Transcript PowerPoint

CAS LX 522
Syntax I
Episode 5b. Agree and movement
5.3-5.4
The Big Picture



Now that we’ve gotten some idea of how the
system works, let’s back up a bit to remind
ourselves a bit about why we’re doing what
we’re doing.
People have (unconscious) knowledge of the
grammar of their native language (at least). They
can judge whether sentences are good
examples of the language or not.
Two questions:


What is that we know?
How is it that we came to know what we know?
History

In trying to model what we know (since it isn’t
conscious knowledge) some of the first attempts
looked like this (Chomsky 1957):



Phrase Structure Rules
S  NP (Aux) VP
VP  V (NP) (PP)
NP  (Det) (Adj+) N
PP  P NP
Aux  (Tns) (Modal) (Perf) (Prog)
N  Pat, lunch, …
P  at, in, to, …
Tns  Past, Present
Modal  can, should, …
Perf  have -en
Prog  be -ing
An S can be rewritten as an NP, optionally an Aux, and a
VP. An NP can be rewritten as, optionally a determiner,
optionally one or more adjectives, and a noun. …
What we know is that an S has an NP, a VP, and
sometimes an Aux between them, and that NPs can have a
determiner, some number of adjectives, and a noun.

Phrase Structure Rules
S  NP (Aux) VP
VP  V (NP) (PP)
NP  (Det) (Adj+) N
PP  P NP
Aux  (Tns) (Modal) (Perf) (Prog)
N  Pat, lunch, …

P  at, in, to, …
Tns  Past, Present
Modal  can, should, …
Perf  have -en
Prog  be -ing

S
NP
Aux
VP

N Modal
V
Pat might eat
NP
N
lunch
History
In this way, many sentences
can be derived, starting from
S.
The tree-style structure is a
way to record the history of the
derivation from S to the words
in the sentence.
We model our knowledge of
English as a machine that
(ideally, when it’s finished) will
generate all of the sentences
of English and no others.
Pat might have been eating
lunch

If you can say
Pat ate
you can say
Pat had eaten
or
Pat was eating
or
Pat had been eating

It looks like the verb can
be past or present alone,
but with have it takes on
an -en (past participle)
form, and with be it takes
on an -ing (present
participle) form.

The first verb or auxiliary
takes on tense forms.


So, Chomsky proposed:
Affix
Hopping
Aux  (Tns) (Modal) (Perf) (Prog)
Tns  Past, Present
Modal  can, should, …
Perf  have -en
Prog  be -ing
Past  -ed
Yielding something like this:

If you build a sentence
this way, things aren’t
in the right order, but
there’s a simple
transformation that can
be done to the
structure to get it right.

Empirically, tense,
perfect have, and
progressive be each
control the form of the
verbal element to their
right.
S
NP
Aux
N Tns
Perf
VP
Prog V
Past
Pat -ed have -en be -ing eat
NP
N
lunch


So, Chomsky proposed:
Affix
Hopping
Aux  (Tns) (Modal) (Perf) (Prog)
Tns  Past, Present
Modal  can, should, …
Perf  have -en
Prog  be -ing
Past  -ed

Affix Hopping
SD: afx verb
SC: verb+afx

The affixes all “hop to the
right” and attach to the
following word.

An ancestor to the kinds of
movement rules we’ve been
exploring, and this
phenomenon specifically is
closely related to the Agree
operation we’ll be talking
about.
Yielding something like this:
S
NP
Aux
N Tns
Past
Pat
Perf
VP
Prog V
NP
N
have+ed be+en eat+ing lunch
History continues


Through the 60s there were
good people working hard,
figuring out what kinds of
phrase structure rules and
transformations are needed
for a comprehensive
description on English.
As things developed, two
things became clear:


A lot of the PSRs look pretty
similar.
There’s no way a kid acquiring
language can be learning these
rules.


Chomsky (1970)
proposed that there
actually is only a
limited set of phrase
structure rule types.
For any categories X,
Y, Z, W, there are only
rules like:
XP  YP X
X  X WP
X  X ZP
X-bar theory

If drawn out as a tree,
you may recognize the
kind of structures this
proposal entails. These
are structures based on
the “X-bar schema”.


XP  YP X
X  X WP
X  X ZP
YP being the “specifier”,
WP being an “adjunct”, ZP
being the “complement”.
Adjuncts were considered
to have a slightly different
configuration then.
XP
YP
X
WP
X
X
ZP

GB
First, a “deep structure” (DS)
tree is built, however you like
but


Around 1981, the view
shifted from thinking of the
system as constructing all
and only structures with
PSRs and transformations
to a view in which
structures and
transformations could
apply freely, but the
grammatical structures
were those that satisfied
constraints on (various
stages of) the
representation.



