-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
1/34
Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=rwrd20
Download by:[123.2.15.242] Date:14 November 2015, At: 02:44
WORD
ISSN: 0043-7956 (Print) 2373-5112 (Online) Journal homepage: http://www.tandfonline.com/loi/rwrd20
Reviews
Alan R. Libert, Joseph Clancy Clements, Martha S. Ratliff, Thomas A. Lovik,Charles W. Kreidler & Jacqueline Anderson
To cite this article:Alan R. Libert, Joseph Clancy Clements, Martha S. Ratliff, Thomas A. Lovik,
Charles W. Kreidler & Jacqueline Anderson (1990) Reviews, WORD, 41:2, 223-255, DOI:10.1080/00437956.1990.11435822
To link to this article: http://dx.doi.org/10.1080/00437956.1990.11435822
Published online: 16 Jun 2015.
Submit your article to this journal
Article views: 3
View related articles
http://www.tandfonline.com/doi/mlt/10.1080/00437956.1990.11435822http://www.tandfonline.com/doi/mlt/10.1080/00437956.1990.11435822http://www.tandfonline.com/action/authorSubmission?journalCode=rwrd20&page=instructionshttp://www.tandfonline.com/action/authorSubmission?journalCode=rwrd20&page=instructionshttp://dx.doi.org/10.1080/00437956.1990.11435822http://www.tandfonline.com/action/showCitFormats?doi=10.1080/00437956.1990.11435822http://www.tandfonline.com/loi/rwrd20http://www.tandfonline.com/action/journalInformation?journalCode=rwrd20 -
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
2/34
eviews
IGOR A. MEL CUK. Dependency Syntax: Theory and Practice. Albany:
State University o New York Press, 1988.
xx
428 pp.
Reviewed by L N R LIBERT
Syntactic theories can be divided into two general types: those
which are based on phrase structure, and those based on dependency.
n North America, theories using phrase structure analysis (particularly
the various versions
o
Transformational Generative Grammar) are far
more popular and widely-known than dependency theories. Indeed
some syntax students may be completely unaware o the existence o
dependency syntax. This book may be seen as an attempt to correct the
situation, and this is a laudable goal. The field o linguistics would be
a healthier one if linguists paid more attention and gave more respect
to their colleagues working in different frameworks. An open-minded
reader can thus easily be sympathetic to Mel cuk s effort, and Depen-
dency Syntax is a generally well-done and convincing presentation
o
one brand
o
dependency analysis, namely the Meaning Text theory
(MTT).
Dependency Syntax may be worthy
o
a detailed commentary and
analysis, but in this short review I can only give a superficial overview
o
the work, in the hope that readers will become interested enough to
read and judge it for themselves. Further, I shall say little about the
MTT theory itself; rather I shall concentrate on Mel cuk s presentation
o
this theory. I shall make some general points about the book as a
whole, and then give an account o the contents o individual chapters.
One should first note that this book does not consist entirely
o
new
material; much
o it is
made up
o
revised versions
o
older papers,
some o which were written with other scholars.
There are several factors that make Dependency Syntax unneces-
223
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
3/34
224
WORD
VOLUME 41, NUMBER 2 AUGUST
1990
sarily difficult to read. Some
of
Mel'cuk's endnotes are very long,
which may make it hard to follow the arguments
of
the main text. For
example, note 5
of
Chapter 7
is
more than three pages, while note 4
of
chapter 5
is
almost five pages long, which must be some kind
of
record. On the other hand, the endnotes are cross-referenced with the
main text, which is helpful. Those readers who do not know Russian
may not be able to fully appreciate the book, as Mel'cuk takes many
of
his examples from that language. t should however be noted that
there does exist an extensive study
of
English in the MTT Framework
(Mel'cuk and Pertsov 1987), for those who would like to see this
framework applied to English. There are a small number of typograph
ical and grammatical errors, not enough to be annoying, but perhaps
enough to be noticeable.
A general problem is that one may judge all types
of
dependency
grammar on the basis
of
this book, and so perhaps not get an accurate
picture
of
dependency theory as a whole. That is, if for one reason or
another, one is not impressed by Mel'cuk's ideas, one may take on a
negative attitude towards all dependency-based frameworks, just as
one might form a prejudice against all frameworks
of
the generative
tradition, just because he found one particular framework within that
tradition (e.g., Generalized Phrase Structure Grammar) to be inade
quate. This may be an unavoidable type
of
problem for a book
of
this
sort, but it should be noted that there are many different points
of
view
under the general heading of dependency theory, and one should not
judge all
of
them on the basis
of
MTT.
Part I ( Dependency Syntax: n Overview ) may be the most
interesting and useful part
of
the book for those unfamiliar with de
pendency grammar. Here Mel'cuk gives a general picture
of
depen
dency theory, including a welcome bibliography
of
work
of
this school
(Chapter 1), and an introduction to Meaning-Text theory (Chapter 2).
The latter chapter presents some difficulties for the reader. First, this
chapter, as well as much
of
the rest
ofthe
book, seems to contain a large
number of items in lists, e.g.
A
generalized lexeme is one of the
following four items (p. 60),
A
DSyntRel [=Deep-Syntactic Re
lation] is one of the following nine binary relations'' (63), ' 'The SSyntR
of a sentence consists
of
four structures (67), The Semantic Com
ponent must be able to perform eight main operations'' (73). This can
make for tedious reading, although I do not know how one could
express the ideas in a better fashion. What are more annoying, and more
difficult to follow, are the figures illustrating the semantic, deep-syn
tactic, and surface-syntactic representations
of
sentences (e.g. Figures
2-3, p.
54).
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
4/34
REVIEWS
5
These are considerably more complex than the phrase structure
trees
of
transformational grammar. While one can not blame Me ' cuk
for the complexity
of
the representations, since they should reflect the
complex interrelations holding among linguistic constituents,
it
would
have been preferable to use simpler sentences than ' 'As the American
press says, the President believes that a duty of the people of the
United States
is
active aid to the development
of
the economy
of
African countries." Mel'cuk does realize the difficulty
of
following
such representations; as he says (53), they
may
stun a newcomer
I had the choice between using a toy SemR [ =Semantic Representa
tion] or exposing the reader to the astonishing complexity
of
the real
thing. Not without hesitation, I opted for the second alternative." I am
not sure whether the right choice was made.
With regard to the MTT framework itself, I shall just mention that
perhaps its most distinctive characteristic, aside from the fact that it
is
a dependency theory, is the fact that it
is
not a generative model, but
rather a mapping system, which links the set of texts to the set of
meanings. Thus it
is
rather different from transformational theories;
nothing
is
transformed or generated; as Mel'cuk says (45), the rules
"simply match" meanings and texts (or the intermediate levels be
tween them). For those used to some of the better known North Amer
ican schools
of
linguistics this may be a new and interesting way
of
viewing linguistic processes.
Part
(''An
important concept of Dependency Syntax: Surface
Syntactic Relations") contains a single chapter
in
which various types
of dependency relation-morphological, syntactic, semantic) are dis
cussed. It
is
important to realize that there are different sorts
of
de
pendencies, and that a dependency
of
one kind does not imply the
existence or directionality
of
dependencies
of
other kinds. This can be
seen
in
Mel'cuk's section on "Possible Combinations
of
Syntagmatic
Dependency" 118-128).
Part Ill begins the examination
of
the practice
of
dependency
theory. This part deals with the difficult problems involved in the study
of
ergative constructions (Mel'cuk rejects the term
ergativity ,
specif
ically with whether there is an ergative construction
in
Dyirbal (Chapter
4) and Lezgian (Chapter 5). A definition
of
subject is clearly crucial to
a decision on whether a construction
is
ergati ve, and so Mel 'cuk spends
some time on the characteristics
of
subjects
in
Dyirbal and Lezgian. His
conclusion
is
that neither language has an ergative construction: the GS
[Grammatical Subject]
of
what had been considered ergative construc
tions
is
a semantic object and not a semantic subject (Mel'cuk's def-
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
5/34
226
WORD,
VOLUM
41, NUMBER 2 AUGUST 1990
mtion
of ergative construction
is ' 'a non-nominative construction
whose GS refers to the 'semantic
subject '),
while Lezgian simply
lacks transitive verbs (and thus can not have an ergative construction).
