What Is A Logical Constant
What Is A Logical Constant
What Is A Logical Constant
Theinferencemarkerviewi
MaraJ.Frpolli
DepartmentofPhilosophy
UniversityofGranada(Spain)
1. Whatislogic?
Mymainaiminthepagesthatfollowistoofferacharacterizationoflogical
constantstakingwhatwe,speakers,dowiththemasthepointofdeparture.Thereare
several definitions of logical constants, but none of them include a comprehensive
account of their meaning, dealing with their syntactic features, semantic value and
pragmaticroleii.Noneoftheaccountsoflogicalconstantsavailableintheliterature
showshowthedistinctaspectsoftheirmeaningarerelatedtotheusegiventothemby
thespeakersandtothegoalsoflogicalreasoning.
Iwillnotdiscussthedifferentexistingdefinitionsoflogicalconstantsinany
detail.They,andalsothestandardcriticismsthatcanbemadeagainstthem,arewell
knowntospecialists. Forspecialists andnonspecialists alike, itis important tobe
awarethat,asWarmbrodsaysinarecentpaper,thereisasyetnosettledconsensusas
towhatmakesatermalogicalconstantorevenastowhichtermsshouldberecognized
ashavingthisstatus(Warmbrod1999:503)iii.
Warmbrod describes the present situation; I, on the other hand, would like to
analyzesomeofitscausesandofferaproposal.Theunsatisfactorysituationconcerning
logicalconstantscanbeattributedtotwomaincauses:(i)thecommonunderstandingof
therelationsbetweenmathematicsandlogic,and(ii)thecommonunderstandingofthe
relationsbetweennaturallanguageandlogic.Contrarytothestandardviewduringthe
pastcentury,logicandmathematicspursuedifferentaims.Themostvisiblepointof
contact between the two disciplines is that they are both considered as formal
sciences;itremainstobeseenwhetherformalhasthesamemeaninginbothcases.
Itisafactthatmodernlogichasbecomemoreakintomathematicaltheoriesthantothe
study of inferential patterns in natural languages. Nonetheless, the legitimate
methodologyofapplyingmathematicaltoolstothestudyoflogicandlanguagesdoes
notsupporttheillegitimateidentificationoftheaimoflogicwiththatofmathematics.
1
Undoubtedly, modern logic developed during the second half of the XIX century
thanks to the work of mathematicians as Jevons, Boole, Peano and Frege, among
others. It evolved from the previous enterprise of applying algebra to the studyof
natural language, which was already a revolutionary theoretical enterprise (see
Goldfarb(1979)).Butmodernlogicwasbornasanindependentsciencewhenlogicians
understoodthepreviousalgebraicrelationsnotasrelationsonsetsbutasrelationson
concepts and conceptual contents. Logic deals with judgeable contents, with
propositions;propositionsarethebasicelementsofarguments,andevenwhenweuse
artificialcalculiinwhichpropositionalstructuresarerepresentedatthesyntacticlevel,
logicalrelationsarenotheldbetweensyntacticitemsasgrammarunderstandsthem,but
betweenthecontentsofsomeofourspeechacts.Uninterpretedsentencesarepurely
syntacticentities,andpurelysyntacticentitiesarenottruthbearers.Thus, afortiori,
uninterpretedsentencescannotbewhatlogic,thescienceofvalidarguments,isabout.
Logicalconstantsarepropositionaloperators;propositions,statements,thoughts
theseexpressionsareequivalent
all
aretheargumentsoflogicalconstants,andnotthe
argumentsaboutanysubjectmatterisright.Thus,theapparentsemanticfeatureofthe
topicneutrality of logic is in fact a syntactic feature about the category of logical
constants,andamountstosayingthatlogicalconstantsarehigherorderfunctions.I
proposetosubstitutethesetraditionalwaysofdescribinglogicalconstantsandothersof
thesamekindbythemoreprecisequalificationofbeinghigherorderpredicableswith
0adicpredicablesasarguments.Thatlogicalconstantsarehigherorderpredicateshas
beenarguedforbyC.J.F.Williams(Williams1992b).
Nevertheless,syntaxdoesnotprovidetherightdemarcation,whichissomething
that medieval logicians already knew. Interjections, exclamations, adverbs, and
punctuationmarksaresyncategorematawithoutbeinglogicalconstants iv;andmanyof
themarealsotopicneutral.Syntaxisnotarelevantlevelofanalysis.Logicalconstants
arelogicallyrelevantexpressionsnotbecauseoftheirsyntacticfeaturesbutbecauseof
theroletheyplayinthegeneraltaskofdrawinginferences.Wordssuchasif,not,
or and the rest have attracted the interest of logicians because the speakers of a
languageusethemessentiallyintheirexplicitinferentialacts. Itistheirfunctionin
ordinaryinferentialpracticesthatmakesthemlogicallyinterestingterms.Duetoallof
this,theprojectofdefininglogicalconstantsexclusivelybyattendingtotheirsyntactic
propertiesiscompletelymisconceivedandassuchcondemnedtofailurev.
2. Inferentialmeaningandinferencemarkers
Myproposalistobringlogicbacktolanguage,itsnaturalhome,andtoplacethe
philosophy of logic within the philosophy of language vi. What logic is cannot be
determinedbybackingoutoftheinferentiallinguisticpracticesofhumanbeings,and
thesamecanbesaidofthetaskofidentifyingthefeaturesthatmakeatermalogical
constant. The semantics and pragmatics of logical words provide us with more
promising insights than the misguiding clues offered by the alternative syntactic
approach.
What is the semantic value of a logical constant? Generally speaking, the
semantic value of an expression is the component it contributes to the proposition
expressedbythesentencesinwhichthetermoccurs.Wittgensteingavetheappropriate
answertothepreviousquestion:none.Logicalconstantsarenotnamesofanythingand
theirsemanticfunctioncannotbetoaddafurtherelementtoacontent vii.Thisdoesnot
meanthattheyhavenomeaning.Semanticvalueisaveryspecificnotion,andthemore
informaltermmeaningismuchbroader.TheWittgensteinianclaimisoftenknownas
Logicalexpressivism.Thereisnogeneralagreementastothecredibilityoflogical
expressivismbut,inspiteofthetheoreticalprotests,everybodyfollowsitinpractice.