Then, adjustments are made to
get the “surface structure” (SS)



Selectional restrictions must be
satisfied
q-roles must be assigned
Etc.
Things more or less like Affix
Hopping, or moving V to v, or
moving the subject to SpecTP.
Further constraints are verified
here: Is there a subject in
SpecTP? Etc.
Finally, the result is assigned a
pronunciation (PF), and,
possibly after some further
adjustments, an interpretation
(LF).
Which brings us to 1993

The most recent change
in viewpoint was to the
system we’re working with
now (arising from the
Minimalist Program for
Linguistic Theory).

The goal of MPLT was to “start
over” in a sense, to try to make
the constraints follow from
some more natural
assumptions that we would
need to make anyway.

The constraints that
applied to the structures in
GB were getting to be
rather esoteric and
numerous, to the extent
that it seemed we were
missing generalizations.

This new view has the
computational system working
at a very basic level, forcing
structures to obey the
constraints of GB by enforcing
them locally as we assemble
the structure from the bottom
up.
Features and technology


The use of features to drive
the system (uninterpretable
features force Merge,
because if they are not
checked, the resulting
structure will be itself
uninterpretable) is a way to
encode the notion that lexical
items need other lexical
items.
What the system is designed
to do is assemble
grammatical structures
where possible, given a set
of lexical items to start with.


A comment about the
technology here:
The operations of Merge,
Adjoin, Agree, and feature
checking, the idea that
features can be
interpretable or not (or, as
we will see, strong or weak)
are all formalizations of an
underlying system, used so
that we can describe the
system precisely enough
to understand its
predictions about our
language knowledge.
Features and the moon

We can think of this initially
as the same kind of model
as this:
m1m2
f G 2
r

The Earth and the Moon
don’t compute this. But if we
write it this way, we can
predict where the Moon will
be.

Saying lexical items have
uninterpretable features
that need to be checked,
and hypothesizing
mechanisms (matching,
valuing) by which they
might be checked is
similarly a way to formalize
the behavior of the
computational system
underlying language in a
way that allows us deeper
understanding of the
system and what it predicts
about language.
The “Minimalist Program”


The analogy 
with the
gravitational
force
equation isn’t 
quite
accurate,
given the
underlying

philosophy of
the MP.
The

Minimalist
Program in
fact is trying
to do this:
Suppose that we have a cognitive system for
language, which has to interact with at least
two other cognitive systems, the conceptualintensional and the articulatory-perceptual.
Whatever it produces needs to be interpretable
(in the vernacular of) each of these cognitive
systems for the representation to be of any
use.
Suppose that the properties of these external
systems are your boundary conditions, your
specifications.
The hypothesis of the MPLT is that the
computational system underlying language is
an optimal solution to those design
specifications. So everything is thought of in
terms of the creation of interpretable
representations.
And now, back to our
Program


With the history laid out in
a little more detail now,
on to the final major
operation and concept in
the computational system
we are hypothesizing.
Operations:






(Select)
Merge
Adjoin
Agree
Move
Hierarchy of Projections:

T>v>V




We’ve covered Merge and
Adjoin.
Select is the name for
picking a syntactic object off
the workbench in
preparation for doing
something.
Agree is what we’re about
to discuss.
We’ve seen examples of
Move already, though we’ll
try to further clarify its
function and restrictions as
a syntactic operation.
Pat might eat lunch.



Pat
v
eat
lunch
might
[N, …]
[uN, …]
[V, uN, …]
[N, …]
[…]
TP
NP
Pat
T [T, uN, …]
Since [uN] needs to be
vP
T
checked on T, and since
might
there are no NPs left to
v
Merge, T looks down into the [T, uN, …] <Pat>
tree, finds the first NP it
v+V
VP
sees, and moves it up.
eat
 “Moves it up” = makes a copy
<eat> NP
and Merges with the root.
lunch
We’ll continue exploring this
in a bit.
Uninterpretable features

Some lexical items come in with
uninterpretable selectional features.



Kick has [uN], it needs an NP.
Merge creates a sisterhood relation.
Uninterpretable features can be checked
by matching features in a sisterhood
relation.

These features are privative— they’re either
there or they’re not.
Uninterpretable features


There is a second kind of uninterpretable feature, an
unvalued feature.
For example, tense features come in two types, past
and present. We say that the tense feature either has
a present value or a past value.


[tense:past] or [tense:present]
An unvalued feature is a type of uninterpretable
feature, but it is not checked by exactly matching a
feature in the structural sister, but is instead checked
by taking on a value from a higher feature of its type.


T[tense:past] … v[utense:]
T[tense:past] … v[utense: past]
Uninterpretable features

Unvalued features


[utype:] is checked on a lexical item Y when a
c-commanding lexical item X has a feature
[type: value], resulting in [utype: value] on Y.
(Privative) uninterpretable features

[uF] is checked on a lexical item Y when its
sister has a matching feature (either [F] or
[uF]).
Feature classes



The feature types that can take on values are the ones
that we’ve been treating as being in classes up to now.
There are tense features. Like past, like present.
There are case features. Like nom, like acc. There are
person features. Like 1st, like 2nd. There are gender
features. Like masculine, like feminine.
So, we can think of this as a feature category or
feature type that has a value.


