Pragmatic sensitivity in NL interfaces and the structure of conversation

P r a g m a t i c s e n s i t i v i t y in NL i n t e r f a c e s
and the structure of c o n v e r s a t i o n
Tom W a c h t e l
Research Unit
i.
Scicon Ltd, L o n d o n
and
for I n f o r m a t i o n S c i e n c e & AI, H a m b u r g U n i v e r s i t y
Introduction
The work r e p o r t e d here i s being conducted as
part
of
the
L O K I p r o j e c t (ESPRIT P r o j e c t
I 0 7 , "A l e g i o o r i e n t e d approach t o knowledge
and
data
bases
supporting
natural
user
interaction").
The goal o f t h e NL p a r t
of
the
project
is
to
build
a pragmatically
s e n s i t i v e n a t u r a l language
interface
to
a
knowledge
base.
By
"pragmatically
s e n s i t i v e " , we mean t h a t t h e
system should
not
only
produce
w e l l - f o r m e d c o h e r e n t and
c o h e s i v e language (a minimum r e q u i r e m e n t of
any NL system designed t o handle d i s c o u r s e ) ,
but
should
also
be s e n s i t i v e
to
those
aspects
o~ user
b e h a v i o u r t h a t humans are
s e n s i t i v e %o over and above s i m p l y p r o v i d i n g
a good r e s p o n s e , i n c l u d i n g producing o u t p u t
t h a t i s a p p r o p r i a t e l y decorated
with
those
minor
and
semantically
inconsequential
elements
of
language
that
make
the
difference
between n a t u r a l
language
and
n a t u r a l n a t u r a l language.
This
paper
concentrates
on
the
representation
of
the
structure
of
c o n v e r s a t i o n i n our system.
We w i l l
first
outline
~he
r e p r e s e n t a t i o n we use
for
d i a l o g u e moves, and then o u t l i n e t h e
nature
of
the
definition
of w e l l - f o r m e d d i a l o g u e
t h a t we a r e o p e r a t i n g w i t h . F i n a l l y , we w i l l
note
a few e x t e n s i o n s t o t h e r e p r e s e n t a t i o n
mechanism.
The ACT s l o t s p e c i f i e s what a c t i s performed
by
the
utterance.
This
will
be e i t h e r a
speech a c t , o r t h e v a l u e ACTION, s i n c e
not
a l l moves need be speech a c t s .
The range of
speech a c t s t h a t t h e
system w i l l
have
to
r e o o g n i s e or' produce i s c l e a r l y s m a l l e r than
t h a t which occurs i n
c o n v e r s a t i o n s between
humans.
Furthermore,
certain
speech a c t s
w i l l be of
primary
importance given
the
domain of a p p l i c a t i o n o f t h e system, namely,
as a f r o n t end t o an e x p e r t system.
We have
therefore
produced
an i n i t i a l
h i e r a r c h y of
potentially
relevant
speech
acts
(Waehtel 1985a),
where
the
major
classification is into requests,
assertions
and commentaries. Some o f t h e s e are r e f e r r e d
t o below.
M a n y o f t h e speech a c t s we use go
one l e v e l below what i s t r a d i t i o n a l l y
viewed
as a speech a c t ( i . e . i n t h e sense of A u s t i n
(~952),
Searle
(1969),
etc.)
and
may be
compared
with
distinctions
that
McKeown
(1985:
9ff.),
f o r example, d i s c u s s e s under
the
c a t e g o r y of
"rhetorical
predicates",
though
they
are by no means t h e s a m e . The
only
speech
acts
discussed
below
a~e
r e f e r r e d t o by t h e f o l l o w i n g a b b r e v i a t i o n s :
REQACT
REQAFF
SUPPAFF
REQCONST
SUPPCONST
REQCONF
SUPPCONF
2.
The HOVE f r a m e
We a r e
assuming a s e v e n - s l o t f r a m e
for
the
r e p r e s e n t a t i o n of
moves w i t h i n a d i a l o g u e ,
w i t h t h e f o l l o w i n g s l o t s : HOVE, USER, AGENT,
TURN, ACT~ B A S E , CONT. Every
move i n a
c o n v e r s a t i o n i s r e p r e s e n t e d by such a frame.
The
HOVE s l o t
uniquely
idenitifies
a
p a r t i c u l a r move i n t h e
c o n v e r s a t i o n by
an
arbitrary integer.
The USER s l o t i d e n t i f i e s
t h e o u c r e n t user of t h e
system.
The AGENT
slot
s p e c i f i e s whether i t i ~ a user move o~'
a system move. The TURN s l o t has
the
value
OPEN n o r CLOSE n, where n i s a number t h a t
r e f e r s t o a p a r t i c u l a r exchange.