These conclusions may still be open to debate, but Mel'cuk has pre
sented good arguments. One is made aware of the importance
of
def
initions of such notions as subject; discussion of ergative constructions
is meaningless until one knows what a subject is. The third and final
chapter of Part Ill is not concerned with a particular language, but is
concerned with defining
ergative construction.
The definition is basi
cally the same as that given before,
in
Chapter 4, but this chapter is of
value because Mel'cuk gives a "typology of predicative construc
tions'',
where these constructions are classified on the basis
of
relations
among three levels, namely semantics, (surface) syntax, and (deep)
morphology. This three-way correlation is significant, because while it
is common to analyze any construction in which the grammatical subject
does not correspond to either the semantic subject or the nominative NP
of a sentence, Me ' cuk reserves the term ergative construction for those
constructions in which grammatical subjects and the semantic subjects
correspond, but do not bear nominative case. Thus Me 'cuk' s definition
is
narrower than those
of
some other linguists.
The two chapters
of
Part IV ("Syntactic Description: Surface
Syntactic Models and Notions") deal with two languages, Alutor
(Chapter 7), and Russian (Chapter 8), with reference to several sig
nificant features which they possess. The former chapter is of value
because so little material is available on the Alutor language, which is
related to the better known Chuckchee. In this chapter we see several
of the surface-syntactic rules posited for Alutor. Some of these rules
describe unusual kinds of agreement present in this language; a matrix
verb can agree with the subject or direct object
of
its subordinate
clause,
or
with the clause as a whole. The next chapter discusses a less
exotic phenomenon, the phonetically null elements which may be pos
ited
in
Russian sentences such as those below (of course similar sen
tences exist in other languages):
(la)
(lb)
Kolya v sosednej komnate
Ulicu zasypali peskom
'Kolya is in the next room.'
'The street was strewn with
sand
(by somebody).
One must be able to argue that these sentences do in fact have
"syntactic zeroes , and Mel'cuk does this for the type of sentence
exemplified by (lb), although I do not know whether I am convinced.
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
6/34
REVIEWS
227
There are two types
of
syntactic zeroes, the zero lex (as
in
la) and the
zero lexeme (as
in
lb). Mel'cuk stresses the distinction between these
zeroes, which
"EXIST
in language and as such are stored in the minds
of speakers" (linguistic
1
zeroes) and "descriptive devices introduced
by the researcher
in
order to make this description look more homo-
geneous, more compact or more elegant" (linguistic
2
zeroes). I am not
sure what he has in mind for the latter class; some examples would
clarify this. I suspect that the PRO pro and traces of Government-
Binding theory may be among Mel'cuk's linguistic
2
zeroes; however,
most researchers working
in
that framework would deny that these
categories have been posited merely to make their descriptions neater.
The last part of Dependency Syntax ("Syntactic Methodology:
Some Thorny Questions of Russian Syntax") consists of three chapters
which attack the problems involved in
making a choice among several
descriptions
of
some construction. Chapter 9 takes
up
the problem
of
how to describe an intriguing class of construction, those where a
"verb of emotion" (i.e. not a "normal verb of speech") introduces a
direct quote, as illustrated
in
(2):
(2) "Ostav'te
menja "-ispugalsja
bufetCik
'"Leave
me
alone "-became
frightened the bartender.'
The questions raised by Mel'cuk are what the (syntactic) relation
is
between the matrix emotion verb and the quote clause and which
constituent of such sentences conveys the idea "utter the given utter-
ance''.
He chooses a ''syntactic'', rather than a ''lexical'' solution, and
so claims that the syntactic relation between verbs of emotions and the
quotations they introduce is not the same as that between verbs such as
say
and the quotes which they introduce. The idea
of
uttering
is
con-
veyed by the
''the
syntactic structure itself ''
in
sentences with the verbs
of
emotion.
In Chapter
1
Me ' cuk argues for the addition of animacy to the
list of inflectional categories relevant for cardinal numerals in Russian,
to account for following type of differences
in
adjectival endings:
3a)
3b)
4a)
4b)
Ja vizu cetyre sosny
Ja vizu cetyrex devusek
Ja vizu krasivyj dom
Ja vizu krasivogo junosu
'I
see four pine trees.'
'I see four girls'
'I see a beautiful house'
'I see a handsome youth'
It
is
surprising that this problem
is
mentioned as little as Mel'cuk
claims; he cites (372) two ''basic reference works on Russian'' in
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
7/34
228 WORD, VOLUM 41, NUMBER 2 AUGUST
1990
which this problem is not brought up . In any case, his solution
appears reasonable, and he provides a list
of
seven arguments for
choosing it over the alternative solution
of
positing another case (the
animate-accusative
case).
The final chapter is also concerned with numerals, specifically
with the question
of
what case
is
borne by certain numerical con
structions (containing animate nouns) occurring after the Russian prep
ositions: v 'in', na 'on', az 'behind', cerez 'through, behind',
po
'each' when these prepositions are used
in
a quantitative sense, as
in
siloj rovno v tri medvedja 'with the power of exactly three bears'. It
has been said that such expressions bear accusative case, but this
answer is not satisfactory because such numerical constructions have a
different form when they occur
in
standard accusative contexts , i.e.
direct object position (e.g. the form required is trex medvedej rather
than tri medvedja . Mel'cuk sets up the choice between the solutions
of claiming that these constructions are actually
in
the nominative case
(what one might see as the obvious solution), and maintaining that they
are accusative, but do not take the expected form because
of animacy
non-agreement between the numeral and the animate noun. Mel'cuk
chooses the second solution. giving four arguments for it. One of his
arguments
is
that Russian prepositions do not govern the nominative in
any other situations, and thus it would violate the spirit
of
the
language to claim that these five prepositions can govern a nominative
noun when used in
a quantitative way. It must be pointed out that it
is
not completely unheard of for a preposition to govern a nominative
in
other languages, and Mel'cuk himself concedes that Russian nomina
tive NPs are not as syntactically independent as has been thought.
Although the earlier chapters of the book may be the most inter
esting for many readers, those who are interested in ergative construc
tions
or
certain aspects
of
Russian syntax will also find the later chap
ters useful. While I do not find that Dependency Syntax is as neutral a
book as Mark Aronoff seems to indicate in the preface, it is a non
polemical introduction to a theory (or set of theories) which will ben
efit those wishing to learn about alternative conceptions of syntax.
Department
o
Linguistics
McGi/1 University
Montreal, Quebec H3A JG5
Canada
REFERENCES
Mel'cuk I. A and N
V
Pertsov.
1987.
Surface Syntax
o
English. Amsterdam: John Benjamins.
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
8/34
REVIEWS
229
DANIELLE CORBIN. Morphologie Derivationelle et structuration
u
lex-
ique. 2 vols. Linguistische Arbeiten Series, Nos. 193-94. Tiibingen: Niemey
er, 1987. vxii 937 Pp.
Reviewed y
JosEPH
CLANCY CLEMENTS
It is unique
in
the current stage
of
development
of
the study
of
derivational morphology (DM) to encounter a theory based on insights
from both lexicology and lexicography as well as from current theory
in
DM. The present two-volume work is a thorough treatment
of
French DM precisely from such a perspective. One of the author s
principle objectives in this endeavor was to help remedy the lack of
detailed empirical analyses in the current literature on DM, which, she
notes, is one reason for the proliferation
of
alternative theories
in
recent years.
The firstvolume is divided into three parts. In the first, Corbin
(C)
argues against the commonly accepted methodological practices for
data collection predominant in DM today. Part is devoted to the
examination of several current frameworks and how they deal with
regularities, subregularities and irregularities in DM. Here, she also
develops what she calls a stratified, associative model and argues for
its superiority over the existing frameworks for the treatment of different
kinds and degrees
of
sub- and irregularities. In part Ill, C presents and
discusses the structure of the lexical component developed in part 11
Volume two contains the footnotes,
16
appendices consisting of data
collections and studies that substantiate arguments put forth in the first
volume along with very helpful subject, author and word indices.