[Gender: masculine]
[Tense: past]
[Person: 1st]
[Case: nom]
Pat ate lunch


So, back to Pat ate lunch.
T here is just T with a past tense feature:



[T, tense:past, …].
We need to make a connection between the tense
feature chosen for T and the tense morphology
we see on the verb.
Here’s how we’ll do it:


Little v has an uninterpretable (unvalued) inflectional
feature [uInfl:]. It will be valued by [tense:past] on T.
It’s “Infl” because we want to include tense, but also
other kinds of features later on. But tense-type
features can check and value unvalued Infl-type
features.
Pat ate lunch.

Pat
v
eat
lunch
T
[N, …]
[uN, uInfl:, …]
[V, uN, …]
[N, …]
[T, tense:past, …]
vP
T
[tense:past,
v
T, uN, …] NP
Pat
v[uInfl:]+V
VP
eat
<eat> NP
lunch
Pat ate lunch.

Pat
v
eat
lunch
T
[N, …]
[uN, uInfl:, …]
[V, uN, …]
[N, …]
[T, tense:past, …]
T [T, uN, tense:past, …
vP
T
[tense:past,
v
T, uN, …] NP
Pat
v[uInfl:past]+V
VP
eat
<eat> NP
lunch
Pat ate lunch.




Pat
v
eat
lunch
T
[N, …]
[uN, uInfl:, …]
[V, uN, …]
[N, …]
[T, tense:past, …]
Last point, how does this
come to be pronounced Pat
ate lunch?
T isn’t pronounced as
anything. It was just a pure
tense feature.
The “past” pronunciation of
eat is ate, so v+V is
pronounced “ate” here.
TP
NP
Pat
T [T, uN, tense:past, …
vP
T
[tense:past,
T, uN, …] <Pat> v
v[uInfl:past]+V
eat
VP
<eat>
NP
lunch
Pat ate lunch.


Pat
v
eat
lunch
T
[N, …]
[uN, uInfl:, …]
[V, uN, …]
[N, …]
[T, tense:past, …]
The idea is that there is a
somewhat separate list of
pronunciation rules that tell
us how to pronounce eat in
the context of a valued uInfl
feature:

eat by [uInfl:past] = ate
TP
NP
Pat
T [T, uN, tense:past, …
vP
T
[tense:past,
T, uN, …] <Pat> v
v[uInfl:past]+V
eat
VP
<eat>
NP
lunch
Pat had eaten lunch

The auxiliary verbs have and be are used in
forming the perfect and progressive,
respectively, which are additional forms that a
verb can take on.


We can’t have two modals, but we can have a
modal and an auxiliary:



Pat has eaten lunch. Pat is eating lunch.
Pat should have eaten lunch.
Pat might have been eating lunch.
Conclusion: Auxiliaries aren’t T, they’re their own
thing. Let’s call have Perf and be Prog.
Pat had eaten lunch


Suppose that Perf can value an Infl
feature, so in Pat had eaten lunch, v+V
has
[uInfl: Perf], pronounced as “eaten”.
But auxiliaries show tense distinctions too,
so they must themselves have an
unvalued Infl feature.


Pat has eaten lunch. Pat had eaten lunch.
Have [Perf, uInfl: ]
Pat had eaten lunch.


Pat
v
have
eat
lunch
T
[N, …]
[uN, uInfl:, …]
[Perf, uInfl:, …]
[V, uN, …]
[N, …]
[T, tense:past, …]
TP
NP
Pat
T [T, uN, tense:past, …]
T
[tense:past, T, uN, …]
PerfP
Perf
vP
[Perf, uInfl:past]
had <Pat> v
NOTE: This tree and the next
one are not quite in final form.
We will soon discuss one
further operation that happens
to the highest Perf or Prog
head.
v[uInfl:perf]+V
eaten
VP
<eat>
NP
lunch
Pat was eating lunch.



Pat
v
have
eat
lunch
T
[N, …]
[uN, uInfl:, …]
[Perf, uInfl:, …]
[V, uN, …]
[N, …]
[T, tense:past, …]
TP
NP
Pat
T [T, uN, tense:past, …]
ProgP
T
[tense:past, T, uN, …]
Prog
vP
[Perf, uInfl:past]
was <Pat> v
Notice that what we have here
is a modern implementation of
Affix Hopping:
Each of T, Prog, and Perf
regulate the form of the
subsequent verbal form by
valuing the next Infl feature
down.
v[uInfl:prog]+V
eating
VP
<eat>
NP
lunch
Hierarchy of Projections

Both have and be (Perf and Prog) are
possible, but just in that order.
Pat had been eating lunch.
 *Pat was having eaten lunch.