Typically,
a move w i t h a v a l u e OPEN n f o r t h e TURN s l o t
might be a r e q u e s t , and one w i t h a CLOSE n
v a l u e t h e c~orresponding r e s p o n s e .
request-for-action
request-for-affipmation
supply-of-affirmation
request-for-constant
supply-of-constant
request-for-confirmation
supply-of-confirmation
The BASE s l o t s p e c i f i e s t h e
current
topic,
i n t h e v e r y r e s t r i c t e d sense cf a p o i n t e r t o
the
node
in
the
~emantic
network
that
corresponds t o
t h e o b j e c t what t h e c u r r e n t
exchange i s a b o u t . This s i m p l i s t i c
v i e w of
topic
i s adopted here as a f i r s t
step o n l y ,
and
serves
to
i~lustrate
the
points
discussed below.
The
CONT s l o t
specifies
the
semantic
r e p r e s e n t a t i o n of
the
utterance,
and
we
envisage using t h e same r e p r e s e n t a t i o n f o r
semantics and
for
actions,
so
that
all
possible
ACT t y p e s
can
be
represented
uniformly
i n t h e CONT s l o t .
In p a r t i c u l a r ,
we w i l l d e f i n e an exchange as
a pair
of
u t t e r a n c e s w i t h t h e same v a l u e f o r t h e CONT
s l o t , f o r t h e t i m e being.
This i s of course
too
strict.
Other
functions
specifying
35
" l o c a l coherence"
in
the
sense
of
Hobbs
(1982:
227)
are
also
r e l e v a n t here.
The
particular
illooutionary
force
of
an
u t t e r a n c e w i l l be a f u n c t i o n of t h e v a l u e of
t h e ACT s l o t and t h e CONT s l o t . Subdialogues
that
are
not
in
the
mainstream of
the
conversation
will
be
identified
by
particular
relationship
of
values f o r the
TURN s l o t between a d j a c e n t
moves,
enhanced
by
the
values
of
the
ACT s l o t s f o r t h e
t h a t any such r e s t r i c t i o n
exists.
We assume t h a t t h e f o l l o w i n g f o u r c a t e g o r i e s
are s u f f i c i e n t f o r t h e r e p r e s e n t a t i o n of t h e
s t r u c t u r e of c o n v e r s a t i o n . The symbols
used
serve
as
mnemonics
for
t h e i r approximate
c o u n t e r p a r t s i n E n g l i s h , but t h e y should n o t
be
strictly
equated
with
them:
CONV
(conversation),
DIAL
(dialogue),
EXCH
(exchange) and MOVE (as discussed a b o v e ) .
moves.
Some examples of t h e use of
this
frame
to
represent
sequences
of
utterances
in
conversations
can
be
found
in
Waohtel
(1985b,
1985o),
including
its
use
to
i d e n t i f y s h i f t s of t o p i c ,
subdialogues
and
relevance,
as
well
as
the
contextual
disambiguation
of s p e e c h acts, w h i c h is
the
m a i n t o p i c of t h e s e w o r k i n g p a p e r s .
3.
The
structure
of
conversation
We assume t h a t i t i s p o s s i b l e t o d e f i n e
the
structure
of
a
possible
c o n v e r s a t i o n by
r u l e . A c t u a l human-human c o n v e r s a t i o n s may
defy
such
a n a l y s i s , as i l l u s t r a t e d
by, f o r
e x a m p l e , the
work
of
Sachs,
Schegloff
&
J e f f e r s o n (1974).
However,
the
possible
ways
in
which
the
c o n v e r s a t i o n s we a r e
dealing
w i t h may go are s e v e r e l y l i m i t e d by
t h r e e f a c t o r s : (a) t h i s i s an
interface
to
an e x p e r t system ( o r some s i m i l a r l y s p e c i f i c
s o f t w a r e ) , which d e l i m i t s t h e p o s s i b l e range
of
topics;
(b)
one of t h e p a r t i c i p a n t s i n
t h e d i a l o g u e i s a machine, which means t h a t
it
w i l l n o t suddenly want t o i n d i c a t e t h a t ,
for e x a m p l e , A l b e r t ' s n i e c e is a f r i e n d f r o m
school,
but t h i s f a c t has no b e a r i n g on the
s u p p o r t i v e t h i n g s being s a i d a b o u t her;
and
(o)
the
other
participant
knows t h a t his
i n t e r l o c u t o r i s a machine, and
will
behave
accordingly.
Therefore,
what
we
n e e d to
model is not a t y p i c a l l y n a t u r a l o p e n
human
conversation,
but
a restricted
type
of
c o n v e r s a t i o n t h a t a l s o occurs between humans
in
c e r t a i n w e l l - c i r c u m s c r i b e d c o n t e x t s . For
example, a c o n v e r s a t i o n between a would-be
passenger and
a t i c k e t c l e r k at a r a i l w a y
s t a t i o n i s c l o s e r t o what we need t o
model,
and i n
such oases i t i s p o s s i b l e t o d e f i n e
w h a t is or is not a w e l l - f o r m e d c o n v e r s a t i o n
by r u l e s of an a b s t r a c t n a t u r e t h a t may w e l l
be i n a d e q u a t e f o r o t h e r n a t u r a l l y
occurring
converations.
