Transcript syntax - Université d`Ottawa

SYNTAX
LIN 1310
• Syntax is the component of grammar that
deals with sentence structure.
• Before we can talk about syntax, we need
to discuss the nature of linguistic
competence and the nature of grammar
The nature of grammar
• Linguistic competence:
Theoretically permits infinite production of
novel utterances, including sentences.
How can our brains handle a system
capable of infinite production?
Answer: The elements of the system are
not infinite. They are finite in number.
Grammar
• A system that permits infinite creativity
with a relatively limited number of
elements, including speech sounds and
the rules for combining them into words
and sentences.
• The rules for making sentences are called
the syntax of the language.
• What are these rules like and how are they
stored as part of competence?
The Rules of Syntax
Does knowing the rules of syntax
involve storing the syntactic
structures of all possible sentences in
our heads as a set of templates with
slots to fill?
No, it does not.
Here are for reasons why not:
Reason 1
1. Rules of syntax allow sentences to
be infinitely long. (The only limits are
imposed by performance.)
Reason 2
2. All possible sentences have not yet
been uttered.
Reason 3
3. Our storage capacity is finite.
Reason 4
4. Sentence interpretation is structure
dependent.
That is, it relies on more than a
simple linear (one word after another)
organization.
(This will be explained further.)
The Nature of Syntax
The following slides present a number
of illustrations of the nature of syntax
and facts about language that a
linguistic theory of syntax must
account for.
Facts syntactic theory must account for:
Fact 1
1. How words in a sentence are assigned
specific roles with respect to one another.
• For example, in English active sentences:
The first noun is the subject or doer of the
action or agent.
The second noun is the object or receiver
(or theme or patient) of the action.
Facts syntactic theory must account for:
Fact 1
• Thus different interpretations for:
The dog chased the cat.
subject verb object
The cat chased the dog.
subject verb object
Facts syntactic theory must account for:
Fact 1
• Word order varies across languages.
• In German, it’s subject, object, verb:
Das Kind wird die Schwester lehren.
The child will the sister
teach.
subject
object
verb
Facts syntactic theory must account for:
Fact 1
• For some languages, like Welsh, the
preferred order is verb, subject, object.
• For example:
darllenais i
y llyfr
read
I
the book
verb
subject object
Facts syntactic theory must account for:
Fact 1
• Other languages may rely more on some
form of morphological marking than on
word ordering to assign roles to words in
sentences.
• However, these languages still have a
most common or canonical word order.
• The other permissible word orders tend to
slightly alter the focus of the sentence.
• See Japanese example on the next slide.
Facts syntactic theory must account for:
Fact 1
•
•
•
•
Canonical word order in Japanese:
‘Yumiko scolded the child’:
Yumiko-ga sono kodomo-o sikat-ta
yumiko-NOM that child-ACC scold-past
subject
object
verb
• Alternative order (focus on object):
• Sono kodomo-o Yumiko-ga sikat-ta
• that child-ACC yumiko-NOM scold-past
•
object
subject
verb
Facts syntactic theory must account for:
Fact 1
• The most typical word order in a language
is called its canonical word order.
• 95% of the world’s languages have one of
the following canonical word orders:
• SVO – Canadians like hockey. (English)
• SOV – Canadians hockey like. (German)
• VSO – Like Canadians hockey. (Welsh)
• VOS OVS OSV – rare or nonexistent
Facts syntactic theory must account for:
Fact 2
2. Sentences with different surface
structures and word orders can have the
same interpretation of who’s doing what
to whom, even in a language with strict
word order and no specific marking of
subjects and objects.
Examine the following sentence pairs
from English:
Facts syntactic theory must account for:
Fact 2
scrambled order:
The boy gave the toy to the girl.
direct
indirect
object
object
The boy gave the girl the toy.
indirect direct
object object
Facts syntactic theory must account for:
Fact 2
active:
The student solved the problem.
(agent/subject)
(theme/object)
passive:
The problem was solved by the student.
(theme/object)
(agent/subject)
Facts syntactic theory must account for:
Fact 2
declarative:
Everyone is happy.
(verb)
yes/no question:
Is everyone happy?
