Linguistics322

Chapter 3

Constituency

Final Rule form of constituents

NP > (C) TP

TP >{NP/CP} (T) VP

VP >(AdvP+) V (NP)({NP/CP}) (AdvP+) (PP+) (AdvP+)

NP > (D) (AdjP+) N (PP+) (CP)

PP > P (NP)

AdjP > (AdvP) Adj

AdvP > (AdvP) Adv

Chapter 4

==

Chapter 6

X-Bar Theory

Introduction

The theories described so far in terms of sentence structure have some inadequacies on close inspection of sentence structure across languages.

Observe:

That big book of poems with the blue cover.

The NP rule NP > NP (AdjP+) N (PP+) assigns this structure to the sentence.

[NP[D the][AdjP [Adj poor]][N man][PP from mars][PP with the red hair]] is on the earth.

This is a flat structure with both PP’s on the same hierarchal level and without any distinction of dominance or c-commanding. So they are flat in respect to [man].

When looking at constituency our tests can create a far more complex structure.

Look at the replacement test:

I saw the poor [man from mars with the red hair] not the rich [one]

Man from mars with the red hair can be replaced by one-replacement but it is not a constituent.

Also one replacement can also find other units that are not constituents

I saw the poor [man from mars] with the red hair not the rich [one] with the green hair

[NP[D the][N’1[AdjP [Adj poor]][N’2[N’3 [N man][PP from mars]][PP with the red hair]]]]

N bar 3 is targeted by [one] in the one replacement of [man from mars] and the [one] replacement of [man from mars with the red hair] targets N bar 2.

 If the [D the] is replaced by [D that] the NP can be modified to account for N bar 1

[NP[D this][N’1[AdjP [Adj poor]][N’2[N’3 [N man][PP from mars]][PP with the red hair]]]]
[not [that one]]

Conjunction also holds to this theory:

Bob is the poor man from mars and the man with the red hair

Get me the man and the dog now

So N' are necessary to explain the things in the sentence that are combined

Bar Level Projections

Here is a revision of the NP rule:

NP > (D) N’
N’ > (AP) N’ or N’ (PP)
N’ > N (PP)

This creates the N bar Node replacing the immediate constituent that [one] replaces.

NP > (D) N’ generates the top NP

It has daughters D and N’

N’ (AP) generates N’1 N’(PP) generates N’2

N’N(PP) generates N’3 as N and PP sisters.

So one replacement is a general process that targets N’ nodes.

Which means that every N’ can be replaced by [one]

This creates binary branching and N’ rule 1 is self recursive

V-bar

In verb phrase syntax a do-so /did-so test (similar to one-replacement) can produce a similar effect

He [made coffee with hot water]

No constituent groups the V and the NP without the PP here but do-so replacement can target this group.

[He [made coffee] with hot water but she [did so] with cold water]

By replacing did-so with a V’ and using revised VP rules:

VP > V’
V' > V’(AdvP)
V' > V'({AdvP/PP})
V’> V (NP)

This creates a structure [VP[Vbar[Vbar[V][NP]][PP]]]

VP >V’ generates the VP and the V’ under it

V’> V’ (PP) opens the first V’ to include another V’ and a PP

And the last V’ generates a V and an NP

So the do-so replacement rule (replace a V’ with (do so/did so/ did too) targets the lower of the two V’ and replaces it with do so

The higher V’ is replaced when sentences such as

Tony [ran water with the hose] and Bob [did so] too

This includes the higher V’ which includes the V the lower V’ the NP and the PP

Also conjunction works in the same way to show instant V’ projection

The house [holds people] and [covers the inside] from the weather

Chapter 7

Extending x-bar Theory to functional Categories

Introduction

X bar theory accounts for an intermediate structure and generalized patterns across categories. It focuses on:

i)	Specifier rule: XP > (YP) X’ or X’(YP)
ii)	Adjunct rule X’> X’(ZP) or X’ > (ZP)X’
iii)	Complement rule X’ > X(WP) or X’ > (WP)X

These rules generate trees across languages and obtain additional properties of hierarchy that we can see within the main constituents.

However what about the status of specifiers?

 The rule states XP > (YP) X’ or XP> X’ (YP)

This means that the specifier needs to be a phrase at the XP level.

So far determiners have been studied that do not seem to be phrasal!

So a new category (called the determiner phrase (DP)) needs to be adopted to be consistent with X’ theory.

Also there are CP and TP rules that need to be addressed.

CP > (C) TP
TP > NP (T) VP

These particular rules do not fit into X’ theory

iv)	Specifier rule: XP > (YP) X’ or X’(YP)
v)	Adjunct rule X’> X’(ZP) or X’ > (ZP)X’
vi)	Complement rule X’ > X(WP) or X’ > (WP)X:

Where the only mandatory element is the head.

But in:

CP > (C) TP
TP > NP (T) VP

The only optional categories are the heads.

So how can they be modified to fit?

Determiner Phrases (DPs)

Determiners like the, a, that, this, those, and these so far have been categorized at the front as specifiers of the NP

This book
Those planes

This is not consistent with X’ theory which states that:

All non-head material must be phrasal.

Determiners are definitely heads in English so they do not seem to be phrasal

*Those these children

In an NP the only this that is not a phrase is the head N so we could say that the D is not in the NP but in its own XP.

This takes determiners out of the NP and makes the NP a complement to the head Determiner.

To show evidence for this we can observe the genitive NP which is the possessive NP. There are two kind of possessive NP’s:

Free genitive or of-genitive
The layer of the cake
The hands of the clock
The letters of the alphabet written in Arabic

These use the preposition [of] to mark the relationship of the possessor For DP evidence there is a more important possessor relationship:

This is the construct or the ‘s-genitive
The cake’s layer
The clock’s hands
The alphabet written in Arabic’s letters

This construction shows the possessive [‘s] coming after the entire possessor It attaches to the whole phrase:

[The alphabet written in Arabic]’s letters
*the alphabet’s written in Arabic letters

This shows that [‘s] is not a suffix but actually it is a minimal word.

Also it cannot appear next to determiners because they are in complementary distribution.
*The cake’s the layer 	meaning 	the layer of the cake
*the clock’s the hands	meaning	the hands of the clock

• the alphabet written in Arabic’s the letters

Meaning 	The letters of the alphabet written in Arabic

Also the [‘s] genitive (unlike the of-genitive) cannot generate a phrase where both the nouns have a determiner so they must be instances of the same thing (this is what complementary distribution means).

So if [‘s] is a determiner and the DP theory is true then we can show evidence that accounts for the positioning of the [‘s] in a sentence relative to the possessor.

The [‘s] sits in the place of the head D and the possessor is in the specifier position.

This still shows a specifier but it is the DP possessor rather than the D’

This opens up a situation where specifiers can associate with subjects of various kinds of constituents.

A Descriptive Tangent into Clause Types

The main properties of a clause are a subject and a predicate phrase.

The simple sentence is a very obvious example.

A bird flew
An apple fell
Tony is a football fan

A root, a matrix, or main clause always stands on it own however there are some examples of clauses that are inside clauses.

[Tony knew [that Bobby fell]]
[Maggie said [she wants to go]]

These are called embedded clauses or subordinate clauses.

As well as subordinate clauses there are also complement clauses, adjunct clauses and specifier clauses.

Complement clauses CP’s are sister to the Verb

Tony showed [that my house is for sale]

The main clause contains the subordinate clause.

Colin knows [if Jonny stole the biscuits]

Adjunct clauses can come in the form of relative clauses

[The apple [with a worm inside]] fell to the ground.

[with a worm inside] is a relative clause and it modifies the head [apple].

Specifier clauses serve as the subject of a sentence.

[[Soldiers on the battlefield] gave their lives for our freedom]
[[With my entire mind] I thought of an idea]
[[For Peter to eat too much] means he might become unhealthy]

So:

There are two different types of clauses:

Main clauses and embedded clauses

There are three kinds of embedded clauses:

The three types are complementizer clauses; adjunct clauses; and specifier clauses.

Clauses are classed into two different types which is dependent on whether they are tensed or not.

A clause with a predicate verb that is tensed is called a tensed clause or a finite clause.

A clause with a predicate verb that is not tensed is called a tenseless or non-finite clause.

I knew [that bobby ate my dinner] is a finite clause
I know [Bobby is eating my dinner] is a non-finite clause.

You can use a number of different tests to differentiate finite and non-finite clauses.

I know [you Speak for yourself] finite
I’ve never seen you [speak for yourself] non finite
I know you spoke for yourself 		finite
*You don’t ever spoke for yourself		non finite

This shows that finite clauses allow tensed morphology but non finite clauses do not This also can be seen by using an –s suffix on the verb

I know he plays football 		is a finite clause
*I’ve seen him plays football		is a non- finite clause and is ungrammatical.

