0022-3514/03/$12.00DOI:10.1037/0022-3514.84.4.857
TheSoundsofSocialLife:APsychometricAnalysisofStudents’Daily
SocialEnvironmentsandNaturalConversations
MatthiasR.MehlandJamesW.Pennebaker
UniversityofTexasatAustin
Thenaturalconversationsandsocialenvironmentsof52undergraduatesweretrackedacrosstwo2-dayperiodsseparatedby4weeksusingacomputerizedtaperecorder(theElectronicallyActivatedRecorder[EAR]).TheEARwasprogrammedtorecord30-ssnippetsofambientsoundsapproximatelyevery12minduringparticipants’wakinghours.Students’socialenvironmentsanduseoflanguageintheirnaturalconversationsweremappedintermsofbaseratesandtemporalstability.Thedegreeofcross-contextconsistencyandbetween-speakersynchronyinlanguageusewasassessed.Students’socialworldsaswellastheireverydaylanguagewerehighlyconsistentacrosstimeandcontext.Thestudyshedslightonamethodologicalblindspot—thesamplingofnaturalisticsocialinformationfromanunobtrusiveobserver’sperspective.
Althoughmostsocialpsychologistswouldagreethatresearchshouldultimatelyleadtoinsightintoreal-lifephenomena,thereisatraditionofbiastowarddecontextualizedlaboratorymethods.Recently,seriousconcernshavebeenraisedthatpsychologyhaslostcontactwithnaturallyoccurringsociallifeandthatthedisci-plinewouldbenefitfromacoursecorrectiontowardamorecontext-andculture-sensitivepsychology(Funder,2001;Hogan,1998;Rozin,2001).
Acentraltenetinthefieldisthatsocialcontextsinfluencebehavior.However,itisalsoclearthatinreallifethereisnorandomassignmenttosocialcontexts.Humansareactiveagentsinselectingandshapingtheirenvironments,socialsituations,andinteractions(Allport,1937;Buss,1987;Swann,1987)—presum-ablytomaximizeperson–situationfit.AccordingtoIckes,Snyder,andGarcia(1997),
Onceindividualsareintheirchosensituation,theirwordsandactions[italicsadded]aregenuinereflectionsoftheirpersonalities,andthefactthattheydisplaythesebehaviorsinsettingstheyhavespecificallychosenensuresasubstantialdegreeofconsistencyintheirbehavior.(p.166)
Substantialprogresshasbeenmadeinmappingeverydayexpe-riences,andfromthisithasbecomeincreasinglyclearthatacoherentpersonalitydoesemergeindailylife—atleastinthemindsoftheresearchparticipants(Csikszentmihalyi&Larson,
1984;Diener&Larson,1984;Epstein,1979;Fleeson,2001;Moskowitz,1994).Inotherwords,whenonelooksatpeople’slivesfrom“their,”thatis,theagent’s,perspective,whatoneseesisasubstantialdegreeoftemporalstabilityandcross-contextconsistency.Adifferent—andcertainlyequallyimportant—ques-tion,however,ishowsociallivesappearfromaperspectivethatmostpeoplenaturallyadoptmostofthetime:Theperspectiveofanunobtrusiveobserver(Funder&Colvin,1997;Hogan,1982).Howstablearepeople’sdailylivesfromthis“outside”perspec-tive?Isconsistencytheexclusivedomainoftheagent,orisitalsoaccessibletothepeoplearoundhimorher?
Inthisstudy,weadoptedthislargelyneglectedperspectivebytrackingpeople’sdailyliveswiththeElectronicallyActivatedRecorder(EAR;Mehl,Pennebaker,Crow,Dabbs,&Price,2001).TheEARisanevent-samplingtoolspecificallydesignedforthenaturalisticassessmentofacousticbehavioraltraces.Ourgoalforthisprojectwastotrackpeople’ssociallivesfromanunobtrusiveobserver’spointofviewandtoidentifythedegreeofstabilityacrosstimeandsituations.Conceptually,weadoptthenotionthatsociallifecanbeunderstoodintermsoftwomajorsources:people’smoment-to-momentenvironmentsandtheirnaturalcon-versationwithpeoplearoundthem.Wefurthermakeanargumentthatthewaylanguageisusedinnaturalconversationcarriesrichandvaluableinformationaboutaperson’ssociallife.First,how-ever,westartoutbypointingtoamethodologicalblindspotinnaturalisticperson–environmentinteractions.
PreparationofthisarticlewasaidedbyNationalInstitutesofHealthGrantMH52391andbyascholarshipfromtheGermanNationalSchol-arshipFoundation.WethankDavidBuss,KennethCraik,SamuelGosling,AnnaGraybeal,andSeanMasseyfortheirhelpfulcommentsonpreviousversionsofthisarticle.WeareindebtedtoGeethaDesikan,KimFisher,BethHenary,AllisonMcClure,JohnMcFarland,KristyOrr,NathalieShook,MonicaSidarous,Mary-BethSylvester,andSaraWebbfortheirhelpincollectingthedataandtranscribingtheaudiotapes.
CorrespondenceconcerningthisarticleshouldbeaddressedtoeitherMatthiasR.MehlorJamesW.Pennebaker,DepartmentofPsychology,UniversityofTexas,Austin,Texas78712.E-mail:mehl@psy.utexas.eduorpennebaker@psy.utexas.edu
857
MethodologicalIssuesinAssessingNaturalistic
Person–EnvironmentInteractions
Becauseoftheirincomparableefficiency,retrospectiveself-reportshavedominatedtheassessmentofeverydaylife.However,ithasalsobecomeclearthatanoverrelianceonpeople’saccountsofthemselvescancomeatratherhighcosts.Askingparticipantstoaccuratelyrecallsettings,activities,interactions,andexperiencesoverextendedperiodsoftimeisataskthatislargelyincompatiblewithhumanautobiographicmemory(Schwarz&Sudman,1993;Stoneetal.,2000;Tourangeau,Rips,&Rasinski,2000).
858
MEHLANDPENNEBAKER
Overthelast15years,experiencesampling(ES)ormomentaryassessmenttechniqueshaveevolvedasvaluablealternativesandgainedsubstantialpowerandpopularity(Csikszentmihalyi&Lar-son,1987;Hormuth,1986;Stone&Shiffman,1994).Peoplearepromptedseveraltimesadayandaskedtoprovideinstantreportsontheirmomentaryexperiences,activities,andenvironments.Thereisgeneralagreementthatminimizingthetimelagbetweentheoccurrenceandtheregistrationofeventsinquestionsuccess-fullybypassesmemory-relatedproblemsofretrospectiveself-reports.Consequently,momentaryassessmentscanbeconsideredrelativelyobjectiveaccountsofaperson’sdailyexperiences.Bydefinition,however,self-reports—bothretrospectiveandmomentary—aresubjecttotwoimportantconstraints:First,anyrecallofeventsisnecessarilyasubjectiveconstrual,areflectionofhowthepersoninterpretedtheevent.Thisisapparentwhentheresearcherseekstodemonstratecoherenceonthelevelofthepsychologicalmeaningofanact,anevent,oraninteraction.However,theoreticallyaswellaspragmaticallyitisequallyim-portanttoasktowhatdegreeaperson’severydaylifeisstableandconsistentfromaneutralobserver’sperspective(Funder&Sneed,1993;Gosling,John,Craik,&Robins,1998;Hogan,1982).Arethereidiosyncratictracesinpeople’sdailysocialworldsthataredetectablefromanoutsidepointofview,ordoescoherencemainlyemergeonthesubjectiveinterpretationallevel?
Asecondconceptualconstraintofanyself-reportassessmentisthatpeoplecanonlyrecallwhattheyareawareof.Peopleareconstantlysurroundedbyaninfiniteamountofpotentialinforma-tion.Althoughsomeofthispotentialinformationisprocessedbythebrain,onlyasmallportionpassessensoryfilterstoenterawareness.Aspectsofpeople’ssocialworldsthatfalloutsidetheirimmediateawarenessareinevitablylost.
Theseconsiderationshaveinspiredobservationalapproachesinthestudyofperson–environmentinteractions.BarkerandWright(1951)pioneeredsuchamethodologyintheirfamouscasestudyofRaymond,a7-year-oldboy,whowasfollowedbyobserversoveranentireday.Craik(2000),inwhathecalledlived-dayanalyses,extendedBarkerandWright’sparadigmbyfollowingapersonwithavideocameraandobtainingacomprehensiverecordoftheperson’ssocialencountersoveranentireday.Naturalisticobser-vationstudiesyieldanimmenseamountofrichinformationaboutaperson’slife.However,theuniqueadvantageisalsothemostcriticalpitfall.Theamountofdatathatiscollectedhasrestrictedthismethodtoideographicanalysis.Anotherproblemwithexist-ingobservationaltechniquesisthattheactofobservationalwaysconstitutesamajorintrusionintoparticipants’worldsandthusreactivelyinfluencestargetbehaviors(Barker&Wright,1951;Craik,2000).
Ourrecentlydevelopedtool,theEAR(Mehletal.,2001),combinesaspectsofbothlived-dayanalysesandESapproaches.TheEARisamicrocassetterecorderthattracksambientsoundsinpeople’senvironmentbyrecording30-ssnippetsevery12.5minoveraperiodofupto4days.Participantswearthedevicewhilegoingabouttheirdailylives.Intherecordings,theEARadoptstheuniqueobserverperspectiveoflived-daystudies.Inrestrictingthedatacollectingtoonlyarepresentativesubsetofeverydaybehav-ior,itincorporatestherelativeeconomicadvantageofESstrate-gies.Thiscombinationallowsforanomotheticstudyofeverydaysociallifewhilepreservingnaturalisticdataattheleveloftheraw
recording.Also,becausetheEARoperatesimperceptibly,mea-surement-inducedintrusionsareminimal.
TheSoundsofSocialLife:
AcousticSourcesofSocialInformation
WhatcantheEARtellaboutaperson’ssociallife?MuchofthesocialworldisrepresentedacousticallyandthusispotentiallydetectablebytheEAR.Twomajorsourcesofinformationarereadilyavailableontherecordingsandareinherentlysocialinnature:people’sdailysocialenvironmentsandtheirnaturalconversations.
Conceptually,people’ssocialenvironmentscanbedissectedintowheretimeisspent,whattimeisspentdoing,andwithwhomtimeisspent(Csikszentmihalyi&Larson,1984).TheambientsoundscapturedbytheEARallowforthereliablecodingofpeople’slocations(e.g.,insideanapartment,inapublicplace)aswellastheiractivities(e.g.,surroundedbysoundsfromTV,alecturer,fromcooking,orcomputertyping).Finally,EARrecord-ingscancaptureimportantinformationaboutpeople’ssocialin-teractions:Aretheyaloneorwithothers?Aretheyengagedinaconversation?Aretheyonthephoneortalkingfacetoface?Obviously,thecategorizationofambientsoundshasitslimitsandcannotreachthedegreeofdifferentiationachievedwithac-tivityandenvironmentquestionnaires(e.g.,Csikszentmihalyi&Larson,1984;Robinson,1985).However,EARdataallowthecalculationoftime-useestimatesonthebasisofacoustictracesuninfluencedbyparticipants’perceptions.Also,theEARcreatestheopportunitytoattendtosubtlesocialphenomena,suchaspersonalinteractionstylesandpreferences,thatparticipantstypi-callyarenotawareofwhenfillinginaquestionnaire.WereadilyacknowledgethattheEARisnotnecessarilyamoreobjectiveevent-samplingtool.Rather,itprovidestheresearcherwithauniqueperspective,thatofanunobtrusivecompanionwhohasaccesstoasourceofsocialinformationthattheparticipantdoesnothave.Thus,differentpicturesemergingfromtheEARascomparedwithtraditionalself-reportdatacallforastudyofhowthetwoviewpointsdifferratherthanforacompetitionforthemostaccurateaccount.