(verb)
Facts syntactic theory must account for:
Fact 2
declarative:
You have been there.
auxiliary
yes/no question:
Have you been there?
auxiliary
Facts syntactic theory must account for:
Fact 3
3.Sentences contain discontinuous elements
which are understood as part of the same
structural constituent, as in:
perfect (auxiliary ‘have’ + suffix ‘-en’):
I have eaten.
progressive (auxiliary ‘be’ + suffix ‘-ing’):
I am eating.
Facts syntactic theory must account for:
Fact 3
Wh question:
What are they looking at?
Object
Declarative:
They are looking at the dog.
object
NB: Although ‘what’ has moved to the front of the question
sentence, it plays the same syntactic role as ‘the dog’
does in the declarative. Both are part of the predicate.
Facts syntactic theory must account for:
Fact 4
4. It is possible for sentences to be
embedded inside other sentences.
The property of grammar which permits
such embedding is called recursiveness.
Facts syntactic theory must account for:
Fact 4
Recursiveness also allows sentence
constituents (parts) to occur inside other
constituents. See the string of adjectives
for the subject noun ‘dog’ in the following
example:
The big gray shaggy friendly dog brought me
the newspaper.’
Facts syntactic theory must account for:
Fact 4
In the following examples, note how a sentence
can grow as more strings of adjectives,
prepositional phrases and embedded sentences
are added:
The dog was sleeping.
subject
The big dog was sleeping.
subject ‘dog’ + one adjective
Facts syntactic theory must account for:
Fact 4
The big ugly guard dog at the factory was
sleeping.
Subject ‘dog’ + string of adjectives +
prepositional phrase
Facts syntactic theory must account for:
Fact 4
The big ugly guard dog at the factory which
produces pillows that some people are allergic
to was sleeping on the road that leads to the
quarry where they found the hobo who had been
strangled with a shoelace that was later shown
to have been stolen from the woman on Main
Street whose house was broken into last week.
subject ‘dog’ + string of adjectives +
prepositional phrases + embedded sentences
Facts syntactic theory must account for:
Fact 5
5. Sentences with the same apparent structural
elements and word orders can have different
meanings or interpretations.
1) John is
easy
2) John is
eager
Subject verb adjective
to please
to please
infinitive
On the surface, the structures of sentences 1
and 2 appear to be the same.
Facts syntactic theory must account for:
Fact 5
1) John is easy to please
2) John is eager to please
In sentence 1, John is unspoken/understood
object of ‘to please’
subject = ? (somebody
In sentence 2, John is unspoken/understood,)
subject of ‘to please’
object = ? (somebody)
Facts syntactic theory must account for:
Fact 6
6. A single sentence may have more than one
meaning or interpretation.
This is called ambiguity.
Structural ambiguity:
The two meanings hinge on different underlying
relationships between the words in the
sentence.
Example:
Visiting professors can be interesting.
Facts syntactic theory must account for:
Fact 6
The structural ambiguity of the sentence rests on
the ambiguous structure of the phrase visiting
professors.
[Visiting professors] can be interesting.
modifier noun
Meaning: Professors who have come here
temporarily from other universities.
or
[Visiting professors] can be interesting.
gerund noun
Meaning: For someone to visit professors.
Facts syntactic theory must account for:
Fact 6
Lexical ambiguity:
Alternative interpretations of a sentence rest
on different meanings of homophonous
lexical items.
Example:
She ate her cottage cheese with relish.
‘relish’ = condiment or enthusiasm.
The two words sound identical, but have
different meanings.
Facts syntactic theory must account for:
Fact 7
7. Not all elements of a sentence appear on
the surface, yet listeners can correctly
interpret the sentence.
For example:
Joe likes pizza and Jack does too.
‘missing’ element: ‘likes pizza’
Both like pizza, but ‘likes pizza’ appears
only once.
Facts syntactic theory must account for:
Fact 7
• Pronouns sometimes stand in for ‘missing’
elements:
Mary bought some ice cream and __ ate it.
Mary bought some ice cream and she ate
it.’
If ‘Mary’ and ‘she’ stand for the same
individual, they are co-referenced.
It is also possible that ‘she’ is coreferenced with someone else.
Facts syntactic theory must account for:
Fact 8
8.There are quite strict rules for forming sentences
which speakers must adhere to.
Syntactically ill-formed sentences are
ungrammatical.
*Eats the food the child.
*Mary refused to allow that the children go to the
concert.
✓Mary refused to allow that type of language.
