Electronic Journal of Human Sexuality, Volume 5, June 16, 2002


The Application of Computer Content Analysis in Sexology: 
A Case Study of Primary Process Content in Fictional Fetishistic Narratives

Andrew Wilson, