Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

Making Thinking Visible with Atlas.

ti: Computer Assisted Qualitative Analysis as Textual


Practices
Author(s): Zdeněk Konopásek
Source: Historical Social Research / Historische Sozialforschung. Supplement, No. 19,
Grounded Theory Reader (2007), pp. 276-298
Published by: GESIS - Leibniz-Institute for the Social Sciences, Center for Historical Social Research
Stable URL: http://www.jstor.org/stable/40981082 .
Accessed: 03/09/2013 14:43

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

GESIS - Leibniz-Institute for the Social Sciences, Center for Historical Social Research is collaborating with
JSTOR to digitize, preserve and extend access to Historical Social Research / Historische Sozialforschung.
Supplement.

http://www.jstor.org

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
- 2007 - No. 19,276-298
HistoricalSocial Research,Supplement

MakingThinking VisiblewithAtlas.ti:
AssistedQualitativeAnalysisas Textual
Computer
Practices

*
ZdenëkKonopásek
Abstract:How is a newqualityof reading, whichwe call
"sociologicalunderstanding", createdduringtheprocèsof
qualitative A
analysis? methodological (conventional) an-
swertothisquestion usuallyspeaksofmental processesand
conceptual work.Thispapersuggests a different view- so-
ciologicalrather thanmethodological; or morepreciselya
view inspiredby a contemporary sociologyof science.It
describes qualitative analysisas a setof material practices.
Takinggrounded theory methodology andtheworkwiththe
computer programme Atlas.tias an example,it is argued
thatthinking is inseparablefromdoingeveninthisdomain.
It is arguedthatby adopting thesuggested perspectivewe
might be better able to speakof otherwise hardlygraspable
processesof qualitative analysisin moreaccountable and
instructable ways. Further,software packages would be bet-
terunderstood notonlyas "meretools"forcodingandre-
trieving,butalso as complexvirtualenvironments forem-
bodied and practice-based knowledgemaking.Finally,
grounded theory methodology mightappearin a somewhat
differentlight:whendescribed notinterms ofmethodologi-
cal or theoretical concepts but ratherin terms of whatwe
practically do with the analyseddata, it becomes perfectly
compatible withtheradicalconstructivist, oreven
textualist,

Addressall communications to: Zdenek Konopásek,CenterforTheoreticalStudy(The


InstituteforAdvancedStudiesat CharlesUniversity and theAcademyof Science of the
Czech
Republic),Jilska1, 110 00 Praha1,Czech Republic;e-mail:zdenek@konopasek.net.
Pieces of thiscontributionhave alreadyappearedin Czech as partsof myreviewsof Al-
tas.ti(KONOPÁSEK 1998,2005a) and in a conference paper(KONOPÁSEK 2005b). This
paperhas beenwritten withintheworkon theframework researchprogramme"Theoretical
Research on Complex Phenomenain Physics,Biology and Social Sciences", MSM
0021620845.

276

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
post-structuralist
paradigms (fromwhichit
ofinterpretation
hasallegedly
departed a
by longway).

1. Introduction
Somecontemporaries oftheprevious versionofAtlas.ti(version4), a software
tool forqualitative data analysis,mayremember thequirkin it. As a new-
comer,whileplayingwithoptionsand menusof theprogramme, you could
havebecometempted to trya verypromising optionofferedin themenufor
workwithtextualdocuments: Relevanttextsearch.Hereit is, you thought,
definitelythekeyfunction in computerised qualitative letus clickon
analysis,
it! Afterchoosingit,however, a smallinfowindowpoppedup withan ironic
replyto yourcommand: "Do youbelievein magic?"And,ifyouwerehappy
enoughto haveyourPC equippedwitha soundcardyou couldalso heara
significative
hawking, indicatingthatyouhadjustdonesomething reallyfool-
ish.
Softwarepackagessuchas Atlas.tisimplycannotdo mental workforyou.It
is alwaysyou,as theanalyst, whohas to do therealanalysis.Becauseonly
humanresearchers can think.The software onlyprovidesmoreor less useful
assistanceand supportto the thinking subject.1It extendstheresearcher's
mentalcapabilities to organise,to remember, andto be systematic.Butwhile
doingso itessentiallyremains a stupidinstrument, whichcannotdo things such
as determining therelevance of a textpassage.Humans,notmachines, do the
crucialworkof codingand retrieving - i.e., decidewhatpassagesof data
shouldbe marked bywhatterms tobe searched andbrowsedlateron.Thehope
thattheprogramme woulddo moreandbe ableto replacetheanalytic mindis
foolish.Onlyhumanresearchers canmakesenseandanalytic use ofotherwise
meaningless operationsof thecomputer - suchwas theunforgettable lesson
givenby thislittle
nastyquirk,incorporated into thedesignofthe programme.

1 Hence thenotionof
CAQDAS, computerassistedqualitativedata analysissoftware, used
forthisfamilyof qualitativecomputing. It shouldbe noted,however,thatthereare other
programmes usefulforqualitativeanalysts,butconstructed quitedifferently,namelyon the
principlesof co-ocurrence analysis.These programmes are explicitlyintendedforgenera-
tionand attribution of meaningon thebasis of computerised analysis(withpracticallyno
directintervention of a cleverhumanmind)of co-wordnetworksin huge bodies of data
(TEIL & LATOUR 1995).Attempts at "intelligent"
computer processing ofqualitative
data
are exploredeven withinthefamilyof classicalCAQDAS tools. Softwarecalled Qualrus
(<http://www.qualrus.com>) recentlyintroduced the conceptof "intelligent coding".The
programme attemptsto proposesuitablecodes forselectedquotationson thebasis of an
analysis- runningas a background processon the computer- of all codingoperations
madeso far(assumingthatquotationscontaining similarwordswouldbe coded similarly)
. . . Butletus leavetheseinterestingdevelopments aside fornow.

277

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
Thiswas an important and muchneededlesson,of course,whichwas de-
to a
signed prevent typicalmisunderstanding aboutCAQDAS. Yet,I am con-
vincedthattheargument was (andstillis) somewhat misleading. Indeed,inthis
paper I would like to that a
suggest makingCAQDAS substantively irrelevant
and purelyinstrumental technicalextension and support of mentalprocesses
was a disservice, something of a pokein theeye of all qualitative research. I
arguethattheentire ideathatsoftware essentially representswhatoccursinthe
analyst'shead strengthened a classical"methodological" view of qualitative
analysis,emphasizing theroleof a researcher who is superior to his or her
research subjectsbyvirtueof specialqualitiesof his or herthinking. Accord-
ingly,thisway of thinking suppressed a non-exclusive, say "ethnomethod-
ological"position,whichhighlights taken-for-granted materialpracticesand
instructability ofknowledge production.
Sucha mentalistic approach, eitherimplicit orexplicit,
hashadtwounhappy
consequences. First,CAQDAS has developedproblematic relationships with
thosetheoretical-methodological in
positions qualitative research, increasingly
influential amongmembers of thecommunity thatdepartedfromobjectivist
methodology. Computer assistedqualitative dataanalysisis seenas noteasily
compatible withradicalconstructivism or post-structuralistunderstandings of
language.Somescholarshaveevenarguedthatunderthedisguiseoftheinno-
vationcalled qualitativecomputing, a conservative ("modernist") approach
reaffirmed itsposition (COFFEY, HOLBROOK & ATKINSON.1996).
Second,a uniqueopportunity forbetter understanding qualitative analysisas
a set of mediations and embodiedpracticeshas been missed.This is really
unfortunate, sincesuchan understanding is pricelessforourabilityto defend,
explain and teachqualitative research. In textson qualitative research,thereis
usually an abundance of descriptions of various paradigms, approachesand
theoretical frameworks; orofdatacollection procedures, fieldwork practices or
research ethics.Butwhenitcomestopractices bymeansofwhicha newqual-
ityofreading(whichwe call sociological understanding) emerges, descriptions
oftenbecomesomewhat vagueandpoor.2Analysisandinterpretation ofquali-
tativedataareoftenseenas performances of "purereason"to suchan extent
thatit is verydifficult to providea clearand practice-oriented accountof it.
Thereseemto be no intermediaries here,justthelucidmindoftheresearcher
contemplating thedata.And it is themindthatis responsible fordeduction,
induction, generalization, conceptualization, -
comparison basicallymental
as
operations ...
Such accountsdo revealimportant thingsaboutqualitative analysis.But
theyareof a limited help.It is especiallytruewhenone has to explainto an
outsider orto a student in whattermsqualitative analysisconsistsof anything
2 This
tendencythatwritingson qualitativeresearch"are long on theirdiscussionsof data
collectionand researchexperiencesand shorton analysis" has been noted also by
STRAUSS (1987, p.xi).

