+ All Categories
Home > Documents > ON GAPPING*

ON GAPPING*

Date post: 27-Mar-2022
Category:
Upload: others
View: 11 times
Download: 0 times
Share this document with a friend
18
The English Society of Japan NII-Electronic Library Service The EnglishSociety of Japan CONSTRAINTS ON GAPPING* Chie Matsui i, Introduction It is well-known that Gapping is a medial ellips!s phenomenon in coordinate clauses where the constituents are ellipted from the tight conjunct under identity with the corresponding parts of the left con- junct, and it tetains stressed contrastive elements with new informa- tion. This isour fundamentalunderstanding of Gapping. However, Gapping cannot occut undet certain conditions as exam- ples (i)-(s) show, though medial ellipsis occurs. Deltasrepresent el- lipted constituents, (i) a. Robert met Mary in ig8o and Tom *A(met Mary in) ig8s. b. John wants to sell a motorcycle and Bill *A(wants to) buy a car, c. Jack calls Joe Mike and Sam *A(calls Joe) Harry. d. John persuaded Mary to help Betty, and Tom *A(persuaded Mary) to help Susie. e, The boy told my girlfriend that he plays baseball, and A(the boy told) my roommate *A(that he plays) tennis. (z) Mary admired John on Monday and Betty scolded *A(John) on Tuesday. (3)Jack calls Joe Mike and Betty *A(calls) Sam Harry, In this paper I would like to discuss what constraints there are on Gapping to reveal the features of gapped clauses, comparing the clauses with left 1 right-peripheral reductions. ' * I would like to thank Roderick A, Jacobs, Shuji Chiba, Yasutomo Arai and anony- rnous SEL reviewers forhelpfu1 comTnents and fruitfu1 suggestions on earlier vetsions of this papet. All inadequacies ate my own. I am also gratefu1 to St. Mary Vivienne and Gregoty Peterson for checking all the exarnples and correcting stylisticerrors and to Sr. Maureen Brinket, St. Jo Ann and St. Jean Schmid for checking a part of theexamples. [S7]
Transcript
NII-Electronic Library Service
Chie Matsui
i, Introduction
It is well-known that Gapping is a medial ellips!s phenomenon in coordinate clauses where the constituents are ellipted from the tight
conjunct under identity with the corresponding parts of the left con-
junct, and it tetains stressed contrastive elements with new informa- tion. This is our fundamental understanding of Gapping.
However, Gapping cannot occut undet certain conditions as exam-
ples (i)-(s) show, though medial ellipsis occurs. Deltas represent el-
lipted constituents,
(i) a. Robert met Mary in ig8o and Tom *A(met
Mary in) ig8s.
b. John wants to sell a motorcycle and Bill *A(wants to) buy a car,
c. Jack calls Joe Mike and Sam *A(calls
Joe) Harry. d. John persuaded Mary to help Betty, and Tom
*A(persuaded
Mary) to help Susie.
e, The boy told my girlfriend that he plays baseball, and A(the
boy told) my roommate *A(that
he plays) tennis.
(z) Mary admired John on Monday and Betty scolded *A(John)
on
Tuesday.
(3) Jack calls Joe Mike and Betty *A(calls) Sam Harry,
In this paper I would like to discuss what constraints there are on
Gapping to reveal the features of gapped clauses, comparing the
clauses with left 1 right-peripheral reductions.
'
* I would like to thank Roderick A, Jacobs, Shuji Chiba, Yasutomo Arai and anony-
rnous SEL reviewers for helpfu1 comTnents and fruitfu1 suggestions on earlier vetsions of
this papet. All inadequacies ate my own. I am also gratefu1 to St. Mary Vivienne and Gregoty Peterson for checking all the exarnples and correcting stylistic errors and to Sr. Maureen Brinket, St. Jo Ann and St. Jean Schmid for checking a part of theexamples.
[S7]
NII-Electronic Library Service
The EnglishSociety ofJapan
s8 Chie Matsui
2. FunctionalUnitConstraint
It is often said that Gapping occurs only in parallel coordinate struc-
tures. For example, Kuno (rg76:3i6) proposes this as one of his four syntactic constraints. The word
` parallel
' is ambiguous. Most lin-
guists understand it as syntactic; i.e,, both conjuncts have the same
syntactic units in the same order as shown in (4).
