Course Notes

advertisement
The Development of X-bar theory
1. What Came Before
a. No notion of phrase in Antiquity because classical languages were morphologically
complex
i.
enough to concentrate on with morphology
ii.
main parts of sentences could be associated with words
A. subject = word bearing nominative morphology
B. subject word agrees with verb and modifying words
iii.
a lot of word order freedom hides phrases
b. American Structuralism
iv.
Operationalism – only observed things can be part of science. BUT most of
linguistics can’t be observed
v.
So everything must be based on what can be observed – i.e. phonetics 
phonology  syntax
vi.
The elements of each higher level must be based on elements of the lower levels
to be rooted in observation
phones  phonemes  morphemes  words  phrases  sentences
vii.
How these are linked to each other is through the notion of distribution
a phoneme is a set of phones which have a certain distribution (complementary
or identical); a morpheme is a sequence of one or more phonemes with a certain
distribution; a word is a sequence of one or more morphemes with a certain
distribution; a phrase is a sequence of one or more words with a certain
distribution.
viii.
This is not exactly the same notion of a phrase we know today as
A. phrases could be discontinuous, e.g. run the battery down
B. meaning was not supposed to be considered (because it couldn’t be
observed)
C. Phrases could be endocentric (replaceable by a word contained in the
phrase) or exocentric (not replaceable by a word contained in it)
a) he [ate the fish] he [ate]
b) he went [to London]
*he went [to] *he went [London]
c. Generative Grammar (Chomsky)
ix. Chomsky heavily criticised Structuralism
A. Operationalism doesn’t work in the mental sphere– too restrictive
c) he argued that language is knowledge, not behaviour
B. Empiricism is wrong – rationalism is the way to go
C. using only distribution as a means of discovering linguistic facts is too
restrictive – ignores a lot of possible data (intuitions)
D. meaning must be considered when hypothesising about syntax – though
they are not the same thing and facts from one cannot be transferred
directly to the other.
x.
He modelled the Structuralist view as a Phrase Structure Grammar, making use
of Re-write rules
A. these weren’t exactly what Structuralists had in mind as they didn’t allow
discontinuous constituents – something which Chomsky then turned into a
criticism of Phrase Structure Grammar: it couldn’t handle real linguistic
phenomena
B. this turned into a whole new branch of linguistics Mathematical
Linguistics, based on the question of what kinds of languages could be
generated by what kinds of rules, e.g. context sensitive vs. context free
rules.
xi. he argued that Phrase Structure Grammar was only part of the linguistic system,
but other kinds of rules (transformations) were also needed
d. Developments in the 1960s
xii. an early development was to separate the Lexicon from the phrase structure
component of the grammar
C. this allowed phrase structure grammar to be context free
D. it also allowed the grammar to be much simpler
2. Remarks on Nominalisation
a. Harris’s raised numbers
i.
in the 1940s Harris (Chomsky’s teacher and a follower of Bloomfield)
introduced a system to account for non-repeatable substitutions
A. On distributional grounds, if X can substitute for Y, then X = Y
B. moreover, if X+Z can substitute for Y, then X+Z = Y
C. but in this case it should be that X+Z = X and X+Z should be able to
substitute for X
ii.
However, this does not always work
A. N+hood = N because boyhood can replace life in sentences like
a) his life was spent on an island in the Mediterranean
B. however, we cannot replace the N in the N+hood sequence with N+hood
as we never get N+hood+hood
iii.
to cope with this Harris suggested that we add raised numbers of a different
kind for those elements which cannot be substituted repeatedly
A. boy = N1
N1+hood = N2
B. then, N2 can replace N1 in all cases except for the equation that defines N2
due to a restriction that says if Nn contains Nm then n > m
iv.
for repeatable substitutions, the same raised number is given to both elements
A. man = N1
old man = N1
B. because ‘man’ in ‘old man’ can be replaced by another A+N sequence (old
grumpy man)
b. On Nominalisation
i.
arguing that not all syntax is transformational
A. issue: nominalisations were seen to be syntactically formed from
underlying sentences:
a) the enemy destroyed the city  the enemy’s destruction of the city
b) John shoots elephants  John’s shooting elephants
B. argument: while gerunds are productive and regular (indicating a syntactic
analysis) derived nominals are neither. Therefore certain processes
(derivational nominalisation) take place in the Lexicon
C. If all the following are lexically formed, then we need phrase structure
rules to generate their structures:
a) destruction of the city
(NP  N (of) NP)
b) his reply to the interviewer
(NP  N PP)
c) their belief that he left
(NP  N S)
D. But this increases the number of rules needed as (derived) nouns will have
the same number of complements that verbs have (adjectives derived from
verbs will go to exasperate the problem)
E. So a first step was to introduce the following kinds of rules:
a) VP  V Comp
b) NP  N Comp
ii.
iii.
c) AP  A Comp
d) Comp  NP/PP/S/...
F. There are two problems with this
a) if these rules are motivated only for derived forms, it casts doubt on the
proposal as a syntactic approach could get by with the VP rules and
changing the VPs into NPs and APs
b) the rules introduce Comp as a category. But this category plays no
significant part in any other syntactic rule and so it has no independent
justification
G. Problem a) is addressed by the observation that there are non-derived
nouns (and adjectives) which require these structures
a) N (of) NP (professor of chemistry)
N PP (pride in his work)
N S (fact that he left)
H. Problem b) can be addressed if we introduce a more general notation
where complements are claimed to be phrases of categories which are
determined by the words they are complements of – the phrase structure
rules only have to stipulate that they are phrases
a) VP  V YP
NP  N YP
AP  A YP
I. But once we have introduced category variables, we can make the whole
system simpler:
a) XP  X YP
the next step was motivated by the fact that in derived nominals, the subject of
the clause is realised as the ‘determiner’ of the NP:
A. the enemy destroyed the city
B. the enemy’s destruction of the city
The subject is separated from the verb inside the VP in the structure of S
S
NP VP
iv.
V
...
Therefore we need a structure like this for the NP
NP
Det
v.
vi.
