E85-1018

THE
BY
R E S O L U T I O N OF L O C A L S Y N T A C T I C A M B I G U I T Y
THE HUMAN SENTENCE PROCESSING MECHANISM.
Gerry Altmann
D e p a r t m e n t of L i n g u i s t i c s
U n i v e r s i t y of E d i n b u r g h
George Square
E d i n b u r g h EH8 9LL.
(GB)
ABSTRACT
o n l y way to m a k e the w h o l e s t r i n g
into
a
sentence
is
to i n t e r p r e t it as a r e d u c e d
form of the passive.
However, w h a t
seems
to
happen
in this (and s i m i l a r ) e x a m p l e s
is that p e o p l e tend to i n t e r p r e t the
word
"sold"
as
the
m a i n verb.
This t e n d e n c y
leads t h e m d o w n a s y n t a c t i c g a r d e n path.
The
resolution
of
local
syntactic
a m b i g u i t y by the H u m a n S e n t e n c e P r o c e s s i n g
M e c h a n i s m is a t o p i c
which
has
provoked
c o n s i d e r a b l e i n t e r e s t in r e c e n t years.
At
is s u e
is
whether
such
ambiguities
are
resolved
on
the
basis
of
syntactic
i n f o r m a t i o n a l o n e (cf.
Minimal Attachment
Frazier,
1979),
or
whether
they are
r e s o l v e d on some
other
basis.
Crain
&
Steedman
(1982)
suggest
that
the
resolution
process
is
governed
not
by
Minimal
A t t a c h m e n t but i n s t e a d by w h e t h e r
or not
a
referring
expression
provides
sufficient
information
with
which
to
identify
a
unique
referent.
Such
an
approach
relies
on
the
provision
of
adequate contextual information, something
which
has
been
lacking
in
experiments
w h i c h have been c l a i m e d to s u p p o r t M i n i m a l
Attachment.
In
this
paper
I
shall
c o n s i d e r a n u m b e r of such e x p e r i m e n t s , and
the
different
patterns
of r e s u l t s w h i c h
emerge
once
contextual
information
is
provided.
Although
the
importance
of
c o n t e x t u a l i n f o r m a t i o n will be s t r e s s e d , I
shall b r i e f l y c o n s i d e r r e a s o n s why p a r s i n g
p r e f e r e n c e s a r i s e in the
absence
of
any
e x p l i c i t p r i o r context.
The c o n c l u s i o n is
that
computational
models
of
syntactic
ambiguity
resolution
which
are b a s e d on
evidence
which
has
ignored
contextual
considerations
are
models
of
something
o t h e r than n a t u r a l l a n g u a g e p r o c e s s i n g .
So the H S P M e x h i b i t s a p r e f e r e n c e for
one
a n a l y s i s over a n o t h e r w h e n f a ce d w i t h
a local a m b i g u i t y .
But why?
A number
of
suggestions
have
been
made
concerning
this.
One suggestion, originally proposed
by
Kimball
(1973)
and
f o l l o w e d up m o r e
r e c e n t l y by
Frazier
(1979)
and
Rayner,
Carlson,
&
Frazier
(1983),
is that the
HSPM
takes
into
account
the
syntactic
structure
of
these sentences.
T h e r e are
two p o s s i b l e
structures
which
could
be
assigned
to
the
ambiguous
sentence
fragment
The
oil
oil t y c o o n sold
tracts . . .
the
off-shore
The
reduced
passive
interpretation
requires
an
e x t r a NP node as c o m p a r e d to
the
main
verb
interpretation.
Kimball
(1975)
and F r a z i e r s u g g e s t that w h e n m o r e
t h a n one i n t e r p r e t a t i o n is
possible,
one
pursues
that i n t e r p r e t a t i o n w h i c h c r e a t e s
the s t r u c t u r e w i t h f e w e s t nodes.
This
is
what
Frazier
calls
the
Principle
of
Minimal Attachment.
This s t r u c t u r a l h y p o t h e s i s
proposes,
then,
t h a t an i n i t i a l d e c i s i o n is m a d e on
g r o u n d s of s y n t a c t i c s t r u c t u r e alone.
If
it
s u b s e q u e n t l y t u r n s out to be the w r o n g
d e c i s i o n (on g r o u n d s of " i m p l a u s i b i l i t y " ) ,
the
alternative
analysis
(which
is
identified
on
the
basis
of
"thematic
s e l e c t i o n " - R a y n e r et al., 1983) is then,
and o n l y then, a t t e m p t e d .