	Nominative		Accusative		Anaphoric
	Singular	Plural	Singular	Plural	Singular	Plural
1st person	i	we	me	us	myself	ourselves
2nd person	you	you	you	you	yourself	yourselves
3rd person Masculine	he	they	him	them	himself	themselves
3rd person Feminine	she	they	her	them	herself	themselves
3rd person neuter	it	they	it	them	itself	themselves

If you look at the case of a noun you can usually tell if the clause is finite or not.

Case means that the subject pronoun is either nominative or accusative.

The subject pronoun in a finite clause will take the nominative case

I said [I] want to play football
I said [he] wants to play football
I said [they] want to play football

The subject pronoun in a non-finite clause will take accusative case:

I want [him] to play football
I want [them] to play football
He wants [me] to play football

If the subject is obligatory then most of the time the clause is finite.

If the subject is optional or not possible then the clause is non-finite.

I know he plays football 	these are finite
* I know plays football		needs a subject
We want (him)(her)(them) to play football
We want to play football		the accusative can be left out

Observing a complementizer can also determine if the clause is finite or non finite.

[for] is only found with non-finite clauses

[if] and [that] are only found in tensed clauses

I’ll play football if he plays football	finite
I’m sure that he wants to play football
[for our game of football] we need [him/her/them] to play
*[for our game of football] we need [he/I] to play

A final test that can be used to understand that finite and non-finite clauses take various types of T units.

The T in a tense clause can have items such as auxiliaries and modals

Will/can/must/may /should /shall/is /have

But the only auxiliary that can occur in a non-finite clause is [to]

We are fairly certain [he will play football]
We are fairly certain [he can play football]
We are fairly certain [he must play football]
We are fairly certain [he should play football]
We are fairly certain [he shall play football]
They want him to play football	non-finite
*They want him will play football
*They want him can play football
*They want him must play football
*They want him should play football
*They want him shall play football

This shows the big difference between finite and non finite clauses and helps to show the distinction in the structure that the T produces.

If the clause has tense then it must be [+- past] or [+-future]

If the clause is non-finite these features do not exist

So clauses are a subject and a predicate phrase.

There are root/main clauses and embedded clauses.

Embedded clauses have complement clauses, adjunct clauses, and specifier clauses.

Clauses can also be finite or non-finite.

Complementizer Phrases (CPs)

TP rules and CP rules stand out as they do not fit X’ theory because the head in TP and CP rules is not always obligatory.

CP > (C) TP
TP > DP (T) VP

To change these rules into X’ theory is fairly simple.

There is a C as the head (keeping with X’ theory) an Obligatory TP complement, and an empty specifier position.

The question is: Do all clauses have CP’s?

Looking generally is seems like only embedded clauses have CP’s:

Tony said that students like to play football
*that students likes to play football
*that student like to play football
Tony said that student likes to play football < CP or no CP?

There is some evidence for the above sentence to have a CP with an obligatory DP

It is called a null complementizer and it has an empty set ∅

Questions that can be answered with yes no or maybe such as

Did you play football?
Have you seen the football?

Require some form of the verb do/does/did to be inserted prior to the subject.

Or the subject and the auxiliary can be inverted

[you have seen the football]
Have you seen the football?

the auxiliary can be inverted [you have seen the football] Have you seen the football? Languages like English which use subject aux inversion do not have special complementizer particles like Irish does.

In Irish to indicate a yes/no question a particle (called Ar) is inserted into a position prior to the verb

Q fall John?
Did john fall?

In English the opposite hold true.

If a language has sub/aux inversion it does not have complementizer particles and if it has complementizer particles then it will not have sub/aux inversion.

Another case of complementary distribution.

In both English and Irish there is a complementizer particle but in English it has nothing there so it is a null complementizer.

There is no phonological form but somehow it needs to have sound when [Q+] so in English the T is moved to the C head.

TP DP You T can VP come
CP C’C Can TP DP you VP come?

This is in correct order and the aux is before the subject.

Languages like Irish don’t do this instead they fill the [+Q] with the particle (such as Ar/An).

English actually does have an overt [+Q] complementizer that is found in embedded questions and if it is present then sub/aux inversion is never present.

Tony wondered if Shaw had hired new employees.
*Tony wondered if had Shaw hired new employees.
*Tony wondered had if Shaw hired new employees.

Because [if] occupies the[+Q] complementizer then sub/aux inversion does not work.

So we can see that in questions there are complementizers and CP’s present even in main clauses due to the fact that in other languages there are overt root complementizers and that sub/aux inversion patterns which are like these overt root complementizers.

Another step in the argument for complementizers being present is that only identical categories can be conjoined.

So if a sentence that shows sub/aux inversion has a null complementizer and that sentence can conjoin to a non question sentence (a statement) then that statement must also have a null complementizer and CP.

In fact you can conjoin statements with questions.

Peter has money to buy a football but will he get one?

This argues that null complementizers are attached to root clauses and are on top of every clause.

Tense Phrases (TPs)

The other rule that does not fit into X’ theory is the S rule TP > DP (T) VP

Where S is replaced by TP at the top the DP sits in [Spec,TP] and the VP is the complement.

This shows a definite instance where the specifier is seen as the subject.

But our problem is that the head of the TP (which is T) is shown to be optional!

In X’ theory heads are mandatory.

What happens in clauses without auxiliaries?

Have/has/had.am/is/are/was/were,do these would normally occupy the head of TP T.

Does the TP suddenly not exist so no T is needed?

Here is an observation

Tense inflection on a verb is in complementary distribution with auxiliaries.

If one is present the other cannot be.

Tony cleans the gun
Tony is cleaning the gun
*Tony is cleaned the gun
*Tony is cleans the gun

Therefore tense inflection is in the same category as auxiliaries.

So T is both auxiliaries AND the inflectional ending on the end of a verb.

Coordination can show this as well as you can only conjoin two items that are in the same category.

Tony cleaned the gun and must shoot the target now

This shows that –ed and had are somewhat the same thing they both involve a T node.

But auxiliaries appear to the left of verbs

Must shoot

And tense inflections appear on the right of the verb

Clean-ed

Also auxiliaries can undergo sub/aux inversion

He must shoot the target
Must he shoot the target?

Can inflectional suffixes? NO! So how is this dealt with?

Here’s a theory

Let’s say that inflectional suffixes and auxiliaries are both created under T

They are different in the fact that auxiliaries can stand alone but inflections need to attach to something (here the verb).

Just like we could move T to C to get the null [+Q] pronounced the suffix cannot be pronounced like an auxiliary so it needs to move for it to be pronounced correctly.

Here is the case of T affix lowering.

The suffix generated in T moves down to join on to the end of the verb so it can be pronounced The motivations for these moves are morphological.

Aux T moves up to C null [+Q] to get pronounced and inflectional T moves down to the verb to be pronounced.

Chapter 8

Constraining X-bar Theory:The Lexicon

Introduction

X’ theory that has been seen so far can accommodate trees for pretty well any language however simply relying on the rules that have so far been determined by X’ theory can produce ungrammatical sentences and sentences that are unacceptable in any language. X’ theory states that complements are optional but when looking at direct objects (the main receiver of the action), direct objects are complements but in many cases they are obligatory and in many cases they are not allowed and in some cases the sentence has to have a direct object and an indirect object!

Tony loves football
¤Tony loves		>needs a direct object
Tony played
Tony blinked
¤Tony blinked the bus
Tony put the ball in the net (a goal)
¤Tony put the ball
¤Tony put in the net (a goal)

This is because certain verbs semantically NEED objects and other verbs do not depending on the verbs property within our stored mental lexicon.

Some Basic Terminology

It is known that different verb types take different numbers of arguments dependent on intransitive, transitive and ditransitive verbs. Intransitive verbs take one

 Tony left

Transitive verbs take a DP subject and a DP object

  Tony opened the door

Intransitive verbs take a DP subject a DP object and a ditransitive verb take a DP subject a DP object and four different types of indirect object depending on the verb.

Tony spared DP peter DP his life
Tony put DP the ball PP in the net
Tony gave DP peter DP the ball/ Tony gave DP the ball PP to peter
Tony told DP Jane DP he cared
Tony told DP his story PP to the newspaper/
Tony told DP Jane CP that he cared

There are also some restrictions based on semantic types and where in the sentence certain things can appear.

#the oven played on the swings
#The oxygen spoke to the dying pencil-case

These are examples of selectional restrictions

Thematic Relations and Theta Roles

Thematic relations are a way to deal with encoding selectional restrictions. These relations are specific terms that describe how the subject and the predicate are used with each other.

i) An agent is the initiator of an action. Usually the agent is the subject but agents can show up in different places within the sentence.

Tony kicked the ball
Tony embarrassingly spoke out of turn

ii) An experiencer can be the subject or the object of the sentence or other argument positions.