??
N
...
Chomsky called the phrase which joins with ?? to make the whole phrase the
Specifier of ??
He systematised the whole thing by making use of something like Harris’s
‘raised number’ notation, but using ‘bars’:
vii.
viii.
A. Χ̿  Spec Χ̅
Χ̅  X ...
This produces a structure
Χ̿
Χ̅
Spec
X
...
3. 1970s developments
a. Inclusion of Prepositions
i.
prepositions were excluded from the ‘on Nominalisations’ because there are no
derived prepositions. They were included into the system about 3 yrs later.
b. Head
i.
prior to X-bar there was no way in PSG to capture the structuralist notion of
endocentric phrases (phrases which could be replaced by one of their elements)
A. XP  Y Z
a) there is no necessary connection between the properties of XP and any
element that it contains, so why any Y or Z can replace XP is entirely
mysterious
ii.
it became apparent that the head (X) in the X-bar scheme had a special property
A. X is necessary as without it we don’t know what X’ or X’’ are
B. Not every X is accompanied by a complement or a specifier
C. therefore X can be the only element in XP
D. so X can replace XP
iii.
another problem solved by the notion of a head is that arbitrary phrase structure
rules predict non-occurring phrases as possible:
A. VP  N PP
a) the question of why VPs (or other phrases) must contain a verb (or
relevant other category) is solved under X-bar theory.
c. how many bars?
i.
Chomsky’s original proposal had phrases with 2 bars simply because this was
all that was needed to do the job
A. the job was to make NPs have a similar structure to sentences
a) sentences have subjects and verbal complements
b) so NPs needed specifiers and nominal complements
B. however, there was no other reason to believe that this was all that was
needed. In the years that followed, various suggestions we followed based
around two questions
a) how many bars are necessary?
b) should all phrases have the same number of bars?
C. X-bar theory does not predict any particular number of bars and so any
limit that is imposed has to be stipulated. Empirical based stipulations are
circular. In principle, a phrase could have any number of bars and all
phrases could be restricted to the maximum needed to account for one type
of phrase – it is just assumed that the extra bars have no purpose in most
phrases. Alternatively we assume that all phrases of a particular kind have
the number of bars necessary for accounting for the most complex example
of that kind and so all phrases may have different numbers of bars
a) e.g. perhaps PP has only one bar (there is no specifier of PP), AP and VP
have two bars and NP has three bars.
d. adjuncts
the original theory did not mention adjuncts at all and pre-X-bar theory didn’t
really have much to say on the subject either (adjuncts were simply included
inside the phrase along with complements). But the special status of
complements given in X-bar theory meant that something needed to be said.
ii.
Jackendoff (1977)
A. he recognised two kinds of adjunct, central and peripheral
a) adverbs of time place and manner are central
b) speaker oriented adverbs (sadly, worryingly, etc.)
B. These were included at different levels to reflect the fact that peripheral
adjuncts are further from the head
C. took the recursive aspect of adjuncts to be due to the unlimited number of
branches that could come from one node. This predicts that adjuncts are
not hierarchically organised.
iii.
Hornstein and Lightfoot
A. pointed out that adjuncts are hierarchically organised
a) semantically: white computer table
b) syntactically: a man in the shop with a beard and one with a moustache –
a man in the shop with a beard and one in the street with a moustache –
this man in the shop with a beard and that one
c) they claimed that a new structure was needed for adjunction which
allowed recursion: N’  N’ adjunct
d) they supported this by pointing out that each new N’ adds a property:
i) a student = one property (of being a student)
ii) a student of linguistics = still one property
iii) a student of linguistics with long hair = two properties (being a
student and having long hair)
iv) a student of linguistics with long hair who I met = three properties
(being a student, having long hair and having been met by me)
B. However, not everyone was convinced and some claimed that this kind of
adjunction (adjunction in the base) should not be allowed
e. the head of S
i.
in the original work Chomsky stated that sentences are introduced by a non-Xbar rule and so X-bar notions such as head, specifier and complement do not
apply
ii.
Jackendoff claimed that sentences are really VPs – headed by the verb. In this
way he made VP structure the same as NP structure with a uniform three bar
limit
iii.
however, others at the same time started talking about the inflection being the
‘head’ of the clause – but at the time this was not really brought into X-bar
theory
f. S and S-bar
i.
in the early 1970 Bresnan argued that complementisers were not included inside
S, but form a constituent with S
i.
A. S can be pronominalised, coordinated, gapped and even moved separately
from the complementiser
ii.
she called the constituent made up of the complementiser and S ‘S-bar’ =
however, this was not meant to indicate that this was part of X-bar theory (S
cannot be a head of anything as it is not a word)
iii.
some took this literally however, and started to propose an S’’ to introduce a
position for the topic – but this was misguided
g. the status of X-bar rules
i.
In the original paper, it was not entirely clear what the status of X-bar rules
should be
ii.
it is clear that X-bar rules cannot account for cross categorial differences – e.g.
V and P take bare NP complements, N and A cannot: V, N and A take sentential
complements, P cannot.
iii.
Jackendoff claimed that such observations necessitated specific Phrase structure
rules (N’  N PP ...). X-bar rules were to be seen as a template for allowable
phrase structure rules.
iv.
However, this was to be challenged in the early 1980s
4. 1980s developments
a. The status of X-bar theory
i.
Jackendoff
A. X-bar theory as a constraint on phrase structure grammar
a) there are cross categorial differences
i) N and A don’t take bare NP complements
ii) V and P do
iii) P cannot take clausal complements
iv) N, V, A do
v) N takes D specifiers, A takes degree specifiers, V has subject as
specifier
b) these cannot be lexical (how could we explain why all lexical elements of
a certain category behave in the same way),
c) they can’t be to do with X-bar either (this doesn’t refer to categories)
d) the only place which gives generalisations about categories is phrase
structure rules.
ii.