(4) a. John ate an apple and Harry A(ate) an otange.
b. John ate an apple and Harry *b(ate)
at midnight.
However, identity of syntactic units does not seem to be a suMcient
condition on Gapping, because (s) meets the condition but it is un-
grammatical.
at home.
We will, thus, reexamine the notion ofparallel. Consider, first, the
sentences in (6-7).
John has studied here and Bill *A(has) played there. -
John has gone to New York and Bill ?*A(has) flown (over) -
to London.
John is studying and Bill ?*A(is) playing. u
John is studying in the room and Bill ?*A(is) playing in the u
open.i
(6-7) with the same syntactic units in both conjuncts are ungrammati-
cal with medial gaps. The perfective ` have+past participle
' or the
fbre, `
have '
or `
be '
cannot be ellipted in Gapping as in ((5--7). `
i Questions like (8a) or negatives Jike (ga) are grammatical because
au... eat fbrms a unit, and if only the verb eat is ellipted as in (8b) or (gb), the gapped clause is ungrammatical.
(8) a. Did John eat meat and A(did) Mary A(eat) fish? .-
t Five of the six informants say that (7a-b) are unacceptable, while one of them says that
they are acceptable. Even the informants considering (7a-b> unacceptable say that the
following is acceptable.
NII-Electronic Library Service
The EnglishSociety ofJapan
fish? k -
(g) a. John didn't eat meat nor/or (*and) Mary A(did (n't) eat)
fish.
b. John. didn)t eat meat nor / or Mary dttt t}:tfE!lse(e t) fish
Consider the fbl!owing along with (!a).
(io) a, John has studied in Canada and Maty b(has studied) in
England.
*A(in) England.
L-."--.-.-i
in England in (io) is a unit; thus, the preposition cannot be ellipted in
Gapping as in (iob). A similar explanation can be given fbr the `modlfier+Head' such as`very fast' as in (ii) or fbr the`Head+ modifier
'i such as
` the book on modern music' in (i2).
(iT) a. John spoke very slowly and Tom A(spoke) very fast.
b. ...and Tom A(spoke) *A(very)
fast -
(i2) a, John read the book on mountain climbing and Mary
A(read) the book on modern music.
b, , . . and Mary A(read) *A(the.eete&]Lel!-e2eek) on modern 7p7asiC
Here I will compare gapped clauses with Ieft-periphera! reductions.
Each of the coordinate clauses in the above contains the same syn-
tactic units, but this is not a necessary factor in left-peripheral reduc-
tlons.
(i 3) a. J ohn is ready and Tom *A(i2]/LyelEwaiting
b. John is ready and A(John is) waiting.
[[NpN] [vp[v?V] [A?Adj]]] and [[NpN] [vi5V [vpV]]] John is a student and Tom
*A(is) studying English now. (i4) a,
b. John is a student and zi(John is) studying English now.
[[NpN] [vp[vpV] [NpDet N]]] and [[N?N] [vp[v?V [viSV]] [NpN] [AdvpAdv]]]
t 1"ahen the modifier comes after the Head as in (i), it may be omitted because the mean-
ing does not change, though the premodifier cannot be ellipted as in (irb). .
A(on tt) . (i) John read the books on rnountain climbing and Tom A(read)
the atticles
NII-Electronic Library Service
The EnglishSociety ofJapan
'6o Chie Matsui
The ungrammaticality of(i3a) or (i4a) is due not only to the fact that iJ evaitiag or is sindyiag constitutes one unit and hence is cannot be el- lipted as stated above, but also to the fact that the right conjunct does not have the same syntactic units as the left conjunct, though the left- peripheral reduction is acceptable as in (i3b) or (i4b). Let us next compare gapped clauses with right-peripheral reductions.
(is) a. John wants to sell a mototcycle and Bill A(wants) to buy a car.
b, (= ib)...and Bill A(wants) 'A(to) bay a car. -
(i6) a. John bought a car, and I want to know how he managed
to a(buy a car).
b. . . . and I want to know how he managed *A(to buy a car).