Tony likes football
Tony watched the football match
Defenders are Tony’s worst enemies

iii) Themes are entities that are changed somehow either by being moved, or experienced, perceived or just are subject to an action.

Tony dropped his backpack
The football skidded along the field
The prime minister said his speech

iv) A goal is where the entity in the sentence is moving towards

Tony went to the laundry room
Tony received a present.

v) Recipients are a unique type of goal that can only be used with verbs that show possessions going form one thing to another.

Tony put the book on the table
My dad gave me some ice cream

vi) The source is the opposite of a goal: this is the thing from which the motion begins

Tony put the book on the table
Tony came home from football practice

vii) The location is the place where the action takes place

Tony is on the football field
They are having a beer at the restaurant

viii) The instrument is the thing that is being used to do the action

Tony smashed the window with a rock
The molars chewed the food

ix The beneficiary is the entity that benefits from the entire event.

Tony gave a mothers day card to his Mum 
The priest said a prayer for the congregation

DP’s can have more than one of these thematic relations

Tony shot a bird with his gun

There is not a one to one relationship between thematic relations and arguments So in linguistics a theta role θ has been created as a special foundation that maps one-to-one with arguments. These are a collection of thematic relations that home in on one argument. So Tony gets two thematic relations (the agent and the source) but is given only one theta role which is a container for the agent and the source.

So to make it clear a theta role θ refers to a cluster of thematic relations but an individual thematic relation would be (for example an agent).

As an example: the ditransitive verb put takes a subject (the agent) a direct object (the theme) and an indirect object (the location).

Tony(agent) put the box (theme) in the cupboard (location/goal).

If for any reason these are changed the sentence is ungrammatical.

¤put the box in the cupboard
¤Tony put in the cupboard
¤Tony put the box
¤Tony put the box the wig in the cupboard
¤The shoe put the air with the cupboard
¤Tony put the box the cupboard

Having too many or too little amounts of arguments = ungrammatical! Thematic roles count on the semantic definitions of the agents etc. shoes cannot put air in cupboards Argument category is important too (only like categories can be conjoined).

The goal of the verb put must be a PP in the cupboard, on the table, under the bridge,

So it requires 3 arguments Which have the theta roles DP agent DP theme PP goal/location. This can be put into a model called a theta grid

Using put as the predicate

Source/Agent DP	Theme DP	Goal PP
i	j	k

The three columns and two rows represent the three arguments the predicate verb put requires.

In each column can be more than one thematic role but only one theta role. Row 1 shows the thematic relations Row 2 shows the index each theta role has

Simply: an argument may have more than one thematic role say  : Source/Agent
But only one theta role index i

This grid can now be applied to the previous sentence

Tony i put [the box] j [in the cupboard] k

Theta roles always come in two different types.

1)The external theta role

This theta role is always assigned to the subject and is underlined in the theta grid.

2)The internal theta role

These are assigned to the object and / or direct object

As a side point there is a semantic reason for the distinction between internal and external theta roles.

The main point is to indicate the subject type by underlining it.

Adjuncts do not count as theta roles as they are entirely optional even though they do have a thematic relation in the sentence.

The reason they do not count as theta roles is because they are not arguments only complements are arguments.

Constraints are necessary to stop the over generation of X’ rules. These are like filters they look at the rules and push aside ones that to not meet the constraints requirements.

By letting X’ theory over generate and produce ungrammatical sentences we can see which sentences are ungrammatical and filter them out.

This filter is called:

'The theta criterion' The theta criterion makes sure that there is a correct match between the amount and types of arguments in a sentence and makes sure it is the same amount that the grid has.

So the rules of the theta criterion are:
1)Each argument is assigned one and only one theta role.
2)Each theta role is assigned to one and only one argument

This then creates a definite one-to-one match between argument DP’s and theta roles because there cannot be more arguments than theta roles and there cannot be more theta roles than DP’s plus theta roles show semantic appropriateness so the arguments must match the semantic categories in the sentence for the filter to let them pass through.

Examples PLAY

Experiencer DP	Theme DP

If the predicate "to play" is produced in a sentence an index is applied to each argument and the index is then matched to the theta roles in the grid.

Tony i plays football j

Experiencer DP	Theme DP
i	j

If the sentence is ungrammatical like:

¤Tony plays

There is no theme argument here so the grid lacks an index

Experiencer DP	Theme DP
i

So the condition 2 of the theta criterion is violated

Each theta role is assigned to one and only one argument

It is not a one to one match so the sentence is filtered out which is the same as saying it is ungrammatical even though X’ rules allow the generation of this sentence.

Oppositely if the sentence has too many arguments:

• Tony i plays football j hockey k

PLAYS

Experiencer DP	Theme DP
i	j	k

There are three arguments but two theta roles violating the criterion a) Each argument is assigned one and only one theta role.

The output of X’ rules can be constrained using theta roles as a semantic tool.

This is a type of filter that removes ungrammatical sentences and the requirement is that there is a one to one relationship between DP arguments and theta roles.

The Lexicon

We now have a model of grammar with 3 standard rules called the X’ rules.

These can generate a hierarchical constituent structure.

The rules are constrained by a filter called the theta criterion which uses the semantic notion of theta roles.

Are these rules actually stored in the mind?

As per Chomsky the mind has a computational component which contains all the rules and constraints which builds sentences and filters bad ungrammatical sentences out.

For the computational component to have the necessary information about the theta roles and other meanings the mind also requires a lexicon which is the dictionary that is stored in the mind. This is where the theta grids are stored.

So information that is necessary for understanding what each predicate means down to each particular word or lexical item) is held in the lexicon.

This lexicon contains entries concerning irregularities of any language, and things about the language that need to be memorized.

Also the lexicon needs:

1)	The meaning of words
2)	The parts of speech
3)	How the word is pronounced
4)	Irregulars
5)	The theta grid or argument structure

Once a word is learned all of these are memorized.

As you can see the lexicon provides information to the computational component that does the combitorial and generation word to make sentences.

The projection principle accounts for the fact that lexical information affects the form of the sentence.

This means that lexical information like theta roles is syntactically represented at all levels

Expletives and the Extended Projection principle

Some predicate verbs such as “weather” verbs don’t work quite the same way as they don’t on the surface assign a theta role.

It hailed
It rained
It snowed

How is the pronoun IT represented by theta roles?

The IT does not appear to refer to anything at all and it is thus a special type of pronoun called an expletive or pleonastic pronoun. They do NOT get a theta role which is counter-intuitive as it violates the theta criterion but in the grid it is unassigned and empty.

Another class of predicate verbs that take an expletive pronoun optionally take a CP subject.

[ CP That Tony will master this subject] is probable.

This predicate “is probable” assign one theta role takes one argument which is the clause.

[ CP That Tony will master this subject]

and clause notation in a theta grid is as follows.

Proposition CP

The theta role is not underlined because the clause bearing this theta role is a complement

It is probable [that Tony will master this subject]

which also contains an expletive IT!

Most of the time expletive pronouns appear in subject position on the occasions that it is in a different position it will most likely have a theta role.

I won it (theme)
I walked over to it (location/goal)

Expletives tend to come up where there is no theta marked DP (or CP) which fills the subject position this is an encoded revision of the principle.

The projection principle accounts for the fact that lexical information affects the form of the sentence.

This means that lexical information like theta roles is syntactically represented at all levels.

Called the Extended projection principle (EPP)

The Extended projection principle (EPP)
All clauses must have subjects.

The EPP is a constraint filter on the output of the X' rules. It requires every sentence to have a subject.

So the rules of the theta criterion are:
1) Each argument is assigned one and only one theta role.
2) Each theta role is assigned to one and only one argument

A method for this is to say that expletives are not generated by X’ rules but by a special expletive insertion rule: That is:

Insert an expletive pronoun into the specifier of TP

This only ever applies when there is no other subject:

Should there be no subject with a theta mark and no expletive subject the EPP will constrain the sentence and discard it as ungrammatical.

To ensure consistency the expletive insertion rule comes after the theta criterion.

The fact that the expletive insertion rule comes after the theta criterion means the theta criterion will not filter the expletive insertion rule out!

Noteworthy: The two types of IT pronouns are neuter pronouns

It saw me coming in the night
And weather predicate expletives
It rained

Chapter 9

Head to Head Movement

Introduction

X’ theory shows us that an object is the complement to Verb (sister to V and daughter to V’ a theory that blocks specifiers and adjuncts from getting in between the complement and the head. However this does occur in some languages; making trees impossible to draw using X’ theory.

This was seen by Chomsky and shows that a phrase structure grammar cannot generate every sentence in a language. To fix this problem a set of rules needs to be devised to change the structure produced by the phrase structure rules. These rules have a name:

 Transformational rules.

These transformations take the output of certain rules like X’ rules and create different trees.

This is done using the following form.

