If all this does not happen then the n e w edge supposedly to be added to the chart is not added: the situation is recognized as a failure.. Again, if all this does not happen, the n
Trang 1COPING WITH DYNAMIC SYNTACTIC STRATEGIES: AN E X P E R I M E N T A L ENVIRONMENT FOR AN
E X P E R I M E N T A L PARSER
Oliviero Stock I.P - Consiglio Nazionale delle Ricerche Via dei Monti Tiburtini 509
00157 R o m a , Italy
A B S T R A C T
A n e n v i r o n m e n t b u i l t a r o u n d W E D N E S D A Y 2, a c h a r t
b a s e d p a r s e r is i n t r o d u c e d The e n v i r o n m e n t is i n
p a r t i c u l a r oriented t o w a r d s exploring d y n a m i c aspects
of parsing It includses a n u m b e r of specialized tools
t h a t consent a n easy, graphics-based i n t e r a c t i o n w i t h
the parser It is shown in p a r t i c u l a r how a c o m b i n a t i o n
of the characteristics of t h e p a r s e r (based on the lexicon
a n d on d y n a m i c unification) a n d of t h e e n v i r o n m e n t
allow a nonspecialized user to explore heuristics t h a t
m a y a l t e r the basica control of the system In t h i s way,
for instance, a p s y c h o l i n g u i s t m a y e x p l o r e i d e a s on
h u m a n parsing strategies, or a "language engineer" m a y
find useful heuristics for parsing within a particular
application
1 I n t r o d u c t i o n
C o m p u t e r - b a s e d e n v i r o n m e n t s for t h e l i n g u i s t a r e
conceived as s o p h i s t i c a t e d w o r k b e n c h e s , b u i l t on AI
w o r k s t a t i o n s a r o u n d a s p e c i f i c p a r s e r , w h e r e the
linguist can t r y out h i s / h e r ideas a b o u t a g r a m m a r for a
c e r t a i n n a t u r a l l a n g u a g e In doing so, he/she can t a k e
a d v a n t a g e of rich a n d easy-to-use g r a p h i c i n t e r f a c e s
t h a t "know" a b o u t linguistics Of course b e h i n d all t h i s
lies the idea t h a t cooperation w i t h linguists will provide
b e t t e r results in NLP To s u b s t a n t i a t e t h i s a s s u m p t i o n i t
m a y be recalled t h a t some of the most i n t e r e s t i n g r e c e n t
ideas on s y n t a x h a v e b e e n developed by m e a n s of j o i n t
c o n t r i b u t i o n s f r o m l i n g u i s t s a n d c o m p u t a t i o n a l
linguists L e x i c a l - F u n c t i o n a l G r a m m a r [ B r e s n a n &
K a p l a n 1982], G P S G [ G a z d a r 1 9 8 1 ] , F u n c t i o n a l
G r a m m a r [Kay 1979], DCG [ P e r e i r a & W a r r e n 1980],
TAG [Joshi & Levy 1982] a r e some of these ideas
I n s t a n c e s of the tools i n t r o d u c e d a b o v e a r e t h e L F G
e n v i r o n m e n t , which w a s p r o b a b l y t h e first of its kind, a n
e n v i r o n m e n t b u i l t b y Ron K a p l a n for L e x i c a l -
F u n c t i o n a l G r a m m a r s , D P A T R , b u i l t b y L a u r i
K a r t t u n e n a n d conceived as a n e n v i r o n m e n t t h a t would
s u i t l i n g u i s t s of a n u m b e r of d i f f e r e n t schools a l l
c o m m i t t e d to a view of p a r s i n g as a" process t h a t m a k e s use of a n unification algorithm
We h a v e developed a n e n v i r o n m e n t with a s o m e w h a t different purpose Besides a n u m b e r of tools for e n t e r i n g
d a t a in g r a p h i c m o d e a n d i n s p e c t i n g r e s u l t i n g structures, it provides a m e a n s for e x p e r i m e n t i n g w i t h
s t r a t e g i e s in the course of the p a r s i n g process We t h i n k
t h a t t h i s can be a v a l u a b l e tool for g a i n i n g insight in t h e cognitive aspects of l a n g u a g e processing as well as for
t a i l o r i n g the b e h a v i o u r of the processor w h e n used with
a p a r t i c u l a r (sub)language
In t h i s way a n a t t e m p t can be m a d e to a n s w e r b a s i c questions w h e n following a n o n d e t e r m i n i s t i c approach:
w h a t heuristics to apply w h e n facing a c e r t a i n choice point, w h a t to do when facing a failure point, i.e which
of the p e n d i n g processes to activate, t a k i n g account of information r e s u l t i n g from the failure?