In
support
of
this
claim,
Rayner
et
al.
collected
r e a d i n g times and eye
movement
data
for
sentences
which,
syntactically speaking,
allow
two
attachment
sites
for
a
prepositional
phrase:
one a t t a c h m e n t , to
an
NP,
requires
an
extra
NP
node
as
c o m p a r e d to the o t h e r a t t a c h m e n t , w h i c h is
to a VP.
There
has
been
much
controversy
recently
surrounding
the
processes
r e s p o n s i b l e for the " g a r d e n
path"
effect
Ln the f o l l o w i n g kind of example:
The oil t y c o o n sold the
off-shore
oil
tracts
for
a
lot
of m o n e y
w a n t e d to kill J.R.
The g a r d e n path e f f e c t a r i s e s here b e c a u s e
the
Human
Sentence
Processing Mechanism
("HSPM") e n c o u n t e r s , d u r i n g the p r o c e s s i n g
of
this
sentence,
a
local
syntactic
ambiguity.
The word "sold" is
ambiguous:
it
can
be i n t e r p r e t e d e i t h e r as a s i m p l e
active, or it can be i n t e r p r e t e d as a past
participle,
in
a
reduced
passive.
The
123
The b u r g l a r b l e w o p e n the
safe
with
d y n a m i t e (Minimal a t t a c h m e n t to VP)
The b u r g l a r b l e w o p e n
diamonds (Non-minimal
the
safe
attachment
S t e p h e n C r a i n did just this: u s i n g an
incremental grammaticality
j u d g e m e n t task,
and
a
class
of
ambiguity
which
is
different
in
form
but
the
same
in
principle
(see
below),
he
showed
that
garden
path
e f f e c t s c o u l d be o v e r c o m e or
induced
depending
on
the
referential
nature
of
the c o n t e x t (i.e. d e p e n d i n g on
w h e t h e r just one "oil t y c o o n " or m o r e t h a n
one
"oil
tycoon"
had b e e n i n t r o d u c e d in
the p r e c e d i n g text).
The
work
of
Crain
suggests
that
a
major
methodological
s h o r t c o m i n g in the w o r k of F r a z i e r (1979),
Rayner
et
al.
(1983)
and
Ford et al.
(1982)
is
the
absence
of
any
context
within
which
the
particular
syntactic
constructions they were studying could
be
a l l o w e d to f u n c t i o n .
T h e r e is thus a v e r y
d e f i n i t e s e n s e in w h i c h the p r o c e s s i n g
of
these
constructions
will
have
been
abnormal.
It f o l l o w s t h a t a s u i t a b l e t e s t
of
R a y n e r et a l . ' s c l a i m s is to r e p l i c a t e
their experiment using
the
same
reading
time
task,
but
with
the
provision
of
c o n t e x t s w h i c h are f e l i c i t o u s w i t h one
or
other
of
the
two
versions
of
their
examples.
the
with
the
to NP)
In the case of the n o n - m i n i m a l l y
attached
version,
the
correct
a t t a c h m e n t (to the
NP) w o u l d
be
attempted
only
after
the
minimal
attachment
to
the
VP had f i r s t
been
tried.
As
they
had
predicted,
reading
times
to
the
non-minimally
attached
versions
were
significantly
~onqer
than
to
the
minimally
attached
versions.
An a l t e r n a t i v e to M i n i m a l
Attachment
is
proposed
by
Ford,
Bresnan, & Kaplan
(1982), who s u g g e s t that t h e s e p r e f e r e n c e s
arise
from
the
order
in
which
l e x i c a l / s y n t a c t i c r u l e s in the g r a m m a r can
be
accessed
(of.
Wanner's
"implementation"
of
Minimal
Attachment,
1980).
Ford
et
a l . ' s t h e o r y of L e x i c a l
P r e f e r e n c e is m o r e p o w e r f u l
than
Minimal
Attachment
because
this o r d e r i n g can, in
part, be d e t e r m i n e d by the a c t u a l
lexical
items
which
are involved.
But t h e s e two
proposals
are
both,
in
effect,
structurally
based.
T h e y take no a c c o u n t
of
the
referential
function
of
the
particular constructions involved.