The Lexicon →→→→→→→→

X' rules

↓

D-Structure

(constrained by the theta criterion)

↓

Transformational rules

↓

S-Structure

Grammaticality Judgments

The sentence is derived at the top and the utterance of the sentence is at the bottom.

Trees are generated by the conspiring of the lexicon and X’ theory; this is known as the base. The end result is called D-structure which never ever is uttered.

This D-structure is also known as the underlying form.

The theta criterion filters out ungrammatical sentences at D-structure.

Transformational rules are then applied to D-structure moving words around in the sentence(similar to subject –aux inversion and affix lowering).

The final output from the transformational rules is known as the S-Structure.

This is filtered by the Extended Projection Principle ensuring a subject is in the sentence finally creating a grammatical sentence.

The two transformation methods to be observed here are movement rules and insertion rules. Movement rules simply move words around in the sentence, and insertion rules add words to the sentence.

Head to head movement allows sentences that X’ theory under-generates by ensuring trees can be created when the specifier or adjunct is between a head and its complement, by moving one head into another.

*Generative power!

Transformations are rules that change how X’ trees are built.

Transformations basically come after the X’ has done its job and change things around by moving, inserting, deleting etc. providing the final output or S-structure.

Transformational rules are very powerful so they need constraints to restrict what they are able to do these constraints are:

Rules must have some form of motivation.
Rules that violate output constraints cannot be written

Verb Movement(V → T)

French

Je mange souvent des pommes
I eat often of the apples
I often eat apples

The word souvent or “often” is supposedly an adjunct here and it occurs between the complement and the head of the verb phrase! Compared to the English version [I often eat apples] often does not do this.

The adverb often and the subject are split by a T which give the inflection to the verb or comes out as an auxiliary.

In French the T is not here but instead the tensed main verb is here:

Je mange souvent des pommes

I	T	Often	Eat	Apples
Je	Mange	Souvent		Des pommes
I	Have	Often	Eaten	Apples
J’	ai	Souvent	Mange	De pommes

This chart shows the relative placement of the major constituents of a sentence in French with a tensed verb, English with a tensed verb, and both languages with auxiliaries.

Auxiliaries are the parts of speech in the category T, and it is usually predicted that they are followed by V’ adjuncts.

[often]
I often eat apples
I eat apples

In French the T position is filled by the tensed verb!

In English the tensed verb follows the adverb.

Using X’ theory form to generate both the English and the French shows the difference of an extra rule in French that moves its verbs out of the VP position into the T slot.

This rule is seen as V > T movement verb movement/ or verb raising.

V > T movement: move the head V to the head T

Why might this rule apply?

This is similar to the affix lowering from T to V in fact they are in complementary distribution as a language either has verb raising or affix lowering.

Verb movement parameter: Either V raises to T (French) or T lowers to V (English).

The solution to this simple problem can be further extended.

I	do	not	eat	apples
je	ne	mange		des pommes
I	have	not	eaten	apples
J'	ne	pas	mange	des pommes

The positioning of not and pas in relation to the verb is the same as with the adverb [often] but in French the main verb precedes the negation and in English the main verb follows the negation.

Many languages other than French have the same alternation of the tensed verb and the auxiliary.

A 	 la	  saka	 li
We	 have	 rice	 eaten
We have eaten rice

When the auxiliary is in place the main verb sits at the end of the phrase.

A    li 	saka
We eat rice

When no auxiliary is present the verb slots in the regular slot.

The motivation for these movements is motivated by the fact that the verb needs to get its inflection.

Irish

Irish is a VSO language and it poses another more difficult problem.

Phog  Maire an lucharachan
Kissed mary the leprechaun
Mary kissed the leprechaun

X’ theory cannot possibly build a tree for this sentence which generalizes across all Irish phrases and of 9 percent of the worlds languages and X’ theory cannot account for these!

An easy way to solve the problem is by using transformation rules.

A nice assumption to make is that at D-Structure VSO languages are in fact SVO.

To create the S-structure transformational rules produce the output VSO.

This can be implemented using obvious verb movement.

Ta Maire ag-pogail an lucharachan
Is Mary ing-kiss	the leprechaun
Mary is kissing the leprechaun

Phog  Maire an lucharachan
Kissed Mary the leprechaun
Mary kissed the leprechaun

The tree for Irish if we account for verb movement still has the structure of SVO not the required VSO how can this be fixed?

The answer could be by using the parameters system.

So in Irish the specifier of TP can be put to the right of TP

But this causes another problem as the sentence is now VOS which is incorrect (in Irish).

This shows that parameters theory is not the solution here.

So what’s left is to say that subjects have been generated in the wrong position all along. Subjects are underlyingly generated in the specifier of VP.

This is what is known as VP-internal subject hypothesis

The assumption that subjects are generated inside the VP and the claim that theta roles are assigned entirely inside the VP shows that the following constraint can be forced.

The locality constraint on Theta Role Assignment

Theta roles are assigned within the projection of the head that assigns them

If this hypothesis is used them creating VSO order is easy.

There is simply a change in movement from V → T

This movement applies to VSO languages but in SVO there would then have to be two movements.

The first movement would be the specifier of the VP (subject) will move to the specifier of the TP and then the Verb will move to the T.

Another type of movement is DP movement Chapter 10

Summary of head to head movement

There are instances where X‘ rules fail to generate the correct sentence orders.

Transformations fix this problem by taking a structure that is generated by X’ and changing it in restricted ways.

One way of doing this is applying the rule V→T

It moves the head verb to the head T so the verb can support inflection.

The mirror of this is affix lowering and it is in complementary distribution with Verb raising as they are tokens of the same rule but used in different languages and are parameterized as such as in the difference between English and French.

Along with this the VP-internal subject hypothesis explains basic VSO word order languages, and produces a much larger generalization.

T-Movement (T → C)

T→C movement is also known as subject-aux inversion. In English yes/no questions auxiliary verbs invert with their subject.

You have played football
Have you played football?

This claim is supported by the null complementizer ∅ [+Q]

Some languages do not have this inversion instead they have a unique form of the initial complementizer. In Irish this is the complement [an] In English however the encoding that is done with a null complementizer is not uttered which poses a problem!

You can’t hear it!

So transformation can be used as a tool to provide a sound in the ∅ [+Q] position around the subject.

This is because subject-aux inversion is in complementary distribution with overt question complementizers.

I asked whether or not you have played football before
*I asked whether or not have you played football before

This should show that subject-aux inversion is triggered It is triggered by complements and similar to V>T movement it is triggered by morphological restrictions, moving one head into another head position. So V→T and T→C are in fact two kinds of the same operation. Head to head movement

On the surface then V→T and T→C do interact with each other.

English shows that auxiliaries are the only parts of speech that occupy the T slot as a free standing unit, and Main verbs do NOT move to T this means that the only units that undergo T>C movement are auxiliaries.

Have you played football?
*Played you football?

In French however main verbs do undergo V>T movement so when sentences in French undergo T>C movement main verbs should invert as they are already in T.

In French main verbs do invert in English they do not

Vous mangez des pommes
You ate the apples
Mangez-vous des pommes
*Ate you the apples?

Do-Support

When no auxiliary is present in English yes/no questions a dummy auxiliary is inserted.

The reason for this is kind of like a way out for T having to raise and lower at the same time. As T lowers to V and then raises back to T in a question is is called do support, and it is a transformation often called do insertion.

The rule is:

insert a dummy verb do into T when the is no other option for supporting inflectional affixes.

There is a trigger for this transformation. The trigger is that nothing else can be transformed these insertions are called last resort transformations.

Do support/insertion is not only in questions but also in negative sentences.

I did not do the crime
I didn’t find her

In these cases the neg NOT blocks the lowering of an affix for some reason.

There are two verbs DO in english
One do verb is a main verb i.e do this or do that do the dishes
The other verb do is a dummy auxiliary which is meaningless and is classified under do insertion.
Do you do the do?
Did you often do the dishes?
The main verb do is not an aux and does not belong in T often the main verb
follows either often or not
Did you not do what I asked you to?

Multiple Auxilaries and Affix-hopping in English

Multiple Auxilaries

If a sentence has more than one auxiliary, such as:

Tony should have been watching football at the pub

Should is in T but what of the other auxiliary verbs? Take negatives for example

Tony shouldn’t have been watching
Tony should never have been watching

The modal auxiliary should cannot be preceded by ‘not’ in english

*Tony not should never have been watching

Therefore they still exist as T so the auxiliary verbs are not in T but many times verbs such as be and have do appear in T due to negation!

Tony has not gone home yet
Tony is not fooling around anymore

This is because auxiliary verbs (not main verbs) raise in English

So auxiliaries should be treated as V’s that take VP complements So the previous verb movement parameter;

Verbs raise To T or T lowers to V when differentiating english and french.