278

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
morethana careful readingofdata,spicedbyprovidential insightsandobser-
vations(ifthereeverareany).As a consequence, itis claimedthatqualitative
research is in factan art,hardlygraspableand transferable (DENZIN 1994,
p.512 and others).It is emphasised, in responseto inquiring questionsabout
"howitis done",thatthereis no singlequalitative method andthatanalysisof
datacan hardlybe separated fromotherresearch-related activities(whichcan
subsequently be describedat length).Qualitative researchis presented as a
complexand context-dependent activity thatresistsa cook-bookstyleof in-
structions.
Similarresponses aresurelynotwrong.Notperse. Buttheyavoidthemain
point.Evenworse:byavoidingthepointtheymakeitevenmoreurgent - how
qualitative analysisactuallygenerates a newknowledge, in a distinctive and
recognizable way?Conventionally, as we haveseen,peoplearetoldthatit is
notbypressing a button intheinterface ofa computer programme. Thiswould
nothelp,itis believed,becauseeverything important happensin ourminds,in
a waythatis difficult to explain.Mypapertakesa different roadthough. I will
try to talk about material and
practices inter-actions, rather than of mental
operations ofan individual. Theambition herecannotbe toexplainthelogicof
(grounded theory)qualitative analysisbetterand deeperthan,e.g., Anselm
STRAUSS in his marvellous"QualitativeAnalysisfor Social Scientists"
(1987). Rather, I wouldonlyliketotakeSTRAUSS moreseriously inthemo-
mentwhenhe notesthatresearch workconsistsof"setsoftasks,bothphysical
andconceptual" (STRAUSS 1987,p.l; italicsaddedbyZK). Andbecausethe
conceptualusuallyseemsto be overrepresented in qualitative methodology
writings, including theSTRAUSS's book,I willfocushereonthephysical.
"Thinking" willbe bracketed out- notbecauseis is unimportant, ofcourse;
but because its presencecannotaccountfordifferences betweenordinary
knowledge practices(e.g.,of research subjects)andqualitative analysisworth
ofthename(on thesideoftheresearcher). Ofcourse,we analysts do think.No
questionaboutthat.Butso do all theothers, including our research subjects.
Therefore itdoesnotmakemuchsenseto groundthesuperiority ofsociologi-
cal knowledge almostexclusively inourmentalqualitiesandintheveryactof
. . . thinking. Ratherwe shouldfocus,as scienceandtechnology studiesdo,on
practical manipulations withvisible,hearableandpalpablepiecesofreality that
havethepowerof makingthefinalsentencestronger and moredurablethan
anyothercompeting statement (LATOUR 1987,oneforall).
In thenextsectionI am goingto briefly explainthisparticular inspiration
takenfromscienceand technology studies.ThenI will discusstheplace of
grounded theorymethodology (GTM) and Atlas.tiin myoverallargument.
Also,I will clarify in whatsenseit is possibleto keepthefocuson material
practices in thevirtualenvironment of a computer programme. The mainpart
thenfollows:an attempt to describetheanalytical workwithAtlas.tiin terms
ofcreation andoperation ofa "textual laboratory".Themostordinary analytic

279

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
proceduressuchas data segmenting and coding,linkingor memoing will be
presentedas practicalmanipulations withobjectsvisibleon thescreen.Pre-
cisely thesemanipulations endow the knowledgearisingfromqualitative
analysiswith that
qualities make it distinctfromordinarymembers'knowl-
edge.Furthermore,they enableus to of
speak qualitative analysisinan instruc-
table,practicalway. The conclusion will discusssome broadertheoretical
consequencesof such of our about
reframing thinking qualitative analysis.

fromScienceStudies
2. The Inspiration
BrunoLATOUR (1995) in hisarticleon a research expedition to theAmazon
forests givesan illustrative exampleofhowcontemporary sciencestudiesun-
derstand theoperation ofscientific work.Thequestionis howitis possiblethat
scientifictextsspeakof reality;whatconstitutes theirreference to thethings
understudy. Thisquestionoftherelation between thewordandtheworldis an
old one.Buttheperspective of sciencestudiescomesoutwitha novelanswer
toit.As a sociologist ofscience,LATOURavoidstheoretical concepts ofepis-
temology and offersan ethnographic account (accompanied by a set of photo-
graphs)of variouspracticesby whichmembersof theresearchexpedition
"translate" theborder betweensavanaandforest somewhere inAmazonia(i.e.,
thephenomenon understudy)intothetextofa scientific report.He emphasises
thattheempirical evidencehe presents contains no tracesofa mysterious jump
fromtheworldto a word;rather, we can follownumerous smallpractical op-
erations bymeansofwhichreality is moreandmoreloadedwithmeaning and
progressively de-materialised so thatitbecomesincreasingly "textual".There
is no directbridgebetweentheworldandtheword,onlychainsoftranslations
- i.e., practicalmanipulations and interventions by whicha piece of natural
landscape is turnedinto a field laboratory with exact parameters and coordi-
nates;by which lumps of soil become sufficientlyrepresentative samples;and
by which qualitiesof these lumps can be substitutedby written codes and
comments so thatthestudied boundary betweensavannaandforest cansucces-
sivelybe inscribed intosomething else, and therefore inhabit/constitutethe
paperrealmoftexts.3
LikeLATOUR,ormanyhiscolleagues, we couldfollowtheseriesoftrans-
lationsmadebyqualitative researchers on themovefromthefieldtotherealm
oftextual data.Forinstance, something (whichhashappened) is narrated byan
interviewee; the narration is recorded; recording transcribed; tran-
the is the
is
script incorporated to a set of data... eachsuchstepmeansthatsomething is
lostand something is gained.In general,it is materiality whatis lost- e.g.,

3 The role of textsand textual in scientificworkis summarisedby


isations(inscriptions)
LAW (1986).

280

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
material oftheinterview
specificities act,suchas thetotality ofvoicemodula-
and
tion,smells,gestures surrounding environment. What is gained?Meaning
is gained,simplyput.Thisis possiblebecausethegradualloss of materiality
bringsaboutnew possibilities. Once realityis narrated, recorded, and tran-
scribedwe can bettermanipulate it- store,transport, compress, mark,juxta-
pose to another cutintopieces,recompose,
realities, reorder, etc.Onlythanks
to thesemanipulations we can see (and show) differences and similarities,
emerging new
patterns, contexts.
Sincewe proceedin sucha waythatitis alwayspossibleto go back,along
thechainoftransformations (i.e.,from a quotation inourpapertothetranscrip-
tion,totherecording, and- withthehelpoffieldnotes andlabelson tapes- to
thesituation of theinterview or even,to someextent, to the"original" event)
we canspeakofreference. HencetheLATOUR's argument, whichhe so nicely
illustratedbythecase oftheresearch expedition to theAmazonforest: scien-
tifictextsspeakofreality notbecauseofa mysterious bondbetween thingsand
words(something philosophers are so busywith), but rather thanks to well-tied
chainsof smalltransformations, duringwhichsomething is preserved while
otherqualitiesarelost.4
However,I am notso muchinterested herein reference as a bondbetween
theworldandthewordwhichwe striveto maintain during themovefromthe
fieldto analyticworkon data.Rather,mytaskis to applysciencestudies'
imagination to a "nextstep",i.e., to thequalitative analysisitself,morepre-
cisely,totheworkwiththeprogramme I wouldliketo showthatwhat
Atlas.ti.
is oftenseen as an achievement of mindcan be perhapsbetterdescribedin
terms ofpracticalmanipulations withbodiesoftexts.

