Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
-
Upload
william-pearson -
Category
Documents
-
view
222 -
download
1
Transcript of Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
1/37
Semantics and Pragmatics: The Boundary Issue
K. M. Jaszczolt
University of Cambridge
To appear in: Semantics: An International Handbook of Natural Language Meaning,
ed. K. von Heusinger, . ortner ! C. Maienborn. "erlin: Mouton de #ruyter.
$ut%or&s address:
'epartment of (inguistics
)aculty of Modern and Medieval (anguages
University of Cambridge
*idg+ic $venue
Cambridge C"- '$
United Kingdom
Tel. /00 122- --3400
)a5 /00 122- --3672
89mail m21;cam.ac.u
1
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
2/37
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
3/37
?1@ T%e anitor left t%e door open and t%e prisoner escaped.
?1a@ T%e prisoner escaped and t%e anitor left t%e door open.
obert Musil and < can mae it clear
in t%e conversation t%at +%en < use t%is description < indeed mean >obert Musil@ or
t%ey can be used attributively, to mean +%oever %appens to be t%e best $ustrian
novelist for t%e person uttering t%is sentence. nce a trut%9conditional semanticist %as
recognised t%e importance of t%ese e5tra9sentential factors and contemplated t%ealteration of t%e obect of study from t%e sentence to t%e utterance, and from sentence
-
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
4/37
meaning to t%e speaer&s meaning, %eDs%e %as t+o options to c%oose from. *uc% a
semanticist can eit%er reect trut%9conditional semantics and loo for an alternative
t%eory, or try to preserve trut%9conditional semantics and adapt it to t%e ne+ obect of
study. (et us assume t%at our semanticist is a devoted supporter of t%e met%ods of
trut%9conditional semantics and opts for t%e latter solution. $t t%is uncture %eDs%e is
faced +it% t+o options again: eit%er ?i@ to advocate t%at t%ere is more to meaning t%antrut% conditions alone and allo+ for a pragmatic overlay to t%e %ereby saved semantic
t%eory, or ?ii@ to propose t%at pragmatic factors contribute to t%e trut%9conditional
content and As%ift& t%e trut%9conditional analysis from sentences to utterances,
claiming t%at t%e really interesting trut% conditions are t%ose of utterances, not t%ose
of sentences. T%e latter avenue opens up t%e option of adding information about
meaning t%at comes from t%e sources ot%er t%an t%e le5icon and t%e structure and
predicating trut% or falsity of t%e total product.
(et us no+ see +%at actually %appened in t%e %istory of t%e
semanticsDpragmatics interface, beginning +it% t%e p%ilosop%er +%o is generally
recognised as t%e founder of modern pragmatic t%eory: aul #rice. #rice pursued t%e
pat% of preserving t%e trut%9conditional analysis but +as rat%er ambivalent as to t%eoptions ?i@ and ?ii@ above: %e allo+ed some pragmatic input in t%e trut%9conditional
content, +%ile eeping most of it outside of t%is content. Eeedless to say, t%is ind of
proposal %ad to generate a %eated debate, a debate t%at is still at t%e forefront of
t%eorising in semantics and pragmatics and t%at is still producing cutting edge
analyses.
1.1. T%e pragmatic +astebaset: #rice on what is saidand what is implicated
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
5/37
but %as t%ree sides.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
6/37
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
7/37
suc% as t%at t%e performance +as a particularly bad one. 2To sum up, implicatures can
arise t%roug% observing or violating t%e ma5ims of conversation.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
8/37
course, t%e conte5t maes it clear t%at t%ey %ave to be cancelled or s%ould not be
computed. T%e implicature in ?0a@ is an e5ample of suc% a generalized prediction,
called by #rice a generalized conversational implicature ?#C
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
9/37
pragmatics but it is an important part of #rice&s o+n perception of t%e boundary: it
s%o+s t%at for #rice trut%9conditional semantics is t%e core of t%e analysis of
meaning, and pragmatic inference is added as and +%en it fits N +it%in t%e trut%9
conditional content ?disambiguation and reference assignment@, or outside it, but +it%
varying degrees of Asemanticity&: from conventional implicatures t%at are a species of
coded content, t%roug% generalized implicatures t%at arise in virtue of suc% codedmeanings, and finally conte5t9dependent particularized implicatures t%at are most
Apragmatic& of t%em all.-
>etrospectively, +e are in a position to say t%at t%e fact t%at #rice ?i@ admitted
t%e output of some pragmatics processes in t%e trut%9conditional aspect of meaning
and ?ii@ arranged implicatures on a scale of progressive detac%ment from t%e semantic
content opened a door for t%e current boundary dispute in t%at once some pragmatic
inference finds its +ay into t%e propositional representation, it becomes a matter of
debate ?and empirical evidence@ %o+ muc% e5actly s%ould be allo+ed in. Moreover,
#rice employed t%e t%eory of implicature to argue against t%e met%odological
superiority of postulating ambiguities +%ere a unitary semantic account can be given.