Stowell
A. X-bar theory as phrase structure grammar
a) if cross categorial differences can be explained by other principles, we
can reduce the role of phrase structure rules
b) if we can do this for all such differences, we don’t need phrase structure
rules at all
c) all we need is X-bar rules, which replace the phrase structure rule rather
than merely constrain them
d) e.g. the fact that N and A don’t have NP complements is just as much a
fact about the distribution of NPs. There are other restrictions on this
distribution
i) NPs don’t go in the subject position of non-finite clauses
ii) this came to be seen as a fact about the Case nominal elements bear:
a. the subject of a finite clause is a nominative position
b. NPs have to occupy Case positions (Case Filter)
c. the subject position of a non-finite clause is not a Case
position and so no NP can occupy it
iii) if we assume that the object positions of nouns and adjectives are
not Case positions either, we can account for the generalisation that
they do not have NP complements
iii.
discussion
A. although Stowell didn’t demonstrate that all categorial similarities and
difference can be reduced to independent considerations, the advantages of
his suggestion outweighed the remaining problems.
B. some of the problems were solved later: why do only NPs have
determiners as specifiers
C. some however still remain: why do Ps not take clausal complements?
b. the head of the sentence
i.
Jackendoff
A. sentence is a VP, so the head is V
B. these assumptions are allowed by Jackendoff’s universal three-bar system
as the subject is in the top most specifier of VP
ii.
Chomsky
A. started to hint at head of S being INFL
a) INFL is either finite or non-finite
b) clauses are finite or non-finite depending on what INFL they contain
iii.
Stowell
A. the first to take Chomsky’s proposals seriously and suggesting an IP
structure (though he didn’t extend it to S’ and complementisers)
B. under this proposal subjects were in the specifier of IP and VP is in the
complement of I
C. everything follows a uniform two bar hypothesis
iv.
Chomsky 1986
A. first to suggest that X-bar theory is universally applicable to all categories
v.
advantages
A. head status of inflection
B. word order SVO
C. fewer bars needed
D. inflection has a complement and specifier – not so in Jackendoff’s system
c. complementisers
i.
Bresnan
A. standard S’ analysis
B. doesn’t fit X-bar theory
ii.
Fassi Fehri
A. first to suggest that complementisers are heads of CP
iii.
Chomsky 1986
A. took IP/CP analysis for all clauses
iv.
advantages
A. uniform analysis for most things
B. complementisers and S’ fit X-bar theory
C. specifier of CP good for wh-phrase (before no real position for these)
d. problems of stopping here.
i.
at this point all lexical categories and half of the functional categories are X-bar
elements
ii.
but this is worse than previously as at least the lexical categories formed a
natural class
iii.
there is no natural class made up of N, V, A, P, I and C
5. DP
a. Background
i.
X-bar theory was partly motivated by an attempt to make NP structure similar to
clause structure
ii.
Bresnan’s S’ hypothesis added a slight complication to this as clausal structure
was extended in a way that NP structure was not (- there was no NP’, unless one
wanted to assume that NP extends for 3 bars, as Jackendoff had proposed, but
by the 1980s no one believed that – besides, S’ was not an X-bar extension of
the clause so even this would not have helped)
iii.
By the time the CP-IP analysis of the clause was proposed, the two structures
were as different as they could possibly be
b. First Step – Szabolsci
i.
noted that Hungarian seemed to have NPs which were similar to S’s in that:
A. there is a position to the left of the definite determiner, which must be
occupied by the wh-element:
a) Jánosnak a kalapja
a János kalapja
b) kinek a kalapja
* a ki kalpja
B. the element which occupies this position can be moved out of the NP, just
as a wh-phrase can move out of S’
a) Jánosnak eset le a kalapja a pólcról
b) * János eset le a kalapja a pólcról
ii.
She concluded that the Hungarian definite determiner is similar to a
complementiser
c. Abney 1987
i.
Starting from Szabolcsi, Abney argued that there is reason to believe that noun
phrase structure is more similar to clause structure in that there is functional
structure built on top of the phrase headed by the noun
A. there are languages with inflections such as agreement within the nominal
phrase – this is similar to the inflection in the clause
B. There are two phrasal positions in front of the noun: possessors and postdeterminers – in that order
a) The post-determiner seems to be in spec NP:
i) it follows possessors
ii) it precedes other N modifiers
iii) unlike other modifiers it is restricted to one
b) therefore the possessor, must be in the specifier of another head
c)
C. the English gerund seems to be a noun phrase containing a VP – but with a
functional projection on top of the nominal this is not such a problem
ii.
He then argued that the determiner is the head of this functional phrase
A. the main obstacle for this is the complementary distribution between
determiner and possessor – if they are in different positions, they should be
able to co-occur.
a) they can co-occur in many languages
b) they can even co-occur in English
i) John’s every move
c) it is possible that the possessive marker ‘s could be analysed as a
determiner, which would explain why other determiners don’t easily fit
iii.
However, the main argument in favour of the DP is that it extends X-bar theory
to the problematic categories of det and deg
A. Abney also proposed a similar analysis to DP for adjectival and adverbial
phrases which places the degree adverb as a functional head above the AP
B. From this perspective, everything conforms to X-bar
d. Remaining Problems
i.
Although Abney’s DP analysis made things symmetrical, Szabolcsi’s
observations are still not fully captured as Abney’s DP is more like IP and there
is no obvious equivalent to CP in the nominal structure.
ii.
it may be, as Szabolcsi first suggested that the definite determiner is different
from others in that it is like a complementiser rather than an inflection
A. but this would be odd as definiteness is related to indefiniteness and this
analysis would put these two things as independent
iii.
Perhaps there is another functional element equivalent to tense and agreement
which sits below the determiner inside the DP
A. that way determiners would be similar to complementisers and the nominal
structure would be the same as the clause
B. not all languages demonstrate agreement between possessors and
possessed. But perhaps there is enough of a similarity between tense and
number to warrant analysing a functional projection above NP but below
DP in the same way that IP is above VP but below CP
6. Problematic analyses: the gerund and coordination
a. Intro
i.