[to [V'P]]
The right conjunct has the same syntactic units as the left conjunct in Gapping as in (isa), but not necessarily in right-peripheral teduction
as in (i6a). Why then is INF te fetained in the reduction? In gapped clauses INF te and bay (a car) cannot be separated as in (isa-b), while to is separated from VP bay a car in the right-peripheral reduction as in (i6a). The left element lo must be retained, which marks the ellipsis of tenseless VP; right-periphetal reduction has often been called VP deletion.
For Gapping the notion ` maximal ptojection of a category
' is em-
ployed; however, it cannot explain the unacceptability of such examples
as (5) Further, such examples as (r7)-(ig), where the right conjuncts
do not contain the same syntactic units as the left conjuncts, are ac- ceptable.
(z7)
(i8)
(i9)
a. The boy is vec)t dever, and the girl A(is) agenias.
b. John was strk and Mary A(was) in bad health. a. The woman left this moraix(g and the man A(left) at mtalaigbt.
b. John is playing eatsidla and Bill A(is playing} in the room. c. The boy walked slowtv and the girl A(walked) zvith great cure.
I want to begin te avrile a ,Pllz),,
and Mary A(wants) to finish readug a noveL
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety ofJapan
' is used functionally.
For example, adverbjals are subcategorized into such classes as Tirne Adv, e.g., at noon, and Place Adv, e.g., at home in (s) or Manner Adv, e.g., sloevly, with great care ifl (i8c). Thus the fbllowing is proposed:
(20)Functional Unit Constraint
Gapping is applied to functional unitsi and gapped clauses
must contain the same functional units as the left nongapped clauses.
Gapped clauses are in a striking contrast to left / right-peripheral re-
ductions. The fbrmer does not necessarily need the same syntactic
units but must contain the same functional units as the left nongapped clauses, while the latter has nothing to do with functional units. It may be conjectured that ellipsis on the basis of functional units is se-
verely restricted, whereas ellipsis on the basis of syntactic units is much freer.
This constraint also accounts fbr the grammaticality and ungram-
maticality of the fbllowing gapped clauses.
(21)
(Z2)
a. Jack calls Joe Mike and Sam Harry. b. (=Tc)...and Sam
'A(calls Joe) Harry.
c. . . . and A(Jack calls) Sam Harry. a. Mary believes John a genius and Tom an idiot.
i We assume the fo11owing phrase structure rules:
(i> S---+NP AUX V,,
(ii) vn-v, Advp
(c --. (;,,P ]) Functional units ate defined as fo11ows;
Subject (Subj) is NP ia (i). Predicate(Pred) is Vi. (o$iS2) Object (O) is NP in (iii). Complement (C) is C in (iii). Adverbial (Adv) is AdvP in (ii).
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety ofJapan
62 Chie Matsui
(2s) a. MaryconsidersJohnhappyandBillunhappy.
b. ...and Bill *A(considers John) unhappy. c, ...and A(Mary considers) Bill unhappy.
(24) a. I saw John running and Mary walking.
b. . . . and Mary *a(saw
John) walking, c. ...andA(I saw) Mary walking,
The gapped clauses in (zib)-(z4b) contain the same functional units as left conjuncts, but they are ungrammatical. The application of
Gapping to functional units also refers to the restriction of ellipted
elements: If more than two units ate ellipted, they must have the same function on the level of ellipsis. Functional units like ca" and
hel7leve are quite different from those like Jbe and ,Tbhn,
which makes
(2ib)-(z4b) ungrammatical.
Kuno (ig76: 3i4) proposes a Simplex-Sentential Relationship, as a
nonsyntactic constraint, saying that ` Two constituents left over by
Gapping are most readily interpretable as entering into a simplex-sen- tential relationship. The intelligibility of a gapped scntence declines drastically if there is no such relationship between the two constit--
uents '.
However, this constraint cannot account for the ungram--
maticality of the gapped clauses like (zTb)-(z4b) because the two con-
stituents retained do indeed enter into a simplex-sentential relationship,
Kuno says in fbotnote 28 (p. 3i6) that in `zo% believe the President
innocent', 2o% and innocent constitute a simplex-sentential relation-
ship.2
Sobin's Subject Condition, that is, ` Other factors being equal, a
t Kuno (ig76: so7) gives the fbllowing (ia) as acceptable.