But neither is obligatory. Thus:

Hierarchy of Projections
T > (Perf) > (Prog) > v > V
Negation


Pat might not eat lunch.
Given everything we have so far, it’s fairly
clear where not must be in the structure.
might is a T
 Pat is in SpecTP
 eat lunch is a vP (with a trace of Pat in
SpecvP)


So, assuming not is a head (of category
Neg), we have a NegP between TP and
vP.
Pat might not eat lunch.





Pat
v
not
eat
lunch
might
[N, …]
[uN, uInfl:, …]
[Neg, …]
[V, uN, …]
[N, …]
[T, …]
Pat might not have eaten lunch.
Pat might not be eating lunch.
Pat might not have been eating
lunch.
Hierarchy of Projections:
T > (Neg) > (Perf) > (Prog) > v > V
TP
NP
Pat
T [T, uN, tense:past, …]
NegP
T
might
Neg
not
vP
<Pat> v
v+V
eat
VP
<eat>
NP
lunch
Pat was not eating lunch.




Now suppose that we
tried to form Pat was
not eating lunch.
This is the result.
HoP says:
T > (Neg) > (Perf)
> (Prog) > v > V
But the words are not
in the right order: be
(is) should be before
not.

What did we do when
we ran into this
problem with give?
TP
NP
Pat
T
[tense:past]
T
NegP
Neg
ProgP
not
Prog
vP
be
<Pat> v
v+V
eat
VP
<eat>
NP
lunch
Pat was not eating lunch.






It seems that be (Prog)
moves up to T.
Pat is not <is> eating
lunch.
Same thing happens
with Perf:
Pat has not eaten
lunch.
TP
NP
Pat
T
[tense:past]
Only the top one moves
over negation:
Pat has not been eating
lunch.
T
NegP
Neg
ProgP
not
Prog
vP
be
<Pat> v
v+V
eat
VP
<eat>
NP
lunch
Pat was not eating lunch.


What we will assume is that
this generally happens (the
closest Prog or Perf moves
to T)…
…having not in the sentence
only revealed that it
happened.
TP
NP
Pat
T
[tense:past]

If there’s no not, you can’t
tell whether Prog has moved
up to T or not.

Just like V moving to v; you
can tell for distransitives like
give because V moves past
the Theme. In transitives like
eat, you can’t hear it move,
but we assume for uniformity
that it does.
T
NegP
Neg
ProgP
not
Prog
vP
be
<Pat> v
v+V
eat
VP
<eat>
NP
lunch
Auxiliaries moving to T


So, we have observed that empirically, have and be
seem to move to T.
However:





Non-auxiliary verbs do not move to T.
Auxiliaries should not have moved to T if there is already a
modal.
This is something special re: have and be.
How can we make this follow from our grammar?
Building on the idea that [uInfl:] on the auxiliaries are
valued by T, let’s say that the [uInfl:] feature on the
auxiliaries is special in that just being c-commanded
by a checker isn’t good enough. It has to be close.
Strong features





We will differentiate between two kinds of
unvalued features, strong and weak.
Strong unvalued features can only be
valued/checked locally.
This will generally require that the lower one be
moved up so that it is local.
Weak features can be valued/checked as we’ve
been discussing so far.
So, the [uInfl:] feature on Aux (Perf or Prog) is
strong if (tense is) valued by T. Otherwise, [uInfl:]
(including on v) is weak.
Pat was not eating lunch.

So, T values [uInfl:]
on Prog, but [uInfl:]
isn’t checked until it
is local to T.
TP
NP
Pat
T
T
[tense:past]
NegP
Neg
ProgP
not
Prog
vP
be [uInfl:*]
<Pat> v
v+V
eat
VP
<eat>
NP
lunch
Pat was not eating lunch.


So, T values [uInfl:] on
Prog, but [uInfl:] isn’t
checked until it is local
to T.
TP
NP
T
Pat
NegP
T+Prog
[tense:past]+[uInfl:past*]
Neg
ProgP
Once Prog has moved
not
up to fuse with T,
<be>
vP
[uInfl:past] is local to
[tense:past] and
<Pat> v
[uInfl:past] can be
checked.
v+V
VP
 As we’ll say next time,
this “fusion” is really a
eat
variation of Adjoin.
<eat> NP
lunch
Closing comment

It turns out that this (moving verbal elements) to
T is common crosslinguistically, and is a point at
which languages vary.


French raises all verbs and auxiliaries to T, Swedish
doesn’t raise either verbs or auxiliaries to T, English
raises auxiliaries but not verbs to T, etc.
In general, this is thought of a variation in feature
“strength”— we’ll discuss this more next time.