Jackendoff in 1977 identified two structures which he considered to be
problematic for X-bar theory: the English gerund and coordination
ii.
Although there have been attempts to provide X-bar analyses of these, they still
remain problematic to this day
b. The English Gerund
i.
the problem of the gerund noted by Jackendoff is that they are nominal phrases
which appear to be headed by a verb
A. we know they are nominal phrases because they distribute like nominal
phrases
a) we spoke about [his behaving badly]/* that he behaved badly
b) as for [his behaving badly]/ * that he behaved badly, I don’t like it
c) does [his behaving badly]/ * that he behaved badly worry you
B. we know the head is a verb because it is modified by an adverb and it can
take a bare nominal complement
a) his behaving badly / * bad
b) his fixing *(of) the car
ii.
Abney’s DP analysis provides a framework which might address this problem
as the DP is not projected from the thematic element but from the determiner.
Thus the fact that there is a verb inside the gerund is not an immediate problem
iii.
But Abney also pointed out other problems
A. even if we suppose that the gerund has a VP inside the DP instead of an
NP, this is a problem as D cannot normally take any other complement
than NP
B. in fact, the problem is worse because there is not just one type of gerund,
but 4:
a) the writing of poems = ing-of
b) his writing poems = poss-ing
c) him writing poems = acc-ing
d) PRO writing poems = PRO-ing
iv.
Abney’s suggested solution to this is to assume that –ing is a nominalising
morpheme which can nominalise at different levels of clausal structure.
A. If it nominalises the verb, then we have a completely nominal structure =
ing-of
B. if it nominalises the VP we get a DP taking an NP complement which is a
nominalised VP so it contains a V head
C. If it nominalises the I, we get a selected VP and no real determiner, but
something that projects a DP
D. if it nominalises the IP, we get a full IP structure which has the external
properties of DP
v.
The problem with this is the proposed behaviour of the nominalising morpheme
as there are no other morphemes which can either modify words or phrases and
the idea of a phrasal derivational morpheme is novel – but restricted to this
particular context
vi.
In BESE this problem is addressed with the help of a more articulated VP
structure, containing more heads that the morpheme can be related to. But the
problem still remains of the uniqueness of the morpheme – no other one
behaves like it: we don’t get adjectivising morphemes which can affect different
levels of the VP or NP etc.
vii.
Thus the gerund still must be viewed as an exception, which is the real problem
for X-bar theory as it predicts uniformality
c. Coordination
i.
Jackendoff’s problem with coordinate structures is of a different nature to that
of the gerund
A. X-bar theory is really a theory of subordination – how phrases can be
inside other phrases and its structure is essentially asymmetric with the
specifier higher than the complement
B. Coordination structures are essentially symmetric and do not involve
subordination
ii.
At the end of the 1980s, X-bar analyses of coordination began to be put
forward. Most of these assumed the coordinating particle to be the head and the
conjuncts that surround it to be in specifier an complement position
iii.
Problems for the X-bar analysis of coordinate structures
A. why do specifier and complement have to be the same category? Heads
impose restrictions on complements, not on specifiers
B. why does the coordinate structure behave like the coordinates, not a phrase
headed by a coordinating particle? If the head determines the category of
the phrase, all coordinate structures should have the same category and this
should not be affected by the specifier and complement
C. how can words and X’s be coordinated? The specifier and complement
positions are phrasal
D. how can there be more than two conjuncts?
iv.
there have been proposals put forward to address these issues, but nothing so far
has satisfactorily addressed them all.
7. The functional explosion: Articulated IP, CP and VP
a. Pollock and the expansion of IP
i.
Pollock (1989) was interested in verb positions in French and English
A. French = sub V adv obj sub V neg obj
B. Eng = sub adv V obj
sub neg V obj
ii.
he claimed that the difference is that French verbs move to I but English verbs
do not (perhaps I moves to V)
iii.
But in French infinitives there is another word order pattern:
A. French infin = sub V adv obj
sub neg V obj
iv.
His claim is that verbs in French infinitives move out of VP as the verb precedes
the adverb. But they do not move as far as I so that the verb does not move past
the negative.
A. So there must be a head position between I and V
v.
He analysed I as split into two parts: tense and agreement
A. these had been taken to be the same node mainly because English does not
separate them – but lots of languages do realise these separately as
independent morphemes
B. Polock had tense as the higher node and Agr as the lower one which the
French verb moves to in infinitival contexts
vi.
But there was disagreement as the standard opinion was that agreement, not
tense, is responsible for assigning nominative to the subject (Portuguese
agreeing infinitives have nominative subjects) and so Agr should be higher
otherwise the verb would precede the subject if it moved to tense.
A. Belletti therefore claimed that Agr was the higher inflection
vii.
Chomsky claimed that both Pollock and Belletti are correct as there is an
agreement morpheme above and below tense – one subject agreement and one
object agreement – therefore I splits into three, not two.
viii.
Part of Pollock’s analysis claimed that negation is a head.
A. Evidence from French: sub ne-V pas obj
B. analysis = ‘ne’ is the head of NegP and V moves through it on its way to
tense (or AgrS). ‘pas’ is in the specifier of NegP and so gets left behind
ix. After Pollock many others suggested that there are other heads between ‘I’ and
V:
A. Asp – possibly breaking up into Perf and Prog
B. Mod – possibly breaking up into epistemic and root (perhaps root breaks
up into deontic and dynamic)
C. maybe tense also breaks up into speaker and reference time
b. Rizzi and the expansion of the CP
i.
Along the lines of Pollock, Rizzi 1997 suggested that what we had been calling
CP should also be divided into smaller phrases. At the top is the place for the
complementisers to distinguish between the force of the sentence (interrogative,
or declatative), below this is a phrase for the topic to appear and below this a
phrase for the focus
ii.
We need all these as they are not in complementary distribution in all
languages:
A. in English a complementiser can be followed by a topic in embedded
clauses
B. in Hungarian a topic and a focus can be preceded by a complementiser in
embedded clauses
C. in English and Hungarian topics precede focus
D. in no language is the relative pronoun preceded by a complementiser
c. Larson and the VP shell
i.
the expansion of the VP actually started before Pollock’s articulated IP analysis.