In
part I, C presents solid arguments against the methological
practices generally employed by morphologists today. With data col
lected on the adverb-forming suffix ment
-ly
from
15
French mono
lingual dictionaries (found in appendices
of
volume two), she argues
that the dictionaries consulted are inconsistent
in
the ordering
of
base
and derived words (DWs) with respect to one another (macrostructural
ordering) as well as in the ordering of the different meanings of a given
DW from the most regular to the most idiosyncratic (microstructural
ordering). Although
C s
study is most thorough, the same should be
carried out for various other affixes of different degrees of frequency
and productivity to ascertain whether her findings hold there as well.
Moreover, even though the same situation were to obtain for dictio
naries in other languages, it seems dangerous to overgeneralize that the
situation
of
French lexica reflects that
of
dictionaries in any language,
which seems to be what C implies.
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
9/34
230
WORD, VOLUME
41. NUMBER 2 AUGUST
1990
In dealing with dictionaries, C draws a useful distinction between
the lexicon o the linguist as opposed to that o the lexicographer.
While the lexicon
o
the latter is present
in
all dictionaries and is the
less restricted o the two in the sense that it contains extra-linguistic
as well as linguistic information, the linguist s lexicon
is
less restricted
in another sense: It must remain unconstrained by anything that is
not strictly linguistic. C also discards the ambiguous notion
o pro
ductivity and suggests availability (my translation o Fr. disponibilite
to denote the capacity o an affix to be used
in
coining unattested
DWs,
andfrequency
(Fr.
rentabilite
to denote the number
o
attested
DWs
a
given affix is involved in forming. Her distinction is reminis
cent
o
Aronoff s (1983) differentiation between
productivity
n fre-
quency.
C also scrutinizes the notion o native speaker intuition, noting
that such
an intuition is actually based on three types o metalinguistic
competence: a) sense o newness o a DW, b) metalinguistic awareness
and c) acceptability judgments. It is not exactly clear why C differen
tiates between the first and second types since she herself (58) con
fesses that they are one and the same. Types a) and b presuppose, she
states, theoretical impossibilities:
I f
there is an intuition o what is or
is
not a neologism, it implies that all speakers have the same lexical
pool from which to draw, which is at odds with the accepted fact that
no one knows the same set o words. Type c), she says, suffers from
a triple ambiguity in that an asterisk marking a DW as ill-formed
can be interpreted either that i it is not attested, ii) it is ill-formed
according to acceptability judgments or iii) according to the word
formation rules (WFRs). The first two o these are cases o accidental
gaps in the lexicon. E.g. the fact thatferroviaire relative to railroads
is normally used instead o ferrovial might say something about the
availability
o
the suffix
-al
to derive new forms, but there are no
linguistic constraints systematically blocking its existence. iii) repre
sents a case
o
a systematic gap. For instance, DWs such as
*demaison
dehouse (N) are impossible because part o the corresponding WFR,
i.e. the category change (N > V), is violated. Systematic vs. acci
dental gaps
in
the body o attested DWs reflect what C refers to as
derivational vs. conventional lexical knowledge. While both are part o
our lexical competence, i.e. the sum o knowledge regarding the lex
icon, the former tells us what is linguistically well/ill-formed while the
latter allows
us
to judge what is otherwise acceptable
or
unacceptable.
She concludes, then, that metalinguistic intuition is necessary but not
sufficient as a source o data for the morphologist. She warns that great
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
10/34
REVIEWS
231
care must be taken to systematically collect a representative inventory
of
data from as many sources as possible, a practice, she claims, that
has not been strictly followed in the discipline.
C categorizes her model as stratified and associative. In part 11
she carefully examines the stratified and non-stratified models in lex
icalist morphology. Her main claim is that non-stratified models, such
as those of ackendoff (197 5), Aronoff (197 6), Booij (1977), Lieber
( 1981) and Selkirk ( 1982), have a lexical component consisting of
essentially only two sub-components: one list of entries where irreg
ularities are listed and one list of rules and/or principles that account
for the regularities. In her stratified model, subregularities, i.e. ones
that cannot be derived by derivational competence alone, are not lo
cated on the same level as regularities, but are subordinate to these.
Moreover, her model deals with the possible as well as the attested
lexicon and can neatly account for this difference in terms
of
system
atic vs. accidental gaps in the lexicon. Unlike the disassociative frame
works which, she claims, separate the treatment of semantic and mor
phological regularities, in C s associative model these treatments are
inseparably linked-an approach which I find to be similar to Zwicky s
( 1987) rule-to-rule relation between morphology and semantic rules.
However, In C s theory, the relation between WFRs, affixal lexical
entries and their respective semantic rules
of
interpretation exhibits a
number of innovative aspects. To understand these, a brief sketch of
her model is necessary.
The Base component contains underived words and affixes. Af
fixes, marked as [Affix] with their own lexical entries that do not
include lexical category specification, are independent of word forma
tion rules (WFRs) but linked to these through certain morphological
operations. One key novelty here
is
that the general notion
of
subcat
egorization (SUBCAT) frames has been done away with. On this view,
a WFR, such as A-> V with the corresponding semantic interpretation
make A , may be used for the prefix en- as
in
riche rich -> enrichir
enrich (make rich) , as well as for the suffix
-iser
as in
industrial
industrial -> industrialiser industrialize (make industrial ) (with
proper allomorphy). Idiosyncratic properties of the affixes are also
posited in the lexical entry. For instance, the suffix -ite -ity , associ
ated with the WFR A
->
N with the interpretation quality/character
istic
of
A ,
is
very restricted as to the bases with which it may combine:
it attaches only to bases containing the suffixes -able -aire -al. -el
-ique. In the Base component, C also distinguishes between complex
underived and complex derived words. The former are words with a
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
11/34
232
WORD,
VOLUME 41, NUMBER 2 AUGUST
1990
recognizable internal structure, of which there are two types: those
whose base is recognizable, as in royaume kingdom from roi king ,
and those whose affix is recognizable, as in carpette small rug , where
the base carpe rug does not exist. Complex DWs, as in chanteur
singer , are not listed in the lexicon and could not be listed since they
are, according to C, theoretically infinite
in
number. A contradiction
emerges here in the interaction of the different components. C states
that the WFRs and all possible DWs are found in the Derivational
Component, the only component with generative capacity. However,
as just noted, the possible DWs are not and, furthermore, cannot be
listed at all. Not listing at least the attested DWs also poses the tech
nical difficulty concerning how they can be used as inputs for the
WFRs to derive further DWs.
In C s
model, the weight
of
the semantic operation falls on the
WFRs and not on the affixes. The structure of a WFR
is
expressed in
the equation WFR = nWSCR + SSCR + MP + CSC + LSI. The
WFR for fertilisation fertilization would be as follows: the Word
Structure Construction Rule (WSCR) is V > N. It corresponds to only
one Semantic Structure Construction Rule (SSCR), which here
is
ac
tion or result of the action of V , and to only one Morphological Para
digm (MP), which
is
a list of all possible affixes/conversions corre
sponding to the WSCR and SSCR in question. These are: conversion
[ + masc] [-A]; conversion
[+fern]
[+A]; -ade; -age; -erie; -ment;
-tion; -ure, etc.
A
= available for forming new DWs). A set of
Categorial Semantic Constraints (CSCs) that restrict the general type
of
base to which the affixes in the MP may attach also form part of the
WFR. For example, the WFR at issue only applies to nonstative Vs.
Other restrictions, such as the constraints on the phonological form
of
the base to which a certain affix may attach, form part of the lexical
entries for affixes. The Lexical Selection and Insertion (LSI) mecha
nism, which handles proper insertion,
is
made sensitive to these con
straints.
Leaving aside the allomorphic specifications, the whole deriva
tional operation for fertilisation would take place, then, in the follow
ing manner. The lexical itemfertiliser fertilize would be inserted into
the structure N
>
[[X]v Y)ar1N whereby
it
would be checked whether
the categorial and semantic constraints associated with the WFR are
met. This operation creates the structure [[[fertile]A (is)]y
Y)ar1N
From among the affixes in the MP
of
the WFR only -tion may be
selected because the form of the base only allows the insertion of this
particular affix. This operation yields [[[fertile]A (is)]v tion)ar1N The
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
12/34
REVIEWS
233
DW is then associated with the meaning action and result of the action
of fertiliser and finally the necessary diacritic features, such as the
feature
[+fern],
are projected onto the DW.