(i) a. Ofthepeopiepolled,8o%believethePresidenttobeguilty,and2o%.(believe
the President) (to be) innocent.
b'. The boy believes Mary to be guilty and the girl *.(believes Mary) (to be) in- nocent,
b'L ...and A(the boy believes) the girl (to be) innocent. The examples like (ia) are problematic. As we notice that (ib') with the same structure
as (ia) is ungrammatical, (ia) is a special case because a special and distinctive word 2o%
is used in clear contrast to 8o%.
2 See footnote i. Kuno does not give an explicit definition of a simplex-sentential
relationship, but
he seems to assume such a relationship as (i)-(iii) in footnote r, p. 6i.
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety ofJapan
Constraints on Gapping 63
gapped structure is most acceptable when an NP (or N2) which is re- maindered by Gapping and which precedes the gap indicates a sub-
ject' (ig8o:ios) also cannot explain the ungrammaticality of the
gapped clause like. (2Tb)-(24b) because .S:am, 1-bmi, Birz] and Macy are
subjects in (2ib), (22b), (23b) and (24b), respectively. He takes only
the grammatical subject into consideration.
Without the help of these nonsyntactic constraints, the constraint
(2o) accounts fbt the ungrammaticality of (2ib)-(24b). The same ex-
p!anation wM be given for the ungrammaticality of the following
gapped clauses.
. , . and Tom *A(gave
John persuaded Mary to help Betty and Tom to help
Susie.
(=id) . . . and Tom *A(persuaded
Mary) to help Susie. . . . and A(John persuaded) Tom to help Susie.
John told Mary that Tom was guilty and Sally that Bill
was lnnocent.
Neijt (ig8i: 7g-gi) points out that the constraints on movement
rules can work on Gapping and that the Tensed-S Condition pro-
posed by Chomsky (ig77: 74) accounts for the ungrammatica!ity of
the gapped clause (28) and the Complex NP Constraint proposed by Ross (ig67; 7o) accounts for that of(2g).
'(28)
(29)
The boy told my girlfriend (t.hft;] he plays baseball and A(the
boy told) my roommate *A((lllhiitl,]
he plays) tennis.
John answered the question of which book Mary should buy
and Tom *A(answered the question) (ob which radio a(Mary
should buy).
iJUithout the
or (2g) by (2o) becauseon
transfbrmational rules, we can
NII-Electronic Library Service
The EnglishSociety ofJapan
64 Chie Matsui
he plays'in (28) or`answered the guestion'in (2g) do not constitute
one functional unit.
It might be claimed that two units are ellipted in (3oa), which is grammatical.
(3o) a. My father visited my uncle last night and my sister A(vis-
ited my uncle) this morning.
b. My father visited my uncle last night and my mother A(vis-
ited) my sister this morning.
The right conjunct of (3oa) and that of (2ib), for example, may be analyzed 2s in (3ia) and (sib), respectively.
(3i) a. [[my sister] [[[visited] [my unclej] [this morning]]]
b. [[Sam] [[calls] Uoe] [Harry]j] The ellipted elements
` visited my uncle
against `this
morning '.
If we compare (3oa) with (3ob), the func- tional leveli of the gapped clauses wil1 be understood. The example
(32) can also be accounted for if we look at the level ofgapped clauses.
(32) John tries to begin to write a novel, and Mary A(tries) ((to begin) to write) a play,2
The surface structure of (33a) is the same as that of(26a). But the gapped c lause in (33b) is acceptable, while that in (26b) is unacceptable. Why is (3sb) acceptable, though two units are emapted?
(3 s) a. John promised Mary to help Betty and Tom to help Susie. b, .. . and Tom A(promised Mary) to help Susie.
`
is acceptable and that the passive ` *Mary
was promised to help Susie '
is unacceptable, while in (z6b) `
*Tom
is unacceptable
p.t67.he functional
2 The
notion ` semi-auxiliaries
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety ofJapan
Constraints on Gapping
In the sentences in which the object of the infinitive mains and the other infinitive elements are ellipted, we
same result as fo11ows :
(34) a. JohnpersuadedMarytohelpBettyandTom*A(persuaded Mary to help) Susie.
b. . . . and A(John persuaded) Tom A(to help) Susie.
The constraint of (2o) applies not only to the gapped elements but also to the remnants. Each of the remnants must fbrm one unit.