Larson suggested that there are really two VPs, one above the other, to account
for double object and multiple complement verbs
ii.
his argument is that when there are more than one complement, another phrase
is needed otherwise we need to assume multiple branching (the binary
branching restriction was proposed by Kayne in 1984 and most people assumed
it to be true from then on).
iii.
moreover the structure of these constructions should be such that the first
complement is higher than the second
A. subjects are higher than objects and can be their antecedents
B. first complements can be antecedents for second complements and so must
be higher than them
C. He suggested two VPs, the highest containing no head but the subject in its
specifier (following Koopman and Sportiche who argued that subjects
originate inside VP). The lower VP has the first complement in its specifier
and the second complement in its complement position. The verb moves
from the lower V position to the higher one.
iv.
Katzer 1996 argued that the higher V position is really filled by a ‘voice’ head
which alternates between active, which provides the agent, and passive, which
does not:
v.
the notion of a light verb (first introduced by Grimshaw in 1990) then became a
popular way to view the VP shell and more light verbs were admitted to account
for different kinds of arguments
8. Critical Evaluation
a. growth or decay?
i.
We have reviewed the development of X-bar theory from 1970 till the early
1990s and it is time to take stock of what has happened. There have been major
changes over the years which can be mapped out as follows:
A. 1970 – start of X-bar
B. 1973 – introduction of S/S’ analysis
C. 1980-4 – introduction of IP/CP analysis
D. 1986/7 – DP hypothesis
E. 1986 – VP shell hypothesis
F. 1989 – articulated IP
G. 1993 – split CP
ii.
The question is whether this development marks steady improvement in our
conception of linguistic structure or a gradual realisation that we are perhaps on
the wrong track.
iii.
In order to assess the situation faced by X-bar theory we will concentrate on
particular notions within the theory and see whether they have become more or
less focussed.
b. Heads
i.
the development to the notion of heads has probably been the most dramatic in
X-bar theory. This is how it looks like:
A. 1970 = N V A
B. 1973 = N V A P
C. 1980? = N V A P I
D. 1982 = N V A P I C
E. 1986 = N V A P I C v
F. 1987 = N V A P I C v D Deg
G. 1989 = N V A P T Agr C v D Deg Neg
H. 1989 = N V A P T Agrs Agro C v D Deg Neg
I. 1990 - = N V A P T Agrs Agro C v D Deg Neg Mod Perf Prog
J. 1995 = N V A P T Agrs Agro C v D Deg Neg Mod Perf Prog Top Foc ...
ii.
While it might seem that this development is one in which the application of Xbar theory has been becoming more general, from about 1987 onwards a serious
flaw started to develop
iii.
the problem is in the notion of a category. Up until 1987 the number of
categories required by the theory was fairly low and could be handled by a
limited theory of categories (±N, ±V).
iv.
Afterwards, the number of new categories required exploded
v.
each of the new heads seemed to be in a category of their own, so very little
generalisation (apart from those of X-bar theory itself) can be made concerning
them
vi.
moreover they also interfere with generalisation we used to be able to make
about other categories.
A. e.g. we used to be able to say that I always took a VP complement.
B. but after the functional explosion we now need to say that it takes NegP,
PerfP, ProgP, ModP, vP, etc. as its complements
C. The same is true of other heads and so we lose generality in the system
vii.
All in all then, although up to 1987 it looked as though the theory was
improving by becoming more general, the process turned out to be rather
problematic as more and more heads we claimed to be subject to the theory.
This lost the ability to maintain a restricted theory of categories and the ability
to generalise about how categories behave.
c. Specifiers
i.
In 1970 specifier could be words or phrases (determiners and possessors)
ii.
The idea of a specifier was more like something with the function of a
determiner – something which specifies another thing which has somewhat
loser reference: man  a property – the man  an individual
iii.
But sometime after the introduction of IP (which had only a phrasal specifier) it
started to become popular to see specifiers as phrasal positions
iv.
this became universal after the DP hypothesis which took away the need to see
words in specifier positions
v.
All this is fairly positive in that now there is a strict distinction between word
and phrase positions
A. only head positions are words
B. all other positions are phrases
vi.
A further generalisation was introduced on the basis of the IP specifier being the
subject
A. Stowell (1984) had claimed that other phrases can have subject in their
specifier positions
B. so the specifier became associated with subjects
C. But Larson’s VP shell hypothesis spoiled this as in this certain objects
were in specifier positions
D. Other suggestions also put non-subjects in specifier position: Abney
claimed post determiners to be in specifier of NP and measure phrases in
specifier of AP
vii.
Other generalisations didn’t hold well
A. specifier of CP and IP are underlyingly empty and landing sites for
movement
B. other specifiers are filled at D-structure (VP, DP, etc.)
viii.
Overall, although there were some positive developments, the notion of the
specifier became less well defined – it seems to be a position for subjects, a
position to move to, a position for arguments, a position for certain modifiers –
this is not a coherent set
ix. Later developments also eroded certain restrictive claims about specifiers
A. One claim that had been made fairly early on was that the specifier is a
single position
B. However, with CP analysis where wh-elements move to specifier position,
there was the need to claim that some phrases can have multiple specifiers
a) some languages can move more than one wh-phrase to the beginning of
the same clause:
i) ki kit szeret
d. Complements
i.
the term complement was used in the 1970 because this was the position that the
complementation of verbs went
A. complement/complementation were traditional terms for the elements on
which the subcategories of verbs were determined (transitive = verb with a
nominal complement, ditransitive, verb with two nominal complements,
etc)
ii.
As X-bar theory developed, this could not be maintained
A. as soon as we envisage functional heads, the meaning of the ‘complement
position’ inevitably changes as functional heads have no complementiation
iii.
There was a time where an uncomfortable double use of the term existed:
complement of lexical head = complementation; complement of functional head
= whatever was situated in that position.
iv.