)or disunction in 8nglis%, for e5ample, it suffices to admit t%e meaning of itseBuivalent from propositional logic as its semantic content and account for t%e
e5clusive reading by means of an implicature: t%ere is no need to postulate semantic
ambiguity on a par +it% t%e le5ical ambiguity of Aban& or Aport&. T%is economy of
senses +as spelled out as a Modified ccam&s >azor ?#rice 14: 04@:
Senses are not to be multiplied be$ond necessit$%
$gain, retrospectively, +e can udge t%at t%is +as a landmar in pragmatic t%eory.
nce unnecessary ambiguities are e5orcised on met%odological grounds, +e are left
+it% semantic meanings t%at cannot al+ays be precisely delineated. $nd t%is +as t%e
foundation stone for t%e radical pragmatics in t%e 146s and for t%e idea of semantic
underdetermination to +%ic% +e no+ turn.
1.2. >adical pragmatics and underdetermination of meaning
T%e boundary bet+een semantics and pragmatics %as been discussed in t%e literature
in terms of t%e follo+ing interrelated distinctions:
?i@ meaning vs. useL
?ii@ conventional vs. non9conventional meaningL
?iii@ trut%9conditional vs. non9trut%9conditional meaningL?iv@ propositions vs. utterancesL
?v@ conte5t9dependent vs. conte5t9independent meaning
?see also (yons 14: 134@. $s +e %ave seen in t%e discussion of #rice&s meaningEE,
all of t%ese distinctions figure in t%e +%at is saidD+%at is implicated dic%otomy and all
of t%em partae in t%e fuzzy boundary bet+een t%ese t+o categories. Ho+ever, +%at
became Buestioned after+ards +as t%e idea t%at most pragmatic processes ?apart from
disambiguation and reference assignment@ belong to pragmatics but not to semantics.
T%e slogan Apragmatics eBuals meaning minus trut% conditions& started to lose its
appeal +%en it +as recognised t%at t%e trut%9conditional content does not correspond
-)or completeness, non9conversational implicatures also %ave to be added. *ee Horn ?1: 121@L
(evinson ?1-: 1-1@.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
10/37
to a clear9cult semantic obect. (et us tae sentential negation for e5ample. *entence
?7@, notorious in t%e semantic literature, seems to allo+ for t+o interpretations,
depending on +%et%er t%e presupposition of t%e e5istence of t%e ing of )rance is
fulfilled at t%e time of t%e utterance.
?7@ T%e ing of )rance is not bald.
T%e reading ?7a@ is a presupposing reading: t%ere is a ing of )rance and %e doesn&t
%ave a property of being bald.
?7a@ T%ere is somebody +%o fulfils t%e property of being t%e ing of )rance, t%ere
is only one suc% person, and +%oever fulfils t%is property is not bald.
r, formally:
?7a&@ 5 ?Ko)?5@ y ?Ko)?y@ y S 5@ "ald ?5@@
T%e ot%er reading is ?7b@.
?7b@ T%e ing of )rance is not bald: t%ere is no suc% person.
(et us try to represent t%is reading formally ust by varying t%e position of t%e
negation operator. ?7b&@ is a possible candidate: it contains a +ide9scope, sentential
negation and states t%at it is not t%e case ?for +%atever reason@ t%at t%ere is a ing of
)rance +%o is bald. T%is reason can be t%at )rance is not a monarc%y or t%at t%e ing
is not %airless.
?7b&@ 5 ?Ko)?5@ y ?Ko)?y@ y S 5@ "ald ?5@@
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
11/37
also came to be no+n as radical pragmatics?Cole 11@. To sum up, +%ile le5ical
and syntactic ambiguity can be represented as t+o independent logical forms
pertaining to t%e ambiguous sentence, semantic underdetermination is t%e case +%ere
one underspecified logical form0ensues in processing and furt%er determinations of
meaning tae place t%roug% pragmatic inference or t%roug% some ot%er pragmatic
process suc% as an application of presumed, salient, default interpretations.Underdetermination and inference are t+o interconnected aspects of utterance
interpretation, as is +ell captured in $tlas&s apt pastic%e of Kant:
Gragmatic inference +it%out sense9generality is blind, but sense9generality +it%out
pragmatic inference is empty. $tlas ?1: 120@.
+icy and *adoc ?143@ presented a battery of tests t%at can be used to tell
ambiguity and underdetermination apart. er%aps t%e most important of t%em is t%e
identity test: conoining reduced constituents s%ould be possible only +%en t%e
conuncts %ave matc%ing readings. )or e5ample, let us tae ?4@ and ?@.
?4@ T%ey sa+ %er duc.
?@ T%ey sa+ %er s+allo+.
T%e conunction reduction to ?@ allo+s only to u5tapose matc%ing senses: t%e action
of ducing +it% t%e action of s+allo+ing, and a duc +it% a s+allo+. Crossed
readings are not semantically +ell9formed, unless t%ey are used for punning effect
?see (ascarides et al. 17@. T%is test is modelled on (aoff&s ?146: -34@ Aand t%e
same goes forR& reduction.
?@ T%ey sa+ %er duc and %er s+allo+.
Mi5ed readings are supposed to be freely available for semantic underdeterminacy.
Ho+ever, as in ot%er aspects of t%e boundary dispute, even %ere +e cannot point to
clear9cut distinctions: some e5amples of underdetermination fare better t%an ot%ers
+it% t%e tests and %ence t%e tests are not conclusive. T%e condition of entailment of
t%e logical form is a muc% safer criterion to follo+.
