Syntax – Week 7, Lecture 1
X-Bar Theory Wrap-up &
Grammatical Relations
1. Possession and the DP Hypothesis
In the last lecture we studied the “DP Hypothesis” whereby a NP with a
determiner is reanalyzed as being a DP headed by a determinative with a
NP complement:
This new analysis of the NP has been driven by two main considerations:
1. That in the ‘s genitive possession construction, the ‘s can attach to
entire phrases:
[John]’s hat.
[The man standing over there]’s hat.
2. Cross-linguistic considerations that strongly suggest that a
determiner heads DP phrase.
However, note that this analysis is still speculative.
1
Syntax – Week 7, Lecture 1
A simpler analysis of possession is
to treat the clitic ‘s as simply
belonging to the N to which it is
attached. Under this analysis, the
possessor is a NP acting as specifier
to a NP.
However, this analysis ignores the
the fact that the ‘s is actually attaching to the entire specifier phrase.
Summary of Possession Constructions:
Representing ‘s genitive possession constructions in English using X-Bar
theory is not trivial, and although various candidate solutions have been
proposed – all have draw backs.
2. Why X-bar Theory?
● We started with a string of words in a language.
We saw various kinds of evidence which told us that a sentence is
not just an undifferentiated string: it represents a pattern of possible
combinations of different categories of words, and within the string
some words cluster together more closely than other parts.
This is the data which syntactic theory tries to account for and
explain.
What evidence did we see?
2
Syntax – Week 7, Lecture 1
(1) “possible combinations of different categories of words”
What tells us words belong to particular, different categories? This
was what the first part of this topic addressed and we saw two types
of evidence = two types of tests:
Substitution test – the distribution of words (what slot they filled
in the string)
Morphological evidence – the kinds of inflections the words
could take.
(2) “and within the string some words cluster together more closely
than other parts”
What tells us that substrings of words belong to larger constituents
(phrases)?
Constituent structure tests – does the substring of interest
behave as a unit in various syntactic contexts?
• To account for these behaviours, we started with a very simple
‘phrase structure grammar’ – where we wrote rules allowing us to
describe structures in terms of their constituency (constituent
structure) and their category labels.
• Interlude: we took an interlude at this point to introduce the notion
of dependency relations and to point out that these are something
that our very simple phrase structure grammar did not account for
or notate. What are dependency relations in a nutshell?
Dependency relations: Are functional relations between the
different elements of a constituent. The different words within
3
Syntax – Week 7, Lecture 1
a phrase or clause have distinct functions within it. We have
seen the following basic functions:
● Head
● Complement
● Adjunct / modifier
● Specifier
Dependency relations are relations which hold between
elements within these functions. They include relations
between heads and complements of:
Agreement [verb agrees with subject in person,
number]
Case Government [verb and preposition require
accusative case complement if
complement is pronoun]
Grammatical Function [complements of a verb are
differentiated in grammatical
function ie subject, object, etc.]
and relations between heads and adjuncts / modifiers or:
Agreement [in some languages adjectives agree
with head nouns in gender, number
etc]
● We then moved on to present the X-bar theory of phrase structure.
The major motivation for X-bar theory is that it allows us to account
for the facts about syntactic behaviour mentioned above – better
than the very simple PSG we initially defined.
4
Syntax – Week 7, Lecture 1
X-bar theory has the following advantages:
We have evidence that we need a level of intermediate constituent
structure between the word and the phrase. We initially looked at
this question for NPs, because it is easy to see within NPs that an
intermediate level is needed:
NP That man with the silly hat
?Intermediate constituent man with the silly hat
N man
We argued for this intermediate level constituent using constituent
structure tests. The question is then what to call this level.
You could just call it ‘NP’ – another kind of nominal phrase or
constituent – but this is not satisfactory because one could not
substitute any old NP for it:
that man with the silly hat
*that the woman
*that she
*that Mary
etc.
The notion of three levels of nominal constituency labelled in terms
of ‘bars’ then arises.
● A secondary part of X-bar theory is the idea that this intermediate
level possibility holds
● not just for NPs but for other categories as well;
● not just for English but for other languages as well.
5
Syntax – Week 7, Lecture 1
This is what is called ‘an empirical question’: in other words, these
are hypotheses, and we need to look and see if they are true or not.