Naturalconversationsconstitutethesecondsourceofinforma-tiondocumentedbytheEAR.Consideringthatawidearrayofpsychologicalconstructsincludepredictionsabouthowindividu-alsinteractwithothers,itissurprisingthatstylisticaspectsofoureverydayconversationshavelargelybeenneglectedinthestudyofperson–situationinteraction(forsomeofthefewexceptions,seeDunbar,Duncan,&Marriott,1997;Emler,2001;Reis&Wheeler,1991).Theideathatlanguageuse—or,morespecifically,wordchoice—canrevealpsychologicalinformationaboutthespeakerisnotnew(forabroadreview,seePennebaker,Mehl,&Niederhof-fer,2003).Asearlyas1942,FillmoreSanfordproposedthatverbalbehaviorcanserveasapowerfulpersonalitymarker(Sanford,1942).Severalresearcherssincethenhavelinkeddifferencesinwordusetoaspectsofpeople’sself-reportedpersonality,includingtheBigFive(Pennebaker&King,1999),dispositionalmoodcharacteristics(Weintraub,1989),self-esteem(Bosson,Swann,&Pennebaker,2000),orself-monitoringandMachiavellianism(Ickes,Reidhead,&Patterson,1986).
Languageusealsoreflectsaspectsofthesocialsituationsthatpeoplearein(e.g.,P.Brown&Fraser,1979;Forgas,1985).In
THEPSYCHOMETRICSOFSOCIALLIFE
859
carefulsociologicalanalyses,ErvingGoffman(1981)identifiedlanguageuseasanimpression-managementtoolthathelpsnego-tiatedifferentaspectsofourselvesindifferentsocialsettings.ForGoffman,thewordsthatpeoplespeakareoftennottheirownbutinsteadreflecttransientsituationalroleconstraintsratherthanstablepersonalcharacteristics.Forexample,thedegreeofformal-ityofasocialsetting(e.g.,R.Brown&Gilman,1960),powerdisparitiesbetweentheinteractants(e.g.,Morand,2000),ortheintimacyofarelationship(e.g.,Weintraub,1989)clearlyaffectfeaturesoflanguageuse.
Takentogether,analysesoflanguageusecanprovideawindowtoaccessabroadspectrumofpsychologicallyrelevantinformationaboutaperson’ssocialworld.Methodologically,thisapproachconstitutesanimplicitassessmentstrategy.Wordchoice—com-paredwithaperson’slocationsandactivities—isasubtlesourceofsocialinformationthatnormallyfallsbeyondaperson’sawarenessandcontrol.Insamplingsnippetsofnaturalconversation,theEARthusprovidesauniqueopportunitytostudyautomaticlinguisticprocessesinnaturalisticsettings.
Insum,thetrackingofpeople’sdailylifewiththeEARcapturesawidespectrumofimportantsocialinformation.Suchasystemhelpsresearchersadoptaperspectivethathaslargelybeeninac-cessibleinthestudyofnaturalisticpsychologicalphenomena.
PurposeoftheStudy
Giventhenoveltyofthemethod,thecurrentprojectsoughttodeterminethebasicpsychometricpropertiesofstudents’everydaysociallivesasassessedbytheEAR.Ontwoseparateoccasions,participantsworetheEARcontinuouslyfor48hronnormalweekdays.Thetwomonitoringperiodswerescheduled4weeksapart.1Threebroadresearchquestionsaroundpeople’severydaysocialenvironmentsandlanguageuseintheirnaturalconversa-tionsguidedtheanalyses:(a)Howstablearepeople’ssocialenvironmentsovertime?(b)Howreliableispeople’slanguageuseovertimeandacrossinteractants?(c)Howstableislanguageuseacrossdifferentsocialcontexts?
AssessingtheStabilityofSocialEnvironments
Baseratesonhow,where,andwithwhompeoplespendtheirdaysareavailablefromnationalandinternationaltime-budgetstudies(Robinson,1977;Szalai,Converse,Feldheim,Scheuch,&Stone,1972)aswellESstudieswithadolescentsandadults(Csik-szentmihalyi&Larson,1984;Larson,1990;Reis&Wheeler,1991).Whereastheseestimatesarebasedonretrospectiveormomentaryself-reportsandthusrepresentpeople’sperceptionoftheirdailylives,thisstudysoughttoestablishbaseratesforpeople’severydaylocations,activities,andconversationsderivedfromunobtrusiveobservationdata.
Ifindividualsactivelychooseandshapetheirsocialworlds,people’severydaysocialenvironmentsshouldbecharacterizedbyasubstantialdegreeofstabilityacrosstime.Researchevolvingaroundtheperson–situationdebatehasunambiguouslydemon-stratedthestabilityofpeople’severydaybehaviorswhenbehaviorisaggregatedinpsychometricallysoundways(e.g.,Diener&Larson,1984;Epstein,1979;Moskowitz,1982;Nezlek,1993;Reis,Lin,Bennett,&Nezlek,1993).Test–retestcorrelationstyp-icallyrangesomewherebetween.50and.80dependingonthe
behaviorinquestionandthemodeofaggregation.Fleeson(2001)recentlydemonstratedthatwhenparticipantsratetheirmomentaryexperiencesintermsofBigFiverelevantstates,theresultingaggregateddistributionsshowalmostperfectstability.Unaffectedbylargedailywithin-personvariability,thedegreetowhichpeopleratedthemselvesasactinginsocial,extraverted,oragreeablewayswashighlyreliableovertime.
However,mostoftheexistingresearchaddressesthestabilityquestionfromwhatwehavedescribedastheagent’sperspective(Hogan,1982).IntheESstudiespresentedbyFleeson(2001),forexample,extravertedstateswerehighlystable,butitisnotclear(a)whatspecificactsdifferentparticipantsconsideredextravertedor(b)howstablethesespecificactswouldbefromtheobserver’sperspective.Thus,Fleeson’sresults,aswithmostoftheresearchinthisarea,addresscoherenceofsocialbehavioronthelevelofsubjectiveinterpretation.WiththeEAR,wecanstarttodeterminethedegreeofstabilityinpeoples’everydaysocialenvironmentsandsocialbehaviorsfromtheoutsideperspective,aperspectivethatpeoplenaturallyadopt,forexample,whentheyformimpres-sionsaboutothers.
AssessingtheStabilityofPeople’sNaturalLanguageUse
Whereassomeresearchisavailableonthecontentofdailyconversations(Bischoping,1993;Dunbaretal.,1997;Emler,2001;Landis,1927;Moore,1922),thereisvirtuallynoinforma-tionavailableonhowpeoplenaturallytalkineverydaylife.BaseratesforwrittenlanguageuseacrossdifferentgenresareavailablefromBiber(1988)andPennebakerandKing(1999).Gleser,Gottschalk,andJohn(1959)presentedfrequencyinformationonaseriesoflinguisticvariablesthatweresampledfrom5-minspeechinalaboratorysetting.Thisstudysoughttoestablishbaseratesforspokenlanguageusederivedfromthewholespectrumofpeople’sspontaneouseverydayconversations.
Onlythreestudiessofarhaveinvestigatedthetemporalstabilityoflanguageuse.PennebakerandKing(1999)haveshownthatwrittenlanguageuseindiaries,classassignments,andevenpro-fessionaljournalabstractsissurprisinglyreliableoverweeksandevenyears—fromtheuseofpronounsandarticlestowordsreflectingemotionaltone.Gleseretal.(1959)foundevidenceforthestabilityofspokenlanguageoveraveryshorttimeperiod—twosuccessive2-minintervalstakenfroma5-minfreespeech.Across21languagecategories(e.g.,wordcount,adjectives,sub-stantives,pronouns,feelings)theaveragecorrelationwas.51.Finally,Schnurr,Rosenberg,Oxman,andTucker(1986)providedfurthersupportforthetemporalstabilityofspokenlanguageusebyreportinghighwithin-personrank-ordercorrelationsbasedon83linguisticvariablesoveraperiodof1week.Itisimportanttonote,however,thatthisstatisticalapproachcanonlybeconsid-1Threewritingsessionsabouteitherapersonaltraumaoraneutralcontroltopicwereinsertedmidwaybetweenthetwomonitoringperiods.Originally,theEARwasimplementedtoidentifysubtlesocialmediatorsofthehealthbenefitstypicallyexperiencedafteremotionalwriting(Penne-baker&Graybeal,2001).Becausepeople’ssociallivesfromtheEARperspectiveconstitutescientificallycompletelyunexploredterrain,wede-cidedtousethedatatofirstlaythepsychometricfoundationsforfutureEARstudies.Resultswithrespecttothewritinginterventionwillbepublishedelsewhere.
860
MEHLANDPENNEBAKER
eredindirectevidenceoftemporalstability,becauseitisinherentlyconfoundedwithnaturaldifferencesinbaseratesbetweenlinguis-ticvariables.Thisstudysoughttoestablishthestabilityofspokenlanguageoveralongerperiodoftime,aperiodof4weeks,andderivedfromarandomsampleofaperson’severydayspontaneousconversations.
Arelatedquestionconcerningreliabilityoflanguageusecon-cernsthedegreetowhichthespeaker’sworduseiscorrelatedwiththetarget’slanguage.Communicationaccommodationtheory(Giles,Coupland,&Coupland,1991)positsthatasmoothcom-municationrequiresthatbothinteractantsadapttoeachother’scommunicativebehaviors.Arecentstudyanalyzedthewaysdyadsusedlanguageinchatroomconversationsand,inthecaseoftheWatergatetapes,howRichardNixoninteractedwiththreeofhisaides.Ineachofthesettings,bothcommunicationpartnerstendedtousesimilarlinguisticforms(Niederhoffer&Pennebaker,2002).Howmuchoverlapistherebetweenpeople’slanguageuseintheireverydayconversationsandthelanguageusedbythosetheyaretalkingto?Thecurrentstudysoughttoaddressthisquestiononacumulativelevelbycomparingparticipants’wordusewiththewordssampledfrompeopleengagedinconversationswiththeparticipants.
AssessingtheStabilityofLanguageUseAcrossSocialContext
Thereisverylittledoubtthatlanguagechangesasafunctionofsituation(P.Brown&Fraser,1979;Forgas,1985;Pennebakeretal.,2003).Notsurprisingly,forexample,havingstudentseaves-dropandrecordsnippetsofconversationsbetweenstrangers,Cam-eron(1969)foundthattheuseofprofanityissubstantiallyelevatedinleisureascomparedwithjobsettings.TheEARallowsassess-mentofhowlanguageusechangeswhenpeoplechangetheirsocialenvironment.Forexample,dopeopletalkdifferentlywhentheyareathomeascomparedwithinapublicplace?Isthelanguagepeopleuseinanamusementcontextdifferentfromthelanguagetheyuseatwork?Dophoneconversationsrequireadifferentlanguagethanpersonalinteractions?Admittedly,ourexaminationofstabilityoflanguageuseacrosssocialcontextisrestricted,giventhewaystheEARallowsustocategorizesocialcontexts.
