In the GPSG framework, passive VP rules are derived by a metarule from active VPs: 1.. the metarule as it stands would apply to VPs which require a sentential subject, like "bother'.. Th
Trang 1P A S S I V E S Steve Pulman
University of Cambridge Computer Laboratory,
Corn Exchange Street
Cambridge CB2 3QG UK
A B S T R A C T
The English passive construction has played a central
role in the to-ings and fro-ings of grammatical theory over
the last 30 years, from the earliest days of transformational
grammar, to more recent, surface oriented theories of syn-
tax The casual reader of the linguistic literature might
therefore suppose that the computational linguist looking
for an off the shelf analysis of passives would be able to
choose from among several competing analyses, each of
which accommodated the facts, but perhaps derived them
from (or from them} different theoretical principles Un-
fortunately, this is not the case as we shall see All of
the analyses that [ am familiar with are incomplete, or in-
accurate in some respects, or simply unprogrammable in
any straightforward form The present paper is an a t t e m p t
to remedy this situation, and to provide such an off the
shelf analysis of the syntax and semantics of passives The
analysis of this central construction will be couched within
a simple and computationally tractable syntactic and se-
mantic formalism, and should translate easily to most cur-
rently popular formalisms It will be quite eclectic, freely
borrowing from several different grammatical theories
T w o u n s a t i s f a c t o r y a n a l y s e s
The original starting point for the analysis here was that
presented in Gazdar et al 1985 (also found unsatisfactory
by Kilbury t986) In the GPSG framework, passive VP
rules are derived by a metarule from active VPs:
1 VP - - NP W ~ VPpas V~ (PPby)
The interpretation of this metarule is as follows: for ev-
ery rule expanding VP which introduces an NP daughter
there is also to be a rule which has the VP marked as pas-
sive, does not contain that NP daughter, and may contain
a PP headed 'by' Feature principles ensure that the verb
heading the VP will have passive morphology in this latter
c a s e
There are several problems with this account An engi- neering problem concerns the interpretation of GPSGs for computational purposes One more or less workable con- strual regards the m e t a g r a m m a r as a set of instructions for producing a 'compiled" object grammar consisting of context free rules augmented with some feature matching mechanism However, this treatment produces large num- bers of such rules When 'slashed' versions of VP rules are also compiled out the multipllcative effect can lead to many hundreds of VP rules in a linguistic description While not fatal, this is still a problem for constructing efficient parsers There are also several descriptive problems As Kilbury points out the metarule as it stands would apply to VPs which require a sentential subject, like "bother' on one of its subcategorisations Thus we will be able to generate junk like:
2 T h a t Kim left was bothered [by Sandy)
Similarly for VPs introducing complements of verbs like
"elect' we will get two outputs from the metarule, only one
of which is a sensible one:
a v P - v(211 ~'m s P ('etecc etc are VI211 )
4 a We elected Kim president
b Kim was elected president
c *President was elected Kim The metarule wilL, however, fail to apply in the case of VPs introducing a sentential object, since there is no NP daughter, failing to generate perfectly good examples like 6b:
5 V P - - V S"
6 a They vehemently denied that there had been a plutonium leak
b That there had been a plutonium leak was vehemently denied
Most of these problems are fairly easily fixable: for ex- amples like 2 it is a reasonable response to say that they are syntactically ok but sortally deviant: the obvious fix for things like 6 is to regard sentential complements of this
Trang 2type as d o m i n a t e d by NP as many o t h e r accounts have
done More serious is t h e fact t h a t the m e t a r u l e t r e a t m e n t
will also fail to get the right results in those instances where
the passivised NP is not a d a u g h t e r of VP T h e r e are several
different cases here: so-called "double passives" like:
7 a Kim was taken a d v a n t a g e of
b A d v a n t a g e was taken of K i m
If 'take a d v a n t a g e o f ' is treated as a c o m p l e x V only
one passive will be derived, for ' a d v a n t a g e ' will not be a
daughter of NP T h e r e are also ' p r e p o s i t i o n a l passives' like:
8 a Kim c a n ' t be relied on
b T h a t meal w a s n ' t paid for
where the "object' NP is actually inside a PP as required
in order to also be able to generate:
9 a On Kim, you can rely absolutely
b For that meal, the c o m p a n y will pay
Passives for which there is no active equivalent will fail
to be derived (by the m e t a r u l e , at least}:
I0 a Sandy was said to be a C N D activist
b *They said Sandy to be a C N D activist
Finally, there is a p r o b l e m a b o u t agent PPs T h e m e t a r u l e
t r e a t m e n t allows for an optional agent phrase as a con-
stituent of the passive VP T h e I D / L P format presupposed
in G P S G allows for some freedom of ordering between P P s
that are introduced by a VP: thus the o u t p u t of the m e t a r u l e
for an input VP VI I, NP PP will allow possibilities like:
11 a A book was given by K i m to S a n d y
b A book was given to Sandy by K i m
But optional P P modifiers of VP are (correctly) intro-
duced by a rule VP - - VP PP T h e r e is thus no way of
accounting for cases where a non-subcategorised-for P P in-
tervenes between verb and agent PP:
12 J o h n was iiiiarrested in the p a r k on Friday!
by the Special Branch
even t h o u g h such cases are freely possible (The s a m e
problem occurs with Bach's (1980) analysis of passives)
Bresnan (1982) presents an analysis of passives within
the LFG framework Lexical entries for passive forms of
verbs are derived from those for the active form via a lex-
ical rule which makes the a p p r o p r i a t e morphological and
semantic changes Then passive VPs are parsed using the
same context free phrase s t r u c t u r e rules as for actives, with
principles of functional coherence and completeness m a k i n g
sure t h a t subcategorisation requirements are met and the
a p p r o p r i a t e i n t e r p r e t a t i o n s arrived at
T h e r e are several problems with the p r o p o s e d lexical
t r e a t m e n t of passives, at least one of which could be re- garded as fatal [t is not clear how passives with no active source are derived, although p r e s u m a b l y the required lexi- cal form could simply be listed Cases where the passivised NPs are not d a u g h t e r s of VP are dealt with by m a k i n g t h e m ambiguous, by stipulation in the "take a d v a n t a g e o f ' case and by a lexical rule in the "prepositional passive' cases: t3 V - - iV Piv
This has the u n f o r t u n a t e effect t h a t the unpassivised,
u n m o v e d versions of these phrases are also syntactically ' a m b i g u o u s ' , i.e they receive two or m o r e parses, corre- sponding to no discernible s e m a n t i c difference:
14 a K i m can be [relied on'
b [On K i m I, you can always irelyi
c You can !irely on] ; K i m !
d You can irely o n K i m /
In the case of those verbs which can take two preposi- tions, the rule must presumably have applied twice:
15 a T h e bed has been t h o r o u g h l y !irolled around v o n ' v
b On the bed the children rolled a r o u n d
c ? A r o u n d on the bed the children roiled giving the curious consequence that the unpassivised version will now be three ways anabiguous:
16 a irolled a r o u n d on the bed
b !irolled around on the bed
c !?rolled around on the b e d Bresnan's lexical rules o p e r a t e within the lexicon and not during a derivation The)- 'express patterns of redun- dancy that o b t a i n a m o n g large but finite classes of lexical entries' (Kaplan and Bresnan 1982 l_'30} This hag the con- sequence that the lexical analysis can only be s u s t a i n e d if there is a finite n u m b e r of passivi~able verbs in English For all practical purpose~, we can suppose that there is, but there is an a r g u m e n t to be m a d e that theoretically, there is an infinite n u m b e r of such verbs, arising as a result
of regular and p r o d u c t i v e morphological processes
A simple version of this a r g u m e n t can be m a d e as fol- lows: there is p r e s u m a b l y no upper limit to the n u m b e r
of proper names in English: we can always make up new ones and we can always c o n c a t e n a t e existing names to form new ones: Slater-Walker Hewiett-Packard etc But we can form verbs using 'ise" from all of these: Bresnanise Hewlett-Packard-ise etc And these verbs are all passivis- able (Thatcherised X,[arks-and-Spencerised) hence there is
a potentially infinite n u m b e r of passive verbs W i t h o u t an infinitely large lexicon the lexical t r e a t m e n t will be unable
to cope It is not clear to me how central the claim q u o t e d
Trang 3above is to the theory of L F G But either it will have to
be abandoned, or s o m e o t h e r way of handling passives will
have to be found w i t h i n t h a t theory
A unification-based analysis
T h e analysis here wilt be couched within a simple unification-
enriched context-free formalism T h e intention is not to
p r o m o t e the formalism, but to use it to describe the anal-
ysis in a way t h a t will m a k e it clear, and easy to trans-
late into your own favourite formalism T h e s e m a n t i c s of
the features in the rules is t h a t provided by o r d i n a r y (i.e
Prolog-style) unification T h e semantic expressions asso-
ciated with the rules are instructions for building logical
forms in a kind of 'intensionless M o n t a g u e ' : a typed higher
order logic, like t h a t in P T Q w i t h o u t the intension and ex-
tension operators S e m a n t i c translations are a s s e m b l e d o n
a rule to rule basis by function application and composi-
tion (I a s s u m e some s e p a r a t e m e c h a n i s m for those cases
where quantifier scope is not d e t e r m i n e d by the syntax.)
