The Analysis of Communication Content: Developments in Scientific Theories and Computer

Techniques by George Gerbner; Ole R. Holsti; Klaus Krippendorff; William J. Paisley; Philip
J. Stone
Review by: Harold D. Lasswell
AV Communication Review, Vol. 18, No. 4 (Winter, 1970), pp. 461-467
Published by: Springer
Stable URL: http://www.jstor.org/stable/30217617 .
Accessed: 22/06/2014 21:17

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.

Springer is collaborating with JSTOR to digitize, preserve and extend access to AV Communication Review.

http://www.jstor.org

This content downloaded from 188.72.127.32 on Sun, 22 Jun 2014 21:17:24 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS : 461

tion are dominant factors in determining what is learned in a

given situation. He argues persuasively that teachers exert a
major influence on the feelings and emotions of children in a
classroom. He notes that they often exert their influence
through tone of voice, facial expression, touch, and posture,
and that "the key behavior is listening and responding with
understanding."These are views not to be taken lightly by in-
structionaltechnologists.
Anyone who studies these three books carefully is likely to
find other points on which instructional technologists such as
Kibler, Barker, and Miles hold views different from those of
educators such as Tyler et al. It is also likely that they may
quarrel with my interpretations of the positions of persons
quoted. However, it is hoped that the quotations given will
identify and illuminate to some degree the fact that a number
of respected professional educators see the educative process
within a different frame than that of most instructional tech-
nologists.
For this reason, the experiencedprofessional as well as the
neophyte in instructional development and technology will find
it profitable to read and reflect on the ideas and recommen-
dations of the authors of the papers and essays in Concepts of
Teachingand ImprovingEducationalAssessment.
It would be highly desirable for all parties concerned, as
well as for American education, if there could be serious and
extended dialog between the instructional technologists and their
colleagues in education whose specializationsare philosophy, the
arts, religion, curriculum,and evaluation.
-PAUL W. F. WITT
InstructionalMediaCenter
MichiganState University

ContentAnalysis THE ANALYSIS OF COMMUNICATION CONTENT: DEVELOP-

Comesof Age MENTS IN SCIENTIFIC THEORIES AND COMPUTER TECHNIQUES,
edited by George Gerbner, Ole R. Holsti, Klaus Krippendorff,
William J. Paisley, Philip J. Stone, New York: John Wiley &
Sons, Inc., 1969. xix, 597 PP. $14.95.
The first important symposium on content analysis was held in
1955 at Allerton House, University of Illinois, and led to the
publication of Trends in Content Analysis edited by Ithiel de
Sola Pool (i959). The initiative was taken by the Committee
on Linguistics and Psychology of the Social Science Research
Council. The initiative for the present symposium came from
Dean George Gerbner of The Annenberg School of Communica-
tions, University of Pennsylvania, Philadelphia, and the editors

This content downloaded from 188.72.127.32 on Sun, 22 Jun 2014 21:17:24 PM

All use subject to JSTOR Terms and Conditions
 AV COMMUNICATIONREVIEW : 462

of the current volume. The general Conference at Annenberg

was in 1967.
The papers are presented in four parts: theories and analyt-
ical constructs, edited by Klaus Krippendorf;aspects of inference
from content analysis, edited by Ole R. Holsti; the recording
and notation of data, edited by William J. Paisley; computer
techniques in content analysis and computational linguistics, by
Philip J. Stone. About a third of the papers are in Part 4,
and most of the remaining papers either utilize the computer
or touchon issues that have been sharpenedby its use. Obviously the
exploitationof the computeris the biggest recent event in the field.
The distinctive frame of reference of those who study "con-
tent" is rather clear. The concern is with "interaction" among
"living forms," and more particularly with "communication."
Yet more narrowly, the frame of reference is the flow of "mes-
sages." Specialists on content analysis are forming a subculture
whose members exchange messages about messages, a meta-
linguistic exercise whose theories and procedures enable them to
adopt a problem-orientedapproach having philosophic, historic
scientific, prophetic, and manipulative-creativecomponents. Most
of the discussion in the present symposium is directed to the
scientifictask. In general:
Content
Specialists A--. messagesx---- B

Other
Communicators C- messagesx----- D

It is understood that the description of a message calls for

references to sequences (including concurrences) of two catego-
ries of events: signs and symbols. Signs are physical events
(e.g., sounds, written letters); symbols are subjective events (im-
ages and moods of the communicatorswhose patterns are me-
diatedby signs).
A message-describer,A, has before him an observational field
of signs. His responsibility is to manage his symbols in such
a way that he produces a sequence of signs that elicit in B
symbols of accurate reference to the relevant symbol events
(usually C's). An initial step is to formulate a question about
the phenomenathat A and B are both interestedin answering. The
numberof such potential questions cannot be predicted.
Techniques of analysis aim at limiting the arbitrariness of
message-describers.All procedureshave two features in common:

This content downloaded from 188.72.127.32 on Sun, 22 Jun 2014 21:17:24 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS : 463

i) they guide the focus of attentionof the classifier;2) they

specifytherecording unitemployedin theclassification.
For instance,the describermay be instructedto focus on a
"contextunit" (sentence,paragraph,article,and so on) and to
record t) the numberof printedwords that correspondto the
persons,groups,organizationslisted in his instructions,and 2)
the numberof statementsthat refer favorablyor unfavorably
to any of these words.The instructionsmay be furtherrefined,
for example,to limit the context to be taken into accountin
performingoperation2 (e.g., limitingthe contextto a sentence).
The scientificobservermay act as his own coder or accept
trained assistance;and he may adopt various techniquesof
learning,of intercoder reliability,andof sampling.
Such rudimentaryproceduresmay be adequatefor a partic-
ular questionat hand, such as whetherin a given period the
terms in a small body of authoritativemyth (a Declarationof
Independence)are occurringwith more or less frequencyin a
limited corpus of elite politicaldiscourse(such as speechesby
topofficialson greatpublicoccasions).
There are, however,a great numberof questionsfor which
the proceduresoutlineddo not provideacceptableanswers.For
instance,corresponding studiesmight be plannedfor all known
bodies politic, past or present, and for elites, mid-elite,and
rank-and-filecommunication, so that theoriesabout the appear-
ance, diffusion,or restrictionof politicalmessages (or message
components)could be confirmedor disconfirmed. The techniques
need to be tightenedz) to reducethe freedomof inferenceby
message-describers,and2) to permittheuseof machines.
All the messagesthatoccurin a givencontextof interactionmay
be describedas a subsystemwithin a more comprehensive system
of "communication" or "socialinteraction"(e.g., of the culture,
socialclass,interest,personality,or levelof crisis)if variouscriteria
aremet (suchas thoseoutlinedby AnatolRapoport).
Whatwoulda comprehensive theoryof messageslook like?One
majorcomponentwouldundoubtedly focus on "syntactic"interde-
the
pendenciesamong signs employedin formingmessageswithin
the sign systemor amongthesignsystemsavailableto the communi-
catorsin a given context.In principle,linguistsare ableto identify
the grammarsby whichpermissiblesentencesare constructed in a
particularnaturallanguageor in naturallanguagesas a whole.Con-
tent analystshave not yet been much concernedwith the syntacticsof
gesture, music, dance, painting, sculpture, architecture,and other
nonverbal systems. Hence we have little experiencewith the char-
acteristicsof nonverbalsystems as a whole or with the way in which
different sign systems affect one another in the context of a single
society. It does not carryus far to be told that in crisis periods "fast
media" (like newsprint)outcompete"slow media" (like architecture)

This content downloaded from 188.72.127.32 on Sun, 22 Jun 2014 21:17:24 PM

All use subject to JSTOR Terms and Conditions
 AV COMMUNICATIONREVIEW : 464

in the total messageflow, or that the largest iconic media (like archi-
tecture)formulatethe most permanentmessagesof a society.
In this connection the analysis of conceptualcategories by James
Deese is seminalwhen the relationshipsbetweenwhat he calls "ideas,
reality and language" are generalizedto all acts of communication.
The messages among content analysts can be words about nonverbal
as well as verbal perspectives(symbols)and signs. Suggestive points
emerge in BarryS. Brook'scontributionconcerningthe purportand
style of music, and in the reportby Paul Ekman,William V. Friesen,
and ThomasG. Taussigon SCAN (systematicclassificationand anal-
ysis of nonverbalbehavior).
Since the study of origins is peculiarlydifficult, message analysts
may not be able to contributemuch to a comprehensivetheory of the
circumstancesin which "information"first transmits "messages,"
although the study of child developmentprovides useful clues. The
problem can also be approachedby investigating intraspecies com-
munication,a burgeoningfield that should be representedat the next
majorsymposiumon contentanalysis.
The problemof selectingfrom a field of interactionthe phenomena
eligible for study by content analysts is a complex one. Usually the
analyst preparesa "text" which is to be the object of further proce-
dures. The computerfrees us from spending an untold number of
manhourson many boringoperations.But some of the most strategic
bottlenecks have not yet been overcomeby adequateautomatic de-
vices. We do not yet have scannersequippedto cope with the varieties
of letter style and format that are found in printed or manuscript
books, periodicals,pamphlets,posters, or correspondence;or which
satisfactorilycover nonprint media; or which cope with translation
problemsin all languagesand sublanguages.However, the improve-
ment has been so dramaticthat it is churlishto harp on the gap be-
tween performanceand perfectionand to lament the necessity of uni-
formizinga text.
Some recordinginstrumentalitiesenlarge the potential of content
analysis by disclosinghitherto concealedsigns that may or may not
be part of the "unconscious"component of message launching or
audience interpretation.Film captures facial movements that come
and go too quickly for ordinarymethodsof observation,and clinical
psychologists are accustomedto rely on indices of pulse rate and
many other "invisible"somaticsigns to delineatethe direction,elabo-
ration, and intensity of a message sender or receiver (and for con-
venience these may be included as componentsof message purport
and style). John A. Starkweatherdeals succinctly with some mea-
surementtechniquesappropriateto oral communication.
Even though it is not possible as yet to control the inferences of
text makers, we can go a long way toward accomplishingthis for
text handlers,especiallywhen the computeris accessible.Responding
to the growing complexity of content analysis, Klaus Krippendorf
identifies three "models"of messages."Association"models are sta-