*emantic underdetermination, a revolutionary idea for t%e t%eory of linguistic
meaning, +as a reaction to a +idespread attempt in t%e 176s and early 146s to give
syntactic e5planations to pragmatic p%enomena. T%is tendency +as called #enerative
*emantics. T%e influence of t%e 5ford ordinary language p%ilosop%y, and most
notably Jo%n (. $ustin, H. aul #rice, eter ). *tra+son, late vie+s of (ud+ig=ittgenstein in Cambridge, as +ell as subseBuent arguments put for+ard by "ruce
)raser, #erald #azdar, Jerry Morgan, Jay $tlas, >ut% Kempson, 'eirdre =ilson, and
ot%ers broug%t about t%e fall of t%e generative semantics and opened up a +ay to
pragmatice5planation of clearlypragmaticp%enomena suc% as various uses of
negation and ot%er sentential connectives.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
12/37
G< read t%e first, introductory paragrap% to *adoc and +icy&s paper, and < t%oug%t
to myself, AT%at&s it.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
13/37
+%ile t%e upper boundary Aat most n& is ust an optional addition e5ecuted t%roug%
pragmatic inference. T%e final product, t%e Ae5actly n& meaning, is t%e output of t%e
t+o. Ho+ever, t%ere are also sentences in +%ic% t%e Aat most n& meaning is dominant,
as in ?16@.
?16@ *%e can miss t%e target t%ree times +it%out dropping out of t%e competition.
$ccording to subseBuent analyses, number terms are ust semantically
underdetermined: semantically, t%ey are neit%er at leastn, nor at mostn, nor exactl$n
?see Carston 1@. "ut more recently, t%is analysis of number terms %as been
Buestioned. 85amples suc% as ?11@ and ?12@ strongly suggest t%at t%e exactl$nreading
is not arrived at pragmatically but instead is t%e straig%tfor+ard coded meaning.
?11@ < %ave at least t%ree pounds in my pocet.
?12@ T%ree men carried t%e piano up t%e stairs.
Vualifying t%e number term by Aat least& or Aat most& strongly suggests t%at t%e basicmeaning is ust At%ree&. *imilarly, ?12@, +%ere t%e reading is clearly collective ?three
men together@, does not allo+ for entailments: t%ree men carrying t%e piano toget%er
does not entail t%at t+o did. Eeit%er does it mean t%at per%aps more t%an t%ree did.
*o, per%aps t%e e5actly ?punctual@ semantics of number terms is correct after all ?see
also Koenig 1-L #eurts 1L "ultinc 2663L Jaszczolt 2663L Horn 12, 2667@.
=%ile e5perimental evidence is still inconclusive, it suggests t%at number terms
be%ave very differently from scalar terms. adical pragmatics taes us to t%e landmar in t%e boundary disputes and
c%ronologically it taes us to t%e 146s. >etrospectively, it is not difficult to assess
+%at %appened ne5t as a natural corollary of t%is landmar. nce a +ide variety of
pragmatic additions +as allo+ed in semantic representation, it %ad to be decided +%at
really matters for t%e boundary: ?i@ t%e distinction bet+een t%e ?by no+ eclectic,
semantico9pragmatic@ semantic representation and t%e separate senses t%at t%is
meaning gives rise to, or ?ii@ t%e distinction bet+een t%e logical form as t%e Apure&
output of synta5 and t%e Apure& +%at is p%ysically uttered on t%e one %and, and t%e
+%ole array of implicatures on t%e ot%er, irrespective of +%et%er t%ey contribute to t%e
semantic representation or to some additional senses t%is primary meaning of t%eutterance gives rise to.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
14/37
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
15/37
T%e trut% conditions of t%e pragmatically enric%ed proposition ?p and then @, t%at is
t%e proposition intuitively understood as t%e one meant by t%e speaer, became t%e
dominant obect of analysis of t%e t%eory of meaning.
1.-.2. T%e e5plicitDimplicit boundary
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
16/37
enric%ments stop as soon as optimal relevance is ac%ieved, +%ere by relevance s%e
means t%e cognitive and communicative principle proposed by *perber and =ilson
?17, 13@, according to +%ic% t%e effort invested by t%e addressee in t%e
processing of t%e speaer&s utterance is offset by t%e so9called Acognitive effect&:
gained information or any ot%er improvement to t%e addressee&s information state.
T%e cognitive principle of relevance says t%at AHuman cognition tends to be geared tot%e ma5imization of relevance& ?*perber and =ilson 13: 276@, +%ile t%e
communicative counterpart says t%at A8very act of ostensive communication
communicates a presumption of its o+n optimal relevance& ?*perber and =ilson
17: 13@. To bring t%e idea do+n to common sense intuitions, it says t%at +e stop
interpreting +%at t%e speaer said once +e %ave reac%ed t%e satisfactory
interpretation. "ut it does not say more t%an t%is intuitive claim: it offers no +ay of
measuring or predicting t%e e5plicature in particular cases.
=%at +e also need is a criterion t%at +ould tell us +%ic% ind of output of
pragmatic inference Bualifies as part of t%e e5plicature and +%ic% %as to be relegated
to implicatures proper. Carston proposes suc% a criterion. *%e says t%at implicatures
arefunctionall$ independentfrom t%e e5plicature. T%is means t%at implicatures, if +e+anted to spell t%em out, +ould %ave to %ave t%eir o+n logical forms +%ic% are
independent from t%e logical form of t%e sentence. T%ey function as independent
premises in reasoning. )or e5ample, in ?13@ repeated belo+, ?13a@ is t%e e5plicature,
+%ile ?13b@ is a possible implicature.
