Sie sind auf Seite 1von 14

1

X-Bar Syntax in VSO languages:


A Generative Approach to Welsh

by

Sandra Hufnagel


Paper submitted in partial fulfillment of the
requirements for a graded credit for the course
Syntax
in Summer Term 2009
Submission Date: 17
th
August 2009
Approved by: Prof. Dr. Jrgen Handke
Philipps University Marburg
2

Contents
1 Introduction ........................................................................................................ 3
2 X-Bar Theory ...................................................................................................... 4
2.1 The X-Bar Scheme ............................................................................................ 4
2.2 Movement .......................................................................................................... 5
2.2.1 Movement and Representation ..................................................................... 6
2.2.2 Movement and Constraints ........................................................................... 6
2.2.3 The Underlying Sentence Structure .............................................................. 7
2.3 Conclusion ......................................................................................................... 7
3 X-Bar Theory in Welsh ...................................................................................... 8
3.1 The Central Problem ......................................................................................... 8
3.2 Flat Structures in Welsh VSO ........................................................................... 8
3.3 A Transformational Approach to Welsh VSO ................................................ 10
3.3.1 V-to-C Movement Hypothesis .................................................................... 10
3.3.2 V-to-I Movement Hypothesis ..................................................................... 11
3.3.3 Verb and Subject Movement ...................................................................... 11
3.3.4 Further Issues .............................................................................................. 12
4 Conclusion ......................................................................................................... 13
5 References ......................................................................................................... 14

3

1 Introduction
X-Bar Syntax and the Government and Binding approach have pushed syntactic theory
forward to a very high extent, since numerous grammatical relations in languages can be
accounted for on the grounds of this theory. However, there are languages where government
and binding occurs without the adjacency of the governing element and the element governed,
namely in languages with verb-subject-object word order (VSO).
This term paper deals with the question of how VSO order can be explained within a
theoretical frame of X-Bar using the example of Welsh, a Celtic VSO language.
In the first section, X-Bar Theory and several constraints connected with it are
introduced. The second part deals with the problem of how to apply X-Bar Theory to a VSO
language like Welsh.
2 X-Bar Theory
2.1 The X-Bar Scheme
Before X-Bar Theory was established, it was common to describe sentence structure
within Generative Grammar (GG) with flat structures (see (1); Roberts 1997:10, modified):
(1) S
NP Aux VP
John is laughing
However, constituent tests proved that smaller units than e.g. S in (1) form a constituent
on their own (Roberts 1997:12):
(2) John [is laughing] and [was dancing].
X-Bar Theory accounts for that by introducing a hierarchical structure which only
permits binary branching. The scheme is claimed to be universal, i.e. applicable to every
language in the world. All types of phrases are presented as in (3), only varying in the
direction of the branching of Specifier and complement (see Roberts 1997:20).
(3) X/XP

Specifier X

X Complement

In (3), X is the lowest possible projection of itself and is therefore called zero projection
(X
0
) (Haegeman 1993:95). The head X forms an intermediate projection X together with its
complement (which is selected by the head); the two are sisters (Culicover 1997:19). X and
another constituent, Specifier (which modifies the head), form together X, or XP, the
maximal projection of the head (Haegeman 1993:95). An adjunct is defined here as having a
sister X and forming another constituent X with it (Haegeman 1993:95).
Accordingly, there are three Phrase Structure Rules, illustrated in (4) (adapted from
Haegeman 1993:95). The order of the X-projection and its sister might vary in the different
languages (see brackets).
4

(4) X Spec X (X X Spec)
X X YP (X YP X) (rule for adjunction; YP = adjunct)
X X ZP (X ZP X) (ZP = complement)

There are five lexical categories which can fill X: adjectives (A), adverbs (Adv), nouns
(N), verbs (V), and prepositions (P) [INT2]. The two central functional categories are
(Haegeman 1993:98-108):
IP (Inflectional Phrase), with inflectional features (agreement and tense), or an
auxiliary in its head
CP (Complementizer Phrase) with a complementizer as if, or that, or an empty
complementizer in its head determining the nature of a clause
Thus, the sentence in (1) would be illustrated as in (5):
(5) CP

C

C IP
5


(empty) NP I

N I VP

N is V

John V

laughing
2.2 Movement
The first section gave an overview of the basic assumptions of X-Bar Theory. In the
following, the transformational component is examined.
6