Of course t h i s kind of e n v i r o n m e n t m a k e s sense only because the p a r s e r it works on h a s some c h a r a c t e r i s t i c s
t h a t m a k e it a psychologically i n t e r e s t i n g realization
2 M o t i v a t i o n o f the p a r s e r
We s h a l l classify psychologically m o t i v a t e d p a r s e r s in
three m a i n categories First: those t h a t embody a s t r o n g
claim on t h e specification of the g e n e r a l control s t r u c t u r e
of t h e h u m a n p a r s i n g m e c h a n i s m The a u t h o r s usually
consider t h e level of basic control of the s y s t e m as t h e level t h e y are s i m u l a t i n g a n d are not concerned w i t h more p a r t i c u l a r heuristics An instance of this class of
p a r s e r s is Marcus's p a r s e r [Marcus 1979], based on t h e claim t h a t , basically, parsing is a d e t e r m i n i s t i c process: only s e n t e n c e s t h a t we perceive as " s u r p r i s i n g " (the so
c a l l e d g a r d e n p a t h s ) a c t u a l l y i m p l y b a c k t r a c k i n g
Trang 2Connectionist parsers are also instances of this category
The second c a t e g o r y r e f e r s to g e n e r a l l i n g u i s t i c
performance notions such as the "Lexical Preference
Principle" and the " F i n a l A r g u m e n t Principle" [Fodor,
13resnan a n d K a p l a n 19821 It i n c l u d e s t h e o r i e s of
p r o c e s s i n g like t h e one e x p r e s s e d by W a n n e r a n d
M a r a t s o s for A T N s in t h e m i d S e v e n t i e s In t h i s
category the a r g u m e n t s are a t the level of g e n e r a l
s t r u c t u r a l preference analysis A t h i r d category tends
to consider a t every m o m e n t of the parsing process, the
full complexity of the d a t a and the hypothesized partial
internal r e p r e s e n t a t i o n of the sentence, i n c l u d i n g , a t
least in principle, interaction w i t h k n o w l e d g e of t h e
world, aspects of memory, and p a r t i c u l a r task-oriented
behaviour
Worth m e n t i o n i n g here is C h u r c h a n d P a t i l ' s [1982]
work which a t t e m p t s to p u t o r d e r in the chaos of
complexity and " c o m p u t a t i o n a l load"
Our parser lies b e t w e e n the second and the third of the
a b o v e c a t e g o r i e s T h e p a r s e r is s e e n a s a
n o n d e t e r m i n i s t i c a p p a r a t u s t h a t d i s a m b i g u a t e s a n d
gives a "shallow" i n t e r p r e t a t i o n a n d a n incremental
functional r e p r e s e n t a t i o n of each processed f r a g m e n t of
the sentence The s t a t e of the p a r s e r is supposed to be
cognitively m e a n i n g f u l a t every m o m e n t o f t h e process
F u r t h e r m o r e , in p a r t i c u l a r , we a r e c o n c e r n e d w i t h
aspects of flexible word ordering This phenomenon is
specially r e l e v a n t in Italian, where, for d e c l a r a t i v e
sentences, Subject-Verb-Object is only the most likely
order - the o t h e r five p e r m u t a t i o n s of Subject Verb a n d
Object m a y occur as well We s h a l l briefly describe the
p a r s e r a n d its e n v i r o n m e n t and, by way of example,
i l l u s t r a t e its b e h a v i o u r in a n a l y z i n g " o s c i l l a t i n g "
sentences, i.e s e n t e n c e s in which one first perceives a
f r a g m e n t in one way, t h e n changes one's mind and t a k e s
it in a different way, then, as f u r t h e r i n p u t comes in,
g o i n g b a c k to t h e p r e v i o u s p a t t e r n ( a n d p o s s s i b l y
continuing like t h i s till the end of the sentence)
3 T h e p a r s e r
WEDNESDAY 2 [Stock 1986] is a p a r s e r b a s e d on
l i n g u i s t i c k n o w l e d g e d i s t r i b u t e d f u n d a m e n t a l l y
t h r o u g h t h e lexicon A word r e a d i n g includes:
- a s e m a n t i c r e p r e s e n t a t i o n of the w o r d , in the form of a
