Pragmatics5 :3.34I-364.
InternationalPragmatics Association
THE INFERENTIAL CONSTRUCTION
Gerald P. Delahuntyr
1. Introduction
Thispaperanalysesthe textual and pragmaticpropertiesof a sentenceform which
has attracted little scholarly attention, presumably because of its infrequent
occunence.The following provides two examplesof the construction:
(1)
{t's not that he hates the pressthe way Nixon did, it's just that he is
insensitiveto the press'role in our society.(L.A.Times l2lI8l83)
The constructionis of interestfirst becauseit appearsto be universal;second
an analysisof its occurrenceand interpretation illustratessome of the
because
pertormed by participantsin discourse,the resourcesthat languageand
activities
languagesmake available for those activities, the principles that govern the
interactionsbetween language and context, and the ways in which speakers can
thoseresourcesfor narrativepurposes;and third, becauseits successful
manipulate
providessupport for the principlesand assumptionsused to accountfor it.
analysis
I will argue that the subordinate clause of the construction represents an
of its local discoursecontext,which in turn licensesthe derivation of
interpretation
a rangeof locally specific inferential connectionsbetween the context and the
clause.Thesecharacteristics
can be accountedfor by Grice'scooperativeprinciple
(CP)and maximsof conversation.
I havebeen looking for examplesof this sentencetype for severalyears,but
haveso far collectedonly 85. The infrequencyof the constructionis matchedby the
with which it is mentionedby linguists.I know of only sevenreferences
infrequency
to it, and in three of these the constructionis mentionedmerely as an appendixto
thediscussion
of another sentenceform. Delahunty (1982)mentionsit as a type of
cleftsentence,a position endorsedby Declerck (1988, 1992),(but seeCollins 199I);
Quirk et al. (1985) mention it in a footnote to their discussionof obligatory
constructions;Kuno (1973) discussesthe Japaneseanalogue,,the no
extrapositive
I My thanksare due to Jim Garvey,
Dick Kitchener,Karl Krahnke,David Lindstrom,
ClnthiaMesh-Ferguson,
and Marna Shillmanfor discussionon the topicsof this paper.The
errorswhich remainare, of course,due to my not taking their good advice.A preliminary
versionof this article appearedin the KansasWorking Papersin Linguistics,Vol. 15,No. 1,
1990,
underthe title 'Inferentials:The story of a forgottenevidential.'
342
Gerald P. Delahunty
desu construction.z
The form had no generallyused name, and I eventuallysettled on inferential.
The reason for this name will, I trust, becomeclear as we proceed.
The presentstudyis basedon a corpusof instancesof the constructionalong
with as much prior and followingcontextas seemsrelevant.The instancesare taken
from a range of text types.Most come from written sources,both fiction and nonfiction, literary, scholarly, and journalistic; a few were spoken, in both radio
editorials and informal conversation.Although the current work focuseson the
English inferential construction, the construction exists in a number of other
languagesand may be universal.3
I begin by providing a brief grammatical overview of the inferential
construction,describinghow it can be modified by negation,modals,adverbs,and
complementizers.Then, I describe the distribution of the constructionin texts.
Third, I describethe construction'svarious contextualizedinterpretations.Fourth,
I briefly discuss negative inferentials. Finally, I relate the interpretation of
inferentialsto Grice's notion of implicature.
2. Grammatical overview of the inferential construction
Inferentials, in English, are sentencesin which a tensed subordinate clause is
embedded as the complement of a form of.beawhose subject is expletive ir. I will
refer to the embedded clause of an inferential, correspondingto that he hates the
pressthe way Nixon did and that he is insensitiveto the press'role in our societyin (1),
as "the clause,"and the part to which the clauseis subordinateas "the matrix,"
correspondingto it's (rtot) (just) in (1).
" According to the October 1988 t,SA Bulletin, Karina Wilkinson was scheduled to read a
'The 'It is
paper on
not that ...' C-onstruction.'Unfortunately I have been unable to obtain a
copy of this.
3
Th" conrtruction occurs in at least the following languages:Finnish, French, German,
Modern Hebrew, Hungarian, Irish, Italian, Japanese,Korean, Mandarin, Portuguese, Russian,
and Spanish. For discussion of the construction in these languages my thanks are due to Paola
Malpezzi-Price (French, Italian), Roger Gilmore and laura Gatzkiewicz (Spanish), Jon and
Barbara Thiem, Roland Wolff, and Ron Walker (German), Tom Mark (Hungarian),7,ev Bar-lrv
(Hebrew), David Argoff and Kaija Wilson (Finnish), Mrs. Gaughran (lrish), Emanuela Mengozzi,
Flaminia Cervesi, and David McCobb (Italian), Charles Miracle (Mandarin Chinese), and W.
Mary Kim (Korean, Japanese). It appears that we are not dealing with just an areal feature of
Western European languages and others, like Hebrew, which are heavily influenced by them.
a
Declerck (1992) claims that the be of inferentials and clefts is "specificational' in that it
specifies a value for a variable. The inferential variable is derived pragmatically from the context
and the clause represents its value. Thus a specificational sentence The bank robber is John
Thomas answers the question Who is the bank robber (Declerck 1988: 6). However the analogous
discourse created from inferentials is ill-formed: Wat isn't it only It isnl onty that I'm nosE. ln
fact wh-questions created by questioning the clause of an inferential seem to allow only an echo
interpretation. It appears, therefore, that it is insufficient simply to classify inferentials as
specificational sentences.
The inferential construction
343
That the il subject of the inferential matrix is expletive, that is, nonreferentialand devoitJof semanticimport, is easilydemonstrated. It cannot
be
questioned
(2), or replacedby a referring expressicin
(3):
(2)
(3)
*What is not only
that I am nosey?
*That/thisis nor
only thar I am nosey.
From a linguisticuniversalspoint of view, it is reasonableto predict that any
.
languagewhich allows either expletive or zero subjects,has a c'opula,
and has
subordinate
tensedclauseswill possessthe inferential construction.Moregver, if
pragmaticprinciplesof interpretationsuch as Grice's (1975) Cp an<J
maxims and
Sperberand Wilson's(1986)principleof relevanceare indeeduniversal,then
the
construction
will havevery similar interpretationsin all languagesthat have it. And
thisis in fact the case.
The constructionallows a number of elaborationsin the matrix that
are
textuallyand pragmaticallysigniticant.
The entire inferentialstructurcmay be (and in actual use,very otien
is)
modifiedby adverbssuch asperhaps:
(4)
Perhaps,it is that women in Irelanclare nor a firrm of prayer?
(Gogarty1968:59)
The matrixmay include adverbssuch as only, just, andsimply:
(5)
It wasjust that it was raining.(lrving l97g: 2j3_4)
Theseadverbscanbe negated(6), and the complementizer
thnt is optional(7) and
(10):
(6)
It isn't only that ['m nosey.(p. cowelr e-mail message4lrglg3)
(7)
"Oh,.it's_I-.!
not pretty enough."(Donohuetranscriptno. 03120,cited
in Kies 1988)
The matrix may also include modals (only may, coukl, ancl migttt occur
in my
corpus)and negation:
(8)
It may be that he lackssomefirrmsof imagination.(Wilder 19g7:174)
(9)
It is not that one fears treachery. (Murdoch 1975: 43)
(10)
It couldn'tbe he'd be goin'in it agin.(sic)(Somervilleand Ross1977:
264)
Modals and negationin int'erentialsmay occur in either or both the
matrix
andthe clause,making possiblescoperelationsand interpretationsdistinct
from
thosein simplesentences:
I
344
Gerald P. Delohurus'*
( 11)
a. I t ma y b e th a t I w i l l h a v eta s k sfb r you as cruelas those the Great
Mother has laid on me. (Bradley 1982:136)
b. *l may will have tasksfor you as cruel as those the Great Mother
has laid on me.