(3s) a, The boy says that John plays baseball, and the girl A(says)
that Mary plays tennis.
b. The boy says that John plays baseball, and the girl hears that Mary
'A(plays) tennis.
c. The boy says that John plays baseball, and the girl A(says) that Mary ?*A(plays) tennis.
d. The boy says that John plays baseball and A(the boy says
that) Mary A(plays) tennis.
'
cannot be considered to be one functional unit and (ssb) is ungram- matical. Similarly,
` that Mary
' may not fbrm one unit and (ssc) is
ungrammatical, which is also accounted for in section 4. The sen-
tence like (ssd), where Gapping occurs within the embedded clause, is acceptable. We do not need the condition proposed by some lin;
guists like Neijt that the verb must be ellipted in the main clause or
the highest S,
To make the above discussion clear, I will revise the consttaint (zo) as follows:
(36) Functional Unit Constraint (revised) Gapping is applied to functional units and
a. gapped clauses must contain the same functional units
as left nongapped clauses
b. the ellipted elements or each of the remnants must form one functional unit at the level of the occurrence of
Gapping.
NII-Electronic Library Service
The EnglishSociety ofJapan
66 Chie Matsui
3. Tensed--Unit Elripsis Constraint
Since the English language is a so-called S V O language and Gap-
ping is a medial or internal ellipsis phenomenon, the first element of
a gapped clause is normally the subject of the clause and the gapped elements normally include the predicate. This fact is shown by the above acceptable examples. Kuno (rg76: 3i8) and many others have claimed that
` gapped elements must include matrix verbs
'. But this
cannot be correct as the matrix verbs appear in the gapped clauses of
(37), which are acceptable.
b,John may walk and Mary A(may) run.
The girl should clean the room and the boy A(should) wash
the dishes.
It is, therefore, clear that ` gapped elements must include the tensed
predicate'. The fo11owing discussion will explain why gapped ele-
ments must include the tensed predicate. Sobin (ig8i : 48g-go) points out that Adv / PP-first reductions need
not comply with Kuno's constraint, giving examples like (38) which
include the matrix verbs:
(38) a. On Tuesday Linda found the bow she had misplaced, and
on Wednesday A(Linda) lost it again.
b. On Tuesday Linda washed the car, and on Wednesday
A(Linda) hroke the car window.
The phenomenon in (38), subject ellipsis, is not the product of Gapping but a left-peripheral reduction, though the fight conjuncts of(38) look like gapped clauses, that is, they are pseudo-gapped clauses. This is- shown by the use of b"t because b"t is not possible in gapped clauses as Hudson (ig82:s48) points out, while it is possible in left/right- peripheral reductions, as in (3g).
(39) a.
b. c.
John drinks tea but Mary doesn't A(drink tea).
If bzat is used in sentences like (38) and the sentences with bat are ac- ceptable, (38a-b) must be left-peripheral reductions, not gapped clauses.
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety ofJapan
Constraints on Gapping 67
(4o) a. On Tuesday Linda fbund the bow she had misplaced, but
on VUednesday A(Linda) lost it again.
b. On Tuesday Linda washed the car, b"l on Wednesday
A(Linda) broke the car window.
From the acceptability of (4o), (4oa-b) are shown to be examples of
Ieft-peripheral reductions.
This is also shown by another fact that identical constituents repre-
sentjng old infbrmation cannot appear in gapped clauses, as in (4i), a!though they appear in left1right-peripheral reductions like (42). dihat is, the remnants of Gapping must express new infbrmation.
(4i) a. The boyi hit the dogl and the boyi/het *a(hit) the cat,
b. ... and the girl "A(hit) iti.
c. ...and the girl A(hit) the cat.
(42) a. The teacher admired the bay and gave him a book.
b. 1)be bay thinks he will win and he will.
The use of it in (s8a) implies that Adv 1 PP-first reductions such as
(38) are not the product of Gapping but left-peripheral reductions.
Are (4sa-b)' counterexamples to this constraint?
(4 s) a. Mary will cook the meals today and Betty may tomorrow.
b. The boy may belong to the club today and the girl will
tomorrow.