This approach has numerous interesting though potentially dis
turbing implications. First, as noted above, the typical concept of
SUBCAT frames for affixes is done away with, the category change
being taken care
of
by the WFR with which a given affix
is
linked.
This is an unusual move since this notion is central in virtually all
models of grammar and all-pervasive in frameworks such as HPSG.
Second, by labeling affixes as [Affix] without lexical category marking
and associating them to a specific WFR as it is conceived of by C, the
notion of head in the DW becomes superfluous. At this stage of lin
guistic theory, where the concept of head is becoming one of the main
links between certain components
of
grammar, the idea
of
discarding
it seems highly undesirable.
Also of note in C s model is that if an affix has several different
meanings, as many corresponding affixes with the same phonological
representation are posited. C thus prefers a homonymous system where
different affixes may have the same form but only one meaning rather
than a polysemic approach in which one affix may have several dif
ferent meanings and only one form. The former approach, she claims,
aids in making the model more constrained. Finally,
C s
model lends
a new twist to the problem
of
bracketing paradoxes. In the associative
model she presents, it is implied that any and every bracketing com
bination for a given DW is theoretically possible as long as it is
permitted by the corresponding the WFRs. Unfortunately, she does not
address this issue at all.
Although the theory C proposes is generally very coherent, there
appear to be several minor inconsistencies
in
her model. The manner
in which she handles the commonly accepted distinction between trans
parent and non-transparent affixes is confusing. She states that only one
semantic operation (SO) corresponds to a given WFR.
n
her view, any
given SO involves three elements: I. a parasynthetic operation,
whereby a thematic role is added upon category change, 2. a lexical
category change operation and 3. a lexical operation. First, the oper
ation in
l. is
considered obligatory and as such must apply both to
derivations like chanteur singer
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
13/34
234
WORD,
VOLUM
41, NUMBER 2 AUGUST 1990
some affixes stand for words as in
-erie
'place',
-ette
'small'. Operation
3. is claimed to be an addition to certain SOs, resulting in a more precise
characterization
of
them. While C
(263-66)
states that these three
operations may appear together, in hierarchical order, associated with
one WFR, it seems that operations 2. and 3. should be mutually
exclusive. It is also unclear exactly how these operations correspond to
DWs. Their application needs to be more constrained. Moreover, the
examples which are used to illustrate the point could be more diverse.
For idiosyncratic information of all kinds, C creates
n
her model
the interesting concept of a pragmatic component, called the Conven
tional Component (CC), in which are found allomorphy rules, an
Idiosyncracy Applier (lA) (Fr.
Applicateur d ldiosyncracies),
de
letion rules, minor semantic rules, a
Selector
(Fr. Selectionneur)
and the conventional (i.e. attested) lexicon; in short, all the material
that a speaker would store n
memory and not be able to derive. The
role
of
the lA is to assign (a) feature(s) to certain DWs in order to
sensitize them to certain minor semantic rules. C is not explicit con
cerning just how the lA knows which DWs to mark. It seems that this
information could simply be included
n
the lexical entry of the affixes.
Nor is it entirely transparent how the Selector operates. From the body
of
all possible DWs, which are not listed, the function
of
the Selector
s to mark with the feature [+Attested] that subgroup of the possible
lexicon which s attested. This s to show the actual state of the lexicon
of
a language at a given moment. The difficulty here appears
to
be that
the possible DWs are not listed and the Selector has again no way
of
knowing which to mark or not to mark. Moreover, the attested lexicon
s already listed
n
the CC. It seems, then, that the content and function
of the semantic operation as well as the functions of the lA and the
Selector need to be reelaborated. With respect to these last two items,
they may well be expendable.
These reservations notwithstanding, this carefully edited and well
written volume s a stimulating and impressive piece of work with a
solid empirical foundation and a healthy, eclectic perspective. It is not
only of import for many issues currently being debated
n
DM but it
offers challanging and penetrating ideas for their development and
eventual solution.
It s
worth the time it takes to read it.
Dept.
o
Spanish Portugese
Indiana University
Ballantine Hall
8/oomington, lnd. 47405
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
14/34
REVIEWS
235
REFERENCES
Aronoff.
M
1983. Potential words, actual words, productivity, and frequency.
Proceedings of
the Thirteenth International Congress
of
Linguists
163-71.
1976 ;
Word formation n generative grammar.
Cambridge:MIT Press.
Booij, G.E. 1977. Dutch morphology. A study of word formation n generative grammar.
Lisse:Peter de Ridder Press.
Jackendoff, R 1975. Regularites morphologiques et semantiques dans
le
lexique , French
translation. Ronat.
M
ed.,
1977. Langue. Theorie genenerative erendue.
Paris:Hermann.
pp
Lieber.
R 1981. On the organization
of
the lexicon.
MIT doctoral dissertation
1980;
reproduced
by
the Indiana University Linguistics Club.
Selkirk
1982.
The syntax
of
words.
Cambridge: MIT Press.
Zwicky,
A
1987.
Transformational grammarians and their
ilk . MIT Working Papers n
Linguistics 25:265-79.
PAOLO RAMAT. Linguistic Typology. Berlin: Mouton de Gruyter, 1987.
Reviewed
y MARTHA
S. RATLIFF
Linguistic Typology is a very welcome English translation and
expansion of a collection of ten articles by the Italian linguist Paolo
Ramat. Chapter 9 is new to this volume. The other nine chapters are
revisions of articles previously published
in
the 70s and 80s, many of
which are
in
a language (Italian, German, French) or in a source out of
reach of potential readers. The essays have been revised with the
coherence of the whole in mind: the result is a clear delineation of
R'
s
theoretical position and practice with regard to typological research.
The book is divided into three parts: theoretical, problematic (Ro
mance and Germanic), and historiographical (Humboldt's typology).
Although the layout is logical, readers may find it most profitable to
read the essays of the second section on Romance and Germanic type
and type change first, to become acquainted with R' s research inter
ests, before reading sections one and three. The translation by A. P
Baldry is to be commended: the simplicity and clarity of R
s
style
is
well preserved and the English seems homegrown.
In the theoretical section, and indeed throughout the book, the
same major themes recur:
I) typological research is inductive and probabilistic, and, like
research into language change, can only suggest what
is
likely rather
than assert what must be so (for example, 35, 159-61);
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
15/34
236 WORD
VOLUME 41, NUMBER 2 AUGUST 1990
2
nevertheless, typological research rests on theoretical assump
tions that shape the inquiry. There
is
a necessary dialectic between
inductive and deductive methodologies. The concept
of type
itself
is not a directly perceivable reality in any language, but an abstract
model for the description and explanation
of
language phenomena''
(24);
3) At its best, typological research exposes networks
of
interre
lated strategies for communication that a given set
of
languages em
ploys-a functional orientation (for example 11-19).
In
the first chapter,
The
problems
of
linguistic typology , R
presents a useful review
of
the definitions
of
typology by modern
European scholars, and follows with an elaboration
of
the points pre
sented in simplified form above. He then discusses the relationship
between the study
of
language type and language universals, the rela
tionship between typology and language change, the explanation of the
existence
of
types, and the classification and quantification
of
typo
logical data in turn.
In
section 1.5.1 R briefly presents certain
pro
posals for language types'' that are consistent with the approach he
suggests (see point three above): Klimov's contentive typology ,
Greenberg's word order typology, and the proposals
of
the Cologne
research project on universals and typology
25-6).
This section de
serves expansion into an essay
of
its own: besides being on the ''right
track , how well do the authors
of
these typologies balance inductive
and deductive methodologies? How insightful and far-reaching are the
networks they reveal? Are they clear in illustrating the limitations
of
their classifications? In general, what kind of evaluation measure are
we to use
in
judging typological statements about
''the
organizational
principles
of
linguistic
data''?