Tosummarize,thepurposeofthisfirstandlargelyexploratoryEARstudywastodeterminethebasicpsychometricsofpeople’severydaysocialenvironmentsandlanguageuseintheirnaturalconversationsasassessedfromanunobtrusiveobserver’sperspec-tive.Forbothsourcesofinformation,wewillpresentbase-rateinformationandtest–reteststabilitycoefficients.Forpeople’snat-urallanguageuse,wewillalsoincludeanassessmentofhowmuchbetween-speakerconsistencyexistsinlinguisticfeaturesofevery-dayconversations.Finally,wewillexplorelinguisticconsistencyacrossdifferentsocialcontexts.
Method
Participants
Fifty-fourintroductorypsychologystudentsattheUniversityofTexasatAustinsignedupforthestudy.Twostudentsdecidednottoparticipate,onebecauseofthe4-weektimecommitment,theotherbecauseofconcernsabouttheEAR.Theremaining52undergraduatestudents(28women,24
men;meanage19.0years,SDϭ1.3)participatedforcoursecredit.Allofthemcompletedtheexperiment.
EARSystem
TheversionoftheEARusedinthisstudyconsistedofamicrocassetterecorder(OPTIMUSMicro-32,FortWorth,TX),anexternalmicrophone(OPTIMUSOmnidirectionalTieClipMicrophone,FortWorth,TX),andacontrollermicrochip.Themicrochipwasprogrammedwitha30son,12.5minoffcycle2andthusproducedroughlyfiveintervalsperhour.Itwasimpossiblefortheparticipanttosensewhentherecorderwasonoroff.ParticipantscarriedtheEARinasmallshock-protectedcaseeitherat-tachedtotheirbelt(similartoacell-phonecase)oraroundtheshoulder(likeapurse).Anexternalmicrophonewasclippedtothelapelofajacketorthecollarofashirt.TheEARwasswitchedoffovernight.Forfurtherdetailsonthedevelopmentandtestingofthedevice,seeMehletal.(2001).
Procedure
ParticipantswereaskedtoweartheEARtwicefor48hrseparatedby4weeks.ThemonitoringsessionswerescheduledeitherfromMondaymorningtoWednesdaymorningorfromWednesdayafternoontoFridayafternoon.Foreachparticipant,bothmonitoringperiodswereoriginallyarrangedonthesameschedule.Thismatchingwassuccessfulfor37ofthe52students(71%).
Firstmonitoringperiod.Participantswereruningroupsof2to4.Onarrivalatthelab,studentswereintroducedtotheprocedureofthestudy.ConsiderableeffortwastakentoprovideathoroughexplanationoftheEAR.Itwasemphasizedthatthesamplingpatternresultedinonly4%oftheirdaybeingcapturedandthattheshortdurationoftherecordingblocksresultedinonlybriefsnippetsratherthancompleteconversationsbeingcaptured.AllparticipantswereencouragedwhentheyreturnedtheirEARstolistentotheirrecordingsanderaseanypartstheyconsideredobjection-ableorthattheypreferredtoremainprivatebeforetheresearcherslistenedtotherecordings.Oncompletionofaquestionnairepackage,participantswerehandedtheEARequippedwitha90-minmicrocassettetape.Becauseofrestrictionsintapelength,participantshadtoflipthetapeafter24hrofmonitoring.All52participantsrememberedtodosowithinatoleranceperiodofϮ2hr.
Whentheparticipantsreturnedtothelab2dayslater,theexperimenterofferedthemtheoptionoflisteningtotheirrecordings.3A7-itemques-tionnaireobtainedfeedbackontheexperienceswiththesystem.Ona5-pointratingscalefrom1(notatall)to5(agreatdeal),participantsratedthedegreetowhichtheywereawareoftheEAR,thedegreetowhichwearingthesystemchangedtheiractualbehavior,andtheextenttowhichotherpeoplerecognizedtheEAR.
Secondmonitoringperiod.Theprocedureforthesecondmonitoringsessionwasvirtuallyidenticaltothefirstone.Attheendofthesession,participantswerethoroughlydebriefedaboutthepurposeofthestudy.
DataPreparation
Participants’socialenvironmentswerecodedfromthecapturedambientsounds.Allrecordedconversationsweretranscribedandsubmittedtolinguis-ticanalyses.Thecodingandthetranscribingweredonesimultaneously.
2Attheonsetofeachrecordingsample,a5-speriodisrequiredfortherecordingvolumetoreachnormallevels.Thisdelay,experiencedwhileinternalrecordercomponentsapproachtheirquiescentoperatingstate,leavesabout25sofusefuldata.3Itisinteresting,consideringthedegreeofintimacyofthedata,thatoutofthe52students,only1decidedtoactuallycheckpartsoftherecording,andthisstudentdidnoteraseanything.
THEPSYCHOMETRICSOFSOCIALLIFE
861
Judges’codingofsocialenvironment.Researchassistantslistenedtothecompleterecordings(bothmonitoringperiodswerecodedbythesameperson).Attheendofeachinterval,theycodedthesocialenvironmentusingtheSocialEnvironmentCodingofSoundInventory(SECSI).TheSECSIisacodingsystemthatcomprisestheperson’scurrentlocation(e.g.,inapartment,outdoors,intransit),activity(e.g.,listeningtomusic,onthecomputer,eating),andinteraction(e.g.,alone,onthephone,talkingtoothers).AlistofcategoriesiscontainedinTable1.
Inadditiontoacousticcuessuchasthenoiseofarunningengine(intransit),thesoundofwindblowing(outdoors),typingnoises(computer),orthevoiceofalecturer(lecture),judgesusedcontextinformationfrompreviousandconsecutiveintervalstoincreasetheiraccuracy.Forexample,ifaperson,afterbeingoncampus(Interval1)andridingonabus(Interval2),entersanapartment(Interval3),itisinferredthatthestudenthasreturnedhome.Thecertaintyofthejudgmentisthenfurtherenhancedbytheinformationfromthesubsequentrecordingperiods,inwhichthepersonmighthaveswitchedontheTVorgonetotherefrigeratortogetsomethingtoeat.
Judges’reliabilitiescalculatedfromatrainingtapeindependentlycodedbyallresearchassistantsyieldedameanCronbach’salphaof.94.Forallcategoriesbutreading(␣ϭ.12)andeating(␣ϭ.64),consistencycoefficientswere.70orgreater.MoredetailedinformationontheSECSIandthetrainingofthejudgeswasgiveninMehletal.(2001).
Tapetranscriptionandlinguisticanalysis.Inadditiontocodingthesocialcontext,researchassistantsalsotranscribedalllanguagecapturedbytheEAR.Theyreceivedspecialtrainingindealingwiththechallengesposedbytranscribingorallanguage,suchashandlingofrepetitions,fillerwords,nonfluencies,orslang.Foreach30-srecordingblock,thelanguagesampleswereidentifiedascomingfromeithertheparticipant(P)oranotherliveperson(O)talkingwithorinimmediateproximityoftheperson.Thetranscribedlanguagesampleswerethensortedbyspeaker(P,
Table1
MappingStudents’SocialEnvironment:Judges’ReliabilitiesandBaseRatesfortheElectronicallyActivatedRecorderRatingsofParticipants’DailyInteractions,Activities,andLocationsAcross4DaysofMonitoring
Baserate(%)
Judges’Category
reliabilityMSDMinMaxInteractionAlone.9368.615.037.097.6Talking.9927.912.81.756.1Toothers.9924.212.31.755.4Onthephone1.003.83.20.013.3Laughing.865.93.90.015.8ActivityMusicon.8913.59.70.041.4TVon.9514.612.21.147.5Computer.968.57.40.038.5Reading.1211.310.20.034.0Working.955.19.70.037.0Eating.642.62.20.011.8Lecture.9811.36.50.029.0Amusement.7114.118.00.071.4LocationApartment.9756.217.911.486.5Outdoors.907.25.21.032.9Intransit.854.14.20.016.0Restaurant
1.002.13.70.024.7Otherpublicplaces
.86
27.7
14.4
5.3
70.0
Note.Nϭ49;reliabilitycoefficientsareCronbach’salphasbasedonsixtranscribersscoring88intervals.Minϭminimum;Maxϭmaximum.
O)andsubmittedtoalinguisticanalysisusingtheword-basedlanguageanalysisprogramLinguisticInquiryandWordCount(LIWC;Pennebaker,Francis,&Booth,2001).
LIWCoperatesbycomparingallwordsofatextdocumenttoaninternaldictionaryconsistingofmorethan2,300wordsandwordstems,fallingintoover70categories(e.g.,first-personsingularpronoun,article,prepo-sition,negativeemotion,socialprocess,presenttense).EachwordisassignedtothoseLIWCcategoriesthatapplyonthebasisofitsoccurrenceintherespectivesubdictionary.Forexample,thetranscribedwordcriedwouldfallintofourcategories:“sadness,”“negativeemotion,”“overallaffect,”and“past-tenseverb”(notethatmostLIWCcategorizesarehier-archicallyorganized).Exceptfortotalwordcount,theLIWCoutputislistedasthepercentageoftotalwordsperpersonpermonitoringperiod.Eachparticipant’srawdataresultedin2(speaker:P,O)ϫ2(firstandsecondmonitoring)linguisticanalyses.
Results
EvaluationoftheEARMethodology
EARfeedback.Attheendofbothmonitoringperiods,partic-ipantscompletedaseven-itemquestionnaireontheirexperienceswearingtheEAR.Consistentlyloweritemmeansduringthesec-ondmonitoringperiodcomparedwiththefirstindicatedhabitua-tionovertime.Evenafterthefirst2daysofwearingtheEAR,however,participants’ratingsofitsinvasivenesswerewellbelowthescalemidpoints.Acrossall4days,alongthe5-pointunipolarscale,where5ϭagreatdeal,studentsratedthatthey(Mϭ2.67,SDϭ0.81)aswellaspeoplearoundthem(Mϭ2.68,SDϭ0.84)weremoderatelyawareoftheEAR.TheyreportedfeelingonlyslightlyuncomfortablewearingtheEAR(Mϭ2.05,SDϭ0.70)andindicatedthedegreetowhichthemonitorchangedtheirbehavior(Mϭ1.43,SDϭ0.52)andtalking(Mϭ1.25,SDϭ0.44)asminimal.Altogether,theEARwasnotparticularlydistractingnordidithaveasignificantimpactontheparticipants’socialbehaviors.
EARcompliance.Participantsshowedahighdegreeofcom-mitmentwearingtheEAR.AllparticipantsreturnedtheEARfortheirscheduledappointment;nostudentforgottoflipthetapeafter24hrofmonitoring;eventhemostprivateconversationsandbehaviorswerecapturedand(surprisingly)noterasedbypartici-pantsattheendofthestudy.