An e x a m p l e set of rules will illustrate:
16 Sitype decl! -~ N P i a g r X V P i a g r X
: NP (VP)
i.e agr on N P m u s t be identical to t h a t on VP: the s e m a n -
tics of the S is w h a t you get by a p p l y i n g the m e a n i n g of
the N P to that of the VP
17 N P i a g r X Detlagr X' N b a r a g r X
: Det (Nbar)
18 Nbar!agr X - - N a g r X
: N
19 V F a g r X ~, ~ V!agr X, s u b c a t npi NP
: V (NP)
A simple set of global default values for features is as-
sumed: thus if a feature does not explicitly a p p e a r on a
category that it is a possible feature for, the default value
if any will be placed on it otherwise it will get a "don't
care' variable
U n b o u n d e d dependencies can be a c c o m m o d a t e d by a
version of the ' g a p - t h r e a d i n g ' m e c h a n i s m described in Kart-
t u n n e n (1986) T h e s e m a n t i c s of u n b o u n d e d dependencies
can be treated in the s a m e way as G K P S 1985: a con-
s t i t u e n t normally of t y p e o, with a gap of t y p e 3, will be
of type ,3 - - cL Gaps are of course NPs, P P s etc which
are missing an N P or PP
So much for background Using this t y p e of m a c h i n e r y
we can obviate the need for a passive m e t a r u l e Essentially
the idea is to c a p t u r e literally the o b s e r v a t i o n t h a t , syn-
tactically, a passive V P is j u s t like an active V P e x c e p t
t h a t the verb has passive morphology, and t h e r e is an NP
missing T h e missing NP is treated as a kind of ' b o u n d e d
d e p e n d e n c y ' In the s a m e way t h a t G P S G style analyses
i n t r o d u c e u n b o u n d e d dependencies at the top of a sentence
for wh-constructions, we will introduce a b o u n d e d d e p e n - dency at the top of a passive VP
We will assume t h a t regular passive verbs are derived
by a productive morphological process a t t a c h i n g a passive affix, e n / e d {See Russell et al 1986 for details of the mor- phological analysis s y s t e m which is used) T h e s e m a n t i c consequences of this are discussed below T h i s process will apply to any verb of the a p p r o p r i a t e class, w h e t h e r or not it
is itself the p r o d u c t of prior morphological processes T h e syntactic effect of this affixation is t h a t passive verbs are marked ~vform passive:, or s o m e t h i n g similar: ' v f o r m ' here
is essentially the s a m e feature used in G P S G 8 5 , a p p e a r i n g also on the VP c o n s t i t u e n t We also i n t r o d u c e a feature distinguishing passive from active VPs: :passive - / - T h i s feature can also occur on NPs for a reason t h a t will be
i m m e d i a t e l y a p p a r e n t T h e default value for passive is -
T h e r e are at least two rules i n t r o d u c i n g passive VPs, one as p o s t n o m i n a l modifiers, and one as c o m p l e m e n t s to
"be" and "get" etc:
20 VP[agr X, vform Y ~
Vbelagr X v f o r m Y]
V P ' v f o r m passive, passive -;-]
: Vbe (VP)
T h e b e h a v i o u r of the passive feature is w r i t t e n into the
VP rules for the different types of v e r b that can passivise (I a m a s s u m i n g a G P S G i s h t r e a t m e n t of s u b c a t e g o r i s a t i o n here)
T h u s a VP rule for a transitive VP might look like:
21 V P % g r X vform Y, passive Z - -
Viagr X, v f o r m Y s u b c a t trans NPipassive Z
: v ( N P )
U n d e r n o r m a l c i r c u m s t a n c e s , the rule will g e n e r a t e or- dinary t r a n s i t i v e VPs, but w h e n a p p e a r i n g as c o m p l e m e n t
to ' b e ' etc will require passive m o r p h o l o g y on the verb, and will c o n t a i n t h a t curious object, an N P m a r k e d [passive +i Such N P s are i n t r o d u c e d by a rule:
22 NP[passive +: -* e : AP _~ x (P x)
A passive NP is an e m p t y NP, but a different t y p e of e m p t y