2.2.1 Movement and Representation
In Government and Binding Theory (GB), it is assumed that every sentence is
represented on different levels, D-structure and S-structure being the central levels
1
(see
Haegeman 1993:280-281): At D-structure, the constituents are ordered in their basic argument
structure given by the lexicon; here, the arguments also receive their thematic roles, such as
Agent and Theme [INT 1]. Between D- and S-structure, movement takes place resulting in a
more superficial representation, S-structure, where all arguments have been assigned case
[INT 1]. Movement in GB is subject to only one rule: move , or move something somewhere
(Haegeman 1993:281-282).
2.2.2 Movement and Constraints
Of course, if any element can be moved anywhere, our theoretical apparatus is far too
strong. Consequently, a certain number of principles are to be introduced which constrain
movement.
2.2.2.1 Structure Preserving Principle
The Structure Preserving Principle states that the structure of a syntactic tree must be
identical for D- and S-structure (Haegeman 1993:312). This also implies that maximal
projections must move into XP-positions, and heads must move into head positions
(Haegeman 1993:312-313). Thus, in a tree as in (5), the NP John could never move to the
head position C; however, the I-head is could move to that position.
2.2.2.2 Trace Principle
The Trace Principle requires that every movement leaves a trace at its original position
(Culicover 1997:181). This implies that no other element can move into the trace position,
and that the trace might also block other processes (compare with Tallerman 1990; in Duffield
1995:50).
2.2.2.3 Head Movement Constraint
The Head Movement Constraint demands heads only to move to their next higher head
position (Roberts 1997:39); e.g. a head V must pass I before landing in C, since I is lower
than C and thus the next higher head position for V.

1
There are also two other levels Logical Form (LF) and Phonological Form (PF), which are omitted here. See: Van Valin
2003:193.
2.2.3 The Underlying Sentence Structure
Another crucial aspect to be considered is how arguments of a sentence are distributed at
D-structure. If we assume the so-called VP-Internal Subject Hypothesis to be correct,
namely that the subject is part of the VP at D-structure (see Roberts 1997:77), a sentence like
John saw Mary would be presented as in (6) (CP not represented):
(6) IP

I

I VP
{PST.
3.SG} NP V

N V NP

N see N

John N

Mary

(6) illustrates that the subject, John, is the Specifier of the VP (Spec.VP), whereas the
object, Mary, is the complement of the VP. The verb see is the head of the VP, and forms a
constituent together with its complement. All information about tense and agreement is in the
head of IP.
2.3 Conclusion
Even though we have not examined more complex sentences in detail, there seem to be
no difficulties with X-Bar Syntax for an SVO language like English. In the following section,
we will attempt to describe Welsh as a VSO language within the same patterns.
7

8

3 X-Bar Theory in Welsh
3.1 The Central Problem
The first question that might arise is where the actual difficulty lies to describe VSO
languages within X-Bar frames. As exemplified in (6), the verb and its object form a
constituent V in English declaratives. However, in VSO languages like Welsh, the verb and
its object apparently are separated by the subject.
There are different hypotheses attempting to give an account of the phenomena in such
languages. Are verb and object only a discontinuous constituent? Or do we have to return to
flat structures assuming that verb, subject and object form one inseparable constituent?
In the first part of this section, we will examine whether flat structures are to be preferred
to layered structures in order to describe VSO order. The second part deals with approaches
using binary branching and a transformational explanation of the phenomena.
3.2 Flat Structures in Welsh VSO
The attempt to explain VSO word order by flat structures is probably the simplest
solution suggesting that VSO is the underlying word order and verb, subject, and object form
one inseparable constituent. Thus, the verb governs subject and object equally as its
complements.
Tallerman (1990; in Duffield 1995:49ff) argues for Welsh to have such flat structures
because of insufficient empirical evidence that could account for a transformational
explanation of VSO. Furthermore, she stresses that the distribution of verb and object do not
provide any direct evidence for being a constituent. Moreover, a certain phonological
phenomenon, soft mutation, should not take place in certain syntactic contexts if there was a
trace left by movement.
Even though there is no direct evidence against those arguments, the attempt seems to be
problematic in various ways. First, there is a theoretical problem concerning the assignment of
thematic roles (see Duffield 1995:52; Sadler 1988:15): In GB, the role Theme is assigned to
the object as the complement of the verb. If, however, subject and object are both
complements of the verb, it is uncertain which of the two should be assigned Theme. Thus,
9