Considerthecustomarywayinwhichoneinterpretssentencesandformulaeinformal
semantics. One does it by defining an interpretation that attaches objects in the
Universetotheindividualconstantsintheformulae,setsofobjectsintheUniverseto
the predicates in the formulae, sets of ordered sets of objects to the relational
expressions.Butlogicalconstantsarenotinterpretedthisway.Onemightretortthat
logicalconstantsdonotneedinterpretationpreciselybecausetheconstancyoftheir
meaning.Thisispartofthereason,indeed,butneitherthecompleteanswernorthe
mostrelevantpartofit.Numerals,forinstance,arealsoconstants;onealreadyknows
theirmeaningandthusthereisnoneedtodecide,ineachnewmodel,whichentities
wouldcorrespondtothem.Theentitiesthatcanbetheirvaluesareofthesametypeas
the entities that are the values of the rest of expressions, i.e. either objects in the
Universeorsetsoftheseobjects.Thecaseoflogicalconstantsisdifferent.Theydonot
correspondtoobjectsofanykind;givenasetofformulaeandaninterpretation,they
help tofind thetruthvalue oftheformulae according to theinterpretation without
adding new entities to the model. This is their specific function and, correctly
understood,thisisthecoreoflogicalexpressivism.
Thepragmaticroleoflogicalconstantscanbeeasilyunderstoodifweconsider
thefollowingillustration.Thepropositionexpressedbyasentencelike()inastandard
context,
()MydaughteriscalledVictoria,
stands in varied inferential relations with other propositions, as for instance, those
expressedbysentences(),(),and()inthesamecontext,
()Ihaveadaughter
()Victoriaisagirl
()Victoriaisahumanbeing.
The propositions expressed by () and () together form the following material
4
inference:
(I)
Victoriaisagirl;Victoriaisahumanbeing.
Thetruthof()followsfromthetruthof().Byasserting(),oneiscommittedto
assentto().
Now, if for some reason one were interested in stressing the inferential
connections between ()and(),onewouldhavetodisplaythe implicit, meaning
based,transitionfrom()to()asaruleofinference,eithersingular,as(R1):
(R1)IfVictoriaisagirl,thenVictoriaisahumanbeing,
orgeneral,as(R2):
(R2)Ifsomebodyisagirl,thensheisahumanbeing.
When the rule of inference is added to the previous material inference, it
becomesaformalinference,asin
(II)
What is the difference between inferences (I) and (II)? They both have the same
conclusion,thatVictoriaisahumanbeing,andthesamepremise,thatVictoriaisagirl.
Theconditionalin(II)isnotafurtherpremise,butaprincipleofreasoning.Bolzano,
Frege and Peirce already paid attention to this distinction between premises and
principles of reasoning. Premises are claims; they are asserted propositions,
judgements.Principlesofreasoningarenotclaimsbutrules.Overlookingthispoint
would lead us to Carrolls paradoxviii. If (I) and (II) share their premise and their
conclusion,inwhichsensearetheydifferent?Theanswerisobvious:theyaredifferent
becausein(II)theprincipleofinferenceusedinbothisexplicitlydisplayed.Todisplay
it,toindicatethatitisaprincipleandnotaclaim,onehastousetheappropriatekindof
words:logicalconstants.In(II)thewordsif,thenservetomakeexplicitan
inferentialconnectionbetweentheantecedentandtheconsequent.Whentheyoccurin
asentence,thesentenceinquestiondoesnotexpressapropositionbutarule.Thesame
effectmighthavebeenachievedbyinsertingthereforebetweenthepremiseandthe
conclusion. Therefore is another logical constant. One might think that with this
explanation, we are committing the sinthat Quine seemed to findat the origin of
modern modal logic, i.e. the sin of confusing use with mention. We are not. The
differencebetweenobjectlanguageandmetalanguageisnotasstraightforwardinthe
actualuseofnaturallanguagesasitisinformalartificiallanguages.Butinanycase,
wearetalkingaboutpropositions,notaboutsentences,andthusthedistinctiondoesnot
apply. The arguments of if, then , understood as means of stressing an
inferentialrelation,arepropositionsandnotsentences,andexactlythesamehappensin
thecaseoftherefore.Nowweareinthepositionofproceedingtothepragmaticrole
oflogicalconstants:speakersusethesewordstodisplaythestructureofaninference.
Logical constants are added to material inferences to demonstrate their status as
inferences; they are not essential to carry out inferential movements but their
involvementbecomesindispensableinordertopresentinferentialconnectionsbetween
propositionsasexplicitinferences.Itisonlywhenwewanttomakethepresenceofan
inferenceexplicitthattheybecomeuseful.
Tobeabletomakeinferentialconnectionsexplicitanexpressiondoesnotonly
require having inferential meaning. All concepts have inferential meaning to some
extent;theinferentialconnectionsbetweenconceptsjustifythematerialinferencesin
whichtheyareinvolved.InthissensetheproposalIamputtingforwardhere,whichI
will call the inferencemarker viewix, goes further than Gentzens and Prawitzs
views. The core of the inferencemarker view is not that the meanings of logical
constantscanbegivenassetsofrules,introductionandeliminationrules,butrather
thatthepragmaticsignificanceoflogicalconstantsistobringanimplicitinferenceinto
theopen.
This pragmatic role does not imply that logical words always indicate valid
inferences.Thespeakeruseslogicalwordstosignalthat,fromherpointofview,the
relevantpropositionssheisexpressingaresomehowinferentiallyconnected.Butshe
mightpresentasaninferenceonethatisinvalid,justbecauseshemightbewrongabout
someaspectsofthecaseinhand.Thisdoesntunderminemygeneralclaimaboutthe
roleoflogicalconstants.If,andtheotherconstants,haveexactlythesamemeaning
whentheyappearindeductive,inductiveorsimplyinvalidinferences.Inductiveand
deductive inferences havedistinctproperties,buttheydonotaffect themeaningof
constants.Andthesamecanbeappliedtoinvalidinferences.Thepragmaticroleofif
6
isconstantacrossitsvarioususes.Ifmeansthesamewhenitappearsinaninstanceof
thefallacyofaffirmingtheconsequentaswhenitappearsinaninstanceofourreliable
ModusPonens.Itisbecauseitsmeaningdoesntchangethatweclassifytheformer,
unlikethelatter,asfallacious.
Alllogicalwordssharethesamegeneralpragmaticrole,butdifferentlogical
wordshavedifferentspecificinferentialmeanings.Ineachcasetherulesthatgovernthe
relevant inferential movements depend on the particular meaning of the logical
constants actually used. Thus, if, not and or codify different inferential
entitlements,thatinformalcalculiarerepresentedbydifferentsetsofrulesofinference
thatdisclosethecircumstancesandconsequencesoftheuseoftheiruse.
I will propose a general definition of logical constants that include their
syntactic status, their semantic characterization and their pragmatic role. However,
unlikemanyproposalsthattakesyntaxasthepointofdeparture,mydiscussionofthe
subjectwilldepartfrompragmatics.Itakewhatwedowithwords,withlogicalwords
inthiscase,asthefoundationlevel.