The constructionallowsthe omissionof redundantmatrix modals.etc. in a
text in which a seriesof clausesare coordinated,as in:
(12)
Conceivablysome other factor, not well accountedfor in the model,
is delaying or counteractingthe warming. It might be that tlrc heat
capacity of the oceens is laryer than cunent models calculate, that the
sun's oLttputhas declinedslightlyor that volcanoeshave injected more
dust irtto the sffalospherethan is cunently known, therebyreducing the
solar enetg)reachingrhe ground.(Schneider1989:76)
The matrix It might be in (12) is fbllowed by three coordinatedclauses,each of
which is in the scopeof the modalmight,which appearsonly once in the sentence.
If the infbrmationin the clauseswere expressed
in a forrnother than the inferential,
the modal would have to be repeatedin eachclause.The inferentialstructurethus
allowsan elegantand parsimoniousparallelism.
Relatedly,it allowsmodals,negation,and adverbsto be positionedso that
the entire clauseis within their scope,includingany negationor modalswithin the
clause.Consider:
(13)
a. [t] is not that what is deniedmust first havebeen asserted.(Horn
1989: 47)
b. What is deniedmust not first have been asserted.
In (13a) the negatortakesscopeover the modal:in the normalizedfrlrm, (13b),the
reverseoccurs,with consequentditferencesin interpretation.
In (la) the inferentialmatrix providesthe sole positionin which only can
modify the clauseand carry the intendedcontrastwith extraordinary:
(14)
Caesarunderstoodwell the propitiatorynature of sacrificeamongst
the Gauls,but of coursethere wasnothingextraordinary
in this custom
in Gaul. or in the wider Celticdomain;it wasonly that the Celts had
retainedarchaicpracticesonce also at home in ltaly, as in Greece,
but now long outmoded.(Powell 19tt3:180)
Of my 85 examples,23 are positiveand unmodalized(two of these are
questions);12 are positiveand modalized;and .50are negative.Five of my examples
are modified by external adverbs:Two each by perhapsand if, and one by thus.
Twenty examplesare internallymoditiedby adverbs:Five by ortly,,ninebyjust, three
by simply, and three by (rtot) so muclt.
The inferential constntction
345
3. The distribution of inferentials
I begindiscussion
of the functionsof inferentialsby examiningtheir distributionsin
texts.Negativeinferentialshave the clearestdistributionalpatterns,so I start with
them.
The textualpatternsin which negativeinferentialsoccur can be describedin
the following taxonomy. The most striking pattern is what I call a "tandem
inferential,"
consistingof a negativeinferentialfollowed by and contrastingwith a
positiveone:
(15) It was not that I felt we had to avoid the Winters at that moment: it
wasjust that it was raining. (lrving 1978:21314)
The secondpattern might be called an "embeddedtandem," becauseit
contains
a singlematrix with (at least) two embeddedclauses,which contrastwith
eachotheron somedimension:
(16)
U]t is not that what is deniedmust first have been asserted,or that
positive facts are more real or more basic than negative ones, but
simply that knowledge of a positive fact counts for more than
knowledgeof a negativeone. (Horn 1989:47)
In thisexamplethe first two clausesare within the scopeof the matrix negation;the
thirdclauseis outsideof the scopeof the negation,but nonetheless
is a complement
of rl rs.
The third pattern consistsof a negativeinferential followed by a positive,
non-inferential
sentenceintroducedby a contrastiveconjunction,typicallybut;
(17)
It is not that mamma caresabout it least in the world, but I know it
is taken notice of by many persons.(Austen 1980[1818]: a7)
The two parts of this type of inferential sequencemay be separatedfrom
eachother by considerabledistances,as illustratedby the two italicizedsentences
in:
(18)
"It is rtot that I would fear discov€ry,"says he interrupting. "I have
never done any seriousdisserviceto either one - at leastnothing they
would be likely to know of. And even if they should have cause to
mistrustme, I am Sir William Cecil'sman, and he has been a patron
to them both. But - yott should wtderstandthiswell enoughand without
knowirtgtoo much - tlrcre n'ould be qttestiorr. There would be, if only
for the sakeof friendshipand good manners,a delay I cannot afford.
(Garrett 1983:173)
The final pattern consistsof a negativeinferentialfollowed immediatelyby
a positivesentence:
(19)
It is not that the model is wrong; there is just not enoughevidence
i r i o i ) i t r t i i { r , L 1 ; i : r . r . i i ti t: .' ( l V l a i i t r i - ri 9 8 9 : 2 5 3 )
I n c a c h o f t h c s c c a s c s .i l r c r i c g . r t i v ei n f ' c r c n t i a ld e n i e ss o m c p r o p o s i t i c l nr e l e v a n t t o
t h c c o n t c r t . w ' h i l e t i r c i i r l k ; r v i n qs c r l t c n c ee l s s c r t a
s n a l t c r n a t i v ep r o p o s i t i o n w i t h t h e
s : l n i c r c l i i t r t l i l : h i Pt t r t i t c t t ) r r i l i \ 1 .
P o s i t i v c i n f c r e n t i i t l so c c u r i n m t i r e r , ' a r i e dc i r c u m s t a n c e sM
. any occur as the
sccond of two contnrstccl L.ropositions.Sclrne of these occur as second parts of
t a n C e m i n f c r e n t i u l s .a s i n ( l 5 ) . O t h e r s f o l l o w n c g a t i v e ,n o n - i n f e r e n t i a lc o n s t r u c t i c l n s ,
c l f w h i c h i h e f o l l o w i n g , , r x u r r i i r lics p a r t i c u l a r l v i n t e r e s t i n g :
1 2 ( t . 1 I i c m c n r t r e r t h a t a s a l i t t l c m a i d c n w h e n I w e r su n h a p p y , w h e n I w a s
c h i d d c n t r v m y m o t h c r o r m y t c a c h e r s ,o r ( a ) w h e n U t h c r - r e t r e l yn o t i c c d n t c t o , . l i s a p p r o v eo f m c , [ . . . ] .
(tr) Not ririrt Uther w'irsevcr unkind to me; (c) il was simply that he
iirtl tt,r yutrtitii!ur iiitert.st irt u girl cltild. My mother was always at the
c c n t c r i l l i i i s i . t c l r r t ,( B r i r r l l c v l ! 1 . 3 2 1: 0 8 )
I h i r v c m u r k c d t h e p a r t s o f t h i s s c q u e n c ea s ( a , b , c ) ,w h e r e ( a ) i s t h e c o n t e x t t h a t
triggcrs thc denial or negation; (b) rs the negativeconstruction;and (c) is the
p o s i t i v c i n t ' c r c n ti a l .