We t h e r e f o r e propose t h r e e r u l e s t h a t d e f i n e
the
notion
of
well-formed conversation in
the present c o n t e x t ,
making
the
following
assumptions. The s t r u c t u r e of a c o n v e r s a t i o n
can be r e p r e s e n t e d as a t r e e s t r u c t u r e .
The
w e l l f o r m e d n e s s of such t r e e s can be d e f i n e d
by r e w r i t e
rules.
The maximal
number
of
levels
of
embedding
in
such t r e e s i s s i x
(see b e l o w ) . I n p a r t i c u l a r , s u b d i a l o g u e s can
be embedded w i t h i n d i a l o g u e s , but t h e r e can
be no embedding
within
subdialogues.
The
last
r e s t r i c t i o n c o n f l i c t s w i t h what people
do.
It
is
one o f
the
restrictions
we
c o n s i d e r n e c e s s a r y , and which can be handled
i n such a way t h a t t h e user w i l l n o t
notice
36
To f o r m u l a t e i n i n f o r m a l terms
the
general
style
and
atmosphere of t h e r u l e s t h a t we
w i l l propose more f o r m a l l y below, l e t us say
that
a
CONV may c o n s i s t
of
one o r more
DIALs, a DIAL may c o n s i s t
of
one
or
more
EXCHs, and
an EXCH c o n s i s t s of two MOVEs,
w i t h each of
these
MOVEs f o l l o w e d
by
an
o p t i o n a l DIAL.
A major p o i n t about c o n v e r s a t i o n s t h a t
must
be handled i n a grammar of t h i s t y p e i s t h e
fact
that
although
MOVEs a r e
the
only
terminal
nodes, and are t h e r e f o r e t h e nodes
t h a t correspond t o t h e u t t e r a n c e s t h a t
are
actually
produced,
with
all
other
nodes
r e p r e s e n t i n g more a b s t r a c t e l e m e n t s , c e r t a i n
features
of
conversation
need
to
be
associated with these
abstract
nodes.
For
example, a l t h o u g h each MOVE i s s p e c i f i e d f o r
who t h e c u r r e n t user of t h e
system
is
and
each
MOVE a l s o
has a p a r t i c u l a r t o p i c , as
discussed
above,
these
notions
properly
belong
to
more
abstract
levels
of
c o n v e r s a t i o n a l s t r u c t u r e . Who t h e
user
is
can
be d e f i n e d
a t t h e CONV l e v e l ( i . e . we
d e f i n e a CONV as
a conversation with
one
user).
The t o p i c
of
an u t t e r a n c e can be
d e f i n e d a t t h e DIAL l e v e l ( i . e . a CONV can
consist
of one o r more d i a l o g u e s , each on a
single
topic).
Furthermore,
a DIAL
can
consist
of
one o r more EXCHs, and i t i s a t
this
point
that
the
content
of
the
utterances that
form
p a r t of t h a t EXCH i s
defined.
L e t us now be more p r e c i s e . We a s s u m e t h a t
some of
the
slots
mentioned above i n t h e
MOVE frame a r e r e p r e s e n t e d as
features
on
the
nodes
in
the
trees
r e p r e s e n t i n g the
s t r u c t u r e of t h e c o n v e r s a t i o n o f
which
the
moves d e s c r i b e d by t h e MOVE frames a r e p a r t .
This a s s o c i a t i o n of
features with
nodes,
plus
the
assumption
that
all
features
trickle
down,
with
a
few
exceptions
discussed
below,
p r o v i d e s f o r t r e e s of t h e
g e n e r a l form shown i n
Table i .
The
lower
case
letters
are constants.
Note t h a t t h e
values
of
the
BASE feature
on
the
subdialogue
nodes
have n o t been s p e c i f i e d .
We r e t u r n
to
this
point
below.
Table I
r e p r e s e n t s a g o a l : t h e s o r t of s t r u c t u r e s we
want t h e r u l e s t o
produce.
The
following
three
rules
generate t r e e s
of t h i s t y p e .
Kleene Plus n o t a t i o n i s used.
The n o t a t i o n
should
be
interpreted
as
f o l l o w s . Roman l e t t e r s as f e a t u r e v a l u e s a r e
constants.
Greek
letters
are
variables
ranging
over
possible
f e a t u r e v a l u e s , and
are t o be i n t e r p r e t e d c o n s i s t e n t l y w i t h i n
a
rule~
but
not
n e c e s s a r i l y between r u l e s .
They are used t o
ensure
that
the
correct
Table
I
CONV
[
DIAL
B SER
DIAL
UBSER :
EXCH
EXCIt
A S E :b
ON~ :
SER:iN
ASH
:
ONLY:
AsIE:~
r
DIAL
I
D IAL
ASlE:_J
I
EXCIt
SE~
ASE
MOVE
MOVE
F:il
A S E :b
ONT :
URN :
A~E :
1
EXCH
it]
EXCH
~kCO~r :
s ~ . . ~]
[
ASE : |
ASE : ]
ONe: ,~1 ]
MOVE
MOVE
a s E : l I nAsE:
O~T:S I ICOm~:
InasE:
ICO~'r:
MOVE
MOVE
nS
MOVE
IBASE: l I~ A S E : I IBASE
ICOm'-m I ICOm~:q ICOm'
F......