Facts syntactic theory must account for:
Fact 9
9.Syntactic and semantic well-formedness
are independent of one another.
The following sentences are syntactically
well-formed but semantically
anomalous or nonsensical:
*Colourless green ideas sleep furiously.
*A verb crumpled the milk.
Transformational/Generative
Approach
• To account for these and other facts, most
linguists adopt a transformational
generative (TG) approach to describing
the syntactic component of grammar.
• A TG approach emphasizes the search for
Universal Grammar (UG).
Transformational/Generative
Approach
• UG presupposes that languages operate
under a shared set of categories,
operations and principles.
• Although languages obviously differ from
one another, their are common principles
governing the way they form sentences.
• These properties of UG are called
language universals.
Transformational/Generative
Approach
• By using the same descriptive approach
for all languages, linguists hope to identify
language universals and gain a more
complete understanding of UG.
• At the syntactic level, there are clearly two
subcomponents found in all languages:
1. a lexicon
2. a computational system
Transformational/Generative
Approach
• The lexicon or mental dictionary lists the
words (and morphemes) in a language.
• The lexicon also includes information
about each entry regarding:
-pronunciation
-meaning
-form (root, affix, bound, free, etc.)
-syntactic category information? (noun,
verb, etc.)
Transformational/Generative
Approach
• The computational system includes the
operations that allow words to be
combined into syntactic structures.
• The computational system has two major
components:
Merge
Move
Transformational/Generative
Approach
• Merge allows the creation of phrases
(parts of sentences) and the combining of
the these phrases into sentences.
• Move allows certain elements to be
transported to a new position within a
sentence.
• An example of a moved element is the
word ‘what’ moving out of object position
into sentence initial position in:
‘What are they looking at?’
Transformational/Generative
Approach
We will examine the elements of syntax as
follows:
1. The lexicon and the notion of syntactic
categories
2. The creation of phrases (phrase structure)
3. The merge operation to form sentences
4. Types of phrasal & clausal complements
5. Move
Transformational/Generative
Approach
• Transformational Generative theory has
undergone many reformulations over the
past 50 years.
• We are operating with a recent version.
• Thus you may encounter terms elsewhere
that are roughly equivalent to the elements
of syntax that we are examining.
• We will discuss these terms at a later date.
The Lexicon/Syntactic Categories
• There is some controversy in current linguistic
theory regarding whether the syntactic category
(noun, verb, etc.) of a word is stored or
syntactically determined. There are essentially
two positions.
1. This information is stored in the lexicon.
2. The syntactic category is only determined when
the word enters into a syntactic structure.
• We will not resolve the issue in this class. We
will assume position 1, although the reasoning
behind position 2 may be discussed.
The Lexicon/Syntactic Categories
• All languages group their words into
syntactic categories.
• We find remarkably similar syntactic
categories across languages
• Your textbook provides example of
the commonly found categories in
Table 5.1.1 on page 133.
The Lexicon/Syntactic Categories
• The Lexical/Nonlexical distinction is
roughly equivalent to the distinction
we used in morphology:
OpenClass/Content or Lexical Words
versus
ClosedClass/Function Words
or Non-lexical
• The chart is reproduced in the next slide
Non-lexical/functional
(Closed Class)
Lexical
Non-lexical/
functional
The Lexicon/Syntactic Categories
The classification of words into syntactic
categories can rely on:
• the type of meaning they express
• what inflectional affixes they take
• the types of structures they fit into
(distribution)
Classification relies on a combination of
these approaches.
Classifying Syntactic Categories
We will apply these 3 types of
classification in the following slides.
This will hopefully allow you to gain a
better understanding of Syntactic
Categories (aka: parts of speech)
Classifying Syntactic Categories
Meaning
• The meanings of Lexical (Open
Class/Content) words are generally much
easier to define that those of Non-lexical
(Closed Class/Function) words.
• Consider the words:
‘the’ (Non-lexical/Closed Class determiner) and
‘brush’ (Lexcial/Open Class - noun)
‘the’ versus ‘brush’
• It’s easier to explain the function of ‘the’,
or where it occurs (before nouns), than
what it actually means.
• It’s easier to describe the concept ‘brush’
than the concept ‘the’.
