised that the same root can form different types of ''verbs''
when combined with verbal morphemes of different types.
In the following paper (''Agree or attract? A Relativized Minimality
solution to a Proper Binding Condition puzzle''), Boeckx examines the
'how likely' paradigm starting with the contrast (=5 in Boeckx)
originally reported in Kroch and Joshi (1985):
1 a. How likely to win is John? but
b. *How likely to be a riot is there?
This contrast has been recently taken as an argument for feature-
movement (Lasnik, 2002). However, Boeckx shows that Lasnik's analysis
is inadequate in a number of ways, both empirically (further facts
going against Lasnik's interpretation are provided) and
conceptually. In contrast to previous solutions to the puzzle, Boeckx
analysis shows that the contrast has nothing to do with remnant
movement or the Proper Binding Condition (''traces must be
bound''). In other words, cases like (1b) should be ruled out prior to
the application of remnant movement. An alternative analysis deeply
implicates Relativized Minimality; in particular, since 'how likely'
is docomposable into a wh-part and indefinite part, it is the latter
that creates an intervention and thus blocks the Agree between T and
NP in (1b). By contrast, the cases like (1a) are rescued by the
presence of a D-feature on the noun phrase being attracted, which
allows Agree to succeed. In general, Boeckx' analysis captures the
basic 'how-likely' paradigm without any appeal to remnant movement,
distinct LF- component, or move-F, and ''[i]n so doing, [it] (...)
lends credence to the conceptually more elegant mechanism of Agree and
the One-cycle model of syntax (p. 58).
Fanselow and Cavar's contribution (''Distributed deletion'') addresses
the problem of NP- and PP-split constructions in Slavic and German,
such as e.g. German. The following example is (5) in F&C): 2
Interessante Buecher hat sie mir keine aus Indien empfohlen
interesting books has she me none from India recommended
''She has not recommended any interesting books from India to me''
It is shown that XP-splits arise in the context of operator movement
only; they can retain (Pull splits) or invert (Inverted splits) the
order of the elements found in the continuous counterpart. The authors
convincingly show that previous analyses (simple movement and base
generation theories, first of all) face serious problems when applied
to these constructions. On the one hand, the former approach cannot
account for the (possible) repetition of phonetic material
(e.g. prepositions) in imperfect splits and for the violation of
standard islands for movement (e.g. PP-islands); on the other hand,
the latter approach assuming base-generation of both parts in situ
fail (among other things) to adequately describe some linear order
facts that in fact call for the movement approach. The paradox is
given solution within the copy and deletion approach to movement (see
e.g. Chomsky 1995). The basic insight of F&C is that ''deletion may
affect BOTH the upstairs AND the downstairs copy, but in a partial way
so, which yields the split XP construction''. Such an approach seems
to be a kind of compromise between simple movement and double
base-generation and provides a unified analysis for both NP and PP
splits.
Frank, Hagstrom and Vijay-Shanker's article (''Roots, constituents and
c-command'') addresses the problem of adequate syntactic
representation and the primitive notions of such representation. The
authors argue for the non-existence of primitive dominance
(''dominance does not figure into grammatical explanation'' p.110);
what they propose instead is the primitive status of c- command. They
show how such notions as roots and constituents that have been
traditionally defined with reference to dominance can be redefined
with reference to c-command only. Such an approach leads to the
distinction of ''the categorial root'' (the node that determines the
category of the tree as a whole) and ''the attachment root'' (the node
that provides the locus of cyclic attachment). With respect to the
notion of constituency much effort is made in order to retain the
validity of the claim that ''All and only constituents can be moved''
(e.g. p. 123). Again, constituency is re-defined without any reference
to dominance. The basic idea of the article may be indeed very
influential for the architecture of syntactic representation; it is
shown that the approach adopted in the paper makes it possible to
capture many generalisations with respect to such phenomena as
movement, ellipsis and conjunction in an elegant and economic fashion.
Finally, it is suggested as a theoretical possibility that ''perhaps
constituency is verified at and useful for PF, while syntax (and even
perhaps LF) are not sensitive to issues of constituency'' (p. 133).
Kural's contribution to the volume (''A four-way classification of
monadic verbs'') deals with the problems that emerge from the more or
less commonly accepted view according to which such verbs can be
divided into unaccusative and unergative verbs, as originally proposed
by Perlmutter (1978). Further research (most notably Burzio 1986) has
shown that unaccusative verbs are generated as internal arguments
inside the VP, while unergative verbs are generated as external
arguments outside the VP; besides, a bunch of tests associated with
this distinction has been proposed in previous studies. In his paper,
Kural takes up some seven of these tests, namely, there-insertion,
locative inversion, subject case (ergative vs. nominative), agreement,
and the possibility of cognate objects, resultatives and
way-construction. It is argued that these tests ''do not all test the
same structural properties'', and that ''the discrepancies in the
behavior of some monadic verbs across these tests can be explained
naturally by positing a four-way classification'' (p. 142). A distinct
syntactic structure is proposed for each type of monadic verbs. While
verbs of being and the change of location verbs have the same VP
design as classic unaccusative and unergative verbs, respectively, the
other two types of monadic verbs distinguished in the paper (namely,
the change of state verbs and verbs of creation) are endowed with a
complex multi-layered structure each involving an extra node
(directional phrase and CAUSE phrase respectively). It is finally
shown that the differences in syntactic structures typical of these
verbs are tightly associated with their broad semantic properties.