However, the VP shell hypothesis also took its toll on the original meaning of
the term (in the same way that it affected the notion of specifier). Not all
complements of verbs appeared in complement positions and not all
complement positions of verbs were occupied by complements (some had
adjuncts in them!)
v.
Another change, virtually the opposite of specifiers, was that the complement
position started off as possibly multiple (double object verbs and complex
transitive verbs seemed to have more than one complement). With the VP shell
hypothesis, the complement position was restricted to a single position
vi.
Over all, the complement position has become more restricted, which from a
theoretical position is a good thing. The main problem is terminology. The term
‘complement’ is now no longer the same as its traditional meaning – it names a
position, not a relationship
e. Adjuncts
i.
adjuncts have always been the least understood of elements – as illustrated by
early disagreements in X-bar as to how to treat them
A. no base adjunction
B. multiple branches at a certain level of projection
C. adjunction to X’
ii.
The main thing that the development of the theory did to adjuncts was cast
uncertainty on where they adjoin.
A. for those who accepted base adjunction, the site of the adjunct was X’
B. but under the DP analysis, it was possible to suggest that AP adjuncts
adjoin to NP not N’
C. It was always uncertain whether VP adjuncts adjoined to V’ or VP (some
adjoin to S – but with IP analysis, do they adjoin to IP or I’?)
iii.
One common claim that began at this time (probably to impose some order on
this chaos) was that adjunction to X’ should not be possible (phrases adjoin to
phrases in the same way that heads adjoin to heads).
iv.
Larson’s VP shell hypothesis attempted to get rid of adjuncts as an X-bar
category, situating all modifiers in complement positions of inner VPs
v.
None of these ideas have been universally accepted and the adjunct remains the
mystery it started off as.
f. Explanation
i.
An overall evaluation of a theory should take stock of how its development
works towards greater restriction and generalisation and therefore explanation.
ii.
X-bar theory certainly started to become more general and yet maintained its
restrictive nature until after the DP hypothesis
iii.
Its generalisation continued after this point, but its restrictiveness did not
iv.
the functional explosion led to a situation in which virtually anything could be
analysed as a head and fitted into X-bar theory.
v.
There was very little empirical support for many of these new heads and their
projected phrases
A. we started with visible heads – i.e. direct evidence of the existence of the
structure
B. then heads were motivated by things that moved either into the head
position itself, or into its specifier position – the head itself need not be
visible – i.e. indirect evidence of the structure
C. some heads were motivated by no observable phenomenon in a language:
there was no visible head, no visible movement to its head or specifier
position. The only motivation for these structures was the fact that there
was direct or indirect evidence that they exist in other languages. The
assumption that X-bar theory is universal and therefore what applies to one
language must apply to them all is the only evidence given for certain
assumed structure
vi.
Clearly this does not add to the explanatory power of the theory
vii.
the overall evaluation of the developments in X-bar theory are therefore not
particularly positive
9. Bare Phrase Structure
a. Minimalist Programme
i.
explanation – generalisation + limitation = reductionist
ii.
where does reduction stop?
iii.
Minimalist answer
A. grammar should contain nothing that isn’t motivated by its function
B. the function of grammar is to link interfaces: meaning and production
C. so everything in grammar is motivated by the need to externalise thought
b. X-bar is not minimalist
i.
a minimalist phrase structure component would take what it is given to start
with and build necessary structures using these and nothing else
ii.
X-bar structures are built from words, but they contain more than just the words
A. they also contain categories and indicators of bar levels
iii.
X-bar theory also contains stipulations which have no reason given by the
interface conditions
A. that there are at most two branches
B. that there are only two projection levels above the word
C. that everything beside the head is a phrase
D. that nothing adjoins to X-bar
c. Bare Phrase Structure
i.
there is one structure building principle motivated by the requirement that the
grammar builds structures (for meaning and expressing linear sounds)
A. Merge: take two elements and make one thing from them
ii.
what do we call this new thing?
A. if all we have to work with is the words we start with, then we must name
it with one of the words = the definition of a head
B. what isn’t the head is the complement
C. how do we know which one to pick?
a) we don’t – we pick one and if it produces something that conforms to the
interface conditions, it is grammatical. If not, it isn’t – that is all a
grammar has to do (note that this is not a model of language production)
iii.
the words that we use as labels are like categories as they are marked as such in
the lexicon – it is only at the bottom of the tree that such words have
phonological and semantic properties
d. Projection levels
i.
how do we distinguish between a word and a phrase if there are no projection
level indicators?
ii.
do we need indicators?
A. can we tell projection levels out of context?
B. No: student could be N, N’, NP
C. in context we can tell
a) the student – student = NP
b) the few students – student = N’
c) the student of linguistics – student = N
D. so we only need context to tell us what projection level we are dealing with
iii.
the contextual definition of bar levels
A. something which is not projected is a head
B. something which does not project is a phrase
C. something which is projected and which projects in an intermediate
projection
e. Consequences
i.
the word we select to name the thing formed by Merge is the head as it is not
projected, but is does project
ii.
the word that we do not select is a phrase, because it does not project
A. it is also a word because it is not projected – so things can be words and
phrases at the same time and we don’t need to represent this in terms of a
structure
B. it follows that everything that isn’t a head must be a phrase because it
doesn’t project
iii.
as all phrases are produced by merge which takes two elements and combines
them, all structures will be binary branching – not unary branching
A. but as words can be phrases, we don’t need unary structures
iv.
there can only be three structural elements : heads, phrases and intermediate
projections therefore there can only be two projection levels
f. What’s the difference between a complement and a specifier?
i.
complements are what we get the first time a head is merged into a structure
ii.
a specifier is everything else that isn’t a head after first merger
A. so there can be multiple specifiers
g. What’s an adjunct?
i.
we don’t really know
10. Optimality Theory and X-bar
a. OT basics
i.