Meanings of the Major Lexical
Categories
Nouns: name entities –objects (book)
–people (Mary)
Verbs: designate –actions (eat)
–sensations (feel)
–states (be, seem)
Adjectives: designate properties and attributes of
what nouns name (heavy book)
Adverbs: Denote properties and attributes of
what verbs designate (eat quickly)
Adverbs tell us ‘how, why, where, when’.
Meanings of the Major Lexical
Categories
• Meaning is not always a clear-cut way of
deciding the part of speech of word.
• Some lexical/content words are more
difficult to define.
• For example, the concepts ‘truth’ or
‘honesty’ do not name entities.
Meanings of the Major Lexical
Categories
• Items from two different categories can
have similar meanings.
For example, the verb ‘hate’ and the noun
‘revulsion’ have very similar meanings:
I hate carrots.
I feel revulsion toward carrots.
Meanings of the Major Lexical
Categories
• Some words, like ‘brush’, can be a noun or
a verb.
I bought a new brush (noun) for the dog.
I brush (verb) the dog every day.
Inflections for Determining
Major Lexical Categories
• So we still need to go beyond simple
meaning to define a lexical (content)
word’s syntactic category.
• We can also look at what inflections a
form takes, as inflections usually attach to
a particular part of speech.
• Note the references to parts of speech in
the following list of English inflections.
The inflectional affixes of English
Inflections for Determining
Major Lexical Categories
Thus we can tell if ‘ship’ is a noun or a
verb by seeing whether it takes affixes that
normally attach to a verb or a noun.
‘ship’ + pl. ‘s’ as in:
I like to sail on ships.
‘ship’ + poss. ‘s’ as in:
The ship’s hull is painted red.
So far, it looks as if ‘ship’ is a noun.
Inflections for Determining
Major Lexical Categories
• To be certain, lets see if adjectival
inflections can be attached to ‘ship’.
• Comparative ‘er’ as in:
*That one is shipper than the first one.
• Superlative ‘est’ as in:
*That is shippest one of all.
So ‘ship’ is not an adjective.
Inflections for Determining
Major Lexical Categories
• What about verbal inflections?
• Past tense ‘ed’ as in:
They shipped the package last week.
• Progressive ‘ing’ as in:
They are shipping the package by air.
So, it looks as if ‘ship’ is a verb as well as
a noun.
Inflections for Determining
Major Lexical Categories
• In fact, many words can belong to more than
one lexical category.
• For example:
brush (noun, verb)
comb (noun, verb)
near (preposition, verb, adjective)
• They got bored near the end.
• They neared the finish line.
• The are nearer to the end than us.
Distribution for Determining
Major Lexical Categories
We have seen that neither the meaning
nor the affix test tells us reliably which
lexical category a word belongs to.
Another more reliable way of determining
lexical category is by looking at a word’s
distribution.
Distribution for Determining
Major Lexical Categories
By distribution, we mean the type of
elements that a word can co-occur with.
The most useful type of elements for
distributional analysis are what functional
categories a word can be used with,
although we can also look at what lexical
categories it can be used with.
Distribution for Determining
Major Lexical Categories
For example:
nouns occur with determiners such as ‘the’
‘the dog’
verbs occur with auxiliaries such as ‘will’
‘will go’
adjectives occur with degree words such
as ‘very’
‘very hot’
Distribution for Determining
Major Lexical Categories
• Verbs do not occur with determiners:
*the go
• Nouns do not occur with auxiliaries such
as ‘will’
*will lamp
• Neither nouns nor verbs occur with degree
words such as ‘very’
*very lamp’ *very go
Homework
• Study Guide: Page 104
• O&A: Page 170-171, Ex. 2
Think about how you could have used
distribution, meaning and inflection tests in
order to identify the syntactic categories
(parts of speech) of the underlined words
and words in the sentences of Ex. 2.
Refer to Tables 5.1, 5.2 and 5.3 in O&A for
help.
Distribution for Determining
Major Lexical Categories
• Distributional analysis helps us to see the
difference between verbs and auxiliaries.
• A verb can co-occur with an auxiliary, but a
verb cannot co-occur with another verb.
We may eat.
We have eaten.
We are eating.
*We play eat.
Distribution for Determining
Major Lexical Categories
• Sentences normally have only one
verb.
• So a sentence with more than one verb is
really more than one sentence.
• Such as two conjoined sentences as in
‘They’re moving and grooving.’
From: ‘They’re moving. They’re grooving.’