3. WhyChooseGTM as an Example?
Butwhatkindofqualitative analysisam I goingto discuss?Thereexistdiffer-
enttraditions
andapproaches toqualitativeanalysis5andmyaccountwillinno
case be "methodologically
neutral".In general,I amgoingtotakeas an exam-
ple thekindofqualitativeresearchthatis closeto whatis knownas grounded
theorymethodology (GTM,see GLASER & STRAUSS 1967).
I shouldstressrightfromthebeginning thatit is not "groundedtheory
methodology" as a labelfora self-containedepistemology thatreallymatters.
Rather,byGTM I refer toa looselydefined setofanalytic theuse of
practices,
4 For a recording
is well-tiedwithitstranslation
intotextbymeansof accurateand
instance,
faithful ASHMORE and REED (2000), amongothers,showthatit is notan
transcription.
easytask.
Identification
and comparisonof differentparadigmsin qualitativeresearchhas becomea
populartopicin books and articles(CRESWELL 1997; GUBA & LINCOLN 1994; GU-
BRIUM & HOLSTEIN 1997).

281

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
whichis verycommonamongsociologists, ethnographers, psychologists or
evenhistorians. HowardBECKER (1993,p.228) saysthat"... generalstate-
mentsofwhatmustbe doneto be scientifically adequaterely,usuallywithout
on
acknowledgement,practical matters in
and, this,theyfollowrather thanlead
everyday practice." This is very close to the position of science studiesin
whichtheperspective ofmethodology is suppressed infavourofa sociological
studyof "science in action".6Therefore, to put it in a rathernon-
methodological way (i.e.,without reference to the established notionsoftheo-
reticalsaturation, axial codingor constant comparison), I am goingto talk
aboutthekindofqualitative research projects whichmakeuse oflargeamounts
ofdata,thatareanalysedin systematic andrigorous ways,andwhichaspireto
provideknowledge different from(and in a way superior to) whatis usually
knownbystudied members orparticipants.
ThereareseveralgoodreasonsforchoosingGTM as an exampleformyar-
gument. First,thechoiceis notsurprising giventhecredittheauthorsof Al-
tas.tithemselves maketo thisparticular approach(MUHR & FRIESE 2004).
Further, whether one likes it or not, GTM enjoyspersisting popularity,espe-
ciallyamongstudents andteachers, andaspiresto be takenas an overallstrat-
egy fornon-deductive researchprojects.Occasionally, if takenas a generic
approach forgenerating theory outofqualitative data,itis evenperceived as a
synonymum forqualitative research.7 The current CAQDAS epidemyeven
strengthens thishegemony. Last and perhapsmostimportant (and in close
relationtotheabove)GTM is nowadaysa challenged andoften misunderstood
qualitativeparadigm. Someregarditas somewhat obsoleteandassociatedwith
modernist adherence to scientific rigourandobjectivity, improper forinterpre-
tivesocialresearch (LINCOLN & DENZIN 1994).Further, software packages
organised aroundtheprocedures ofcodingandretrieving contributedtoa more
orlessimplicit conviction thatgrounded theory methodology is nothingbutan
application of thecode-and-retrieve principle. This is an unfortunate misappre-
hension (STRAUSS & CORBIN 1994),whichis difficult tocombat.
On theotherhand,however,the"ecumenical"focuson something-like-
grounded-theory-methodology is relativelyarbitrary. In fact,we couldtryto
followotheranalyticpractices- e.g., in conversation analysisor narrative
analysis- equallywell(perhapswiththeriskofbeinglesswidelyunderstood,
sincethesepractices arefamiliar tofewersocialsciencepeople).
It shouldbe also stressed thatI am notgoingto comeoutwithsomenew
and specificanalytical procedures. No newanalytical techniques and no new
featuresofAtlas.tiwillbe proposed.Instead,I suggest justan alternative"the-

6 A similar
emphasison researchpracticesand a reservetowardtheoriesof qualitativere-
searchcan be foundin SEALE (1999).
7 After the titleof the
all, alreadyquotedbook on GTM by AnselmSTRAUSS (1987) is
AnalysisforSocial Scientists"
"Qualitative (withoutfurther
qualification).

282

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
ory"andpractice-oriented
accountofveryordinary we all
andbasicprocedures
do
usually as analysts.

4. Reality,Virtuality
andPractices
A questionmight appear:ifwe aretounderstand materialpracticesofqualita-
tiveanalysis,why notto look at a pre-CAQDAS researcherworking withreal
things suchas sheetsand of
pieces paper,printers, colour scissors,
pencils, glue
and cardfiles?Sucha focuswoulddefinitely be verypossible.Andat some
moments itcouldbe pretty illuminating.
In comparison to thatwhenan analystworkswitha specialisedcomputer
programme, theonlythinghe orshecanmanipulate seemstobe pureinforma-
tion- bitsandbytesthatarethought to representideasin researcher's
mind.
Indeed,ifwe consider a computer to be a directextensionofhumanthinking,
we couldhardlytalkaboutmaterial practicesat all.8But computers can be
vieweddifferently.Theyhavekeyboards, mouses,speakersandmonitors. And
on screensof monitors we can create,see and manipulate variousobjects.
Theseobjectscanbe ofdifferent sizesandshapes;theycanbe hidden, moved,
split,colourized, and
grouped regrouped, forgotten andrediscoveredon unex-
pectedoccasions.In short,computers providea virtualised environment in
whichwe can notonlydo all theoperations availableto thepre-CAQDAS
researcherequippedwithpaper,scissorsandpencils,butmuchmore.Virtual
objectson thescreenare evenmoreshapeableby andembeddedin practices
thanrealones.

4.1 Thecreation
andbasicoperation
oftextual
laboratory
Whatdo researchers do withAtlas.ti
practically whenanalyzing theirdata?Let
mepickupjusta fewkeymoments oftheprocess.I willproceedfromwhatis
typicalforthebeginning
of theprojectto whatusuallytakesplace at later
stages.

8 Of
course,even workingwithcards,scissorsand colourpencilswithinthe "old" paper-
pencilmodelcan be (and usuallyis) viewedas a directextensionof mentalprocesses.But
still,mostpeople would probablythinkthatpickingcomputers withtheirvirtual,"non-
material"environment as an exampleis notthebestwayhow to overcomethismentalistic
orrepresentationist
approach.ButI believetheoppositeis true:CAQDAS is an opportunity
to graspan alternative
view of qualitativeanalysisas a setof practicalmanipulations
with
data.

283

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
4.1.1Assigning documents
primary
In Atlas.tia research projectis definedbya setof"primary documents" (PD).
Theseareourdata.Anddata,so is thecommonbelief,arewhatwe gatherin
thefield.True,butthisis onlya halfofthestory. Becausedataarealso every-
thingthatwe strivetoputon oneplace, on onetable.Or,moreexactly, intoa
single textual laboratory - which has thepower to shrinktime and spacedis-
tancesbetweenobservable phenomena so thateverything important is present
andundercontrol.9
We can better understand thepointwhenwe imaginewhathappenswhen
primary documents are assignedto a project(to a "hermeneutic unit",as it is
called in Atlas.ti).Addingnew documentshas important practicalconse-
quences:oncewe openthehermeneutical unitnexttimewe immediately have
all thematerials at hand.Thesematerials can havevariousformats - theycan
be texts,photographs, scanneddocuments, audioorvideorecordings. Theycan
evenbe physically locatedonvariousmedia- on harddisks, opticaldiscs,local
network ortheInternet. Butmostimportantly, thesedocuments can havetheir
originin a variety of timesandplaces. Theyreferto different sitesand mo-
ments.
Interviews, recorded and transcribed,
couldhavebeenmade,forinstance,
during last two years in dozens of housholds and officesin severalmiddle-
sizedcities.Buttheinterview or
transcripts, other data"from thefield",arenot
theonlydocuments thatmaybelongto theprimary documents of ourproject.
Otherprimary documents, dependingon circumstances, could be: excerpts
fromliterature on thetopic,written downactuallyduringourentireprofes-
sionalcareereitherat homeor duringstudytripsabroad;scholarlyarticles
downloaded fromonlinedatabasesand coveringseveraldecadesof relevant
research;selectednewspaperarticleson theproblem,publishedin the last
decades;relatedofficial documents obtainedfrom theInternet orcoaxedfroma
range of involved a
authorities;projectproposal ofour research written almost
threeyearsago; e-mailexchanges withcolleagueshomeandabroadthattook
placewhentheproject proposalwasprepared. Andso on.
So nowwe haveall thisin sightandwithin arm'sreach.Or rather, we have
all thisavailableforscrutiny withthehelpof a fewclicksof thecomputer
mouse.Whilebrowsing primary documents of theproject,we travelin time
andspace.It is unbelievably easyandfast:click,click.An interview withMr.
Millerfromthecityof Plzen,May 2005: we talkedabouthownewcivilor-

9
Laboratory has been a prominent topicwithincontemporarysociologyof sciencebothin
the literalsense (i.e., in a numberof laboratorystudiessuch as KNORR-CETINA &
MULKAY 1983; LATOUR & WOOLGAR 1986; LYNCH 1985) and morewidelyin the
senseof laboratory as a basic instrument controland visualisation
for(scientific) (GIERYN
2006; MILLER & O'LEARY 1994).