I n ( b ) n o t e t h e p o s i t i o n o f r u t t a n d t h e t b r m o f t h e c l a u s ei n i t s s c o p e . ' T h e
clausc is subordinate in f'ornr tlut there is no overt matrix ciause in which it is
enrbcddcd and substituting It vt,usrutt thut Utlrcr wos ever uttkirul to me for (b) does
n o t a l t e r t h e t c : x t ' sw e l l - l i r r m c d n e s so r i t s i n t e r p r e t a t i o n . "
Sornc posrtive inferentials frllkrw assertions ior which they provide
r e f o r m u l a t i o n s o r r c i n t e r p r e t a t i o n s( a s I d i s c u s si n m o r e d e t a i l b e l o w ) :
(21)
Hc hiid qot past the stage ot reason, even his power of mocking at
himself wirs de ad. or perlnp.s it w,as thot there seemed no longer
utt\'thiilc tltut c'ouldhe ntockad at. (Somerville and Ross 1917: 209110)
'fhese
s h o u l d n o t b c t a k c n a s a n c x h a u s t i v el i s t o f t h e c o n t e x t sw h i c h p r e c c d e
p o s i t i v e i n f e r c n t i a l s . I t i s n o t p o s s i b l e t o p r o v i d e a c o m p l e t e c h a r a c t e r i z a t i o no f
thosc contcxts. l'his is bcc:au,;cslrciikers mzry interpolate an inferential wherever
t h c y l ' c e l t } : c n e c c lt o e x p l i n d u p o n w h a t t h c y h a v ej u s t s a i d - e i t h e r t o e x p l a i n i t , o r
r e i r t t c r ; r r e ti t . o r d c n y l r n i n l ' c r c n c cd c r - i v a h l ef r o m i t . " l ' h e t e x t u a l p o s i t i o i r sa t w h i c h
-'Htrngarian
n , : g a l i v t i n f e r c r l t i r r l sc u n h c o l t h i s l r ) r r i l - n c g , a t o rf o l k ; $ c c l - . y ' l ts u b o r d i n a l c
c l a u s c- r v i l h n o c o p u l l ro r c x p l c t i v ci i r l h c m l t r i x ,
6
D c c l c r c k ( i 9 q : ) a r [ . u c st h a t t h c l i t t t t l t a t c o n s l r u c t i o ni s n r l t i n h c r e n t l v : , l l c c : i i i c a l i o n a i ,
a l t h o u . q hh c c o n e c d c st i l r l t i t c r u i h c u s c ds p c c i f i c r t l ( ) n a i -i !a s i n ( 2 ( ) )- ' i f t h c c o n t c : <cl t ; n r a i n sa
r a r i r t h l c . " l ' { s d r r c sn o t i d c t t i i t _ ivh c v u r i a o i c i t i l : , i 1 9 1f1r ; r i h c s p c c i f i e a t i o n : trlc a t l i n g o f t l i c r t o l t h u t
- \ o n ' c sL o c l a r i l v t h c r c l c v a n c co f t h c
c o n s t r u f t i o n i n 1 2 t ) ) .h u t c l i l i r n st h a t t i r e i . ' ( ) n 5 t r u { . t i o" n
p r c c c d i n q s p c c e ha c 1 "( i n ( ) ; . ( i i r , c n t h c p r o h l c n s r n h c r r l r l i n a p p h ' i n t s p c c c ha c t t h c o r y t o t c x t s
s u c h l t s t i t o s t rl ' r r ; mu ' t i i c h i i r i s C l r i at s d r a l v n ,t r ' u c o u l i l r c l s o n l h l v s u h s t i t u t c" p i c c c o l t e x t ' f r t r
" s p c c c ha c t . " T h i s t t l r o t . t ! s Li s r r - . ! 1r ' i o : l cl t t h o * ' I , i t t t c h i m t t r q t h a t t h c i n f c r e r i t i a iw o r k s . I d o
a q r c t * ' i t h i i c c i c r c ' k l h : r i n ( ) / r l i r i i r i i i r l t l i . t n t t l r l , , r /r , r n s t r u c t i i i n sa r c n o t a l w a y si n t c r c h a n g t ' l b l e .
'fite
inlcrentiol consmtctton
347
' l ' hey
th esp eak er
will f eel t h e s eu rg e sa rc u n p re d i c ta b l e .
are not di sti ngr-ri shabl
ase
"contain[ing]
a variable"as Declerck(1992)cliiims.In fact.texts"contrrin"indefinite
numbersof potential "variables."However, manv of thesc sequenccsmay be
analysed
as composcdof three parts. The first part consistsof a context which
promptsa denial from the speakcr:7the second (which may be ?Lnegative
inferential)
rejectsa propositionas either not true or not locallyrelevant;and the
third(positiveinferential)introducesa propositionwhich contrastswith the second
in beingpresentedby the speakeras true or relcvant,or providesthe narrator's
reasonsfor rejecting the proposition in the second part. In all cilses the
interpretation
of the inf-erential
requiresreferenccto the local context.
4. Interpretationsof the inferential construction
In this sectionI consider the range of meaningsdisplavedtry thc inferential
constructit-ln
in contextand demonstratethat th';:intr:ny't'iutionof un inferentialin
co n te xist indet er m in a teI .c l a i m th a t th c ra n g eo i mr:.ri i i ngs
rcl )rescnts;'natural
r
class.I end this section bv consideringthe roles th:rt the inf'erentialplays in
discourse
inferencinq.
4.1.Possibility
and retlection
Quirket al. (19f15:1392)in the footnotc in which tlic-!'rctc:rto it, claim that the
co n str uc t ion
irnd(especi al l v)ti l r retl ecti ve
m aybe' us c dfo r e x p rc s s i o nosf p trs s i h i l rtrr
questions.'
Their (constructed)examplesare:
(22)
a.
b.
It may be that she no longer irusts vou.
Could it bc that you lrl't th'.' k,'..'sit: vnur oil'icc'.)
N o t i c e t h a t h o t h o f t h c s e c o n t a i n m o d e r l si r r r r it i r ; r t t h e s e c o n r l i s i r c l u e s t i o n .
In contrast,the two intererrtials in my corpus whosrr rlirtrix cl;iuses rlre positive,
declarative
a ,n d c o n t a i n n e i t h e r m o d a l s n r l r a d v c r b sd o i r o t r n d i c a t c p o s s i o i l i t y .T h e
i n f e r e n t i ailn ( 2 3 ) m a k e s a c a t e s o r i c a la s s e r t i o n :
(23)
A p r o b l e m l i k e t h i s g e n t l e m a n t l i k e , , j ; r b t r u ts o o p e n l y m i g h t b e j u s t
a normal pattern for him. and thzrt coullle would fecl better if that
w o m a n k n e w i t w a s h i s n o r m a l p a t t r - ' r r rs,h c n : i g h t f i n d i t m u c h e a s i e r
'Olt, lr
to accept than if she th<luqht.
it
I'm nol prett)' eruntgh.'
( D o n o h u e t r a n s c r i p t n o . 0 3 1 2 0 ,c i t c d i n K i e s 1 ! ) E 8 )
Example(2t)c)"which contains an advcrb, asserts thut LJther Pr:ndragcln hild ncl
particularinterest in a girl child. These examplcs su[qcst that Quirk ct al.'s choclsing
modalizedforms misled them and that the tornr itselt tlocs not exprc'sspossibility.
'
'hearer'
The terms 'spcaker' and
are intcndcd to incluilc writers and rcadcrs as
appropriate.
348
Gerald P. Delahunty
the form is quite compatiblewith the expressionof possibility,
Nonetheless,
as the modalizedfbrm in Qa) shows:
even in declaratives,
(24)
It may be that you have receivedreportof her deathfrom other sources.
It may also be true - and we pray that it will be so - that by the time
this letter is safe in your hands,her servantswill have been set free.
( G ar rc tt l 9 tt3 : 1 4 0 "
As I ncltedearlier, an inferentialmay be modified by an adverb such as
perhups,thus providinganothermeansof indicatingpossibility,as in (21) and:
(25)
'He groanswhen a really good-lookinggirl meets him. The prettier
the worse it takeshim. Sometimeshe's damnedrude.'
'Perhapsit is that women in lreland are not a
form of prayer?' (Gogarty
1968:58-9)
Reflection is distinct from possibility.It is the process of attempting to
interpret some phenomenon.Ref'lectingmay involveaskingquestions,as in:
(26)
Morgausesaton the t1oor,leaningher headagainstViviane'slap, and
Igraine saw that the sulky eyeswere filled with tears.She has us all
in her lnnd. How cctrtshe havesttchpower over us allT Or is it that slte
is the only motlrcrMorgausehaseverknown?She was a grawn woma,t
wlrcn Motgause was bont, she has alwaysbeen motlter, as well as sister,
to both of us.Their mother, who had been too old for childbearing,
had died giving birth to Morgause.(Bradley 1982:10-11.Italics in
original;emphasisadded.)
It can also involve drawing inferences,either categorically,as in (23), or
tentatively,as in (25).The inferentialconstructionmay representan interpretation,
reflection, or conjecture about an issue relevant to the local context. Its mood,
modals,or adverbsindicatethe degreeof certaintywith which the interpretationis
proposed.
4.l. | . ReinterpretationsIreformula tiorts
The inferential clausemay also representa reinterpretationor reformulation of an
immediately precedingpiece of text, more or lesstentativelyproposed.
8
Declerck (1992: 2}7ft) claims that the sentenceitalicize<lin Q\ is not an inferential but
'the extraposed version of that you hav'e receivedreport ... nny be
ItrueJi this appears, amongst
othcr things, from the fact that the next sentencebegins with il nny also be tnte that ...' It seems
to me that replacing the inferential with Declerck's paraphrase,extraposedor not, subtly but
significantly, changesthe meaning (and well-formedness)of the text. I have checked this intuition
with several colleagues(all native speakersof English, though not all linguists). All agreed that
the inferential and ils tnrc congener are not substitutablc in this contexl.