Onthebasisofatargetedmonitoringtimeof15hrperday(assuming8hrofsleepandanadditionalhourofnotwearingtheEARbeforegoingtobed[30min]andaftergettingup[another30min])and4.8recordingperiodsperhour(12.5-mincycles),theupperlimitofintervalspermonitoringperiodwas144.Inoursampleof52introductorypsychologystudents,theEARrecordedanaverageof102(SDϭ34)intervalsduringthefirstmonitoringperiod.Overthesecond2daysofmonitoring,thismeanincreasedto113(SDϭ33)timeblocks.Thelowernumberforthefirstmonitoringperiodreflectstechnicalproblems(unreliabletrigger-ing)experiencedintheinitialphaseofthestudy.Estimatesbasedondebriefinginterviewsanddocumentedrecordingtimessuggestthatacrossall4daysofmonitoring,onaverageabout15intervalsperparticipantwerelostbecauseoftechnicalproblems(e.g.,failuresofmicrochiptriggers),andabout20intervalsbecauseofparticipantsnotwearingtheEAR(mostlyduringsports,exercise,andpersonalhygiene).Only2participantsmentionedhavingturnedofftheEARduringintimateconversationswiththeirpart-ners(bothforlessthananhour).
862
MEHLANDPENNEBAKER
AnalysesofParticipants’SocialEnvironments
Thejudges’codingsoftheparticipants’socialenvironmentswereaveragedacrossallintervalsofeach2-daymonitoringperiod.TheSECSIvariablesthusrepresentpercentagesofallintervalsinwhichacertaincategoryapplied.4Threeparticipantsprovidedfewerthan50intervalsononeoftheirtwomonitoringperiods.Theirdatawereexcludedfromtheanalysesbecauseofinsufficientsampling.Thefinalsamplefortheanalysisofstudents’dailysocialenvironmentscomprised49participants(22men,27women).
Baserates.Table1depictsthedescriptivestatisticsfortheSECSIcategoriesaggregatedover4days.Itwassurprisingtolearnthatstudents’activesociallifeisrestrictedtoroughlyathirdoftheirwakinghours.Theaverageparticipantspentmorethantwothirdsoftheirwakingtimesurroundedbynothingbutbackgroundnoises.InonlyabouteveryfourthintervaldidtheEARcaptureaparticipanttalkingtoanotherperson.Studentsonaveragewereonthephoneinanother3.8%oftheintervals.Interestingly,alookatthevariabilityofthemeasuresalsorevealedstrikingindividualdifferences.Whereas1participantwaswithothers67%ofthetime,2participantsspentvirtuallytheentire2daysalone.Asimilarpictureemergedfortheotherinteractioncategories,“talk-ingtoothers”and“onthephone.”
Insomewhatlessthanathirdoftheintervals,participantsexposedthemselvestomediasuchastheTVorradio.Lectureattendancewascapturedin11.3%ofthecasesand—evenonweekdayswithinthesemester—14.1%ofthesoundsampleswerejudgedbytheratersascapturingactivitiesdeemedasamusement(e.g.,gameparlor,sportingevent).
Intermsofwhereparticipantsspenttheirdays,judgeslocatedtheminsidetheirapartmentsinmorethanhalfofthecases;outdoorsin7.2%oftheintervals;intransit4.1%ofthetime;insidebars,restaurants,orcoffeeshopsin2.1%oftheintervals;andinotherpublicplaces,includingthecampus,arcades,malls,orgro-cerystores,in27.7%ofthecases.
Four-weekstability.Toassessthereliabilityofpeople’sen-vironments,simplecorrelationswerecomputedbetweenamountoftimespentwithineachSECSIcategoryatTime1andthetimespentatTime2.AscanbeseeninFigure1,correlationsforthesocialcontextcategoriesshowedthatstudents’socialenviron-mentswerecharacterizedbyahighdegreeofstabilityovera4-weekperiod.Theaveragetest–retestcorrelationwasrϭ.54fortheinteractioncategories,rϭ.50forthedailyactivities,andrϭ.65forparticipants’locationvariables.Allbutonecorrelation(eating)wasstatisticallysignificant(pϽ.05,one-tailed).Over-all,12outofthe17correlationswereequaltoorgreaterthan.50.Interestingly,inherently“social”variablessuchasbeingalone,talkingtoothers,laughing,oramusementwereamongthemoststableacrossthe1-monthinterval.Itisalsonoteworthythatstudents’lectureattendance—despitethematchingoftheweek-daysforthefirstandsecondrecordingperiod—turnedouttobeonlymoderatelystable.
AnalysesofNaturalLanguageUse
Aswiththesocialenvironmentanalyses,aminimumnumberofsamplingsinanygivenmonitoringperiodwasrequiredtoensurereliabilityofthelinguisticanalyses.Participantsforwhomthe
EARcapturedfewerthan50wordsina2-dayperiodwereex-cluded.Thelanguagesamplesof5participantsdidnotmeetthiscriterion.Oneofthese5studentswasalsoexcludedfromtheanalysesofsocialenvironment.Notethattheremaining4studentsdidprovidesufficientrecordingintervalsbutspoketovirtuallynooneduringthe2-dayperiod.Threeofthese4“nontalkers”(andalsothefifth“nonsampler”)weremen.Thefinalsampleforthelinguisticanalysisconsistedof47participants(20men,27women).
Fromthepotential70ϩLIWCvariables,the16variablesiden-tifiedbyPennebakerandKing(1999)assufficientlyreliablewereretainedforthepresentstudy.ThreeadditionalLIWCvariableswereincludedinthecurrentprojectbecauseoftheirrelevancetospoken(asopposedtowritten)language:swearwords,nonfluen-cies,andfillerwords.Finally,thecompletesetofpersonalpro-nouns(first,second,andthirdpersoninsingularandplural)wasincludedbecausereferencestopersonsprovideimportantinfor-mationaboutpeople’srelativepositionsintheirsocialnetworks.Thefinalsetof23LIWCcategories,showninTable2,wassubjectedtothelinguisticanalysesofparticipants’dailynaturallanguageuse.
Baserates.Table2depictsparticipants’languageuseinev-erydayconversations.Overa4-dayperiod,theEARsampledonaverageslightlymorethan1,000wordsperperson,rangingfrom164toalmost3,000.Roughlyeveryseventhspokenwordwasapersonalpronoun,withalmosthalfofthemreferencestoself(I,me,my).Participantsusedswearwordsin5outof1,000words.However,thevariabilitybetweenparticipantswasastonishinglyhigh.Whereas1personusedswearwordsatthesameratethattheaveragepersonusedpositiveemotionwords,11participants(47%)didnocursingatall.Fillerwordssuchaslike,well,youknow,orIdon’tknowweresampledatarateof1.7%.Becauseofthenatureofeverydayconversations,participantsusedsocialwordsandpresent-tenseverbsatahighrate.
Four-weekstability.The4-week-stabilitiesforthelinguisticcategoriesareshowninFigure2.OnthebasisofasamplesizeofNϭ47,16outof23correlationspassthethresholdofsignifi-cance.Thetest–retestcorrelationsforthe12standardlinguisticcategoriesaveragedrϭ.41,withespeciallyhighcorrelationsforcategoriesuniquetospokenlanguage(swearwords,nonfluencies,fillerwords).Theonlyexceptionamongthestandardlinguisticcategorieswasintheuseofsecond-personpronouns(rϭϪ.10).TheaverageretestcorrelationofthesevenLIWCcategoriesindi-catingpsychologicalprocesseswasrϭ.24andmainlyreflectedaratherstableuseofpositiveandnegativeemotionwordsovertime.Inaddition,theuseofpresenttensewasmarkedlymorestablethantheotherthreerelativitycategories.
4Atthispointadecisionhadtobemadeonhowtohandle“empty”intervalswhere,becauseofalackofambientsounds,nojudgmentofcontextcouldbemade.In609of10,944intervals(5.6%),judgeswereabletodecipherneithertheparticipants’locationsnortheiractivitiesorinter-actions.Separateanalysiswithandwithouttheseintervalsyieldedcom-parableresults,withconsistentlyslightlyhigherbaserateswhencalculatedfromthereducedsetofintervals.BecauseemptyintervalsconstituteamethodologicalchallengefortheEARratherthananonbehavioronbehalfoftheparticipant,weconsideredtheresultsbasedonthenonemptyintervalsonlymoreaccurateproxiesofaperson’srealbehavior.
THEPSYCHOMETRICSOFSOCIALLIFE
863
Figure1.Four-weekstabilityofjudges’ratingsofsocialcontext:Test–retestreliabilitiesforall17SocialEnvironmentCodingofSoundInventorycategories.Nϭ49.
Table2
Students’LanguageUseinNaturalConversations:BaseRatesforSelectedLIWCCategoriesAcross4DaysofMonitoring
LIWCcategories
StandardlinguisticdimensionSampledrawwordcountWordsofmorethan6lettersFirst-personsingularpronounsFirst-personpluralpronounsTotalsecond-personpronounsTotalthird-personpronounsNegationsArticlesPrepositionsSwearwordsNonfluenciesFillerwords
PsychologicalprocessesPositiveemotionsNegativeemotionsCausationInsight
DiscrepancyTentative
SocialprocessesRelativityPasttensePresenttenseInclusiveExclusive
Examples
M1,064.4
8.86.91.03.72.83.43.98.90.50.41.73.21.41.22.22.32.411.14.515.94.94.1
SD559.41.71.80.61.01.11.21.01.40.70.51.41.20.80.50.60.70.82.01.52.00.90.9
Min146.05.23.20.01.50.81.61.52.80.00.00.00.50.00.30.80.80.95.52.010.93.01.9
Max2,943.013.211.12.65.75.79.26.211.73.42.45.76.84.12.83.64.75.116.87.820.26.86.0
I,me,myWe,us,ourYou,yourShe,him,theirNo,not,neverA,an,theTo,with,aboveDamn,bastard
Uh,er
Yaknow,like,ImeanHappy,goodHate,uglyBecause,effectRealize,knowWould,shouldPerhaps,maybeFriend,talkWas,wentIs,goWith,andExcept,but
Note.Baserates(exceptrawwordcount)areexpressedinpercentoftotalwordsacross4days.Fillerwordsorphrases(e.g.,likeoryouknow)wereidentifiedbytranscribersandconvertedintouniquelyidentifiedsinglewords(e.g.,yaknow,Imean)thatwerecapturedbyLinguisticInquiryandWordCount(LIWC).Minϭminimum;Maxϭmaximum.
864
MEHLANDPENNEBAKER
Figure2.Four-weekstabilityofthelinguisticcategories:Test–retestreliabilityforthe23LinguisticInquiryandWordCount(Pennebakeretal.,2001)categories.Nϭ47.WCϭwordcount;SLWϭwordsofmorethansixletters.
Stabilitybetweenspeakers.Totestforlinguisticsynchronyinpeople’severydayconversationsonagenerallevel,wecalculatedthedegreeofsimilaritybetweentheoveralllanguageusebytheparticipantsandtheirinteractionpartners.Recallthatinadditiontotheparticipants’language(P)wealsotranscribedwhenanotherperson(O)wastalkingwithorinimmediateproximityoftheparticipant.TheseO-languagesamples,irrespectiveoftheactualpersontheyoriginatedfrom(e.g.,friend,romanticpartner,stranger)wereanalyzedcumulativelyandresultedinoneoveralllinguisticprofileofpeopletalkingwiththeparticipant.
