Syntax – Week 6, Lecture 2
1
X-Bar Theory, Part IV
Readings: Radford 1988 Transformational Grammar Ch. 4, 5, 6.
1. Revision – IP
In the last lecture, we examined English auxiliaries. We noticed:
● the first auxiliary in a verbal element carries the tense of the entire
clause
● each auxiliary adds a feature to the clause such as tense, aspect
modality and/or voice
● each auxiliary acts like the head of a phrase, subcategorizing for the
following verbal complement
Over the years, there have been many ways of dealing with the intricacies
of English auxiliaries. The most commonly accepted solution now is to treat
the first auxiliary as belonging to a different category I (short for INFLection,
since this is the position associated with tense inflection), which is the head
of the clause. This allows us to integrate S into X-Bar theory, which is the
other problem we were wanting to deal with. What are the phrase
structure rules which will generate these IP trees?
IP → NP I̅
I̅ → I VP
VP → V̅
V̅ → V VP
The key thing to notice is that a VP may ultimately
contain another VP. This is how additional auxiliaries are
added to the phrase structure tree.
As a sister to I, we put a VP which in turn is headed by the next auxiliary
or the main verb. For example:
Syntax – Week 6, Lecture 2
2
Exercise: Draw the phrase structure tree for “She might have been driving
the car”
Single Verb Sentences
When there is just one main finite verb, it occupies the main V position
and I is not filled:
For our purposes, we will assume that the I
node is not optional, and that it is still there
even if there is no word occupying this node
(this is more in keeping with the Chomskyan
version of X-bar, though we haven’t gone as
far as to suggest abstract features like
[+Past] occupy the I node).
Syntax – Week 6, Lecture 2
3
In Chomsky’s (80s-90s) theory of syntax, there
are two levels of structure, ‘deep structure’ and
‘surface structure’. In deep structure the I
position is filled by the tense inflection of the
verb. This tense inflection is then moved to the
end of the verb in surface structure (in a
transformational rule called ‘Affix Lowering’).
This is the kind of analysis used in the Secret Life
of Language textbook. But we won’t be following
it here.
Summary
Our clause is now an IP. When the clause has one or more auxiliaries, the
first fills the I node followed by a VP complement headed by the main
verb or next auxiliary. This process continues until there are no verbs left
in the verbal unit. When the clause contains only a main verb, the I node
is empty, and the V heads a VP which is daughter to I’. These structures
are summarized below:
John ate spam John has eaten spam John has been eating spam.
Syntax – Week 6, Lecture 2
4
2. Specifier to V’
Why does VP always go to V’ first in our rules? Is it possible for a VP to
have a specifier? One candidate for Spec of VP is an element such as all as
in:
● The boys are all waiting for us.
● The boys may all be waiting for us.
● The boys may have all been finished.
● The boys may have been all finished. Etc.
3. IP’s with only a main verb
In the last lecture we noted that when a
sentence has only a main verb, this verb is
daughter to V and not I.
For example, in the sentence “John ate
spam”, the verb ate fills the V node and not
the I node. You should still include and
expand the I node, but leave it unfilled.
We follow this analysis because auxiliary verbs have different syntactic
properties when compared to lexical verbs. These are:
● Auxiliary Inversion
● Auxiliary Negation
Syntax – Week 6, Lecture 2
5
Auxiliary Inversion
Auxiliaries can undergo inversion in direct questions while main verbs
cannot:
You can speak Japanese ↔ Can you speak Japanese?
It is still raining ↔ Is it still raining?
Compare this with lexical (or main) verbs:
You fetched the water ↔ *Fetched you the water?
Caesar conquered Gaul ↔ *Conquered Caesar Gaul?
Note that direct questions of lexical verbs in English are formed using the
dummy operator “do” (this same operator also carries the tense):
Did you fetch the water?
Did Caesar conquer Gaul?
The main exception to the generalisation about auxiliary verbs being
different to main verbs concerns those verbs which can function as both:
be, have and do. For these three verbs, we get variable results:
Inversion is natural with be:
I am an undercover cop ~ Am I an undercover cop?
But not so much with have:
The VC has a secret underground passageway
? Has the VC a secret underground passageway?
Does the VC have a secret underground passageway?
Syntax – Week 6, Lecture 2
6
And not at all with do:
You did the dishes already.
*Did you the dishes already?
Did you do the dishes already?
Auxiliary Negation
Auxiliaries can be directly negated by the addition of n’t or not following
them – this is not the case for main verbs:
Jill could come ↔ Jill couldn’t come
I might see her ↔ I mightn’t/not see her
I detest spam spread ↔ *I detest not spam spread
You travelled to Europe ↔ *You travelled not to Europe
To negate a main verb, once again, we make use of a dummy operator:
I do detest spam ↔ I don’t detest spam
You did travel to Europe ↔ You didn’t travel to Europe
This syntactic evidence suggests that the auxiliaries (and the dummy
operator do) occupy a different place in our tree structures when compared
to main or lexical verbs. Because the main verb behaves differently to a
tensed auxiliary verb, it is placed under a V node in our tree structures, and
the first tensed auxiliary is placed under the I node.