In
Universals and typology , R describes the different ways the
term ''universal'' has been used: ''essential universals' ' (after Coseriu)
proceed from a definition of language (such as the property of linearity)
and are arrived at deductively, objective universals are discovered
by empirical research and are based on physiological or psychological/
communicative constraints shared by all (such as the markedness
of
OVS order), and subjective universals are those R considers lin
guists' analyses raised to the status
of
reality (such
asS->
NP VP, where
a particular linear order
is
given preference at the universal level ).
As
to the relationship:
The
purpose
of
typology
is
to create a
universally valid means
of
describing languages which will necessarily
be based on the essential, constituting (=universal) properties. For the
creation
of
an effective model of typological research, such concepts as
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
16/34
REVIEWS
237
'sentence' (and 'operation' 'grammar', and even 'structure',
'opposition', and so on are
relevant-and
these concepts do belong to
the domain of universals. (42).
The next chapter, The typological level: predicates and argu-
ments follows smoothly from the preceding one (and from the intro-
duction
of
this concept of deep structure given in 1.2.2). At the deepest
level common to all languages certain communicative functions lie
(universals-of
the
objective
type). The typological level is inter-
mediate between this level and the level of the manifestation of these
functions in any particular language. For example, all languages are
able to express the deictic function, but only a particular type will have
a cateogry of deictics (64). R criticizes modern treatments of the pas-
sive, in which the passive is derived from the active, in this light: at the
level of predicate and argument they are the same, and empirical study
demonstrates that one
is
not necessarily more basic than the other.
n
Crisis in formalism? Theory
of
grammar and empirical
data ,
R sets forth a description of the
' 'two
paradigms'' of linguistic research
that have separated linguists counterproductively into opposing camps:
''One formal paradigm considers a language (=a grammar) as an
abstract object, so that grammar is consequently seen as a set
of
formal
rules
of
syntax to be applied regardless
of
the possible meanings and
the possible uses of the syntactic structures described. A second takes
language primarily as an instrument of social interaction which estab-
lishes a communicative relationship within a society.'' (71) There is a
concomitant methodological difference: The formal paradigm is
(mainly) deductive, the functional paradigm is (mainly) inductive.
(72). Among many others, the former is exemplified by Hjelmslev and
Chomsky (in terms of metholodology), the latter by Bloomfield and
Dik. The challenge is to find a convergence
of
these two paradigms to
yield a more integrated theory
of
language 96). R sees hope in the
recent broadening of the generative model to account for more lan-
guages than English and in the introduction
of
the notion
of
param-
eter which in practice [accounts] for the existence of phenomena
that lead to the divergences that typological studies feed on. 96)
Nonetheless, the bulk
of
the chapter is critical
of
various generative
analyses which do not hold
up
to confirmation by empirical data from
a wide range of languages: two areas of discussion that are given
consideration at length are the representation
of
word order in grammar
which grows, R claims, from the inclusion
of
categorical notions in the
deep structure, implying universal validity for them 76-79) and gap-
ping phenomena in different languages, the causes for which include
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
17/34
238
WORD, VOLUME 41, NUMBER 2 AUGUST
1990
psycholinguistic as well as syntactic strategies in a complex and often
contradictory mix of
principles (80-88). Again, in place
of
these
analyses, R advocates an approach which studies the strategies differ
ent groups of languages employ to accomplish the same communica
tive task (Chapter 3).
The second section contains the following five articles: Toward
a typology
of
Common
Germanic ,
The birth
of
new morphological
categories: the case
of
the article and relative pronoun in Germanic
languages'', ''Towards a typology of Pompeian
Latin'',
' 'An example
of reanalysis: periphrastic forms in the Romance languages' verb
system ,
and Sentence Negation in Germanic and Romance lan
guages''.
I found two chapters of special interest in this section. R' s re
working
of
his oft-cited
1st das Germanische eine SOV-Sprache? ,
here Toward a typology of Common Germanic , effectively refutes
Vennemann's analysis of the cause for the change from SOY to SVO,
which involves topicalization of 0 and the subsequent movement of V
to second position to disambiguate the two NPs. R proposes instead
that the V2 position, already a tendency in Indo-European ( Wacker
nagel's
Law ),
was generalized (albeit inconsistently) in Germanic.
Sentence Negation in Germanic and Romance languages , a
joint project with linguists Molinelli and Bernini, examines type (es
pecially word order type) as one factor in the curiously parallel devel
opment of the position of the negative from preverbal to discontinuous
to postverbal (the latter two less common positions for the negative in
the world's languages) in both Romance and Germanic. A wealth of
data is presented to indicate that type may have influenced the histor
ical development of the negative in these languages, but that a possible
common origin for these two branches, language contact, and the
complex histories of individual languages in the two families have
played roles as well (the language contact situation is revealed in an
interesting map on 87 comparing both the structure of negative-verb
constructions and the lexical items involved).
The final essay
of
the collection,
The
language typology
of
Wilhelm von Humboldt'' , originally a contribution to an issue of
Lingua
e Stile
dedicated to H, seeks to correct the impression created by
Chomsky's
Cartesian Linguistics,
among other writings, that H was
primarily a Rationalist: there is no doubt that the idea of a
Universal Form or Grammar, underlying the diversity
of
various lan
guage forms and the consideration of language as creativity
(energia)
'which makes an infinite use
of
finite means' are concepts which
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
18/34
REVIEWS
239
characterize
H's
linguistics But as we have seen, behind H's
complex linguistic approach . . . lies one main goal: the study of
(historical and anthropological) causes and kinds of language differ
ences H thus wrote
Uber die Verschiedenheit des menschlichen
Sprachbaues
not
Uber die Gleichheit
and repeatedly stressed that the
'spirit of the people' is the real explanatory principle and basis for the
difference between languages
(198-99).
To demonstrate that H
was a Romantic, a product of his times,
as
much or more than a
Rationalist, R points out
Since
H agreed with Schlegel that some
languages (namely the inflectional ones) are better than others, his
typology, like Schlegel's has a finalistic teleological outlook, a must
be approach to languages, and thus becomes a critical yardstick.
( 192).
H s threefold classification of languages is based on the methods
different languages use to express unity
in
a sentence and are the ( l)
isolating, (2) incorporating, and (3) agglutinating-inflectional types
206-7).
R' s mastery
of
the literature on typology, his knowledge
of
current
theoretical debates in general linguistics both in Europe and the United
States, and his facility
in
the fields of Germanic and Romance philol
ogy are impressive. Although R s theoretical choices are evident and
his criticisms
of
generative grammar
in
Chapters 4 and
lO
are pointed,
he is not polemical, and constantly strives for synthesis and commu
nication among those who practice different brands of linguistics.
Both
[deduction and induction; theory-oriented and data-oriented]
positions are methodologically valid and necessary . . . Both derive
from a pretheoretical choice, often made on the basis of personal
inclination and, ultimately, ideological convictions, usually uncon
sciously accepted and hence not expressed explicitly-but nevertheless
still operating.''
71
). The volume will be of interest, then, to those
who wish to set a framework broad enough to encompass and relate the
variety of kinds of linguistic research that is going on today. It will be
crucial to those involved
in
typological research who need to sharpen
their understanding of the methods and goals of their research.
Dept. o English Linguistics Program
Wayne State University
Detroit M/. 48202
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
19/34
240
WORD
VOLUME 41, NUMBER 2 AUGUST
1990
ALBRECHT SCHONE, ed. Kontroversen alte und neue. Textlinguistik con-
tra Stilistik? Wortschatz und Worterbuch. Grammatische oder pragmatische
Organisation
vo
Rede?
Eds. Waiter Weis, Herbert Ernst Wiegand and
Marga Reis. Akten des VII. Internationalen Gennanisten-Kongresses Got
tingen 1985. Band 3. Tiibingen: Niemeyer, 1986. x + 403 pp. (32.50 DM.)
Reviewed by THOMAS A LovtK
This is the third
of
an 11-volume series
of
papers presented during
the 7th International Congress of Germanists held in Gottingen from
Aug.
25-31,
1985. Entitled Controversies, old and new, the par
ticipants, all Germanists trained in both literature and linguistics, ad
dressed primarily literary topics. Nonetheless, several sessions dealt
with more linguistically oriented topics, including the 45 articles on the
three topics included in this volume:
1
Textlinguistics or Stylistics? 2)
Lexicon and Dictionary, and 3) Grammatical or pragmatic organisation
of speech? Volumes 4 and 6 address language norms and dialects, and
women's language and literature, respectively.