How would we do this?:
● we would look at other phrasal categories within English and
use constituent structure tests to tell us whether it is justified
to postulate an intermediate category
● we would look at a whole range of other languages and see
whether, if that language is ‘configurational’, there is evidence
for an X-bar type of structure
Finally, X-bar theory addresses this question of the differences
between the words within a phrase with respect to their function.
The hypothesis is that there is a structural distinction parallel to the
functional one. That is, there is something special from a structural
point of view about:
● Heads
● Complements
● Adjuncts
● Specifiers
That is:
● We need to distinguish between the Specifier and the
rest of the phrase.
● There is evidence that complements are sisters to their
heads.
● And adjuncts / modifiers seem to be sisters of something
different (the complement + head complex).
6
Syntax – Week 7, Lecture 1
In summary:
X-bar theory is more than just a sophisticated form of the phrase structure
description which we started off with. It:
● accounts for the fact that we need intermediate constituents
● explains the different syntactic behaviour of complements and
adjuncts
● seems to offer a more general and explanatory description in that it
recognises the similarities across the structures of different
categories of phrases
● seems to offer a more general and explanatory description in that it
recognised the similarities across phrasal structure in different
(configurational) languages
Where does the ‘X’ bit come from and what does it mean?
It just means that you can substitute any lexical category for ‘X’ and get a
good description of the structure of a phrase of that category type:
Replace ‘X’ (and hence X’, X’’) with
N N’, N’’
A A’, A’’
V V’, V’’
P P’, P’’
I I’, I’’ etc.
(with the proviso that some people argue that it works for major category
types but not minor)
7
Syntax – Week 7, Lecture 1
2. Types of Grammatical Relations & Coding Systems
Reading: Kroeger Ch 7, Dixon (94), Blake (94)
Examples of Grammatical Relations are the relations between verbs and
their complements and subjects or between nouns and their dependents.
Over the past few weeks, we have examined structural ways that these
relations can be encoded in English. Now we are going to examine the
different ways languages can encode these grammatical relations.
Revision
Recall that this discussion came about because we realized that a
language needs a way of defining who did what to whom.
Two grammatical relations, subjects and objects are often referred to as
Terms or Direct Arguments because of the prominence so many languages
place on these two arguments. In English, we saw that subjects and
objects can’t be defined semantically – i.e.,
the subject is not always the doer of the verb:
she underwent surgery
the object is not always the patient of the verb:
I miss her.
Instead, these functions are defined grammatically:
subject: specifier of IP
object: NP complement to V
8
Syntax – Week 7, Lecture 1
The entities engaged in various semantic roles (such as agent, patient etc)
defined by the verb are mapped to the grammatical functions. This
mapping must be learnt as part of the meaning of the word. For example:
hit (agent,patient)
the agent is the subject
the patient is the object
was_hit (agent,patient)
the patient is the subject
the agent is an oblique
As we saw, subjects, objects and obliques differ in their syntactic behavior.
What are Grammatical Functions?
Arguments: complements and subject. These form the core grammatical
functions (subjects and objects, usually NP’s).
Within the core grammatical functions, we can also talk about:
external arguments (i.e., the subject – external to VP)
internal arguments (i.e., complements)
Internal arguments are subcategorized for by the verb and are in a
dependency relation with it.
9
Syntax – Week 7, Lecture 1
Arguments and their prototypical functions
Subject
That functional element in the structure of the clause that
prototypically expresses:
(i) the semantic role of agent
(ii) the presentational status of topic (we will come back to this)
Direct object
A grammatically distinct element of clause structure which in canonical
agent-patient clauses expresses the patient role.
Indirect object
A distinct element of clause structure characteristically associated with
the semantic role of recipient.
(In case marking languages, prototypically marked by Dative case.)
Predicative complement
A grammatically distinct complement of the verb characteristically
expressing a semantic predicate (which applies to the subject or direct
object).
Languages across the board have a way of grammatically coding these
types of information. The study of this is called Linking Theory – every
grammar must have an account of the way grammatical roles are linked to
the functions in a language.
