Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
The
evidential
meaning
of
modal
parentheticals
Corinne
Rossari
*
UniversityofFribourg,DomaineFrançais,13av.Beauregard,CH-1700Fribourg,Switzerland Received26September2011;receivedinrevisedform5July2012;accepted17September2012
Abstract
Thearticleexploreshowtheevidentialmeaningof‘inference’,‘perception’and‘hearsay’transmittedresp.bythefollowingFrench markers:fautcroire[oneshouldbelieve],ondirait[oneshouldsay]andparaît[itseems]isrelatedtotheirparentheticaluse.Toaddress this question,wefirstanalyzethesyntacticandpragmaticfunctionofthesethree parentheticalconstituents,byshowingthattheir preferenceforthefinalpositionistobeputinrelationwithadialogicalvalue.Second,wewillseethattheirevidentialmeaningcanbe reconstructedstepbystepbytakingintoaccountsuchadialogicalvalueaswellasthelexicalmeaningoftheverbtheyarecompounded with.Moregenerally,ourstudywillshowthatanevidentialmeaningisnotagenuinevalue,butisobtainedbyapluralityoffactors.Among them,thepragmaticfunctionduetotheirparentheticaluseplaysasignificantrole.
©2012PublishedbyElsevierB.V.
Keywords: Pragmatics;Syntax;Parenthetical;Evidentiality;Lexicalization
1. Introduction1
Theobjectiveofthisarticleistohighlightacaseofevidential2meaningderivedfromtheparentheticaluseofmodal
verbs.Weessentiallyfocusourattentiononmarkersbuiltwithverbsconveyingverydifferentlexicalmeaning:fautcroire
[oneshouldbelieve]builtwithaverbexpressinganobligation‘faut’andaverbexpressinganepistemicstate‘croire’,on
dirait[onewouldsay]composedwithadicendiverbusedintheconditionalmoodandparaît[itseems]whichconsistsinan
impersonalperceptiveverbconjugatedattheindicativepresent.Theirparentheticaluseinfinalpositionobliteratesthese
different lexicalmeaningsand givesrisetoasimilarevidentialvalue,as illustratedinthefollowingexamples3:
(1) Àdiredumalainsi,médire,mépriser,menacer,çaleurfaisaitdubien,fautcroire.Tospeakilllikethat,todespise,
tothreaten,it madethemfeelgood, oneshouldbelieve4(Céline,1932)
www.elsevier.com/locate/pragma
JournalofPragmatics44(2012)2183--2193
* Tel.:+41263007800;fax:+41263009752. E-mailaddress:corinne.rossari@unifr.ch.
1Thiscurrentstudyisacontributiontotheresearchproject‘‘TheevidentialsystemofFrenchandItalian’’fundedbytheSNF(Nos.
101512--117645).ItextendstheworkinitiatedinRossari(2012)forfautcroire,ondiraitandparaît.Thepresentanalysisreliesonthesyntacticand pragmaticfunctionofthesemarkers,whereasmyformerresearchwasfocusedontheinteractionbetweenmodalandevidentialmeaning.
2Thenotionof‘‘evidentialmeaning’’isusedinreferencetothewayithasbeenintroducedinpragmaticsfortheanalysisofmodalmarkersof
differentlanguages,asexposedintheseminalstudyofDendaleandTasmowski(2001).
3Ourcorporaarebasedonanyauthenticoccurrenceregardlessofdiaphasicvariation.Thereforewetakeintoaccountdatabases,asFrantext
(fortheauthor'sexamples)orInternet(forthespontaneouswrittenoccurrences).Astudyfocusedonwherethesewaysofsayingcomefromisto bedone.Howeverwehypothesizethattheyoccurinwrittentextswiththepurposetoimitatethespontaneouslanguage.
4TheFrenchdiscoursesandexpressionsarealwaystranslatedwithaword-for-wordtranslation,evenwhenitisnotcorrectinEnglish.Inthe
caseofthemarkersanalyzedtheword-for-wordtranslationisquitefarfromthemeaninginFrench.
0378-2166/$--seefrontmatter©2012PublishedbyElsevierB.V.
(2) Lesavionssontmaintenanttoutprès.Ilsfrôlentlesarbres,ondirait.Theplanesarenowveryclose.Theynearly
touchthetrees,onewould say(Cavanna,1979).
(3) Maintenant,jevaischezPascal.Etaprès,chezlapatronnequiveutmeparler,paraît.NowI’mgoingtoPascal.And
after,tothebosswhowantstotalkwithme,itseems.(Duhamel,1938).
Thethreemarkersindicatethatthespeakerhasonlyindirectevidenceforthestateofaffairs(SoA)expressedinthe
hostutterance.
Ouranalysiswillconcentrateontwoissues:ontheonehand,thesyntacticandpragmaticfunctionbetweenthemarker
andthehostclause;ontheotherhand,thesemanticsofeachmarker.Bothanalyseswillallowustoassesshowtheir
generalevidentialmeaning(indirectevidence)andtheirparticularone(‘inference’forfautcroire,‘vagueperception’foron
dirait and‘weakhearsay’forparaît)occurin thewaytheycontributetotheinterpretationofthehostutterance.
2. Thesyntacticandpragmaticfunctionbetweenthemarkerandthehostclause
2.1. Thesimilaritiesanddifferenceswiththereducedparentheticalclauses
Thethreeevidentialexpressionshavemanyfeaturesincommonwiththereducedparentheticalclauses(RPC)analyzed
inSchneider(2007).SomeoftheFrenchexpressionsincludedinSchneider'sRPClistareverysimilartotheitemsunder
consideration:‘‘jecrois,disons,ilfautdire,jedirais,paraît-il,mesemble-t-il’’(p.78).Accordingtothisauthor,theRPCare:
‘‘neitherthemainclausenorasubordinateclause,butareinsertedintooradjoinedtotheendofthesentenceinaway
similar with sentence adverbs.Their position isfree and there isno overt syntactic linkbetween them and the host
sentenceorpartsofit.Theyarerelatedtothehostonlybyadjacencyandbythefactthattheirmissingargumentcanbe
recoveredfromthehost.Asotherperipheralelements,theyareoptional,i.e.,theycanbeaddedaswellasdroppedfreely
without endangeringthegrammatical acceptability ofthehost.Nevertheless, theyarepragmaticallyconnected toit.’’