the flat structure approach cannot account for why there is a semantic asymmetry of the
subject as the carrier of Agent and the object as the carrier of Theme.
Furthermore, there is a range of indirect empirical evidence in favor of an underlying
SVO approach. Duffield (1995:50ff) argues against Tallerman (1990) that in both Welsh and
Irish, another Celtic VSO language, there are subject-object asymmetries, e.g. concerning
binding of anaphors. The examples in (7) show asymmetry concerning reflexives in Welsh:
subjects can serve as antecedents to reflexives in object position; however, objects cannot be
antecedents to reflexives in subject position (from Borsley/Tallerman/Willis 2007:51):
(7) a. Gwelodd Gwyn ei hun.
see-PST.3.SG Gwyn 3.M.SG self
Gwyn saw himself.

b. *Gwelodd ei hun Gwyn.
see.PST.3.SG 3.M.SG self Gwyn
Himself saw Gwyn.
If subject and object were both complements of the verb, this should not be so.
Additionally, the adjacency of non-finite verb and object in periphrastic (see (8b)) and
aspectual non-finite structures (see (8c)) in Welsh also hints to an underlying SVO order
where verb and object form a constituent (see brackets) (from Tallerman 1990, in Duffield
1995:49-50), as well as SVO order in embedded non-finite clauses in (8d) (from Ouhalla
1999:337):
(8) a. Gwerthodd Huw gar.
sold-3.SG Huw car
Huw sold a car.

b. Gwnaeth Huw [werthu car].
did-3.SG Huw sell-INF car
Huw sold a car.

c. Mae Huw [wedi gwerthu car].
is-3.SG Huw ASP sell-INF car
Huw has sold a car.

d. Dymunai Wyn i Ifor [ddarllen y llyfr].
wanted Wyn for Ifor read the book
Wyn wanted for Ifor to read the book.

Thus, it seems more plausible to assume that VSO word order in Welsh is derived from
an underlying SVO word order.
3.3 A Transformational Approach to Welsh VSO
If SVO is underlying in Welsh, which are the movements taking place between D- and S-
structure? Most analyses propose an operation of V-fronting; however, they diverge in their
assumptions about the landing site of the verb and about the question whether subject
and/or object remain in situ or not. According to the Head Movement Constraint, there are
only two options for the verb to move: either to C or to I. In the following, we will discuss
which hypothesis of movement appears to be adequate for Welsh syntax.
3.3.1 V-to-C Movement Hypothesis
Since V-to-C movement is assumed for Germanic and Romance languages, it might be
applicable to Welsh, too. However, in the former, there is a blocking of verb-movement if a
complementizer is present, whereas in the latter, it is not (from Roberts 2005:20-21):
(9) a. When will they read this book?
b. *I wonder if will they read this book.
c. Tybed a geith hi ddiwrnod rhydd wythnos nesa.
I-wonder PRT will-get she day free week next
I wonder if shell get a free day next week.
In (9), the particle a has a similar function in Welsh as if in English (Roberts 2005:21).
As the C-position cannot be double-filled, V seems to be excluded from moving to C (as
illustrated in (10)). Thus, it appears more plausible to move V to I.
(10) CP

C

C
a
VP

10

V

V

geith
11

3.3.2 V-to-I Movement Hypothesis
Since VSO order in Welsh is related to finiteness (Roberts 2005:9), it seems plausible to
assume that V moves to I triggered by the inflectional features in I. However, also here a
problem arises if we consider that verbs must be adjacent to their subjects in VSO order (see
(11)) (Roberts 2005:10). If V moves to I and the subject remains in Spec.VP, there are
positions available to adverbs like yfory (Roberts 2005:10), which should not exist. Thus, it
seems implausible to assume the subject to remain in situ.
(11) *Mi welith yfory Emrys ddraig. (from Roberts 2005:10)
PRT will-see tomorrow Emrys dragon
V X S O
Tomorrow Emrys will see a dragon.