Can be modified based on multiple auxiliaries

Option1:  All tensed verbs raise to T
Option 2: tensed Auxiliaries raise to T and T lowers to tensed main verbs

In English the second option is used

Affix-hopping

Another property of English that is noticed by observing multiple auxiliary constructions is the form that verbs take after different auxiliaries.

The base form is the way a verb looks after a modal auxiliary verb like “should” that sits in T

Should play
Should write

The verb is a non finite verb in its base form

*should walked

When the verb follows the auxiliary have in its perfective usage (there is a clear ending to an event) the verb usually ends in a suffix –ed, or a suffix –en

This is the verb in its past participle form Has growen>morphological change to > grown

Tony has ridden
*Has walk

If an action is an ongoing action something is in the midst of being done the verb takes the progressive form or the gerund.

Tony is riding
*Tony is ridden/
*Tony is play

The aspect is called imperfect as there is no clear ending

If the verb follows the auxiliary verb “be” the verb takes a passive form, it uses the past participle like the perfective

Tony was bitten

Every different inflection here in English takes a distinct verbal form! The entire meaning of the expression is determined by the suffix and the auxiliary combination. If the verb is simple past or present it is a tensed main verb and undergoes the appropriate affix morphology: -s, -ed or –null

If a modal such as can, or will, or should appears the verb is in its base form with no morphology.

With perfect structure the auxiliary have changes the morphology on the verb to –en progressive pairs the verb ending in –ing with be, and the passive pairs the verb ending in –en with the verb be.

So

Simple tense >>>>>>>>>>>>>>V+ed/V+s/ V+null
Modal >>>>>>>>>>>>>>>>>>>>>>modal V in base
Perfective >>>>>>>>>>>>>>>>>>have V+en
Progressive >>>>>>>>>>>>>>>>>be V+ing
Passive>>>>>>>>>>>>>>>>>>>>be V+en

If the patterns are combined together there is a systematic distribution of information about aspect and voice.

Tony should       be    be   +     ing    bitt   +   en
[Should is to base form][be is to progressive][be is to passive]

Tony has>>> be+   en          be   +   ing                     beat    +    en
[Has is to perfective –en] [be is to progressive ing ] [be is to passive –en]

These are called interleaving patterns where the suffix that attaches to the verb is due to the preceding auxiliary verb that is associated with it.

This is known as affix hopping a phenomenon termed by Chomsky

Passive cannot be simply described as following the verb be as the progressive also follows that verb or either categorized as ending in –en as do the perfectives, passives are only decoded when the form of the verb ends in en and follows the verb be.

To account for this we can split the verb into two syntactive units

By using a transformation V>T will help result in correct word order.

Summary

So subject aux inversion, cross linguistic word order, and do support generalizes a theory of language showing that it takes more than just X’ theory.

There are also transformational rules that apply

By using a transformation V→T will help result in correct word order.

Appendix

Tests for Determining if a Language has V → T or Affix Lowering Here are three tests to help determine if a language shows verb raising or not

A)If the language shows S V +often+  O it has V→T
If the language shows S +often+ VO it has affix lowering
Tony often plays football
B)If a language shows SV not O it has V→T
If a language shows S not VO it has affix lowering
C)If main verbs undergo T→C movement then the language also has V→T

Chapter 10

DP Movement

Introduction

Transformations also are applied to NP’s and DP’s. This phenomenon is called DP movement. Head to head movement is motivated by word orders that cannot be generated by using standard X’ theory, but DP transformations take typical X’ structures and transforms them into altered X’ trees. These movements are motivated by the fact that DP’s can appearing places that theta roles do not assign them to.

A puzzle for the Theory of Theta Roles

The theta grid for the verb to leave

Agent DP

i

Just takes one obligatory argument which is the agent.

He i left
John i left work

There can be other arguments but they are not mandatory so they are not assigned a theta role.

The agent argument must be assigned to the verb leave in the same clause to prevent a crash.

• john said [that left]

This is due to the locality constraint on theta role assignment

Theta roles are assigned within the projection of the head that assigns them.

This means that the constraint requires that the DP getting the theta role be local to the predicate that assigns it.

The predicate left is in a different clause than the DP agent that assigned it making it ungrammatical.

But:

[Johni is likely [to leave]]

This is a direct violation of the locality constraint rule as the agent is far away from the predicate that assigns it.

The reason is because a transformation occurred moving the lower DP to the higher DP.

So only one argument is included in the theta grid for the predicate likely which is an embedded clause.

Preposition CP

j

[That john will leave] j is likely] In this sentence there is nothing about John that is likely but what john is doing is likely which violates

Theta roles are assigned within the projection of the head that assigns them.

So the theta roles are assigned prior to the transformation.

So the subject DP which is located in the embedded VP clause is assigned a theta role and moved into the empty TP specifier position.

This is called raising because you are raising the DP from a lower clause up into a higher clause.

The theta grid for likely includes just one argument the embedded clause.

[That john will leave] j is likely]

With this being said then the sentence

[Johni is likely [to leave]]

John is not getting its theta role from is likely because there is nothing likely about John it is the leaving that is likely.

So theta roles need to be assigned in the initial clause and then movement occurs.

This movement is DP movement and it is done by:

DP movement
Move a DP to a specifier position

If the TP specifier position is empty it can go there.

For transformations to occur there must be a motive for the movement

The extended projection principle forces the expletive IT into specifier TP as it needs to be filled by something.

It is likely that john will leave

So if this spot has nothing in it DP movement is motivated to raise the DP into specifier TP

John is likely to leave

If there are two TP’s this rule happens twice

[CP [ TP T + is [VP t v [AdjP likely [CP[TP to [VP john leave]]]]]]]

First john moves from its theta position in the VP to the lower TP to satisfy the EPP for that TP then moves again to the higher TP to satisfy its EPP requirement

Passives

In traditional grammar the sentence

The man drove the car

Is an active sentence and the man is the topic of the sentence

This can be contrasted with

The car was driven by the man

This is a passive sentence and the topic of the sentence is the car

Even though they do describe the same event that is taking place and the participants are the same. There is some driving happening

The driver (agent)

And the thing being driven the car (theme)

In the passive sentence the agent is represented by a prepositional phrase headed by the word [by].

This prepositional phrase is an adjunct and adjuncts are not included in theta grids as they are not subject to the theta criterion.

This means the agent should be optional:

The car was driven

So actives and passives must have different thematic properties This is not a syntactic difference but a morphological difference.

In English the two passive affixes are –ed and –en

The operation to change an active to a passive

Drive > Drive +en

This operation affects the pronunciation, the entire meaning, and the theta grid of the verb If –en is present there is no DP as its place has been taken up by the affix.

Drive

Agent DP	Theme DP

Agent DP	Theme DP
-en

In active sentences the theme argument appears in the object position in the passive the theme appears in subject position.

This could be because the theme is always generated in object position in both types of sentence and them moved to subject position in passive sentences.

In the tree for the passive then the D-structure would have the was+agent in the VP making was+en the main verb drive and the DP the car.

This leaves the specifier TP empty and EPP requires that the specifier TP must be filled so The DP moves up to specifier TP the copular verb was moves to T and the affix lowers to join the main verb and attaches to it.

When the passive has the adjunct PP that begins with the word [by] the PP joins on to the V’ like other adjuncts do.

Case

One way to satisfy the EPP

all clauses must have subjects > the specifier TP must be filled

is to insert the expletive IT

but in raising environments

*it is likely john to leave

Also the object DP cannot move to satisfy the requirement only the subject DP

Car i is likely Tony to drive t i

In passives the same thing happens

*It was driven the car

To show why this does not work we have to explain CASE

Many languages put markers on nouns to indicate the grammatical relation it has Subjects, objects, indirect objects etc.

This is not the same as a thematic relation as thematic relations represent meaning and noun markings represent how a DP is functioning in the sentence syntactically.

These suffixes represent CASE

Two main case markings indicate the nominative case (found with subjects) and the accusative case (found with objects).

In English sentences that have full DP’s there is no obvious markings only the noun position indicates the grammatical relation.

Tony called Nancy
Nancy called Tony

Pronouns on the other hand do show morphological differences He ([NOM] subject) called her ([ACC] subject) She ([NOM] subject) called him ([ACC] object)

Here is a list of the Nominatives and Accusatives in English

CASE		1sg	2sg	m3sg	f3sg   n3sg	1pl	2pl	3pl
Nominatives:     I 	 you	  he	she	it	we	you	they
Accusatives:    me	  you	 him	 her	it	us	you	them

Even though the verb forms of the first and second person do not have a morphological suffix

I speak you speak we speak they speak

The inflection comes from the morphology of the pronoun (in the present tense).

This means in the present tense there has to be a null (zombie) category that is unpronounced.

This leads to a realization that as well as null tense suffixes in English there can also be null CASE suffixes.