3.Syntax,Semantic,andPragmaticsofLogicalWords
Thedefinition:
[DEF]
predicablesasarguments.Theydontnameanykindofentitybutratherarenatural
languagedevices formakinginferentialrelations amongconcepts andpropositional
contentsexplicit.
DEF involves a syntactic claim, that logical constants are higherorder; a
semanticclaim,thattheydonotname;andapragmaticclaim,thatbyusingthema
speakersignalsthepresenceofaninference.Thesemanticclaim,logicalexpressivism,
andhasbeendefendedbyJohnBuridanandAlbertofSaxonyintheMiddleAges,and
byWittgenstein(1922)andBrandom(1994)intheXXcentury.Thepragmaticview
hasbeensupportedbyRamsey(1928),Ryle(1956)andBrandom(1994),andIhave
calledittheinferencemarkerview.
Beinghigherorderisonlyoneofthenecessaryconditionsforbeingalogical
constant.Andthesamehappenswiththesemanticaspect.Inlanguage,therearemany
differentexpressionsthatdontnameanythingx,andshouldntbecataloguedaslogical
constantsforthisreasonalone.Butaslogicisthescienceofvalidarguments,logical
constants are essentially inferencemarkers; particular logical constants express the
specificinferencialmovementstheyallow.
A0adicpredicableisapredicablewith0argumentplaces,i.e.aproposition.
Thatlogicalwordsarehigherorderpredicablesthathavepropositionsasarguments
shouldbeobviousifonerecallsthatthebasicnotionoflogicisvalidity xi,thatvalidity
is a property of arguments or of inferences, and that arguments and inferences,
consideredinanobjectivesense,aresetsofpropositions.Inferences,infact,shouldbe
consideredasmovements,transitionsfromsetsofpropositionstoaproposition.They
haveadynamicsense,somethingthatfromdifferentpointsofviewhasbeenrecently
stressedbyDubucsandMarion(2003),byMartiLf,andbySundholm,amongothers.
Itisbecauseaninferenceismovement,thatgenuinelogicalconstantshaveanaspectof
theirsignificancethatisdynamic.Theyshowinferentialbridgesbetweenconceptsand
propositions.
An immediate objection to my definition is that, although it fits sentential
connectiveswell,itignoresidentityandfirstorderquantifiers.Iwilldiscussthispoint
later.ButapossiblewayoutisprovidedbythefactthatDEFcanhavetworeadings,
oneweakerthantheother.Theyarethefollowing,
[DEF]weak Logical constants are higherorder predicables that may admit 0adic
predicablesamongtheirarguments.Theydontnameanykindofentitybutratherare
naturallanguagedevicesformakingexplicitinferentialrelationsamongconceptsand
propositionalcontents.
[DEF]strong Logicalconstantsarehigherorderpredicableswhoseargumentsare0adic
predicables.Theydontnameanykindofentitybutratherarenaturallanguagedevices
formakingexplicitinferentialrelationsamongconceptsandpropositionalcontents.
[DEF]weakpredicateslogicalconstanthoodoftypes,while[DEF]strongpredicatesit
oftokens.Undertheformer,atype,sayaquantifier,isalogicalconstantif,among
othercharacteristics,ithastokensthatarefunctionsof0adicpredicables.Propositional
8
quantification would be an obvious case that would provide quantifiers with the
required feature. In any case, quantifiers are not an homogeneous kind, and it is
reasonabletoassumethatdifferenttypeswithdifferentfunctionsmaybedistinguished.
Underthestrongerdefinition,whatisclassifiedasalogicalconstantisatoken,i.e.,a
particular instance of a type together with its particular aspects. If one selects
exclusively[DEF]strong,itmakesnosenseaskingwhetherquantifiersoranyotherkind
ofexpressionareorarenotlogicalconstantsornot.
Fortunately,itisnotnecessarytochooseoneofthetwooptionsandrejectthe
other.Wecanassumethecharitablepositionofclassifyingtypesaslogicalconstantsin
aweaksenseif,andonlyif,theyhavetokensthataresoinastrongsense.
DEF assembles three aspects that are individually necessary and jointly
sufficientforbeingalogicalconstant.Itisinterestingtoseethatthetwofirstaspects
followfromthepragmaticroleanditsrequirements.Removingsomeaspectswhile
leavingothers,wegetbroadersetsofexpressionsstillrelatedtotheclassoflogical
constants.Table1showsdifferentexpressionswithdifferentsetsofproperties.
HIGHERORDERPREDICABLES
MonadicA
BinaryB
1. With 0adic Negation, epistemic operators, Conjunction, disjunction,
functions
arguments
binary
quantifiers, nonstandard
(most, seldom, a
few)binaryquantifiers
AlloperatorslistedinB.1
meaning
3. With inferential negation
Disjunction, conditional,
significance as an
binaryquantifiers
inferencemarker
TABLE1
AllexpressionsinTable1haveatsometimebeenconsideredlogicalconstants,and
therearereasonsforthis.Allofthem havefeatures alsopossessedbythegenuine
logicalterms;butmostofthemfallshortofhavingthemalltogether.Accordingtomy
definition,onlynegation,disjunction,conditionalandbinaryquantifiersare,astypes,
logicalconstantsinaweaksense,becausesomeoftheirtokenshavepropositionsas
argumentsandactasexpressiveinferentialmarkers.Therestoftheexpressionslisted
sharesomeaspectswiththem,buttheyarenotlogicalterms.Logicalconstantshaveto
befunctions;ifIamright,thisrules outtruthvalues,andtheuniversal andempty
classes,consideredlogicalconstantsbymostinvariantists.Thestatusofanexpression
asaninferencemarkerrequiresittobehigherorder.
Mydefinitionrulesout:
(i)
(ii)
(iii)
(iv)
Monadicsentenceformers,suchasmonadicquantifiers
Nevertheless,
(v)
(vi)
(vii)
DEF doesnt imply that quantifiers are not logical constants. Rather, it
10
distinguishesdifferentkindsofquantifiers.Monadicquantifiersdontactas
inferencemarkers,butbinaryquantifiersusuallydo.Thisdoesnotmeanany
rejectionofFreges(1884)account.Fregerightlyunderstoodthenatureof
numerical expressions as higherorder concepts, and correctly defined
existenceasanexpressionofquantity.Numericexpressionsandexistence
are monadic higherorder functions whose arguments are concepts. They
indicate sizes of concepts extensions. Nevertheless, they dont act as
markersofinferencesandsotheyarenotlogicalconstants.