Figure3showstheP–Ocorrelationsforthe23selectedLIWCcategoriesonthebasisoftheentire4daysofmonitoring.Withanaveragecorrelationofrϭ.35,thestandardlinguisticLIWCvariablesshowthehighestcorrespondencebetweenspeakers.Cat-egoriesuniquetospokenlanguageagainrangeamongthecatego-rieswiththehighestcorrelations(swearwords,nonfluencies,filler
Figure3.Similaritybetweentheparticipant’sandhisorherinteractionpartners’overalllanguageuse:Person–othercorrelationsforthe23LinguisticInquiryandWordCount(Pennebakeretal.,2001)categories.Nϭ47.WCϭwordcount;SLWϭwordsofmorethansixletters.
THEPSYCHOMETRICSOFSOCIALLIFE
865
words).Bothoverallwordcountswerehighlycorrelated(rϭ.72).Theuseofsecond-personpronounsbetweenspeakerswasunre-lated(rϭϪ.16).Interestingly,theuseoffirst-personsingularpronounsfailedtoshowP–Ocorrespondence(rϭ.06).
NaturalLanguageUseandSocialContext
Students’dailysocialenvironmentsaswellastheirlanguageuseinnaturalconversationarereliableovertime.Isthereliabilityoflanguageusedue(atleasttosomeextent)tosituationalcontextsdrivingcertainkindsoflanguageuse?Ifso,linguisticreliabilitywouldemergeasanepiphenomenonofsituationalreliability.Unfortunately,astrongtestofthisquestionwasrestrictedbytwofactors:ourabilitiestocategorizesituationsandthecomparativelysmallsamplesize.Thefollowinganalyses,then,shouldbecon-sideredpreliminary.
Forthecontextanalysesoflanguageuse,anewdatasetwascreatedwherethelanguageutteredbytheparticipantduringeachintervalwassubmittedtoaseparateLIWCanalysis.5Thenewinterval-basedLIWCinformation(e.g.,percentageofemotionorswearwordsduringthe30-sEARrecording)wasmatchedwiththeinterval-basedinformationaboutstudents’socialcontextderivedfromthejudges’SECSIratings.Thisdatafilewasthenaggregatedbyparticipantsandsocialcontexts,sothataveragedLIWCdatawereavailableforeachparticipantonthesocialcontextswesoughttocompare.Theanalysesarebasedontheparticipantswhowereincludedinboththelinguisticaswellasthesocialcontextanalysis(Nϭ45).
Languageuseacrosslocation.Towhatextentdoparticipants’locationsconstraintheirlanguageuse?Forthis,theoriginalSECSIlocationcategories“outdoors”and“intransit”werecombinedinto“outside,”and“restaurant”and“otherpublicplaces”weremergedinto“inpublicplaces.”Thenewcategorieswerethencontrastedwiththecategory“apartment.”Aseriesofwithin-subjectanalysesofvariance(ANOVAs)onthe23LIWCvariablesrevealedsig-nificantcontexteffectsonlyforwordcount,F(2,80)ϭ4.34,pϽ.05,andfirst-personpluralpronouns,F(2,80)ϭ4.21,pϽ.05.Participantsutteredsignificantlyfewerwordsina30-speriodwheninpublicplaces(Mϭ18.5%)thanwhenintheirapartments(Mϭ23.4%),t(40)ϭ3.28,pϽ.01,oroutside(Mϭ22.1%),t(40)ϭ2.16,pϽ.05.Also,theyusedfirst-personpluralpronounsatasubstantiallyhigherrateinpublicplaces(Mϭ1.0%)ascomparedwithathome(Mϭ0.7%),t(40)ϭ3.94,pϽ.01.Languageuseacrossactivity.Asecondquestioniswhetherstudents’languageusechangedasafunctionoftheiractivitywithinaspecificlocation.OnthebasisoftheSECSIinformation,talkinginpublicplacesoccurredmainlywhileworkingorpursu-ingactivitiesdeemedamusements.Otheractivitycategories,suchas“onthecomputer,”“reading,”“eating,”or“attendingalecture,”hadinsufficienttalkingbaseratestobeincludedinthecontextanalysis.Accordingly,withinthecategory“inpublicplaces,”thecategories“working”and“amusement”werecontrastedwitharesidualcategory,“otheractivities.”Unfortunately,only13par-ticipantsprovideddataonallthreeactivitycategories.Onthebasisofthisveryrestrictedsample,within-subjectANOVAsfortheLIWCvariablesrevealedsignificantmeandifferencessolelyinfirst-personsingularpronounuse,F(2,24)ϭ5.16,pϽ.05.Participantsreferredtothemselvesmorethantwiceasmuchin
amusementcontexts(Mϭ10.2%)thanwhileworking(Mϭ3.8%);t(12)ϭ3.63,pϽ.01(Motherϭ6.2%).6Languageuseacrossmodeofinteraction.Finally,wewereinterestedintheextenttowhichthemodeofinteraction,talkingtoapersondirectlyversusonthephone,determinedlanguageuse.Toaddressthisquestion,languagecapturedbytheEARwhilepar-ticipantswereonthephoneinsidetheirapartmentswascontrastedwithtranscriptsofrealconversationsthatwerecapturedathome.Notethatwewantedtolookathowlanguagechangesasafunctionofsocialcontextwithinagivenlocation.Theanalysesrevealedthatlanguageusewasindependentoftheactualmodeofinterac-tion.Statisticallysignificantdifferencesonlyoccurredin4outofthe23LIWCcategories.Onaverage,inphoneascomparedwithrealconversations,participantsutteredmorewordsperinterval(Mϭ30.1%vs.Mϭ21.5%),t(39)ϭ3.70,pϽ.01;usednon-fluenciesatahigherrate(Mϭ1.0%vs.Mϭ0.4%),t(39)ϭ2.11,pϽ.05;usedfewerfirst-personpluralpronouns(Mϭ0.5%vs.Mϭ0.8%),t(39)ϭϪ2.03,pϽ.05;andusedfewerinclusivewords(Mϭ3.4%vs.Mϭ4.2%),t(39)ϭϪ2.02,pϽ.05.Summary.Inthecross-contextanalyses,languageuseemergedasstableacrosslocations,activities,andmodesofinter-action.Contrastingparticipants’wordchoiceathome,outside,andinpublicplaces;theirlanguageinamusementversusworkcon-texts;andtalkingstylesinpersonalconversationsversusphonecallsathomerevealedonlyaverylimitednumberofsignificanteffects.Theresults,however,shouldbeinterpretedcautiouslyfortworeasons:First,thecontrastswerelimitedbytherestrictedcontextinformationavailablefromtheSECSI.Second,theywerebasedonarestrictedsamplebecausenotallparticipantsprovidedlanguagedataacrossallcontexts.
OtherRelevantData
Althoughnotthefocusofthisarticle,thedatasetallowedustoperformexploratoryanalysesongenderdifferencesinstudents’dailysocialenvironmentandlanguageuseinnaturalconversa-tions.Relativelyfewdifferencesinpreferencesforsocialsettingsandactivitiesemerged.Overaperiodof4days,femaleascom-paredwithmaleparticipantsspentsignificantlymoretimeonthephone(Mfemaleϭ4.8%vs.Mmaleϭ2.5%,MpϽ.05)andlesstimeonthecomputer(Mfemaleϭ6.1.%vs.maleϭ11.4%,pϽ.05).AllotherSECSIcategoriesyieldedstatisticallyequalmeanbaserates.Comparedwiththerelativelyfewdifferencesinpreferencesforsocialsettingsandactivities,thelanguageusedbymaleandfemaleparticipantsintheireverydayconversationsdifferedinanumberofways.Consistentwithpreviousresearch(Pennebaker&King,1999),menusedsignificantlymorebigwords(wordsmorethansixletterslong;Mmale4.4%ϭvs.9.4%Mvs.Mfemaleϭ8.3%,pϽ.05),morearticles(Mmaleϭfemaleϭ3.5%,pϽ.01),fewerfirst-personsingularpronouns(Mmaleϭ6.2%vs.Mfemaleϭ7.5%,pϽ.01),andfewerdiscrepancywords(Mmaleϭ2.0%vs.Mfemale5FollowingastrategyadoptedbyNiederhofferandPennebaker(2002),werestrictedtheinterval-basedLIWCanalysistoutterancesofatleastthreewords.In214outof2,449intervals(8.7%),participantsweresayingfewerthanthreewords.Theseintervalswereexcluded.6Theresultsoftheposthoctestsarebasedonthe13participantswhowereincludedintheoverallANOVA.Analysesusingpairwiseexclusionyieldedsimilarresults.
866
MEHLANDPENNEBAKER
ϭ2.5%,pϽ.05)thanwomen.Also,men’stranscriptscontainedfourfoldtheamountofswearwords(Mmaleϭ0.9%vs.Mfemaleϭ0.2%,pϽ.01),considerablyfewerfillerwords(Mϭ2.1%,pϽ.05),andfewerreferencesmaleϭto1.3%positivevs.Mfemaleemotions(Mmaleϭ2.8%vs.Mfemaleϭ3.5%,pϽ.05).
Somewhatsurprisingly,maleparticipantsusedsignificantlymorenegativeemotionwordsintheirdailylanguage(Mmaleϭ1.9%vs.Mfemaleϭ1.1%,pϽ.01).Moredetailedanalysesrevealedthatthiseffectwasexclusivelyproducedbymenusinganalmostthreefoldamountofangerwords(M0.4%;pϽ.01).Nodifferenceemergedmaleinϭthe1.1%usevs.Mfemaleϭofanxiety-andsadness-relatedwords.Also,contrarytothewidelyheldnotionofwomen’shighersocialrelatedness,women’severy-daylanguagedidnotcontainmorereferencestosocialprocessesormorefirst-personpluralpronouns.
Discussion
Thepurposeofthisstudywastodeterminethebasicpsycho-metricpropertiesofpeople’ssociallivesasmeasuredbytheirdailysocialenvironmentsandtheirlanguageuseintheirnaturalconversations.UsingEARmethodology,weadoptedtheuniqueperspectiveofanunobtrusiveobserver.Theresultsindicatethatstudents’sociallivesshowaremarkabledegreeofstabilityevenfromthisoutsiderperspective.Theysupporttheideathatcoher-enceinpersonalityisnotmerelyanexperientialphenomenoninpeople’smindsbutisalsodrivenbyconsistencyinobjective—inthesenseoftraceable—aspectsofpeople’severydaysociallives.
Students’DailySocialLivesFromanAcousticPerspective
EARbaseratesforstudents’sociallives.Oneofthegoalsofthisstudywastodeterminebaseratesforpeople’ssocialenviron-mentsandnaturalconversationsfromtheperspectiveofanunob-trusiveobserver.Byandlarge,theestimatesthatemergefromthisstudymapontobaseratesobtainedfromtime-budgetstudies(Robinson,1977,1985;Szalaietal.,1972)—predominantlybasedonretrospectiveself-reports—andESstudies(Csikszentmihalyi&Larson,1984;Larson,1990).ThisisparticularlytruefortheSECSIlocationandactivitycategories.Here,substantialdiffer-encesonlyemergedforwatchingTV(EAR14.6%vs.ES7.2%)andlisteningtomusic(EAR13.5%vs.ES1.4%)andarelikelydrivenbyinterpretationaldifferencesinwhatconstituteswatchingTVorlisteningtomusic.ItisimportanttonotethattheSECSIlocationandactivitycategoriescomprisehighlyobservable,thatis,acousticallydetectable,behaviorsandevents.Observabilityhasbeenshowntofacilitateself–otheragreement(Funder&Colvin,1997;Goslingetal.,1998).