MOVE
HOVE
ASE:Pl
[BASE :
[CONT:
[~URN :
MOVE
pASE:
o~T.~I Ices,:
[TURN:o_] [jUURN:
2
CONV
-~
DJ:AI.,d+
DI~L d
ABE : b [
pass:
]
~D:-
.--)
BASl!: :?[
~u~D :
_
EXCHjL
suBD: r I
.
~om:o
d
3
EXC}~
-->
MOVE
BASE:~ /
BASE : ~
ICON'.e: 6
co~T:~ /
<(DIALd)>
n[
|
|
feature
values
triokle
down i n t h e r i g h t
oases. Node s u b s o r i p t s
distinguish
between
different
i n s t a n o e ~ of t h e same node. Note
that
Greek
letters
are
also
used
a~
variables
for
node
subsoripts (rule (3)).
Round b r a o k e t ~ i n d i o a ~ e o p t i o n a l i t y
in
the
usual
way.
Angle
braokets
( r u l e ( 3 ) ) are
in the
same
way
as u s e d
by e.g.
( ( D I A L d d I)
[~URN : CLOSI?
[~ul~N : O e E ~ J
used
MOVE n
Labor
(1972: oh. 8)
in
the
study
of
~ooiolinguistio variation.
They i n d i o a t e
a
relationship
between elements
on the l e f t
and on t h e r i g h t o f t h e r u l e . The expansion
on t h e ~ i g h t i s o n l y p o s s i b l e i f t h e f e a t u r e
on t h e
l~!ft
of
the
rule
has
the
value
speoified,
if
both
are
enalosed i n angle
b r a c k e t s . I n t h e p r e s e n t ease, t h e y a r e used
t o p r e v e n t t h e expansion ~f a s u b d i a l e g u e as
a f u r t h e r s u b d i a l o g u e . The f e a t u r e
ESUBD=-3
i s i n t r o d u o e d by t h e r u l e t h a t expands CONV.
The r u l e e;¢panding DIAL o o p i e s t h i s
feature
with
this
value.
The r u l e expanding EXCH
allows
an expansion of
EXCH tQ
include
(optionally)
one o r
two DIALs, but i f t h e
DIALs appear, then t h e y
oarry
the
feature
[SUBD:+].
The expansion of such a DIAL by
rule
(2)
oopies
this
f e a t u r e with
this
value,
as b e f o r e , when t h e DIAL i s expanded
t o one o r more EXCHs. However,
sinoe
the
EXCHs
so
generat,ed
oarry
the
feature
[GUBD:+3, t h e r u l e t h a t
expands EXCH w i l l
not
a l l o w t h e p o s s i b i l i t y of f u r t h e r DIALs,
beoause any suoh
e~pansion
is
oonditional
upon
the
EXCH having t h e f e a t u r e [SUBD:-3,
as s p e o i f i e d in r u l e ( 3 ) .
The v a l u e o i t h e f e ~ t u r e TURN i s e i t h e r OPEN
or
CLOSE p l u s a o o n s t a n t t h a t r e f e r s t o t h e
r e l e v a n t EXCH. Note t h e
use
of
the
6reek
37
Table 2
CONV I
~SER: Max~
f
DIAL I
]
ASE : IJCA_~
UBD :
-
f
EXCH I
FSE~: Max q
IBASE : IJCAI I
ISUBD :
CmONT: cI 3
l
J
MOVE I
'
]
SER: Max
|
SE: IJCAI |
NT: CI
|
RN: OPEN I
ENT:USER
|
T: REQCONST~
I
~SER: Max
IBASE: EXCH
~UBD: +
I
'
MOVE 2
OVE: 2
I~
J
EXCH2
~SER: Max
f
MOVE 6
IOVE : 6
q
SER: Max
|
ASE: IJCAI
|
ONT: CI
|
URN: CLOSE I |
GENT:SYSTEM |
CT : SUPPCONST~
I
q
|SUBD:
LC_0NT:
C2 _ J
/___
]
__ __ MOVE 3
u s s a : Max
SER: Max
ASE: EXCH I BASE: EXCH 1
CONT: C2
ONT: C2
URN: OPEN 2 TURN: CSOSE 2
GENT:SYSTEM /AGENT:USER
T~QCON~
EXCH4
FsER: Max -1
r-~
FSER, Mox ]
|suED: +
"~
Isu sD: +
IBASE: E x c e l I
~ONT [ c 3
/
__1
[s3)
__ __ MOVE 4
I
I
I
/
/
USER: Max
BASE: EXCU 1
CONT: C3
TURN: OPEN 3
IAGENT:SYSTEM
L e t us now p r o v i d e a c o n c r e t e example o f t h e
structure
that
these
rules
assign
to
a
d i a l o g u e such as ( 4 ) .
lEASE: EXCH
LC0NT i C4
......