Distribution for Determining
Major Lexical Categories
Or, two verbs might be a clue that there is
a main sentence containing an embedded
or subordinate sentence or infinitival
complement as in
‘We are going to play.’
‘go’ is the main verb and ‘play’ is the verb
in the infinitival complement ‘to play’.
Distribution for Determining
Major Lexical Categories
• Further distributional analysis can show us
that there are actually three types of
auxiliary in English:
• Modals (will, would, shall, should, can,
could, must, may, might)
• Perfect (have –en)
• Progressive (be-ing)
Distribution for Determining
Major Lexical Categories
• Lets look at modals first.
• Modals can co-occur with verbs:
He will eat. He may eat.
• Modals cannot co-occur with modals:
*He may will eat.
Distribution for Determining
Major Lexical Categories
• You cannot replace a verb with a modal
as in
*’We are maying.’
• Nor can you replace a modal with a verb
as in
*’We wish be eating.’
Distribution for Determining
Major Lexical Categories
• Modals can co-occur with perfect ‘have’.
He may have eaten.
• Modals can co-occur with progressive ‘be’.
He may be eating.
Notice that when modals co-occur with
any of these elements, the modal always
comes first.
Distribution for Determining
Major Lexical Categories
• Lets try a substitution test to determine
whether the following are modals or verbs:
• Try substituting each of the following
words for ‘will’ in the following sentence:
can, play, could, shall, see
• He will eat.
• Which ones are OK?
Distribution for Determining
Major Lexical Categories
• Now try substituting the same words for
the verb ‘eat’ in
‘He will eat’.
can, play, could, shall, see
• Which ones are OK?
• So our substitution test tells us that:
‘can’, ‘could’ ‘shall’ are modals
‘play’ and ‘see’ are verbs
Distribution for Determining
Major Lexical Categories
• Modals always occur before ‘perfect’ and
‘progressive’ (when they are actually
present) and before a ‘verb’.
• Verbs occur after all possible elements in
the auxiliary.
Distribution for Determining
Major Lexical Categories
• How do we know that perfect ‘have -en’
is not a modal?
• Try the substitution test on:
He will have eaten.
• Can you substitute a modal for ‘have’?
*He will may eaten?
• Can you reverse the order of ‘will’ and
‘have”?
*He have will eaten?
Distribution for Determining
Major Lexical Categories
• How do we know that progressive ‘be -ing’
is not a modal?
• Try the substitution test on:
He will have eaten.
*He be having eaten.
Distribution for Determining
Major Lexical Categories
• How do we know that progressive ‘be –ing’
and perfect ‘have –en’ are separate
categories?
• Try reversing them in:
He may have been eating.
*He may be having eaten.
Still not convinced about modals?
• Lets apply the meaning test.
• Modals express the speaker’s attitude or
intentions with respect to the sentence.
Modals can express possibility, probability,
necessity, intention, etc., as in:
• I will go.
• I may go.
• I must go.
Still not convinced about modals?
• Lets apply the inflection test.
• Unlike verbs, modals do not take the
present tense ‘s’ inflection in the 3rd person
singular verbs.
He eats at nine.
*He wills eat at nine.
*He mays eat at nine.
Homework
• We can apply the distribution and meaning
tests to identify perfect and progressive as
distinct functional categories.
• The meanings of sentences with perfect and
progressive are affected in interesting ways.
• Think about the meanings of the sentences with
perfect and progressive on the following slide.
• Be prepared to discuss them in your DGD.
Perfect and Progressive
John has studied karate, but he is more interested in kickboxing.
John had studied karate for ten years before he got his black belt.
I lived in Florida for ten years. Then I moved to Texas.
I have lived in Florida for ten years, but I still miss Canada.
I had lived in Florida for ten years before I saw my first alligator.
I can’t talk while I am driving.
I didn’t use my cell while I was driving.
I am driving to Montreal tomorrow.
Think about the concept of tense in English and what it expresses:
I am exercising right now.
(present progressive - ongoing)
I exercise every day.
(simple present - habitual)
I exercised every day last week. (simple past - completed)
I was exercising when you called me last week. (past progressive - ongoing
in past)
Perfect
• Perfect have-en expresses the duration of an event as in ‘I have
lived in Florida for ten years.’ In the past tense it can also express
the duration of an event with respect to a certain point in the past, as
in ‘I had lived in Florida for ten years when they discovered I was an
illegal alien.’ Perfect have-en can also be used to refer to an event
that took place over some indefinite period in the past, even when
the sentence is actually inflected for the present, as in ‘I have
studied karate’.