ThepicturelooksdifferentfortheSECSIinteractioncategories.Bothtime-budgetandESstudieshavefoundthatthetimeadultsspendalone(andawake)fallssomewherebetween4hrand6hrperday,whichisequivalentto25%–38%oftheirtimeawake,whereastheEARyieldedanestimateof68.6%.ESstudieshavepinpointedstudents’amountofdailyconversationsataround6hr—about38%oftheday(Reis&Wheeler,1991)—whereastheEARdatasuggest27.9%.Interpretationaldifferences,suchaswhatitmeanstobealoneorengagedinaconversation,againemergeasapotentialexplanation.TheEAR,forexample,identi-
fiesapersonstudyingatacoffeeshopas“alone”becausethere,thepersonis“a‘single,’apartyofone,apersonnotinawith,apersonunaccompanied‘socially’byothersinsomepublicunder-taking”(Goffman,1981,p.79),whereastheparticipantsthem-selvesmostlikelywouldreportbeingwith—inthesenseof“sur-roundedby”—people(e.g.,Larson,1990).
Anotherpossibilityisthatthetimingofthesamplingexplainsthedifferencesintimeestimates.Thatis,themidweek-onlysam-plinginthisstudyversustheweekend-alsosamplingintypicaldiarystudies(e.g.,Fleeson,2001;Larson,1990;Reisetal.,1993)couldhaveresultedindifferentialpatternsofsocialinteractions.Surprisingly,althoughcertainqualitativeaspectsofpeople’ssocialinteractionshavebeenshowntochangeovertheweekend(K.W.Brown&Moskowitz,1998;Gable&Reis,1999),noempiricaldatahaveyetbeenpublishedontheactualratesofinteractions.Aninternalanalysisofarecentlycompleteddatasetwhere15peopleworetheEARcontinuouslyfor10daysrevealednosignificantdifferencesbetweentheweekdayaverageoranymidweek2-dayperiodandtheweekendfor“timespentalone”and“timespentinconversation.”Therewas,however,atrendforrelativelymoredyadicandfewergroupinteractionsovertheweekend(Mehl&Pennebaker,inpress).Thus,therehasbeenlittleevidencetodatethatpeoplechangetheiroverallfrequencyofsocialinteractionsovertheweekend.Still,theinfluenceofsamplingdecisionsonthetimeestimatescannotberuledout.People’sweekendsocialinter-actionsclearlyareaninterestingtopicforfutureEARaswellasESresearch.
Overandaboveinterpretationalambiguitiesandpotentialsam-plingeffects,however,itisinterestingthatthesocialinteractionbaseratesdivergeinawaythatmakespeople’severydaysociallivesappearmoresocialfromtheagent’sthantheobserver’spointofview.Thequestionthenariseswhetherthisisarandomeffectorduetocharacteristicsinherenttosocialinteractions.Socialinteractionsarethelandmarksofsociallifeandstandoutaspunctuatedfiguresonaratherhomogeneousgroundofdailyactivities.Craik(2000)reportedaninterestingphenomenoninhislived-dayanalyses:Judgesexperienceastrongimpulsetopressthefast-forwardbuttonwhenwatchingnonsocialperiodsofthere-cordingsofaperson’sday.Thisanchoringofpeople’sperceptionsofeverydaylifearoundtheirsocialinteractionsthencouldpoten-tiallymakesocialepisodesthetargetofwell-knownbiasesinhumaninformationprocessing(Schwarz&Sudman,1993;Stoneetal.,2000;Tourangeauetal.,2000).Althoughobtrusiveaspects,suchaswithwhomandaboutwhatonehadaconversation,probablyposenoproblemforaccuraterecall,moresubtleaspects,suchasthedurationofdailyconversations,runtheriskofbeingmisrepresentedand—becauseoftheirhighsalienceandself-relevance—mostlikelyoverestimated.
ThisunderscoresourbasicconceptualargumentfortheEAR:Itisnotagenerallymoreobjective,inthesenseofaccurate,event-samplingtool.Rather,itisaresearchtoolthatallowsustoobtaininsightintopeople’ssociallivesfromaunique(i.e.,theunobtrusiveobserver’s)pointofview.Thisperspectivecanyieldinformationsimilartoorquitedifferentfromtheperson’sinsiderperspective,dependingontheeventorbehaviorunderinvestigation.Divergenceislikelyforeventsoraspectsofeventsthatdonotnaturallycaptureaperson’sattention(as,e.g.,insubtle,low-frequency,orcounterhabitualbehavior).AstudycollectingbothEARandESdataonthesameparticipant
THEPSYCHOMETRICSOFSOCIALLIFE
867
couldtestthisideasystematicallyandwouldbeextremelyvaluablefordevelopingabetterunderstandingofthepsycho-logicalprocessesinvolvedinself-andotherassessment(Funder&Colvin,1997).
Individualdifferencesinstudents’sociallives.Thedegreeofindividualvariabilityinstudents’everydaysocialliveswasstrik-ing.The52participantsdifferedtremendouslyintheirpreferencesforsocialsettings,activities,andinteractions.Someparticipantsspoketovirtuallynooneduringtheentire4daysofmonitoring;otherswereengagedinconversationsthemajorityoftheirwakinghours.Thewayspeopleusedlanguageweresimilarlyvariable.Somesworequiteabit,manynotatall.Infact,1participantusedprofanitymoreoften(3.4%)thananotherpersonmadeself-references(3.2%).Thesameappliedforemotionalwords,cogni-tivewords,fillerwordsandmostoftheotherlinguisticmarkers.Obviously,someofthebaseratesforpeoples’socialenvironmentandlanguageuse,suchasthetimespentathomeortheuseofswearwords,wouldvarywiththetargetpopulation.Futurere-searchneedstoaddressthisissuebyidentifyingmoderatorssuchasaperson’ssocioeconomicstatus,age,andculturalorsubculturalbackground.Themostimmediatenextstep,however,istolendpsychologicalmeaningtotheindividualdifferencesinpeople’ssociallives.Inafollow-upproject,wearecurrentlylookingforpersonalitycorrelatesofpeople’ssocialenvironmentselectionandnaturalwordchoice.Doextravertsgooutmorefrequentlythanintroverts?Dodepressedpeoplesociallywithdrawbystayingathome?Doneuroticswhineandcomplainintheirdailyconversa-tions?Donarcissistsshowalinguisticself-focus?Thiswillulti-matelyenrichpersonalitypsychologybyprovidingcontextualizedpersonalitycorrelatesderivedfromnaturalisticobservationdata(Funder,2001;Hogan,1998).
Asanaside,itisinterestingthatinthisstudygenderaccountedforsolittlevarianceinstudents’dailysocialenvironments.Al-thoughwomenspentmoretimeonthephoneandmenwereonthecomputermoreoften,allotheraspectsofdailylocations,activities,andinteractionsyieldedcomparableestimates.Femaleandmaleparticipants,however,differedsubstantiallyintheireverydaylan-guageuse.Overall,thedifferencesconvergewithfindingsforwrittenlanguageusefromthisandotherlabs(Lakoff,1975;Mulac,Bradac,&Gibbons,2001;Pennebaker&King,1999)—withtwointerestingexceptions.Thisstudydidnotfindsupportforthenotionthatwomeningeneralrefertoemotionsmoreoften(forareview,seeMulacetal.,2001).Althoughwomenusedmorepositiveemotionwordsintheireverydaylanguage,menhadahigherprevalenceforangryutterances.Also,contrarytothewide-spreadideathatwomen’slanguageismoresociallyengaging(Maccoby,1990;Tannen,1990),andalsocontrarytofindingsconcerningwrittennarrativesfromourlab(Pennebaker&King,1999),femaleparticipantsdidnotmakemorereferencestosocialprocesses.
Clearly,researchonpeople’severydayspontaneouslanguageuseisinitsinfancy.Futureresearchwillhavetoadoptamoredifferentiatedperspectiveandsystematicallydistinguishbetweensame-sexandopposite-sexinteractions(e.g.,Thomson,Murach-ver,&Green,2001)aswellasaddressbasicpragmaticandfunctionaldifferencesbetweenwrittenandspokenlanguageuse(e.g.,Biber,1988;Pennebakeretal.,2003).
StabilityofStudents’SocialLives
Temporalstability.AnalysesofthevariousSECSIcategoriesrevealedadegreeofstabilityinstudents’dailysocialenviron-mentsoveraperiodof4weeksthatiscomparabletothestabilityfoundinself-reportandobservationalresearch(Diener&Larson,1984;Epstein,1979;Fleeson,2001;Moskowitz,1982;Nezlek,1993;Reisetal.,1993).Theratherlowstabilityforlectureattendancewassurprisingconsideringthematchingofweekdaysforbothmonitoringperiodsandthehighintercoderreliability.Overall,however,thedataclearlyrevealedthatpeople’severydaylivesarenotonlycoherentfromtheagent’sperspectivebutalsoshowahighdegreeofconsistencyfromanoutsider’sperspective.Theaveragestabilitiesforthelinguisticcategorieswerelowerthanthoseforthesocialenvironmentcategories.ConsideringthattheEARsampledlanguageacrossawidespectrumofconversa-tions,withsomebeingveryprivateandsomenotmorethanacasual“Howareyou?”theirmagnitudewassurprising.ModeratetohighcorrelationswerefoundbetweenTime1andTime2fortheuseofarticles,prepositions,first-personpronouns,andpresent-tenseverbs.Theuseofpositiveandnegativeemotionwordswasequallyreliable.Thevarietyoftopicsofpeople’sdailyconversa-tionssuggeststhatthisstabilityreflectsastabilityoflinguisticstylemorethanlinguisticcontent.Thefactthatthestabilitycoef-ficientsfortheuseofswearwords,nonfluencies,andfillerwordsexceededtheonesforanyotherLIWCcategoryshowsthatspokenlanguageprovidesvaluablestylisticinformationoverandbeyondwrittenlanguage(Pennebaker&King,1999).