Schneider(2007:1).
Ourthreeexpressionsfitverycloselytothelatterdescription. However,theydifferontwoimportantpoints:
(i)Theyoccurinmostcasesinthefinalposition:
(4) Etc’estmêmepasbonàmanger;lerougedespoissonsrouges,c’estdupoison,paraît.Itisnotevengoodtoeat;
theredoftheredfishes,itispoison,itseems.(Fallet,1956)
(5) ??Etc’estmêmepasbonàmanger;lerougedespoissonsrouges,c’est,paraît,dupoison5Itisnotevengoodto
eat;theredofthe redfishes,it, itseems,ispoison.
Aparentheticalsuchas paraît-il[doesitseem]6canbeinsertedinanyposition:
(6) C’est,paraît-il, dupoisonItis,doesitseems,poison
(ii)Theydo nothaveamissingargument.They canbeusedasindependent utterances:
(7) Est-cequeçaleurfaisaitdubiendemédire,mépriser,menacer?Diditmakethemfeelgoodtospeakill,todespise,
tothreaten?
-Fautcroire/Ondirait /ParaîtOneshouldbelieve/Oneshouldsay/Itseems
Theforms withsubjectinversionare alsoRPCsaccordingto Schneider,but they cannot beusedas independent
utterances:
(8) Est-cequeçaleurfaisaitdubiendemédire,mépriser,menacer?Diditmakethemfeelgoodtospeakill,todespise,
tothreaten?
??[Faut-ilcroire/Dirait-on/Paraît-il/Mesemble-t-il]Shouldonebelieve/Shouldonesay/Doesitseem/Doesit
seemtome
5‘‘??’’indicatesthatthereisnoattestedoccurrenceoftheformandthatthesentencesoundsodd.
6Englishmakesnodifferencebetweentheformwithsubjectinversion(paraît-il)andtheregularone(ilparaît).Wewillusetheinterrogative
Otherwise,mostofthecriterialistedbySchneiderarerelevantforourthreeexpressions.Thelackofovertlinkwiththe
hostclause,thesententialityandthepossiblelinkwithamainclause useddescriptivelyarepropertiessharedby our
markers.Theyarenotfull-fledgedclauses,i.e.theyaregrammaticallyandpragmaticallyreducedandexpressionswith
thesameverbscanbeusedasmainclauses:ilfautcroirequeP[oneshouldbelievethatP]orilfautlecroire[oneshould
believeit], ondiraitqueP[onewouldsaythatP]etilparaîtque P[itseemsthatP].
2.2. Thepossiblerelationshipswiththehostclause
TheauthorsquotedbySchneider(2007)putforwarddifferenthypothesestoexplainthesyntacticrelationshipandthe
pragmaticfunctionoftheseexpressions.Whileallofthemnotethelackofanovertlink,manyauthorstrytoestablisha
correspondencebetweentheRPC'sconstructionandthesameconstructionusedasamainclause.Thislinkisbasedon
theideathatthereissomesortofaderivationprocessbetweenthesubordinatestructure(VerbthatClause)andthe
parenthetical one (Clause, Verb). The notion of ‘‘recteur faible’’ (‘weak governing verbs’) introduced by
Blanche-Benveniste (1989) and reanalyzed in Blanche-Benvenisteand Willems (2007)and Willems and Blanche-Benveniste (2010)iscommonlyacknowledgedtocharacterizethisparticularrelationshipbetweenthetwostructures.Themainideais
that the subordinate clause in the (Verb that Clause) structurecanbe interpreted as the maincontent ofthe whole
structure,aspointedoutintheseminalworkofUrmson(1952).Suchaninterpretationisconsistentwiththelossofthe
subordination link, resulting in the parenthetical use of the verb. RPC is thus seen as maintaining ‘‘some kind of
relationshipwiththehostatahigherlevel’’Schneider(2007:29).Thedisagreementsregardthenatureoftherelationship.
Someauthorsconsider thatthereisanimplicit hypotacticrelationship betweentheparentheticalandthehost.For
instance, Damouretteet Pichon(1933)seethe‘‘incise’’assubordinateclauses:
‘‘Celui-là,ditLalieenmontrantBernard,n’estplusmonfrère.Thatoneoverthere,saysLalieindicatingBernard,isn’t
my brotheranymore
‘‘Celui-làn’estplusmonfrèrequeditLalieenmontrantBernard.Thatoneoverthereisn’tmybrotheranymore,that
saysLalieindicatingBernard’’ Schneider(2007:31)
Suchananalysisisnotrelevantforfautcroire asanevidentialmarker,butispossibleforparaîtandondirait:
(9) Cen’estplusmonfrère,fautcroire /ondirait/paraîtThatoneisn’tmy brotheranymore,one should
believe/one wouldsay/itseems
(10) #Cen’estplusmonfrèrequ’ilfautcroire7Thatoneisn’tmy brotheranymore,thatone shouldbelieve
(11) Pasde goguesici,qu’ilparaît.Norestroomshere,asitseems(Degaudenzi,1987)
(12) Lecœurvaluisauterqu’ondirait.Hisheartwillexplode,as onewouldsay(Lacretelle,deGuéritte,T.,Sarn
[trad.],1930).
Brinton(2007)sustains,likeDamouretteetPichon,thattheparentheticaluseofImeancouldcomefroman‘‘adverbial
finiteclause(e.g.,asyouknowinwhichashasarelativefunction,orasIseeitinwhichashasasubordinatorfunction)’’
Brinton(2007:54).However,suchananalysiscouldnotbeusedtocharacterizethelinkbetweenfautcroire,ondirait,
paraîtand thehostclause.Noitemcouldbeusedwithcomme:
(13) *Cen’estplusmonfrèrecommeilfautcroire/commeilparaît/commeondirait.8Thatoneisn’tmybrother
anymore,asone shouldbelieve /asonewouldsay/asit seems
Brintonfinallyselectsanother scenarioforexplainingtheemergenceofImean asapragmaticparenthetical:
‘‘InitiallyImeangovernsaphrasalelement{NP,VP,AP,PP,AdvP}andhasscopewithinthesentence.Thebonds
betweenImeanandthephrasalelementareweakenedorloosened,andImeancanbegintobepostposedtothephrasal
element. The phrasal element is then reanalysed as an independent element, and I mean as a syntactically free
parentheticalwithscopeoverthesentence.Atthispoint,Imeanisextendedtothecontextofclausesandcanbepre-or
postposed toclausalelementsas well,thusacquiringscopeoverdiscourse.’’Brinton(2007:58).