284

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
ganisations in theCzech Republichad been established in thebeginning of
1990s.Click,click.A resolution of thegovernmental councilforNGOs ap-
provedone monthago: it suggestsa redefinition of thelegal statusof some
non-profitorganizations.Click,click. My own excerpts froma bookon envi-
ronmental movements, published in 1984: I made the excerptsroughly five
yearsago in Pariswhen I was writing a short
note on new social movements.
Now,in thecontext oftheseexcerpts, whatexactlydidMr.Millersay?Click,
click,andherewe are.10
WhenI was reading thebookon environmental movements, I didnotknow
aboutMr.Miller'scivilassociation. I was noteveninterested in it.I had no
ideathatI wouldengage,severalyearslater,in a research projecton expertise
and democracy, forwhichI wouldalso needinterviews withlocal activists.
And at themoment whenI was doingmyinterview withMr. MillerI only
vaguelyrecollected whattheauthors ofthebookhadsaid.Thetwoeventswere
too distantfromeach other.Bothtemporally and geografically. And also in
termsof theirnature, sincetheformer concerns"theory" (and broader socio-
logicalcontexts) while the latteris about the of
production "empirical evi-
dence"(andmyowndata).Butatthepresent moment are
they juxtaposed, next
to each other,rightat hand:Parisand Plzen 1993and 2005 (referring to the
early1980sand early1990s)- hereand now.The distancebetweenthetwo
piecesof realityis verysmallat thepresentmoment, measurable by a few
clicksofthecomputer mouse.Theycanbe carefully compared andconfronted.
As primary documents, theyhave standardized headers(e.g., in comments
attachedto eachPD) thatenableus tokeepthereference totheoriginal distant
timesandplaces.

4.1.2Defining
quotations: PDs intopieces
Cutting
Butitis difficult
tojuxtaposeentirePDs. Theyaretoo large.Thereusuallyis
nopractical to
way squeezetwofulltranscripts intoa singleunifying
view.We
can see morethanone PD at onceonlyas a listof itemsora setof iconsin a
window,arranged in variousways.Evenmore,it is hardto see - at one mo-
ment- a singleentirePD. Bothourvisualfieldandthesize ofthescreenare
limited.We canalwayssee buta coupleofparagraphs.11
We needa different kindof objectto be able to closelystudyour data.
Something smaller.Thatis whywe marksomeparagraphs or sentencesof
interest
particular as "quotations".
In thefirstviewit lookslikemarking rele-
vantpassageson themarginof a book.But thevirtualenvironment allows
more:in fact,bymarking a pieceof data,we notonlymodify andextendthe

10For the sake of the examplesand namesare not authentic,


simplicity, howevertheyare
inspiredbyrealwork.The sameis truefortherestoftheexamplesinthispaper.
Thereare some differences if we consideraudio recordings
or imagesinsteadof textual
PDs, butbasicallytheargument wouldbe similar.

285

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
original PD, butalso createa newanalytical object- a pieceofdataseparated
andfreedfromitsoriginalcontext. The separation is nevercompletethough.
We canalwaystracebackthequotations totheiroriginal location.
Whatis theadvantageof havingthemarkedquotations at handas self-
contained objects? We now have our data in a form that better fitsthescreen
and,in itsvariety and multiplicity, ourfieldof view.12 The references to the
original PDs are preserved: this is what Mr. Miller said, thatis what the gov-
ernmental resolution stated,andherewe havea sociologicalobservation from
Butnowwe canworkwithall thesetextual
literature. piecestogether, sincethe
dataaretransformed in a doubleway.First,theyarereducedinnumber so that
we can focusonlyon whatwe havefoundrelevant so far.13
Second,theyare
reducedinsize so thattheybecomegraspable piecesofdata.Onlynowcanwe
arrange, onthescreenofa monitor, unprecedented rendezvous thatoccurunder
ourdirectvisualcontrol: a pieceof a legaldocument (a particularparagraph)
meetsa pieceofan interview ora passagefroman olderresearch report.14 Do
theysupport each other? Do they contradict? In what sense?Now we are in a
goodposition to startarguing about all that.
Indeed,quotations are elementary unitsof analysisnotonlybecausetheir
are
meanings reasonably contained and therefore accessibleto ourmindsand
mentalprocessing; theyarealso ofreasonable physicalsize to be graspedand
processedin a materialway - by eyes,hands,listsand boxes,computer
screens. Hencethegeneralpointofthispaper,i.e.,toshowthatanalytical work
is inan important sensea material praxis(andviceversa).
Thereis a big"but"in thisthough. Themorequotations we have,themore
distantfromeach othertheyagainbecome.Theyare so numerous thatone
easilygets drowned indata. It takesa and
long painful journey to find a way,or
eventheway,theconnection, fromone piece to another. Two relevant pas-
sages are oftenseparatedfromeach otherby hoursof carefulreadingand
browsing.

12A number ofpracticalrecommendations ofPDs andfontsettings, both


regarding formatting
aimedat good arrangement couldbe givenhere.Fromtheperspective
of visualisation, pre-
sentedhere,thesewouldbe directly relevantforanalyticprocedures.
13The listof
quotationsis not fixed,of course,it growsas we processthedata; and some
quotationsmaybe deleted.Also, we oftengo back to theoriginaldocuments and look for
otherrelevantpassages.Butin anycase,bycreating we createa selectionofdata
quotations
thatallowus to lookat themin greater detail.
14Let us also note:the commentof Mr. Millerwas highlighted as important last
particular
Tuesday,whiletheparagraph oftheresolution hadbeenmarkedas relevanttenmonthsago,
beforethefieldwork started;and theargument hasjustbeennoted.(We im-
fromliterature
mediatelysee theseproceduraldetailswhenwe look at thequotations- thedate of their
creationor modificationis an automatically generatedpartoftheirheaders.)Thus,notonly
variousdocumentsmeetin frontof oureyes at thismoment, butalso variousmomentsof
ourownpreviousanalyticalwork.

286

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
4.1.3Codesandcoding:Reintegrating
thepieces
Piecesofdata,quotations, needtobe somehowordered tobecomemanageable
eveninlargequantities. Thisis wheretheprocedure knownas codingcomesin
as a usefulstrategy.15
By codingwe linkcertain quotationstogetherandform
thematic groupsof data-pieces. Codes are namesforsuchgroups,indicating
whatkindofquotations canbe foundin eachparticular bundle.Herethegath-
ereddocuments, interviews, excerpts, scholarlypapers,projectproposals, and
mediaarticlesspeak,forinstance, about"money", hereabout"legislation" and
hereabout"negotiation". Withthehelpofcodes(andthevirtual Atlas.tienvi-
ronment), we can see thethematic countours of eachgroupof quotations16 as
wellas thesizeofthegroups.
Butcodesarenotjustnames,conceptual labels.Theyarealso usefulhan-
dles withwhichwe can graspand manipulate respectivegroupsof data-
pieces.17Codes can be selected,commented, ordered, filtered,
moved, renamed,
split,andlinkedtoeachother. Theycanbe viewedinlists,hierarchies, network
viewsoras particular occurrences whenbrowsing
(instances) throughourdata.
Anytime we aredoingan operation witha code(e.g.,whenwe arelinking itto
another code orjust selecting it) we do someindirect workon all associated
quotationsas well.
Now,insteadofhavingto freely digthrough andthrough an unsorted heap
of quotations,we can proceedmoreeffectively. By meansof coding,quota-
tionsgainrelevanceand meaningfulness. Somegroupsof quotations become
closerthanothers.Coded data selectively shrinkanalyticdistancesbetween
somepiecesofdata,making theseelementary unitsmoremanageable. In short,
they allow fora kind of more efficient, or
thematicallysemantically organised
reading.18