Whataretheimplicationsofthefactthatpeople’ssociallivesarestablefromanobserver’spointofview?Itiseasytoseethatapparentlyunrelatedbehaviorssuchasbeingatacoffeeshop(reading)andtypingonthecomputer(writingapaper)canbecoherentlyinterpretedas“studying”onasubjectivelevel.Thisstudy,however,showsthatoverandbeyondthis“higherorder”experientialconsistency,thereisrelativestabilityinpeople’sso-cialeventssuchaswhere,withwhat,andwithwhompeoplespendtheirdays.Demonstratingthisobjectivestabilitymeansestablish-ingreliabilityforoneofthecorecriteriaofnaturalisticpersonalityresearch:people’severydaybehaviors.Real-lifebehaviorscanonlybepredictedfrompeople’sdispositionsifthesereal-lifebehaviorsfollowsoundpsychometricproperties—notonlyintheagent’seyesbutalsointheeye(orEAR)ofaneutralobserver.Demonstratingtemporalstabilityofpeople’severydaysociallivesfromtheobserver’spointofviewisalsoconceptuallyim-portantforresearchonimpressionformation.Naturally,peopleformimpressionsbyseeingotherpeopleactintheirnaturalenvi-ronments.Self-reportedconsistencyhasbeenshowntoreliablyfacilitateself–otheragreementinpersonalityassessment(Funder&Colvin,1997).Onceconsistencyinpeople’sobjectiveeverydaybehaviorsandsocialenvironmentshasbeenestablished,onecanstarttodeterminewhichcueslaypeopleuseinformingimpres-sionsaboutothers.People’severydaysocialenvironmentsandtheirlanguageuseinspontaneousconversationscertainlyaregoodcandidatesforbeingsocialcuesthatpeoplenaturallydrawonwhenfiguringoutwhoapersonis(e.g.,Gifford&Hine,1994;Gosling,Ko,Mannarelli,&Morris,2002).Relatedtothisidea,itwouldbeinterestingforfutureresearchtoidentifywhatkindof
868
MEHLANDPENNEBAKER
sociallifeandnaturallanguageusemakepeoplebeseenaslikeablebyothers.7Between-speakerstability.Transcribingutterancesbyboththeparticipantsandtheirinteractionpartnersmadeitpossibletotestbetween-speakerstabilityinlanguageonabroadlevel.Commu-nicationaccommodationtheory(Gilesetal.,1991)identifiescon-vergenceanddivergenceasstrategiesthathumansadopttoman-ageconversations.Ouranalysesrevealedvaryingdegreesofoverlapinwordchoicebetweenspeakers.Swearwords,nonflu-encies,andfiller-wordcategoriesshowedahigherdegreeofcon-vergencethanmostothervariables.First-personsingularandsecond-personpronounsfailedtoshowconvergence,mostlikelyreflectingthattheyaretypicallyusedcomplementarilyindyadicconversations.Thetotalwordcountsforbothsetsoftranscriptswerehighlycorrelated(.72).OurfindingsconfirmonacumulativelevelwhatNiederhofferandPennebaker(2002)foundonaturn-by-turnlevel:Naturalconversationsarecharacterizedbyasub-stantialleveloflinguisticsynchrony.Unfortunately,thisstudydidnotallowustoidentifysourcesofthissynchrony.Dopeopleprefertointeractwithlinguisticallysimilarothers?Dopeoplesubtlyimposetheirlinguisticstyleontotheirinteractionpartners?Orissynchronysimplyareflectionoftheconversationalcontext?Bywhateversourcesynchronymayfinallybedriven,itcertainlyfosterstemporalstabilityinpeople’slinguisticstyles.
Cross-contextstability.Finally,theEARdataallowedustolookathowparticipants’languagevariedacrosssocialcontexts.Clearly,languageuseisconstrainedbysituationalfactors(Forgas,1985;Pennebakeretal.,2003).Ouranalysesofwordchoice,however,alsopaintapictureofrelativestabilityacrosscontext.Thecomparisonsoflanguageuseinpublicplacesversusathome,inamusementversusworkcontexts,andonthephoneversusindirectpersonalinteractionsdidnotyieldsignificantlinguisticdifferenceseffectsoverandbeyondwhatcanbeexpectedbychance.Theseresultsincombinationwiththestabilityoflanguageuseacrosstimeprovidestrongsupportfortheideathatpeopleexpressthemselvesinidiosyncraticwaysusingdistinctivelinguis-ticstyles.Note,however,thattheanalysesareexploratoryinnaturefortworeasons:(a)Theselectionofcontextswasseriouslyconstrainedbywhatcouldbecodedfromambientsounds,and(b)becausenoteverystudentprovidedlanguagesamplesfromallcontexts,somecontrastswerebasedonasmallnumberofobservations.
EthicalandLegalIssuesinConductingEARResearch
Clearly,recordingsnippetsofpeople’ssociallivesraisessomecriticalethicalandlegalquestions.AsmentionedintheMethodsectionhereandinmoredetailelsewhere(Mehletal.,2001),thisstudyimplementedseveralsafeguardstoprotectparticipants’pri-vacyandensuremaximumdataconfidentiality.However,themostseriouslegalandethicalconcernsrevolvenotaroundthepartici-pantsthemselves,butratherpotential“secondary”participants,bystanderswhoarenotdirectlyinvolvedinthestudybutwhosevoicesandbehaviorsarecapturedbytheEAR.AlthoughforfuturestudiesanoptimizedEARsystemisavailablethatmaximizessoundemittedbytheparticipantswhilesuppressinginformationaboutpeoplearoundthem(bymeansoflowsensitivityrecordingwithamodifiedunidirectionalmicrophone),thisstudydidcaptureanduseinformationfromotherpeople.
IntheUnitedStates,thereareveryfewrestrictionsaboutre-cordingpeople’sutterancesinpublicplaces.Thesituationcon-cerningtherecordingofprivateconversationsisfarmoreambig-uous—bothlegallyandethically.Inmoststates,includingTexas,wherethecurrentstudywasconducted,recordingscanbemadelegallyifatleastoneoftheinteractantshasknowledgeoftherecordingdevice.Asmallnumberofstates(e.g.,Maryland)onlyallowrecordingsifallinteractantshaveknowledgeoftherecord-ing.Eveninthemostlegallyrestrictivestates,however,unautho-rizedrecordingsareonlyaproblemiftheparticipantsareidentifiable.
Inthepresentstudy,participantsdidweartheEARintheirownoratfriends’houses,andotherpeoplewererecordedinprivatesettings.However,participantswereaskedtoincorporatetheEARintotheirdailylivesasnaturallyaspossible.ThismeanstheywereencouragedtowearthemicrophonevisiblyandopenlytalkabouttheEARanddiscussitwithothersintheirconversations.Anec-dotalinformationfromthedebriefingsessionsrevealedthatpar-ticipantsregularlyinformedtheirfriendsandpartnersabouttheEAR.
Irrespectiveofnotification,anonymityofotherpeople’sutter-ancesisofparamountimportance,becausetheirspeechsamplesarecollectedwithoutinformedconsent.Inthisstudy,anonymityofthedatawasmaximizedbylimitingrecordingsto30-ssnippets.Innocasedidthesoundsamplesrevealthefull,thatis,firstandlast,nameofaperson.Also,the30-sintervalsgenerallycapturedthelastwordsofanutterance,oneortwomorecompletesentences,andthebeginningofanotherutterance.Substantialcontextinfor-mationisnecessarytoinferthemeaningofasentencewithinaconversation.Itisthushighlyunlikelythatthisstudyviolatedprivacyrightsofapersonwhowasinadvertentlyrecorded.
Asfurtherprotections,theaudiotapesweretranscribedeitherbyprofessionaltranscribersorresearchstaffwhodidnothaveaccesstotheidentitiesoftheresearchparticipants.Inaddition,theyweretrainedtoomitanypotentiallyidentifyinginformationfromthetranscriptsthemselves.Recallalsothatparticipantsweregiventheoptionoflisteningtotheirtapesanderasinganyportionsbeforetheyturnedthemovertotheexperimenters.
Despiteallofthestepsthatweretakentoensuretheconfiden-tialityofparticipantsandanonymityofnonparticipants,EARresearchersmustbeawareoftheethicalimplicationsofrecordingothers.Throughcarefulinstructionsandinformedconsentproce-dureswithparticipants,briefrecordingperiods,detailedtrainingoftranscribers,andtheremovalofallidentifyinginformation,thepresentprocedureadherestobothethicalandlegalstandardsforresearch.
PotentialsforFutureEARResearch
TheEARhasgreatpotentialforresearchersinsocialandpersonalitypsychology.Currently,itistheonlyavailablemeth-odologythatallowsthesamplingofbehavioraldatafromanunobtrusiveobserver’spointofview.AlthoughmanyoftheEARdataoverlapwithwhatisobtainedfromretrospectiveormomen-taryself-reports,theEAR’suniquestrengthliesinthetrackingofsubtleacousticallydetectableevents.Verbalandparaverbalbe-7Wethankananonymousreviewerforbringingupthiscreativeidea.
THEPSYCHOMETRICSOFSOCIALLIFE
869
haviorinspontaneousconversationsclearlyemergesasthemostimmediatearenafortheEAR.However,avarietyofotherreal-lifephenomenathatusuallyfalloutsideaperson’sawarenessarealsoinherentlyimportanttopersonalityandsocialpsychologists:Sub-tlesignsofemotions,suchaslaughter,sighingorwhistling;unobtrusivehealthindicators,suchascoughing,sneezingorsniff-ing;orevenambientsoundlevelsinaperson’senvironmentcanbedetected,quantifiedandanalyzed.Also,theEARtechnologyposesonlypragmaticconstraintsonhowmanydatapointscanbeob-tainedfromaperson.Doublingoreventriplingthesamplingrate,forexample,tomoreadequatelycapturelow-frequencybehavior,constitutesnoadditionalburdenonbehalfoftheparticipant(eth-icallythough,itprobablyrequiresproportionallyshorterrecordingintervals).Finally,asanarchivalrecord,theEAR—unlikemanymethodologiesinthesocialsciences—producesrichrecordsofpeople’sdailylivesthatarereadilyavailableforotherresearchersandquestionsthatgobeyondtheinvestigators’primaryinterests.ItisconceivablethatgenerationsoffuturescientistscouldhaveaninterestinanalyzingtheEARrecordsusingtheir—maybethenverydifferent—interpretivelenses.
EARresearchishigh-investmentresearch.Thedatacollectionandpreparationprocessistime-consumingandlaborintensive.TheEAR’spotentialasatoolforunobtrusivenaturalisticobser-vationmakesitaneededcomplementtopsychologists’long-standingrelianceonself-reports.Itprovidesauniquewindowintoawidevarietyofreal-lifesocialphenomenathatultimatelycon-stitutetheendpointofourresearchendeavors.Initsunobtrusive-nessanditsoutsiderviewpoint,theEARcallsforthestudyofthosesocialphenomenathathavetoolongexistedintheshadowofthereportable.
References
Allport,G.W.(1937).Personality:Apsychologicalinterpretation.NewYork:Holt.
Barker,R.G.,&Wright,H.F.(1951).Oneboy’sday.Aspecimenrecordofbehavior.NewYork:Harper&Row.
Biber,D.(1988).Variationacrossspeechandwriting.Cambridge,En-gland:CambridgeUniversityPress.
Bischoping,K.(1993).Genderdifferencesinconversationaltopics,1922–1990.SexRoles,28,1–17.
Bosson,J.K.,Swann,W.B.,Jr.,&Pennebaker,J.W.(2000).Stalkingtheperfectmeasureofimplicitself-esteem:Theblindmenandtheelephantrevisited?JournalofPersonalityandSocialPsychology,79,631–643.Brown,K.W.,&Moskowitz,D.S.(1998).Dynamicstabilityofbehavior:Therhythmsofourinterpersonallives.JournalofPersonality,66,105–134.
Brown,P.,&Fraser,C.(1979).Speechasamarkerofsituation.InK.R.Scherer&H.Giles(Eds.),Socialmarkersinspeech(pp.33–62).Cambridge,England:CambridgeUniversityPress.
