Professional Documents
Culture Documents
DOI 10.1007/s11135-005-5077-3
IOANNIS TSOUKALAS
Department of Social Anthropology, Stockholm University, 106 91 Stockholm, Sweden,
E-mail: ioannis@socant.su.se
Abstract. The current article is concerned with the presentation of a novel method for study-
ing social representations. Social representations are a key concept within social science
but, as with many other social phenomena, notoriously difficult to grasp and study in a
systematic way. The survey method presented though a formal one is not dependent on pre-
specified and pre-graded answers as most questionnaires do. Instead it creatively utilizes the
elicitation and elaboration of free associations on a given topic in order to shed light on
the semantic field and cognitive organization of a given social representation. As a neces-
sary complement to such a methodological exposition the article also presents some of the
theoretical background that informs the method as well as the tradition of studying social
representations in general. In the course of the article a research design for the method is
presented accompanied by a complete script of the survey instrument. This is followed by a
discussion of the merits of the method as well as some suggestions of analytical approaches
that can be applied to the results. The article concludes with a short discussion of potential
areas of application beyond the field of social representations.
Key words: analytical suggestions, formal method of study, free associations, innovative
questionnaire, research design, social representations
1. Introduction
The present article is concerned with the issue of how to study social rep-
resentations, a key concern within social psychology and social anthropol-
ogy. In this spirit it hopes to further the methodological options available
to social scientists by proposing a relatively new method for systematically
gathering information on the incidence and character of social represen-
tations within a group or larger population. Since the article is mainly
a methodological exposition the discussion of the theoretical background
and epistemological underpinnings supporting the method will be kept to
a minimum. For those wanting to look deeper into such issues references
to relevant literature will be given.
representations and social facts (see Cuff et al., 1992). More recently this
heritage from classical sociology has been taken over and developed by
a group of French social psychologists working under Serge Moscovici
(Jodelet, 1989; Moscovici, 1989). This tradition also has some counter-
parts within social anthropology where the concept of social representa-
tions has figured in one way or other (Holy and Stuchlik, 1983; Sperber,
1984).
Social representations besides being a respectable field of investigation
in themselves also have affinities with other important theoretical tradi-
tions. In European social psychology for example – as developed within the
Social Identification Theory – social representations have a prominent the-
oretical status and are considered to be closely related to issues of social
identity and group stereotypes (Hogg and Abrams, 1988; Moscovici, 1989;
Billig, 1995). Finally the theory of social representations also has affini-
ties with studies of native classification systems (Lévi-Strauss, 1966; Bloch,
1990; Eriksen, 1995) and cognitive science (Sperber, 1984; Keesing, 1989;
Bloch, 1990; Karmiloff-Smith, 1994; Strauss and Quinn, 1997).
As can be seen from the above presentation social representations are a very
complex phenomenon, formed in the tension field between people’s cogni-
tive abilities and their social interactions, and prone to several interesting
968 IOANNIS TSOUKALAS
Knowledge
science
abstract framework
reified knowledge
construction interpretation
epistemic material
psychological social
collective expression symbolizing human
ideal
Practice
experience
action
embodiment
3. The Method
The method of surveying social representations to be presented below is
based on the use of free associations. The fact that it is based on the use of
free associations means that it is less structured and more open-ended than
most traditional survey methods. For example it does not depend on pre-
specified and pre-graded answers as most questionnaires do. At the same
time however it retains many of the advantages of the questionnaire, like
for example its formal character and the ability to gather large amounts of
quantifiable data.
The subjects are asked to produce a number of associations in response
to a given mother-word. These associations are then successively sorted-out,
contextualized and elaborated through the use of several exercises. The
resulting pattern of associations helps elucidate and pin-point the distribu-
tion of a social representations semantic field, its social construction as well
as its basic features and central core. Getting valid knowledge about these
four aspects of a social representation is important in order to be able to
understand its meaning.
The method although a formal one can easily be used in most set-
tings and is sufficiently non-technical in its construction and application
STUDYING SOCIAL REPRESENTATIONS 969
4. The script
The survey instrument has the following outline (see next page):
Part I
1. Think shortly of what it means to be a social anthropologist. Then write down fifteen (15) words or short
sentences that come to mind with respect to being an anthropologist. It is preferable that you produce words;
use sentences only if you have to. The words could stand for thoughts, feelings, behaviours or anything else.