10
Syntax – Week 7, Lecture 1
Examples
In a language like English, we can make the following distinctions with
respect to grammatical relations:
Subject
Jill saw the platypus.
subject (direct) object
Phil slept.
subject
Object (direct and indirect objects may be distinguished)
Jill saw the platypus.
subject (direct) object
Phil gave Jill the book.
subject (indirect) object (direct) object
Predicative complement
Jill is happy.
subject predicative complement (subjective)
Jill considers Mike a fool.
subject (direct) predicative complement
object (objective)
11
Syntax – Week 7, Lecture 1
Oblique
She leant the plank against the wall.
subject (direct) object oblique
I bought this table from the Brotherhood for $5.
subject (direct) object oblique oblique
Adjunct
Jill saw the platypus at Whittlesea on Tuesday
adjunct adjunct
Object of Preposition
up the hill
object of preposition
Defining Grammatical Relations
Configurational Languages (review) — In some languages GRs are
configurationally defined:
• Subject Specifier to IP (usually NP)
• Object NP complement to V, daughter of V’
• Object of P NP complement of P, daughter of P’
• Oblique PP complement of V, daughter of V’
• Adjunct Phrasal sister of V’ daughter of V’
This approach requires a VP for example. In other languages, GRs cannot
12
Syntax – Week 7, Lecture 1
be configurationally defined. Several languages are examined below:
1st Case: ‘VSO’ languages, where V and object NP do not form a
constituent e.g. Scots Gaelic:
Bhàsaich an cù
died the dog
‘The dog died’
Chunnaic Calum an cù
saw Calum the dog
‘Calum saw the dog’
or Welsh:
Lladdodd y dyn y ddraig
killed3sg the man the dragon
‘The man killed the dragon’
2nd Case:
So-called ‘non-configurational’ languages, where word order is free,
clause structure is ‘flat’, and grammatical relations are typically indicated
by CASE, e.g. many Australian Aboriginal languages such as Jiwarli (NW,
WA).
Kurlki juma yananyja
girl small went
‘The small girl went’
13
Syntax – Week 7, Lecture 1
or equivalently:
Kurlki yananyja juma
Juma kurlki yananyja
Juma yananyja kurlki
Yananyja kurlki juma
Yananyja juma kurlki
Kurlkingku jumangku thuthu parnka nhanyanyja
girl small dog large saw
‘The small girl saw the large dog’
or equivalently:
Kurlkingku thuthu nhanyanyja jumangku parnka,
Jumangku thuthu parnka nhanyanyja kurlkingku,
Nhanyanyja thuthu parnka kurlkingku jumangku
or any other possible permutation of these five words.
3rd Case: ‘polysynthetic’ languages with the arguments indicated by
AGREEMENT inside the verbal word e.g. Wubuy
ngarra-mani-nyung na-walyi-nyung ngunu-lharrma-yn
F-woman-Hum.Sg M-man-Hum.Sg S.3.f.sg/O.3.m.sg-chase-PP
‘The woman chased the man ’
Also: na-walyi-nyung ngarra-mani-nyung ngunu-lharrma-yn
14
Syntax – Week 7, Lecture 1
Case
The form of a nominal varies according to its grammatical relation. Such
morphological variation is called case or case marking. For example,
nominals in Jiwarli (and in many other languages with extensive case
marking) have a case paradigm (also called a declension). For example,
the following are traditional case labels for different GRs (see readings):
Case label GR
nominative subject
accusative direct object (of verb)
dative indirect object, object of certain
prepositions
ablative location in space or time
genitive possessors
Latin ‘gate’
porta
portam
portae
portā
portae
etc…. (some languages have many more cases than this –
Finnish has 14)
Note that verbs and prepositions are said to govern their complement NPs
and assign case to them – the particular case form assigned may depend
on the specific lexical verb or preposition.
Agreement: the form of the verb varies with the grammatical or semantic
features of one or more of its arguments. Such morphological variation is
called verb agreement.
Verbs normally agree only with core (or term) grammatical arguments and
typically, verb agreement is limited to a maximum of two entities. Verb
agreement normally reflects the features of person, number, and/or
15
Syntax – Week 7, Lecture 1
gender.
Summary — the main ways in which GRs may be encoded are:
● by word order / configurationally
● and/or by case marking
● and/or by verb agreement.
16