NP from u n b o u n d e d d e p e n d e n c y gaps (This prevents the same NP from being both passivised and w h - m o v e d in the same VP) It means, roughly "something' All o t h e r NP~ default to p a s s i v e -
Syntactically then a passive version of a t r a n s i t i v e VP looks just like the active, except t h a t the object is empty, Notice that the features g u a r a n t e e t h a t the passive N P is empty if and only if the verb is in the passive form T h e
a t t r a c t i o n of this t r e a t m e n t is t h a t it is the S A M E rule t h a t generates both the active and the passive versions: no e x t r a rules are involved
Trang 4We do similar things with the o t h e r types of VP which
can passivise:
(i) verb-particles:
23 V P ; v f o r m X, passive Y -*
V v f o r m X s u b c a t prt I
P UPipassive Y:
-giving things like:
24 The fight was switched off
Notice t h a t we can choose w h e t h e r it is the moved (NP P)
or the u n m o v e d (P UP} version which is capable of pas-
sivising: but only one of t h e m , for otherwise passives will
get two parses
(ii) phrasal verbs:
25 VP vform X passive Y -
V vform X s u b c a t p h r
P
P
NP passive Y -giving:
26 J o h n was looked up to by his children
(iii) the raised version of 'object raising' verbs:
27 V P i v f o r m X, passive YI -*
Vlvform X, s u b c a t objr 1 NP!passive Y!
V P (iv) both types of dative:
2g VP vform X passive Y
V vform X s u b c a t dat
UP passive Y
PP
29 VP vform X, passive Y -~
V:vform X, s u b c a t d a t m v t :
N P ' p a s s i v e Y:
UP
We prevent passive from applying where it should not by
simply leaving out the passive feature on the relevant rules:
it then defaults to value -
For passives t h a t have no active equivalent, we rely on
the same mechanism There are two types of case, those like
"said', ' r u m o u r e d ' etc and those like "surprised at', ' a s t o n -
ished at' For the 'say' type cases, the passive version will
be derived by the object raising rule above T h e i r passive
entry will be listed directly in the lexicon with the relevant
subcategorisation T h e r e will be no entry for the active ver-
sion on that subcategorisation T h e absence of the actixe
version guarantees that we will not generate things like:
30 ~They r u m o u r e d him to be a spy because the only lexical e n t r y for ' r u r n o u r ' with the a p p r o - priate s u b c a t e g o r i s a t i o n is the passive form, and the fea- tures g u a r a n t e e t h a t this c a n n o t cooccur with a full NP
in this s t r u c t u r e T h e familiar ~promise persuade" alterna- tion is precisely" the inverse of this: we can simply a r r a n g e for the lexical e n t r y for "promise" on this s u b r a t e g o r i s a t i o n
to be marked as not undergoing affixation by the passive
m o r p h e m e T h u s we will get the following p a t t e r n :
31 J o h n promised persuaded Bill to leave
32 Bill was " p r o m i s e d : p e r s u a d e d to leave For the ' s u r p r i s e d ' cases, we assume t h a t there are actually two different verbs, with different semantics: the o r d i n a r y transitive verb denotes an event, and behaves regularly:
33 J o h n surprised B i l l
34 Bill was surprised by J o h n The o t h e r denotes a state and does not have an active form: it subcategorises for "at" and is listed directly as a passive, with the a p p r o p r i a t e semantics:
35 "Tile noise was surprising at Bill
36 Bill was surprised at the noise
37 VP vform passive - -
V ' v f o r m passive, s u b c a t s r p r s P a t U P Now we turn to the 'rely on" t y p e of case Here the p r o b l e m
is that the missing NP is not a d a u g h t e r of the VP: a fatal problem for the m e t a r u l e t r e a t m e n t O u r solution is to pass
on the b o u n d e d U P dependency down t h r o u g h a P P :
38 V P : v f o r m X passive Y'
V ' v f o r m X, s u b c a t rivl P P i p a s s i v e Y!