Formulate yourself in a clear way so that it will be easy for a future reader to understand what you mean
(make it easy to separate positive from negative evaluations, subject from object, present from future tense
etc.). Don’t think about it to much; just write down whatever words pop-up in your head.
2. Look for a moment at the words you have produced in the previous section and try to figure out which
ones best capture the meaning of the mother-word. Which are most representative of the meaning of the
mother-word? Select three such words. Transfer those three words to this page and write them down under
the mother-word.
Social anthropologist
——– —————— ——–
3. Couple each of the above words with their mother-word to create three pairs. Think shortly about the
meaning of these pairs. Then write down three words or short sentences that come to mind in association
with each of the pairs. The words could stand for thoughts, feelings, behaviours or anything else. They
could be new words or the same as before.
Pair one: Pair two: etc . . .
Associated thoughts: Associated thoughts:
4. Couple each of the above words with their mother-pair to create nine triplets. Think shortly about
the meaning of these triplets. Then write down three words or short sentences that come to mind in
association with each of the triplets. The words could stand for thoughts, feelings, behaviours or anything else.
Triplet one: Triplet two: etc . . .
Associated thoughts: Associated thoughts:
Part II
1. Go back to the previous sections and gather all the associations (single words) you have produced under
exercise 2, 3, and 4 and transfer them to this page. Use the copy and paste function of your computer
for convenience. They should amount to a total of 39 words. Select 32 of them by discarding seven words
that you find less important.
Look at the 32 words and think about them for a while. Now try to group them in a way that feels
meaningful for you. Place all the words that you think are related in the same group. You can create as
many or as few groups as you want as long as they are more than one. You don’t have to use all 32
words; leave some words out if you like.
Selected words (32): Group 1: Group 2: etc . . .
2. Now that you have placed the words into different groups describe in a few sentences what considerations
lead you to group them in this particular way. Try to explain in what way the words are related to each
other and to the initial mother-word (“social anthropologist”). You should offer one explanation for every
group you have made. Formulate yourself in a clear way so that it will be easy for a future reader to
understand what you mean.
3. Finally give a suiting name/heading to every group.
Part III
1. Go back to Part II (page x) and transfer the 32 words you selected to this page. Use the copy and
paste function of your computer for convenience. Divide the words into two equal groups (16 words in
each): one for those words that are most characteristic of the mother-word (“social anthropologist”) and
one for those that are least characteristic. Take the group consisting of the most characteristic words and
repeat the procedure. Divide it into two new equal groups (8 words in each): one for those words that are
most characteristic of the mother-word and one for those that are least characteristic. Repeat the procedure
until you just have one word remaining in the “most characteristic group”.
Selected words (32): First round: Second round: etc . . .
Most characteristic words: Most characteristic words:
Least characteristic words: Least characteristic words:
2. Go back to the previous part and gather all the associations you have assembled into groups and transfer
them to this page. Write them down in the form of a list for every group. If you had three groups in the
previous part then you should make three lists of words, if seven groups then seven lists etc. The order of
the words does not matter.
Now think of the mother-word we used in the beginning, the word from which all the above associa-
tions (single words) originate from. Think of the meaning of this word. What does it mean to be a social
anthropologist? Then look at the lists of words which you have produced. These words are all derived from
the mother-word by association. They are part of what it means to be a social anthropologist.
Look at the first word in the list. What happens if you take it away? Is it still possible to be a
social anthropologist without it? Is the idea of an anthropologist viable without it? If it is possible to be
an anthropologist without it put the word in the non-essential category (copy it from the list and paste it
there). If the word is a necessary condition for being an anthropologist put it in the essential category. In
other words you should try to figure out which words are an integral part of being a social anthropologist.
Repeat this procedure with the other words in the list. You should end up with just a few essential words;
the ideal would be to have just one essential word per list. Repeat this procedure for the other lists as well.
List 1: List 2: etc . . . .
Non-essential word(s) of list 1: Non-essential word(s) of list 2:
Essential word(s) of list 1: Essential word(s) of list 2:
STUDYING SOCIAL REPRESENTATIONS 973
does not differ substantially from those to be expected when conducting survey
investigations in general they will not be given a special treatment here. The
interested reader could look into a standard methodology textbook for more
information on this issue (see for example Dane, 1990).