semantic n e t s h r e d ;
- static s y n t a c t i c information, including the category,
features, i n d i c a t i o n of l i n g u i s t i c f u n c t i o n s t h a t a r e
bound to p a r t i c u l a r nodes in the net One p a r t i c u l a r
specification is the Main node, h e a d of t h e s y n t a c t i c
c o n s t i t u e n t the word occurs in;
- dynamic syntactic i n f o r m a t i o n , including impulses to connect pieces of s e m a n t i c i n f o r m a t i o n , g u i d e d by syntactic constraints Impulses look for " f i l l e r s " on a
given search space (usually a s u b s t r i n g ) T h e y h a v e
a l t e r n a t i v e s , (for i n s t a n c e the word TELL h a s a n impulse to merge its object node with the " m a i n " of
e i t h e r a n NP or a s u b o r d i n a t e clause) An a l t e r n a t i v e includes: a c o n t e x t u a l c o n d i t i o n of a p p l i c a b i l i t y , a category, features, m a r k i n g , side effects ( t h r o u g h which, for e x a m p l e , c o r e f e r e n c e b e t w e e n s u b j e c t of a subordinate clause a n d a function of the m a i n clause can
be indicated) I m p u l s e s m a y also be d i r e c t e d to a different search space t h a n the n o r m a l one (see below);
- m e a s u r e s of likelihood These are m e a s u r e s t h a t are used for d e r i v i n g a n overall m e a s u r e of likelihood of a
p a r t i a l a n a l y s i s M e a s u r e s a r e i n c l u d e d for t h e likelihood of t h a t p a r t i c u l a r reading of the word and for aspects a t t a c h e d to a n impulse: a) for one p a r t i c u l a r
a l t e r n a t i v e b) for the relative position the filler c) for the overall necessity of finding a filler
- a c h a r a c t e r i z a t i o n of idioms involving t h a t word (For a description of the p a r t of the p a r s e r t h a t deals with the
i n t e r p r e t a t i o n of flexible idioms see [Stock 1987]) The only o t h e r d a t a are in the form of s i m p l e (non
a u g m e n t e d ) t r a n s i t i o n n e t w o r k s t h a t o n l y p r o v i d e restrictions on s e a r c h spaces where impulses c a n look for fillers In more t r a d i t i o n a l words it deals with the distribution of constituents A d i s t i n g u i s h i n g s y m b o l ,
$EXP, indicates t h a t only the occurrence of s o m e t h i n g expected b y preceding words (i.e for which a n impulse was set up) will allow t h e transition
The p a r s e r is b a s e d on of the idea of c h a r t p a r s i n g [Kay
1980, K a p l a n 1973] [see Stock 1986] W h a t is r e l e v a n t here is the fact t h a t "edges" correspond to s e a r c h spaces Edges are complex d a t a s t r u c t u r e s provided with a rich
a m o u n t o f i n f o r m a t i o n i n c l u d i n g a s e m a n t i c
i n t e r p r e t a t i o n of the f r a g m e n t , syntactic d a t a , p e n d i n g impulses, a n overall m e a s u r e of likelihood, etc D a t a on
a n e d g e a r e " u n i f i e d " d y n a m i e a l l y as indicated below
An a g e n d a is provided w h i c h i n c l u d e s four k i n d s of tasks: lexical tasks, traoersal tasks, insertion tasks, virtual tasks A lexieal task specifies a possible r e a d i n g
of a word to be i n s e r t e d in the chart A t r a v e r s a l t a s k specifies a n active edge a n d a n inactive edge t h a t can extend it A n i n s e r t i o n t a s k specifies a n o n d e t e r m i n i s t i e unification act, and a v i r t u a l t a s k involves extension of
a n edge to include a n inactive edge far a w a y in t h e
s t r i n g (used for long distance dependencies)
Trang 3L A
4 ~ ,
~ ~ o
l ~L p , k p , , , =
I ~ ~ t P a a R ~ to:
• + t l l l [ x : 4 •
I '~ l ++ l ? i $ , i k K t O 6
l I v P I ' ( P I / ~ R K t , ~
i
t 14 ~,bUH[I~II i ~ 5
I I ~ P ~ ( P v o ?