The inferential construclion
(21)
349
He had got past the stageof reason,even his power of mocking at
himself was dead, or perhaps it was that there seemedrto longer
anythingthat could be mockedut. In spite of his knowledge of the
world the position had an aspectthat was so seriousand beautiful as
to overpowerthe others,and to becomeone of the mysteriesof life
into which he had thought himselftoo cheap and shallowto enter.
(Somervilleand Ross 1977:209110)
In this instancethe inferential may be seen as an alternativeformulation of the
character'sstate of mind. Moreover, the narrator, by using an inferential and
introducingthe inferentialwith perhaps,indicatesa limitation on her understanding
of the characterwhose emotionsshe is describinq.
4.2.ExplanationsIaccounts
Perhapsthe most frequent use of the inferential is to suggestan explanationfor
This interpretationoccursamongstall
are under discussion.
whatevercircumstances
threetypes,positiveunmodalized,modalized,and negativeinferentials.
The inferentialin (23) is an explanationfor the couple'ssexualdifficulties.
'why European
The inferentialin (25) is a tentativelyproposed explanationtbr
womenare utterly without interest for' the author's friend: The autobiographer,
OliverGogarty,is discussinghis friend Mcloren's sexualdifficultieswith one of his
'informants.'Mcloren as a young man stationedwith the British army in India
apparentlyhad sexualrelationswith a sacredtemple courtesan('a form of prayer'
in traditional Indian culture), which were of such intensity and duration that
'European
womenare utterlywithout interest'forhim. The inferentialinvitesus to
infer that Mcloren's sexualproblemsare to be explainedby the assumptionthat
Irishwomenare not a form of prayer.The adverbperhapsand the question mark
indicatethat the inferenceis tentativeand reflective.
The negative inferential in (28) rejects I did not thirtk of it or desire it as
for Wentworth'snot writing to Anne Elliot:
explanations
(28)
'Tell
me if', when I returned to England in the year eight, with a few
thousand pounds, and was posted into the l-aconia, if I had then
written to you, would you have answeredmy letter? would you, in
short, have renewedthe engagementthen?'
'Would I!'was all her answer;but the accentwas decisiveenough.
'Good God!'he cried, 'you wouldt It is ttot that I did nat tltirtk of it, or
But I was proud,
desireit, as what wouldakne crownmy other success.
too proud to ask again.'(Austen 1980:233)
That the inf'erentialcan conveyan explanatoryinterenceis confirmed by the
factthat when the sentencebeginningThe explarntionwas. . . in (29a) is replaced
by an inferential,as in (29b), the two versionsof the text are well-tbrmed and
synonymous.
350
GeraldP. Delahunty
(ze)a.In
England a new ruling element had displaced much of the existing
landholding class. but had been content to divert to itself the labour
and produce of the indigenous rural population. When the Normans
had expanded into Wales, however, they had established new
'Englishries',
communities, which came to be known as
in the valley
floors, while confining the Welsh inhabitants, by and large, to the
uplands. The explanatiotr was, not tlnt tlre Normarts were more hostile
towards tlrc Wel.slt than lownrds tlte Atryk> Saxorts, but rather that
Welslt rural society, witlt its pastorul emphasis, was ttot geared to the
satisfactory workirtg of arable land. (Frame 1981: 77)
(2e)b.It was not that the Normans were more hostile towards the Welsh
than towards the Anglo-Saxons, [...].
The ditt'erencebetweenthe versionsis that in the original(29a) it is made
explicit that the propositionsin the italicizedsubordinateclausesare to be
interpretedas an explanationfor the differences
betweenthe Norman colonizations
of England and Wales.This interpretationis lefl implicit in the adaptedversion
(2eb).
The claim that the clauseof an inferentialsentencemay be interpretedas
an explanationtor some situationor event is further supportedby Kuno's account
of the Japaneseno desu construction.He claims that rto da (informal), no desu
(polite), and tto de aru (formal writing) can be "roughly translatedas 'it is that"'
(1973: 223). Desu and its variants are copulasand no is a nominalizingparticle.
Kuno (1973:223, no. 1) presentsthe followingpair of exampleswithout and with
no desu:
( 30)
a.
b.
Kaze o hiita
cold drew
'l've
caught a cold'
Kaze o hiita no desu
'(lt)
is that I have caught a cold'
Kuno translatesthe majority of his examplesof the no desu constructionas 'The
explanationfbr X is that S.' (Althoughin an earlier section(p.202) he translates
th e m as ' I t is t he ca s eth a t S.' ) F o r e x a mp l e :
(31)
Kaze o hikimasita. Ame ni
hurarete
nureta no desu
cold drew
rain
fallen-being got-wet
'I have caught cold. (Lit.) The
a
explanationfor my havingcaughta
cold is that I was rainedon and drenched.'(Kuno 1973:221,,no. 3a)
Kuno discusses
interrogativeforms of the !o desuconstructionand translatesthem
as requestsfor the hearer'sexplanationof some situation.Ka is an interrogative
particle.
( 32)
K aoiro
ga warui desune. Byooki na no desu ka?
complexion bad
sick
'You don't look well. Is the
explanationfor your not looking well that
The inferential constntction
351
you are sick?' (Kuno 7973:225, no. 4a)
In a number of his examples Kuno adds 'or evidence'to the gloss of the
construction.
For example:
(33)
Byooki desu. Taizyuu
ga ryuppondo
hetta no desu
sick
weight
ten-pounds
lessened
'I am sick. The explanation(or evidence)
for my being sick is that I
have lost ten pounds.'(Kuno 1973:226, no. 6a)
Someof theseJapaneseexamplescan be felicitouslytranslatedinto English
usingthe inferentialconstruction.For example:
(34)
(35)
(36)
You don't look well. Is it that you are ill?
I caughta cold. It's that I was caughtin the rain and drenched.
?I"am ill. It is that I have lost ten pounds.
The Japaneser?odesu constructionseemsto have a somewhatdifferent and
largerrange of interpretationsthan the English inferential. It may be that the
Japanese
constructionhas conventionalizedthe interpretation of explanation or
evidence.
This is consistentwith the tact that it contrastsdirectlywith the kara destt
construction,
which seemsto havethe conventionalinterpretationof "it is because..."
For example:
(37)
Kaze o hikimasita. Ame ni hurarete nureta kara desu.
cold drew
rain by fallen-being wet-got
'I have caughta
cold. It is becauseI was rained on and drenched.'
(Kuno 1973:226,,
no. 5)
Further,the no desaconstructioncan be usedeither to provide an explanation(or
evidence)
for what has been said or for what has been observed.In other words it
doesnot requirea prior linguisticcontext.The interrogativeform particularlyseems
to be able to occur without one, for example:
(38)
Dokoka
e iku no desu ka?
somewhere go
'(l see that you
are preparingto go out.) Is the explanationfor this
that you are going somewhere?'(Kuno 1973:225, no. 4c)
It is unlikelythat an Englishspeaker,upon seeingsomeonepreparingto go out,
would say 'ls it that you are going somewhere?'rather than 'Are you going
somewhere?'
or 'Where are you going?' In this context the no desu construction
marksthe informationin the clauseas an interpretationof its contextand thus may
providean indirect way of asking where the addresseeis going or why they are
leaving.Indeed,Kuno (1973:233)suggests
that "[i]n the immediateenvironmentin
whichthe speaker has made some observation,questionsabout the observation
withoutusing the no desu constructionare often out of place. For example, in
talkingto a girl who is crying the (a) sentencesare out of place:
352
Gerald P. Delahunty
(3e) a.
b.
*Naite-imasuka?
Are you crying?
*Naze naite-imasuka?
Why are you crying?
Naite-iru no desuka?
Is it that you are crying?
Naze naite-iru no desu ka?
Why is it that you are crying?"