More unique problems of the Representational Survey have to do with
its special use of free associations. The data collected through the Repre-
sentational Survey might for example give rise to several semantic ambi-
guities. For one thing the plentitude of synonyms in most languages may
result in respondents denoting (or connoting) the same thing but by differ-
ent lexical means, by generating different words. How could such a situ-
ation be remedied? The easiest solution is to decide on a standardization
of recurring denotations (or connotations) by way of a book of synonyms.
Thus for example the words “brother” and “sister” could be standardized
into the synonym word of “sibling”.
Another semantic difficulty that might occur is the use of sentences
instead of words. The method allows this in order not to interrupt the
spontaneous meaning making of the respondents but of course prefers
words since they are easier to handle analytically. In this case a possible
solution is to nominalize the verbs supporting the meaning of a sentence.
Thus for example the sentence “they cared for the weak” could be nomi-
nalized into “caring” or “protection” depending on what the context (both
of the collected associations and of the respondent’s life situation) suggests.
Of course nouns thus produced can in case of semantic overlap, different
words denoting or connoting the same thing, be standardized by the use of
synonyms as indicated above.
Finally one ought to mention that although care is taken so as to semanti-
cally contextualize the associations produced by the respondents, via a whole
series of successive exercises and elaborations, the output data of the question-
naire are still mostly words and words, even in relatively good company, can be
misleading and ultimately meaningless. Thus additional care must be taken to
contextualize the results of the questionnaire, both for the individual respon-
dent, for example via interviews, and the group he belongs to, via for example
focus groups or participant observation.
Overall the questionnaire gives rise to several interpretative possibilities,
something which is not necessarily a weakness. The possible fuzziness of
some answers can most of the time be remedied thanks to the inbuilt logic
of the questionnaire which allows several ways to trace and even recon-
struct the original meaning of the responses.
6. Analytical Considerations
The Representational Survey can be analyzed in several ways and all its
parts are capable of giving useful information. The words produced from
976 IOANNIS TSOUKALAS
the exercises of Part I and listed in Part II (32 words) provide the most
general and inclusive overview of the semantic field of a social representa-
tion. The information in Part I is thus well suited for getting a broad pic-
ture of the ideational variation exhibited by the whole group.
The words to be found in the “groups” of Part II and the “rankings” of Part
III provide a more personal and exclusive rendition of the key meanings inher-
ent in the semantic field of a social representation. Part II and Part III thus
offer the most condensed and focused information in the survey. These are the
parts where some of the randomness, the unavoidable “noise”, of the responses
is worked through and sorted out, leaving a more systematized and thus more
comprehensible result. The outcome of these parts could be considered as the
conceptual backbone of the respondent’s opinion about a subject, the central
meaning of the social representation in question.
The results from the survey can be analyzed in several ways. The sim-
plest quantitative analysis would be to calculate the average frequency and
rank of the words for the given sample. Frequency and rank can further
be compared to each other in order to ascertain if there is any system-
atic correlation between them and whether that correlation is significant.
The frequency of the words can also be compared to their incidence in any
potential interviews so as to corroborate their importance. The statistical
correlations that can be made between various data of the Representational
Survey are several. To name a few
• frequency in the RS to ranking in the RS
• frequency in the RS to central core in the RS
• frequency in RS1 and RS2 to central core in RS1 and RS2
• frequency in the RS1 to frequency in RS2
• ranking in RS1 to ranking in RS2
• central core in the RS1 to central core in RS2
Commonalities in the frequency, ranking and central core of a social represen-
tation between respondents are indicative of a “stream-lining” (social construc-
tion) due to group processes. For this reason a Representational Survey should
not be conducted early on in the life-cycle of a group. Time should be allowed
to pass in order for the potential social construction to take place.
Also more advanced statistical methods could be used, like for example
cluster analysis, but in that case the researcher must take care in the scaling
and scale transformations of the received data so as not to make unjusti-
fiable operations or invalid inferences. The survey instrument as it is gen-
erates mostly nominal and ordinal data and is therefore not well suited for
advanced statistical processing. Still the yield of data that it gives is suffi-
ciently rich to allow such methods, if they are deemed necessary.