15
Codingis preciselythe momentwhenan objectionmay easily arise: semanticrelevance
cannotbe assessed by a computerprogramme such as Atlas.ti;the crucialanalyticalas-
sessmentsand decisionsnecessaryforthecodingprocesshave to be made by a thinking
subject.Butagain,I do notdenythatqualitativeanalystshave to think.I onlysay thatthe
practicalinstructivevalue of an appeal such as "Think!Thinkmoreand better!"is rather
low. Furthermore, just takingnoticeof a semanticrelationship does notbring,in itself,any
analyticutility.Such an observation becomeseffective onlytogether withits inscription
intoan analyticalobject("link")thatallowsforitsfurther use.
16Thatis
whyit is so important to chooseappropriate namesforcodes. If we choosebadly,
we do notsee thecontent ofquotationaggregates clearlyenough.
Sincewe can use severalsuchhandlesat once,we shouldkeepcodes simpleandrefering to
a singlething- we can alwayscombinethemfreelylater. J
lu
The possibilities
oforganisedreadingare further enhancedbytheabilityofAtlas.tito make
complexqueries:we can,forinstance, view all thequotations thatspeakbothabout"nego-
tiation"and "legislation"(and studyhowexactlyinall thespeficities).

287

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
4.2 Makingthetextual reallyuseful:Beyond
laboratory
code-and-retrieve
Fourprinciples, onwhichAtlas.tioperates, areintroducedintheuser'smanual
(MUHR & FRIESE 2004,pp.3-4):visualization, integration, and
serendipity,
exploration. Exploration is a verygeneraltermand can be appliedto almost
anything we do in qualitative analysis.19The remainingthreeprinciplesare
morespecific, andthusmoreinteresting.
It can be said thattheprevioussectionof thispaper,focusedon thebasic
logicofcoding-and-retrieving, withwhatMUHR andFRIESE
dealtprimarilly
calledintegration - i.e.,withhowitis practicallyaccomplished bytheanalyst
thatheterogeneous pieces of dataare heldwithinreach,kept undercontroland
becomemanipulable. Indeed,coding-and-retrievingrefersto notmuchmore
thanthemerepossibility oforganised andefficientreading.No matterthatso
manypeoplecannotimaginethatqualitative analysiswouldconsistofanything
morethanprecisely thisprocedure, a rigorous
analyticknowledge in
originates
something else otherthanonlythe coding-and-retrieving And this
activity.
something elsehassomething todo withtheothertwoprinciples, visualization
andserendipity}0

4.2.1 Writing comments


Eachoftheanalytic objectswe createin Atlas.ti- PDs, quotations, links,and
network views- can be accompanied by a comment. There also are "free"
comments, calledmemos, thatcan be attached eitherto more or less thanone
(kindof) object.The ways in whichcomments are used maybe different,de-
pending on thekind of commented object and chosen strategy.For instance,
comments toindividual PDs maycontaindetailedinformation aboutthesource
ofdata.Codecomments wouldtypically,butnotnecessarily, be descriptions or
explanations of namesgivento less obviousor less descriptive codes.In the
case of quotationsor links,comments mightprovideexplanations of whywe
havecreated these -
objects i.e.,what was so about
interesting them.
Memosarea specialcase. Theirimportance andanalyticaluse is typically
growingtogether with the progressof our analysis.In memos we integrate

19The authors was includedamongtheprinciplesmainlyto


jokinglyadmitthat"exploration"
obtaina niceracronym (VISE).
20Visualisationand visual has been extensivelydebated in contemporary
representation
science studies(LYNCH & WOOLGAR 1990; LATOUR 1986; SNYDER 1998). By
meansofvisualisation forcontrolling,
we createconditions manipulating, andaccumulating
small pieces of knowledge,oftenmeaninglessin themselves,and integrating theminto
moreelaboratedand complexstatements. of visualisationin qualitativere-
Nice strategies
searchcan be foundinMILES andHUBERMAN (1994).

288

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
partialobservations. The integration is notjustan abstract mentaloperation. It
corresponds with the abilityof memos to be attached to several codes,quota-
tionsandothermemosat once.We can therefore imaginememosas embryo-
paragraphs or -pagesof a future research report,alreadywell-founded in em-
pirical data and embedded in a broader argument (in the structure of other
the
memos).Ideally, report should be at leasthalf-written within Atlas.ti:
much
ofwriting thereport ina textprocessor (outsideofAtlas.ti)wouldthenconsist
ofediting, associating andcompleting piecesoftextscontained in memosand
associatedanalyticalobjects,especiallyquotationsand variousothercom-
ments.However,sucha denseand empirically grounded network of Atlas.ti
analytical objectsdoesnotappearoutofnothing. It is theresultofa long-term
workwhichgoes through and beyondtheabove-described code-and-retrieve
operations. Whatkindofwork?
It is generally thought thatthemainpurposeof commenting on analytical
objects is to help one's memory. The best to
way prevent the ideas emerging
fromourreadingthedatabecominglostto ourmindsis to writetheseideas
down.Again,thisis a conventional view,in whichtheuse of software pro-
motesand extendsour mentalcapabilities.But thereare otherbenefitsof
commenting.
Firstofall,itis important tonotethatcommenting is oneofthekeymoves
thatconstitute interpretation of data.By meansof writing comments there-
searcher inscribes him-or herself intothestudiedmaterial so thatitbecomes
moreandmoreundercontrol. In thebeginning, almosteverything we have"on
thetable"is whatothers say;as timegoes,theothers'accountsareextended by
ourowninterventions andadditions. Brackets thatmarkquotations emergeon
themargin ofthetext;codenamesareattached to someofthequotations; and,
aboveall,we addourcomments hereandthere. After sometime,we arestudy-
ingnotexactlythesameoriginal data,buta muchrichermixture ofvoices,our
ownvoicebeingincreasingly pervasive among them. This is how sociological
textis produced outofthetextofdata.No suddenswitchfrom theempirical to
thesociological is possible,onlyslowgrowing ofthelatter intotheformer.
Comments shouldtherefore notbe seen onlyas toolsforpreservation of
ideas,butalso (andperhapsrather moreimportantly, sincetheaimofanalysis
is nottojustpreserve ideas!)as a spaceinwhichsociologicaltextis gradually
born.As suchtheyshouldbe madewhenever possible.
Our abilityto add a comment to a possiblenew freequotation or a link
couldevenbe welltakenas a testofwhether thecreation ofcertain newobjects
is legitimate. It is typicalthatbeginners producenew analyticalobjectsof
Atlas.tiin a rather free-and-easy way.Seducedbytheeffortlessness andspeed
withwhichnewquotations or linkscanbe made,theysoonhavethousands of
codedquotations andhardly anyitemunlinked toanything else,without having
an idea whatto do withthesehugequantities of connected objects.Careful
consideration is in place, especiallywhennon-trivial, "stronglinks"are at

289

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
stake.21Butwhatcouldbe a feasiblecriterion fordecisionsaboutwhether to
linktwo quotations or codes or not?Some wouldsuggestvariouskindsof
rational butI recommend
criteria, a pragmatic (andalmostmechanical) one:is
thereanything worthofputting downaboutthisparticular textpassageorcon-
nection? Ifyes,thenletus createthelinkwithconfidence andmaketherespec-
tivecomment. Butifwe areunableto writea comment on theconsidered link
atthetime,andonlyhavean uncertain "feeling" or"sensation", thenwe should
hesitate.If theoryis to be grounded in empirical datathenpracticaldetails,
suchas linksgrounded inarguments (notmentally, butvirtually, intheformof
writtenlinkcomments), areobservable procedural elements ofit.
Creation ofquotations is somewhat different inthis.Themostcommon pur-
pose forcreation ofa newquotation is theneedto code a pieceofdata.Often
creationofa newquotation andcodingcanbe considered a singleoperation.22
Nonetheless, freequotations (unlinked - at the moment of creation- toa code)
canbe a veryusefultool.We canimaginea procedure technically analogousto
creationoffreecodes,whichwouldconsistofmarking outonlyfreequotations
duringan initialreadingof data,without thinking of anycodes (forthetime
being). Strategically,the procedure might be understood as an alterna-
tive/complement to whatSTRAUSS and CORBIN (1990, p.62) call "open
coding".Andprecisely fortheproduction of freequotations we mightuse a
similarruleas forlinks:commented freequotations are fullylegitimate, un-
commented onlyas exceptions.