Kuno offers no explanationfor why the (a) questionsare out of place. Perhapsit
is becausethey do not mark the girl's cryingas an interpretation,unlike the two (b)
questions,which do, thereby adding a degree of indirectnessand politeness.Or
perhaps the no desu construction has grammaticalized the explanation
interpretation.
'Is it that you are crying?'is infelicitousin English in
Notice of course that
this context,presumablybecauseEnglishspeakersdo not normally treat what they
directly observeas an interpretation,even for politeness'sake.
As Kuno does not discuss modifications to the no desu construction
analogousto the Englishadverbs,modals,and negation,it is impossibleat this stage
of our understandingof theseconstructionsto be certain of how similar they are in
Japaneseand English.We can, however,note that the tensedclausein inferentials
is in subject position, and that no expletiveoccursin these sentencesin Japanese
(or Korean). These constructionsmay be used to make assertionsthat implicate
causeand explanation.And indeed the English examplesdiscussedhere translate
straightforwardlyinto Japanese.For example,(40) translatesIr's that I'm not pretty
enough.
(40)
Watashi-ga kiree-de-nai-kara
pretty-is-NEG-since
I-SM
na-no-desu
is-COMP-ise
Closely related to explanationsare reasonsand causes.
4.2.1.Reasons
In (41), the clauseis interpretedas a reasonfor him being cruel to her.
(41)
e
Perhaps,if she had not been so frightened and rebellious in those
days,she might have seenthat he was eager to pleaseher then too.
He had not been cruel to her, or if he was,i/ wasonly that he seemed
to know liltle of women'sbodiesand how to use them. (Bradley 1982:
24\
I u. grateful to W. Mary Kim of the U.S. Foreign Service Institute for translating several
of the English examples into Japaneseand Korean and for her comments on this construction in
those languages (personal communication). I am, of course, entirely responsible for the
interpretation of her comments representedin this paper.
The inferential construction
353
4.2.2.Causes
In (23) the clause,'I'm not pretty enough'representsthe causeof 'his problem.'h
(42)threepotentialcausesof the delayin globalwarmingare proposedin the three
coordinated
inferentialclauses.
(42)
If the observedtemperatureincreasereally is a greenhousewarming
and not just "noise"- a random fluctuation - one might account for
the disparityin variousways.Perhapsthe modelsare simply twice too
sensitiveto small increasesin greenhousegasses,or perhaps the
incomplete and inhomogeneousnetwork of thermometers has
underestimatedthe global warming.Conceivablysome other factor,
not well accountedfor in the models,is delayingor counteractingthe
warming.It might be that the heat capacityof the oceonsis largerthan
cuffenr modelscalculate,that the sLut'soutput has declinedslightly or
that volcanoeshave injected more dust into the stratospherethan is
cuffentty known, therebyreducingthe solar energ/ reaching the ground.
(Schneider1989)
43. Conclusions
The clausemay representa conclusiondrawn from the context.
(43)
Lichardus'model is a variant of a broader explanationof the cultural
changeseen throughout both Northern and Central Europe in the
late Neolithic. [Extensive discussionof the shortcomingsof the
model.] It is rtot that the model is wrong; there is just not enough
evidenceproposedto evaluateit. (Mallory 1989:253)
In (a3) the inferentialinvitesthe inferencethat the conclusionto draw from all of
thecounterevidenceadducedis not that the model is wrong,rather there is just not
enoughevidenceto evaluateit. Beyondthis we might add that givenall the counter
evidencethat the author presents,a reader might be forgiven for concludingthat
the author believesthat the model is wrons.
4,3.I.Results
Iconsequences
Closelyrelatedto conclusionsare resultsand consequences.
(44)
Someone- Rubinstein,maybe- once said,when askedif he believed
in God: "Oh, no, I believe
in something much bigger." And
someoneelse- was it Chesterton?- saidthat when men stop believing
in God, it istt't that they then believe in nothing: They believe in
354
Gerald P. Delahunty
ev er y th i n g(E
. c o l 9 fi 9 :6 2 0 )' o
ln (44) the inferential licensesthe inference that the result of mcn stopping
believing in God is not that they then believe in nothing.
4.4. Relevant alternative comparisons lcontrasts
The clause. especially when the matrix contains just or only, mtty represent an
alternative comparison or contrast to one expressed in the krcal c<tntext.
(45)
'Do
you leave Kent on Saturday'l' said she.
'Yes if Darcy does not put it off again. But I am at his disposal. He
arranges the businessjust zrshe pleases.'
'And
if not able to please himself in the arrangement, he has at least
3::il,'f:::':J".:h1ill;ffi'il';i,l; il:iil-,i?,Tll xl:
Darcy.'
'He
likes to have his own way very well,' replied Colonel Fitzwilliam.
'But
so we all do. 1r is orilv tltat he hus better meons o.f hovirtg it than
motv ollrcrs, becuttse he is ich, urtd mart,v others ore poor. I speak
t-eelingly.A younger son, you know, must be enurecl to self-denial and
d e p e n d e n c e . (' A u s t e n 1 8 1 3 [ 1 9 6 3 ] : 1 5 3 1 1 )
In (a5) Fitzwilliams' interential proposes an interpretation of Darcy's behavior or
personality that contrasts with Eliza Bennett's, namely that Darcy is ric:her than
many others and so can indulge himself more than they can.
The inferential in (46) expressesan alternative interpretation of the histrtrical
facts that contrasts with the interpretation suggestedin the immecliately prirlr text.
(46)
It would be a mistake, nevertheless,to describe the lrish Church from
the sixth century on as a monastic Church. There were bishops
throughout Ireland who carried out functions specitic to them aktne.
The fact that there wits no late Roman diocesan structure which could
be taken over by the Church was not just the case in lreland; it was
evident in all areas outside the boundaries of the Roman Empire. 1r
is jtut that Ireland was the first coLntry in Weskm Ettrope to feel the
fitll impact of the,secircumstances.(Richter 198i1: 61-2\
4.5. Indeterminacy of interpretation
Many of these. examples can be entered under several ditferent interpretive
l0
E x a m p l e ( 4 4 ) i s a t r a n s l a t i o no f t h c o r i g i n a l l t a l i a n , w h i c h h a s t h e following infercntial
corresponding to the italicizcd English: non i che non credanopiu a nulla (Eco 1988: 492). Note
t h e n u l l s u b i e c t a n d t h e o t h e r l e x i c a la n d s v n t a c t i cc l c m e n t sc h a r a c r c r i s t i co f i n f e r e n t i a l s .
The inferential construction
355
What this meansis that we cannotbe sure of our interpretationof the
categories.
sentences
in their contexts;that is, we cannothe certainthat the implicatureswe
deriveare actuallvthe ones intendedby the speaker.This of courseis exactlyas
Gr ic epr edic t ed( 1 9 7 5 :5 fl ):
"Sincetclcalculatea conversational
implicatureis to calculatewhat hasto be
supposed
in order to preservethe suppositionthat the cooperativeprinciple
is heing<lbserved,
and sincethere may be variousspecificexplanations,
a list
of which'mavbe open, the conversational
implicatumin such caseswill be
disjunctionof suchspecificexplanations,
and if the list of theseis open,tlrc
implicatumwill hut,ejust the kind of indetenninacytltat many actual implicata
do in fact seamIo heve."(My italics.)
Iror example,(23) It'sI'm not pretry'
enough... can be interpretedas a reason
for.or a causeof, rlr an explanationfor. or an interpretaticln
of the man's pattern
or problem.We haveno way of knowingjust whichof thesepossibilities
is intended
bythespeaker.indced,there is no reasonto assumethat the speakerintendedany
specific
oneof them,as opposedto simplypromptingzilland leavingit to the hearer
to makea chclice(or not).
More significantly,
we canview all of thesecategories
asrepresenting
a single
naturalciass:They all representaspectsof interpretation.So the inferentialalways
represents
an interpretationof the krcal contcxt.its preciseinterpretationas an
explanation,
reflection,etc. depcndsupon its particularcontext.Indeed the range
of particuiarimplicaturesscems to be limited to those which have to do with
interpretation.