The answer is no, though they appeat to be gapped sentences, that is,
pseudo-gapped clauses. This is because in (43) but can be used as in (44a) and anaphora representing old information can appear as in
(44b), when (4sa) is examined.
(44) a. Mary will cook the meals today bitt Betty may tomorrow.
shei may and today b. Maryi will (surely) cook the meals
tomorrow.
where a tensed constituent like may always appears.
It might be claimed that adverbials on Il7leduesdy in (38) and temorL
t According to Goodall (ig87: 8i-4), (43 a-b) are gapped sentcnccs, but the application
of his rule makes them unacceptable.
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety ofJapan
68 Chie Matsui
roav .in
(43) are disregarded in spite of their existence. Since the right conJuncts in (s8a-b) and those in (4sa-b) have been shown to be lefr- peripheral reductions and right-peripheral reductions, respectively, it can be assumed that optional adverbials do not enter the process of ,ellipsis.i On this basis, the ellipsis in (s8) will be considered as initial and that in (43) as final rather than medial; that is, the fbrmer are left- peripheral reductions and the latter right-petipheral reductions.
It is clear that the right conjunct of (4sc), which is one of the two interpretations of (4sa), is a left-peripheral reduction.
(4s) a. My father visited my unele last night and my sister this
mormng.
b. (= soa) . . . and my sister A(visited my uncle) this morning.
c. . . . and A(my fathet visited) my sister this morning.
Then, is the tight conjunct of (4sb), the other interpretation of (4sa), a gapped clause or a case of right-peripheral reduction? According .to the above assumption, it is considered to be a right-peripheral re- duction because this monnii(g is an optional adverbial. However, it is quite diflerent from a right-peripheral reduction where a tensed con-
ptituent like neay in (43) always appears, because the tensed predicate is ellipted in (4sb), which complies with the constraint of Gapping
proposed above. Thus, though (4sb) may be consideted to be a right-
peripheral reduction on one hand and to be a gapped clause on the other hand, it is much closer to being a gapped clause because it mani- fests another characteristic of Gapping-it always expresses new infor- mation explicitly.
i When
the same optional adverb appears medially, it is ellipted wjth the verb as in (i). (i)
SimonquicklydroppedthegoldandJackA(quicklydropped)thediamonds. Uack--
endoff ig7i : 23)
However, when the adverbs, especlally manner adverbs, are quite diflkirent in the medial position, Gapping cannot occur as in (ii). (ii)
'Simon euic)eb dropped the gold, and Jack suzidenb the diamonds.
When they appear finally, they are more acceptable.
(iii) (P) Simon dtopped the gold quickly, and Jack the diamonds suddenly.
iJV'hen
diffetent time adverbs such as .freeuenily, semetimes and t!f]fen 2ppear medially as shown in (iv), Gapping can occur.
(iv) John frequently beats his dog and Bill sometimes his cat. The adverbs must belong to the same functional class.
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety of Japan
Constraints en Gapping 6g
So far I have argued not only against Kuno's constraint that ` gap-
ped elements must include matrix verbs ', pointing out the examples
(s7), but also against Sobin's claim that `Adv
1 PP-first reductions need
right-peripheralreductions. Here I will demonstrate the incorrect- ness of the constraint proposed by Kuno.
(46) (=s7) a. John may walk and Mary A(may) run.
b. The girl should clean the room and the boy
A(should) wash the dishes,
(47) The girl will clean the kitchen and the boy may
*A(clean)
the living room.
The gapped elements in (46) are not matrix verbs but modal auxiliaries.
In section 2 Gapping has been shown to apply to functional units,
and the perfective ` have+past participle
' and the progtessive
` be+
ptesent participle ' to be units, that is, the predicate. From (46), we
see that modals have different features from ` have
', a part of the per-
fective and ` be
', a part of the progressive, because modals can be
gapped elements. Let us therefbre call modals auxiliary predicates. The fbllowing example shows that the auxiliary predicate (AuxPred) and the predicate (Pred) are ellipted.
(48) The boy may read the book and the girl a(may read) the .
magazlne.
(4g) Tensed-Unit Ellipsis Constraint
Gapped elements must include a tensed unit, that is, the
tensed predicate.
Based on the fundamental properties of Gapping defined in the In-
troduction, the constraint implies that the tensed predicate which is
the same as the predicate in the left nongapped clause must be ellipted,
so that it does not allow the derivation of (soa), but allows that of
(sob).
fish.