Publishing the papers from a conference on a specialized topic,
such as this one in Gottingen, has its advantages and disadvantages.
On the one hand, the volume includes contributions by many major
researchers in the field. On the other hand, some contributions do not
always represent new ideas or research, but instead report on research
which may have appeared elsewhere in print. For the most part, how
ever, the quality of papers presented in this volume is very good and
the issues addressed are truly of a controversial nature. Since it
is
not
possible to discuss each paper in detail, some brief comments on the
general themes addressed in the presentations will have to suffice.
The session on Textlinguistics
or
Stylistics? brought together
13
specialists from the fields
of
textlinguistics, stylistics and literary
studies. The first seven papers by G. Michel, H-W Eroms, H. Aust, B.
Sandig, G. Lerchner, K. Weissenberger, and A. Obermayer all con
sider the theoretical dichotomy between textlinguistics and stylistics or
rhetorics and treat a variety
of
topics, e.g., I) textual and stylistic
norms, rules, and typologies;
2
stylistic variation, and 3) textual mod
els. Their data is taken from several non-literary text types, including
marriage announcements found in newspapers and weather reports
(Sandig) and historical descriptions of cities (Eroms).
Four of the remaining six papers involve stylistic/textual analyses
of literary texts. J. Goheen looks at allegorical structure in Gottfried' s
middle high German courtly epic Tristan und Isolde; F. Simmler char-
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
20/34
REVIEWS
24
acterizes and clarifies the function of syntax in the literary fairy tales
of German romanticism and contrasts this with the genre
of
orally
transmitted fairy tales represented by the brothers Grimm; H. Rupp's
interpretation of Paul Celan's poem Heimkehr demonstrates the limi
tations of a textlinguistic analysis sans interpretation; and W Weiss
demonstrates the interface of stylistics and textlinguistics in the prose
of Robert Musil. In a clear departure from the other papers, E. Schulz
and E.W.B. Hess-Liittich investigate the use of the spoken language as
a group marker by young people.
Not surprisingly, most of the papers begin with a discussion of the
relationship between sytlistics and textlinguistics. Style or stylistics
the older discipline,
is
seen as a subjective evaluation of esthetics,
which considers primarily stylistic devices. A stylistic analysis seeks to
provide depth, and remains, by nature, fuzzy.
Textlinguistics on the other hand,
is
descriptive, more objective,
more certain
of
its task, and has a distinct methodology, which is based
on taxonomic and algorithmic procedures. The strategies for text cre
ation,
e.g.,
narrating, describing, arguing, and directing, in short the
speaker/writer intention, are critical.
Despite the controversy reflected in the session title Textlinguistik
contra Stilistik? there
is
general agreement among these presenters at
least that textlinguistics and stylistics are not mutually exclusive.
In
fact, textlinguistics may even have something to offer stylistic studies
Cf. Goheen on allegory and Weiss on Musil's use of metaphor. To this
end, Michel (6) and Sandig
passim}-in
her extremely lucid style
subsume stylistics under textlinguistics.
The major shortcoming of traditional (literary) stylistics according
to G. Michel
is
its failure to adequately differentiate between the key
issues of normative
rule
vs. stylistic
regularity.
Current linguistic the
ories and methods, e.g., speech act theory, conversational analysis and
text linguistics, in conjunction with more empirical data from text type
studies, he suggests, are better suited to determine the most likely
regularities of the elusive stylistic norms.
Eroms finds it much easier to distinguish a stylistic analysis from
a textlinguistic analysis, whereas Aust notes that a text is much easier
to grasp than style. According to Aust the basic difference between
textlinguistics and stylistics is not what the two disciplines investigate,
but rather the questions each poses. Furthermore, a stylistic analysis is
more interested
in
the relationshhip of individual phases within a speech
activity, while a textlinguistic analysis
is
interested in the genesis of
texts and the identification of constitutive elements. In his concluding
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
21/34
242
WORD
VOLUME
41, NUMBER 2 (AUGUST
1990
commentary,
W.
Weiss applauds the emphasis in the presentations on
the centrality of the text versus that of the origin, reception or context
of
a text, thereby exposing his own methodological bias.
It should come as a surprise to
no one that the field of lexicogra
phy, which enjoys a distinguished tradition in German linguistic schol
arship, and lexicology should include
17
papers, an introduction
(Wiegand), a podium discussion and a conference summary (Fleis
cher), addressing controversial theoretical and practical issues
of
lex
icography,
e.g.,
word class inclusion/exclusion, monolingual vs. bi
lingual definitions, and audience needs.
In the lead article G. Harras, who speaks from years of lexico
graphical experience at the Institut fiir deutsche Sprache in Mannheim,
agrees with Wiegand (1985) that dictionary definitions need to be
written for the layman. Drawing on the work of H. Putnam (197 5,
1978), Harras recommends word definitions 1 that utilize stereotyp
ically marked semantic features as reflected in the Sprachgemein-
schaft
and
2
that avoid pseudo-scientific terminology. Kucera, on the
other hand, supports the notion that some degree of technical infor
mation that is comprehensible to the nonspecialist must be included in
definitions. Piischel, who reviews the merits of Johann August Eber
hard's contribution to the history
of
German thesaurus writing, reiter
ates the old notion, that it is unclear who uses a thesaurus, despite the
necessity that any thesaurus be written with the user clearly in mind.
Some of the most stimulating papers propose inclusion in dictio
naries of traditionally omitted word classes, e.g., modal particles
(Wolski), hedges (Kolde), formulaic expressions (Kiihn) and idioms
(Fleischer).
Three papers describe ongoing or planned dictionary projects.
Kirkness proposes abandoning the familiar German term Fremdwort
'foreign word' and adopting the English technical term 'hard word,' as
has been done at the IdS/Mannheim in the project
Schwerworterbuch
for German. E. Firchow H. Fix report on a successful computerized
lexicon for the Old Islandic Elucidarius, and Triib attempts to reconcile
difficulties presented by semasiological and onomasiological dictio
naries for Swiss German. L Zgusta broaches the delicate topic of
reputed plagiarism in M. Monier-Williams 1872 edition
A
Sanskrit
English Dictionary (Oxford: The Clarendon Press, 1989 1951.)
Acknowledging that dictionary writers today regularly borrow from the
work
of
others, he predicts an increase
in
this trend with the increased
computerization of dictionary projects.
Wiegand's paper on author lexica and his subsequent chairing of
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
22/34
REVIEWS 243
the podium discussion with West German authors Giinter Grass and
Helmut
e i ~ e n b i i t t e l
(Wolski,
228-236)
proved to be a highly con
troversial topic. Giinter Grass, who has strived for and developed a
pluralistisches SprachversHindnis in post-war West Germany, crit
icized Germanists in general and particularly Wiegand for utilizing a
Germanistensprache Lehrersprache Funktioniirssprache (229),
which is difficult (for the layman not to mention G. Grass himself) to
comprehend.
Several papers address the language used in definitions. Reiterat
ing an idea that has been around for several years, H.-P. Kromann
pleads for two different types
of
bilingual dictionaries for Germanis
t ik-a
passive reading reception dictionary that fosters Heriiber-
setzen i.e. from L
2
to L
1
and an
active
production dictionary that
enables Hiniibersetzen i.e., from L
1
to L
2
K. Sunaga discusses the
necessity for eliminating culturally-bound interference problems in
German-Japanese dictionaries. Similarly, H. Nikula criticizes the de
ficiencies of using authentic language examples. J. Korhonen outlines
the problems of semantic and syntactic verb descriptions
in
a historical
dictionary
of
German, and F. J. Hausmann criticizes the major German
thesaurus, the Schiilerduden (Muller) for failing to provide a) adequate
context in its examples,
b
pragmatic information
of
use to non-native
speakers and c) foreign words as an aid in meaning.
Forum 4 of the congress, entitled Grammatical or pragmatic
organization of speech?'' consisted of 13 papers, including two official
responses to two
of
the papers presented (Lenerz to Braunmiiller, and
Hohle to Zemb). As M. Reis indicated in her introductory remarks, the
presentations represented very different opinions on the (in)depen
dence of the linguistic system and pragmatics.