The following contexts
were
devised
for
an
e x p e r i m e n t (Altmann, f o r t h c o m i n g ,
and Altmann & Steedman, forthcoming):
T__oo ~ n d u c e
S t e p h e n C r a i n (Crain, 1980, d i s c u s s e d
in C r a i n & S t e e d m a n ,
1982) n o t e d that w h a t
m a n y of the g a r d e n - p a t h s e n t e n c e s h a v e
in
common
is
that
of
the
two
possible
a n a l y s e s , one is
functionally
equivalent
to,
or
identical
with,
a
restrictive
r e l a t i v e clause.
N o u n p h r a s e s a r e used
by
the
speaker
to
refer
to
objects.
The
f u n c t i o n of a r e s t r i c t i v e r e l a t i v e
is
to
give
additional
i n f o r m a t i o n as to who or
what
is
being
talked
about.
This
additional
information
is
necessary
because
without
it,
there
would
be
insufficient
evidence
with
which
to
d e t e r m i n e who or w h a t was
being
referred
to.
So
restrictive
relatives
are used
b e c a u s e if w e ' d just h e a r d the
expression
"the
oil
tycoon"
or "the safe" we w o u l d
not have k n o w n just w h i c h
candidate
"oil
tycoon"
or
which
candidate
"safe"
was
intended.
But wi%ere
do
these
different
" c a n d i d a t e ....oil t y c o o n s " and "safes" come
from?
N o r m a l l y , they m u s t
presumably
be
introduced
into
the
d i s c o u r s e some time
before
these
target
sentences
are
encountered,
and
represented
by s p e a k e r
and h e a r e r in some kind of
model
of
the
discourse.
T h e r e is a s e n s e in w h i c h all
of t h e s e e x a m p l e s
are
unnatural
because
each
sentence
is p r e s e n t e d in i s o l a t i o n .
We refer to
"the
oil
tycoon"
and
"the
off-shore
oil
tracts",
but
we've never
m e n t i o n e d them
before.
To
control
for
this,
we
should
really
present
these
t a r g e t s e n t e n c e s e m b e d d e d in a c o n t e x t .
attachment
to NP
A b u r g l a r c a r r y i n g some
dynamite
broke
into
an
h e i r e s s ' s house.
O n c e i n s i d e he f o u n d
two
safes.
One
of
them
had
some d i a m o n d s
inside
whilst
the
other
had
several priceless emeralds.
To
induce
attachment
to VP
A burglar carrying some
dynamite
broke
into
an
h e i r e s s ' s house.
O n c e i n s i d e he f o u n d a safe and a
j e w e l r y box. One of them had some
d i a m o n d s i n s i d e w h i l s t the
other
had s e v e r a l p r i c e l e s s e m e r a l d s .
Minimal
(V_~P) a t t a c h m e n t
The b u r g l a r b l e w o p e n
w i t h the d y n a m i t e .
Non-minimal
(NP)
the
safe
attachment
The b u r g l a r b l e w o p e n
w i t h the d i a m o n d s .
the
safe
These examples
are
"minimally-different"
to
the
extent
that
the o n l y d i f f e r e n c e
between
them
is
the
change
from
"two
safes"
to
"a
safe
and
a j e w e l r y box".
This is a c h a n g e which, in theory, a f f e c t s
only
the
cardinality
of
the
set
of
"safes".
It was
124
found
that
there
was
a
strong
effect
of r e f e r e n t i a l c o n t e x t on the nonminimally
(NP)
attached
targets
(230
msec) .
Furthermore,
reading
times
to
these
were
considerably
faster
than
reading
times
to
the
minimally
(VP)
a t t a c h e d v e r s i o n s (there was a
difference
of
348
msec
in
the
"NP-inducing"
c o n d i t i o n s , and 190 m s e c o v e r a l l I ).
This
is
of c o u r s e the r e v e r s e of w h a t w o u l d be
expected
on
a
Minimal
Attachment
or
I,exical
Preference
account.
N e i t h e r of
these
could
account
for
this
effect
without
having
to
discount
the
experimental
evidence
which
currently
s u p p o r t s them.
been s a t i s f i e d by the p r e c e d i n g d i s c o u r s e ,
t h e n that c o n s t r u c t i o n
will
be
favoured
over
a
construction
whose
associated
presuppositions have not
been
satisfied.