Suchanexplanationisalsoplausibleforfautcroire. Ithasthecapacitytogovernaphrasalelementwhichcanbe
expressedbyaclitic:ilfautlecroire.Butthepresenceofthecliticobjectisneverattestedwhenfautcroireisusedinthe
finalpositionandifever usedtheutterancesoundsodd:
7‘‘#’’indicatesthatthesentenceisgrammatical,buthasanotherinterpretation:cen’estpasmonfrèrequ’ilfautcroiremaismonami[itisnotmy
brotherthatweshouldbelieve,butmyfriend].
8‘‘*’’indicatesthatthesentenceisnotgrammatical.Commeondiraitispossible,butitmeans‘‘itisawayofsaying’’.Ithasnoevidential
(14) Rosaliereprit:Rosaliesaid:
-M.Paulviendradèsl’enterrementfini.Demainàlamêmeheure,fautcroire.Mr.Paulwillcomeassoonasthe
burialisover.Tomorrowatthesametime,oneshould believe.(Maupassant,1883).
(15) M.Paulviendradèsl’enterrementfini.Demainàlamêmeheure,??ilfautlecroireMr.Paulwillcomeassoonas
theburialisover.Tomorrowatthesametime,one shouldbelieveit.
Paraîtandondiraitcantheoreticallygovernaphrasalelement(ondiraitX/ilparaîtX)butinbothcasesthephrasal
elementiseither aclauseorareducedclause.
(16) Ilparaîtque Paulestvenu.ItseemsthatPaulcame
(17) On diraitPaul->Ondirait quec’estPaulOnewouldsayPaul->OnewouldsaythatitisPaul
ThereforetheanalysiscannotbetheoneassumedbyBrintonforImean.Forotherwise,itwouldimplythefollowing
scenario:
‘‘Ithinkthathewillwin>Ithinkhewillwin>Hewillwin,Ithink’’, Brinton(2007:53).
ThelatterisnotadoptedforexplainingtheemergenceofImeanasapragmaticparenthetical.Furthermore,thereisno
empiricalgroundtoadoptitforexplainingthesentencefinalpositionofourthreeexpressions:itneitherexplainsthehigh
preferenceforfinalposition,northefactthattheyareindependentclauseswhereasparentheticalwithsubjectinversion
arenot:cf.ex.(8).OnthesameparadigmofilfautcroirequeP,wecanbuildilfautpenserqueP[oneshouldthinkthatP],il
fautenvisagerqueP[oneshouldforeseethatP],butnoneoftheseclauseshasdevelopedauseinsentencefinalposition
as theoneoffautcroire.
(18) Ilfaut envisagerquece n’estplusmonfrèreOneshouldforeseethatthatoneisn’t mybrotheranymore
(19) ??Cen’estplusmonfrère,fautenvisagerThatone isn’tmybrotheranymore,oneshouldforesee
(20) Ilfaut penserquecen’estplusmon frèreOneshouldthinkthatthatoneisn’tmy brotheranymore
(21) ??Cen’estplusmonfrère,fautpenserThatone isn’tmybrotheranymore,oneshouldthink
Suchacontrastalsoremainsunexplainedwiththeideathatthereisalinkbetweenthe‘‘recteurfaible’’useandthe
parenthetical one.
Atthepragmaticlevelthedescriptionsaremoreunified.TheviewofUrmson(1952)--accordingtowhichthemain
functionofRPCistolimitspeaker'scommitmentregardingthetruthoftheassertedproposition--iswidelyadopted.The
position(insertedorfinal)oftheRPCisusuallynotconsideredas relevant.Ourdescriptionwillconfirmthepragmatic
functionofmitigationbut willshowthatthepositionisnotirrelevanttofulfillsuchafunction.
MorelandDanon-Boileau(1998)considerparenthesisfromatextualpointofview.Theynotethatitenablesspeakers
tocommentontheirdiscourse(cf.Schneider,2007:34).
Ourhypothesiswillreinforcethelatterassumption.Wewillseeindeedthatthefinalpositionisparticularlyefficientfor
addingmaterialthatallowsthespeakertocommentonher/hisowndiscourse.But,incontrastwithmoststudies,werefuse
theideaofalinkbetweentheparentheticaluseoftheitemsanalyzedandthenotionof‘‘recteurfaible’’characterizingthe
(VthatClause)construction.Theuseofthesameverbinboth structuresisnotareasontohypothesizesomesortof
derivation process from the ‘‘recteurfaible’’ to the parenthetical useof the sameverb. Instead,we assumethat the
parentheticals are linked to their host at discourse level through a relationship between their dialogical uses as
independentutterances.
Inotherwords,wepostulatethattheexpressionsusedinthestructureexemplifiedby(7)endorseapragmaticfunction
similartotheonetheyhavewhentheyareusedasparentheticalsinmonologicaldiscourses.Thepreferenceforthefinal
position isa consequenceof thisdialogicalfunction. However,wedonot exclude that theycanbeinserted in other
positionswithasimilarfunction,sincesuchafunctionbecomes partoftheirmeaning.
2.3. Thecluesofalinktothehostclauseatdiscourselevel
Thethreeevidentialmarkersarelexicalizedinthesensethattheyhavepartiallylosttheircompositionality.Theverb
hasnoinflexion(#fallaitcroire[onehadtobelieve]/#ondira[onewillsay]/#paraîtra9[itwillseem])andthecliticscanor
havetoberemoved.Asalreadyseen,fautcroireisungrammatical insentencefinalpositionwiththecliticobjectand
unattestedwiththecliticsubject. However,paraîtisattestedin bothversionsinsentence finalposition(cf.(3)).