4.2.2Howtosee relevance?
Let us assumethatourdataare segmented and codedcarefully andwithcir-
cumspection.Segments andcodesarelinkedto eachotherbyvariouskindsof
relationswhereappropriate. Comments are attachedto createdobjectsand
objectstoo),which- as I havejustargued-
links(thatare,in fact,analytical
enhancesthequalityandargumentative groundedness of ourwork.In short,
a
largenumber of partialand limitedanalyticconsiderationshavebeenmaterial-
ised(orrathervirtualised)intheformofobservable andmanipulableobjects-
codes,quotations,comments andlinks... So farso good.Butthissurelycan-
notbe theendofanalysis, butratherthebeginning.
Whatnextthen?Whatto startwith?Thereareso manypotential pointsof
so manypossiblequestions.
interest, We nowneedtobecomefocused.Andwe
also needto reduceourempirical materialand workfurther onlywithsome
of
parts it, themost relevantones.

21Atlas.tiallows creationof kindsof relation)either


qualifiablelinks (links of different
betweenindividual orbetweencodes- thesearecalled"stronglinks"(MUHR &
quotations
FRIESE 2004,p.212).
22Some software
packages,such as Ethnograph origi-
(<http://www.qualisresearch.com>),
nallyevendidnotallowcreationoffreequotations.

290

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
Buthowcanwe recognize a relevant pieceoftext?Howtoidentify themost
relevant codesormemos?Somewouldsuggesta really,reallydeepthink. It's
timeto stepoutfromthesomewhat mechanistic worldofcomputer processing
and finallystartdoingtrueintellectual work... butI don'tthinkso. On the
contrary, thisis themoment whenwe shouldsticktothecomputer andask for
an answerfromAtlas.ti. No, I do notbelieveinmagic(as might be impliedby
thequirkmentioned in theintroduction). I onlybelievein relevanceas an
emergent andrecognizable property ofmyentire workup tonow.
Indeed,a glanceat themonitor andfewclicksofmouseareenoughin At-
las.tito see whichquotations are mostrelevant and thusmostpromising for
furtheranalytical scrutiny.Provided we haveproceeded as described above,we
caneasilyhavea lookateverything we havethought ofourdata.Whatexactly
is worthy ofnotice?Simplyput,an especiallyimportant pieceofourdatais a
quotation for which we have a comment; and/or which is connected to several
codes; and/or which has been linked to (an)other quotation(s), preferably with
commented (argued)links; and/or which has in
appeared noteworthy network
views... But wait,whichnetwork views- amongall thesavedones - are
noteworthy? Again,it is thesameprinciple: thosewithcomments, thosecon-
taining relevant quotations and important codes. Important codes? Yes, those
codesthatare associatedwithhighernumbers of quotations; thatkeepa spe-
cificpositionin theschemeofcodes;thatareusedforclassification ofquota-
tionsinkeyPDs (suchas a projectproposal);and/or thatarelinkedtorelevant
memos.Relevant memos?Yes, again,thosememosthatarelinkedto interest-
ingquotations andcodes(andtherefore areconceptually andempirically satu-
rated);andthosethatarealso linkedto othermemosso thattheyparticipate in
thestructure ofan overallargument.23
All thesequalitiesarewelland easilyvisiblein Atlas.ti.Densityandnature
oflinksespeciallycanbe seenalmostimmediately. Whenyoulookat respec-
tivelistsofobjects,youbecomeoriented in a fewseconds.Recentversions of
theprogramme even offernice summarizing previewsof co-ocurrences of
codesinthedataset.Possibilities ofvarioussynoptic viewsareoverwhelming.
Of course,you cannotstartyouranalyticworkwithAtlas.tiby pressing the
magicbutton marked "Relevant textsearch";butafteryouhavefruitfully spent
sometimeon yourdata,manyAtlas.ti buttons becometruly magical:justclick
onthebutton thatopensa smallquotation manager window and thenmakeone
moreclicktosortyourquotations by the number of links to otherobjects- and
voila,hereon thetopwe have candidatesforthepositionof mostrelevant
piecesofthedata.In thesamemanager we immediately see whichquotations
arecommented andwe canevenfilter outuncommented ones.The listofcan-

23The criteriashouldbe A quotationmaybe consideredas relevantand


appliedsensitively.
suitableforfurther
attention evenifitis codedbya singlecode orno commentis attached-
provided,forinstance,it is coded by an especiallyimportant
code. Simplyput,thecriteria
arenotstrict,buttheystillprovidegoodorientation.

291

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
didatesgetsmorenarrowand solid.Thereare severalwaysto findouthow
manycodes (and whichcodes) are associatedwiththecandidatequotations.
Aretheseimportant codes?Ifyes,therespective quotation shouldbe elevated
intheranking Andso on.24
ofcandidates.
You can see all thisquicklyandeasily,without seriousor deepconsidera-
tionsinvolved. not
Well, really. But the important ofthinking
acts havealready
happened, in countlessmoments of our coding,segmenting, commenting,
linking...; andnowit is sufficientto onlytakea brieflookandmakeuse of
thesenumerous smallacts materializedand visualizedin a powerfulsum.If
youtrust yourjudgement, as it has beenappliedduringthelongterm andde-
tailedworkwithindividual PDs, quotationsand otherobjects,you can com-
fortablyrelyuponthecriteria outlinedabove.Theyhelpto crownyourentire
effort.

4.2.3Readingdataina newway
Fromthesuggested pointof view,thequalityand relevanceof conceptsand
theirempirical content are theresultsof theongoinganalytical work,notits
precondition. Relevance is made. And it is made notexactlyby ourthinking
alone.Rather, as something thatcan easilybe seen,it is produced bymaterial
practices, in whichthe virtualenvironment of thecomputer plays crucialrole
a
of mediator. Atlas.tiprovidesan interface in whichandthrough whichwe do
thinking.
We couldsimilarly describepracticalcounterpartsofsomeothermentalop-
erations. Let us take,forinstance, thesituation whenwe needto temporarily
lookawayfromtheoretical conceptsused up to themoment and look at our
data"withneweyes".Thisis a difficult taskforone's mind,requiring a lotof
self-disciplineandrenunciation. Butithas a verypractical dimension. We can
arrange ourworking environment, ourvirtual scene,so thatthesoftware takes
on (at leastpartly)theburdenoftheabove-mentioned intellectual challenge.It
is possible,witha fewclicksof mouse,to simplyfilter outall therespective
codes- i.e.,thecodesthatembodytheabove-mentioned theoreticalconcepts.
As a result, theycompletely disappear from thevirtual
desk. They can be found
neither in thecode managernorin theobjectexplorer. Thesecodes are re-
movedevenfromthemargin area.Simplyput,theytemporarily ceaseto exist.
Andthisis howit is practically donethatthestudieddocuments areread(as
muchas possible)"anew",without theconceptual burdenofpreviousanalysis.
Outofsight, outofmind.