This
notion
is
important
because
its
a p p l i c a t i o n to a n o t h e r c l a s s of
ambiguity
phenomena
suggests
that
evidence
p r e v i o u s l y t h o u g h t to
favour
lexical
or
structural
accounts
of
the
resolution
p r o c e s s does not
bear
on
the
issue
of
a m b i g u i t y r e s o l u t i o n at all.
In the f o l l o w i n g
ambiguous
example,
the
complement
clause
analysis
of
the
" t h a t - c l a u s e " is p r e f e r r e d to the r e l a t i v e
clause
analysis
(e.g.
Wanne~, Kaplan, &
Shiner, 1974).
H o w e v e r , it was a l s o f o u n d that t h e r e
was
n__oo e f f e c t of c o n t e x t on the m i n i m a l l y
(VP) a t t a c h e d t a r g e t s (the
difference
in
reading
time
across
the
two
context
c o n d i t i o n s was o n l y 78
msec).
This
was
surprising
given
that
the " V P - i n d u c i n g "
c o n t e x t s h o u l d h a v e been
felicitous
with
this target, and the " N P - i n d u c i n g " c o n t e x t
infelicitous.
It
became
apparent,
however,
that
neither
of t h e s e c o n t e x t s
was in fact f e l i c i t o u s w i t h V P - a t t a c h m e n t .
The boy
told
the
liked the s t o r y
were
he
With regard
to
the
preference
for
complements
over
relatives,
Crain's
original
demonstration
of
referential
context
effects
used
examples
which
exhibited
this
same
class
of
local
ambiguity.
However, by the n a t u r e of the
t a s k he e m p l o y e d , C r a i n
did
not
address
the issue of c o m p l e x i t y .
Restrictive relatives
provide
Given
i n f o r m a t i o n , and the i n f o r m a t i o n c o n t a i n e d
within
the
relative
must
therefore
be
matched
against information which already
exists
in
the
hearer's
model
of
the
universe.
This
matching
process
presumably requires a
certain
amount
of
inferencing,
or " b r i d g i n g "
(of.
Haviland
& Clark, 1974; S a n f o r d
&
Garrod,
1981).
Complement
clauses
require
no
such
m a t c h i n g p r o c e s s , and are
therefore
less
complex.
The i n f e r e n c i n g p r o c e s s can o n l y
W h a t seems to be i m p o r t a n t , then,
is
not
so
much
the
structure
of
a
construction,
but
rather
the
presuppositions
which
are
i m p l i c a t e d by
its use.
If
these
presuppositions
have
I
that
F u r t h e r m o r e , e v e n w h e n the r e l a t i v e c l a u s e
analysis
is
initially
chosen,
these
examples take longer
to
process
(via
a
reading
time
measure)
than
when
the
complement
analysis
is
chosen
(e.g.
Wanner,
Kaplan,
& Shiner, 1974; A l t m a n n ,
forthcoming),
In
other
words,
the
relative
clause
a n a l y s i s is not just the
least preferred,
but
is
also
the
more
"complex".
The
general
explanation
is
that,
in
the
example
above,
the
noun
phrase
"the
girl"
is
preferentially
t r e a t e d as a s i m p l e NP,
and
not
as
the
first
NP
constituent
in
a
c o m p l e x NP.
W a n n e r et al.
(1974)
and
Wanner
(1980)
m o d e l t h e s e e f f e c t s u s i n g an ATN, and s h o w
that
they
can
be
made
to
arise
from
peculiarities
of
the o r d e r in w h i c h arcs
leave certain
states.
Frazier
&
Fodor
(1978) cite this o b s e r v a t i o n in s u p p o r t of
Minimal Attachment,
whilst
Ford
et
al.
(1982)
would
predict
this e f f e c t on the
basis
of
their
theory
of
lexical
and
s y n t a c t i c p r e f e r e n c e s , in w h i c h the s i m p l e
NP e x p a n s i o n is o r d e r e d b e f o r e the c o m p l e x
NP e x p a n s i o n .
The f u n c t i o n of a PP w h e n a t t a c h e d to
an
NP,
in
t h e s e examples, is to p r o v i d e
a d d i t i o n a l and n e c e s s a r y i n f o r m a t i o n
with
which
to
i d e n t i f y a p a r t i c u l a r o b j e c t in
the d i s c o u r s e model.