In the following paper (''On agreement. Locality and feature
variation'', by Luis Lopez) a new approach to the operations Agree and
Move is offered, that makes it possible to get rid of some unnecessary
assumptions proposed earlier (e.g. ''freezing effects'' and Global
Economy framework, that allows operations that can ''look ahead'' at
future derivational steps). Instead, the principle of Locality of
Agreement is maintained, that limits Agree to elements not separated
by a maximal category. Quite traditionally, Agree is understood as an
operation that co-values two sets of features. However, Lopez admits
the possibility of Agree between two terms with unvalued features:
''[i]f both probe and goal have unvalued features of the same type,
they will remain unvalued, with a twist: since they are now involved
in the Agree relation, these features will be co-valued'' (p. 172).
Contrarily to Chomsky (1999) and in agreement with the Locality of
Agreement principle, movement is understood as a consequence of the
''tension'' in the system created by the inability of Agree between
two items that are too distant for probing. These theoretical
assumptions are successfully used in the analyses of structural case
assignment, expletive constructions (a sketchy typology of those is
provided) and movement chains. It is shown that feature co-valuation
underlies these phenomena and that ultimately Case assignment is only
a variant of the agreement relation.
Mateu and Rigau (''A minimalist account of conflation processes.
Parametric variation at the lexicon syntax interface'') undertake a
minimalist analysis of conflation processes; they take Talmy's (1985)
original analysis as a starting point of their investigation and
acknowledge its descriptive adequacy. In particular, they acknowledge
the fact that various languages may conflate semantic components like
Figure, Motion, Path, Manner, or Cause into the verb in different
ways. Thus, for instance, ''conflation of motion with path is argued
to be typical of Romances [sic!] languages (...), whereas conflation
of motion with manner is typical of English'' (p. 211). Mateu and
Rigau's insight is to give these facts a basically syntactic
explanation within a minimalist framework. For that purpose, they
introduce a number of phonologically null verbs such as GO, CAUSE
etc. that must be conflated (by means of simple Merge) with some other
element with phonological properties. Conflation of a full verb
(e.g. 'dance') with one or another phonologically null verb will lead
to the different syntactic structures that in their turn correspond to
some diagnostic constructions analyzed by Talmy. While the former
verb expresses the manner component, the latter verb determines the
cause/motion/state meaning of the construction. As a consequence, the
cross-linguistic variation with respect to the conflation processes as
analyzed by Talmy, is reinterpreted by Mateu and Rigau as an
epiphenomenon of (un)availability of the relevant empty heads
(CAUSE/GO/BE). The various syntactic phenomena involving path
constructions and existential locative constructions are explained
within this approach.
Romero (''Morphological constraints on syntactic derivations'')
addresses once again the problem of the sources of cross- linguistic
variation. He argues that languages differ in the formal features they
encode, although there is a universal set of features available for
the faculty of language. This selection is idiosyncratic, although not
completely arbitrary. The empirical data Romero discusses include the
cases when Person-Case Constraint (PCC) is at work; this constraint
states that dative arguments are ungrammatical if the accusative is
first or second person, and agreement is overtly realized. Thus,
Romero argues that PCC shows up only if there are agreement features
involved. Romero's analyses of particular linguistic data aim at
getting rid of ''looking-forward'' properties triggering computations,
which is a desirable way of thinking for any minimalist study, since
''[i]f being (un)interpretable is an interface property, then it
cannot be relevant troughout [sic] the derivation'' (p. 255).
In the last contribution to the volume (Sabel, ''Intermediate traces,
reconstruction and locality effects''), a Constraint on Adjunction
Movement (CAM) is introduced, according to which ''[m]ovement may not
proceed via intermediate adjunction'' (p. 260). To put it in a
somewhat different way, this constraint suggests that whenever an
element is moved to an adjoined position this element cannot move
further and is ''frozen in place'' in this position. Or, to rephrase
the constraint once again, movement can only proceed via specifier
position. The article is organized in such a way that first, a bunch
of phenomena are introduced that have been traditionally analyzed as
cases of successive-cyclic movement via intermediate adjunction.
These phenomena include weak crossover effects, locality phenomena and
reconstruction properties of moved elements with respect to scope and
binding properties. However, the in the following section the author
argues that if ''the intermediate adjunction hypothesis holds, several
ad hoc devices are needed to constraint the unrestricted use of this
mechanism'' (p. 269) and that these devices are not independently
motivated. Thus, an alternative interpretation for various movement
types are offered, including wh-movement, quantifier raising,
scrambling, A-movement, empty operator movement, and head
movement. Some cases that have been traditionally used as an empirical
basis for intermediate adjunction are reanalyzed with the help of
multiple specifier positions.