(22) -Bonjour,Lucas,souritMarie-Andrée,onnousaenvoyésici,tuypassestoustesaprès-midi,ilparaît.
HelloLucas,says Marie-Andrée,smiling,weweresenthere,youspendhereallyourafternoons,itseems
(Lang,2001)
The three expressions can be usedas independent utterances withthe sameformal properties. They can occurin
dialogicalcontextstoexpressa reactiontoaconfirmationrequestsharingthesamefeaturesoflexicalization.
(23) «Bonjourpetite!Eh!commetuvas!-Jesuispressée:ilyadesjourscommeça!-Fautcroire!»‘‘Hello,baby!
Well,howfastyougo?-Iaminahurry:therearedayslikethat!-Oneshouldbelieve’’(Bazin,1899).
(24) -Hein!Quoi?Ilsesauve!Well!What!Herunsaway!
Mongicourt,assissurle canapé,d’untonblagueur.-Ondirait!Mongicourt,sittingonthesofa,jokingly.-One
wouldsay(Feydeau,1914).
(25) -Etill’aimetoujourscettebonnefemme?Andhestilllovesthiswoman?
-Paraît.Itseems.(Queneau,1944)
Theverb hasnoinflexion:
(26) «Bonjourpetite!Eh!commetuvas!-Jesuispressée:ilyadesjourscommeça!-*Fallaitcroire!»‘‘Hello,baby!
Well,howfastyougo?-Iaminahurry:therearedayslikethat!-Onehad believed’’
(27) -Hein!Quoi?Ilsesauve!Well!What!Herunsaway!
Mongicourt,assis surlecanapé,d’untonblagueur.-*ondira/*-nousdirions!
Mongicourt,sittingonthesofa,jokingly.-onewill say/-wewouldsay.
(28) -Etill’aimetoujourscettebonnefemme?Andhestilllovesthiswoman?
-?Paraîtrait.10Itwouldseem.
Other lexicalized expressions used as confirmation in dialogical structures, have the ability to occur in the final
position.11Forinstance,amongthe‘‘weakverbs’’(ilmesemble,[itseemstome]ilparaît,[itseems]j’ail’impression,
[Ihavethefeeling]ondirait,[onewouldsay]jedirais[Iwouldsay])analysedbyWillemsandBlanche-Benveniste(2010),
j’ail’impressionandilsemble(andnotilmesemble)couldbeaddedtoourlist.Bothcanbeuseddialogicallyorinthefinal
positionwiththesamefeaturesoflexicalization:
(29) Lafourrure,cen’estpastellementtonstyle,j’ai l’impression.Furisnotreallyyour style,Ihavethefeeling.
(Mréjen,2004)
(30) Je reviensvousembêter carlesarticlesde Reviewer.frnesontànouveaupaspris encompteparGoogle
Actualitésdepuiscemidi.Ya-t-ildesralentissements?Iamcomingbacktobotheryoubecausethearticlesof
Reviewer.frareonemoretimenottakenintoaccountinGoogleActualitiessincemidday.Isthereanyslowdown?
-J’ail’impression,oui.Ihavethefeeling,yes.
(https://productforums.google.com/forum/?fromgroups#!forum/news-fr,05-25-2012)
j’ail’impressionusedasalexicalizedformcannotbemodifiedbypeut-être[perhaps].Whenthelatterisused,itselectsthe
full-fledgedinterpretationoftheconstruction.
(31) Je reviensvousembêter carlesarticlesde Reviewer.frnesontànouveaupaspris encompteparGoogle
Actualitésdepuiscemidi.Ya-t-ildesralentissements?Iamcomingbacktobotheryoubecausethearticlesof
Reviewer.frareonemoretimenottakenintoaccountinGoogleActualitiessincemidday.Isthereanyslowdown?
??J’aipeut-êtrel’impression,ouiIhaveperhapsthefeeling,yes/??J’aipeut-êtreeul’impression,ouiIhave
perhapshadthefeeling,yes
(32) Ilyalongtempsquejevisici.J’ail’impression...commentdire?...d’unrétrécissement...cemotpeutsembler
péjoratif...Non,riendetel.C’esttrèsdifficileàdire...Ihavebeenlivinghereforalongtime.Ihavethefeeling...
HowcouldIsay?...ofanarrowing...suchawordcouldseempejorative...No,itisnot.Itisdifficulttosay
(Bianciotti,1985)
10‘‘?’’indicatesthatthereisnoattestedoccurrenceoftheform.
11WedonotexcludethatmanyotherRPCscouldintegratethisparadigm,suchasjepense,jecrois...Thepossibleextensionoftheparadigm
(33) Ilyalongtempsquejevisici.J’aipeut-êtrel’impression...commentdire?...d’unrétrécissement...cemotpeut
semblerpéjoratif...Non,riendetel.C’esttrèsdifficileàdire...Ihavebeenlivinghereforalongtime.Ihaveperhaps
thefeeling...HowcouldIsay?...ofanarrowing...suchawordcouldseempejorative...No,itisnot.Itisdifficulttosay
Inthefollowingexamples,il sembleislexicalized. Itcouldneither bemodifiedby‘‘peut-être’’,norcanthetensevary:
(34) -N’a-t-ilpasdespaquets?Hashenobag?
-Ilsemble.Itseems(Romains,1913)
??Ilsemble peut-êtreItseemsperhaps/??IlasembléItseemed
(35) -Est-ilassezgrand,ouiounon,poursavoircequ’ilaàfaire?Isheoldenough,yesorno,toknowwhathehastodo?
-Ilsemble.Itseems(Romains,1922)
??Ilsemble peut-êtreItseemsperhaps/??IlasembléItseemed
(36) Etpuis,vousavezété recommandé,ilsemble. Andyouhavebeenrecommendedby someone,itseems
(Sollers, 1987)
(37) Etpuis,vousavezétérecommandé,??ilsemblepeut-être/??ilasemblé.Andyouhavebeenrecommended
by someone,itseemsperhaps/Itseemed
Instead,Ilmesemblehasnofeatureoflexicalization.Whenuseddialogically,itisinterpretedasafull-fledgedclause:
(38) -Tucroisqueçasignifie quejesuisfaitepourlesfemmes?YoubelieveitmeansthatIpreferwomen?