As such it
must
be
providing
G I V E N i n f o r m a t i o n (cf.
Clark &
Haviland, 1977).
The
function
of
a
PP
w h e n a t t a c h e d to a verb, is to p r o v i d e N E W
i n f o r m a t i o n a b o u t the
action
denoted
by
the
verb:
the b u r g l a r d i d n ' t s i m p l y b l o w
open the safe, he blew it
open
with
the
dynamite.
This in turn, p r e s u p p o s e s t h a t
the a c t i o n
denoted
by
the
verb
("blow
open")
is
GIVEN.
In the s o - c a l l e d "VPinducing" context, this was not the
case:
the
co--text (that is, the p r e c e d i n g text)
was not f e l i c i t o u s w i t h the V P ~ a t t a c h m e n t .
A
second
e x p e r i m e n t was run in w h i c h the
" b l o w i n g open" was k n o w n a b o u t by s u b j e c t s
in
advance
of
the t a r g e t s e n t e n c e (i.e.
was Given), and this time, s t r o n g
effects
of
context
were
found
on b o t h k in d s of
t a r g e t (113 m s e c for
NP-attached
targets
across
the two c o n d i t i o n s of context, and
358 m s e c
for
the
VP-attached
targets).
Once
again,
the
non-minimally
attached
t a r g e t s were s i g n i f i c a n t l y f a s t e r t h a n the
minimally attached targets
(486
msec
in
the
NP-inducing
condition,
and 245 msec
overall).
All r e p o r t e d d i f f e r e n c e s
f i c a n t at least at p<O.05.
girl
signi-
125
be c o n t r o l l e d for if the
materials
under
s t u d y are p r e c e d e d by f e l i c i t o u s c o - t e x t s .
To a s s e s s the c o n t r i b u t i o n of
inferencing
to
p r o c e s s i n g time, an e x p e r i m e n t was run
(Altmann, f o r t h c o m i n g ) u s i n g the f o l l o w i n g
examples
(which are s i m i l a r to t h o s e used
by Crain, 1980).
In
the
absence
of
any
preceding
discourse,
there
can
e x i s t no d i s c o u r s e
model
within
which
to
integrate
the
information
c o n t a i n e d w i t h i n the i s o l a t e d
sentence.
In s u c h cases, n o t h i n g
can
be
succesfully
interpreted
as
Given
information.
It f o l l o w s t h a t all i n c o m i n g
m a t e r i a l m u s t be t r e a t e d as if it p r o v i d e s
New information.
If the i n c o m i n g m a t e r i a l
is
ambiguous,
between
a
reading
which
promises
New
information
(e.g.
a
complement
c l a u s e ) and one w h i c h p r o m i s e s
Given
information
(e.g.
a
relative
clause),
then
in
the
null
c o n t e x t the
f o r m e r i n t e r p r e t a t i o n m u s t be chosen.
In
general,
if
there
is a c h o i c e b e t w e e n a
complex
NP
analysis,
which
implicates
additional
Given
i n f o r m a t i o n by w h i c h to
identify
the
intended
referent,
and
a
simple
NP
analysis,
then
in
the
null
context2the simple
NP
analysis
must
be
chosen.
"INFERENCING"
A policeman
was
questioning
two
women.
He
was s u s p i c i o u s o f one
of them but not of the other.
"MINIMAL INFERENCING"
A policeman
was
questioning
two
women.
He
had
his d Q u b t $ a b o u t
one of t h e m but not a b o u t the other.
RELATIVE CLAUSE TARGET
The p o l i c e m a n told the w o m a n
that
h_ee had
his
d o u b t s a b o u t to tell
the truth.
S t r u c t u r a l l y b a s e d t h e o r i e s of
local
s y n t a c t i c a m b i g u i t y r e s o l u t i o n can a c c o u n t
for the
null
context
data,
but
cannot
a c c o u n t for the d a t a c o n c e r n i n g c o n t e x t u a l
effects
on
ambiguity
resolution.
The
present
account
can
be
a p p l i e d to b o t h
sets of data.
COMPLEMENT CLAUSE TARGET
The p o l i c e m a n t o l d the w o m a n
that
he had his d o u b t s a b o u t her c l e v e r
alibi.
(The u n d e r l i n i n g was not
present
in
the
experimental
items.)