Furthermore, the (second) negation element in Welsh, which is believed to be adjoined to
IP, must follow the subject (see (12); Roberts 2005:10-11). This implies that the subject must
move to a higher position than Spec.VP in which it precedes the negation element. Thus, both
verb and subject seem to move out of VP.
(12) Cheisiodd Gwyn ddim ateb y cwestiwn bob tro.
tried Gwyn neg answer the question each turn
Gwyn didnt try to answer the question every time.
3.3.3 Verb and Subject Movement
Assuming the subject to move, we have to consider the Structure Preserving Principle
demanding the subject to be raised into an XP-position, in this case into Spec.IP. If we,
however, attempt to position subject and verb in a syntactic tree, there seems to be a trap: If
V only moves to I, the resulting word order is no longer VSO, but SVO. If V moves to C, the
same problem as in (11) arises: the C-head cannot be double-filled in embedded clauses. It
appears that the only solution is to either split CP or IP.
3.3.3.1 Split-CP Hypothesis
For reasons of space, there will be no discussion of a Split-CP Hypothesis here. Briefly,
there are distributional difficulties if we assume V to move into the lower head of the split CP
(and the complementizer into the higher head of the split CP), e.g. the behavior of the Welsh
auxiliary bod (see Roberts 2005:33ff).
3.3.3.2 Split-IP Hypothesis
Thus, a syntactic tree as in (13) (adapted from Roberts 2005:43, slightly modified) might
be appropriate for Welsh: IP is split into an Agreement Phrase displaying the agreement of
subject and verb, therefore AgrSP, and a Tense Phrase, TP. The verb initially moves from V
to T, then from T to AgrS. The subject moves from Spec.VP to Spec.TP. As there does not
remain a slot for adverbs between AgrS and Spec.TP, the requirement for the position of
adverbs in Welsh is also accounted for.
(13) Mi welais i Megan. (from Roberts 2005:7)
PRT saw I Megan
I saw Megan.

CP

C

C AgrSP

Mi AgrS
12


AgrS TP

welais
v
NP T

i
k
t
v
VP

t
k
V

t
v
NP

Megan

3.3.4 Further Issues
As it might be expected, there are also other problematic points in describing Welsh
syntax. As an example, Roberts (2005:44) mentions the so-called Anti-Agreement Effect in
Welsh and other VSO languages (like Classical Arabic), a systematic disagreement, which
occurs in VSO structures despite the proposed fact that the verb is the head of the agreement
phrase (AgrS). Thus, Welsh syntax theory remains a disputable area.
PRT: predicate
AgrSP: Agreement of subject in verb
PRT: predicative particle in predicate adjectives and
nominals
Mi o fe se sitan frecuentemente antes de los verbos
conjugados para indicar que son enunciativos. En el
presente e imperfecto del verbo bod (ser), se emplea yr.
13

4 Conclusion
In the sections above we have discovered the framework of X-Bar Theory and
transformations and have observed that the scheme possesses explanatory adequacy dealing
with SVO word order.
Although Welsh as a VSO language is a challenge to the explanatory frame of X-Bar
Theory, we have shown that X-Bar proved to be more adequate than the traditional system of
flat structures, and have been able to at least account for a number of phenomena of Welsh
syntax. Still, there remain several weak points within the scheme demanding more
satisfactory alternatives to the one above, which might force us to make the scheme more
complex (see Roberts 2005). On the other hand, attempts have been made to explain Welsh
syntax in terms of non-derivational constraints within Lexical Functional Grammar, and
Head-Driven Phrase Structure Grammar (see Borsley/Tallerman/Willis 2007:52ff). Both
proposals may be starting points for further research in VSO phenomena.

14

5 References

Borsley, Robert D. / Tallerman, Maggie / Willis, David. 2007. The Syntax of Welsh. Cambridge:
Cambridge University Press.

Culicover, Peter W. 1997. Principles and Parameters: An Introduction to Syntactic Theory. Oxford:
Oxford University Press.

Duffield, Nigel. 1995. Particles and Projections in Irish Syntax. Dordrecht: Kluwer.

Haegeman, Liliane M. V. 1993. Introduction to Government and Binding Theory. Reprint. Oxford:
Blackwell.

Ouhalla, Jamal. 1999. Introducing Transformational Grammar: From Principles and Parameters to
Minimalism. Second Edition. London: Arnold.

Roberts, Ian G. 1997. Comparative Syntax. London: Arnold.

Roberts, Ian G. 2005. Principles and Parameters in a VSO Language: A Case Study in Welsh.
Oxford: Oxford University Press.

Sadler, Louisa. 1988. Welsh Syntax: A Government-Binding Approach. London: Croom Helm.

Van Valin, Robert D. 2003. An Introduction to Syntax. Reprint. Cambridge: Cambridge University
Press.

Internet Sources

[INT1] www.linguistics-online.de; module: Syntax/Transformations I, accessed 7 August 2009.

[INT2] www.linguistics-online.de; module: Syntax/X-Bar Syntax, accessed 15 August 2009.

Das könnte Ihnen auch gefallen