W. Abraham performs a major task by sketching out the domain
of
the term pragmatics in the introductory articles. Citing data from
right-brain/left-brain research (Bayer) he concludes that a definitional
separation
of
pragmatics from semantics is supported by the biological
separation.
Given the unique characteristics of German word order, it is un
derstandable why most
of
the papers discuss the relationship
of
prag
matics and syntax and make frequent use of
valence grammar. Within
this framework
P.
Mrazovic examines the extremely problematic clas
sification
of
noun/verb compounds,
e.g.,
in Abrede stellen as a verb
plus adjunct or as full verb. K. Braunmiiller discusses principles
of
German word order as typological patterns that have crystalized out
of
contextually bound strategies.
In
his response to Braunmiiller Lenerz
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
23/34
44 WORD
VOLUME 41, NUMBER 2 (AUGUST 1990)
rejects B s notion, that grammar is merely frozen pragmatics, because
it fails to explain, among other things, the conditions why certain
patterns should emerge but not others. J-M. Zemb questions the notion
of syntactic fields in German, particularly the Mittelfeld i.e., the area
between subordinating conjunction and finite verb
in
German, for
being too metaphorical and lacking descriptive power. This is but one
of the notions that T. Hiihle systematically refutes in his response to
Zemb. H. Glinz, who sees
no
conflict between grammar and pragmat
ics, proceeds to discuss the sentence as a pragmatic notion, reserving
the English term clause for pragmatics, since German uses Satz
in
both
cases. P Valentin discusses subordinate clauses which look like sub
ordinate clauses,
i.e.,
they are verb-final, but do not function as nor
mal sentence elements, e.g., enn du Durst hast Bier ist im Kuhl
schrank. If you are thirsty, beer is in the refrigerator.'
The remaining five papers all treat topics other than the relation
ship of pragmatics to linear syntax.
I
Rosengren dispenses forthwith
of the notion of a one-to-one relationship between speech act and
syntactic type and proposes instead that the semantic as well as syn
tactic structure determine expressions. In a discussion of pragmatics
and sentence stress, A Fuchs rejects the notion that the most important
word in a sentence is accented and establishes instead three text-se
mantic, functional dimensions for placing accent. G. Ohlschlager re
jects the current definition of modality, i.e., speaker attitude toward a
proposition, as too broad. At the same time attributing modality to
linguistic expressions alone is too narrow, he states. W Koch presents
an extremely valuable description of modal verb usage in a corpus of
German business correspondence, and H. Weydt concludes the volume
with yet another fine contribution toward understanding the modal
particles in German, in this case the pair denn and eigentlich.
As the title of this volume states, the papers presented here raise
a plethora of new and naggingly familiar controversial questions of
importance, not only to Germanists but to general linguists and lan
guage practitioners in general.
Dept. o Linguistics Germanic Asian African Languages
Michigan State University
Wells Hall A-615
East Lansing. M/. 48824-/027
REFERENCES
Bayer, Josef. 1985. Neurophysiologie und modulare Sprachbeschreibung. Ms. Aachen: Tech
nical University.
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
24/34
REVIEWS
245
Miiller. Wolfgang. Ed. 1977.
Schulerduden. Die richtige Wortwahl. Ein vergleichendes Wort-
erbuch sinnverwandter Ausdrucke.
Mannheim: Bibliographisches lnstitut.
Putnam, Hilary. 1975. Philosophical Papers.
Vol
2: Mind Language and Reality. Cambridge:
Cambridge U.P.
. 1978. Meaning and the Moral Sciences. London: Routledge K Paul.
Wiegand, Herbert Ernst. 1985. Eine neue Auffassung der sog lexikographischen Definition. Karl
Hyldgaards-Jensen Arne Zettersten Ed. Symposium on Lexicography 11 Proceedings of
the Second International Symposium on Lexicography May 16 17 1984 University
of
Copenhagen. Tiibingen. =Lexicographica. Series Maior 5. Pp. 15-100.
GERHARD LEITNER, ed.,
The English Reference Grammar: Language and
Linguistics Writers and Readers. (Linguistische Arbeiten 172). Tiibingen:
Max Neimeyer Ver1ag, 1986. ii + 450 pp.
Reviewed y CHARLES W. KREIDLER
This is a collection of papers given at a conference on ''English
Grammar-English Grammars, held at the Freie Universitat Berlin in
July 1985. Participants
in
the conference were scholars and teachers
concerned with the writing of grammars (of English) and/or the use of
such grammars in secondary and university-level English courses. Leit
ner points out in his Introduction
( l-3)
that a linguistically sound and
comprehensive grammar is not necessarily, or usually, the same as one
which is specifically aimed at students' needs and which concentrates
on the most common patterns of usage. Several of the participants ir
the conference, and contributors to this volume, are authors or co
authors of such pedagogical grammars, published or in progress.
The 24 papers in the volume are divided into two parts, sixteen
of
them in Part I, Contemporary and Future Reference Grammars, and the
remaining eight in the more homogeneous Part Il, Historical and Na
tional Profiles of English Grammars.
The general topic
of
Part I is What should a pedagogical reference
grammar be? Three papers, however, deal with more specific matters.
Charles-James N. Bailey, 'Where English cannot put a preposition
before a relative or interrogative pronoun,' goes over the intricacies
of
constructions like 'the hill up which they ran' vs. 'the bill which they
ran
up,
adding a lot
of
good points to a topic which has been heavily
worked over. One might wish his terminology was less idiosyncratic
and his references to other scholars less limited. Robert Burchfield,
The end of the alphabet: Last exit to grammar,' reaffirms, with ex-
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
25/34
246
WORD,
VOLUM
41, NUMBER 2 AUGUST 1990
amples from recent writings, the fact that subject-verb concord exists
in English and that variation occurs after nouns like group. Derek Davy,
Implications
of
the emergence of new standards of English for the
writing of English grammars, discusses problems which may arise in
attempts to describe English
in
all its native and non-native varieties.
Elsewhere in the volume there are a few pious statements about the need
to consider the varieties of World English, but in reality most of the
contributors show greater concern for finding a pedagogical minimum,
a search for the most common ways of expressing, for example, future
intentions.
Three of the participants describe texts which they have co-au
thored. Sidney Greenbaum,
in The Grammar o Contemporary En-
glish and the Comprehensive Grammar o the English Language,
explains why the authors
of
Quirk et al. 1972 decided to rewrite rather
than revise that book. The rewrite, Quirk et all985,
is
nearly twice the
size of the 1972 work,
is
said to contain new findings, new theoretical
discussions, a more elaborate treatment of tense, aspect, and modality,
and improvements in terminology and indexing. Jan Svartvik, A Com-
municative Grammar
o
English,
makes the claim that Leech and
Svartvik 1975
is
a new approach to English grammar, based on lan
guage in use. Future pedagogical grammars, he maintains, will be
based on new means of surveying actual usage, will show greater
acceptance of language variety, and will be more oriented toward
semantic and pragmatic descriptions. Quite different
is
the approach of
Gottfried Graustein, English Grammar: A scholarly handbook in
teacher-training in the
GDR,
who elaborates a much more extensive
analysis of syntax, as dealt with in his work, 1984.
Part of deciding what to teach involves the knowledge of what is
most used. John McH. Sinclair, First throw away your evidence,
maintains there
is
need for much larger
corpora-l
million words or
more-than
have previously been used as a basis for grammar-writing
or else descriptions are not likely to reflect the real language, and he
illustrates what can be done with present-day computers. A similar
note is struck by Dieter Mindt, Corpus, grammar, and teaching En
glish s a foreign language, who says that instruction should be based
on a fresh examination of data. He exemplifies the fresh approach with
a study of what standard grammars have to say about the comparative
frequency of different ways of expressing future intention
(will,
shall,
be going to,
simple present, et al.), how German textbooks deal with
the same points, and what actually emerges from a study of the con
versations in Svartvik and Quirk (1980). Mindt concludes that peda-
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
26/34
REVIEWS
47
gogical materials need to recognize that relative frequency of alternate
expressions is not the same
in
different kinds of discourse. n a similar
vein Wolfgang Zydatiss, 'Grammatical categories and their text func
tions-Some
implications for the content of reference grammars,''
examines the use, or interpretation,
of
the present perfect in various
kinds of texts-advertisement, financial report, sports story, et al .-to
illustrate his point that a reference grammar should give an account of
the basic meaning
of
a linguistic form and show the modifications or
extensions of that meaning
in
various kinds of texts.