• Perfect have-en is rarely used to indicate a completed act. Instead
English uses the simple past tense as in ‘I lived in Florida for ten
years.’ Compare that to the sense of duration in ‘I have lived in
Florida for 10 years.’ Note also that the simple present in English
cannot express duration as in *I know John for ten years. However,
the simple present in English does express habitual actions as in ‘I
eat breakfast at seven.’
Progressive
• Progressive be-ing refers to events in progress in
either the present or the past. In the present, the event is
understood as happening while the sentence is being
uttered. In the past, the event is understood as
happening while another event in the discourse was
happening. Progressive can also communicate
anticipation of an event which will happen as in ‘I am
going shopping later.’ There is a narrative discourse style
in English that uses the present progressive and the
simple present to relate events that occurred in the story,
as in ‘I’m walking down the street last night and this guy
jumps out from behind a parked car and starts yelling at
me (etc.)’
A word about auxiliaries
• English has four different functional
categories that can be described as
auxiliaries:
tense, modal, perfect, progressive
If modal, perfect and progressive are all in
a sentence, they must occur in that
specific order for the sentence to be
grammatical.
We will discuss tense placement later.
From Word Categories to Phrases
• Sentences are made up of sub-parts called
phrases.
• Every language has a set of rules for forming
these phrases.
• They are called Phrase Structure Rules
• The Phrase Structure Rules are a set of rules
that allow us to map out the structures of
phrases in a language.
• We diagram these structures as inverted trees.
From Word Categories to Phrases
• The inverted trees reflect the hierarchical
arrangement of phrases.
• Sentences consist of a series of phrases
also joined together in a hierarchical
manner.
Merge: X’ (X-bar)
Merge is the part of the syntax in which of words
are fitted together into phrases and in which
phrases are joined together to form sentences.
Merge combines words into phrases using the X’
schema that we will be discussing shortly.
The Phrase Structure Rules determining these
phrases used to be fairly elaborate, but in the
current formulation of the theory have been
essentially streamlined to the X’ schema.
Merge: Subcategorization
Phrases are also the product of an
interaction between the X’ schema and the
subcategorization properties of words.
• Subcategorization refers to the type of
complement structures that certain words
must or can appear with.
‘to hate’ (verb) requires a direct object
*I hate.
I hate carrots.
I hate carrots.
Merge: D-structure
Merge results in what used to be called
the Deep (D) Structure of a sentence
The D-structure of a sentence very closely
resembles the canonical word order in the
language.
D structure is not, however, the final form
of the sentence.
Move
Not all sentences have the canonical
word order in the language.
Thus another component of syntax, called
Move, moves elements to where they
belong in the surface structure or Sstructure of the sentence.
• Compare:
• John is eating an apple. (canonical)
• Is John eating an apple? (non-canonical)
Why D- and Surface Structure?
• We just saw that not all sentences follow
canonical word order.
• But why not allow Merge to create these
structures in the first place?
• Why do we need Move?
• I will provide some reasons in the next few
slides and point out others as we explore
the syntax more thoroughly.
Why D- and Surface Structure?
• By adhering to strict canonical word order, Dstructure gives us important information about
the semantic roles of the elements in a
sentences, especially the main nouns with
respect to the verb.
• Thus, for English, the first Noun Phrase
constituent in the tree represents the subject of
the sentence.
• The subject of the sentence has a special
relationship to the verb, as it is the doer of the
action or agent.
Why D- and Surface Structure?
• The first Noun Phrase immediately
following a transitive verb is its direct
object or theme.
• In semantics, these relationships of nouns
with respect to verbs and their subjects
and objects are called thematic roles.
The students read the book.
agent
theme
Why D- and Surface Structure?
• The thematic roles of the constituents of
the sentence would not be so apparent at
D-structure if non-canonically ordered
trees were permitted.
• The move component of the syntax can
operate on D-structure trees to create
these non-canonical orders as required.
Why D- and Surface Structure?
• At D-structure it is clear when two
sentences have the same thematic
relationships between their words.
• Recall the sentences we looked at a few
classes ago:
Why D- and Surface Structure?
declarative:
Everyone is happy.