Note we can use the tests above (and other tests) to identify a subset of
verbs called operators which fill the I node rather than a V node. For
example: dare (in a slightly old-fashioned usage):
Dare we tell Uncle Quentin?
Syntax – Week 6, Lecture 2
7
Also, for some speakers, better:
We better tell him.
We better not tell him.
“You’d better get up then, bettern’t ya?”
c.f., You can eat Jam, can’t you?
3.1 Case Study – Icelandic
Verbs in Icelandic can undergo inversion:
Þú talar íslensku – You speak Icelandic.
Talar þú íslensku? – Do you speak Icelandic?
Hann kemur í kvöld – He comes tonight.
Kemur hann í kvöld? – Does he come tonight?
This suggests that main verbs in Icelandic belong under I. This analysis
predicts that negation ought to be possible by adding a negator directly
after the main verb:
Ég kasta skeiðar – Ég kasta ekki skeiðar
I throw spoons – I do not throw spoons.
Ég sé hrafn – Ég sé ekki hrafn
I see the raven – I do not see the raven.
Syntax – Week 6, Lecture 2
8
4. Possession, NP’s and DetP’s
Thus far, we have examined only simple
instances of possession such as “my old hat”.
Even this raises questions – for example, the
determinative is the only specifier we have seen
that is not phrasal. In this section we are going to
finalize our analysis of determinatives and
possession.
Consider the following:
My cabin
Ash’s old cabin
In these two phrases, there are two entities – one which possesses
something, and the other is the thing possessed. We call the ‘s clitic the ‘s
genitive possession construction. Note the distribution of the underlined
phrases below:
a. The old cabin.
b. Ash’s old cabin
c. *Ash’s the/this/that old cabin
This ‘s genitive construction exists in complementary distribution to
determinatives. This suggests that despite being different phrases,
determinatives and ‘s genitive possession constructions have the same
distribution (i.e., are in the same position in phrase structure tree)
We call the ‘s a clitic (rather than a morpheme) because like most other
clitics, the ‘s attaches not just to words, but to entire phrases. Specifically,
it attaches to the end of a NP, not to the head N:
Syntax – Week 6, Lecture 2
9
[Ash]’s old cabin
[That man and woman]’s child
[The man standing over there]’s old cabin
Not: *[The man’s standing over there] old cabin
Currently, there are two competing analyses of these types of
constructions. We will develop the following sketch of possession in
English (largely following Kroeger’s analysis in the textbook).
4.1 The NPPOSS analysis:
We note the following features of these constructions:
1. The possessor is a complete NP:
[The smart woman]NP’s favourite book
2. The possessor (along with the ‘s clitic) exist in a complementary
distribution with determiners:
The smart woman’s / her / the favourite book
This suggests that the possessor phrase is a full NP that acts as a
specifier within the larger NP:
Syntax – Week 6, Lecture 2
10
Note that the possessor NP is subscripted ‘poss’. The idea is that this is an
NP with a specific grammatical function — possessor — like the subject NP
or the object NP in a clause. Because the NP bears this function, it is
morphologically marked with the possessor clitic ‘s, which happens to be
realised at the end of the NP. This is essentially the analysis proposed in the
textbook (pp. 92-97).
This analysis makes a prediction – the thing possessed has to be an N’ – it
should be able to have adjuncts and complements, but not a determiner;
also, the thing possessed can’t be a pronoun (because pronouns aren’t N’
constituents):
[The woman]’s favourite book with adjunct
[The army]’s destruction of the city with complement
*[The smart woman]’s the book with det.
*[The man and woman]’s her/he/him/she pro.
This prediction is born out, leading the following analysis of possession
constructions:
NP → NPposs N’
The analysis developed above captures most of the intuitions we have
with respect to possession in English. But it’s not the only analysis…
Syntax – Week 6, Lecture 2
11
4.2 The DP Hypothesis
I will introduce a competing theory of possession called the DP hypothesis
that is more applicable cross-linguistically and is more widely accepted in
the literature.
In this theory, we note that
determinatives should be heads
of their own phrases:
Here, ‘D’ is the determinative (often called a determiner) head. YP is its
complement, which we’re going to say is a NP. And XP is its specifier,
which is where (it is argued) possessive NPs go.
In the case of simple determinatives, there’s not a lot of difference.
Compare below our new analysis and our old analysis for the phrase “my
hat”:
DP → (XP) D’
D’ → D NP
NP → det N’
N’ → N
New Old
But what about something like Ash’s old cabin?
Syntax – Week 6, Lecture 2
12
The proposal is that ‘s is a determinative and therefore the head of the DP.