EVALUATION
Overall, this volume is a valuable contribution to the study of the
architecture of Universal Grammar (UG) and of the sources of
cross-linguistic variation. Taken together, the contributions to the
volume cover a wide range of theoretical problems (probably, at the
expense of evident unity between the papers). Some of the papers
provide very interesting insights in the study of particular syntactic
phenomena (e.g. Arad, Kural, Mateu and Rigau) drawing attention to the
previously unnoticed regularities in natural languages. The stress of
some other papers is on the theoretical rather than empirical side;
e.g. the papers by Boeckx, Frank & Vijay-Shanker, Lopez and Sabel
address the issues that are very relevant for the development of the
minimalist program; their theoretical claims may have a very
far-reaching impact on the basic matters of the UG and can potentially
appear to be very influential for the future research. This
distinction is not meant to imply that theoretical issues in the
contributions are kept apart from discovering challenging empirical
regularities (a weighed balance between the two sides can be found in
e.g. Fanselow and Cavar's contribution). Although there is a lot of
controversy between the implications of particular papers
(e.g. Sabel's contribution employs ''freezing effects'' in order to
maintain his Constraint on Adjunction Movement, while one of the aims
of Lopez's analysis consists in getting rid of these effects as they
are considered to be unwelcome for the general ideology of the
minimalist program), the volume is significant and useful in a variety
of ways.
My main reservation with respect to the volume is two-fold. First, I
am not quite sure that the title of the volume (''Theoretical
approaches to universals'') is fitting for its content. Indeed, the
very word ''universal'' (as a noun) does not appear in the majority of
papers; moreover, the question of the universality of particular
phenomena is not directly addressed to in some of them,
either. Rather, it is the problem of cross- linguistic variation that
is discussed explicitly in some papers (Arad, Romero) and it is only
through the analysis of particular constraints imposed upon this
variation that the problem of universality is tackled. Of course,
universality of the basic architecture of language is a prevailing
principle for any minimalist researcher, but still the title of the
volume may seem somewhat misleading.
Second, the cross-linguistic dimension of the volume is not entirely
convincing. It is explicitly stated in the introduction by the editor
that the goals that are chosen in the volume can only be reached by
comparative study of language. However, there are some papers that are
almost exclusively concerned with the data from English (e.g. Boeckx'
contribution). Many other contain basically a comparison of English
and one or two more languages (e.g. German or Spanish). In general,
the list of languages referred to in the volume does not have but some
25 entries, out of which only 10 are non-Indo-European languages and
with only 7 languages spoken outside Europe, and only 3 outside
Eurasia. In general, dealing with particular languages seems to be
somewhat unprincipled and ad hoc in many cases. Moreover, there are a
lot of inaccuracies in this respect. E.g. on page 117 we read ''(:) in
certain Slavic languages, including Bulgarian and Romanian [sic]''. It
may be further mentioned that both Bulgarian and Romanian are missing
in the language index, although there are examples from both in the
volume (besides, the example on the same page 117, and many others
e.g. on page 251 are not properly introduced, as it is not indicated
what is their language). Being a native speaker of Russian, I checked
the three passages of the volume where Russian data are discussed and
found that ALL OF THEM are not perfect (namely, the root is udiv-
rather than udivl- in examples on page 34; a diacritic is missing in
example (32) on page 225 and the unstarred examples on page 284 are at
best to be marked as questionable according to my speaker's intuition.
Of course, these critical remarks do not undermine the fact that the
book under review could be highly recommended to anyone interested in
the recent advances in the basic issues of the generative grammar.
REFERENCES
Burzio, L. 1986. Italian syntax. Dordrecht: Reidel publishing.
Chomsky, N. 1995. The minimalist program. Cambridge, MA: MIT Press.
Chomsky, N. 1999. Derivation by phase [MIT occasional papers in
linguistics 18]. Department of linguistics and philosophy, MIT.
Kroch, A. and A. Joshi. 1985. The linguistic relevance of tree-
adjoining grammar. Ms., University of Pennsylvania.
Lasnik, H. 2002. Feature movement or agreement at a distance? In:
A.Alexiadou, E. Anagnostopoulou, S. Barbiers and H.-M. Gaertner
(eds.). Dimensions of movement: From features to remnants.
Amsterdam: John Benjamins.
Perlmutter, D. 1978. Impersonal passive and the unaccusative hypothesis.
In: Proceedings of the 4th annual meeting of the Berkeley
linguistic society. UC Berkeley.
Talmy, L. 1985. Lexicalization patterns: Semantic structures in lexical
forms. In: T. Shopen (ed.). Language typology and syntactic
description. Vol. 3. Grammatical categories and the lexicon.
Cambridge: CUP.
ABOUT THE REVIEWER
Sergey Say is a post-graduate student/assistant of ILI RAN (Institute
for linguistic research of the Russian Academy of Sciences), Saint-
Petersburg. His academic interests include ellipsis, word order,
semantics and structure of the argument structure, and other
(morpho-)syntactic matters in the Russian and Baltic languages. His
basic intent is to study these phenomena in an anthropologically-
oriented, typologically-backgrounded perspective.