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety ofJapan
7o Chie Matsui
The predicate fbrm of the right conjunct in (sob) is only teflected by the subject and shows the same tensed unit as that of the left conjunct, It also pxevents the derivation of(si) because the tensed unit, scokilea
appears in the gapped clause.
(si) (=2) Mary admired John on Monday and Betty scolded
*b(John)
b, Who bought the magazine, and who the book?
Though the tensed unit is ellipted in (s2a), it is unacceptable because it is not the medial ellipsis, while (s2b) is acceptable.
4. No-three-obligatory-unitConstraint
The units retained in the process of Gapping are `stressed
con-
trastive units ' expressing new information as shown in (4ic).
Kuno (ig76: 3i8) and Hudson (ig8z: s48) claim that only two con-
stituents are possible in the gapped clauses, However, consider (s3) as well as (3ob).
(s3) a. Mary played the piano on the shrge on Mbndlev and Betty
A(played) the violin in thepit en IPledueJziZy, .
(Sobin ig8i: 49o) b. The man went to Tblpo igv train lasl week and his daughter A(went) Xo Gaami byplaneptesberdLiL7. (s3a--b), with two or three optional adverbials, are acceptable. There- fore, more than two constituents are allowed in the gapped clauses. Examine (s4)-(s6)
(s4) a. Mary put the book on the desk, and John ?'A(put) the ball in the box.
b. ...and a(Mary put) the ball in the box. (ss) a, Betty treats the dog well and Bill ?*A(treats) the cat badly. b. ... and A(Betty treats) the cat badly. (s6) a. Mary chose Bill chairman and John *A(chose) Tom vice-
chairman.
The English Society of Japan
NII-Electronic Library Service
The EnglishSociety of Japan
Constraints on Gapping 7r
The acceptability of sentences (s4a) and (ssa) is very low because the
place adverbial in (s4) and the manner adverbial in (s s) are strictly sub-
categorized by the verbs pat and trea4 respectively. The three units
retained in (s4) and (ss) are, thcrefore, obligatory units, VCXhen three
obligatory units are retained in gapped clauses, the gapped construc-
tions are ungrammatical, which is also shown in (s6a) and (3). But why then is (46b) acceptable when the three obligatory units are
retained in the gapped clause? To explain the acceptability of (46b), we must consider the functional level of the gapped clauses. (46b) is
considered to have such a functional structure as ` Subj AuxPred
Pred." When AuxPred is ellipted, only two obligatory units, `
Subj
Pred'are retained. That is, when the AuxPred is ellipted, the level
on which it appears is considered, and similarly when the Pred is el-
lipted, the level on which it appears is considered. Therefore, it seems
clear that no three obligatory units are retained.
Further, consider the fbllowing sentences.
(s7) a. John gave a book to Mary and Tom A(gave) a notebook
to Betty.
*A(gave)
c, John gave a book and Tom A(gave) a notebook.
(s7a) provides the same infbrmation as (s7b), but the fbrmer is gram- matical, while the latter is not. This is because (s7a) has functional
structure ` Subj Pred O Adv
' where Adv is optional as (s7c), which
l:a.3C,?gAzb,is,S.h?ygk. `
,Els,%og.e.xer,gsgi'i&",,aLss,sl}gw.sdib,Rg.
`
2o.:g:g.l On the other hand, once`Betty'has the function of the indirect ob-
ject in (s7b), it becomes an obligatory unit. Therefbre, three obliga-
toty units ` Subi O O
' are retained in the right conjunct in (s7b), and
'(s7b) is ungrammatical.
The gapped clauses in (s8a) and (sga) are out because the remnants
are of three obligatory units.
(s8) a. John begged Sue to see Bill and Tom *a(begged)
Mary
NII-Electronic Library Service
The EnglishSociety ofJapan
';72 ,Chie Matsui
b. . . . and Tosn A(begged) Mary A(to see Bill).
(sg) a. The boy asked Mary to visit Bill and the giri *A(asked)
Sue to visit John. b. ,., and the girl A(asked) Sue A(to visit Bill).