?13@ < %ave eaten.
?13a@ T%e speaer %as already eaten lunc% on t%at day.
?13b@ T%e speaer is not %ungry.
"y t%e criterion of functional independence, sentence ?1b@ repeated belo+ is t%e
e5plicature of ?1@.
?1b@ T%e anitor left t%e door open and as a resultt%e prisoner escaped.
?1@ T%e anitor left t%e door open and t%e prisoner escaped.
Ho+ever, functional independence is not a sufficiently precise criterion. =%en +e try
to formulate it in terms of entailment, problems arise. T%e criterion t%at t%e
implicature must not entail t%e e5plicature +ors for most e5amples: ?1&&@ entails ?1@
and %ence it +ould not be plausible to assume t%at t%e interlocutors store bot%
propositions in t%e mindL t%e propositions are not functionally independent. "ut
entailment does not +or +%en construed in t%e ot%er direction: it is perfectly normalfor an e5plicature to entail an implicature, as in ?1@. "&s ans+er entails and
implicates t%at " boug%t some flo+ers.
?1@ $: 'id you buy flo+ers for Mary&s birt%day
": < boug%t some roses.
n some scenarios, it is also problematic to tae functional independence to mean t%e
relation of entailment even +%en it proceeds from implicature to e5plicature ?see
>ecanati 1 and Carston 1 for responseL see also Carston 2661@.
Astensive communication& means communication by s%o+ing an intention to mae some
assumptions no+n to t%e addressee. Cf. *perber and =ilson ?17: 0@.
17
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
17/37
Eo+, as can be seen from e5amples ?13@ and ?17@, t%e sentence uttered by t%e
speaer can be perfectly complete, %ave a complete logical form, and yet pragmatic
inference can tae place. T%is is an important observation in t%at it %as sometimes
been argued t%at pragmatic enric%ment can be e5plained in terms of filling in slots in
t%e logical form: AImuc% syntactic structure is unpronounced, but no less real for
being unpronounced& ?*tanley 2662: 132L see also *tanley and *zabo 2666L King and*tanley 2663@. ecanati ?1: @.
He distinguis%es pragmatic processes of filling in slots in t%e logical form as in t%e
case of t%e assignment of reference to personal pronouns ?saturation@ suc% as Aecanati 2660, 2663@, e5emplified in 1, 13914 above. *trengt%ening, or
modulation, is a Atop9do+n& process: it is independent of t%e constitution of t%e logical
form, +%ere t%e latter is t%e output of t%e processing of grammar. He admits t%at
postulating slots in t%e logical form is t%eoretically possible but encounters a
Buandary: in order to postulate t%e necessary slots to be filled +e already %ave to
no+ +%at is saidW *o, +e %ave circularity in t%e e5planation.Having obected to compulsory syntactic slots as +ell as to some aspects of
t%e functional independence principle, >ecanati ?1@ proposes %is o+n criteria for
delimiting +%at is said. =%at is said is to be specified intuitively and corresponds to
Apre9t%eoretic intuitions&, as %is $vailability rinciple states:
Gecanati ?1:
167@.
14
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
18/37
T%is principle, being rat%er general, is adopted in conunction +it% t%e *cope
rinciple, adapted from Co%en ?141@ +%o, nota bene, used it to t%e opposite effect:
to argue for t%e ric% lexicalmeaning of sentential connectives:
G$ pragmatically determined aspect of meaning is part of +%at is said ?and, t%erefore,
not a conversational implicature@ if N and, per%aps, only if N it falls +it%in t%e scopeof logical operators suc% as negation and conditionals. >ecanati ?1: 110@.
*o, to invoe Co%en&s o+n e5ample, in ?1@ belo+, t%e temporal Aand t%en& meaning
of conunction andin 8nglis% is part of +%at is said because in t%e scope of
implication ?if(then@ it is necessary to mae sense of t%e assertion.
?1@ ecanati&s solution is %is claim t%at +%at is said is t%e
smallest constituent available consciously. ecanati 2660: -900@. =e s%all return to t%e topic of automatic
enric%ment in t%e discussion of default interpretations in *ection 2.
$ll in all, t%e trut% conditions of t%e utterance depend on t%e interplay of a variety of
sources of information. $s a result, +e obtain t%e t%eory of Trut%9Conditional
ragmatics:
GRvarious conte5tual processes come into play in t%e determination of an utterance&s
trut% conditionsL not merely saturation N t%e conte5tual assignment of values to
inde5icals and free variables in t%e logical form of t%e sentence N but also freeenric%ment and ot%er processes +%ic% are not linguistically triggered but are
pragmatic t%roug% and t%roug%. T%at vie+ < +ill %encefort% refer to as ATrut%9
conditional pragmatics ?TC@. >ecanati ?2662a: -62@
>ecanati advocates a rat%er strong vie+ of pragmatic enric%ment ?modulation@. He
claims t%at suc% conte5tual modulation is alwa$spresent: At%ere is no level of
meaning +%ic% is bot% ?i@ propositional ?trut%9evaluable@ and ?ii@ minimalist, t%at is,
unaffected by top9do+n factors&. T%is vie+ is called contextualismand it is currently
t%e subect of %eated debates +it% t%ose +%o +ould rat%er eep semantics simple,
Aminimal&, close to +%at is p%ysically uttered. T%e debate is discussed in *ection 1.0.