To ret-lecton, explain,clr identifythe consequences
of somesituation
(or text) is to interpret it. By analysingthe inferential as representingan
interpretation
of its local context,we can accountnaturallyfor this limitation.
4.6. Therolesof the inferential atnstruction in disuturse inferencing
I a s s um e
t hat t he i n tc rp re ta ti o no f d i s c c l u rsbey i ts parti ci pantsi s a processof
- from the languageusedand its interactionwith its context. In
derivinginferences
thissectionI will briefly cclnsidcrin mclregeneralterms the roles played by the
propositionexpressedbv the ciause of the inferential construction in the
i n t er pr et at ion
of di s c o u rs eT.h i s p ro p o s i ti o nc a n tu ncti onas ei thera concl usi onor
a premise.
If thc tnf'erence
representsa conclusion,then it is one that can be derived
tiom the irrintcdiatelyprior text. either alone or in conjunctionwith implicated
premises
or bridgingassumptions.
The conclusionalso can be aflirmed, qualified,
o r denied.E x am p l e(4 7 ) i s a c l ra r c a s eo f a n i n f' erredconcl usi ctn.
(.ai1
Lichardus' model is a variant of a broader explanation of the cultural
c h a t t g e ss e e n t h r o u g h i l u t b o t h N o r t h e r n a n d C e n t r a l E u r o p e i n t h e
Llte: Ncolithic.
[Paragraph continues detailing problems with
Lichardris' mildcl.] It is ttttl thut tlw tnrrlel is wrong; there is lust not
eni;ugh criiJcncc pir.rltitscrtto cvulu;rtc it. (l\'laliory 1989:253)
Gerald P. Delahunty
356
The problems attributed by Mallory to the Lichardusmodel along with the
assumptionthat the enumerationof a model's problems are usually suftlcient
*u..unt to concludethat (the author thinks that and expectsthe reader to conclude
that) the model is wrong. The author rejects this conclusionand postpones
evaluationof the model.
If the inference represents a premise (usually a minor one) then the
propositionsrepresentedby the (immediately)prior text, or derivablefrom it, can
te deduced from the inferential proposition in conjunction with other, usually
implicated (rather than overtlystated)premisesor bridgingassumptions.Inferences
inttrpreted as premises are often, though by no means solely, interpreted as
The premisecanbe affirmed,qualified'or denied.
explanationr,,urr"., or reasons.
interpretedas a premise.
an
inference
represents
Example (48)
(4g)
'Tell me if, when I returned to England in the year eight, with a few
thousand pounds, and was pclstedinto the l-aconia, if I had then
written to you, wcluld you have answeredmy letter? would you, in
short, have renewedthe engagementthen?
'Would I!'was all her answer;but the accentwas decisiveenough.
'Good God!' he cried, 'you wouldr.It is not that I did not think of it, or
But I was proud,
desire it, as what would alone crown my othersLrccess.
1980
(Austen
again.
to
ask
too proud
[1818):233)
that
Frorn the text prior to the inferential we can derive the assumption
'I did not
engagement.
Wentworth did not write to ask Anne Elliot to renew their
think of it, or desireit' representsa (minor) premisefrom which,with other readily
and thereby
available assumptions,
^hadn'tWentworth's not writing to ask would follow,
written. Wentworth summarilyrejectsthis explanation'and
explain why he
providesan alternative('I was proud, too proud to ask again').11
5. Negativeinferentials
Becausethey do not entirely parallel their positivecounterparts,inferentialswith a
negated matrix require a brief separate comment. We should expect negative
infErentialssimply io deny the inferenceslicensedby their positives,and for the
most part this ishow they function.Clearly,negativeinferentialscan reject the truth
1l
The inferencing can be laid out as the following syllogism:
If Wcntworth did not think of it, or desire it, then he would not write'
Wentworth did not think of it, or desire it.
Therefore Wentworth did not write.
The proposition derived from the clause of an inferential may be entertained eithcr as a premise
or as a conclusion, or both:
lf Wentworth did not write then he did not think of it, or desire it.
Wentworlh did not write.
Therefore Wentworth did not think of it, or desire it. The indeterminacy of the role
played by rhe proposition is a further sour@ of indeterminacyin their interpretation. My thanks
to Jim Garvey for discussionon this topic.
The inferential construction
357
of the informationin the clause:
(49)
"Nance was sayin' l.ambert was gone to Dublin again, but what
signifieswhat the likes of her'd say;it couldn't be he'd be goin' in it
agin and he not home a week from it." (sic) (Somervilleand Ross
1977:264)
However,they may reject,not the truth, but the relevanceof an assertionor
an assumptionat the point in the discourseat which they occur:
(50)
On principle I usuallyavoidintroducingmy friendsand acquaintances
to each other. It is not that onefears treachery,though of course one
does. What human fear is deeper? But endlesslittle unnecessary
troublesusuallyresult from suchintroductions.t2
(Murdoch 7975:43)
Similarly,the negativeinferentialin (a7) is compatiblewith either the model
beingright or beingwrong. Its function is to forestallthe reader'sinferencethat the
author'slitany of the model'sinadequaciesleadsto the conclusionthat it is wrong.
Negativeinferentialsmay also deny an inference which would provide a
plausible
explanationrelevantin the context.In (50) the negativeinferential rejects
the proposition that the narrator fears treachery as the explanation for his
reluctanceto introduce his friends and acquaintancesto each other, a perfectly
plausibleexplanationfor his behavior,and an inferencethat might be made by any
reasonable
audience.The sentenceimmediatelyfollowing concedesthis fear. And
thethird sentenceprovidesthe explanationthat the narrator wants the audienceto
accept.
A negative inferential may also deny a plausible interpretation of, or
extrapolationfrom, its context, in the following case,the extrapolation that demon
loversare usuallygrosslycruel:
(51)
It was ... as if she had died long before and come back to me as a
demon lover. Demon lovers are alwaysrelentless,however kind in
life. And it was sometimesas if t could 'remember' Christian's
kindness,though all now was spite and demonry.It was not that she
was usually,tltottglt slrc was sametimes,grosstycruel. (Murdoch I975:
e1-z).
Only one of my collection of negativeinferentialscontains a modal, (49)
above.The negativetakesscopeover the modal and we can paraphrase(49) as the
negationof a possibility:
(52)
It is not possiblethat he would be going there again.
12
Replacing the inferential in this text with a sentence such as It is not the casethat one
fearsteachery renders the text contradictory. This supports my contention that the inferential is
specializedin function and contrasts with sentencessuch as those with predicates such as 'true'
or'the case'in the matrix.
358
GeraldP. Detahunry
Many of the inferentialsin the corpus,both positiveand negative,contain
matrix adverbs,typicallyonly,just,,or simply.These limit the domain of a contrast.
In (45), the contrastbetweenDarcy and'we all' is initiallyproposedto be that he
likes his own way, but is subsequently
limited to his merely being richer than the
others, and so better able to afford to indulge himself.
In exampleswhich containboth matrix adverbsand negation,the adverbfalls
within the scopeof the negative.When the adverbis only,the exhaustiveness
of the
inferenceis denied. The following passageassertsthat Anne Elliot could never
accept Mr. Elliot not only becauseher feelings were adverse to him but also
becauseher judgmentwas againsthim:
(53)
She never could accepthim. And it wasrtot only that herfeelingswere
still udverseto any matt save otrc; her judgment, on a serious
consideration
of the possibilities
of sucha case,wasagainstMr. Elliot.
( A us te n 1 9 8 0 [1 8 1 8 1]:5 2 )
When the adverb isjast or simply,,the denial may be either of exhaustiveness
rurof a limitation of the domain of contrast:
(54)
'I wish, I
wish she hadn't met Arnold.'
'You're very attached Arnold,
to
aren't you?'
'Yes.'
'lt's rtot just that you core what he thinks?'13
'No.' (Murdoch 1975:117)
So, there is a general,thoughnot perfect,parallelismbetweenpositiveand
negative inferentials.The negativedenies the various inferenceslicensedby the
positive.Typically, negativeinferentialsoccur in the context of contrastingclaims,
most clearly exemplifiedin the tandem constructions.We can reasonablyinterpret
'The
the pattern as:
inferenceto draw from (the utterance)of this expressionis not
that p; rather it is that q) ln this respect inferentialsare akin tu repairs in
conversation.