(viii)
Althoughpoints(i)(viii)willnotbecompletelyclarified
asinglepaperwouldnotbe
enoughforsuchataskthatwouldeventuallyinvolvearevisionofalltermsthathave
ever been proposed as logical, some comments would help. Point (i) is hardly
controversial: there are many authors who do not count first order identity or
membershipamonglogicalconstants(see,forinstance,Peackocke(1976;p.234andp.
237)andWarmbrod(1999)).Point(ii)referstoarelevantissue,thatsomeusesofthe
words that are commonly accepted as logical terms actually have a combinatorial
function. Some uses of negation, conjunction and disjunction have predicative
expressionsasarguments.Whenthishappens,theirfunctionishelpingtobuildcomplex
conceptsoutofsimpleones.Complexconceptssuchasunhappy,honestpolitician,
married woman, homeless are composed of more basic concepts by means of
negation and conjunction. Although nowadays this function of concept construction
seemstohavebeenforgotten,medievallogicianswereperfectlyawareofit(seeKlima
2006). Point (iii) implies that some monadic sentential functions are circumstance
shifting operators, operators that, although dont contribute a component to the
propositionexpressedbythesentenceinwhichtheyoccur,arerelevanttothetaskof
evaluatingthepropositionsthatactastheirarguments.Modal,epistemicandtemporal
operatorshaveexpressivemeaning,buttheydontworkasinferencemarkers.Saying
thatanexpressionhasexpressivemeaningisonlyanegativecharacterization.Inthe
positive,therearedifferentwaysinwhichanexpressionmayhaveinferentialmeaning.
11
Definingmodal,epistemicandtemporaloperatorsascircumstanceshiftingoperators,
i.e. operators that indicate in which circumstance the propositions that are their
arguments must be evaluated, is a way of specifying the negative characterization.
Modal,epistemicandtenseoperatorsareallofthis kind.Point(iv)demonstrates a
significant feature of some quantifiers. All of us consider the Fregean treatment of
existence in (Frege 1884) as the first step towards the correct understanding of
quantifiers. In 53, Frege says: In this respect existence is analogous to number.
Affirmationofexistenceisinfactnothingbutdenialofthenumbernought.Although
thisisaccurateinrelationtoexistence,itisnotinrelationtogenerality.Itisrelevant
heretoacknowledgethatquantifiersmaybemonadichigherorderoperatorsorbinary
higherorderoperators.Themonadicexistentialquantifierindicates,asFregesaw,that
theextensionoftheconceptthatisitsargumentisnotempty.Butthemonadicuniversal
quantifierhasaslightlydifferentmeaning:itindicatesthescopeoftheconceptthatis
its argument. In natural languages, both quantifiers standardly are binary operators.
Fregealsosawthis:Itmustberemarkedthatthewordsall,any,no,someare
prefixed to conceptwords. In universal and particular affirmative and negative
sentences,weareexpressingrelationsbetweenconcepts,weusethesewordstoindicate
thespecialkindofrelation.Theyare,thus,logicallyspeaking,nottobemoreclosely
associatedwiththeconceptwordsthatfollowthem,butaretoberelatedtothesentence
asawhole.(Frege1892:48)Whentheyarebinaryoperators,theirmeaningisdynamic
and they indicate an inferential connection between the two concepts that are their
arguments.FromthequotedtextitfollowsthatFregethoughtthatbothuniversaland
existentialquantifiershaveusesinwhichtheyarebinary.Itotallyagree.Inartificial
languageswearefreetodefinethestatusoftheoperatorsthatweintroduce,butin
natural languages we are not. These operators, as the rest of our expressions, are
supported by the tasks the speakers use them for, and standardly the existential
quantifierisusedasamonadicoperatortoexpressthenonemptinessofanextension,
and the universal quantifier is used as a binary operator to express a principle of
reasoning.Iwillsaysomethingmoreonthistopiclater.
4.GenuineLogicalConstants
12
ThemainlinesthatdefinethepositiononlogicandlogicalconstantsIstandfor
havebeenalreadyintroduced.NowIwillapplythemtothenotionsthathavebeen
commonly accepted as logical constants, and wewill then see whether they are so
accordingtomydefinitionandwhichoftheirusesjustifyustoclassifythemassuchin
astrongsense.Inallcases,asinferencesareattheirbasetransitions,genuinelogical
constantsmustpossesssomedynamicaspectsintheirmeaningthatcanaccountforthe
tasktheyhavetoperform.
Negationcanbecombinedwithpredicablestoformnegativepredicables,asin
(1),
(1) Themeetinghasbeenunsuccessful,
andalsowithpropositionsxii,asin(2)and(3),
(2) ItisnotthecasethatVictoriaisathomexiii,
(3) Noteverycustomerwassatisfied.
In(2),thenegationoperatorItisnotthecasethatfunctionsasawayofexplicitly
vetoingtheenteringofacontent,thatVictoriaisathome,intotheacceptedrepertoireof
propositionsreadytoactaspremisesininferentialacts.Inthiscase,itsforceisthe
oppositetoassertionsxiv and,asanoperator,itistheconverseoftruthinitsuseasa
sentential operator, i. e. it is the converse of It is true that. In (3), the scope of
negationisagainthewholesentence,anditdirectly applies tothesentences main
operator,i.e.thebinaryquantifier.Itsfunctionisblockingtheinferentialmovement
represented,aswewillsee,bythequantifier(togetherwiththehiddenconditional).It
doesntprohibitit,butonlysuspendsit.Inthesetwolatteruses,negationisalogical
constant.
Conjunction and disjunction are less similar than the standard view has
traditionallyregardedthem.Theyarealikeinthattokensofthemaremeanstobuild
complexpredicates,tokensofbothcanactasbinarysententialfunctions,andtheylack
representationalmeaning.Theyshare,sotospeak,someoftheirsyntacticandsemantic
aspects. They are similar in that they both may have 0adic predicables, i.e.
propositions,andalsonadic(n>0)predicablesasargumentsandinthattheymaybe
13
usedonnadic(n>0)predicables toconstructcomplexnadic(n>0)predicables.
Conjunction ina sentence such as there are alligators that are tame and shyand
disjunctioninasentencesuchasallintegersareeitheroddorevenhavetheafore
mentionedconstructivefunction.FromthepragmaticpointofviewIamdrawingon
here, nonetheless, they have differences that are worth noticing. In particular,
disjunctionhasfunctionsthatconjunctionlacks
Whendisjunctionappearsasagenuinesententialoperator,asin(4),
(4) Either her husband abandoned her and she became an alcoholic or she
becameanalcoholicandherhusbandabandonedher;Imnotsure,
itproposestwopossiblestatesofaffairs,orcoursesofaction.Itdisplaysapossible
bifurcationintheinferencethatahearermaydrawwhenpresentedwith(4).Whenused
with this function, it can be seen as a binary circumstanceshifting operator. The
propositionalargumentsofdisjunctionarenotsaidtobetrueintheactualworldbut
theyratherareaskedtobealternativelytakenastrueandcarrytheinferenceon.When
disjunction takes the force of a conditional, as it happens in (5), it is a standard
circumstanceshiftingoperator,
(5) Sheiseitherathomeoratwork.