Grammatical principles conflict e.g.:
A. verbs should support inflections
B. adverbs precede verbs
C. adverbs follow inflections
ii.
if grammaticality were absolute, there would be nothing grammatical
iii.
so grammaticality must be ‘soft’
A. principles can be violated without causing ungrammaticality
iv.
but there has to be a system to this – we can’t just violated principles because
we want to – there would be no grammar at all
v.
therefore, there must be a ‘reason’ for violating principles
vi.
the conflicting nature of principles suggests that the reason we violate some
principles is so that we can satisfy others
vii.
which do we satisfy, which do we violate?
A. apparently there are some principles which are stronger (more important)
than others: we violate the weak ones in order to satisfy the strong:
a) adverbs can precede inflections if this allows the adverb to precede the
verb and the verb to support the inflection:
i) verb support infl > adverb precede verb > adverb follow infl
ii) this is called ‘ranking’
b) note if the verb does not need to support the inflection (because
something else does) then the adverb can follow inflection and the
weakest principle is satisfied (in fact it must be satisfied:
i) * –s always win
ii) * wins always
iii) always wins
iv) But
v) has always won
vi) * always has won
vii) * has won always
b. What OT is good for
i.
it can explain exceptions
A. exceptions are not uncommon – they are in fact quite normal
B. most grammars take exceptions to be bad
C. from OT perspective, they are a natural part of the system
ii.
it can explain existent variation patterns
A. we can see that principles which hold in some languages don’t in others,
but then there are other languages where the principle holds in some cases
but not in others
a) some languages are configurational (conform to X-bar theory)
b) others are non-configurational (don’t conform to X-bar theory)
c) others (e.g. Hungarian) conform to X-bar theory with some structures but
not others
i) SVO, SOV, OVS, OSV, VSO, VOS are all possible meaning that
sentences appear to be non-configurational
ii) but NP/DPs don’t allow free word order, so they are configurational
B. this shows that variation isn’t an ‘on-off’ thing (parameters of GB and
minimalism)
C. OT claims that languages differ in terms of ‘re-ranking’
a) configurational principles high in English
b) configurational principles low in Warlpiri
c) configurational principles in the middle for Hungarian
c. How an OT grammar works
i.
we start with an input = words and stated relationships between them
ii.
a structure generator generates possible structures from the input
iii.
the set of structures are evaluated against ranked constraints
iv.
the structure which best satisfies the constraints is optimal and grammatical
A. a structure best satisfies the constraints if for every constraint it violates it
satisfies a higher ranked constraint that all other structures violate
d. OT and X-bar theory
i.
most work in OT assumes that X-bar theory is part of the structure generator,
which means it stands outside the OT system (is inviolable)
ii.
Grimshaw assumes that some aspects of X-bar are OT based
A. head first
B. specifier first
C. complement first
D. the ranking of these constraints can determine the order of elements within
an X-bar structure
iii.
Hordos argued that the notion of projection was subject to OT principles – the
gerund has a verbal head but projects a nominal phrase
iv.
Gaspar argued that coordinated structures were not organised on X-bar
principles but on similar grounds to the organisation of clauses
A. John likes ---- and Bill hates wine
B. John likes wine and ---- hates beer
C. the missing element in the coordinates has its antecedent placed depending
on where is it positioned in the clause: at the front or back
e. The potential of OT in X-bar theory
i.
the notion of a head is not so straightforward in some assumptions
A. although Grimshaw claims that I is the head of the clause, she also
considers IP to be an extension of VP (so V is the ultimate head of the
clause, like Jackendoff claimed).
a) verbs which take CP complements, also take IP complements because
both CP and IP are extensions of VP
B. Abney came to a similar conclusion in the DP analysis
a) why can adjectival modifiers of nouns not have a complement?
b) because they are heads taking NP as complement
c) but D takes NP complements, not APs
d) this is because the AP is an extension of the NP
e) we might also say that DP is an extension of NP
C. Thus, if IP and CP are really projected from V and AP and DP are
projected from N, it seems that two aspects of ‘head’ are separated
a) the category of the functional phrase is projected from the thematic
complement
b) the X-bar structure is projected from the functional head
11. Problems with the notion of structure
a. the notion of structure and where it came form
i.
Recall that phrase structure started as a notion which allowed American
Structuralists to maintain their radical empiricist stance (nothing that cannot be
directly observed can be included in a theory) in the face of an obviously
abstract phenomenon such as language
ii.
These days, no one accepts this assumption.
iii.
But most people still accept the concept of structure
iv.
While it is true that false premise can lead to valid conclusions, these
conclusions should at least be re-examined in the light of the undermining of
their original foundation.
b. classic tests for structure and what they show
i.
replacement test: if a string of words can be replaced by a single word in all
contexts they appear in, they form a phrase – e.g. the man can be replaced by
John
A. but suppose what is happening here is that the noun man is being preplaced
by John.
B. We know that there is a restriction on the use of singular nouns without
determiners and that (in English) proper nouns generally don’t appear with
determiners
C. These are ‘dependency relations’
D. but this by itself does not necessarily lead to the conclusion that
determiners and the nouns they accompany form a phrase
E. For a phrase to be formed, the grammar must have rules which does
something with the phrase
ii.
pronominalisation test: if a set of words can be replaced by a pronoun, then it
must be a phrase
A. This is the same as the distribution test and similarly to that, it does not
necessarily conclude that there are phrases
iii.
distribution test: if a string of words has a set of positions in which we can find
it in various expressions, it is a phrase because this indicates that the grammar
‘does something’ with that set of words – i.e. positions them in sentences.
A. but suppose what has distribution is words
B. if the distribution of words is given with respect to the distribution of other
words (on which they are dependent), then they will tend to distribute
together
C. this does not require that the grammar works with the notion of a phrase at
all, hence distribution is not a conclusive argument for the existence of
phrases.
iv.
coordination test: two words can be coordinated only if they are the same; two
sets of words can be coordinated, so they must be the same. But two things can
only be the same if they are ‘things’, hence they must be phrases.
A. again, though, it may be that words are being coordinated and they have
dependants, so these must accompany the coordinated words.