Brown,R.,&Gilman,A.(1960).Thepronounsofpowerandsolidarity.InT.A.Sebeok(Ed.),Styleinlanguage(pp.253–276).Cambridge,MA:MITPress.
Buss,D.M.(1987).Selection,evocation,andmanipulation.JournalofPersonalityandSocialPsychology,53,1214–1221.
Cameron,P.(1969).Frequencyandkindofwordsinvarioussocialsettings,orwhatthehell’sgoingon?PacificSociologicalReview,12,101–104.
Craik,K.H.(2000).Theliveddayofanindividual:Aperson–environmentperspective.InW.B.Walsh,K.H.Craik,&R.H.Price(Eds.),Person–environmentpsychology:Newdirectionsandperspectives(pp.233–266).Mahwah,NJ:Erlbaum.
Csikszentmihalyi,M.,&Larson,R.(1984).Beingadolescent:Conflictandgrowthintheteenageyears.NewYork:BasicBooks.
Csikszentmihalyi,M.,&Larson,R.(1987).Validityandreliabilityoftheexperience-samplingmethod.TheJournalofNervousandMentalDis-ease,175,526–536.
Diener,E.,&Larson,R.J.(1984).Temporalstabilityandcross-situationalconsistencyofaffective,behavioral,andcognitiveresponses.JournalofPersonalityandSocialPsychology,47,871–883.
Dunbar,R.I.M.,Duncan,N.D.C.,&Marriott,A.(1997).Humanconversationalbehavior.HumanNature,8,231–246.
Emler,N.(2001).Gossiping.InW.P.Robinson&H.Giles(Eds.),Thenewhandbookoflanguageandsocialpsychology(pp.317–338).NewYork:Wiley.
Epstein,S.(1979).Thestabilityofbehavior:I.Onpredictingmostofthepeoplemuchofthetime.JournalofPersonalityandSocialPsychol-ogy,37,1097–1126.
Fleeson,W.(2001).Towardastructure-andprocess-integratedviewofpersonalitytraitsasdensitydistributionofstates.JournalofPersonalityandSocialPsychology,80,1011–1027.
Forgas,J.(Ed.).(1985).Languageandsocialsituations.NewYork:Springer-Verlag.
Funder,D.C.(2001).Personality.AnnualReviewofPsychology,52,197–221.
Funder,D.C.,&Colvin,C.R.(1997).Congruenceofother’sandself-judgmentsofpersonality.InR.Hogan,J.Johnson,&S.Briggs(Eds.),Handbookofpersonalitypsychology(pp.617–656).NewYork:Aca-demicPress.
Funder,D.C.,&Sneed,C.D.(1993).Behavioralmanifestationsofpersonality:Anecologicalapproachtojudgmentalaccuracy.JournalofPersonalityandSocialPsychology,64,479–490.
Gable,S.L.,&Reis,H.T.(1999).Nowandthen,themandus,thisandthat:Studyingrelationshipsacrosstime,partner,context,andperson.PersonalRelationships,6,415–432.
Gifford,R.,&Hine,D.(1994).Theroleofverbalbehaviorintheencodinganddecodingofinterpersonaldispositions.JournalofResearchinPersonality,28,115–132.
Giles,H.G.,Coupland,N.,&Coupland,J.(1991).Accommodationtheory:Communication,context,andconsequence.InH.Giles,J.Cou-pland,&N.Coupland(Eds.),Contextsofaccommodation:Develop-mentsinappliedsociolinguistics(pp.1–68).Cambridge,England:Cam-bridgeUniversityPress.
Gleser,G.C.,Gottschalk,L.A.,&John,W.(1959).Therelationshipofsexandintelligencetochoiceofwords:Anormativestudyofverbalbehav-ior.JournalofClinicalPsychology,15,183–191.
Goffman,E.(1981).Formsoftalk.Philadelphia:UniversityofPennsyl-vaniaPress.
Gosling,S.D.,John,O.P.,Craik,K.H.,&Robins,R.W.(1998).Dopeopleknowhowtheybehave?Self-reportedactfrequenciescomparedwithon-linecodingsbyobservers.JournalofPersonalityandSocialPsychology,74,1337–1349.
Gosling,S.D.,Ko,S.J.,Mannarelli,T.,&Morris,M.E.(2002).Aroomwithacue:Personalityjudgmentsbasedonofficesandbedrooms.JournalofPersonalityandSocialPsychology,82,379–398.
Hogan,R.(1982).Asocioanalytictheoryofpersonality.NebraskaSym-posiumofMotivation,30,55–89.
Hogan,R.(1998).Reinventingpersonality.JournalofSocialandClinicalPsychology,17,1–10.
Hormuth,S.E.(1986).Thesamplingofexperiencesinsitu.JournalofPersonality,54,262–293.
Ickes,W.,Reidhead,S.,&Patterson,M.(1986).Machiavellianismandself-monitoring:Asdifferentas“me”and“you.”SocialCognition,4,58–74.
IckesW.,Snyder,M.,&Garcia,S.(1997).Personalityinfluencesonthechoiceofsituations.InR.Hogan,J.Johnson,&S.Briggs(Eds.),
870
MEHLANDPENNEBAKER
Handbookofpersonalitypsychology(pp.165–195).NewYork:Aca-demicPress.
Lakoff,R.T.(1975).Languageandwoman’splace.NewYork:Harper&Row.
Landis,C.(1927).Nationaldifferencesinconversation.JournalofAbnor-malandSocialPsychology,21,354–357.
Larson,R.(1990).Thesolitarysideoflife:Anexplanationofthetimepeoplespendalonefromchildhoodtooldage.DevelopmentalRe-view,10,155–183.
Maccoby,E.E.(1990).Genderandrelationships:Adevelopmentalac-count.AmericanPsychologist,45,513–520.
Mehl,M.R.,&Pennebaker,J.W.(inpress).Thesocialdynamicsofaculturalupheaval:SocialinteractionssurroundingSeptember11,2001.PsychologicalScience.
Mehl,M.R.,Pennebaker,J.W.,Crow,M.,Dabbs,J.,&Price,J.(2001).TheElectronicallyActivatedRecorder(EAR):Adeviceforsamplingnaturalisticdailyactivitiesandconversations.BehaviorResearchMeth-ods,Instruments,andComputers,33,517–523.
Moore,H.T.(1922).Furtherdataconcerningsexdifferences.JournalofAbnormalandSocialPsychology,17,210–214.
Morand,D.(2000).Languageandpower:Anempiricalanalysisoflin-guisticstrategiesusedinsuperior–subordinatecommunication.JournalofOrganizationalBehavior,21,235–248.
Moskowitz,D.S.(1982).Coherenceandcross-situationalgeneralityinpersonality:Anewanalysisofoldproblems.JournalofPersonalityandSocialPsychology,43,754–768.
Moskowitz,D.S.(1994).Cross-situationalgeneralityandtheinterpersonalcircumplex.JournalofPersonalityandSocialPsychology,66,921–933.Mulac,A.,Bradac,J.J.,&Gibbons,P.(2001).Empiricalsupportforthegender-as-culturehypothesis:Aninterculturalanalysisofmale/femalelanguagedifferences.HumanCommunicationResearch,27,121–152.Nezlek,J.B.(1993).Thestabilityofsocialinteraction.JournalofPer-sonalityandSocialPsychology,65,930–941.
Niederhoffer,K.G.,&Pennebaker,J.W.(2002).Linguisticstylematchinginsocialinteraction.JournalofLanguageandSocialPsychology,21,337–360.
Pennebaker,J.W.,Francis,M.E.,&Booth,R.J.(2001).Linguisticinquiryandwordcount(LIWC)[Computersoftware].Mahwah,NJ:Erlbaum.Pennebaker,J.W.,&Graybeal,A.(2001).Patternsofnaturallanguageuse:Disclosure,personality,andsocialintegration.CurrentDirectionsinPsychologicalScience,10,90–93.
Pennebaker,J.W.,&King,L.A.(1999).Linguisticstyles:Languageuseasanindividualdifference.JournalofPersonalityandSocialPsychol-ogy,77,1296–1312.
Pennebaker,J.W.,Mehl,M.R.,&Niederhoffer,K.G.(2003).Psycho-
logicalaspectsoflanguageuse:Ourwords,ourselves.AnnualReviewofPsychology,54,547–577.
Reis,H.T.,Lin,Y.-C.,Bennett,M.E.,&Nezlek,J.B.(1993).Changeandconsistencyinsocialparticipationduringearlyadulthood.Developmen-talPsychology,29,633–645.
Reis,H.T.,&Wheeler,L.(1991).StudyingsocialinteractionwiththeRochesterInteractionRecord.AdvancesinExperimentalSocialPsychol-ogy,24,269–318.
Robinson,J.P.(1977).HowAmericansusetime:Asocial–psychologicalanalysisofeverydaybehavior.NewYork:Praeger.
Robinson,J.P.(1985).Thevalidityandreliabilityofdiariesversusalternativetimeusemeasures.InF.T.Juster&F.P.Stafford(Eds.),Time,goods,andwell-being(pp.63–92).AnnArbor,MI:InstituteforSocialResearch.
Rozin,P.(2001).Socialpsychologyandscience:SomelessonsfromSolomonAsch.PersonalityandSocialPsychologyReview,5,2–14.Sanford,F.H.(1942).Speechandpersonality.PsychologicalBulletin,39,811–845.
Schnurr,P.P.,Rosenberg,S.D.,Oxman,T.E.,&Tucker,G.J.(1986).Amethodologicalnoteoncontentanalysis:Estimatesofreliability.Jour-nalofPersonalityAssessment,50,601–609.
Schwarz,N.,&Sudman,S.(Eds)(1993).Autobiographicalmemoryandthevalidityofretrospectiveself-reports.NewYork:Springer-Verlag.Stone,A.,&Shiffman,S.(1994).Ecologicalmomentaryassessment(EMA)inbehavioralmedicine.AnnalsofBehavioralMedicine,16,199–202.
Stone,A.A.,Turkkan,J.S.,Bachrach,C.A.,Jobe,J.B.,Kurtzman,H.S.,&Cain,V.S.(Eds.).(2000).Thescienceofself-report:Implicationsforresearchandpractice.Mahwah,NJ:Erlbaum.
Swann,W.B.(1987).Identitynegotiation:Wheretworoadsmeet.JournalofPersonalityandSocialPsychology,53,1038–1051.
Szalai,A.,Converse,P.E.,Feldheim,P.,Scheuch,E.K.,&Stone,J.P.(1972).Theuseoftime:Dailyactivitiesofurbanandsuburbanpopu-lationintwelvecountries.TheHague,theNetherlands:Mouton.
Tannen,D.(1990).Youjustdon’tunderstand:Womenandmenincon-versation.NewYork:Morrow.
Thomson,R.,Murachver,T.,&Green,J.(2001).Whereisthegenderingenderedlanguage?PsychologicalScience,121,171–175.
Tourangeau,R.,Rips,L.J.,&Rasinski,K.(2000).Thepsychologyofsurveyresponse.Cambridge,England:CambridgeUniversityPress.Weintraub,W.(1989).Verbalbehaviorineverydaylife.NewYork:Springer-Verlag.
ReceivedDecember21,2001RevisionreceivedAugust9,2002
AcceptedAugust22,2002Ⅲ
因篇幅问题不能全部显示，请点此查看更多更全内容