38 PP~passive X ~ P U P passive X i However, this is as far as the passive feature can go, unlike true u n b o u n d e d dependencies:
39 a On J o h n , you can d e p e n d
b J o h n , you can d e p e n d on
c ,John can be d e p e n d e d on
d John you can depend on the promises of
e ' J o h n can be d e p e n d e d on the promises of This can be simply achieved by not m e n t i o n i n g the passive feature anywhere else
A notorious problem for m a n y analyses of passive is the case of verbs like "sleep" and "walk" which a p p e a r to be subcategorised as intransitives, bur occur in passives like the following:
40 T h i s bed was slept in by the queen
41 T h e new grass s h o u [ d r ( t be w a l k e d over
Trang 5Apparently, an NP inside an optional modifier can be pas-
sivised A simple account of this can be given by a d d i n g
the following rule:
42 V P [ v f o r m passive, passive -] *
Vivform passive, s u b e a t intrl P
(We d o n ' t need to b o t h e r looking for an NP which is always
passive} T h i s claims t h a t any intransitive v e r b can b e h a v e
in this fashion, which seems a p p r o x i m a t e l y correct:
43 T h e plane was first flown in last year
44 T h e film was snored t h r o u g h / s n e e r e d at by
most of the audience
However the p u t a t i v e PP modifier has been flattened, into
a P NP passive - sequence (i.e just a P): this is in o r d e r
to facilitate the semantic t r e a t m e n t of this c o n s t r u c t i o n
and has no adverse syntactic effects It can be t h o u g h t of
as an i m p l e m e n t a t i o n of the "reanalysis" t r e a t m e n t of phe-
n o m e n a like this often a d v o c a t e d within the C;overnment
and Binding framework
This t r e a t m e n t has the added a d v a n t a g e of simplifying
our s t a t e m e n t of affixation of the passive m o r p h e m e , which
now might as well apply freely to any verb i n d e p e n d e n t l y
of its subcategorisation Of course, the result might not be
well-typed, as in the case of these intransitive verbs: we
will return to this m a t t e r when discussing their s e m a n t i c s
below Passive forms of o t h e r verbs which really d o n ' t pas-
sivise can never figure in a VP given the rules, features and
subcategorisation regime we are assuming
A r e m a i n i n g p r o b l e m is t h a t of "double passives' like
45 a A d v a n t a g e was taken of J o h n
b J o h n was taken a d v a n t a g e of
T h e r e are several solutions one might explore here We
could have a rule for j u s t this handful of verbs of the form:
(keep tabs on, lay siege to, take pity on)
46 VP vform X, passive Y
V ' v f o r m X s u b c a t idiom:
NP passive Y
P P where the NP must be as required by the verb T h e n
for the other passive we could assume a c o m p l e x lexical
entry for "taken a d v a n t a g e of" which is s u b c a t e g o r i s e d as an
o r d i n a r y transitive This is the suggestion m a d e by m a n y
linguistic t r e a t m e n t s W i t h i n the feature s y s t e m used here
it is in fact possible to do all this by b r u t e force: a s s u m e
t h a t the NP rules p e r c o l a t e up from their head a feature
' n f o r m ' which has as value the actual stern of the noun
T h e n we have two rules:
47 V P ; v f o r m X, passive Y
V[vform X, s u b c a t idiom, needs Z]
NP:passive Y, nform Z
P P
48 V P v f o r m passive, passive - -
V vform passive, subcat idiom, needs Z
N P n f o r m g
P P p a s s i v e -
T h e n this idiomatic sense of 'take" is e n t e r e d in the lexi- con as V subcat idiom, needs a d v a n t a g e etc T h e active form only gets parsed by rule 47, but b o t h passive versions are accepted (Incidentally, the idea of m a k i n g different fea- tures share variable values can enforce correct v e r b - p a r t i c l e combinations, p a r t i c u l a r required PP forms, etc)
This concludes the list of some of t h e s y n t a c t i c prob- lems faced by any analysis of verbal passives, a n d solved