I
variable.
The same c o n s t a n t i s used as t h e
v a l u e of t h e f e a t u r e BASE i n
subdialogues.
What t h i s
amounts t o i s a s t i p u l a t i o n
that
t h e t o p i c o f a s u b d i a l o g u e i s t h e EXCH t h a t
i t i s p a r t o f , which seems t o be a b o u t r i g h t
intuitively.
This
is
what
makes
them
metalinguistio
in
character.
Furthermore,
note
that
this
is
a
case
where
a
feature/value
pair
does
not t r i c k l e
down.
This i s t a n t a m o u n t t o s t i p u l a t i n g
that
DIAL
is
a
BASE-bounding
node:
it
creates
"islands" with
BASEs t h a t
do
not
extend
upwards
to
the
main d i a l o g u e , b u t w i t h o u t
o v e r w r i t i n g t h e BASE of
the
current
main
dialogue.
Again,
this
seems intuitively
correct.
38
I
EXCH3
DIAL3
~SER: Max ]
IBASE: EXCH
~SUBD: +
I
I
I
_ __ MOVE 5
__ M O V E
lUSER: Max
lEASE: EXCH 1
/CONT: C3
/TURN: CLOSE 3
/AGENT: USER
4
Ui
Si
U2
B2
U3
$3
U4
$4
I
7
__ __ MOVE 8
//USER: Max
//BASE: EXCH ~
//CONT: C4
'
/ /TURN: OPEN 4
/ F GENT:usER
]/US~R:
//EASE:
/
Max
|
EXCH 1 |
C4
/
/ /TURN:
/c°NT: CLOSE 4/
/ / AGENT=SYSTEM /
When i s t h e n e x t J I C A I meeting?
I presume you mean " I J C A I "
Yes
Do you mean t h e n e x t c o n f e r e n c e
o r t h e n e x t c o n v e n e r s ' meeting?
Conference
12 August
19857
Yes
The
structure
ie
given
as
Table 2.
The
values
for
the
feature
CONT a r e g i v e n as
constants
rather
than
as
full
semantic
representations,
and
the
constants rather
t h a n as f u l l
semantic
representations,
and
the
constant
IJCAI
is
used f o r t h e BASE,
which i s t h e a c t u a l c o n f e r e n c e due
to
take
p l a c e on 12 August 1985. This v a l u e has been
g i v e n i n a l l r e l e v a n t oases,
thus
glossing
over
the
identified
triggeeed
however,
fact
that
the BASE could not be
immediately,
which
is
what
the
subdialogues.
We
add,
neetain
features
that
wece
discussed
earliee,
s u c h as MOVE, AGENT and
ACT, t o o l a e i f y h o w t h e f i n a l
focm
of
the
frame
eepresenting
each
MOVE
is
deeived.
T h i s is a o o n v e e s a t i o n
in
which
there
is
only
oFie
main
dialogue
arid o n l y o n e m a i d
exchange within that dialogue. I
hope
that
it
is
clear
how these additional elements
w o u l d be i n c o e p o r a t e d
into
the
structure,
and
how
the
appropriate
v a l u e s foc B A S E ,
CON?
and
SUBD
would
be
iaaintained
or
changed.
It is i n t e e e s t i n g
to note that the nodes
iri
conveesation
trees
of
this
sect
have
a
conceptual
validityl
in t h a t d J . ~ f e c e e t
node
types
(Jo~'eespond
to diffeeent
aspects
of a
~enve~°satierio
Thus a CONY node
eoc~'esponds
to
"this
(Iollve~'sation
with
this usec",
a
DIAL
[:SUBD: - 3
node
eoccespends
to
"this
topic"~
a n E X C H [ S U B D : --3 n o d e c o r r e s p o n d s
to
"this
point"
~
a
DIAl
[BUBD ~-~']
node
eoreesponds
to
"a
point
that
needed
olai'ifieation",
arld a n
EXCH
ESUBD:÷]
iiede
ooe~;'espends to "what was unuleac",
E a c h HOVE
ill]de 12eprl~sel]t5
~311 u t t e e ~ J i i l ] e ,
(11> [~oLll,~e.