[ I b P I ( P I ~ R ~ f G 7
Y ( I I L K : 8
$1,¢~ J ~ E | 4 7 v e I
m ~ I l L
~ i , m l t
- ' =
m ~ ' "
m
• + P - P U I c , Pe, P - & | - T ~ C ' O ; ~
¶
I
The parser works asymmetrically with respect to the
"arrival*' of the M a i n node: before the M a i n node
arrives, an extension of an edge has almost no effect O n
the arrival of the Main, all the present impulses are
"unleashed" and m u s t find satisfaction If all this does
not happen then the n e w edge supposedly to be added to
the chart is not added: the situation is recognized as a
failure After the arrival of the Main, each n e w head
must find an impulse to merge with, and each incoming
impulse must find satisfaction Again, if all this does not
happen, the n e w edge will not be added to the chart
4 O v e r v i e w o f t h e e n v i r o n m e n t
W E D N E S D A Y 2 a n d i t s e n v i r o n m e n t a r e
i m p l e m e n t e d on a X e r o x L i s p M a c h i n e T h e
environment is composed of a series of specialized tools,
each one based on one or more windows (fig 1)
Using a mouse the user selects a desired behaviour from
menus attached to the windows We have the following
windows:
Fig I
- the m a i n W E D N E S D A Y 2 window, in w h i c h the sentence is entered Menus a t t a c h e d to t h i s w i n d o w specify different modalities (including "through" a n d
"stepping", "all parsings" or "one parsing") a n d a number of facilities;
- a w i n d o w w h e r e one c a n v i e w , enter a n d m o d i f y transition networks graphically (fig 2)
- a w i n d o w where one can view, enter and modify the lexicon A s a word e n t r y is a c o m p l e x o b j e c t for
W E D N E S D A Y 2, entering a n e w word can be greatly facilitated by a set of subwindows, each specialized in one aspect of the word, "knowing" how it m a y be and facilitating editing The lexicon is a lexicon of roots: a morphological a n a l y z e r a n d a lexicon m a n a g e r are integrated in the system Let us briefly describe t h i s point A lexicalist theory such as ours requires t h a t a large quantity of information be included in the lexicon This information has different origins: some comes from the root and some from the affixes All the information must be put into a coherent data structure, through a a
p a r t i c u l a r l y constrained u n i f i c a t i o n b a s e d p r o c e s s
Trang 4\ ' x
,Z,°,,T
Fig 2
V l E ~ 3 - P P - O i - - 0 ~ '
YERI~-O! l~-{l~l
1-01 - IIIF-OI~I V1E]~3- I I O - A O C
~1 PJ~O~ZXS ~ m
iii!ii~i
C t N I L N I L
ll-OOd ,'(3 N I L [31BJ X2 N I L
Test ~llP B e f e r e l l k e / ~ o ¢ i l F e 4 t l l c e l M 4 r k ~ d e t f e c t
aN
I~ER
( A - O b J ) ( ( T PP/mARK I ~JL = ( o e a )
( ( T NP 1 NIL NIL N[ (SUBJ)
(NUST 8 ) ( ( T NP ~ N | L N I L NI
Furthermore w e must emphasize the fact that, just as in
LFG, p h e n o m e n a such as passivization are treated in
Fig.3
the lexicon (the Subject and Object functions and the related impulses attached to the active form are
Trang 5rearranged) This is s o m e t h i n g t h a t the morphological
analyzer m u s t deal with The i n t e r n a l b e h a v i o u r of the
m o r p h o l o g i c a l a n a l y z e r is b e y o n d t h e scope of t h e
p r e s e n t paper We s h a l l i n s t e a d b r i e f l y discuss t h e
lexicon m a n a g e r , the role of which will be emphasized
here
The lexicon m a n a g e r deals with the complex process of
e n t e r i n g d a t a , m a i n t a i n i n g , a n d p r e p r o c e s s i n g t h e
lexicon One notable aspect is t h a t we h a v e a r r a n g e d the
lexicon on a hierachical baseis according to inheritance,
so t h a t properties of a p a r t i c u l a r word can be i n h e r i t e d
from a word class a n d a word class can i n h e r i t aspects
from a n o t h e r class One consequence of this is t h a t we
can introduce a graphic aspect (fig 3) a n d the user can