(verb)
yes/no question:
Is everyone happy?
(verb)
Why D- and Surface Structure?
Wh question:
What are they looking at?
object
theme
Declarative:
They are looking at what.
object
»
theme
Why D- and Surface Structure?
scrambled order:
The boy gave the toy to the girl.
direct indirect object
theme goal
The boy gave the girl the toy.
indirect direct
goal
theme
Why D- and Surface Structure?
• By having the alternative word orders created by
Move, there is no need for extra phrase structure
rules that would create essentially the same
structure in two different places in the tree.
• So we don’t need a rule in Merge permitting a
verb to occur at the beginning of a tree as well
as after the subject:
Is everyone happy? Everyone is happy.
• We also don’t risk confusion about whether the
role of the nouns ‘everyone’ and ‘what’ in the
question sentences.
A schematic of syntax
Merge
Deep Structure
Move
Syntactic Surface Structure
Is that all there is to the grammar?
• In the previous slide we saw that Merge creates
D-structure which is acted upon by Move which
in turn creates Syntactic Surface Structure.
• Is that it for the grammar? No.
• The syntactic surface structure does not
represent the final spoken form of the utterance.
• The rules of phonology and semantic
interpretation (logical form) must apply before
the sentence is fully derived.
From theory to practice
• Merge is the product of the X’ schema.
• This basically sets out the structure of a
typical, generic phrase as:
»
XP
X’
(Specifier)
X
(complement)
From theory to practice
The abbreviations in the generic X’ phrase structure tree
X=part of speech of the head of the phrase
P=phrase (e.g. NP = noun phrase)
Specifier=phrase boundary marker, makes
meaning of head more precise
Complement=phrases which provide
information about the meaning of the head. The type
of complement a head can take is part of the
information stored with it in the mental lexicon.
NB: Parentheses ( ) mean that an element is optional in
some phrases and with some heads.
Types of X - Heads
• Heads can be: Nouns, Verbs, Adjectives,
Prepositions
• At the sentence level, heads are called:
I = Inflection C= Complementizer
We will discuss sentence heads later.
Specifiers (abbreviations in bold)
• The following can act as Specifiers:
Determiners (for nouns N):
e.g. ‘the, a, this, these, those, no,’
Adverbs (for verbs V):
e.g. ‘never, perhaps, often, always
Degree word (for Adjectives A, Prepositions P)
e.g. ‘very, quite, more, almost’
Complements
• Phrases that add information about
entities and locations implied by the head
and for which the head is subcategorized.
• In the following example, the NP ‘carrots’
is the complement of the verb ‘hate’.
‘to hate’ requires a direct object complement
*I hate.
I hate carrots.
Other Complement Examples
•
1.
2.
3.
4.
•
•
•
prepositional phrases as in:
I often eat [at that restaurant].
I never approved [of that purchase].
I was certain [of his loyalty].
The destruction [of the city] angered me.
1 and 2 show verb complements,
3 shows an adjective complement.
4 shows a noun complement.
Practice drawing phrase structures
• Keep in mind that all phrases follow the x’
schema.
XP
X’
(Specifier)
X
(complement)
He often reads tons [of books]
He often reads tons [of books]
He [often reads tons of books]
How to recognize a phrase
• Look for typical heads, such as:
noun, verb, adjective, preposition
• Work from right to left, since English
branches right and your lowest phrases in
the tree will likely be found to the right.
• Consider possible structures of phrase
and see how X’ applies
Complements
• N B: See O&A Charts 5.5 (p143), 5.6, 5.7,
5.8 (p144) for types of complements
Verb Complements
Noun, adjective, preposition complements
Phrase Tests
• You can also try the phrase tests of:
substitution
movement
coordination
Substitution Test
• nouns can be replaced by ‘they’ ‘it’
[The boys] bought [a firecracker].
• verbs can be replaced by ‘do so’
The children can [play].
• prepositional phrase can be replaced by
‘there’
He went [to the bank].
Movement Test
• prepositional phrases can often be moved
without compromising the grammaticality
of the sentence
The children sang [in the chapel].
[In the chapel] the children sang.
Coordination Test
• a phrase can be conjoined with another phrase
of the same type using ‘and’
They [mowed the lawn].
They [mowed the lawn] and [raked the leaves].
Sally saw the car on the hill.