-Jene saispas.Ilmesemble...Qu’enpenses-tu?Idon’tknow.Itseemstome...Whatdoyouthink?
(Monferrand,1991)
(39) -Tucroisqueçasignifie quejesuisfaitepourlesfemmes?YoubelieveitmeansthatIpreferwomen?
-Jenesaispas.Ilm’asemblé/Ilmesemblepeut-être...Qu’enpenses-tu?Idon’tknow,Itseemedtome/It
seemsperhapstome...Whatdoyouthink?
Thesetestsshowthatthethreeconstructions,whenuseddialogicallyasindependentutterances,shareexactlythe
samefeaturesastheexpressionsusedasRPCinmonologicaldiscourses.Soinsteadofestablishingalinkatasyntactic
levelwiththehostclause,weestablishalinkatthediscourselevel.ThespeakerusestheRPCinmonologicaldiscourses
as atool togiveadialogicalflavortoher/hisdiscourse.Itallowsthespeakertoreacttohis/herowndiscourseand to
confirmtheSoApreviouslyexpressed.Inthissense,itfulfillsthefunctionof‘comment’highlightedbyMoreland
Danon-Boileau(seeabove).ThefinalpositionisaconsequenceofthefundamentaldialogicalvalueoftheseRPCs.Thewaythey
workmakethemappeartobesimilartotheclausesof‘‘justificationsconfirmatives’’[confirmativejustifications]identified
byBéguelin(2010)intheinventoryofthefunctionsofparatacticsequences:‘‘[...]lasecondeclause[quicorrespondraità
nosRPC]reprendenécho,maisdemanièreabrégée,lecontenudelapremière,produisantuneffetdeconfirmationou
d’insistance’’(Béguelin,2010:21).[Thesecondclause[whichcorrespondstoourRPC]repeatsinasummarizedformthe
contentofthefirstone, producingaconfirmativeoranemphaticeffect].
However,inthecaseofourRPCs,thereisnoexplicitreiterationofthecontentofthefirstclause.Theuseoftheclitic
objectle[it],whichindicatestherepetitionofthecontent,makesthesentenceillformed,asalreadyseenwithexample
(15).Thisisbecausethepresenceofthecliticwouldunderlineatextuallinkbetweenbothclauseswhichwouldcancel
theirfundamentallydialogicalrelationship.Inshort,weassumethatthesalientpropertiesoftheevidentialRPCsaretheir
abilitytostressalinkatthediscourselevel.ThehostclauseisconceivedasaconfirmationrequestandtheevidentialRPC
isusedtoaddressan answertothisrequest.Sucha discourseprocess involvesamitigation ofthehostclause:the
confirmationexpressedbytheRPCjeopardizesthereliabilityoftheSoAexpressedinthehostclause.12Itisthroughthis
functionthattheyacquireanevidentialvalue,as itwillbeseeninthenextsection.
3. Theevidentialvalue
3.1. Thecontributionofthedialogicalfunctiontothegeneralevidentialmeaning
Thedialogicalfeaturesoftheseexpressionsdetermineinparttheirevidentialmeaning.In‘real’dialogicalcontexts,
theyallowthespeakertoconfirmaSoAby makinganallusiontothecommonknowledge.Inthisdialog:
12Thediscoursepattern--rhetoricalquestionandanswer --crystalizedwithsomeRPCsisaverycommonwayof introducingtopicsin
monologicaldiscourses:‘‘18mai2012--Aurait-ilsurestimésescapacitésvocales?Ehbienoui!’’Wouldhehaveoverratedhisvoice?Well,yes! (http://www.lexpress.fr/culture/musique/donna-summer-l-hommage-2-0_1115905.html05-18-12).Thespeakerpretendstodoubtthevalidityof theideaexpressedinthequestion,andthen,she/heanswersthequestionhe/shehasherself/himselfasked.
(40) -Mincealors,fit-il, l’autresalaudluiaraconté?Damn,hesays,thebastardhastold himthestory?
-Fautcroire. Oneshouldbelieve(Clavel, 1962).
Fautcroireoranyotherexpression(ondirait,paraît,ilsemble,j’ail’impression)conveysaninterpretationwherethe
speakerconfirmstheSoAexpressedinthequestionbymakinganallusiontosomesharedknowledge.Ifthespeaker
knowsthatsomeonetoldthestorytotheprotagonistbecausehe/shewaspresentduringthediscussion,she/hewouldnot
sayfautcroire/ondirait/paraît/j’ail’impressionorilsembletoanswerthequestion.Forinstance,theseformscouldnot
beusedinthefollowingdialog:
(41) -Mincealors,fit-il, l’autresalaudluiaraconté?Damn,hesays,thebastardhastold himthestory?
-??[Fautcroire./Ondirait./Paraît/J’ail’impression/Ilsemble],j’yétais.Oneshouldbelieve/Onewouldsay
/Itseems/Ihavethe feeling/Itseems,Iwaspresent
Thespeakercoulduseoui[yes]instead:
(42) -Mincealors,fit-il, l’autresalaudluiaraconté?
-Oui,j’yétais.Yes, Iwaspresent
Tosumup,theevidentialmeaningoffautcroire/ondirait/paraît/j’ail’impressionorilsemblecanconfirmaSoAby
indicatinganindirectsharedknowledge.
Inmonologicalcontexts,theevidentialvalueisnearlyidenticaltotheoneindialogicalcontexts.Thespeakerindicates
that he/shehasindirectsharedevidenceto communicatewhatissaid inthe hostutterance.Such anindicationisa
projection of their dialogical use. The following constraints characterizing their use add supporting evidence to this
evidentialmeaning.
1.TheSoAmodifiedbythem isaconjecturenotafact thatcanbe directlyobserved.
(43) [Paulestdevantmoi]??Paulestarrivé,fautcroire/ondirait/paraît/j’ail’impression/ilsemble.13[Paulisin
frontofme]Paulhasarrived,one shouldbelieve...
2.TheSoAcannotbeaninferencethatonlythespeakerisabletomake:
(44) Jemedisais:‘‘SiPhilippesaitquePetitTitouavul’intérieurducoffre,ildoitêtreinquiet,ilestrestédansles
paragespourl’observer.[...]’’Iwasthinking:‘‘ifPhilippeknowsthatPetitTitouhasseenwhatisinsidethechest,
hemustbeworried;hestayedaroundsomewheretokeepaneyeonhim(Japrisot,1966).