Therefore, I will propose the fbllowing:
(6o) No-three-obligatory-unitConsttaint
This consttaint is cioseiy connected with the assumption that optfonal
adverbials do not enter the process of ellipsis as stated in section 3.
7. Conclusiolt
I have investigated what constraifit Gapping has. To cover a wider
scope of Gapping phenomena, I have proposed and shown three sim-
pie constraints on the fundamental understanding of Gapping as de- scribed in the !nttoduction, They are (i) Functional Unit Constrairrt,
which distinguishes between Gapping and left 1 right-pefipheral reduc--
tions 2nd restricts not only ellipted elements but a!so the semnants,
covering Kuno's nonsyntactic consttaint ` Simplex-Sentential Rela-
tiofiship ' and Sobin's Subject Condition, (ii) Tensed-Unit ffllipsis Con-
straint, which is vital fbr the grammaticality of gapped clauses and
discriminates between gapped clauses and pseudo-gapped clauses ( =a
kind of left / right-peripheral reduction that meets Fun ¢ tiofial Unit Constraint), and (iii) No-three-obligatory-unit Constraint, which is closely connected witk the assupapt,ion that optional adverbials do not
cnter the process of eMpsis.
It seems that the condition of recoverability and perceptual strate-
gies operate on gapped sentences, contributing to the dithculty of
processing them. This in tutn is reflected in these sentences as a (b) part of Functional Unit Constraint attd No-three-obligatory-unit Con- straint. The no-three-obligatory-unit constraint may, especially, be integrated into or be a part of the condition of recoverability.
I hope this paper will serve as a springboard for furthcr sndy on
Gapping.
NII-Electronic Library Service
University Press. Doctoral dissertation (!g84), University of California,
San Diego, '
Gazdar, G., G. K, Pullum, I. A. Sag, and T. Wasow. ig82. " Coordination
and Transformational Grarnmar." Lingn'stic lveur'c), i3, 663-76 Hankamer, J. ig73.
"Unacceptable Ambiguity." Liirgaisrk lvevriry s, i7--
68.
and Surface Anaphora." Liizguistic
"Incomplete Conjuncts." Lingn'srk lvevaic)r i3, s47-so.
JackendoffR. S. zg7i. " Gapping and Related Rules." Lingnt'stic ineva'cy 2,
2I-5j.
Kajita, M. ig68. A Generaime--T>'an{fbnvational StndJ of Semi-A")c:iliare'es in
Present-D`y,AmericanEagldrh. Tokyo: Sanseido. Koutsoudas, A. !g7o.
" Gapping, Conjunction Reduction, and Coordinate
Deletion." IFbandutibns ofLairguage 7, 337-86 Kuno, S. :g76.
"Gapping: a Functional Analysis." Lii{grvisrk IinptcJt 7,
3oo-I8. Levin, N. S. and E. F. Prince. ig86.
"
Maling, J.M. ig72. "On
Li,rgwatistic bvexiry 3, ioi-8
Neijt, A ig7g. GaLPvbiirg: a Contn'bntien to Sentence Gramnear. Dordrecht: Foris Publications.
. ig8i. "Gaps
Gapping." LinguistibAnabsis8,6g-g3
Quirk, R., S. Greenbaum, G. Leech, and J. Svartvik. ig8s. .`'{{ (bmpre-
hensive GrammaroftheEngmaLai({lutqge. London: Longman.
Ross J. R. ig67. (lonstrnints en Vlariables in .ijrntax. Doctotal dissertation, MIT,
Sag, I.A. ig76. Deletibn andLagi'cal Fonv. Doctoral dissertation, MIT.
Schachter, P. ig77. "
Constraints on Coordination." Laizgfftrge ss, 86-xo3.
Siegcl, M. E. A. ig84. " Gapping and Interpretation." Li)zg"istic ])7ezicr is,
523-30.
The English Society of Japan
NII-Electronic Library Service
" The Formulation of Gapping in English as Evidence
for Variable Types in Syntactic Transformations." Liagzistic Anaipsis i,
247-73
Stowell, T. :g8s. " Subjects across Categories." Zbe Li/rgwistic Recriew 2,
28s-3r2,
Webber B. L. ig78. A Fonval xt{ipProaab te Disco"rce Anapthera. Doctoral dissertation, Harvard University.
Received October is, ig87

Recommended