Inthestandardtruthfunctionalaccount,disjunctionandconjunctionareinterdefinable
vianegation;but,letmeinsist,itistoorestrictingtoidentifythecontributionoflogical
words to the communicational acts in which they occur with their truthfunctional
meaning.Onlywhendisjunctionisusedwiththeforceofaconditionalshoulditbe
understoodasagenuinelogicalconstant.
Conjunctiondoesnothavetheseconditionaluses,forinfactitdoesnotpossess
any dynamic aspect in its meaning, i.e. any aspect that suggests a movement, a
transitionbetweentwocontents.Whatisessentialtologicalconstanthoodisbeingused
todisplayaninference.Inferencescanbedrawnwithouttheinvolvementoflogical
words,butwithoutlogicalwordswecouldntputforwardinferencesasinferences.On
thisscore,conjunctionisnotalogicalconstant.Conjunction,asdisjunction,canactasa
means of building complex predicables. It is worth highlighting this feature of
conjunctionandconnectingittomyformerclaimthattheexistentialquantifierisnota
14
logicalconstant.Existentialsentencesinnaturallanguages,suchas
(S1)Therearepoliticiansthatarehonest,
arestandardlytranslatedintothepredicatecalculusbymeansofanexistentialquantifier
andaconjunction.Thefunctionofconjunctioninthesecasesisconvertingthetwo
previous predicables, honest and politician in (S1), into a single complex
predicable, politician thatis honest,abletobethesole argumentofthemonadic
quantifier.Itisenlighteningtocontrastthissituationwithwhathappensinthecaseof
universalquantifiers.Innaturallanguages,universalquantifiersarestandardlybinary
operators,andgeneralpropositionsarestandardlytranslatedintothepredicatecalculus
usingconditionals.Howeverconditionals,unlikeconjunctionanddisjunction,cannotbe
used to construct complex predicables; conditional concepts do not exist. The
conditionalparticleservespreciselytoshowtheinferentialtransition.Forthisreason,
binaryquantifiersandconditionalsarelogicalconstants,andexistentialquantifiersand
conjunctionsarenot.
Conjunction is a peculiar operator since it is semantically and pragmatically
dispensable.ThereisnothingthatA&BpossessesthatthesimplejuxtapositionofA
andBlacksxv;norevenitstemporalandcausalaspects,whichtheGriceantraditionhas
placedinwhatmightbepragmaticallyimplicatedbyconjunction.Letusconsiderthe
followingexamples,
(6) Victoriafelldownthestairs.Shegothurt
(7) Thecustomersprotested.Shewasfired
(8) TheSunroseoverthehorizon.Theairwarmedup.
(6) suggests a causal link, (7) succession of events, and (8) simultaneity. If we
substitutedconjunctionfortheperiod,theresultwouldbethesame.
Theonlyrealtaskthatconjunctionperformsissyntactic.Amerejuxtaposition
ofsentencesisnotasentence,andsometimesweneedtheinformationgatheredina
single sentence, as for instance when this information is itself a component of an
operator with wider scope. The two sentences in (8) cannot be, as they stand, the
antecedentofaconditional.Itisnecessarytojointhemtogetherinasinglesentenceto
obtainawellformedconditionalsentence.
15
(8*)IftheSunroseoverthehorizonandtheairwarmedup,thereisachance
thattheywillsurvive.
4.1Hypotheticalsandhypotheticalvariables
Fortunately,conditionalsandquantifiersaremuchmoreinteresting.Andmuchmore
complex.Therearemanytypesofconditionals,whichareusedfordifferentpurposes,
and some of their tokens are genuine logical constants. But others are not. In the
standardinterpretationofthefollowingexamples,conditionalsarenotlogicalconstants,
(9) Ifitisraining,itsrainingheavily
(10)
Ifsheisinnocent,Iwilleatmyhat
(11)
IfthePresidenttoldthetruth,IamthePope
(12)
IfhehadntkilledthePresident,somebodyelsewouldhave.
Allmenaremortal
(14)
MostSpaniardslovegarlic
(15)
FewBritonsenjoycooking
Victoriaisagirl
(24)
Victoriaisahumanbeing
orinourknowledgeofthesurroundingword,asinthetransitionfrom(25)to(26),
(25)Victoriaisathome
(26)ShehasmissedherEnglishlessons.
17
the antecedent false. But again, mode doesnt have any effect on the status of a
conditionalasalogicalconstant.
Allthisisstillveryvague,buttheprincipalideaundertheinferencemarker
view,thatIamproposinginthispapershouldbeclearbynow. Tosumup,logical
constanthoodisafunctionalconcept.Itappliestotokensofexpressionsdependingon
theroletheyperform.Thecentralnotionoflogicisnottruthbuttruthpreservation,and
truthpreservation,i.e.,validity,isapropertyofarguments,i.e.ofsetsofpropositions.
Thisisthestandardexplanation,andthecorrectone.Myproposaltakesitseriouslyasa
guideintotheinquiryaboutlogicalconstants.Andtheresultisthat,inastrongsense,
onlysomeusesofnegation,disjunction,conditionalandquantifiersaregenuinelogical
constants,their types being logical constants inaweaksense.Inadditionthereare
several kinds of expressions that play a role in the practice of drawing inferences,
althoughtheirfunctionisnotpresentinginferencesassuch.Thesekindsaretypically(i)
operators on predicative expressions that help forming complex predicables out of
simplerones(someusesofnegation,conjunction,anddisjunction),(ii)propositional
operators, both monadic and binary, and among them, (ii. a) circumstanceshifting
operators(modalandtenseoperators,forinstance),and(iii)binaryfirstorderidentity
andmembership.Allthesekindsdeservethelogiciansattention,althoughfordifferent
reasons.
Ifthisconclusionsoundstoounpalatablethereisstillthepossibilityofrelaxing
therequirementsandconsideringanyoperatorwitharelevantroleinthegeneraltaskof
drawinginferences as alogical constants.This would allow adding theexpressions
described in (i), (ii) and (iii). Still, the perspective should be pragmatic and the
characterization shouldattend thetaskperformedratherthanthesyntactic category.
Thisbroadercharacterizationwouldbehighlyimprecise,andprobablytooliberal,butit
wouldpermitwelcomebackonboardthefamiliarsetofwords.Idontobjectasfaras
weremainawarethat,amongthehospitableset,severalwelldefinedtypesofoperators
canbedistinguished.