Many negated inferentialsraise the issue of what prompts the inf-erence
rejectedby the speaker.In somecases,e.g.(49),the inferentialclausemerelyrejects
13
Declerck (1992:221, fn 8) paraphrasesthis infercntial as'the narLtreof your feelingsfor
Amold" is not just that you care what he thinks?, a paraphrase which is hardly grammatical and
certainly not consistent with his analysis.He claims (p.220) that the expressionsthat replace the
'automatically
cxpletive i/ are
taken from the group rcason/cause/explanation/interpretation,"
a
claim which cries out for explanation. Clearly the narure of your feelingsfor Amold does not fit
naturally into this class.The current analysissuggestsa paraphrasesuch as the following, which
fits the text better: I ant not to interpret vour agreeingthat you are very attached to Amold to nrcan
just that you care what he thinks?
In fact, Declerck's analvsisis not nearly restrictive enough. There are many nouns that
form a natural scmantic class with reason, explaruilion, etc., which can appear in the copular
structure'NPrr that',and which expressa notion whose contents can be specified by a thatclause, e.g. et'idence (see my discussion of Kuno's analysis),fact, allegation,forecast, prediction,
position, but which do not appear to figure in the interpretation of inferential il rs (not) that
sentences.This limitation follows naturally from the current analysis.
The inferential constntction
359
a proposition expressed in the prior sentence. But most cases are not so
straightforward.
Some propositionsare derived from generallyheld beliefs,others
from stereofypes.
The interentialin (18) derivesfrom the assumptionthat spiesfear
discovery;
that in (55) representsan inferencebasedon the assumptionthat rich
peoplewho live 'not immodest'lives are merely pretendingto be 'ordinary chaps':
(55)
I shouldmake it clearthat Arnold was not in any crude sense'spoilt'
by success.He was no tax-dodgerwith a yacht and a house in Malta.
... He lived in a fairly large, but not immodest suburban villa in a
'good class'housingestatein Ealing.
His domesticlif'ewas,even to an
irritating extent, lacking in style.It was not iltat he put on an act of
being 'the ordinary chap'. (Murdoch 1975: 3I)
Narrators may produce a negative inferential when they wish, sincerely or
manipulatively,
to introducean interpretationof an aspectof the contextwhich they
wantto reject.
However, some examples contain information which is not likely to be
inferredby the hearer from the local context.The negativeinferential acts as an
instructionto the hearer to establishand then reject an inferential connection(eg.
explanation,
conclusion,etc.) with the local context.
Why would a speakermention a propositionwhich is unlikely to be interred
by the hearer only to reject it immediately?To amuse and disarm, as in the
followinge-mail messagefrom the chair of my department:
(56)
Be sure to let me know when you are out of town on University
business. It isrt't ortly tlrut I'm nosey. Apparently for Worker's
Compensationcoveragein case of an accident,we have to know
you're on the job. (P. Cowell,C.S.U.Englishdept. e-mail message
4lIele3)
More interestingly,becausethe form revealsinferencesmade by its speaker,
it canbe usedby authorsto revealthe thoughtsof a fictional characteror narrator.
of my examples(e.g.(50)) are producedby the narrator of Murdoch (1975)
Several
andhelp to underminehis reliability.la
6. Inferentials:An analysis
The hypothesisto be explored here is that the inferential can be viewed as a
pragmaticinstructionto its audienceto regard its clauseas an interpretationof its
local context, that is, to be about, rather than of, its context. The specific
interpretations,
describedin section4,,are to be seen as particular interpretations
of particularcontexts.Consjderthe following pair:
la
Whether relevanceas opposeclto truth is alwaysat issue with inferentials will be dealt
with in the next section. Certainly (47) and (50) dcmonstrate that truth per se is not always at
issue.
360
Gerald P. Delahunty
(s7) a.
b.
Women in Ireland are nclt a form of prayer.
It is that women in Ireland are not a form of prayer.
(57a) merely reportsthat women in lreland are not a form of prayer;(57b) on the
other hand, invites the inferencethat the propositionwomen in lreland are not o
for of prayer has some special relevanceto the local text. Utterances seem in
general to have two very distinct types of roles in discourse:They can be either
relevant.That is, they can fit with their contextsin
contextuallyor metacontextually
wayswhich we might characterizeas ordinary,e.g.,addinginformation,or they may
have an extraordinaryrelevanceby commentingon their contexts.In this section I
will account for the strictly local relevanceof inferentials,and argue that their
interpretation involvestwo stages.The first stageinvolvesthe interpretationof the
implicaturelicensedby the
inferentialconstructionas a generalizedconversational
construction'sform, namely that the proposition representedby the clause is an
interpretation of the local context. The second stage involves calculating the
potentialparticularizedimplicaturesof a constructionin its specificcontext;that is,
calculatingjust how the inferentialmight be an interpretationof its specificcontext.
That the inferentialrelatesto the local contextfollows from Grice's CP,
which requiresthat a participant'scontribution'be suchas is requiredat the stage
at which it occurs,'from the submiuim of manner "Be orderly,"and from the fact
that there is nothing to indicatethat the sentencehas a non-localrelevance.Any
speaker/writerintendinga non-localrelevancewould be in violation of the submaxim of manner "Avoid obscurity."Note, of course,that although the inferential
interprets the immediately prior text (nr immediate context, in the case of
Japanese),
the extentof that text (or context)is not fixed.This is anothersourceof
the indeterminacyof the meaningof inferentials.
becauseof its form, licensesa
In supportof the claim that the construction,
implicature,note that suchimplicaturesare detachable
generalizedconversational
and calculable(Levinsctn19t13:12U).15
As we haveseenfrom the data presentedin thispaper,inferentialsentences,
whether positiveor negativc,interrogativeor modalized,all instructtheir hearers
to regard the clauseas an interpretationof the local context;that is, the same
generalinferencefollowsfrom all inferentials.
'l'his
inference can be detached,as is generallythe casewith implicatures
dependentupon the form of the utterance.If we removethe matrix,we may lose
the inferentialinterpretation.Compare(57a) and (57b).
The inf'erentialeftect can be calculatedas there are iust two waysin which
15
Usually, conversaticlnalimplicatures are canccllablc.However, it seems to me that a
speaker producing an expressionwhosc form w<luld ordinarily license an implicature is unlikely
to be believed if they claimed that the form has no relevanceat all. For example, a parent who
spells a messageout to a spouse in the hearing of a preliterate child can hardly deny that they do
so with some intention, such as that of excluding the child from the intended audience. Likewise,
the generalized conversational implicature licensedby the inferential construction seems not to
be cancellable. In this respect it is like a conventional implicature, and may in some languages,
such as Japanese,pei'haps,have becomc conventionalized.However, if the construction licensed
conventional rather than convcrsational implicatures,we could not explain how it licensesthe
same implicatures in languagc aftcr language.
The inferential construction
36I
a new propositionis integratedwith those that have gone before it: It can have
eitherordinaryor extraordinaryrelevance.That is, its relevancecan only be either
contextualor metacontextual.The calculationmight go as follows.
A speakerhasusedan inferentialsuchas (57b)whoseconventionalmeaning
non-inferential(57a).
is the sameas that of the corresponding
So,givena choicebetween(57a)and (57b),why would a speakerchoosethe
consistingof a that clausesubordinatedto semanticallyempty it and be?
expression
(57b) cannot mean just (57a) becausethis would violate the maxim of
manner,specificallythe injunction to be brief. It would also violate the maxim of
relation,as the matrix would have no relevance.
Nor can (57b) convey less than (57a) because (57a) represents the
meaning of (57b), the minimal information representedby both. If
conventional
(57b)meantlessthan (57a), a hearer could not work out the significanceof (57b)
the matrix, having no conventionalmeaning,gives hearers no clue as to
because
whatinformationin (57b) to disregard.So a speaker using (57b) to convey less
informationthan (57a) would be in violation of the maxim of manner'sinjunction
againstobscurity.