The possessive NP is its specifier (which must be another DP or NP!). The
‘thing possessed’ is a NP complement to the head D:
This may feel odd, but the motivation comes from substitution evidence:
recall that we can either get a NP with clitic ‘s (like Ash’s) OR we can get a
simple determinative like this, not both – we can’t say *Ash’s this old
cabin. This tells us that there’s a complementary distribution here.
Something substitutes for a determinative like this, and it’s either got to
be Ash’s or just the clitic ‘s (with Ash in SPEC position). We know that Ash
is a N though, so this leaves the clitic ‘s as the only thing that we could
possibly analyse as a determinative.
Importantly, an advantage of this analysis is that it allows us to treat all
specifiers as fully phrasal – we never just have a single word like a det. as
a specifier in this approach.
Another example:
Syntax – Week 6, Lecture 2
13
Note that there are various problems with this analysis, which we won’t
discuss in detail. But here’s just one. The structure above, where:
DP→(DP) D’
D’→D NP, etc.
predicts that we could get more than one word of category D in a
possessive construction. There’s nothing to rule out, for instance:
*The smart woman the favourite hamster
In your assignments and exam, unless otherwise instructed, we will not be
using the DP analysis – and we’ll stick to Det/NP[POSS] analysis.
Syntax – Week 6, Lecture 2
14
5. Wrap-up of X-bar Theory
We went from this:
Syntax – Week 6, Lecture 2
15
To this:
Syntax – Week 6, Lecture 2
16
The key insight we made
was all the phrases
conform to the same
general configuration:
However, X-bar theory is still an imperfect description of languages. I have
listed some limitations below:
1. We argued that the I in an IP carries the tense of the clause – however,
modals like “may/might” don’t really seem to carry tense:
I may/might see her tomorrow
I see/*saw her tomorrow
2. What happens when a speaker does manage to get an adjunct between
the head of a phrase and its complement?
I argued for several hours with the council, but to no avail.
More generally, some linguists argue that something like X-bar theory is
central to all languages. This position faces some challenges though:
3. Languages with VSO word order don’t have an apparent VP. For
example Welsh:
Gwelodd Sion ddraig
see.3sg.pst John dragon
“John saw the dragon”
Adjunct Position
Complement Position
Syntax – Week 6, Lecture 2
17
4. Some polysynthetic languages look like they have little to do with X-bar
theory. For example, in a discussion on Navajo Ken Hale provides the
following deep form for the structure of (a) below:1
a. Bich’odeeshniił
3.1sg.take_the_side_of
“I’ll take his side”
The claim is that at their hearts, all languages look like English. Would we
have reached this position if most syntacticians spoke Navajo rather than
English?
Successes:
1. Very frequently, other languages can often be described by X-bar
theory. For example, the requirement that complements come before
adjuncts is seen in many languages:
Icelandic:
Eyðing borgarin-nar um morgun-inn
destruction city.f.sg.gen-the in morning.m.sg.acc-the
“The destruction of the city in the morning”
?Eyðingin um morgun-inn á borginni
destruction in morning.m.sg.acc-the of city.f.sg.dat
“The destruction in the morning of the city”
*Eyðingin um morgun-inn og á borginni
destruction in morning.m.sg.acc-the and of city.f.sg.dat
“The destruction in the morning and of the city”
1Hale, Ken, Andrew Barss, Ellavina Tsosie Perkins, and Margaret Speas (1989) Logical Form in Navajo. In E.-D. Cook
and K. D. Rice, eds., Athapaskan Linguistics: Current Perspectives on a Language Family, pp. 317-334. Trends in
Linguistics, State-of-the-Art Reports 15. Berlin: Mouton de Gruyter.
Syntax – Week 6, Lecture 2
18
2. There is no reason to expect that NP’s and VP’s would have the same
structure, but we showed that they do. This pattern is observed in many
other languages.
3. A syntactic account of structural ambiguity, adjuncts and complements.
Appendix: English Verb Examples
I might wait for you.
Finite modal verb + main verb in base form.
We have seen him.
Finite perfect have+ main verb in –en form.
The students are protesting about the war.
Finite progressive be + main verb in –ing.
My bag was stolen yesterday.
Finite passive be + main verb in –en form.
She might have waited for you.
Modal + perfect have in base form + main v. in -en
She might be waiting for you.
Modal + progressive be (base) + main verb (-ing)
My bag might be stolen.
Modal + passive be (base) + main (-en)
I have been waiting for you.
Finite perfect have + progressive be (-en) + main verb (ing)
Syntax – Week 6, Lecture 2
19
My bag has been stolen.
Finite perfect have + passive be (-en) + main verb (-en)
I might have been waiting for you.
Modal + perfect have (base) + progressive be (-en) + main verb (-ing)
My bag might have been stolen.
Modal + perfect have (base) + passive be (-en) + main verb (-en)
My bag might have been being stolen at the time.
Modal + perfect have (base) + progressive be (-en) + passive be (-ing) +
main verb (-en)