This content downloaded from 188.72.127.32 on Sun, 22 Jun 2014 21:17:24 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS : 465

tistical correlationsbetweenobservedvariables(such as Dollard and

Mowrer's DiscomfortRelief Quotientwhich was computedfrom the
frequencies of two classes of words and indicators of a speaker's
state of stress). "Discourse"models consider contents as linguistic
referents and "realize them in denotationsand connotations."This
method is not yet well exemplified in much completed research.
"Communication"models depend on control processes "within dy-
namic systems of interaction"and are also more programaticthan
fully actualized.However,the "conversational"expositionby David
G. Hays, the "dyadic" analysis by George Psathas, the "interna-
tional relations"results obtainedby Hosti and his associates,are all
highly pertinent.The carewith which the three proposedmodels are
explicatedwill undoubtedlyraise the level of theoreticalanalysis in
the field.
I shall not dwell on the many fascinatingdisplays of technicalin-
genuities and theoreticalinsights found in the papersthat summarize
researchprogramsin many conventionallyseparablefields of knowl-
edge. For example, it is tempting to underlinepoints in Calvin S.
Hall's analysis of dreams,Daniel M. Ogilvie's study of folktales, or
Howard P. Iker and NormanHarway'sattemptto avoid a priori cate-
gorization.
ContentAnalysis Rather, I propose to commenton the latent significanceof content
as a Policy analysis as a policy science. The policy science approachdoes not
Science focus on topics that hold the headlines, although it may have much
that is timely to offer about such items (as, for example,William J.
Paisley's examination of style-"minor encoding variations"--in
the Kennedy-Nixon debate). The conception of the policy sciences
emphasizesthe analysis of choice and the mobilizationof knowledge
for choice and also the essentiality of operating with a contextual
model of social process, a problem-orientation,and a polymethodo-
logical strategy.
The present symposiumindicates that content analysis is emerg-
ing as an identifiablefield of knowledge, although not many spe-
cialists are fulltime professionals.The theory of message analysis and
management is on the verge of articulatinga distinctive frame of
referencewith a professionalcorps of devotees. For example, in the
broad field of communicationit is distinguishablefrom information
engineering.
A remarkabout timeliness:our knowledge of the world has been
enormously expandedand deepenedby two great intellectualrevolu-
tions. One is the spreadof mass-energyequivalentsas generalizedby
Einstein and the super-rapidstate changes formulated by Planck.
The second innovation is crucialfor the life sciences. Although in-
formation sequencesutilize energy, this is not the crucial character-
istic. The importantfeatureis patterning(genetic information,etc.).
A third revolutionconcernssequencesof events that are only in part
describablein physical and informationalterms.They differfrom in-
formation patterning since they introduceanother and much richer