browse t h r o u g h the lattice (the lexicon a p p e a r s as a tree
of classes where one h a s s p e c i a l i z e d e d i t o r s a t e a c h
level) W h a t is even more r e l e v a n t is the fact t h a t one
can factorize knowledge t h a t is in the lexicon, so t h a t ff
one p a r t i c u l a r p h e n o m e n o n n e e d s to b e t r e a t e d
differently, the c h a n g e of i n f o r m a t i o n is i m m e d i a t e for
the words concerned Of course t h i s m e a n s also t h a t
t h e r e is a space gain: the s a m e i n f o r m a t i o n needs not to
be duplicated: complete word d a t a are reconstructed
w h e n required
T h e r e is also a m o d a l i t y by which one c a n e n t e r t h e
syntactic a s p e c t s of a word t h r o u g h e x a m p l e s , a la
TEAM [Grosz 19841 The r e s u l t s a r e less precise, b u t
m a y be useful in a more application-oriented use of the
e n v i r o n m e n t
- a window showing the p r e s e n t configuration of t h e
c h a r t ;
- a window t h a t p e r m i t s zooming into one edge, showing
several aspects of the edge, including: its s t r u c t u r a l
a s p e c t , i t s l i k e l i h o o d , t h e f u n c t i o n a l a s p e c t , t h e
specification of u n r e a l i z e d impulses etc
- a window d i s p l a y i n g in g r a p h i c f o r m t h e s e m a n t i c
i n t e r p r e t a t i o n of a n edge as a s e m a n t i c net, o r , if o n e
prefers so (this is usually the case w h e n the n e t is too
complex) in logic format;
- a window where one c a n m a n i p u l a t e the a g e n d a (fig 4)
A t t a c h e d to t h i s window we h a v e a m e n u including a set
of functionalities t h a t t h e t a s k s included in t h e a g e n d a
to be m a n i p u l a t e d : MOVE BEFORE, MOVE A F T E R ,
DELETE, SWITCH,UNDO etc One j u s t points to the
two p a r t i c u l a r t a s k s one wishes to o p e r a t e on with the
mouse a n d t h e n to t h e m e n u e n t r y In t h i s way t h e
desired effect is o b t a i n e d The effect c o r r e s p o n d s to
a p p l y i n g a different s c h e d u l i n g function: the t a s k s will
be picked up in the order here p r e s c r i b e d by hand This
tool, w h e n the p a r s e r is in the " s t e p p i n g " m o d a l i t y ,
L T vertex: 8 ¢~Lt: PREPMARK - : 1
I"T A : 9 a:15 t~WL.N: 5 6 NEWTr,
L I vertex: 4 caC PREPART eel: 1
Llr vertex: 2 ~:ax: PREP LM: 1
G 4 ] ~
mmK-4rl~
s k ~
m m
R
Fig 4
p r o v i d e s a v e r y e a s y way of a l t e r i n g t h e d e f a u l t
b e h a v i o u r of t h e s y s t e m a n d of t r y i n g o u t n e w strategies This m o d a l i t y of s c h e d u l i n g by h a n d is
c o m p l e m e n t e d by a s e r i e s of c o u n t e r s t h a t p r o v i d e control over the p e n e t r a n c e of these s t r a t e g i e s (The
p e n e t r a n c e of a n o n d e t e r m i n i s t i c a l g o r i t h m is the ratio
b e t w e e n t h e steps t h a t lead to the solution and the steps
t h a t are carried out as a whole in t r y i n g to o b t a i n the solution Of course t h i s m e a s u r e is included b e t w e e n 0
a n d 1.) Dynamically, one tries to find sensible s t r a t e g i e s , b y
i n t e r a c t i n g w i t h t h e a g e n d a W h e n , a f t e r
e x p e r i m e n t i n g formalizable heuristics h a v e b e e n t r i e d out, t h e y can be introduced p e r m a n e n t l y into the s y s t e m
t h r o u g h a given specialized function This is t h e o n l y place w h e r e some knowledge of LISP and of the i n t e r n a l
s t r u c t u r e o f W E D N E S A Y 2 is required
We s h a l l now briefly discuss a processing e x a m p l e t h a t
we h a v e b e e n able to u n d e r s t a n d using t h e e n v i r o n m e n t
m e n t i o n e d a b o v e The following e x a m p l e is a good
i n s t a n c e of flexibility a n d p a r s i n g problems p r e s e n t in Italian:
a N a p o l i preferisco R o m a a Milano
The complete sentence reads "while in Naples I prefer Rome to Milan" The problem arises d u r i n g the p a r s i n g process w i t h the fact t h a t the "to" a r g u m e n t of "prefer "
in I t a l i a n m a y occur before the verb, a n d the locative preposition "in" is a, the s a m e word as the m a r k i n g preposition corresponding to "to"
Trang 6The r e a d e r / h e a r e r first t a k e s a Napoli as a n a d v e r b i a l
location , then, as the verb preferisc9 is perceived, a
Napoli is clearly r e i n t e r p r e t e d as an a r g u m e n t of the
verb, {with a sense of surprise) As the sentence proceeds
after the object Rorna, the new word a_ causes t h i n g s to
change a g a i n a n d we go back with a sense of surprise to
the first hypothesis
The following t h i n g s s h o u l d be noted: - w h e n t h i s
second reconsideration takes place, we feel the surprise,
but this does not cause us to reconsider the sentence, we
only go back a d d i n g more to a n hypothesis t h a t we were
already working at; -the surprise seems to be caused not
by a h e a v y c o m p u t a t i o n a l load, b u t by a s u d d e n
r e a d j u s t m e n t of the weights of the hypotheses In a sense
it is a m a t t e r of memory, r a t h e r t h a n computation
We have been in a position to get WEDNESDAY 2 to
perform n a t u r a l l y in such situations, t a k i n g a d v a n t a g e
of the e n v i r o n m e n t The following s i m p l e h e u r i s t i c s
were found: a) try solutions t h a t satisfy the impulses (if
t h e r e are a l t e r n a t i v e s consider likelihoods); b) m a i n t a i n
viscosity (prefer the p a t h you are a l r e a d y following); a n d
c) follow the a l t e r n a t i v e t h a t yields the edge with the
g r e a t e s t likelihood, a m o n g edges of comparable lengths
The likelihood of a n edge depends on: 1) the likelihood of
the "included" edges; 2) the level o f o b l i g a t o r i n e s s of the
filled impulses; 3) the likelihood of a p a r t i c u l a r relative
position of a n a r g u m e n t in t h e string; 4) the likelihood of
t h a t t r a n s i t i o n i n t h e n e t w o r k , g i v e n t h e p r e v i o u s
transition
The critical points in t h e s e n t e n c e are the following
(note t h a t we d i s t i n g u i s h b e t w e e n a P P a n d a " m a r k e d
NP" possible a r g u m e n t of a verb, where t h e preposition
has no s e m a n t i c s asociated:
i) A t the beginning: only t h e P P edge is expanded, (not
the one including a ~ m a r k e d N P ' , b e c a u s e of s t a t i c
preference for the f o r m e r expressed in the lexicon a n d in
the t r a n s i t i o n network
ii) After the verb is detected: on the one h a n d t h e r e is a n
edge t h a t , if extended, would not satisfy a n obligatory
impulse, on the o t h e r h a n d , one t h a t would p o s s i b l y
satisfy one The ~marked NP" a l t e r n a t i v e is chosen
because of a) of the above heuristics
iii) After the object Roma: w h e n the preposition a_ comes
in, the edge t h a t m a y e x t e n d the sentence with a P P on
the one hand, a n d on the o t h e r h a n d a cycling active
edge t h a t is a p r o m i s i n g satisfaction for a n impulse are
compared Since this relative position of the a r g u m e n t is
so favourable for the p a r t i c u l a r v e r b preferisco (.9 to 1
for this position c o m p a r e d to the a n t e c e d e n t one), the
p a r s e r proceeds with the a l t e r n a t i v e view, t a k i n g a
Nap.o!i as a modh']er So it goes on, after r e e n t e r i n g t h a t
w o r k i n g h y p o t h e s i s T h e o b j e c t is a l r e a d y t h e r e ,
a n a l y z e d for the o t h e r r e a d i n g and does not need to be reanalyzed So a Milano is t a k e n as the filler for the impulse and the analysis is concluded properly