The
seA;
of
HOVE
nodes
dominated
by EXC'.H
[UI.IBO:"LI e o r e e s p e n d s
t o " w h a t was s a i d ;
t',he
gelrf~l,a!
l iln-~ of t h e O C i l V e c s a t i o n " ,
alld t i l e
s e t o t HOVE n o d e s d o l , i i r i a t e d
by EXCFI ['31l[)1):+3
corresponds
t(i "the
!iubdialogees".
t.J. k e w i s o ~
£H~'I;~1 [if
t i t tic>r
il o d (,~~
ClOl2ices p o n t l
i;o
o51iel?
hclladec ~lollients
el 6~ O C l i V e ~ ' s a t i o r l =
The set
07 C()NV nod(.~>s o o r ' ~ ' e i l d s t t l " a l l
the di]ltJ~-erlt
~lerivi!IcfJ{~];ierl~
)[ l i c i t
ill
thi~,~ s e % s i o l l w i t h
d.i.~t~rerit
tlsol,'s",
the set ill
1)i~%1. [ B U B D . ' - ' J
7iod{-#~5 (ltl!. e R ~ p o n d t ;
t o " t l i ~ ! to~] il:;f7 (i(.iv()l~ed ,i i
tli~,~ s e t of F;XCH F B U B D ' - ]
ned~:q~ ~ei:,i.,espsnd+~
tu
"the
lJoin~s
discussed",
a n d +Je mi. By
takin(t
J.n t n
account
oenf iUu~'ai, iolts
o~
I eatue~
~It
liodes ~
one
oaii
isolate,
th~:~'efo~e, s u c h e l e m e n t s
a~
"the
last
but
c n e topi¢~ d i s c u s s e d
by tl~o previous
u s e r " or
"the
fleet
point
i~i t h i s
ooeve~'saticm
that
needed
o l a r i l i{lation".
Let us now turn to
two
e)(tensions
o~
the
above
system~
hypebhetieal
moves
arid
anticipatocy
moves,
~'equired
by
certain
dialogue
phenomena.
4.
Hypothetical
U:
S:
U:
S:
On
formal
grounds,
the
oonversation
closed
a f t e r t h e f o u r t h . m o v e (U: Y e s ) ,
is
with
a l l OPENed moves having been
CLOSEd. What,
then,
t r i g g e r ' s t h e s y s t e m ' s p r i n t i n g of t h e
eequieed
text (MOVE
5) ~ a n d
what
happens
after
that,
Since
an
odd n u m b e c of m o v e s
~annot
constitute
a well-formed
conversation
aeoecding
to the grammar
peesented
above?
It
is o l e a e t h a t it is t h e m e a n i n g
of
MOVE
3
t h a t is t h e k e y . T o h a n d l e
this fo~mally,
we
pcopose
the
use
of
the
notion
of
"hypothetical
move"
and the cepresentation
speech
aot.~ n e t a s a t o m i c
elements
Hilt a s
struotuces.
Thus
we
wi I ]
ue p r e s e n t
a
supp Iy -o~-a~ f i~mation
in
response
to
o
c e q o e s t - l o c - - a f ~ i r m a t i o n as SUPPAFF(REgAFF)o
A hypothetiual
m o v e is a m o v e t h a t d o e s
not
a~tually
oeeur
ilr
the
oeilversation $
but
w h i c h t h e ~iystem o o s s t c u u t s
on lille b a s i s
o~
very
specific
clues,
a n d whic:h a l l o w it to
(~ontinue
the
eo~iw.~rsat i o n
appcopi~ia~ie i y °
They
ooz, c e s p e Y i d
in
some
way
to
a
~q~pcesentation
c~ G e i e e ' ~
(19/5)
llotien
o~
implioature.
Foe
,E,~xample~ a m o r e d e t a i l e d
analysis
o:~
(5)
reveals
that
(6)
omits
several
important
details.
T i l e f i r s t iilOVE! i s
a b t u a l ly alailigtAous
betwei:)n
a
~'equost-:focaffirmation
a n d a ceqoe!~t-foz'-acM~ioli~
What
we would
like) t h e systelii t o do i s te
supp).y
bbe
a f f il'iilatiori
te
the
r ' e q u e s t - ~ f o r ,-~
allil~'mati(3n
part,j
arid t o r e t l u ~ s t
alfiPiliation
c o n c t ~ c i l i f i g t h e eequest--lel.-:aetion
pact.
Th~
important
p o i n t is t h a t a p[<epeP a n a l y . q i B of
"Oo
y o u ~#ant to s e e t h e ! m ? " s h o u l d
r'epPesent
the
J[aet
tl'iat
this
is
response
to
the
potential
~'equest.- f or~aotion
int, ez'p['et at ion
of " C a n y o u p r i n t
the minutes
of
the
last
meeting?"0
The
upshot
of
tliis is t h a t a
m o r e pL~ecise
~ep~'eseritation
of
the
~i['st
l o u t m o v e s ef (5) is (7), cathez' t h a n (6),.
i: U, ()PEN i, R E Q A F F / R E Q A C T
2: S, C L O S E i~ ~ U P P A F F ( R E Q A F F )
3: B, OPEN 21 REQAFF(REQACT)
4: U, CLOSE 2~ SUPPAFF(RESAFF(~E@ACT)
innocuous
COilVepsatiens
may
cause
ti~oob i e on
Can you print the minutes
Of
the last meeting?
Y e s . Do y o u w a n t to s e e t h e m ?