This content downloaded from 188.72.127.32 on Sun, 22 Jun 2014 21:17:24 PM

All use subject to JSTOR Terms and Conditions
 AV COMMUNICATIONREVIEW : 466

configuration,which is referentiality.Symbol events are not distribu-

tions of "bits"but of "ments" (i.e., mentions of time-spaceand fig-
ure).
Message analysis is central to the emerging "third revolution."
As indicatedbefore,the scientifictask is to explain symbol - sign dis-
tributions and successions. Suppose we describe a given configura-
tion of perspectives(identities,values, expectations).Can we account
for the immediatelypreceding and succeeding configurations?Our
explanatorypropositionswould need to accountfor the intersymbolic
consequencesfollowing changes in a particularcategory of symbols.
And this is not a hopeless task. For instance, we are not surprisedto
find that symbols of adversefactual referenceto a political party are
often followed by shifts among symbols that assert identity with
competingparties; and that symbols of demandfor particularvalues
follow in the wake of factual referencesto threateneddeprivations
in terms of such values (wealth, etc.). Existing knowledge, though
largely unsystematized,can carry us a long way toward formulating
scientificinterdependencies amongsymbols.
The scientifictask is only one of the five intellectualoperationsre-
quiredby a problem-orientedapproachto content analysis. The pres-
ent proceduresand projectsavailable for describinghistorical trends
are summarized,amongothers, by GerardSalton and KennethJanda.
Futureprojectionsare neglected;but particularmanipulativealterna-
tives are partly dealt with (e.g., the "pedagologic" of Edwin S.
Schneidman). Goal models for message systems are left implicit,
although Gerbner'scontextualpresentationcould be adaptedto pos-
tulatedachievementcriteria.
A perfectionist'sdreamfor messageanalysis as a policy sciencefield
would emphasizeits role in referenceto the five problem-oriented
tasks mentioned above: I) the clarificationof valued goals for the
flow of messagesin a social context (such as the criteriafor the con-
tent of the public media that would inform and mobilize effective
participationby the membersof the context); 2) the descriptionof
past and currentmessagesin the relevantsocial process in ways that
reveal the degree to which valued goals have been approximatedor
failed of approximation;3) the analysis of the factors that have con-
ditioned the messagesdescribed;4) the projectionof the probablese-
quence of future messages under probablecontingencies;5) the in-
vention, evaluation,and selectionof intermediatemessages and strat-
egies that, if they occur, would result in realizing a message flow
most closely approximatingthe goal criteria, at least net cost and
risk.
As shown by the survey of education in content analysis by F.
EarleBarcusthe existing coursesor course-includedpresentationsof
the subjectare spreadacross the entire map of knowledge in the so-
cial sciencesand the humanities.Contentanalysis is a label for mis-
cellaneoustechniquesthathave been improvisedby many scholarsfor
highly diversifiedpurposes. Slowly, however, the distinctive char-

This content downloaded from 188.72.127.32 on Sun, 22 Jun 2014 21:17:24 PM

All use subject to JSTOR Terms and Conditions
 BOOK :
REVIEWS 467

acter of the enterpriseis emergingfrom the state of latency. Systems

of referentialityhave much in common,and the explanationand ma-
nipulation of referentialityis the frame of referencethat is coming
graduallyinto the open. This, at least, is the perspectiveof some who
have a vision of the desirable,even probable, evolution of content
analysis. One awaits the symposium of 1979 (if not earlier) with
vivid-and heightened-anticipations as a result of this superbcom-
pendium.
-HAROLD D. LASSWELL
Yale University

Communication THE MEDIA AND THE CITIES, edited by Charles U. Daly. Chi-
of Violence: cago: The University of Chicago Press, 1968. 99 pp.
Implications
VIOLENCE AND THE MASS MEDIA, edited by Otto N. Larsen. New
York: Harper and Row, 1968. 310 pp.
Do the media shape or reflectAmerican society? This question has
plagued educators,sociologists, and those engaged in media activi-
ties for many years. These two books make no attempt to answer
this question, but they do review the various facets of the problem
and perhaps add a few new insights.
Each of these volumes is a compendiumof several papers. The
Media and the Cities is an outgrowth of a conferenceheld by the
University of Chicago Center for Policy Study in May, 1968. The
33 participants,including scholars, membersof Congress,represen-
tatives of the media, and others, were asked to review portions of
the Report of the National Commission on Civil Disorders (the
Kerner Commission) and prepareappropriatepapers. As might be
expected, the participantswere divided in their views of society and
of the media.
Violenceand the Mass Mediais a collectionof 31 articles,speeches,
news reports, and congressional reports dating from 1954 to 1966.
An idea of the wide scope of this collection may be gained from the
fact that articles have been taken from publications ranging from
The Annals of the AmericanAcademy of Political and Social Sci-
ences to The LadiesHomeJournal.
The first book concentrateson the effect of the variousmass media
on violence in the United States, naturally giving most of the em-
phasis to civil disorders.While there is evidence of some feeling that
the media may contributeto civil disorder,the consensusappearsto
be that this is not so-that the media merely report.This is not sur-
prising when one remembersthat most of the paperswere prepared
by practicingjournalistsfor eitherprintor electronicmedia.
This does not mean that the writers have shunned responsibility.
In fact, there is a recurringtheme in most of the papersto the effect
that the media have failed to really communicate,especially so far

This content downloaded from 188.72.127.32 on Sun, 22 Jun 2014 21:17:24 PM

All use subject to JSTOR Terms and Conditions