This means that all nouns do get case but we just cannot see it.

This is abstract case

Case has a grammatical function associated with DP’s and being a syntactic property it will overtly be triggered by something in the structure

Chomsky’s theory of case says:

DP’s are given case iff they appear in certain positions in the sentence
[NOM] case is assigned in the specifier of T
[ACC] is assigned as sister to the transitive verb

The case phenomenon serves as the trigger for DP movement

So to get CASE the DP has to go and get it from somewhere and then it can be checked giving it license to be in the sentence. Without this license the DP cannot function correctly in the sentence.

This checking mechanism is called:

The case filter The case filter says:

All DP’s must be marked with a case
If a DP doesn’t get Case the derivation will crash

Feature checking is a fairly standard procedure of case checking and derives from phonology This idea states that words have atomic distinctive features

Take the word He:

HE
masculine
3rd person
singular
nominative

Case assigners can also bear these features

T (is)
present tense
3rd person
singular
nominative

I both of the last (matrices)

They share the case of nominative and the case filter need the [NOM] noun [he] close to the [NOM] Case assigner T (is) to ensure it has the correct features.

[TP [DP he{NOM} [T’[T is{NOM}]]]
[ VP[V’[V loves{ACC}[DP him{ACC}]]]]

Raising: Reprise

Consider the following sentences Bearing in mind the EPP (Extended projection Principle)

All clauses must have subjects > the specifier TP must be filled

a)  It is certain that Patrick left. >>>EPP satisfied with the expletive IT
b)  That Patrick left is certain. >>>Tensed clauses can satisfy EPP
c) *Patrick is certain that left. >>> SSR not possible with finite (Tensed clauses)
d) *It is certain Patrick [to leave]. >>> Verb needs to be finite to satisfy EPP
e) *Patrick [to leave] is certain. >>> Tenseless clauses cannot satisfy EPP
f)  Patrick is certain [to leave]. >>> SSR is possible with tenseless clauses

In clauses like the last 3 that are tense-less (finite) DP’s are assigned [NOM] case in specifier of finite T

Thus non finite T does not have [NOM] feature and finite T does

So Patrick cannot get [NOM] case in the specifier of the embedded clause.

So Patrick to leave cannot assign Patrick [NOM] case violating the case filter crashing he derivation so making the sentence ungrammatical

But Patrick can be assigned [NOM] Case with the finite main clause [is certain] when he is moved up to the specifier DP of the finite main clause adhering to the case filter thus making the sentence grammatical.

This is the journey of the [NOM] DP noun from DP in the non finite tense-less verb specifier position to obtain a theta role to the upstairs specifier of the TP to get case whilst satisfying the EPP along the way.

Here is the sentence taken from Ling322 lecture notes

Tony is likely to understand her

Firstly the meaning of the sentence is derived by the noun obtaining a theta role from the predicate verb

Understand

Agent DP	Theme DP
i	j

Tony is the agent that is doing the understanding and [her] is the one being understood Tony is also a [NOM] in subject position (Tony] can be replaced by [he]

So Tony has to be in the same clause as the predicate non finite verb to receive the theta role

This is the first step:

Meaning is first understood by the assignment of a theta role

The problem is that Tony in the DP of the verb cannot get case from a tenseless verb in non finite form so he has to travel up to a T to satisfy the EPP (all main clauses must have a subject in specifier position).

T must be filled

The first specifier position of T that Tony reaches is right above the non finite verbs preposition [to]. Tony must stop here to satisfy the EPP but he cannot obtain a Case check to as there is no case in this position.

So he has to go up to the next specifier position of T to satisfy the EPP (all main clauses must have a subject in specifier position) where he sits alongside a finite T (is)[NOM] which is able to check [NOM] case.

The finite verb checks the case and satisfies the case filter so Tony remains in the subject position of the TP.

This process is standard and will always hold for all DP's

DP’s always move from positions where they cannot get case but they can obtain a Theta role,
to a position where they can get case, stopping off at any specifier of TP
(even if it cannot assign  case) to satisfy the EPP.

With the first three sentences the DP Patrick,

a) It is certain that Patrick left. >>>EPP satisfied with the expletive IT
b) That Patrick left is certain. >>>Tensed clauses can satisfy EPP
c) *Patrick is certain that left. >>> SSR not possible with finite (Tensed clauses)

can already be assigned case so the motivation to move is non existent and as the 3rd sentence shows if the move is made case is checked twice and crashes the system

Passives Reprise

Passive verbs cannot be assigned accusative case only active transitive verbs allow this.

Remember the accusative pronouns

Me you him her it us you them

To try and assign accusative case to a passive verb will result in an ungrammatical sentence

I drove the car
The car was driven me
He drove the car
The car was driven him

Burzito’s Generalization:

A predicate that assigns no external theta role cannot assign accusative case!

The passive absorbs the ability for the verb to assign accusative case and the passives suffix absorbs the verbs ability to assign an external theta role

Drove

Agent DP	Theme DP
i	j

Driv-en

Agent DP	Theme DP
-en

Closing up a loose end

With some languages subjects get generated in specifier of VP not TP.

In English subjects DP’s are in front of the T and the reasoning is because the subject DP is moved upstairs into specifier of the finite T to get case!

When this happens with actives and transitive verbs the noun comes from the specifier of VP but with passives the movement comes from the object position.

So the difference between SVO languages like English and VSO languages like Irish is the positioning of where NOM subjects get their case from.

English (SVO): NOM is assigned in the specifier of T
Irish (VSO): NOM is assigned in the specifier of V so it is immediately c-commanded by the finite T.

Chapter 14

Raising,Control,and Empty Categories

Introduction

These two sentences look similar but are different in structure

Tony is likely to die
Tony is reluctant to die

The first sentence is a subject raising sentence and the second sentence is a control sentence with no DP movement.

The second sentence has a null DP in subject position of the embedded clause.

The special DP is known as PRO

PRO is a null pronoun

Tony i is likely [ti to die]

The subject Tony gets its theta role from [die]

Agent DP

i

And raises to get case from T

Tony is reluctant [PRO to die]

Consider:

Tony wants john to die
Tony persuaded Harry to jump

John is the one Tony wants to die so john is in the clause of the verb to die and moves up to the object of the verb wants. This is subject to object raising In the second sentence Tony and Harry are both arguments of.

Persuade

Agent DP	Agent DP
i	j

So there is no movement here but a null DP in the V specifier of the verb to leave Once again this is a null pronoun called PRO

Tony i persuaded Harry j [PRO to leave]

Raising vs. Control

Two kinds of Theta Grids for Main Predicates

[That Tony jumped] is likely		clausal subject
It is likely [that tony jumped]	extraposition

[That Tony jumped] is a proposition that acts as the predicates subject with the predicate[ likely] only taking this as the argument [that tony jumped]

In the second sentence there is an expletive IT in subject position and likely takes the proposition as a complement in an embedded clause.

Anything that can take an expletive does not assign the expletive a theta role

Jump assigns a theta role to the AGENT

jump

Agent DP

i

• it is jump cannot take an expletive IT

It seems
It is likely

Has the expletive IT in subject position but is not marked in the theta grid so the theta roles are CP propositions

is likely

Proposition CP

In D-structure the two sentences are exactly the same

In the first sentence the CP

[that Tony Jumped]

starts downstairs to obtain the theta role and moves upstairs to get case from the causal [is] which moved to T to get a tense marker.

In the second sentence the CP stays downstairs and an expletive IT is inserted into [Spec,TP]

As [likely] does not assign a theta role but does get nominative case the upstairs TP is available for case checking this is why the DP [Tony] with the non finite embedded clause can move upstairs to get case.

[Tony ti  tv is [likely to die]

Here the DP in the VP gets its theta role from the non-finite verb [die] and obeys the EPP by moving to the first available open position in subject of TP but cannot get case there so it moves up again to get NOM case from the T [is].

Contrasting with [reluctant]

Tony is reluctant to die

Is reluctant

Experiencer DP	Proposition CP
i	j

Reluctant takes two arguments the person that is reluctant to do something and what they are reluctant about so here another theta role is assigned. You can test to see this is true by trying to insert an expletive IT in the subject TP

It is likely that tony died
That tony died is likely
*It is reluctant that tony died
*That tony died is reluctant

These are both missing an experiencer

Marsha i is reluctant to believe j [that tony died]<adjunct
The car i is reluctant k to move j

The car is theta marked by is reluctant but there is also a non finite predicate verb [move] that also seems to be assigning a theta role to the car?!

This violates the theta criterion

The theta criterion: A theta role can only be assigned to one and only one DP

move

Agent DP

M

So what is the solution to get a one to one system and hold true the theta criterion?