The material generated by the survey can also be analyzed by more
qualitative techniques. One such approach would be to use the various
STUDYING SOCIAL REPRESENTATIONS 977
7. Suggested Uses
Besides its obvious uses within the established tradition of studying social
representations the Representational Survey presented above can also be
adapted to other purposes and other traditions within the social sciences.
One such obvious candidate is the study of memory. Especially complex
memories that have taken on a “public life” and circulate widely within
a population (e.g., collective traumas) could easily be studied with this
method. The fact that these kinds of memories are moulded through
various social processes makes it very interesting to study their social con-
struction, selection and dissemination performed by a population and its
subgroups. Generally collective memories stand very close to the concept
of a social representation as defined above and should thus not present any
special difficulties for the Representational Survey.
Another candidate for study by the Representational Survey is native
classification systems. The fact that classificatory concepts have clear affin-
ities with social representations (see relevant footnote) makes them amena-
ble to study by this method. In cases where the degree of literacy is low
in the studied population or familiarity with formal assessment techniques
is lacking the Representational Survey could be creatively adapted to suit
the local conditions. It is easy for example to change it from a paper-and-
pencil exercise to a more audiovisual method for gathering information. In
the later case one could use pictures, objects or just oral accounts in order
to get to the semantic network underlying a social representation.
A third candidate that could benefit from using this method is semi-
otics and discourse analysis. The Representational Survey could for exam-
ple be used to conduct media production and audience reception studies in
various communicative sectors (newspapers, literature, TV-programs etc.) as
well as looking into the important question of how these two are related.
In this purpose the Representational Survey could help elucidate both the
coding and the decoding of a message and showcase how the socially con-
structed and contextually specific arrangement of its semiotic components,
its denotations and connotations, are affected by successive disseminations
and filterings, from production to consumption, from arrangement to re-
arrangement and vice versa (on the coding-decoding hypothesis see Hall,
1999). The method could thus add interpretative depth and formal strin-
gency to discourse analysis as it is practised in the social sciences.
978 IOANNIS TSOUKALAS
8. Concluding Comments
The method that has been presented above offers new options to the
researcher interested in the systematic study of social representations. It
can be used separately or in combination with other methods customarily
used in this field and in any case offers several advantages. The first and
most important one becomes obvious when considering the form and func-
tion of more standard questionnaires. Standard questionnaires are usually
dependent on very complicated methodological assumptions, construction
principles and analytical procedures, often perplexingly so, and this creates
an unnecessary distancing from ordinary experience and understanding and
makes their use a cumbersome and parochial business. In contrast the pres-
ent questionnaire is crafted in such a way as to stay close to the intui-
tive and common-sense meaning of a social representation – both when it
comes to its findings and the analytical options they suggest – while simul-
taneously being open to a multitude of more systematic data analysis meth-
ods. This is an unmistaken advantage which makes the questionnaire more
transparent and its use more meaningful, for respondent and researcher
alike.
Another advantage of the method is that the use of free associations instead
of pre-given questions and pre-specified answers offers more freedom and ini-
tiative to the respondents to express their own opinions on the subject matter.
Still the successive exercises performed on their first round of responses offers
the researcher the possibility to get under the surface of their answers thereby
STUDYING SOCIAL REPRESENTATIONS 979
getting a thorough picture of their underlying conceptual world. Thus the open
ended character of the questionnaire does not impinge negatively on the need
for a systematic approach and the collection of formal data.
Finally the method offers a high degree of flexibility to the researcher that
chooses to use it. This is also a clear advantage. The various parts of the ques-
tionnaire can for example be used and evaluated both together and separately
depending on what the situation calls for. Further the results of the method can
be subjected to both a qualitative and a quantitative analyses and evaluation,
again depending on what the situation calls for. It is thus up to the particular
researcher to choose what suits his needs or purposes better.
The method presented above has been preliminary tested in a pilot study
with encouraging results. This together with the fact that it has well documented
antecedents in the scientific literature (see Abric, 1994) as well as obvious com-
parative advantages makes it a good candidate for any researcher wanting to
supplement his methodological options. In the near future the method is going
to be tested within a larger interdisciplinary research project and this “larger
run” will hopefully reveal and correct any shortcomings or misfiring hitherto
undetected. In the meantime the method is deemed sufficiently interesting and
safe to be worthy of announcement and a peer review.