Four of the papers undertake to say what a reference grammar
should be. Egon Werlich, in
'The
relevance
of
a text (type) grammar
in foreign language teaching: With a note on text type switches,'
makes the point that a text-oriented type of grammar is superior to a
sentence grammar, but it needs to
be
complemented by a text typo
logical component (i.e. something which deals with different kinds of
discourse). Rene Dirven, 'Towards a pedagogical grammar,' states
that the making
of
a pedagogical grammar consists in selecting the
grammatical items which are problems for the learner, and what
is
selected is to be presented in two phases. First, the descriptive matter
concentrates on what is different from the student's native language;
then an additive
phase
provides more specific detail. Jochen Nie
meyer, 'Teachers, grammar teaching, and grammar books: Some de
siderata,' deals with the role of grammar in the classroom. He points
out that school grammars and also recent standard reference works fail
to deal with, or give conflicting opinions on, several problems, such as
concord, co-occurence of certain adverbials with the present perfect
and past tense, solved already by earlier 20th century grammarians,
who should not be ignored. One point of this article is that the usual
teacher
of
English in Germany depends heavily on a reference gram
mar and does not have a whole shelf of such books nor subscribe to
teacher journals. Friedrich Ungerer, 'Guidelines for a multi-purpose
teaching grammar,' discusses the effects
of
a multi-purpose approach,
that is, one which takes into account student production of texts,
comprehension, and analysis, and he proposes a reduction and stan
dardization of grammatical terms.
Three other articles are concerned with the audience, the nature of
students for whom such grammars are intended. Dietrich Lange, 'Is it
the schools' fault if students don't use grammars?,' searches for the
reasons for difficulties in learning grammar often experienced by Ger
man university students of English. His conclusions are forthright:
grammar teaching is confined to lower levels, where it is unduly sim-
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
27/34
248 WORD VOLUM
41, NUMBER 2 (AUGUST 1990
plified, and at upper levels didactic efforts are concentrated too exclu
sively on productive skills and the teaching
of
literature. Wolfgang
Mackiewicz and Harald Preuss,
The
role
of
scholarly grammars in
course design at university level, conclude from a study they made that
use
of
grammatical handbooks leads to student acquisition
of
termi
nology but not to student performance in the target language. However,
such handboods have played an important role
in
the design
of
the
courses. A. Adler, H. Hirschmiiller, G. Leitner, K. Priifer, and G.
Schnorr, Grammars of
English versus students of English, maintain
that though most grammars select content and terminology on the basis
of presumed needs
of
their readers, no study has been made of how the
intended audience actually uses such books. They offer a pilot study
conducted at Freie Universitat as an entry into such further exploration.
The papers in Part differ in scope but all are worthy contribu
tions to the history of grammars
of
English. Robert H. Robins,
The
evolution of English grammar books since the Renaissance, traces
four centuries
of
change in the models which grammarians have fol
lowed, from the traditional Latinate to 17th century writers who partly
discarded the Latin model, to the 19th century approach of Henry
Sweet and the German Anglisten, to 20th century structuralists and
generati vists.
John Algeo, in A grammatical dialectic, sees the history
of
grammar writing and teaching in the USA
s
a sequence of eight
developments: Latinate grammars, nativist grammars, clause-focused
grammars, historical grammars, utilitarian functionalism, structural
ism, the Students Rights movement, and the Back to Basics movement.
Though somewhat overlapping, these developments have generally
been related in a continual thesis-antithesis-synthesis course. Charlotte
Downey,
The
constants and variables which guided the development
of
American grammar writing in the 18th and 19th centuries, shows
that grammar writing in America was at first strongly imitative
of
British
models, with emphasis on memorization
of
rules and definitions and on
parsing, changing by mid-19th century to more inductive exercises and
sentence-building. Goold
Brown-The
American grammarian
of
grammarians in the nineteenth century, by Kurt Wachtler, attributed to
that pedagogue the perpetuation
of
the authoritarian tradition of cor
rectness and blames his view of language behavior for the ill-founded
linguistic stereotypes and cliches which are still with us.
Flor Aarts, English grammars and the Dutch contribution:
1891-
1985 , shows with an interesting time-line (364) how the Dutch gram
marians Poutsma, Kruisinga, Zandvoort, Aarts Aarts, and van Ek
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
28/34
REVIEWS
49
Robat have influenced or been influenced by Sweet, Jespersen,
Curme, Long, Quirk et al, and Hiddleston.
Aleksandra Jankowska, in Polish grammars
of
English, dis
cusses seven Polish grammars
of
English published in the period
1948-1980,. scholarly and pedagogical, and traces their development
from traditional, word-class-based grammars to an approach which
is
meant to be communicative.
Bertil Sundby,
in
Parallelism and sequence in early English pre
scriptive grammar, claims that 18th century authoritarian views on
sequence
of
mood, tense, voice, etc. and on structural parallelism are
not merely opinions but reflect to some degree syntactic constraints
still in the language. These observations, he writes, serve to demon
strate the linguistic potential
of
a projected historical dictionary
of
English normative grammar, such as the one currently being prepared
at the University
of
Bergen (see Sundby 1980).
Gerhard Leitner provides the retrospective summation in English
grammars: Past, present and future, retracing major developments in
English grammar writing and concluding with a brief outline of the
future challenges: text/discourse, data base, native/non-native En
glishes and usage patterns.
The volume concludes with summaries
of
articles, descriptive
sketches
of
the contributors, a name index, and a very thorough key
word index.
Department o Linguistics
Georgetown University
Washington. DC 20057 US
REFERENCES
Graustein, Gottfried et al. 1984.
English grammar: A university handbook.
3rd ed. Leipzig:
Enzykopiidie.
Leech, Geoffrey, and Jan Svartvik. 1975. A communicative grammar o English. London: Long
man.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1972. A grammar o
contemporary English. London: Longman.
. 1985. A comprehensive grammar o the English language. London: Longman.
Sundby, Bertil. 1980. A dictionary o English normative grammar 19700 1800. A preliminary
report.
Linguistic Project Reports, Department
of
English, University
of
Bergen.
Svartvik, Jan, and Randolph Quirk, eds. 1980.
A corpus o English conversation.
Lund: Gleerup.
Dow
nloadedby[123.2.1
5.2
42]at02:4414November2015
-
7/24/2019 Review of Dependency Syntax: Theory and Practice by I. Melcuk
29/34
250
WORD,
VOLUM
41. NUMBER 2 AUGUST 1990
JONATHAN FINE and ROY
0.
FREEDLE eds.). Developmental Issues n
Discourse. Advances in Discourse Proccesses, X Norwood, NJ: Ablex,
1983. xi-315 pp.
JONATHAN FINE ed.).
Second Language Discourse: A Textbook
o
Cur-
rent Research. Advances
n
Discourse Processes, XXV). Norwood, NJ:
Ablex, 1988. xi-214 pp.
Reviewed by JACQUELINE ANDERSON
The Advances in Discourse Processes Series, published by Ablex
under the general editorship of Roy 0. Freedle, continues to make a
significant contribution to the field of linguistic analysis by bringing
together ideas, authors, and approaches from a number of disciplines,
all sharing a common interest
in
discourse. The strength of Volumes X
and XXV, like others in the series-and ironically, their most notable
weakness-is their characteristic emphasis on empirical testing. The
multi-frame models provided by the authors included in the two books
strongly encourage teachers to construct theory-driven language de
velopment and intervention programs. But they do so by stressing the
kinds of logarithmic analyses that are frequently inaccessible to the
very classroom instructors who might most benefit from the models
being suggested.
The aim of these two volumes is to present diverse theoretical
approaches, methods, and applications of discourse analysis as they
relate to development with