by that presented here I have not to d a t e e n c o u n t e r e d any other e x a m p l e s of passives in English which will not yield to s o m e c o m b i n a t i o n of the m e t h o d s used in the pre- ceding While I would be the first to concede t h a t these analyses leave a great deal to be desired in t e r m s of el- egance, e x p l a n a t o r y power, and the o t h e r g r a n d criteria
by which syntactic theorie~ can be j,Ld~ed, they are con- ceptua]ly and c o m p u t a t i o n a l [ y quite Mmpie and a p p e a r to
be descriptively a d e q u a t e , a~though s o m e w h a t Iongwinded:
a more e c o n o m i c a l g r a m m a t i c a l f o r m a l i s m m i g h t express things m o r e succinctly
I have said n o t h i n g a b o u t adjectival passives: these seem to be of two types, those t h a t are already lexicalised as distinct items, like 'closed', and those p r o d u c e d by (fairly)
p r o d u c t i v e derivationat rules, where the s u b c a t e g o r i s a t i o n
of the verb (minus the passivised NP) is inherited by the adjective:
49 T h e door r e m a i n e d o p e n closed
50 T h e bottle r e m a i n e d e m p t y filled with wine
It is simple to i n c o r p o r a t e a [exical t r e a t m e n t of this phe-
n o m e n o n into the analysis here and so I will say n o t h i n g more a b o u t t h e m (see Levin and R a p p a p o r t 1986 for a de- tailed study)
S e m a n t i c s o f P a s s i v e s
I t u r n now to the semantics of passives We have been assuming t h a t the passive form of a verb, unless it is irreg- ular, is derived by a m o r p h o l o g i c a l rule of affixation T h e semantic effect of passive m o r p h o l o g y on a verb is to switch around its first two a r g u m e n t s T h u s a t r a n s i t i v e verb in simplified form would be r e p r e s e n t e d as say:
5l hit: Aeso (hit e s o)
Trang 6(where a A is followed by a sequence of variables, this
is to be regarded as a s h o r t h a n d for a ' c u r r i e d ' expression:
i.e Axyz Ax Ay Ax .) T h e first variable in 51 is
an "event' variable: I a m a s s u m i n g the Davidsonian(1980)
analysis of verbs here: more on this below I a s s u m e an
affixation rule s o m e t h i n g like:
52 V - - V A f : A f ( V )
Affixes are in general (polymorphic) things which take verbs
to verbs: the relevant ones here i n t r o d u c e tenses and the
passive
(i) p a s t / ' p r e s e n t = AVe (V e) /~, ( p a s t / p r e s e n t e)
(ii) passive is of type {ez-(e~-(e~-a)))) ;,- (e~-(e~(e~-a))))
For transitive verbs passive a m o u n t s to AVexy (V e y x)
Intuitively, tenses are predicates on events, and passive is
an o p e r a t o r that has the effect of switching round the first
two (non-event) a r g u m e n t s of the verb it applies to T h e
easiest way to see how all this fits t o g e t h e r is to give s a m p l e
derivations from the following little g r a m m a r (I will o m i t
the feature specifications on rules for simplicity):
S -* NP VP : -qe (NP (VP e)}
: the event variable is b o u n d at the top level
NP ~ Name : AP (P Name)
: the rule raises the type
VP -* Vtr NP : Aea (NP l a b (V o a b)))
: VPs are of type (e~ ( e ~ t ) )
V P - - Vbe VP : Aea (Vbe e) " ((VP e) a)
; assume t h a t "be' etc j u s t carries tense
VP - - VP PP : Aea ((VP e) a) ,' ( e e e)
; P P modification is t r e a t e d as a predication
; on the event
P P * P NP : Ax (NP (Ay (P x y)))
; P P s are of t y p e (e>-t}
Given these rules, and lexical entries, a V P like 'hit Bill'
will be translated, after s o m e variable r e n a m i n g a n d a few
rounds of b e t a reduction, as:
53 Aea (hit e a Bill) A (past e)
Modifying this VP with a P P like 'in C a m b r i d g e ' will give
a VP with translation:
54 Aea (hit e a Bill) ," {.past e)
• (in e C a m b r i d g e ) Incorporating this into a sentence with subject ' J o h n ' the
above rules will get us:
55 _~e (hit e J o h n Bill) " (past e)
.', (in e CambridgeJ
as a translation of ' J o h n hit Bill in C a m b r i d g e ' : "there was
a hitting by J o h n of Bill event, in the past in C a m b r i d g e '
In the case where we have a passive like "Bill was hit' application of the passive affix to the verb produces:
56 !AVexy (V e y x)! (Aeso (hit e s o)) reducing to:
57 Aexy (hit e y x}
T h e VP c o n t a i n i n g the e m p t y passive NP will t r a n s l a t e
a s :
58, Aea (!AP (3i (P i))]
(Ab ([Aexy (hit e y x)l e a b))) Notice that the passive m o r p h e m e has changed the o r d e r
in which the verb expects its a r g u m e n t s T h i s beta-reduces
tO:
59 Aea _:i (hit e i a)
I n c o r p o r a t i n g this with the VP that introduced the passive
VP as c o m p l e m e n t to "was' gives us:
60 ,~ea _:i (hit e i a) ; (past e)
If we now c o m b i n e this with the subject we will get, after reduction:
61 _:el (hit e i Bill) "., (past e) ' T h e r e was a past hitting by s o m e t h i n g of Bill e v e n t ' Notice t h a t agent phrases for passives are treated in ex- actly the s a m e way as any optional V P - m o d i f y i n g PP So
a VP like "was hit by John" - given s o m e obvious a s s u m p - tions a b o u t the translation of agentive "by' and some way
of selecting the translation a p p r o p r i a t e to the sentence (as opposed to a locative or t e m p o r a l "by') - will t r a n s l a t e as:
62 Aea _=i (hit e i a) ' (past e) ' (agent e J o h n ) Notice that agentive PPs are not required to be adjacent
to the passive verb correctly T h e r e is thus no syntactic connection between the presence of an agent phrase and passive morphology This means t h a t a sentence like:
63 J o h n hit Bill by ['red
on the agent reading of the PP is treated as syntactically well-formed, but thematically incoherent in the s a m e way that:
64 J o h n hit Bill with a h a m m e r with a chair
Trang 7is where t h e P P s b o t h have i n s t r u m e n t readings
We need an a x i o m a c h e m a to make the t r a n s l a t i o n s of
' J o h n hit Bill' and 'Bill was hit by John" inter-deducible
This is not s o m e t h i n g e x t r a d e m a n d e d by this analysis
however: it is already needed to establish the connection
between agents and certain types of events to account for
the i n t e r p r e t a t i o n of agent phrases in nominatisations where
the passive is not involved:
65 T h e h i t t i n g of Bill by J o h n was an accident
For the m o s t part this s e m a n t i c analysis e x t e n d s s t r a i g h t -
forwardly to the o t h e r cases of passives discussed earlier
T h e i r are three cases which need f u r t h e r c o m m e n t , how-
ever For datives, I a s s u m e t h a t the NP PP and NP NP
forms have different s u b c a t e g o r i s a t i o n s which are related
by an obvious r e d u n d a n c y rule in the lexicon However we
can assume t h a t the verb has the s a m e s e m a n t i c s in both
cases:
66 kexyz (give e x y z)
Associated w i t h the rule t h a t generates the d i t r a n s i t i v e
form will be a ' d a t i v e ' o p e r a t o r , defined thus:
67 kVexyz (V e x z y)
This has t h e effect of switching r o u n d the final two argu-
ments of t h e verb T h e rules will be:
68 V P ~ Vdat NP PP
: Aex ( P P (Az (NP (Ay ( ~ d a t e x y z)))))
69 VP - - V d m NPi NPj
: Aex (NPj (Az (NPi (Ay (V e x y z)l)))
where V is actually the d a t i v e o p e r a t o r applied
to V d m
I assume t h a t a r g u m e n t PPs like those associated with
datives t r a n s l a t e as s o m e t h i n g having the type of an NP
r a t h e r t h a n a PP, as befits their i n t e r p r e t a t i o n T h i s can
be i m p l e m e n t e d simply by m a r k i n g these P P s as a r g u m e n t s
and m a k i n g the t r a n s l a t i o n of a P P c o n s t i t u e n t so m a r k e d
consist simply of the d a u g h t e r NP: the preposition con-