Notes
1. This is how Holy and Stuchlik define a group:
‘group’ usually refers to a plurality of individuals bounded by some principle(s) of
recruitment and by a set of membership rights and obligations . . . Everybody fulfill-
ing the recruitment criteria is a member of the group and every group member auto-
matically has the rights and discharges the obligations characterizing membership . . .
Interactions of people are seen, not as those of individuals, but as those of group mem-
bers, i.e. occupants of specific statuses. Only when a particular individual does not fulfil
his membership obligations does his action constitute a problem, a discrepancy to be
explained on contingent grounds.
(Holy and Stuchlik, 1983: 111)
2. A less enumerative and more operational definition of culture is offered by Maurice Bloch.
According to his view culture “can be defined as that which needs to be known in order to
operate reasonably effectively in a specific human environment” (Bloch, 1990: 183).
3. An alternative, more technical, definition of a representation is offered by Dan Sperber
(1984: 76): ‘A representation involves a relationship between three terms: an object is a
representation of something, for some information processing device’. Thus according to
this view a specific digital series of 0 and 1 – corresponding to a sequence of electrical
currents in an integrated circuit – is a representation of a programming operation for
the processing device called a computer. Similarly closing our eyes and imagining a car
– which presumably corresponds to a specific pattern of neuronal activity – produces a
representation of a car outside our head appropriate for the processing device called a
human mind.
4. By the same process a technical definition of a cultural institution can be given in repre-
sentational terms:
980 IOANNIS TSOUKALAS
Some sets of representations include representations of the way in which the set should be
distributed. An institution is the distribution of a set of representations which is governed by
representations belonging to the set itself. This is what makes institutions self-perpetuating.
Hence to study institutions is to study a particular type of distribution of representations.
This study falls squarely within the scope of an epidemiology of representations.
(Sperber, 1984: 87)
5. Here we are not using the term synesthesia in its strictly technical sense, meaning inher-
ently multimodal representations, a rare occurrence even within the field of neurology, but
in a looser sense, meaning conditionally multimodal representations. Synesthesias of the
later case are commonplace and are more akin to a spatiotemporal association of sensory
input through social conditioning, intentional or accidental.
6. As the perceptive reader might have noticed social representations resemble two other
analytical concepts currently used in social science. The first one is that ofschemata or
scripts used in cognitive science. A schema is defined as a “chunked network of loose pro-
cedures and understandings which enable us to deal with standard and recurring situa-
tions, for example ‘getting the breakfast ready’, that are clearly culturally created” (Bloch,
1990: 185). However the two concepts also differ in important respects. Thus schemata
are in general applied on the micro and intermediate level whereas social representations
are generally more relevant to the macro level; schemata are more individual and con-
text specific whereas social representation are by definition widely shared and distributed;
schemata fall back on and are used in connection with issues of learning, memory and
expert systems whereas social representations are more closely affiliated with issues of
power, ideology and discourse; schemata are informed by cognitive theory whereas social
representations by social psychology; schemata are more of a technical term whereas
social representations are more familiar to the general population. For more details see
Anderson (1990), Strauss and Quinn (1997), Jodelet (1989), and Bloch (1990).
The second is that of classificatory concepts used within anthropology and ethnology.
Classification “entails dividing objects, people, animals and other phenomena according
to socially pre-established categories or types” (Eriksen, 1995: 220). Here too there are
differences between the two concepts. Thus classificatory concepts are in general under-
stood as verbal and propositional entities whereas social representations are generally
more visual and imaginary in their character; classificatory concepts are seen as more
inert and sacrosanct whereas social representations are more labile and transmutable;
classificatory concepts are more hierarchically organized whereas social representations
are organized in loose networks. For more details see Lévi-Strauss (1966), Jodelet (1989),
Keesing (1989), and Bloch (1990).
7. The central core can be thought of as the element or combination of elements that is
peculiar to the social representation. In other words the part of a social representations
semantic content that is not shared with other similar representations. The essentialism
implied in the term “central core” is not a literal one but a socially constructed one. So
anti-essentialists, do not panic . . .