24It shouldbe notedforthosewho arenotfamiliarwithprogrammes suchas Atlas.tithatone


can easilyskipfromlistsand overviewsto quotationsthemselvesand theiroriginalloca-
tions.In otherwords,we notonlysee whethera quotationis commented or not or how
readitin fulllength.
manylinksithas to otherobjects,we can immediately

292

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
Makingthingstemporarily invisible, or rather something we couldcall se-
lectivevisualisation, is an important aspect of the visualisation principle.It
in
occurs, fact, all the time. Imagine the most ordinary situation when we
browsequotations ascribedto a code or severalcodes.Sucha procedure sub-
stantiallytransforms our reading of the data. We do not read individualdocu-
mentsas usualanymore, i.e.,one afteranother. Insteadof studying theinter-
viewwithMr.Miller,thenthelegaldocument, thena sociologicalarticle, then
another interview, andso on,we proceedtransversely. By listingandviewing
all thequotations coded,e.g.,bythecode "money", we construct - outo/the
originaldataandinaddition -
tothem a newcomposite andmulti-vocal texton
financial matters. Thiscomposite textis another embodiment of ourprogres-
sivemovingfromoriginalcontexts andmeanings to a sociologicalargument.
As a newelement,newlya created object, belongsa littlelessto ourrespon-
it
dentsanda littlemoretous,analysts.
WhenI speakoftheconstruction of a newtextI do notmeanitas a meta-
phor. What we have here is a quite real sequenceofsentences andparagraphs,
whichcanbe readon themonitor fromthebeginning to endandwhichcanbe
savedas a newdocument orevenprinted on paper.We canevenassignsucha
newly created document as another PD to ourproject(hermeneutical unit)and
treatitas material to be further analysed.25 . . . Whyshouldwe? Becauseonce
thepiecesofdataarecutofffromoriginalcontexts andputto other(themati-
callydefined) relationships, theytella storyunheard sofar.Whatseemedtobe
important at firstmaysuddenly appearas a minorissue;conversely, whatwe
originally considered as marginal maygainimportance, since,forinstance, it
becomesclearhowoften different people mention it.A space fornew insights
andideasopensup,whichbrings aboutnewtextual additions (comments, links,
codings), andthusalsonewrelevances ... theserendipity principle inaction.
Whatthenconstitutes thenewqualityofsociologicalreadingofdata?How
is a newunderstanding ofreality born?Initially it seemsthatinterpretation of
qualitative data involves a of
range manipulations with textual units- manipu-
lationsthatstemfromrepeatedreadingof one and thesamesetof collected
data.A closerlook,however, revealssomething else. The researcher in fact
manipulates thetextsofdataso thatnewtextsareprogressively created(writ-
ten)outof theold onesand alongsidethem.It is nota linearprocess,buta
tangledand intermittent procedure. As its result,a numberof new accounts
emerge, in which the voices of studied actorsare stillpresent, butmoreand
moreso is thevoiceoftheresearcher. Thesenewaccountsofferandprovoke
newperspectives andinsights. Sucha textual practice,basedas muchon writ-
ingas onreading, is theprimary vehicleoftheproduction ofa newunderstand-
ing.

25This
mightbe considered
as a "systemclosure"(RICHARDS & RICHARDS 1994).

293

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
Once we startconsidering "natural"activitiesof consciousness such as
thinkingor seeing26
as embodiedmaterial practices,we betterrealizeon what
trulyseparatesfromordinary
groundssociologicalinterpretation social inter-
pretations.
Sociologistsscore not
primarily by bright and refined mindsor
sharpeyes, but rather
by everythingtheypractically perform with their data.
Whatappearsas readingone and thesamedatain a newway,whichcan be
takenas a desireable
outcomeofanalysis,is in factan effectoftheprocedure
in whichwe artfullyproducenewandnew(versionsof) textsandreadthem
withbasicallyoneandthesamesetofeyesandmind.

5. Conclusions
It canbe saidthatCAQDAS has brought aboutextraordinary easiness,speed,
andreliabilitywithwhichwe canmovethrough andthrough extensive datasets
andwithwhichwe areabletoremember, recollect
andthink. Butprogrammes
suchas Atlas.tioffermuchmorethanthat.Theyenableus tosee fromvarious
what
perspectives (we believe)happensin ourminds.The sophisticated inter-
faceof thesesoftware toolsis important notonlyto allowintuitive andcom-
fortableoperation, butalso becauseit bringsa rangeof mutually relatedde-
vicesofvisualisation.
Atlas.tithereforeenablesresearchers to thinkin a visibleway.Visualised
thoughts or mentaloperations can easilybe stored,recollected, classified,
linked,filteredoutin greatnumbers . . . andmademeaningful insum.Visuali-
sationimplies, forinstance,
thatcodesarenotonlymentalentities orconcepts,
butalso namedelements ofvarioussize andcolourthatcanbe manipulated by
handsand controlled by vision.Thus,thinking madevisibleis by thesame
tokenthinking mademoreaccountable andinstructable.27
is
Thinking inseparable doing.Thisis an important,
from butneglected les-
son forqualitative It is
analysis. paradoxical thatso many texts on qualitative
methodology ignorethelesson,giventhefactthatitwas introduced andelabo-
ratedwithin severalrelatedintellectual traditions
thatconstitute thetheoretical
background of key qualitative approaches.28 The adventof CAQDAS even
deepenedtheparadox.Software packagesforqualitative analysisare often
26
Seeing as instructedand materialpracticehas been nicelydemonstrated, froman eth-
nomethodological in a recentconference
perspective, paperby LAURIER and BROWN
(2005).
27 was recentlyhighlighted
Instructability by GARFINKEL (2002) as a key conceptfor
ethnomethodological understanding of practicalactionwhichcan neverbe fullydependent
on rule-following,
butwhichis stillunderstandable andaccountable.
28I refer
here, above all, to phenomenology(MERLEAU-PONTY 2002 [1945]),
ethnomethodology (GARFINKEL 1967; HOLSTEIN & GUBRIUM 1994), post-
structuralism(DERRIDA 1976; DENZIN 1994, 1995), and constructivism (BERGER &
LUCKMANN 1967; SCHWANDT 1994).

294

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
presentedas tools thatcan extendand supportcapabilitiesof researcher'smind,
butthatcannot"reallythink".As such,theyreaffirm thementalistic, essentially
of
methodologicalconception knowledge.
Inspiredby contemporary science and technologystudies,I have attempted
to show CAQDAS and qualitativeanalysis in a different light.Insteadof de-
scribingordinary moments of qualitativeanalysisand interpretation in termsof
specific mental operations(representedin the software'sinterface),I have
emphasisedmaterialpracticesand manipulations.The analyticworkwithAt-
las.ti is especiallysuitableforsuch reframing. Indeed,it mightbe arguedthat
qualitativecomputingis misunderstoodinsofaras softwarepackages are not
seen as virtualenvironments or media forembodiedand practice-basedknowl-
edge making. Inseparably of thinkingand doing in qualitativeanalysis is
hardlyobservablebetterelsewhere.
Groundedtheorymethodology(broadlydefined),thismore or less explicit
alterego of CAQDAS, has been reframedtoo. When describednot in termsof
methodologicalor theoreticalconceptsbut ratherin termsof what we practi-
cally do withtheanalyseddata,groundedtheorybecomes perfectly compatible
with the textualist,post-structuralist paradigm(fromwhich it has allegedly
departedconsiderably).As ZygmuntBAUMAN summarises(1992, pp.BOf):
One of themostimportant boundariesthatcannotbe drawnclearlyand that
generateambiguity in theveryprocessof beingcompulsively drawnis that
betweenthetextand itsinterpretation. The centralmessageof Derridais that
interpretation is butan extensionof thetext,thatit"growsinto"thetextfrom
whichitwantsto setitselfapart,and thusthetextexpandswhilebeinginter-
pretedwhichprecludesthepossibility of thetexteverbeingexhaustedin in-
terpretation.
And this is preciselywhat we have seen. The way analystsmanipulate,
transform and extendPDs in Atlas.ti(or withscissors,glue,and colourpencils)
might be taken as an empiricaldemonstration of this post-structuralist argu-
ment.To put it differently, GTM looks desparately"modern",scientistic,and
faraway fromwhatwas broughtaboutby thetextualturnin the social science
only insofaras its proceduresare interpreted "immaterially", i.e., as basically
conceptual work on data. Once we take seriously STRAUSS' s statement,
quoted at thebeginning,thatqualitativeanalysisshouldbe understoodas both
physicaland conceptualsets of tasks,GTM becomes open to all post-structural
and radicalconstructivist sensitivity.
Such understanding of GTM, however,does not implya loss of normativity
and instructability. The contraryis true.While GTM has always been popular
amongteachersand studentsforitsrelativeabilityto be formulated as practical
and understandableguidelinesforaction,the proposedreframing would only
enhancethisvirtue.