Yes
(p~ints)
MOVE i i USER~ OPEN 1 , REQAFF
MOVE 2: SYS'FEM, CLOSE i , SUPPAFF
M O V E 3: SYSTEM, O P E N 2, R E Q A F F
MOVE 4: USER, CLOSE 2, SUPPAFF
MOVE
HOVE
MOVE
MOVE
moves
"rheee are appaeently
suoh
as
(5)
which
formal
grounds.
This
is
a
five-move
conversation
(the
system's
"Yes"
c o u n t s as a s e p a c a t e m o v e ) .
The g~ammac
would assign
the steuotuee
shown
informally
as (6) t o t h e f i r s t
four moves.
We n e w h a v e a
way
of
amcountillg
fo~
the
system' s
eext
more
(pminting),
and
fec
explaining
why a n y t h i n g
at
all
happens.
MOVE 4
i s a SUPPAFF'(REQAFF(REQACT)), i . e ° a
supply
of
affirmation
in
response
to
a
request
foc
affirmation
in
response
to a
request
foe aotion.
It
seems
quite
gloat
intuitively
that
t h i s complex s t r u c t u r e i s
e q u i v a l e n t "to a REQACT, and w e propose
that
this
t y p e of r e d u c t i o n
should
t a k e p l a c e by
rule.
39
However, t h i s r u l e must n o t
over-write
the
o r i g i n a l i n t e r p r e t a t i o n of the i l l o o u t i o n a c y
f o r c e o f t h e move, which
must
be r e t a i n e d
foc
the
dialogue
to
be w e l l - f o r m e d w i t h
r e s p e c t t o t h e grammar. We propose t h a t
the
e~feot
of t h i s t y p e of r u l e (an i m p l i o a t u r e
redundancy r u l e ) i s t ~ c r e a t e a h y p o t h e t i c a l
move
immediately
following
it
of
the
a p p r o p r i a t e t y p e . I t s e f f e c t is t o a l t e r the
s t r u c t u r e of t h e c o n v e r s a t i o n i n e x a c t l y t h e
same way as i f t h e user ( i n t h i s
case)
had
actually
uttered
something l i k e " I r e q u e s t
you t o p ~ i n t t h e minutes
now",
except
for
the
fact
that
it
i s noted t h a t t h i s i s a
h y p o t h e t i c a l move.
We now have a
formal
entity
t h a t can t r i g g e r t h e p r i n t i n o of t h e
required t e x t , since t h i s is
a CLOSure of
the
hypothetical
move. I f no p r i n t i n g t o o k
place,
then
the
d i a l o g u e would
be
illformed, s i n c e i t would c o n t a i n one OPEN t h a t
had n o t been CLOSEd. T h i s ,
the
system
is
behaving
as
if
the
use~
had
made a
p a r t i c u l a r move t h a t d i d n o t a c t u a l l y o c c u r .
(The
notion
'as
if'
is
central
to
V a i h i n g e r ' s (1935) t h e o r y of f i c t i o n s = I t i s
al~o c r u o i a l t o 6 r i o e a n i m p l i c a t u r e . )
The r e s u l t i s t h a t (4) i s now analysed as
a
s i x - m o v e d i a l o g u e , w i t h t h e s t r u c t u r e shown
a~ (8).
MOVE
MOVE
HOVE
MOVE
MOVE
HOVE
5.
i:
2:
3:
4:
5:
&:
U,
S,
S,
U,
U,
S,
OPEN I , REQAFFIREQACT
CLOSE I , SUPPAFF(REQAFF)
OPEN 2, REQAFF(RE~ACT)
CLOSE 2, SUPPAFF(REQAFF(REQACT))
OPEN 3, RE~ACT, h y p o t h e t i c a l
CLOSE 3, ACTION
A n t i c i p a t o r y moves
Another type
of
ir=ealis
move
is
an
anticipatory
move,
where on
t h e b a s i s of
s p e c i f i c c l u e s t h e system a n t i c i p a t e s
what
the
user's
next
move will
be.
The
d i f f e r e n c e between these
and h y p o t h e t i c a l
moves i s
that
no a c t i o n
i s t a k e n by t h e
system u n t i l t h e r e has been a r e a c t i o n
from
t h e user t h a t e i t h e r c o n f i r m s o r d i s o o n f i r m s
t h e c o r r e c t n e s s o f t h e move t h a t
has
been
anticipated.
The use o f such moves w i l l be
of
assistance
in
the
interpretation
of
c r y p t i c f o l l o w - . u p s , as i n ( 9 ) .
u:
S:
U:
S:
Can you p r o v i d e p r o g r e s s r e p o r t s on LOKI
subprojeots?
Yes. Do you want t o see them?
LOKA
(prints)
The u s e r ' s
second
utterance
must
be
interpreted
as
a r e q u e s t - f o r - a c t i o n , which
is
difficult
on
formal
grounds.
Without
going
i n t o t o o much d e t a i l , we propose t h a t
in
such
oases t h s
system
should
have
f o r m u l a t e d an a n t i c i p a t i o n
of a f o l l o w i n g
40
r e q u e s t ~ f o r - a c t i o n before
the
user'~
next
utterance.