8. If random samples cannot be taken the participants in the representational survey should
be matched (in gender, nationality, class etc.) to the total population.
References
Abric, J.-C. (ed.) (1994). Méthodologie de recueil des représentations sociales, Pratiques so-
ciales et représentations. Paris: Presses Universitaires de France.
STUDYING SOCIAL REPRESENTATIONS 981
Anderson, J. R. (1990). Cognitive Psychology and its Implications, 3rd edn. New York: W.H
Freeman and Company.
Bignell, J. (1997). Media Semiotics:An Introduction. Manchester University Press.
Billig, M. (1995). Banal Nationalism. London: Sage.
Bloch, M. ( 1990). Language, anthropology and cognitive science. Man (N.S.) 26: 183–198.
Bourdieu, P. (1990). The Logic of Practice. Cambridge: Polity Press.
Cuff, E. C., Sharrock, W. W. and Francis. D. W. (1992). Perspectives in Sociology, 3rd edn.
London: Routledge.
Damasio, A. (1999). The Feeling of What Happens. Body and Emotion in the Making of Con-
sciousness. New York: Harcourt Brace and Company.
Dane, F. C. (1990). Research Methods. California: Brooks/Cole Publishing Company.
Depraz, N. (1999). The phenomenological reduction as praxis. In: F. Varela and J. Shear (eds.) The
View from Within: First-person Approaches to the Study of Consciousness. Imprint Academic.
Eriksen, H. E. (1995). Small Places, Large Issues. An Introduction to Social and Cultural
Anthropology. London: Pluto Press.
Geertz, C. (1993[1973]a). The impact of the concept of culture on the concept of man. In
The Interpretation of Cultures. London: Fontana Press.
Geertz, C. (1993[1973]b). The growth of culture and the evolution of mind. In: The Inter-
pretation of Cultures. London: Fontana Press.
Hall, S. (1999). Encoding, decoding. In: S. During (ed.), The Cultural Studies Reader.
London: Routledge.
Holy, L. and Stuchlik, M. (1983). Actions, Norms and Representations. Foundations of
Anthropological Inquiry. Cambridge: Cambridge University Press.
Hogg, M. A. and Abrams, D. (1988). Social Identifications. A Social Psychology of Inter-
group Relations and Group Processes. London: Routledge.
Jodelet, D. (1989). Les représentations sociales: un domaine en expansion. In: D. Jodelet
(ed.), Les représentations sociales. Paris: PUF, pp. 30–61.
Karlsson, G. (1993). Psychological Qualitative Research from a Phenomenological Perspective.
Stockholm: Almqvist & Wiksell International.
Keesing, R. M. (1989). Exotic readings of cultural texts. Current Anthropology 30(4): 459–479.
Karmiloff-Smith, A. (1994). Précis of beyond modularity: A developmental perspective on
cognitive science. Behavioral and Brain Sciences 17: 693–745.
Kvale, S. (1996). InterViews. An Introduction to Qualitative Research Interviewing. Sage Pub-
lications.
Lévi-Strauss, C. (1966). The Savage Mind. The University of Chicago Press.
Moscovici, S. (1989). Des représentations collective aux représentations sociales: éléments pour
une histoire. In: D. Jodelet (ed.), Les représentations sociales. Paris: PUF, pp. 62–86.
O’Connor, J. and Seymour J. (1993). Introducing NLP: Psychological Skills for Understand-
ing and Influencing People. London: Thorsons.
Skinner, B. F. (1971). Beyond Freedom and Dignity. New York: Alfred A. Knopf.
Sperber, D. (1984). Anthropology and psychology: towards an epidemiology of representa-
tion. Man (N. S.) 20: 73–89.
Strauss, C. and Quinn, N. (1997). A Cognitive Theory of Cultural Meaning. Cambridge:
Cambridge University Press.
Tomasello, M., Kruger, A. C. and Ratner, H. H. (1993). Cultural learning. Behavioral and
Brain Sciences 16: 495–552.
Tomasello, M. (1999). The human adaptation for culture. Annual Review of Anthropology 28:
509–529.
Varela, F. J., Thompson, E. and Rosch, E. (1991). The Embodied Mind. Cognitive Science
and Human Experience. Cambridge: The MIT Press.