295

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
References
Ashmore, Malcolm & Reed, Darren (2000). Innocence and Nostalgia in
ConversationAnalysis:The Dynamic Relationsof Tape and Transcript[45
paragraphs].Forum QualitativeSozialforschung / Forum: QualitativeSocial
Research[On-lineJournal],7(3),. <http://qualitative-research.net/fqs-texte/3-00/
3-00ashmorereed-e.htm> [Date ofaccess: December15,2006],
Bauman,Zygmunt (1991). IntimationsofPostmodernity. London:Routledge.
Becker,HowardS. (1993). Theory:The NecessaryEvil. In David J. Flinders&
GeoffreyE. Mills (Eds), Theoryand Concepts in QualitativeResearch:
PerspectivesFromtheField (pp.218-229).New York:TeachersCollegePress.
Berger,PeterL. & Luckmann, Thomas(1967). TheSocial Construction ofReality:
A TreatiseintheSociologyofKnowledge.Harmonds worth:PenguineBooks.
Coffey,Amanda;Holbrook,Beverley& Atkinson,Paul (1996). QualitativeData
Analysis:Technologiesand Representations. SociologicalResearchOnline,7(1).
^ttp^/www.socresonline.org.uk/l/lM.htm^ [Date ofaccess: July18, 1998].
Creswell,JohnW. (1997). QualitativeInquiryand ResearchDesign: Choosing
AmongFive Traditions. London:Sage.
uenzin,iNorman r. (iyy4j. ine An ana ronticsoi interpretation, in iNorman k.
Denzin& YvonnaS. Lincoln(Eds), HandbookofQualitativeResearch(pp.500-
515). London:Sage.
Denzin, NormanK. (1995). The Poststructural Crisis in the Social Sciences:
LearningfromJames Joyce. In Richard Harvey Brown (Ed), Postmodern
Representations: Truth, Power,and Mimesisin theHumanSciencesand Public
Culture(pp.38-59).Urbana& Chicago:University ofIllinoisPress.
uerriaa,Jacques(iy /oj.uj urammatoiogy. «aminore,mu: jonnsnopKinsrress.
Garfinkel,Harold(1967). StudiesinEthnomethodology . EnglewoodCliffs:Prentice
Hall.
Garfinkel,Harold(2002). Ethnomethodology 's Program:Working OutDurkheim 's
Aphorism. Oxford:Rowman& Littlefield Publishers, Inc.
Gieryn, ThomasF. (2006). Cityas Truth-spot: Laboratories andField-sitesin Urban
Studies.Social StudiesofScience,36('), 5-38.
Glaser,Barney& Strauss,AnselmL. (1967). TheDiscoveryof GroundedTheory:
forQualitativeResearch.New York:Aldinede Gruyter.
Strategies
Guba,Egon G. & Lincoln,YvonnaS. (1994). Competing Paradigmsin Qualitative
Research.In NormanK. Denzin & Yvonna S. Lincoln (Eds.), Handbookof
QualitativeResearch(pp.105-117). London:Sage.
Gubrium, JaberF. & Holstein,JamesA. (1997). TheNew Languageof Qualitative
Method.New York:OxfordUniversity Press.
Holstein,James A. & Gubrium,Jaber F. (1994). Phenomenology,Ethno-
methodology, and Interpretive Practice.In NormanK. Denzin & Yvonna S.
Lincoln(Eds.), HandbookofQualitativeResearch(pp.262-272).London:Sage.
Knorr-Cetina,Karin & Mulkay, Michael (Eds.) (1983). Science Observed:
Perspectiveson theSocial StudiesofScience.London:Sage.
Konopásek,Zdenëk (1997). Co si pocit s pocitacemv kvahtativnim vyzkumu:
program ATLAS/tiv akci [Whatis Computer AssistedQualitativeData Analysis
Good For?Atlas.tiinAction].Biograf12, 71-110.

296

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
Konopásek,Zdenék(2005a). Abymysleníbylodobrevidèt:Nad novouverzípro-
gramuAtlas.ti[MakingOur Thinking Visible:A Reviewof theNew Versionof
Atlas.ti].Biograf,37, 89-109.
Konopásek,Zdenék(2005b). Co znamenáínterpretovat text?[WhatIt Means to
Interpret a Text of QualitativeData?] In Michal Miovsky,Ivo Cermák &
VladimirChrz (Eds.), Kvalitativní prístupa metodyve vedácho clovëku- IV:
Vybranéaspekty teorie a praxe [QualitativeApproachand Methodsin theHu-
man Sciences- IV: TowardSome Aspectsof Theoryand Practice](pp.85-95).
Olomouc:FFUP.
Latour,Bruno(1986). Visualisationand Cognition:Thinking withByesand Hands.
Knowledgeand Society:Studiesin theSociologyof CulturePast and Present,6,
1-40.
Latour,Bruno(1987). Science in Action:How to Follow Scientistsand Engineers
through Society.MiltonKeynes:Open University Press.
Latour,Bruno(1995). The „Pédofil"ofBoa Vista:A Photo-philosophical Montage.
CommonKnowledge,4(1), 144-187.
Latour,Bruno& Woolgar,Steve (1986). LaboratoryLife: The Construction of
ScientificFacts (2nd.edition).Princeton:PrincetonUniversity Press.
Laurier,Eric & Brown,Barry(2005). Methodand Phenomena:Learningto See
Fish and Flying Objects. Paper preparedfor the International Institutefor
Ethnomethodology and ConversationAnalysis conference „International
Perspectives", BentleyCollege,Waltham, MA 02452,USA.
Law, John(1986). The Heterogeneity of Texts.In MichelCallón,JohnLaw & Arie
Rip (Eds.), Mappingthe Dynamicof Science and Technology:Sociologyof
ScienceintheReal World(pp.61-S3). London:Macmillan.
Lincoln,YvonnaS. & Denzin,NormanK. (1994). The FifthMoment.In Norman
K. Denzin & Yvonna S. Lincoln (Eds.), Handbookof QualitativeResearch
(pp.575-586).London:Sage.
Lynch,Michael (1985). Artand Artifact in LaboratoryScience: A Studyoj Shop
Workand Shop Talk in a ResearchLaboratory.London: Routledge& Kegan
Paul.
Lynch,Michael & Woolgar,Steve (Eds.) (1990). Representation in Scientific
Practice.Cambridge: MIT Press.
Merleau-Ponty, Maurice(2002 [1945]). Phenomenology ofPerception.London&
New York:Routledge.
Miles,MatthewB. & Huberman, MichaelA. (1994). QualitativeData Analysis:An
ExpandedSourcebook(2nd.edition).ThousandOaks: Sage.
Miller,Peter & O'Leary, Ted (1994). The Factoryas Laboratory.Science in
Context, 7(3), 469-496.
Muhr,Thomas & Friese,Susanne (2004). User's Manual forAtlas.ti5.0 (2nd.
edition).Berlin:Scientific SoftwareDevelopment.
Richards,Tom & Richards,Lyn(1994). UsingComputers in QualitativeResearch.
In NormanK. Denzin & Yvonna S. Lincoln(Eds.), Handbookof Qualitative
Research(pp.445-462).London:Sage.
Schwandt,ThomasA. (1994). Constructivist, st Approachesto Human
Interpretivi
Inquiry.In NormanK. Denzin & Yvonna S. Lincoln (Eds.), Handbook of
QualitativeResearch(pp.118-137).London:Sage.
Seale, Clive (1999). TheQualityofQualitative Research.London:Sage.

297

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions
Snyder,Joel (1998). Visualizationand Visibility.In CarolineA. Jones& Peter
Galison (Eds.), PicturingScience - ProducingArt (pp.379-397). London:
Routledge.
Strauss,AnselmL. (1987). QualitativeAnalysisfor Social Scientists.Cambridge:
Cambridge UniversityPress.
Strauss,AnselmL. & Corbin,Juliet(1994). GroundedTheoryMethodology:An
Overview.In NormanK. Denzin & Yvonna S. Lincoln (Eds.), Handbookof
Research(pp.273-285).London:Sage.
Qualitative
Teil, Geneviève& Latour,Bruno(1995). The Hume Machine: Can Association
NetworksDo More Than Formal Rules? StanfordHumanitiesReview,4(2),
47-65.

298

This content downloaded from 143.88.66.66 on Tue, 3 Sep 2013 14:43:51 PM


All use subject to JSTOR Terms and Conditions

You might also like