(45) SiPhilippesaitquePetitTitouavu l’intérieurdu coffre,ilestinquiet,??[fautcroire/ondirait/paraît/j’ai
l’impression/ilsemble]ilestrestédanslesparagespourl’observer.Iwasthinking:‘‘ifPhilippeknowsthatPetit
Titouhasseenwhatisinsidethechest,heisworriedoneshouldbelieve...;hestayedaroundsomewhereto
keepaneyeonhim
3.They cannotbe insertedin anutterancethatthe speakerrefusestoendorsedirectlyafterwards:
(46) Onditquelarévolutionasupprimélanoblesse;maisc’esttoutlecontraire,elleafaittrente-quatremillionsde
nobles...Onesaysthattherevolutionhasabolishedthenobility;butonthecontraryitmadethirty-fourmillionsof
noblemen(Michelet,1846).
(47) ??Larévolutionasupprimélanoblesse,paraît/ondirait/fautcroire/j’ail’impression/ilsemble;maisc’est
toutlecontraire,elleafaittrente-quatremillionsdenoblesTherevolutionhasabolishedthenobilityoneshould
believe...;butonthe contraryitmade thirty-fourmillionsofnoblemen
Onditque,whichalludestoanothersourcefromthespeaker,enablesher/himtorejectthepointofviewborrowed,in
his/hersubsequentdiscourse.Thisisnotthecaseofourmarkers.Theymeanthatthespeakerdoesendorsewhathe/she
communicatesin thehostutterance,evenif he/sheputssomedistance withit.
13Anironicinterpretationmakestheexamplenatural,butinthiscasethespeakermakesfunofPaulactingasifhe/shehasnotnoticedhis
Alltheseconstraintsareconsequencesoftheirbasicdialogicalfunction.Thehostclauseshouldbeinterpretedasan
assessmenttobeconfirmed.Therefore,itcannotbeapersonalinferencefromthespeaker,sinceitmakesnosenseto
confirmone'sowninferences.Itcanneitherbeadirectperception,becauseitcannotbeinterpretedasanassessmenttobe
confirmed.Finally,itcannotbeanutterancethatthespeakerrefusestoendorseinher/hisfollowingdiscourse,becauseour
markersindicatethatthespeakerconfirmtheinformationconveyedinthehostclause.Thesecharacteristicsarecommonto
allthesemarkers.Buteachofthemalsoexpressessomeparticularevidentialmeaningthatwewillbeexaminedinthenext
section.
3.2. ThecontributionofthesemanticsoftheverbtotheevidentialmeaningofeachRPC
TheconfirmationconveyedbyfautcroireisbasedonanindicationthattheSoAcanbeassumedbecauseitisfunded
on obvious clues. This iswhy the RPCseems to endorse an evidentialmeaning of ‘inference’. Forinstance, in the
followingexamples,thespeakercomments truismswithfautcroire:
(48) On acequ’onméritefautcroire!Wehaveonlywhatwedeserve,one shouldbelieve!
(http://blogues.lapresse.ca/therrien/2012/04/12/mongrain-sen-va/04-13-2012)
(49) Jamaisjen’auraisimaginérevivreuncertaintempsensaprésence,maisilnefautjamaisdirejamaisfautcroire.
Iwouldneverhaveimaginedtoliveagainsometimewithhim,butoneshouldneversayneveroneshouldbelieve
(http://mamancoach.blogspot.ch/2011_05_01_archive.html05-14-2011)
TheeasyuseoffautcroirewithtruismsisduetothefactthatitallowsthespeakertoactasiftheSoAwereworthtobe
said.Withoutfautcroire,theutteranceor‘‘onaquecequ’onmérite’’[wehaveonlywhatwedeserve]or‘‘ilnefautjamais
dire jamais’’ [oneshould never saynever] isinterpretedas non-informative. Withfaut croire,it isreinterpreted asan
assumption(eveniftheSoFexpressed isself-obvious)thattheuseoffautcroire confirmsretroactively.
Theindicationofobviousnessislinkedtothelexicalvalueoftheverbfalloir,alwayspresentintheRPC.Thesequence
withbothclitics(subjectandobject)ilfautlecroire[oneshouldbelieveit]canbeusedwithacompositionalmeaning.In
suchause,itexpressesthenecessitytobelieveinaparticularSoA,becauseofthepresenceofclues(mentionedornot)
thatprove thetruthofit.
(50) -Est-ceque,parhasard,pensais-je,j’enseraistombéamoureuxaujourd’hui,sanslesavoir?Ilfautlecroire,
puisquevoilàHurielquim’ypousse,etdontl’œil aurasaisilavéritésurmafigure. Didby anychance,Iwas
thinking,Ifallinlovewithhertoday,withoutknowingit?Onehastobelieveit,becausehereisHurielwhomakes
medo it,andwhoseeyewillhavenoticedthe truthon myface(Sand,1865).
Thecliticpronounle[it]referstothecontentofthequestionandthecluesonwhichtheanswerby‘‘ilfautlecroire’’is
basedaregivenjustaftersince.The necessityverbfalloirhasitsfulllexicalvalue.When thesequence isusedasa
lexicalized item,theobviousnessmeaningstandsinforthenecessitynuance.
Theconfirmationexpressedbyondiraitisbasedonasensorialperception(visualinmostcases).ThisiswhytheRPC
seemstoendorseanevidentialperceptivemeaning.Whenthespeakercommunicatesanassessmentthatcanonlyrely
onavisualperception,ondiraitismorenaturalthanthetwootherevidentialmarkers.Inthefollowingcontext,thespeaker
looksatherfriend andthinksthatshelookspregnant.
(51) Maistuesenceinte,ondirait!Butyouarepregnant,onewouldsay
Using fautcroireouparaîtwouldbeverystrangein thiscontext:
(52) Maistuesenceinte,#[fautcroire/paraît]14!Butyouarepregnant,one shouldbelieve/itseems
Inthenextexample,thecontextindicatesthattheSoAisobtainedbyaphysicalperception.