] ' h i s c o u l d e i t h e r be an e ; ¢ p l i c i t
r e q u e s t = f o r - a c t i o n ("Please p r i n t
the
LOKA
progress
report"),
o r ~imply "Yes" ( i . e . a
SUPPAFF(REQAFF(REgACT))),
which
would
trigger
a system r e q u e s t f o r c l a r i f i c a t i o n ,
perhaps, o r a n y t h i n g e l s e a t
all
tha~
can
serve
to
identify
the
BASE
of
the
a n t i c i p a t e d r e q u e s t - f o r - a C t i o n . This i s
the
important
point
about
the
anticipatien~
Anything. at
all
that
can
fill
in
the
unspecified
~Iots
in
the
BASE of
the
anticipated request-for-action will
confirm
t h a t t h i s u t t e r a n c e i s i n t e n d e d as a REQACT.
For t h i s
reason~
the
bare
n a m e LOKA i ~
enough
t o g e t t h e r e p o r t p r i n t e d . Any o t h e r
sufficiently
i d e n t i f y i n g d e s c r i p t i o n of
the
relevant
s u b p r c j e o t would have achieved t h e
same, such
as
any
of
the
following
<as
appropriate):
the
one
based
in
Hamburg,
Hamburg, NL,
Max'~
projeot~
most
ceeent~
etc.
6.
Conclusions
The processes and f o r m a l i s m s o u t l i n e d
above
are
all
t e n t a t i v e i n n a t u r e , and represent
part
of
an approach
to
the
problem of
pragmatic s e n s i t i v i t y ,
r a t h e r than p u r p o r t e d
s o l u t i o n s t o t h e problem. We e n v i s a g e
then
as
being
part
of
a system t h a t
uses a
m u l t i - l e v e l parsing technique,
with
mutual
a s s i s t a n c e between d i f f e r e n t subcomponents
of t h e p a r s e r , so t h a t p r a g m a t i c i n f o r m a t i o n
can
i m m e d i a ti e l y
be
used t o a s s i s t par~ing
f o r s y n t a x , and so
on.
We a l s o
see
that
par~ing
will
involve
not
only
~entence
p a r s i n g , but a l s o c o n v e r s a t i o n p a r s i n g ,
in
that
the
appropriate
structure
of
a
c o n v e r s a t i o n must be b u i l t up a t each
step.
This
is
simply
one
further
part
of t h e
g e n e r a l p a r s i n g process~
but
one
that
we
envisage
as
being
of
assistance to other
p a r s e r suboomponents, as w e l l as
fulfilling
i t s p r i m a r y f u n c t i o n of making sure t h a t t h e
system i s something of a c o n v e r s a t i o n a l i s t ,
rather
than
just
being
a communicative
plodder.
REFERENCE~
Cole, P. & J. L, Morgan (eds.) 1975.
Syntax
& Semantios~
volume 3:
S p e e o h a o t s . New
Yock= Aoademio Press
Findle~
N. V.
(ed.)
1979.
Assooiative
networks:
representation
and
use
of
knowledge by oomputers". New York:
Aoademio
Press
Given, To ( # d . ) 1979.
SyntaH & Semantios~
volume 12: Oisoeuvse and s y n t a x . New York:
Aoademio Pcess
Gcioe, H. P. 1975 "Loglo and o o n v e r s a t i o n "
I n : Cole & Reagan ( e d s . ) , 1979,~41-58
Hobbs, 3. R. 1982. "TowacdB an understanding
el
ooherenoe in
d i s o o u v s e " . I n : Lehnert &
Ringle ( e d s . ) , 1982, 223-243
Labov~ N. I772o
Sooiolin~uistlo patterns.
Philadelphia:
U n i v e r s i t y of
Pennsylvania
Press
Lehne='t~ |~. & M. H.
Ringle
(ed=.)
1982.
S t r a t e g i e s f o F n a t u r a l language prooessing.
London= Erlbaum
MoKeown, I(. R. 1985.
"Discourse s t r a t e g i e s
for
g e n e r a t i n g n a t u r a l - l a n g u a g e t e x t " . In:
Artifioial
Zntelligenoe~27~ I - / , I
V a i h i n g e r , H. 1935. The philosophy of
'As
if'
-a
system of
the
theoretioal~
p r a o t i o a l and r e l i g i o u s f i o t i o n s of mankind.
London: Routledge & Kegan Paul
Waohtel~ T.
1985a.
LOKI Report N L I - I . I ~
"Disoourse s t c u o t u c e " .
U n i v e r s i t y of Hamburg
Waohtel~
T.
1985b.
"Guided
speeoh
aot
assignment
in
'oan
you'
utteranoes"
Unpublished L O K I working
paper
TW-II,
U n i v e r s i t y of Hamburg
Waohtel~ T. 1985o.
"Contextual faotocs
in
speeoh aot
assignment".
Unpublished LOKI
working paper TW-13, U n i v e r s i t y of Hamburg
41