The
amount
of
inferencing
required
to
process
the
relative
target
was
manipulated
by
c h a n g i n g the ( u n d e r l i n e d ) w o r d i n g
in
the
p r e c e d i n g c o - t e x t from "was s u s p i c i o u s of"
("inferencing")
to "had his d o u b t s
about"
("minimal
inferencing").
Apart
from
f i n d i n g s t r o n g e f f e c t s of c o n t e x t ( t h e r e b y
replicating
C r a i n ' s e x p e r i m e n t but w i t h a
d i f f e r e n t m e t h o d o l o g y - r e a d i n g time),
it
was
found
that
there
was
n__q a b s o l u t e
difference between complement targets
and
relative
targets
once
context
@nd
i n f e r e n c i n g w e r e c o n t r o l l e d for
(only
31
msec
in
the
"minimal
inferencing"
condition
vs.
385
msec
in
the
"inferencing" condition).
Many
studies
have,
in
the
past,
viewed
s y n t a x as b e i n g c o n c e r n e d w i t h the
relations
which
hold
within
single
sentences.
S u c h a v i e w is s h o r t - s i g h t e d ,
however.
W i t h i n a d i a l o g u e , the s p e a k e r ' s
intention
is
to
e v o k e a s p e c i f i c set of
2
A l t h o u g h this
explains
the
preference,
in the null c o n t e x t , for c o m p l e m e n t
c l a u s e s o v e r r e l a t i v e c l a u s e s , it d o e s n o t
explain
the i n c r e a s e d c o m p l e x i t y of r e l a tive clauses.
This is e x p l a i n e d
as
follows:
the
relative
clause intepretation
violates more presuppositions
(concerning
the s t a t e of the h e a r e r ' s d i s c o u r s e m o d e l )
t h a n d o e s the c o m p l e m e n t c l a u s e
interpretation
(see
Crain
& Steedman,
1982, and
Altmann & Steedman, forthcoming,
for
discussion).
The r e p o r t e d e x p e r i m e n t s d e m o n s t r a t e d t h a t s u c h v i o l a t i o n s lead
to
increased
reading
times.
The
relative
clause interpretation induces longer reading
t i m e s b e c a u s e of the i n c r e a s e d n u m b e r
of t h e s e v i o l a t i o n s .
It w o u l d appear, then, that
much
of
the
evidence
cited
in
the
literature
concerning
the
resolution
of
local
syntactic
ambiguity
has
been
misinterpreted because these studies
have
~ailed
to
consider
the
referential
f u n c t i o n of the c o n s t r u c t i o n s in q u e s t i o n .
An
account
of
syntactic
ambiguity
resolution
has
been
a l l u d e d to w h i c h is
ba s e d on a n o t i o n of
referential
context
and
d i s c o u r s e models.
More specifically,
it is an
account
based
on
the
act
of
establishing
what
is,
and
w h a t is not,
a l r e a d y k n o w n to the hearer.
But a l t h o u g h
it
can e x p l a i n p a r s i n g p r e f e r e n c e s w h e n a
target
sentence
is
embedded
in
a
discourse,
can
it
also
a c c o u n t for the
preferences
exhibited
in
isolated
sentences
(the
"null
context"
cf. the
o r i g i n a l "oil t y c o o n " e x a m p l e ) ?
3
It is a r g u e d in A l t m a n n ( f o r t h c o m i n g )
and A l t m a n n & S t e e d m a n ( f o r t h c o m i n g ) ,
that
an
account
based
on
the
distinction
between
what
is
and w h a t is not a l r e a d y
k n o w n to the h e a r e r / r e a d e r
(here
defined
as
the d i s t i n c t i o n b e t w e e n G i v e n and New)
will a l s o g e n e r a l i s e to the e x a m p l e s w h i c h
have,
on
" s t r u c t u r a l " a c c o u n t s , b e e n explained
by
Right
Association
(Kimball,
1973) and L a t e C l o s u r e (Frazier,
1979).
126
processes
in
the
hearer.
This
is
achieved, in part, by way of the s y n t a c t i c
c o n s t r u c t i o n s w h i c h the s p e a k e r c h o o s e s to
adopt.
The role of these p r o c e s s e s is to
establish
a
relationship
between
the
i n f o r m a t i o n c o n v e y e d by the u t t e r a n c e , and
the
information
already
known
to
the
hearer.