There are now three theta roles

Agent DP i[The car] experiencer DP m [?] and proposition k [to move]

The answer is the inaudible argument called “big PRO”that gets the third theta role

That sits in the subject position of non-finite tenseless clauses where case cannot be assigned and refers to Tony without question.

summary

The main differences between control structures and raising structures is that in raising constructions the predicate does not assign an external theta role, the subject is caseless so it needs to go upstairs to obtain CASE and it also stops off at every Spec TP to satisfy EPP.

In control structures the main clause predicate does assign a theta role to an external argument.

Then raising is not needed and the theta role is assigned to an inaudible caseless PRO

Distinguishing raising from control

Control and raising structures are very similar on the surface but on closer inspection there are some noticeable differences

To help distinguish the differences some tests can be used which only make sense to each type of structure

The most dependent feature of each structure is the main clause predicate

likely
reluctant

The best and most useful way to make the distinctions is to figure out the theta grids associated with the non dominated matrix predicates.

If it assigns an external (underlined) theta role in the subject position

is reluctant

Experiencer DP	Agent DP
i	j

Then it is a control structure

Tony is likely to sneeze
Tony is reluctant to sneeze

What is likely is the sneezing there is nothing about Tony that is likely

Tony’s reluctantcy is what is attributed to the predicate reluctant

there are many more reluctant "control predicates" and it is likely that there are a "lot less raising verbs"

is apparently is in complementary distribution with is reluctant as is is likely

*everyone is apparently is reluctant to leave
*everyone is reluctantly is apparently to leave
*Everyone is apparently is likely to leave
*Everyone is reluctant is likely to leave

Idioms are a good way to test for the difference between raising and control structures Idioms are a way of stating a hidden meaning in what seems like a literal sentence Here are some famous idioms to compare

The cat got out of the bag
The pot called the kettle black
A Bird In The Hand Is Worth Two In The Bush
A Leopard Can't Change His Spots
Bite Off More Than You Can Chew
Everyone here’s a card short                    means everyone’s a bit insane in their own way
Every

If any of these are broken up only the literal meaning exists

The cat  thinks it is out of the bag
The pot tried to call the kettle black
A rare Bird that can land In my Hand Is Worth looking for In The Bush
A Leopard running around in circles Can't Change  His Spots
Bite Off some of my sandwich  but not More Than You Can Chew:

So the subject of the idiom must be localized with the rest of the idiom for it to have the hidden meaning (idiosyncracy) In the D-structure of a raising construction the surface subject of the main clause starts in the specifier of the embedded TP so in raising constructions (at D-structure) the subject of an embedded sentence is local to its predicate.

So the subject of the idiom must be localized with the rest of the idiom for it to have the hidden meaning (idiosyncracy)

In the D-structure of a raising construction the surface subject of the main clause starts in the specifier of the embedded TP so in raising constructions (at D-structure) the subject of an embedded sentence is local to its predicate.

[_______is likely][tony to die]

If we internally represent idioms at d-structure then we should also get theidiomatic meaning from the raising constructions but control structures are not local to the embedded clause so the idiosyncracy should disappear.

Tony is reluctant to die

Lets see

The cat likely got out of the bag
The pot likely called the kettle black
A Bird In The Hand Is likely Worth Two In The Bush
A Leopard likely Can't Change His Spots
Bite Off More Than You Can likely Chew
Everyone here’s likely to be a card short

The cat clawed out of the bag
The pot tried to call the kettle black
A Bird biting me In The Hand Is Worth Two hours of pain In The Bush
A Leopard Can't eagerly Change His Spots
Bite Off More tablespoons of toffee Than You Can Chew
Everyone here's a clubs card short

if the idiosyncracy remains then raising is involved if an expletive it can be put in subject position then raising is involved

What is PRO?

PRO is a DP suprisingly without case because it is a very special silent DP this is why it appears in a caseless position Without PRO an extra theta role would have to be assigned to a role that already has been assigned one which violates the theta criterion Although it seems to be a way to cover up a technical problem with a mask but it is the best theory so far.

Two kinds of control

when movement applies in a sentence:

a)Tony is likely[ti to leave]
b)Tonyi is reluctant [PROi to leave] subject control Tony is co-referred with PRO

a) is a raising construction Tony gets a theta role from the tense-less verb [to leave] and moves to [spec,TP] to get case.

b) gets a theta role from [is reluctant] doesn't move and there is an empty PRO which is caseless.

Tony wants Peteri [ti to leave]
Tony persuaded Peteri [PROi to leave] object control Peter is co-referred with PRO

leave has one theta role

Agent
 DP
 i

want has two theta roles

Experiencer     Proposition
   DP               CP
   i                 k

Peter is the argument of leave but not an argument of want

Chapter 11

Wh- movement

Introduction

Two main types of questions are:

i)the familiar yes/no question where the only answer can by yes/no/maybe etc. and no other answer will work.

did you turn on the lights?

ii)the other type of question is a wh question who what when where which and how. These questions cannot be answered with yes or no.

When does the plane arrive from Boston?

Movement in Wh-questions

Wh phrases appear in a position of the sentence away from the place where the theta roles are assigned.

Tony edited the magazine

Here Tony is the agent and the magazine is the theme and is in object position

What did Tony edit?

Here Tony is the agent and the question what is the theme and it is at the start of the clause!

Edit takes two theta roles edit

Agent<\n><cen>DP<\cen>	Theme<\n><cen>DP<\cen>
i	j

In this sentence

What did the producer say Tony had to edit?

Here [what] is still the theme of [edit] but it is very far away from theta role assignment position(even further than before).

Locality constraints and case assignment rules seem to be violated here

Theta roles are assigned within the projection of the head that assigns them.

Accusative case is assigned when a DP is the sister to a V.

Tony [VP paid him ACC]

However in wh questions like who

The accusative form is not sister to v

Who ACC did Tony pay?

So from these observations wh questions are not in their theta role positions and they are not in their case positions either which brings us back to movement.

So where do the wh phrases move to?

The empty slot [Spec, CP]

This movement is the movement of an entire phrase

[Which way] did you say the supermarket was?

So because an entire phrase is being moved it is not head to head (V>T)(T>C) movement it must not move to a head position which is the empty sec CP

These elements that can go into spec CP can be either DP’s PP’s or AdvP’s Because of the movement into spec CP this feature shows that wh questions also involve T>C movement.

What team are you playing today?
*What team you are playing today?

The aux [are] is positioned in C and the wh question is higher than this spot so it can only fill spec CP.

Irish allows wh phrases along with an overt complementizer

Cad   a    ta sa     seomra?
What C-wh  is in-the  room?

In English the only thing that is allowed to appear in C is an inverted auxiliary any other complementizer is not allowed.

*I wondered what that you wrote?
I wondered what it was that you wrote

This assumes that the only complementizer that is compatible with wh movement in English is null.

Whereas in other languages, the complementizer has a phonological particle that is inserted into C.

On the same lines as T to C movement being triggered by a [+Q] feature

And DP movement being triggered by the DP needing case so it moves from spec,VP to spec,TP to get ACC case from the Head of TP T.

So what triggers wh movement?

The feature can be called [+WH] and it sits in the C of a wh phrase.

Some languages actually have special types of complementizer particles which represent this feature Irish

[-Q,-WH] go ; [+Q,-WH] an ; [+Q,+WH] aL

If the sentence is not a yes/no or wh question the C gets a [go]

If the sentence is a yes/no question the C gets a n [an]

If the sentence is a wh question the C gets a aL

The wh phrase moves to [spec,CP] to get close to the [+WH] feature.

This is a bit like case checking the DP for NOM but for wh checking instead.

Wh movement
Move a wh phrase to specifier of CP to check a [+WH] feature in C

Who did Tony shoot?

In D structure this would be as follows:

Tony and who both get theta roles from the verb kiss in the VP. T has a suffix –ed

And C is showing [Q,+WH]

[Who] also gets ACC case in the VP base position DP

Then the subject Tony needs to get case checked by the suffix of shoot which is in the T position of the TP so Tony moves up to the spec TP position and collects NOM case.

Do needs to be inserted to support the pronunciation of –ed and then do is moved to C to fill the null [+Q] complementizer T > C movement.

Next Wh movement applies to check the [+WH] feature

Resulting in

Who did Tony shoot?

Here is a more complicated wh movement example

Who was arrested?

The d structure for this sentence is detailed as follows

The spec CP is empty

C is null complementizer [+Q +WH]

Spec TP is empty

T gives NOM case for past tense

The V in the first VP is split for the verb to be and –en gives ACC case

The V in the next VP has the verb arrest and the DP has the wh question [who][+WH, NOM] which is complement to the verb

The only argument in the sentence is who which is a theme the en lowers down to the verb which absorbs the agent theta role and the verbs ACC case. Also the aux verb [be] raises to T to support the tense and becomes was.

[Who] now needs to move to the [spec, TP] to get NOM case because the en has absorbed the case in the base position.