1
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
19/37
1.-.0. T%e middle level: an impliciture
ecanati do. n t%e one %and, it is some+%atcounterintuitive to assume t%at in ?13@ repeated belo+ t%e speaersaidt%at %e or s%e
%ad eaten lunch on that da$.
?13@ < %ave eaten.
n t%e ot%er, if +e ept +%at is said minimal, close to +%at +as p%ysically uttered,
+e +ould %ave to admit t%at +%at is said considerably differs from +%at t%e speaer
intended to communicate. T%e e5plicit content of ?13@ +ould %ave to be ?13c@, +%ere
At%e time prior to t%e time of t%e utterance& can be instantiated by any past occasion of
eating, even several years before.
?13c@ T%ere is a time prior to t%e time of t%e utterance at +%ic% t%e speaer ate.
n most occasions, t%e speaer, assuming %eDs%e is a %uman being rat%er t%an a robot
in a science fiction story, +ould deny any intention of conveying suc% content. T%ere
is anot%er difficulty: for #rice, saying somet%ing entails meaning it. *o, +%ile saying
%as to be close to t%e conventional meaning of t%e sentence, at t%e same time it %as to
fulfil t%e meaning entailment.
Kent "ac% ?10, 2661@ recognises t%is difficulty and offers a more intuitively
acceptable solution: t%ere is +%at is said and +%at is implicated, but t%ere are also
parts of t%e content t%at are implicit in what is saidand yet are not implicatures
proper. He acno+ledges t%at people often spea loosely, non9literally, and t%at fact
is not a deficiency of %uman communication but rat%er a positive trait: it is more
efficient to do so and to rely on t%e addressee to recover t%e missing aspects. )or
e5ample, t%e speaer of ?13@ used t%e sentence non9literally: +%at t%e speaer said
differed from +%at t%e speaer meant. *uc% non9literality is different from using
figures of speec%: no part of t%e sentence +as used figuratively, and yet t%e result +as
sentence non9literality, in t%e sense of Aloose tal&.*imilarly, +%en a mot%er reacts to
a c%ild&s crying about a cut finger in saying ?26@, s%e uses t%e sentence non9literally:
?26@ Fou are not going to die, eter.
?from "ac% 10: 274@. =%ile t%e content of t%e sentence ?called t%e minimal
proposition@ is t%at eter is going to live forever, t%e implicit constituents inferred by
%im from %is mot%er&s utterance ensure t%at t%e message is somet%ing to t%e effect of
?26a@.
?26a@ Fou are not going to die from t%is cut, eter.
*o, +%at is meant by t%e speaer is t%e e5pansions of suc% a minimal proposition,
e5emplified in ?13a@ repeated belo+ and in ?21@.
?13a@ T%e speaer %as already eaten lunc% on t%at day.)or a discussion of e5ample ?13@ see "ac% ?10: 27@.
1
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
20/37
*imilarly, sentences t%at are semantically incomplete in t%at t%ey do not %ave clear
trut% conditions suc% as ?14@ repeated belo+, reBuire implicit constituents of +%at is
said.
?14@ Tom is not e5perienced enoug%.
*entence ?14@ does not e5press a propositionL it only e5presses apropositional
radical+%ic% is in need of pragmatic completion in order to produce t%e intended
meaning, suc% as, for e5ample, ?14a@.
?14a@ Tom is not e5perienced enoug% to lead a Himalayan e5pedition.
*o, +e %ave t+o similar p%enomena %ere: sentence non9literality, +%ere t%e minimal
proposition reBuires e5pansion, or, as "ac% calls it Afles%ing out&, and semantic
underdetermination, +%ere t%e propositional radical reBuires completion, or Afilling
in& ?see "ac% 10: 27@. *uc% e5pansions and completions constitute a Amiddleground& bet+een +%at is said and +%at is implicated: t%ey do not belong to +%at is
said and t%ey are not implicatures eit%er. T%ere is no clear boundary to discern. $
proposition enric%ed t%roug% e5pansion and completion %e calls an impliciture:
G
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
21/37
GT%e semantics9pragmatics distinction is not fit to be blurred. =%at lies on eit%er side
of t%e distinction, t%e semantic and t%e pragmatic, may eac% be messy in various
+ays, but t%at doesn&t blur t%e distinction itself. Taen as properties of sentences,
semantic properties are on a par +it% syntactic and p%onological properties: t%ey are
linguistic properties. ragmatic properties, on t%e ot%er %and, belong to acts of
uttering sentences in t%e course of communicating. *entences %ave t%e properties t%ey%ave independently of anybody&s act of uttering t%em. *peaers& intentions do not
endo+ t%em +it% ne+ semantic propertiesR "ac% ?2660: 24@.