B. So coordination does not provide arguments that there must be phrases.
v.
structure dependency of syntactic processes: processes depend on structure to
work the way they do. For example, auxiliary inversion involves the finite
auxiliary of the main clause, which is not necessarily the nearest auxiliary to the
front:
the one who can run fastest will win 
a) will the one who can run fastest win
b) * can the one who run fastest will win
Thus there must be structure for this to work like this
A. But dependency relations introduce structure, without necessarily
introducing constituent structure
a) if A is dependent on B and C is dependent on A, there is a structural
hierarchy between these elements, but no necessary requirement that A
and B form a phrase which is used by the grammar
B. we can use dependency structure to limit grammatical processes without
concluding that sentences are structures in themselves
c. distributional paradoxes and the need for linear conditions
i.
Suppose that constituent structure is all that is needed to account for the
distribution of an element (word/phrase) – this is what the structuralists assumed
ii.
The following should be true:
A. if A is in complementary distribution with B and if B is in complementary
distribution with C, then A should be in complementary distribution with C
iii.
But there are many cases where this does not hold:
A. inverted aux, complementisers and wh-phrases
a) * had if I known ...
b) * a man who that I met
c) who had Mary seen
B. Hungarian negative, preverb and focus
a) * nem elment
b) * JANOS elment
c) JANOS nem ment el
C. determiners, quantifiers and possessors
a) * the every wish
b) *John’s the wish
c) John’s every wish
D. Hungarian focus and fronted negatives
a) * JANOS A PAPIRT olvassa el
b) * semmit JANOS nem olvassa el
c) senki semmit nem olvassa el
iv.
These observations require explanations other than structural ones to account
for:
A. auxiliaries and complementisers go in the same position, but wh-phrases
don’t. Therefore there is something else which blocks a wh-phrase and a
complementiser from appearing together = the doubly filled comp filter
B. possessors don’t go in determiner position, but something blocks the
possessor from being followed by most determiners = ???
v.
Such conditions are linear restrictions X must/cannot be followed by Y
d. the complexity of structural theories
i.
clearly it is not possible to account for distribution in terms of structural
conditions alone
ii.
so current theories account for distribution in terms of structural and linear
conditions together
iii.
but this is obviously a complex theory
e. the alternative?
i.
what hasn’t yet been investigated, but which should have been, is a theory in
which only linear conditions are used to account for distribution
12. Alignment Syntax
a. Linear Processing
i.
Clearly language is not organised on an absolute linear way (position 1 >
position 2 > etc.
ii.
A more sophisticated linear organisation is a transition network (finite state
grammar)
A. this allows you to return to a position from other positions in order to
simulate recursion
iii.
Chomsky (1957) showed this was not a good model for language as it could not
cope with discontinuous constituents
A. entering on position (state) cannot influence what goes on in another
iv.
Thus for a long while linear systems were thought to be inadequate for
modelling language and were not investigated
v.
However, this view of linear systems was dependent on an absolute view of
grammar. With OT came the view of relative grammaticality, which throws a
different light on linearity
b. Relative linearization
i.
relative linearization sees the linear order of an expression as a series of linear
relations between some of the elements of the expression rather than one which
holds of all the elements.
A. suppose we have four element a, b, c, d. These can be organised by placing
a and b with respect to c and c with respect to d
B. this will inevitably bring the elements into conflict with each other (which
is why we can’t do this with an absolute set of rules)
C. so the organising principles have to be soft and which element is placed in
its preferred position will depend on the ranking of the rules
a) if it is more important for a to be immediately in front of c than it is for d
to be, then the order b > a > c will be established and if it is more
important for c to be immediately after d than it is for a to be immediately
in front of c then the order b > a > d > c will be the result.
D. such an approach could also handle discontinuous ‘constituents’:
a) as there are no constituents from this point of view, the fact that the
relevant elements are separated by others is not in principle a problem
b) we can handle the phenomena in the following way: suppose there are
two elements, a and b, which are related to each other. Then there are
other elements, c and d, which are not directly related to a and b.
However, the grammar could arrange that, for some reason, c and d are
closer to a than b is, hence we would get two related elements separated
c. Alignment Syntax
i.
types of linear relation
A. ordering
a) x precedes y
b) x follows y
ii.
c) although these can produce the same orders, these conditions are not
equivalent as one element is the target and the other the host. If x must
precede y and there is no x, then the condition is vacuously satisfied, but
if x must precede y and there is no y then the condition is necessarily
violated.
B. Adjacency
a) x is adjacent to y
b) note that ordering relations do not entail adjacency and adjacency does
not entail ordering
i) if x must precede y, it can do so at any distance from y
ii) if x must be adjacent to y it can be so on either side of y
some phenomena
A. elbowing
a) this happens when two or more elements are in competition for a position
immediately at one side of a host and it is more important to maintain the
side than it is adjacency
b) suppose aPb > aAb > cPb > cAb – in this case a will be immediately
infront of b and c will be ‘elbowed’ into a more distant position but still
in front of b: c a b
c) e.g.
i) ioFv > ioAv > oFv > oAv
ii) the object wants to be in a position immediately after the verb, but so
does the indirect object. when there is just an object, it will be in this
position = saw Bill yesterday - * saw yesterday Bill
But when there is an indirect object this will win for the post verbal
position and the object will be elbowed into second place: = gave Bill
money - * gave money Bill
B. side switching
a) this happens when two or more elements are in competition for a certain
position and it is more important for the loser to be adjacent to the host
than it is to maintain its preferred order
b) suppose aAb > aPb > cAb > cPb – in this case a will be immediately in
front of b and c will be immediately behind b: a b c
c) e.g.
i) compFhead > compAhead > adjAn > adjPn
ii) the complement of a head wants to follow the head and an adjective
wants to be adjacent to the noun it modifies and preferably will be on
the left of the noun: tall man
but if the adjective has a complement, it cannot be adjacent to the
noun as its complement must follow it and hence separate the noun
from its adjectival modifier. In this case the adjective will switch
sides to be adjacent to the noun but on the right: a man fond of beer
13. Review
Download