Next [Who] moves up to [spec CP] after getting the NOM case to get checked by [+WH] and the aux [be]moves up to C (T>C movement)for [+Q].

Surface now becomes

Who was arrested

Islands

There is a barrier that wh movement may encounter and this is to do with what categories it can move out of.

A simple complement CP

What did Tony witness[CP that he had  to tell ti the police about]?

If you try to move a wh out of a complement clause

Why i is it [CP that I make ti mistakes to jeopardize my career]?

but if you try to move out of a DP

*Whati did Tony witness [TP the crime [he saw [CP that he had to tell tithe Police about]]?

looks very bad

Basically wh movement can occur out of simple complement CP but not out of a clause that contains a DP as the DP is an island.

This is known as the complex DP constraint.

Other islands can also be present in clauses.

Wh islands are one of them.

When you try and move a wh question to the [spec, CP] when the P is [+WH] this is possible.

I wonder [CP howi C[-Q+WH] [TP Tony managed ti to pick ]]

Also moving an additional wh-phrase to [spec, (main)CP]

[CP Whichk one do [TP you think [Tony will pick from the selection tk]]]?

But when you try to do both of these things in one phrase

*[CP Whichk one do [TP you wonder [CP howi [CP Tony managed to pick ti tk]]] ?

It all goes wrong even though the two transformations that were done are both correct.

This does not mean that two wh questions cannot be in a sentence.

It is just they cannot both be transformed (moved)

So acceptable movement is moving either the subject or the object to the [spec, CP]

I wonder why Tony went to the store.
Where did Tony go?

*[CP1 Where k did [TP you wonder [CP2 why i [TP Tony went ti to the store tk]]]]

This should show that when a wh phrase is moved into the [spec, CP] that CP then becomes an island.

I asked [CP whyi Tony went to the store ti]

If you move out of this island the result becomes ungrammatical.

So the wh constraint says you cannot do wh movement and skip around a CP that has another wh phrase in the specifier of that CP.

Another type of island is the subject.

[TP [CP that Tony was going to the police station ] made sense.

*Wherei  was [TP[CP that Tony was going ti] make sense.

This is the subject condition.

There is a CP in subject position and trying to move the wh [where] to mean the police station the sentence becomes bad.

A final island to bring up is the coordinate structure island.

Tony has some pencils and papers
*Whati did Tony have some pencils and ti?
*Whati did Tony have some ti and papers?

If either of the DP’s undergo wh movement the result is not grammatical.

Also if any conjoined structure undergoes wh movement the result is not grammatical.

Tony [VP played football] and [VP scored a goal]
*whati did Tony play ti and score a goal?
*whati did Tony play football and scorei ?

So there are four different instances where wh movement will result in grammatical errors. Complex DP’s Subjects, CP’s that have wh words in spec, CP already and wh movement out of one conjunct in a coordinate conjunction.

A possible explanation for these problems and the effects is a constraint called the minimal Link Condition MLC.

The minimal link Condition

Wh-islands and the minimal Link Condition

Movement of wh phrases is only possible in questions when each wh phrase is independently moved but if combined movement occurs the result is very bad.

Syntactic operations usually like to be locally done within their own clause (such as binding anaphors in their binding domain) or they like to create localities (such as DP movement to get case), (affixes move to get close to the word they are assigned to and wh phrases move to get close to [WH+].

So grammars like close relationships.

So with wh islands in mind when wh phrases move to a spec, CP to be close to [+WH] in the C a further restriction can be formulated.

Movement must always target the closest potential position.

This is the minimal link condition (MLC).

MLC: move to the closest potential landing site.

In the phrase:

*[CP1 Wherek did [TP you wonder [CP2 whyi [TP ti went  to  tk]]]]?

There are two CP’s but both wh phrases start of in the embedded clause.

[TP Tony went ti to the store tk]

So for both wh phrases the closest possible landing site is CP2.

In d structure here is a phrase with the landing sites underlined.

[CP1 _ C [+WH][TP you ø[pres] wonder [CP2_C [+WH] [TP why Tony -ed  go where]]]]?

Let’s move where to the first possible landing site which will then check its [+WH] feature.

[CP1 _ C [+WH][TP you ø[pres] wonder [CP2 wherek C [checked WH] [TP why Tony -ed  go tk]]]]?

This move has been done in accordance with the minimal link condition as it moved to the closest potential landing site.

it can work with the other wh phrase too.

[CP1 _ C [+WH][TP you ø[pres] wonder [CP2 whyi C [checked WH] [TP ti Tony -ed  go where]]]]?

So now the other wh phrase needs to check it’s [+WH] features but the closest possible landing site already has been landed in by the [where] so to check case it would have to go to CP1 but this violates the minimal link condition so it cannot happen because it cannot skip the first POTENTIAL position which is still CP2.

[CP1 whyi C [+WH][TP you ø[pres] wonder [CP2 wherek C [checked WH] [TP ti Tony -ed  go tk]]]]?

Hence it cannot be checked for wh feature [+WH] regardless of order of wh phrase to be moved.

This MLC explanation for wh islands:

Multiple wh phrases that require movement, movement of at least one wh phrase will be
blocked by  MLC  because the closest possible landing site is occupied by the other wh phrase.

What about long distance movement with wh phrases like

[CP1Whyi do you think[CP2  ti[TP ti Tony went to the store]]]?

This movement should violate the MLC as the movement is not local.

The wh word why gets its theta role downstairs from the embedded clause and it finishes in the highest spec, CP!

This is because wh movement can cross boundaries and does so in hops first it lands in potential spec, CP2 then hops up to spec CP1.

So the reason it can do this is because MLC constraints require local movement so the movement has to hop to abide by the rules.

This is cyclic movement and you can have many CP’s in a tree and a wh word can hops onto every one of them all the way to the top.

[CP1 [DPi Whyi][C’[C [+WH]] do [TP you think] [CP2 that john believed Mary’s story
[CP3 that Paul ate the piece of fish [CP4 that was  caught by the fishing company
[CP5 that was fined for selling poisonous fish ti]]]]]?

The wh phrase stats ar the lowest cp to get a theta role and hops onto every spec, CP all the way to the top satifiying the MLC all the up.

Not all island effect are accounted for by the MLC only wh islands.

The MLC in DP movement and Head movement

The MLC can be used to account for locality effects in DP and head movement.

Recalling that [is likely] and [seems] do no have subject positions so they allow for subject to subject raising.

Tonyi is likely [ti to win] tony gets a theta role from win
Tonyi seems [ ti to have won]  tony gets a theta role from win

If one of these sentences is embedded into the other DP movement can only occur in the lower of the two case positions.

d-structure of the two sentences embedded.

_seems [that _ is likely[Tony to have won]]

The DP can shift in to the lower local position and an expletive it can be inserted into the upstairs.

It seems [that Tonyi is likely [ti to have won]]

If the DP shifts up any higher the result is ungrammatical even when the expletive it is inserted into the lower DP position.

*Tonyi  seems[that is likely [ti to have left]
*Tonyi seems [that it is likely  [ti to have left]]

If two case positions are available in d-structure MLC can account for this movement constraint Move to the closest potential landing site to get checked.

If the DP moves up higher the position is already occupied by an expletive “case island”

A formalized argument for the minimal link condition.

Movement of X item can target Y position of type Z iff:
i)	Y c-commands X
ii)	There is no W of type Z such that Y c-commands W, and W c-commands X
iii)	 Z is defined as:
    a)	 a head if X is a head
    b)	 the specifier of TP if X is a DP with unchecked [NOM]
    c)	 the complement of the V if X is a DP with unchecked [ACC]
    d)	 the specifier of CP if X is a wh phrase with unchecked [+WH] feature.

Closeness is defined in terms of c-commanding where there is not any intermediate c-commanding landing site intervening between the item moving and its landing site which is very close to binding theory. MLC is assumed to be a successor to binding.

Echo Questions (Wh-in-situ) in English

MLC prevents the movement of multiple wh phrases but this does not mean a clause cannot have more than one wh phrase!

Who went where?
Tony found what?

Even though the second wh [where] doesn’t move the result is still grammatical this is called

 wh-in-situ “in place”

The reason why the wh does not move to get checked for [+WH] features is because these are NOT wh questions so they are [-WH] instead they are echo questions.

Echo questions do not seek new information they confirm known information.

I saw the policeman hit the old man
The policeman hit who?
I told you, the old man
The policeman did what? (to the old man)
I told you he hit (the old man)

Echo questions do not involve movement so they do no get constraints from the MLC instead they require a phonological audible raising and stress.

This is very similar to questions that do not have sub/aux inversion

Tony spent the night in his car?

This is a request for confirmation and infers doubt in what really happened This is different in meaning to

Did Tony spend the night in his car?

This is a request for information it has sub/aux inversion and do support

Source

Syntax http://www.blackwellpublishing.com/carnie/