- ‘hill’ is a noun, so it must be part of an NP
- ‘the’ is determiner, so it must be the
specifier of the NP ‘the hill’
Sally saw the car on [the hill].
Sally saw the car on the hill.
- ‘on’ is a preposition, which is the head of a
PP
- PPs have NPs as their complement, so
‘the hill’ is the complement of the PP with
‘of’.
Sally saw the car [on the hill].
Sally saw the car on the hill.
- ‘car’ is a noun, so it must be the head of
an NP
- ‘the’ is a determiner, so it must be the
specifier of the NP with ‘car’
- ‘on the hill’ is a PP which acts as a
complement to ‘car’
Sally saw [the car on the hill].
Sally saw the car on the hill.
-’saw’ is a verb, which acts as the head of a
VP
-’saw’ is subcategorized for an NP
complement
-the NP complement of ‘saw’ is ‘the car…’
Sally [saw the
car on the hill].
Sally saw the car on the hill.
’Sally’ is a noun, which is the head of an NP
[Sally]
saw the
car on
the hill.
But how does it all fit together?
Using X’, the theory treats the subject NP
as the specifier of a new phrase called IP
The theory treats VP of the sentence as the
complement of this phrase.
The head of the phrase is the tense
inflection of the sentence I.
Remember that English has only two tenses:
past (+pst) present (-pst)
Sally
saw the car
on the hill.
Try: The dog bit the cat.
The dog bit the cat.
IP
IP allows us to deal with the tense and
modal parts of the auxiliary of the
sentence.
The remaining parts of the auxiliary
(perfect ‘have’ and progressive ‘be’)
are handled a little differently, as we will
see.
Modals in IP
• The theory treats models has the lexical part of
I of the IP.
• This is because models have inherent tense.
• We never add a tense inflection to make them
past or present.
He will eat. (-pst, NB: no ‘s’ in 3rd pers. sg.)
He would eat if he had money (-pst)
He would eat whenever anyone fed him. (+pst)
Modals in IP
• You judge the tense of the modal based on
the rest of the sentence or discourse.
• If there is a modal, the verb is not marked
for tense.
The dog will bite the cat.
The dog bites the cat.
Modals in IP
• Note that if there is no Modal, the tense in
the PS tree matches the inflection on the
verb.
• This may seem bizarre, but syntactic
theory has had a long-standing conflict of
over whether and how to handle
inflectional morphology.
• Compare:
Modals in IP
• Any modal could take the place of ‘will’ in ‘The
dog will bite the cat’.
• The dog will bite the cat.
• The dog would bite the cat.
• The dog may bite the cat.
• The dog can bite the cat.
• The dog could bite the cat.
• The dog should bite the cat.
• The dog must bite the cat.
• The dog might bite the cat.
• The dog shall bite the cat.
Perfect and Progressive
• If the sentence contains the auxiliaries
perfect or progressive or both, these are
treated as verbs.
• This will give us successive VPs as
follows:
1
2
Modal, Perfect and Progressive
• In ‘The dog has bitten the cat’, note that
tense is on perfect ‘have’, not on the verb.
• Tense with perfect ‘have’ is relative to the
other events in the discourse.
• Compare a sentence with modal and
perfect.
• The dog will have bitten the cat.
• Here tense is expressed through modal
‘will’, rather than on ‘have’.
Modal-Perfect-Progressive
• In older versions of the theory, the affixes
associated with perfect ‘have’ (-en) and
progressive ‘be (-ing) were shown together with
them as unattached inflections in the PS.
• Movement took place to attach them where they
belonged.
• Nowadays, syntacticians do not want
morphological operations within syntax.
• Thus they are glossed over in the PS trees.
Whole Sentences as Complements
• What if our complement is a whole
sentence embedded in the bigger
sentence.
• We use the term clause for sentence.
• Hence – embedded clause.
• A clause must have a verb.
Embedded Clauses
• I know [that Mary has left].
• Compare it to:
• I know [the answer].
• The [ ] structures are both complements
of the verb, but one is a clause (aka:
sentence).
• The embedded clause acts like an NP.
Embedded Clauses
• Embedded clauses that replace NPs are
marked by the words ‘that’ or ‘whether’ or
‘if’.
• These are called complementizers.
• They form the head of a CP
(complementizer phrase) that goes into
the complement position of the VP.