It should be noted t h a t the Final A r g u m e n t Principle [Fodor, K a p l a n and B r e s n a n 1982] does not work with the flexibility characteristic of Italian (The p r i n c i p l e would cause the r e a d i n g "I prefer [Rome [ in Milan]] to Naples" to be chosen a t point iii) above)
C o n c l u s i o n s
We h a v e i n t r o d u c e d a n e n v i r o n m e n t b u i l t a r o u n d
W E D N E S D A Y 2, a nondeterministic parser, o r i e n t e d
t o w a r d s e x p e r i m e n t i n g with dynamic s t r a t e g i e s The
c o m b i n a t i o n of i n t e r e s t i n g t h e o r i e s a n d s u c h tools realizes both m e a n i n g s of the word " e x p e r i m e n t a l " : 1)
s o m e t h i n g t h a t i m p l e m e n t s new ideas in a prototype; 2)
s o m e t h i n g b u i l t for the sake of m a k i n g e x p e r i m e n t s We
t h i n k t h a t t h i s a p p r o a c h , possibly i n t e g r a t e d w i t h
e x p e r i m e n t s in psycholinguistics, can help increase our
u n d e r s t a n d i n g of parsing
A c k n o w l e d g e m e n t s
F e d e r i c o Cecconi's help in t h e g r a p h i c a s p e c t s a n d lexicon m a n a g e m e n t h a s b e e n precious
References
Church, K & Patil, R Coping with syntactic a m b i g u i t y
or how to p u t the block in the box on the table American Journal of Computational Linguistics, 8; 139o149 (1982)
F e r r a r i , G & Stock,O S t r a t e g y s e l e c t i o n for a n A T N
s y n t a c t i c parser Proceedings of the 18th Meeting of the Association for Computational Linguistics, P h i l a d e l p h i a
(1980)
Ford, M., B r e s n a n , J & Kaplan, R A competence based
t h e o r y of s y n t a c t i c closure In B r e s n a n , J , Ed The Mental Representation of Grammatical Relations The MIT Press, C a m b r i d g e , (1982)
Gazdar, G P h r a s e s t r u c t u r e g r a m m a r In J a c o b s o n a n d
P u l l m a n (Eds.), The Nature of Syntactic Representation
Dordrecht: Reidel ( 1981 )
Trang 7Grosz, B TEAM, a transportable natural language interface system In Proceedings of the Conference on Applied Natural Language Processing, Santa Monica
~1983~
Joshi, A., & Levy, L Phrase structure trees bear more fruits then you would have thought American Journal
of Computational Linguistics,8; 1-ll (1982)
Kaplan, R A general syntactic processor In Rustin, R {Ed.), Natural Language Processing Englewood Cliffs, N.J.: Prentice-Hall (1973)
Kaplan,R & Bresnan,J Lexical-Functional Grammar: a formal system for grammatical representation In
B r e s n a n , J , Ed The Mental Representation of Grammatical Relations The MIT Press, Cambridge, 173-281 (1982)
Kay, M Algorithm Schemata and Data Structures in Syntactic Processing Xerox, Palo Alto Research Center (October 1980)
Kay, M Functional Grammar In Proceedings of the 5th Meeting of the Berkeley Linguistic Society, Berkeley, 142-158(1979}
Marcus, M An overview of a theory of s y n t a c t i c recognition for n a t u r a l language (AI memo 531) Cambridge, Mass: Artificial Intelligence Laboratory, (1979)
Pereira, F & Warren, D., Definite Clause Grammars for language analysis A survey of the formalism and a comparison with Augmented Transition Networks
Artificial Intelligence, 13; 231-278 (1980)
Small, S Word expert parsing: a theory of distributed word-based natural language understanding (Technical Report TR-954 NSG-7253) Maryland: University of Maryland (1980)
Stock, O Dynamic Unification in Lexieally Based Parsing In Proceedings of the Seuenth European Conference on Artificial Intelligence Brighton, 212-221
(1986)
Stock, O Putting Idioms into a Lexicon Based Parser's Head To appear in Proceedings of the 25th Meeting of the Association for Computational Linguistics Stanford,
Ca [1987]
Thompson, H.S Chart parsing and rule schemata in GPSG In Proceedings of the 19th Annual Meeting of the Association for Computational Linguistics Alexandria,
Va (1981)