(53) Jeretournepasàlamaison.Lesmurstremblentcommetout.J’aitroppeurquelesplafondsmedégringolent
dessus. Idon’t comebackhome.Thewalls areshaking.Iamtooafraidthat theroofmightfalloverme.
-Çasecalme,ondirait. Itisgettingquiet,onewould say(Bory,1945)
14Bothofthemcouldbeusedbutwithadifferentinterpretation:fautcroireindicatesthattheclueisnotwhatthespeakerhasobservedbutsome
otherindicationthatleadsnaturallytothisremark.Paraîtdoesnotcommunicateanyindicationontheclueonwhichthespeakerbaseshis/her remark.
Neitherfautcroirenorparaîtarenatural insuchacontext:
(54) Jeretournepasàlamaison.Lesmurstremblentcommetout.J’aitroppeurquelesplafondsmedégringolent
dessus. Idon’tgo comebackhome. Thewallsareshaking.Iamtooafraid thattheroofmightfalloverme.
-Çasecalme,??[fautcroire /paraît].Itisgettingquiet,oneshouldbelieve/itseems
Intheexample below,anacousticperceptionisattheoriginoftheinformation:
(55) Jecollel’oreille.Plusrien...[...].Unefenêtres’ouvretoutàcoup.Au-dessusdenous,ondirait.Iamlistening.No
morenoise.Suddenly,awindowopens.Upstairs, onewouldsay(Boudard,1963).
Theuseoffautcroireorparaîtcancelstheacousticsourceleadingtotheconjecture:
(56) Jecollel’oreille.Plusrien...[...]Unefenêtres’ouvretoutàcoup.Au-dessusdenous,fautcroire/paraît.Iam
listening. Nomorenoise.Suddenly,awindowopens.Upstairs, oneshouldbelieve/itseems
The perceptionconveyedby on dirait comes fromitsdicendi meaning. On dirait literallycommunicates whatone
shouldsay,ifaparticularSoAwerenoticed.Theconditionalbringsanindicationofcautionwhichgivesrisetothemeaning
ofvaguefeeling.
(57) Sil’onsavaitcela,l’ondirait queMadeleineestfolle.Ifweknewthis,wewouldsaythat Madeleineiscrazy
(Gautier, 1836)
Inthisexample,ondiraitisinterpretedasadicendiverbusedintheconditionalmood.Butwithouttheif-clause,itcan
alsobeinterpretedasanevidentialverb.Inthelattercase,theinterpretationcorrespondsto:onal’impressionque...
[wehavethefeelingthat...].Inexample(2)ondiraitisusedasaRPC,butitisalsoambiguous.Itcanbeinterpretedas‘‘si
onvoyaitlesavionsdeprès,ondiraitqu’ilsfrôlentlesarbres’’[ifwerelookingcloselyattheplanes,wewouldsaythatthey
nearlytouchthetrees]orasanevidentialmarker.InthiscaseitisequivalenttotheRPC‘‘j’ail’impression’’:‘‘Lesavions
sontmaintenanttoutprès.Ilsfrôlentlesarbres,j’ail’impression’’[Theplanesarenowveryclose.Theynearlytouchthe
trees,Ihavethefeeling].Sincethesequenceislexicalized,thedicendimeaningisoverandtheRPCexpressesasimilar
meaningtoj’ail’impression.
Theconfirmationexpressedbyparaît,isbasedonavisualperceptionwhoseoriginisdissimulated.Inacontextwhere
thesourceisapreciseclue(forinstancethespeakerseesthelightinPaul'sflat)she/hewouldnotsay:
(58) #Paul estlà,paraît.Paulhasarrived, itseems
Withparaîttheutteranceisinterpretedasalludingnottothelight,buttosomeotherinformation,thesourceofwhich
remainsundetermined.Suchanevidentialmeaningissimilartoaweak‘hearsay’.Onditqueforinstance,whichconveys
a‘real’‘hearsay’,allowsthespeakernottoendorsethetruthofthepropositionalcontent.Suchaninterpretationisnot
possible with paraît, as seen in example (47). The example below shows that paraît does not allude to the visual
perceptionofthespeakerbuttosomeinformationwhoseoriginisnotcommunicated.
(59) C’estainsiqu’ellepritl’habitudedevenirlevoiràtouteheure.Aquelquetempsdelà,iladvintqu’elleredescendit
avecunedecesbabiolesqu’onlesportecommebracelets,c’estlemotpourcettesorted’objet,mêmequ’ilen
avaitd’enroulésàsesbras,lemort,unbraceletwampum,etunbeau,paraît.Thussheusedtocomeandsee
him,atanytime.Later,shesometimeswentbackdownstairswithsometrinketthatwewearlikebracelets,thatis
thewordforthissortofthing,thedeadhadoneofthesearoundthearm,abraceletwampum,andaniceone,it
seems.(Aragon,1926).
Thenextexample highlightstheambiguous interpretationofparaîtbetweenperceptiveand evidentialmeaning:
(60) -T’esun sacrétireur,garsClaudius,paraîtbien!You areahellofshooter,chapClaudius,itseemswell
(Chevallier,1934)
Theperceptivemeaningisstillpresentwiththeuseofbien,butparaîtcanalsobeinterpretedasconveyingaweak
Theindeterminationoftheperceptionconveyedbytheevidentialmeaningofparaîtcomesfromtheimpersonaluseof
theverb.Paraîtreusedas afull-fledgedverbindicatesthatthespeakerhasapersonalinterpretationofher/hisvisual
perception.Inanutterancesuchas:
(61) PaulparaîtpâlePaullookspale
thespeakersaysthathe/shehasthepersonalassessmentofinterpretingthecolorofPaul'sfaceaspale.Whatis
surprisingwiththeevidentialvalueisthatparaîtnolongerindicatestheperceptionofthespeaker.Intheexamplebelow:
(62) Paulest pâle,paraîtPaulispale,itseems
theperceptionofPaul'sfacecannotbeinterpretedastheoneofthespeaker.She/Heconveysanassessmentthatshe/
he hasnot establishedherself/himself.Such an apparentsemantic change isdueto theimpersonal useof theverb
paraître.Thepersonattheoriginoftheperceptiondisappears.Whenonesays‘‘Paulparaîtpâle’’[Paulseemspale],the
hearerknowsthatthespeakerisattheoriginofthevisualperception.Butwhenonesays‘‘IlparaîtquePaulestpâle’’,[it
seemsthatPaulispale]thehearerhasnoinformationontheoriginofit.Thisiswhytheevidentialmeaningcorrespondsto
aweakhearsay,as perfectlyillustratedbytheexamplebelow:
(63) [...]maiscettemaisonappartenaitdoncàM.DeSaint-Méranavantquevousnel’achetiez,Monsieurlecomte?