Such
processes
must
therefore
a d d r e s s i n f o r m a t i o n w h i c h is b o t h i n t e r n a l
and
external
to
the u t t e r a n c e .
Studies
which
purport
either
to
investigate
syntactic
processing
empirically,
or to
model
it
computationally,
should
not
ignore
the
role
or
the r e q u i r e m e n t s of
these processes.
To do
so
is
to
study
something
other
than
natural
language
processing.
Ford, M.,
Bresnan,
J.,
&
Kaplan,
R.M.
1982.
A
Competence-Based
Theory
of
Syntactic
Closure.
In
J.
Bresnan
(1982).
Frazier,
L.
1979.
O__nn C o m p r e h e n d i n g
Sentences:
Syntactic
Parsinq
Strategies.
Ph.D.
Dissertation,
U n i v e r s i t y of C o n n e c t i c u t .
Frazier, L.
&
Fodor,
J.D.
1978.
The
Sausage
Machine:
A new 2 - s t a g e P a r s i n g
Model.
In C o g n i t i o n , 6, 291-325.
H a v i l a n d , S. & Clark, H.H.
1974.
What's
New?
Acquiring
New
Information
as a
P r o c e s s in
Comprehension.
J.V.L.V.B.,
13, 512-21.
ACKNOWLEDGEMENTS
Kimball,
J.
1973.
Surface
Structure
2, 15-47.
The w o r k r e p o r t e d
here
was
carried
out
w h i l s t in the S c h o o l of E p i s t e m i c s on
an
S.E.R.C.
postgraduate
research
studentship.
My t h a n k s to the C e n t r e for
S p e e c h T e c h n o l o g y R e s e a r c h and
the
Alvey
Large
Scale
Demonstrator
Project,
for
providing
additional
financial
support,
and
to my s u p e r v i s o r s E l l e n Bard and M a r k
Steedman, for p r o v i d i n g
additional
moral
support.
K i m b a l l , J.
1975.
Predictive
Analysis
and O v e r - T h e - T o p Parsing.
In J. K i m b a l l
(ed) S y n t a x and Semantics, V o l u m e 4.
Rayner, K., Carlson,
M.,
&
Frazier,
L.
1983.
The
Interaction
of
S y n t a x and
Semantics
during
Sentence
Processing.
In J o u r n a l of V e r b a l L e a r n i n g and V e r b a l
B e h a v i o u r , 22, 358-374.
Sanford,
A.J.
Understanding
REFERENCES
Altmann,
G.T.M.
Forthcoming.
Reference
and
the
Resolution
of Local S y n t a c t i c
A m b i g u i t y : The E f f e c t of C o n t e x t
during
Human
Sentence Processing.
PhD Thesis.
University
of
Edinburgh.
To
be
Submitted.
1981.
Wiley.
Wanner, E., Kaplan, R., & Shiner, S. 1974.
Garden
Paths
in
Relative
Clauses.
U n p u b l i s h e d ms.
Harvard University.
1982.
The
Mental
o_~f G r a m m a t i c a l R e l a t i o n s .
Clark,
H.H.
&
Haviland,
S.E.
1977.
Comprehension
and
the
Given-New
Contract.
In
(ed.)
R.O.
Freedle
D i s c o u r s e P r o d u c t i o n and C o m p r e h e n s i o n .
Crain, S.
1980.
Contextual
on
Sentence Comprehension.
U n i v e r s i t y of C o n n e c t i c u t .
&
Garrod,
S.C.
Written Language.
Wanner, E. 1980.
The A T N and the
sausage
Machine:
which
one
is
baloney?
In
Cognition, 8, 209-225.
A]tmann, G.T.M.,
&
Steedman,
M.J.
The
(]arden
Path
in
Context: R e f e r e n c e and
the
Resolution
of
Local
Syntactic
Ambiguity.
In p r e p a r a t i o n .
Bresnan,
J.
Representation
MIT Press.
7
Principles
of
Parsing.
Cognition,
Constraints
PhD Thesis.
Crain, S. & Steedman, M.J.
1982.
On
not
being led up the g a r d e n path: the use of
c o n t e x t by the p s y c h o l o g i c a l parser.
In
(eds.)
Dowty,
D.,
Kartunnen,
L.,
&
Zwicky,
A.
Natural
Language
Parsing:
psychological,
computational,
and
theoretical perspectives.
In press.
127