Here "ac% taes issue +it% t%e conte5tualist idea t%at semantics must produce trut%
conditions and +%en it does not produce t%em as in ?14@ or produces A+rong ones& as
in ?13@, conte5t Aintrudes& into t%e domain of semantics in t%e form of pragmatic
inference and supplies missing information. *o, %e taes issue +it% relevance9
t%eoretic and >ecanati&s positions. )or "ac%, t%e semantic content of sentences and
t%e intended content of utterances %ave to be ept apart. *entences uttered normally
underdetermine +%at t%e speaer means but t%ere is not%ing e5traordinary about t%is
fact: semantics ends +it% t%e sentence meaning, even if it is counterintuitive orincomplete, and pragmatics begins +it% utterance meaning.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
22/37
"ac%&s middle level of meaning comprises all inds of e5pansions of completions:
t%ose t%at can be inferred from t%e conte5t, as +ell as t%ose t%at can be assumed to
conform to normal, standard, scenarios t%at t%e interlocutors encountered before or
t%at belong to common bacground no+ledge base.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
23/37
resumptive meanings are t%us #C
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
24/37
M9%euristic: A=%at&s said in an abnormal +ay isn&t normal.&
T%is %euristic +ors as a complement to t%e
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
25/37
resemblance to t%e sentence uttered. )or e5ample, in e5ample ?2@ repeated belo+, t%e
most salient meaning of "&s response is somet%ing lie ?2"a@.
?2@ $: *mit% doesn&t seem to %ave a girlfriend t%ese days.
": He %as been paying a lot of visits to Ee+ For lately.
?2"a@ *mit% may %ave a girlfriend.
n t%e 'efault9*emantics construal, ?2"a@ is t%e primary content of "&s utterance and
it is t%is content t%at enters into t%e merger representation of ?2"@. *o, semantics is
understood %ere as t%e t%eory t%at provides t%e intuitive trut% conditions of utterances,
but t%ese trut% conditions are even Amore intuitive&, so to spea, t%an t%ose of trut%9
conditional pragmatics discussed above. =e drop t%e restriction t%at one of t%e
sources of speaer&s meaning, namely t%e logical form of t%e sentence ?t%e =*@ %as
priority over t%e ot%ers and t%at t%e contribution of inference and presumed, salient
enric%ments is to be limited to embellis%ments of t%is logical form. T%is vie+,
alt%oug% intuitively correct, does not come free t%oug%. ecanati ?2660@ already
observed t%at, +it% intuitive trut% conditions of trut%9conditional pragmatics, +e need
a more Apragmatic& approac% to compositionality: composition of meaning +ill %ave
to proceed not on t%e level of t%e logical form of t%e sentence but rat%er on t%e level
of t%e semantic representation of t%e intended proposition. T%is idea is furt%er
developed in 'efault *emantics and in its merger representation conceived of as an
interaction of t%e four sources of meaning. =%at a semantic t%eory needs is an
algorit%m to s%o+ %o+ all t%e sources of information about meaning interact. $not%er
?related@ conseBuence of e5tending +%at is said in t%is +ay is t%at +e need a ne+
criterion for distinguis%ing +%at is said ?and a merger representation@ from
implicatures proper.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
26/37
conte5tualism is a position in t%e p%ilosop%y of language t%at a sentence e5presses
fully determined content only in t%e conte5t of a speec% act. T%is vie+ did not arise in
a void. ecanati ?2660, 2663@, literalism. 85amples of semantic underdetermination andsentence non9literality discussed in *ection 1.-.0 testify to t%e deficiencies of t%is
vie+. Eo+, according to >ecanati&s ?2663@ %istorical overvie+, literalism %as gone
t%roug% a Agradual +eaening&: it %as become less and less radical, until it gave rise to
stronger and stronger forms of conte5tualism.
ecanati At%e syncretic vie+&, %olds t%at
t%ere is a minimal semantic content, but on t%e ot%er %and t%ere is also intuitive
utterance content. *ince t%e latter can be freely modulated to reflect t%e speaer&s
intended meaning as assessed by t%e addressee, t%e first can be ept even Amore
minimal&: slots provided by deictic and ot%er conte5t9dependent e5pressions %ave to
be filled but t%ere is no temptation to postulate any ot%er contentious slots ust in
order to capture t%e intuitive meaning of t%e utterance: sentence9type meaning and
speaer&s meaning can be safely ept apart. $ syncretist, 8mma "org ?2660@, claims
in %er t%eory ofMinimal Semanticst%at semantic t%eory is unaffected by any
intrusion of aspects of meaning t%at are not present in t%e sentence itself. *emantict%eory is a t%eory of Aliteral linguistic meaning& and its tas is only to provide Apure&
sentence meaning. *%e assumes t%at t%e understanding of sentence meaning is
modular and s%ould be ept apart from t%e understanding of speaer&s intentions and
from any non9deductive inference. *entences %ave trut% conditions even if it +ould
not be possible to tell +%at situation +ould mae t%e sentence true. T%ere is no need
for t%e intrusion of t%e conte5tual enric%ment. Trut% conditions are not conditions of
verification as conte5tualists %ave it. $ similar orientation is presented in Cappelen
and (epore&s ?2663@Insensiti&e Semantics: a trut% condition can be produced for a
sentence even if +e are not in a position to discern possible situations t%at +ould
verify it. *o, on t%is account, Tarsi&s T9sentence ?20@ produces a trut% condition ust
in virtue of %aving a form of a T9sentence ?see "ac% 2667: 0 for criticism@.