Ipreferthestrongercharacterization.Inmyview,negationandconditionality,
bothsingularandgeneral,constitutethecoreofourlogicalapparatus.Andcorrectly
understood,thisviewdeeplyrespectstradition.ItisfaithfultoFrege,Ramsey,and
19
Peirce,toWittgenstein,Prior,andWilliams,toSellarsandBrandom,andingeneralto
allthosewhoconsiderwhatwedowithwordsasthebasiclevelofanalysis.
References
Adams,E.W.(1965),TheLogicofConditionals,Inquiry,8,pp.16697
Adams,E.W.(1975),TheLogicofConditionals,Dordrecht,Reidel
Barwise,J.(1989),TheSituationinLogic.Stanford,CenterfortheStudyofLanguage
andInformation.
Brandom, R. (1994), Making it Explicit. Reasoning, Representing, and
DiscursiveCommitment.HarvardUniversityPress
Carroll,L.(1895),WhatAchillessaidtotheTortoise.Mind,15,27980
Carston, R. (2002), Thoughts and Utterances. The Pragmatics of Explicit
Communication.Oxford,Blackwell
DeSwart,H.(1996),(In)definitesandGenericity.InKanazawa,M.,Pin,C.andde
Swart,H.(1996),pp.171194.
Dosen,K.(1989),LogicalConstantsasPunctuationMarks.Notre
DameJournalofFormalLogic30(3):362381
Dubucs, J and Marion, M. (2003), Antirealism and substructural logics. In A.
Rojszczak, J. Cacho, G. Kurczewski (dir.), Philosophical Dimensions of Logic and
Science. SelectedContributedPapersfromthe11th InternationalCongressofLogic,
MethodologyandthePhilosophyofScience.Krakow,1999,Dordrecht,Kluwer,2003,
pp.23549
Dudman, V. H. (1991), Interpretations of if sentences. In Jackson (ed.),
(1991),pp.202232
Dummett, M. (1973), Frege: Philosophy of Language. Cambridge, Mass., Harvard
UniversityPress
Dummett, M. (1981), The Interpretation of Freges Philosophy. Cambridge, Mass.,
HarvardUniversityPress
Etchemendy, J. (1983), The doctrine of logic as form. Linguistic and
20
Philosophy6,pp.319334.
Evnine,S.(1999),BelievingConjunctions.Synthese118:201227
Feferman, S. (1999), Logic, logics, and logicism. Notre Dame Journal of
FormalLogic,vol.40,pp.3154
Frege,G.(1879),ConceptualNotation.Aformulalanguageofpurethoughtmodelled
upontheformulalanguageofarithmetic.InGottlobFrege:ConceptualNotationand
RelatedArticles.TranslatedandeditedwithaBiographyandIntroductionbyTerrell
WardBynum,Oxford,ClarendonPress,1972,pp.101203
Frege,G.(1884),TheFoundationsofArithmetic.Oxford,BasilBlackwell,1980
Frege, G. (1892), On Concept and Object. In Geach, P. and Black, M. (eds.),
Translations from the Philosophical Writings of Gottlob Frege. Oxford, Basil
Blackwell,1952.
Geach,P.(1962),ReferenceandGenerality.AnExaminationofSomeMedieval
andModernTheories.IthacaandLondon:CornellUniversityPress.
Gentzen,G.(1969),CollectedPapers.Amsterdam,NorthHolland
Goble,L.(ed.)(2001),TheBlackwellGuidetoPhilosophicalLogic.Oxford,Blackwell
Publishers
Goldfarb,W.(1979),Logicinthetwenties.Thenatureofquantifiers. Journalof
SymbolicLogic44,351368.
GmezTorrente, M. (2002), The problem of logical constants, Bulletin of
SymbolicLogic,vol.8,pp.137
GmezTorrente,M.,forthcoming,ConstantesLgicas.EnM.J.Frpolli(ed.):
FilosofadelaLgica,Madrid,Tecnos.
Goodman,N.(1947),Theproblemofcounterfactual conditionals. TheJournalof
Philosophy,44,pp.11328.IncludedinJackson(ed.),(1991),pp.927
Haack,S.(1978),PhilosophyofLogics,CambridgeUniversityPress
Higginbotham,J.(2000),Onsecondorderlogicandnaturallanguage.InG.Sherand
R.Tieszen(eds.),BetweenLogicandIntuitionism.EssaysinHonorofCharlesParsons.
CambridgeUniversityPress,pp.7999
Hilpinen, R. forthcoming, Conditionals and Possible Worlds: On C. S. Peirces
ConceptionofConditionalsandModalities.InHaaparanta,L.(ed.),forthcoming
21
Jackson,F.(ed.)(1991),Conditionals.NewYork,OxfordUniversityPres
Jackson,F.(1998), Mind,Method,andConditionals.SelectedEssays.London
andNewYork,Routledge
Jackson,F.(1998a),Onassertion.InJackson(1998),pp.326
Jackson,F.(1998b),ClassifyingconditionalsI.InJackson(1998),pp.2742
Jackson,F.(1998c),ClassifyingconditionalsII.InJackson(1998),pp.4350
Jackson,F.(1998d),Postscriptontruthconditionsandassertibility.InJackson
(1998),pp5154
Kamp, H. and Reyle, U. (1993), From Discourse to Logic. Introduction to
Modeltheoretic Semantics of Natural Language, Formal Logic and Discourse
RepresentationTheory.PartI.Dordrecht,KluwerAcademicPublishers.
Kanazawa,M.,Pin,c.anddeSwart,H.(1996),Quantifiers,Deduction,andContexts.
CSLIPublications
Keenan, E.L. (1987), Unreducible nary quantifiers. In P. Grdenfors (ed.),
Generalized Quantifiers. Linguistic and Logical Approaches. D. Reidel Publishing
Company,1987.
Klima, G. (2006), Syncategoremata. Elseviers Encyclopedia of Language and
Linguistic,2nd.Ed.EditedbyKeithBrown.Elsevier:Oxford,vol.12,pp.353356.
Koslow,A.(1992),AStructuralistTheoryofLogic.Cambridge,CambridgeUniversity
Press
Lapierre,S.(1996),Conditionals andQuantifiers.InvanderDoesandvanEijck
(1996),pp.237254
Lewis,D.(1976),ProbabilitiesofConditionalsandConditionalsProbabilities.