It follows that (57b) must have more significancethan (57a). This extra
cannotbe conventionalbecauseif it were the speakerwould be being
significance
obscureand so in violation of manner,and perhapsalso in violation of quantity in
as we are given no clues
not supplyingsufficientinformation for the circumstances,
asto what that conventionalextra might be.
Consequently,the extra significanceassociatedwith (57b) must have to do
with the relationshipof the information representedin it to its context, i.e. its
relevance,because,ex lrypothesi,there are only two possibilities:Either the
propositionrepresentsinformation that is integratedinto the interpretationof the
text as ordinarilytopical, or it has some special,extra relevance.It is relevant as
informationabout the context.
Significantlyin this regard, becauseits interpretationcrucially involvesthe
maximof relation,it is the proposition'srelevance,rather than its truth, which is
important in the context. Example (50) pointedly illustrates this characteristic.
Negativeinferentials deny, not the truth, but the (special) relevance of the
propositionrepresentedby an inferential'sclause.Truth, although necessaryfor
is certainlynot sufficient.
relevance,
In sum, the generalizedconversationalimplicatureof the inf-erentialis that
the clauserepresentsan interpretationof the local discoursecontext.
On the other hand, the further interpretationof the clauseas a reflection,
explanation,conclusion,etc., must be derived from the interpretation of the
inferentialas a comment on the context, and the particularsof that context, in
with Grice'sCP and maxims.In other words,this interpretationconsists
accordance
of a set of particularized conversationalimplicatures.They are particularized
to the problem of discoveringthe specificspecialrelevanceof a particular
solutions
inferentialsentencein a particular context.They have the characteristicsof such
They are vague and must be calculated,as Grice claimed.They are
implicatures:
alsodefeasible:The audiencecan never be certainthat the particular implicatures
theyderiveare the ones intended,and consequentlyany particular inferencescan
be denied.
Finally, the construction appears to have the same or very similar
362
GeraldP. Deluhunry
interpretationsin all the languagesin which it occurs.This universalitycan only
result from universalprinciplesof pragmaticinterpretation,such as the CP and
maxims. It fbllowsthat the interpretations
of inferentialsmustbe non-conventional.
If we assumethat hearersdraw inferencesasthey createmeanings,and if we
also assumethat a speaker'sgoal is to ensurethat the hearer entertainsonly the
interpretationhe or she intends,then we can view the inferential torm as an
instructionto the audienceto infer an interpretationalrelationshipbetween the
clause and the local context.If the inferentialcontainsmatrix negationthen it
indicatesthat an infercncewhich may be plausiblein the contextis not intendedby
the narrator.The adverbsand modalswhich occurin manyinterentialsindicatethe
scope of the relevanceof the inf-erential
propositionsand the degree<tffaith the
sp e ak erhas in t he m.
7. Conclusion
What inferentialsdo is inclicatethut the clauseis tclbe regardedas an interpretation
of the local context. Negative int'erentialsdeny the interpretation; positive
inferentialsaftirm it; and inferentialswith adverbsor modalsindicatethe degreeof
co nf idenc et o be p l a c e di n th e i n te rp re ta ti c l nTsh. i s functi oncl fthe i nterenti ali s ti r
iiccnse a _eene
ralizcd conversationalimplicature which is consequentlythe
re s pons ibilitof
v t h e s p c a k e r.
Infcrcntialsconversiitronallv
licensethe particularizedint'erencesthat the
cluuscis (or rsnot) a conclusion.
account.explanation,
reason,possibility,
retlection,
'l'he
ctc.
r-:hoicc
amongstthesc int'erences
is the responsibility
of the hearer.
The interactionbctweeninterpretations
for whichthe speakeris responsible
and interprctationsfor which thc heareris responsible
can give the impressionthat
the speakerappearsto be makingassumptions
about inferencesbeing madeby the
hearer in the interprctiveprocess.By revealingwhat the speakerappearsto think
that the hearerassurnes,
the constructioncanbe usedby skilledwritersto revealthe
concernsof narrators,somctimcsto suggesttheir limitationsand unreliability.
Given that a speakermustguidean audiencealongan interpretationalpath,
licensingcertain inferencesand rejectingothers,it should not be surprisingthat
languagesprovide specializedsentencestructureswhose function is to indicate
interpretationsof local context. Nor should it be surprisingto find that these
constructions
occurin identifiablediscourse
patternsdesignedspecifically
to indicate
that the speakerwants the hearerto draw, not this interence,but that other one.
And, given the universalityot thr: pragnraticprinciplesupon which the inferential
co n s t r uc t ionr elies .i t i s n o t s u rp ri s i n ge i th e r th a t the constructi onhas the same
t'unctionsacrosslanguages.
References
A u s t c n , J . ( 1 9 6 3 ) [ 1 U l 3 l P r i d c a n d p r e l u d i c e .[ - o n d o n : D e n l .
A u s t e n , i . ( 1 9 8 0 ) [ 1 8 1 8 i P c r s L t L t s i <Ot n
x f. o r d : O x l i r r d P a p c r b a c k s .
The infere:ntial construction
363
Bradley,M.Z. (1982) The nists of Avalon. New York: Ballantine.
Collins, P.C. (1ry1) Cleft and pseudo-cleft consfructionsin English. L.ondon: Routle<lge.
[-ouvain: Lcuvcn Univcrsity
Declerck,R. (1988) Srudieson copular sentences,t:leftsand p.seudo-cleft.s.
Press.
Declerck,R. (l!xl2) The inferential ir rs lhrzl-constructionand its conscncrs.Lingua u7: 2()3-2.30.
cltft sentences.Bloomington:
Delahunty,G.P. (1982) Topics in lhe syntarand sentantics<;j'L,n,gii.th
IULC.
Eco, U. (1988)II pendolo di Foucault. Milano: Bonipiani.
Eco, U. (1989) Foucauld's pendulunt. Translated b.,'"W. Weavcr. San Diego: Hilrcourt, Brac€,
Jovanovich.
Frame,R. (1981) Colonial lreland 1169-1369.Dublin: Hclicon.
Garrett,G. (1983) The succession.New York: Doubleday.
Gogarty,O.St.J.(1968) As I was going dow'n Sack:'illeStrter. Lrndon: Spherc.
Grice,H.P. (1975) Lngic and conversation.In P.Cole and J.L. Morgan (eds.),Syntru and sennntics:
Speech
ncts.Vol..l. New York: Academic Prcss.
Horn, L.R. (1989) The narural history of negation.Chicago: University of Chicago Press.
Irving, J. (1978) The 158-pound nnrriage. New York: Pocket Books.
Kies, D. (1988) Marked themes with and without pronominal rcinforcement: Their meaning and
distribution in discourse. In E. Steincr and R. Vcltman (eds.), Pragnmtics,discourse and text.
Norwood: Ablex.
Kuno, S. (1973) The stntcture of the Japaneselanguage. Cambridge, MA: MIT Press.
lrvinson, S. (1983) Pragnatics. Cambridge: Cambridge Univcrsity Press.
Mallory, J.P. (1989) In search of the Indo-Europeans. l.ondcln: Thames antl Hudson.
Murdoch,I. (1975) The black prince. Ncw York: Penguin.
Powell,T.G.E. (1983) The Celts. krndon: Thamcs and Hudson.
grantnnr of the English
Quirk, R., S. Greenbaum, G. Lrech, J. Svartvik (1985) A contprehensive
lnngunge.lnndon: Longman.
Richter,M. (1988) Medieval lreland. Dublin: Gill and Macmillan.
Rosch, E. (1975) Cognitive rcpresentations of semantic catcgories. Journal of Expeintental
Prychologt:General 104: 192-233.
Schneider,S.H. (1989) The changing climate. ScientilicAnrcrican (Scpt.).
Somerville,E. and M. Ross (1977) The reol Charlorte.Lrndon: (luartct.
364
Gerald P. Delahunty
Sperbcr, D. and D. Wilson (19tt6) Relevance:contnunicalion and cognition. Cambridge, MA:
Harvard.
Wilder, T. (1987) The ides of March. New York: Harpcr and Rowc.