-Ilparaîtqueoui,réponditMonte-Cristo.ButthishousebelongedtoMr.DeSaint-Méranbeforeyouboughtit,Mr.
theCount?-- Itseemsthat yes,answered MonteCristo.
-Comment,ilparaît!Vousnesavezpasàquivousavezachetécettemaison?How,itseems!Youdonotknow
fromwhomyouboughtthishouse?
-Mafoinon,c’estmonintendantquis’occupedetouscesdétails.Well,no,itismytreasurerwhotakescareof
thesedetails.(DumasPère,1846)
The reaction(How, it seems!)of the speaker showsthat he is fully awarethat Monte Cristo, using il paraît que,
dissimulateshisknowledge.Yet,theinformationintroducedbyilparaîtqueisstillnotdubious,becausetheideaofvisual
perceptionremainspresent.ItexplainswhythespeakercannotrejecttheSoAafterwardscf.(47).
Inshort,theevidentialmeaningofeachmarkerisconveyed,ononehand,bytheirdialogicalpropertiesandonthe
otherhand, bythelexicalmeaningoftheirverb.
4. Theoreticalissues
Ourstudyhasshownthattheevidentialmeaningcannotbeconsideredasaprimitivesemanticvalue,butoccursin
relationtoapluralityoffactors.Syntax,semanticsandpragmaticsplayaroleinthe constitutionofthismeaning.
Ontheonehand,thefinalpositionoftheRPCsanalyzedisconsistentwiththeillocutionaryfunctionofconfirmationon
anassessment,whichgivesrisetoanevidentialmeaningofindirectsharedknowledge.OnecanonlyconfirmaSoAthat
he/she hasnotestablished by himself/herself.Thisfunction obliteratesthe different lexicalmeaning ofeachmarker.
ThereforenoneoftheseRPCscanoccurinacontextwherethespeakerhasadirectaccesstothereportedknowledgecf.
(43).
On the other hand, the semantic kind of modality allows them to convey one particular evidential meaning:
‘obviousness’forfautcroirewhichgivesrisetoaninferentialevidentialmeaning,‘vagueperception’forondiraitwhich
givesrisetoaperceptiveevidentialmeaning,and‘dissimulationoftheexactsource’forparaît,whichgivesrisetoaweak
hearsayevidentialmeaning.
Acknowledgements
Ithankthe twoanonymousreviewersfortheirhelpfulsuggestionsandconstructivecriticismsofan earlierdraft.
References
Béguelin,Marie-José,2010.Noyauxprédicatifsjuxtaposés.In:Béguelin,M.-J.,Avanzi,M.,Corminbœuf,G.(Eds.),Laparataxe,Tome1.Lang, Berne, pp.1--34.
Blanche-Benveniste,Claire,1989.Constructionsverbales‘‘enincises’’etrectionfaibledesverbes.RecherchessurleFrançaisParlé9,53--73. Blanche-Benveniste,Claire,Willems,Dominique,2007.Unnouveauregardsurlesverbesfaibles.BulletindelaSociétéLinguistiquedeParis102
Brinton,LaurelJ.,2007.ThedevelopmentofImean:implicationsforthestudyofhistoricalpragmatics.In:Fitzmaurice,S.,Taavitsainen,I.(Eds.), MethodsinHistoricalPragmatics.MoutondeGruyter,Berlin, pp.37--80.
Damourette,Jacques,Pichon,Edouard,1933.DesmotsàlaPensée.EssaideGrammairedelaLangueFrançaise.1911--1933.TomeTroisième: Morphologieduverbe,StructuredelaPhraseVerbale,Infinitive.D’Artrey,Paris.
Dendale,Patrick,Tasmowski,Liliane,2001.Introduction.Evidentialityandrelatednotions.JournalofPragmatics33(3),339--348. Morel,Mary-Annick,Danon-Boileau,Laurent,1998.GrammairedeL’intonation.L’exempleduFrançais.Ophrys,Paris.
Rossari,Corinne,2012.Valeurévidentielleet/oumodaledefautcroire,ondiraitetparaît.LangueFrançaise173,65--81.
Schneider,Stefan,2007.ReducedParentheticalClausesasMitigators.StudiesinCorpusLinguistics27.Benjamins,Amsterdam/Philadelphia. Urmson,JamesO.,1952.Parentheticalverbs.Mind61,480--496.
Willems,Dominique,Blanche-Benveniste,Claire,2010.Verbes‘faibles’etverbesàvaleurépistémiqueenfrançaisparlé:ilmesemble,ilparaît, j’ail’impression,ondirait,jedirais.In:Iliescu,M.,al,(Eds.),ActesduXXVeCongrèsInternationaldeLinguistiqueetdePhilologieRomanes, Septembre2007,TomeIV,DeGruyter,Innsbruck, pp.565--579.
CorinneRossariisAssociateProfessorofFrenchlinguisticsattheUniversityofFribourg(Switzerland).Shehaspublishedandeditedseveral booksonlinguisticchangeanddiscoursemarkers.ThelatestareGrammaticalizationandPragmatics:facts,approaches,theoreticalissues (StudiesinPragmatics5,Elsevier,2009),andLesmoyensdétournésd’assurersondire(Pressesdel’UniversitédeParis-Sorbonne,2007),both writtenwiththecollaborationofherPh.D.students(AnneBeaulieu-Masson,CorinaCojocariu,AnnaRazgouliaeva,ClaudiaRicci,andAdriana Spiridon).Herresearchinterestsliemainlywithinthefieldsofpragmatics,semanticchanges,rhetoric,argumentationandvarietiesofFrench.