27
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
27/37
?20@ ATom is not strong enoug%& is true if and only if Tom in not strong enoug%.
Cappelen and (epore claim t%at it is a mistae to assume t%at a semantic t%eory
s%ould account for speaers& intuitions about t%e content of t%e utterance, i.e. about
t%e speaer&s meaning. T%ey argue t%at t%ere is no strong connection bet+een t%e
content of speec% acts and t%e content of sentences and t%at t%ey s%ould be ept apart."y t%e same reasoning, semantics and pragmatics s%ould be ept apart:
G*emantics is about %o+ best to specify t%e semantic value of t%e le5ical items and
t%eir contribution to t%e semantic values of comple5 e5pressions and sentences in
+%ic% t%ey occur.
n t%e ot%er %and, +%en +e t%in abut and describe +%at people say, i.e.,
+%en our aim is to represent or articulate +%at&s said by an utterance, +e aim to
c%aracterize a speaer&s act ?t%at utterance@, and in so doing our aim is to determine
somet%ing about a particular act in a particular conte5tR Cappelen and (epore
?2663: 3@.
T%e boundary dispute +ould be simple indeed to solve if semantics and pragmatics
could be delimited in t%e +ay Cappelen and (epore suggest. Ho+ever, t%ey seem to
cut off parts of eac% discipline in order to eep t%e boundary clear: composing
sentence meaning is not a simple enterprise of combining clear9cut +ord meanings in
t%e +ay provided by t%e grammar. Eeit%er is pragmatics merely a description of +%at
people say. ost9#ricean pragmaticists aim at generalizations about t%e speaer&s
meaning t%at stem out of t%e principles of rational conversational be%aviour. Most of
t%em are also interested in t%e psyc%ology of utterance interpretation and in t%e
application of suc% principles during t%is process. *o, even leaving aside t%e debate
concerning t%e unit of +%ic% trut% conditions s%ould be predicated, it is evident t%at
one must resort to un+arranted simplifications in order to force semantics and
pragmatics into separate moulds.
*emantic minimalism gave rise to ample criticism from t%e conte5tualists. $s
+e no+ from t%e earlier discussion of >ecanati&s version of conte5tualism, t%e main
obection to t%e first version of minimalism is t%at t%ere is no evidence for t%e inds
of syntactic slots t%at %ave to be postulated in order to maintain t%e vie+ t%at syntactic
form alone accounts for t%e intuitive trut% conditions. Many of t%em seem to %ave
been postulated by its follo+ers ust in order to save t%e t%eory. )or e5ample, t%e
argument slot for t%e location for +eat%er predicates suc% as A
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
28/37
t%eoretical sense, not to mention t%at it is useless as a proposed stage in utterance
processing: meanings c%ange online, before t%e addresseeDprocessor reac%es any
proposition at all.
Conte5tualists& main obection to t%e second version of minimalism, t%e
syncretic vie+, is t%e lac of practical utility of a minimal proposition. T%e syncretists
+%o opt for verification9free trut% conditions are also attaced on t%e count ofredefining trut% conditions:
GT%is move stries me as an unacceptable +eaening of t%e notion of trut%9condition.
T%e central idea of trut%9conditional semantics ?R@ is t%e idea t%at, via trut%, +e
connect +ords and t%e +orld. ecanati ?2663:
13@.
T%e trut% conditions of syncretists do not %ave to fulfil t%is reBuirement of
Aconnecting +ords +it% t%e +orld&: t%ey only provide a formal procedure for a t%eory
of meaning. T%e debate is at present in t%e focus of attention of most post9#riceansand is liely to occupy us for some time. *uffice it to say t%at obections to +eaening
trut% conditions s%ould not be taen lig%tly.
T%ere is anot%er vie+ t%at +e s%ould consider in t%at it s%ares some
assumptions +it% t%e above versions of minimalism, +%ile reecting ot%ers. =e %ave
seen in *ection 1.-.0 t%at Kent "ac% advocates a clear9cut boundary bet+een t%e tass
of semantics and t%ose of pragmatics. He does so by claiming t%at t%e semantic
analysis of a sentence need not result in a trut%9conditional content: not every
declarative sentence e5presses a proposition, even if it is free from inde5ical
e5pressions.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
29/37
currently in t%e forefront of t%eorizing, it is t%e ne5t logical step from t%e conte5tualist
stance represented in trut%9conditional pragmatics and ot%er approac%es t%at allo+
free, top9do+n pragmatic enric%ment.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
30/37
(evinson&s notion of default is rat%er strong. *uc% defaults are, as +e sa+ in *ection
1.-.3, attac%ed to +ords, p%rases, or ot%er sub9sentential e5pressions and as a result of
t%eir locality, t%ey are often cancelled as conversation develops. $lt%oug% it is
generally agreed t%at t%e addressee uses some s%ortcuts t%roug% costly inference,
t%ere is no e5perimental evidence in support of suc% strong local defaults ?see e.g.
Eovec and *perber 2660@.