PhilosophicalReview,85,pp.297315.IncludedinJackson(1991),pp.76101
Lewis,D.(1986),Probabilities ofConditionals andConditionalsProbabilities
II.PhilosophicalReview,95,pp.5819.IncludedinJackson(1991),pp.10210
McGee,V.(2000),Everything.InSherandTieszen(eds)(2000),pp.5478
Mellor.D.H.(1978), Foundations:Essays inPhilosophy, Logic,Mathematics,and
Economics.London,RoutledgeandKeganPaul
Mostowski,A.(1957),OnaGeneralizationofQuantifiers.FundamentaMathematica
44,pp.1236
22
Peacocke,C.(1976),Whatisalogicalconstant?.JournalofPhilosophy73:221240
Peirce,C.S.(1932), CollectedPapersofCharlesSandersPeirce.EditedbyCharles
Hartshorne and Paul Weiss. Volume II. Elements of Logic. Cambridge, Harvard
UniversityPress
Prior,A.(1976),Whatislogic?.PapersinLogicandEthics.Amherst,Universityof
MassachusettsPress
Ramsey,F.P.(1928),Universalsoflawandoffact.InMellor(1978),pp.12832
Ramsey,F.P.(1928),Generalpropositionsandcausality.InMellor(1978),pp.135
51
Read, S. (2003), Logical consequences as truthpreservation. Logique et Analyse,
183184,pp.479493
Recanati,F.(2001),LiteralMeaning.CambridgeUniversityPress
Sher,G.(1991),TheBoundsofLogic.AGeneralizedViewpoint.Cambridge,Mass.The
MITPress
Sher,G.andTieszen,R.(eds.)(2000),BetweenLogicandIntuition.EssaysinHonorof
CharlesParsons.CambridgeUniversityPress
Stalnaker,R.C.(1968),ATheoryofConditionals.StudiesinLogicalTheory,
AmericanPhilosophicalQuarterly,Monograph2,pp.98112.IncludedinJackson
(ed.),(1991),pp.2845
Stenius,E.(1960),WittgensteinsTractatus.Oxford,BasilBlackwell
Tarski,A.(1966):WhatAreLogicalNotions?,HistoryandPhilosophyofLogic,
vol.7,1986,pp.143154.
VanBenthem,J.(1984),Foundationsofconditionallogic.JournalofPhilosophical
Logic,13,no.3,303349.
VanBenthem,J.andterMeulen,A.(eds.)(1997),HandbookofLogicandLanguage.
Cambridge,Mass.,TheMITPress.
VanderDoes,J.andvanEijck,J.(eds.)(1996),Quantifiers,Logic,andLanguage.
CSLIPublications.
VanEijck,J.andKamp,H.(1997),Representingdiscourseincontext.Invan
BenthemandterMeulen(eds.)(1997),pp.179237
Wansing,H.(2001),Negation.InGoble(2001),pp.41536
Warmbrod,K.(1999),LogicalConstants.Mind,vol.8,July,pp.503538
23
Williams,C.J.F.(1989),WhatisIdentity?OxfordUniversityPress
Williams,C.J.F.(1992a),Being,IdentityandTruth.OxfordUniversityPress
Williams, C. J. F. (1992b): Towards a Unified Theory of HigherLevel
Predication.PhilosophicalQuarterly,42(169),pp.449464
Wittgenstein,L.(1921),TractatusLogicoPhilosophicus.
Wittgenstein,L.(1978),RemarksontheFoundationsofMathematics.EditedbyG.H.
vonWright,R.Rhees,G.E.M.Anscombe.Revisededition,Cambridge,Massachusetts.
TheMITPress.
NOTES
24
ThisworkhasbeenfinanciallysupportedbytheResearchProjectHUM200400118oftheSpanish
MinistryofEducationandScience.VersionsofthisworkhavebeenpresentedatthePhilosophy
Department,UniversityofMiami,atthePhilosophyDepartment,FloridaInternationalUniversity
(Miami,USA),PhilosophyDepartment,SouthFloridaUniversity(Tampa,USA),andalLogica
06,Hejnice(CzechRepublic).IamdeeplygratefultoProfessorSusanHaackformanyhelpful
discussionsonthetopicofthispaper.
ii
TheclosestcharacterizationIhavefounddoesnotbelongtoasinglephilosopherbuttoabroad
tradition.Itistheanalysisanddiscussionofsyncategorematagivenbythemedievallogiciansand
grammariansoftheXIIIandXIVcenturies:WilliamofSherwood,PeterofSpain,William
Ockham,JohnBuridanandAlbertofSaxony.Allofthem,andotheroftheircontemporaries,
studiedlogicaltermsattendingtotheirordinarymeaninganduse.SeeKlima(2006).
iii
Forexpositionsandcriticismsoftheseveralaccountsonlogicalconstants,seeforinstance
Peacocke(1976),Haack(1978:310),Feferman(1999),Warmbrod(1999),GmezTorrente
(2002),andGmezTorrenteforthcoming.
iv
Dosen(1989)arguesforaviewofconstantsaspunctuationmarks.Irejectthisviewsince
punctuationmarksdonothaveanyespecialsignificanceforourinferentialpractices.
ThisappliestoTarskisproposal(Tarski1966)andthelonglistofsequelsthatarenowknownas
invariantisttheories.Wecallanotion`logical,Tarskisays,ifitisinvariantunderallpossible
oneonetransformations ofthe worldontoitself (1966:149).SherusesMostowskis ideas to
defend the thesis that a logical constant has to be invariant under oneone transformations in
universesofthesamecardinality(Sher1991p.14).Thegrainoftruthbehindthisclusteroftheories
is that logical notions cannotbeusedtodistinguish amongindividuals.Butthis grainoftruth
cannotalterthefactthatbeinginvariantasapropertyofafunctionhasnoinferentialsignificance,
andafortiorinologicalrelevance.
Nevertheless,myproposalisnottobringlogicbacktogrammar.Fregefreedlogicfromgrammar
vi
inhisconceptualnotation,andIdonotattempttoundoFregesachievement.In Begriffsschrift,
Fregesituatedthetaskoflogicintheinferentialrelationsofconceptualcontents;hetooklogicfrom
grammartoplaceitintherealmofcontent,orofthought,inanonpsychologicalsense.Thisalso
wasaninterestofPeircessemiotics.MyproposalgoesalongwithFregeandPeirce,andseeksto
understandlogicasanaspectofwhathumansdowithwords.
vii
Asimilarviewisfoundinmedievallogic:syncategorematadonotsignifybutmerelycosignify.
Theydonotrepresententities,butatmost,modesofpresentationofentitiesandmodesof
combinationofconcepts.BuridanandAlbertofSaxonydefendedthisviewaboutthemeaningof
logicalterms,whichseemstoproceedfromtheAristoteliananalysisofthecopula.See,Klima
(2006).
viii
WhatisnowknownasCarrollsparadoxisnotoriginaltotheBritishlogician.Itappearsin
Frege(Frege1903),andinBolzano(Bolzano1837)beforehim.SeeSundholm(1999).
ix
IowethisnametoSusanHaack.