tributes n o t h i n g to t h e m e a n i n g In the case of the Vdat
rule, when the verb is in the passive, things are exactly anal-
ogous to t h e earlier cases ( m o d u l o differences caused by the
['act t h a t t h e verb is of a different type): the passive mor-
p h e m e simply switches round the a r g u m e n t s c o r r e s p o n d i n g
to subject and direct object In the case of the Vdat rule,
w h e n in the active, the d a t i v e o p e r a t o r shifts the final two
a r g u m e n t s , so that e v e n t u a l l y the i n n e r m o s t t e r m contain-
ing the verb will be of the form give e x z y In the
passive, what the dative o p e r a t o r applie¢ to is of the form
give e y x z, because of the prior result of a t t a c h i n g the
passive affix T h u s t h e result of t h e dative o p e r a t o r is of
t h e form give e y z x
I will spare you the sequence of b e t a r e d u c t i o n s involved but with the rules and lexical entries given the right re- sults are achieved (For those w i t h long linguistic m e m o r i e s the sequence of l a m b d a m a n i p u l a t i o n s involved may seem strongly reminiscent of the s t a n d a r d theory T G t r e a t m e n t
of c o n s t r u c t i o n s like this)
T h e t r e a t m e n t of a r g u m e n t P P s here is also needed for the 'rely on" type cases T h e s e m a n t i c s of the rule is simple:
70 VP - - Vr P P : Aex ( P P ray (Vr e x y)))
T h e P P here also has the type of NP
T h e final wrinkle concerns the a p p e a r a n c e of intran- sitive verbs in passives A p p l y i n g a passive affix to an in-
t r a n s i t i v e verb directly results in s o m e t h i n g t h a t is not well typed: intransitives are here of t y p e (e>-(e>-t)) T h e sim- plest course is to a s s u m e t h a t u n d e r these c i r c u m s t a n c e s the passive affix is s i m p l y ignored T h e n we c a n associate
w i t h the relevant rule t h e s e m a n t i c s as follows:
71 V P p a s , Vintr P : Aex {_=i (Vintr i) (P e x)) Given that tile m e a n i n g of "sleep' is Aex (sleep e x), this wilt produce a t r a n s l a t i o n of "This bed was slept in r e c e n t l y ' as:
72 - e i b (sleep e i) ' (bed b) " {past e) ,' (in e b) :' (recent e)
' T h e r e has been a past sleeping of s o m e t h i n g event and
t h a t event was in this bed and recent'
While this may seem a little clumsy, it seems to p r o d u c e acceptable results No o t h e r analysis [ a m familiar w i t h has
a n y t h i n g at all to say in detail a b o u t the s e m a n t i c s of these
C a s e S
A C K N O W L E D G E M E N T S
T h e analysis presented here has been c o m p l e t e l y imple-
m e n t e d w i t h i n a p a r s e r - i n t e r p r e t e r r u n n i n g in C a m b r i d g e This work was s u p p o r t e d u n d e r S E R C g r a n t s G R , C ' 7 9 l t4 and G R / D / 5 7 7 1 3 T h a n k s to Hiyan Alshawi A n n a b e l Cor- mack Mike G o r d o n Bob M o o r e and G r a h a m Russell
Trang 8R E F E R E N C E S
Bach, E 1980 In de/ense o/ Passive, Linguistics a n d
Philosophy, 3, 297-341
Bresnan, J 1982 The Passive in Lezical Theory in her
(ed.) The Mental R e p r e s e n t a t i o n of G r a m m a t i c a l Relations, Cambridge, Mass: MIT Press
Bresnan, J and Kaplan R 1982 Lezical Functional Gram-
mar, in Bresnan ed., op.cit
Davidson D 1980 The logical form of action sentences
reprinted in his Essays on Actions a n d Events Oxford: Clarendon Press, (originally appeared 1967)
Gazdar, G., Klein E Pullum G Sag, [ 1985 Gen- eralised P h r a s e S t r u c t u r e G r a m m a r , Oxford: Basil Blackwell
Karttunnen, L 1986 D-PATR: a development environ-
me,t for unification-based grammars, in Coling 86: Bonn,
AC L p74-80
Kilbury, J 1986 Category Cooecurrence Restrictions and
the Elimznat,on of Metarules, in Coling 86: Bona, ACL
p50-53
Russell, G., Pulman, S Ritchie, G., Black, A 1986 A
dictionary and morphological analyser for English in Col-
ing 86: Bonn, ACL, p277-279