$s far as t%e conte5t9bound vs. conte5t9free opposition in delimiting defaults is
concerned, Horn&s ?2660: 093@ claim supporting #C
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
31/37
roposals concerning default interpretations are ample in semantics,
pragmatics, logic, and p%ilosop%y, to mention only "ac%&s ?10@ default reasoning,
$s%er and (ascarides&s ?266-@ r%etorical structure rules, constraints of ptimality9
T%eory ragmatics ?"lutner and eevat 2660@, as +ell as t%e vie+s discussed above:
(evinson&s presumptive meanings, Horn&s acno+ledgement of #Cecanati&s
automatic modulation, and Jaszczolt&s cognitive and social9cultural defaults. T%ere isno uniBue Adefault model& +e can oppose to t%ose in favour of Anonce9inference&.
=%ile t%e rationale for suc% salient meanings is strong, t%e properties of suc%
interpretations are still in need of furt%er investigation, especially +it% respect to t%e
follo+ing properties:
?i@ availability +it%out conscious inferenceL
?ii@ s%orter time reBuired for t%eir processing as compared +it% t%e time in +%ic%
meanings inferred from t%e assumptions about speaer&s intentions are
processedL
?iii@ cancellability ?called defeasiblit$@L
?iv@ locality, i.e. t%eir availability pre9propositionallyL
and
?v@ conte5t9independence.
nly +%en t%e terms Adefault&, Asalient&, Aunmared&, Apresumed& are paired +it% clear
e5planations in terms of semantic and pragmatic c%aracteristics and t%e psyc%ology of
processing, +ill +e be able to assess t%eir raison d&tre.
-. *ummary: T%e significance of t%e boundary dispute
T%e current state of t%e debate concerning t%e interface bet+een semantics and
pragmatics is t%e ups%ot of t%e revolutionary period in t%e study of meaning no+n as
radical pragmatics, aided by t%e vie+s of ordinary language p%ilosop%ers. T+o
relatively separate disciplines, t%e formal study of sentence meaning and t%e relatively
informal study of t%e properties of speec% acts became more and more intert+ined as
a result of t%e adoption of semantic underdetermination and t%e admittance of
pragmatic inference about t%e speaer&s intentions, as +ell as some ot%er conte5t9bound information, into t%e semantic content. T%is facilitated t%e s%ift of t%e centre of
attention from t%e sentence to t%e utterance. Ho+ever, t%e direction of c%ange %as not
been steady t%roug%out t%e past t%ree decades. T%ere are attempts to eep semantics
and pragmatics apart eit%er t%roug% denying t%at semantics %as to provide
propositions and %ence trut%9conditional content, or t%roug% eeping t%e obectives of
semantics and pragmatics apart and stressing t%e t%eoretical utility of t%e sentence&s
trut% conditions, lie minimalists of t%e syncretic flavour do. T%e dominant
orientations are %o+ever various forms of conte5tualism. T%is state of affairs is
undoubtedly aided by t%e overall desideratum to stay fait%ful to speaers& intuitions
about meaning and to t%e vie+ t%at t%e aim of semantic t%eory is to cater for t%ese
intuitions. =%et%er conte5tualism +ill retain its po+er, succumb to minimalism, or
-1
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
32/37
evolve into a radical form of occasion9meaning of meaning eliminativism remains to
be seen.
0. >eferences
$s%er, Eic%olas ! $le5 (ascarides 266-.Logics of *on&ersation. Cambridge:Cambridge University ress.
$tlas, Jay '. 144. Eegation, ambiguity, and presupposition. Linguistics and
+hilosoph$1: -219-7.
$tlas, Jay '. 14. Ho+ linguistics matters to p%ilosop%y: resupposition, trut%, and
meaning.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
33/37
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
34/37
van 'eemter, Kees ! *tanley eters [email protected]$ and 9nderspecification.
*tanford, C$: C*(< ublications.
Dowty, David R., Robert E. Wall & Stanley Peters 1981.Introduction to Montague
Semantics. Dordrecht: D. Reidel.
Ducrot, Oswald 1972.Dire et ne pas dire. Paris: Hermann.
#eurts, "art 1. *calars.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
35/37
Horn, (aurence >. 2667. T%e border +ars: $ neo9#ricean perspective.
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
36/37
(evinson, *tep%en C. 14. Minimization and conversational inference. eader ! 'yer. )ourt% edition.
Musolino, Julien 2660. T%e semantics and acBuisition of number +ords: ecanati, )ran[ois 1-.)irect 7eference: 0rom Language to 1hought. 5ford:
"lac+ell.
Recanati, )ran[ois1994. Contextualism and anti-contextualism in the philosophy of
language. In: S. L. Tsohatzidis (ed.). 1994. Foundations of Speech Act Theory:Philosophical and Linguistic Perspectives.London: Routledge, 156-166.
>ecanati, )ran[ois 2661. =%at is said. S$nthese 12: 4391.
Recanati, )ran[ois2002a. Unarticulated constituents.Linguistics and Philosophy25:
299-345.
>ecanati, )ran[ois 2662b. 'oes linguistic communication rest on inference.Mind
and Language14: 1639127.
>ecanati, )ran[ois 266-. 8mbedded implicatures.+hilosophical +erspecti&es14:29--2.
-7
-
8/12/2019 Jaszczolt. K - Semantics and Pragmatics - The Boundary Issue
37/37
>ecanati, )ran[ois 2660.Literal Meaning. Cambridge: Cambridge University ress.
>ecanati, )ran[ois. 2663. (iteralism and conte5tualism: *ome varieties.