Austindefendedthispluralistviewinhis1962Lectures:Ithascometobeseenthatmany
speciallyperplexingwordsembeddedinapparentlydescriptivestatementsdonotserveto
indicatesomespeciallyoddadditionalfeatureintherealityreported,buttoindicate(notto
report)thecircumstancesinwhichthestatementismadeorreservationstowhichitissubject
orthewayinwhichitistobetakenandthelike.Tooverlookthesepossibilitiesintheway
oncecommoniscalledthedescriptivefallacy.(Austin1962:3).
Therearetwomainwaysoflookingatthetaskoflogic.Oneisunderstandinglogiconaparwith
xi
therestofsciences,withtheaimofsystematisingsomefeatures,themostgeneralinthiscase,of
reality.Inthissense,logicwouldbeadescriptivescience.Thisoptioncanstilladopttwoversions,
thePlatonicversionfollowedbyFregeinsomeofhiswritings,andtheempiricist,defendedby
Quine.Thenotionoftruth,oflogicaltruth,isthecentralconcepthere.Thereisanalternativeway
oflookingatlogic,whichwasmadefashionablebyGentzenandhisfollowers.Logic,theypropose,
doesnt represent the world, but rather represents inferential connections among propositions.
Naturaldeductioncalculidonthaveaxiomsbutonlyrulesofinference.Thebasicnotionoflogicis
nolongertruthbuttruthpreservation.Thisisthestandardviewinlogichandbooks.Nevertheless,
itisnotuncommonthatlogiciansdevelopaviewonlogicmorecongruentwiththetruthviewwhen
theythinkabouttheirsubjectmatter.AnexampleisHigginbothamwhodefineslogicasthemost
generaltheoryofthetrueandthefalse.(2000,p.79).Fregebeganoneofhislogicalinvestigations
withasimilarclaim.ForFrege,however,thetruewasatechnicalexpressiontounifyhisviewof
thesenseandreferenceofsaturatedexpressions,andshouldntbetakenintheordinarymeaningof
theword.InMybasiclogicalinsights,Fregesays:Trueonlymakesanabortiveattemptto
indicatetheessenceoflogic,sincewhatlogicisreallyconcernedwithisnotcontainedintheword
trueatallbutintheassertoricforcewithwhichasentenceisuttered.Brandomcommentsonthis
passageandexplainsthatwhatreallymattersinthepracticeofdrawinginferencesisnottruthbut
thepragmaticattitudeoftakingtrueorputtingforwardastrue(Brandom1994)p.82.InLogic
(1897)Fregedefineslogicasthescienceofthemostgenerallawsoftruth,whichisnotthesame
asthescienceoftruth.IconsiderBrandomsinterpretationofFregesviewtobeaccurate,andboth
BrandomsandFregesviewsare,inmyopinion,correctexpressionsofthetaskoflogic.Logicis
notthescienceoftruthbutoftruthpreservation.Itisnotconcernedwithwhichpropositionsare
trueandwhichfalse,butwiththelinkbetweenwhataspeakerassertsandthecommitmentsshe
undertakesbysodoing.SeeRead(2003)foradefenceofthisviewoflogic.
xii
Contemporary neoaristotelians (see Wansing 2001) defend the view that negation is not a
sententialoperator;thisviewhasitsgrainoftruthinit,butitisnotthewholetruth.Whennegation
actsastheoppositetoassertion,itsargumentisthewholesentence,andalsowhenitappliesto
complexsentences.
xiii
Iusethisratherartificialformofnegation,Itisnotthecasethat,tostressthatitsintended
scopeisthewholesentence.Ifwerestrictthemeaningofanoperatortoitstruthconditionaleffects,
predicativenegationandpropositionalnegationofthiskindareequivalent.Butthereismoreto
humanlinguisticactivitythantruthfunctionalcalculation.Neglectingthedistinctionbetweenthe
roleofnegationin(8)andin(9)wouldimplyneglectingthedifferencebetweenpredication(ofa
negativepredicate)andrejectionofacontent.AccordingtoStenius,negationintheTractatuswas
understoodastheoppositeofassertion.
xiv
Stenius, in his classical comment to Tractatus, attributed this view about negation to
Wittgenstein.
Thedebateabouttheexpressivemeaningofconjunctionisfound,forinstance,inCarston(2001)
xv
andEvnine(1999).
Iborrowexamples(20)(22)fromDeSwart(1996).
xvi
KampandReyle,proponentsoftheDiscourseRepresentationTheory,explaintherelationshipas
xvii
follows:Thereexistsanintimateconnectionbetweenconditionalsandthewordsevery,alland
each.Wegetaglimpseofthisconnectionwhenwecompare,say
(2.41)IfafarmerownsaMercedeshethrives
and
(2.42)EveryfarmerwhoownsaMercedesthrives.
Althoughthesesentencesmaynothaveexactlythesamemeaning,theirmeaningsare
certainlyveryclose.Inparticular,eachprovidestheinformationonthebasisonwhichwecaninfer
thatanindividualthrives,wheneverweencounterthisindividualwhomwerecognisetobeafarmer
andinpossessionofaMercedes.(1993,p.166)Theydonotclaimthatconditionalsandbinary
quantifiersbelongtothesamegenerallogicalcategory,butcomeclosetoit.
xviii
SeeDudman(1991)p.206,fn.3,foracriticismoftheimportancecurrentlyputonthe
distinctionbetweenindicativeandsubjunctiveconditionals.
xix
Ramsey (1928), Goodman (1947), Adams (1965), Lewis (1976) Jackson (1998a), (1998b),
(1998c),(1998d),andBennett(2003),areprimarilyconcernedwiththeproblemofjustification.
Evenauthorsthatdontprimarilyaddressit,suchasStalnaker(1968),considerthequestionof
decidingamongdifferentvaluationsanessentialtasktoberesolved.Iamnotconcernedwiththe
problemofhowtojustifyorwhenareweentitledtoassertaconditional,butwiththequestionof
whatlanguagesusersusethemfor.
Dudman expresses what I have called the inferential significance of conditionals in an
xx
illuminating way. He says: [H]ypotheticals are paradigm for their unaffirmed independent
messages.Theyalsoenjoyawelladvertisedrangeofdeductiveaccomplishments[].Thisisall
becausetheyarecondensedarguments,ofcourse.Toaffirmoneistoresorttotheargumentwhile
withholdingcommitmentfromapremise.(Dudman1991:228).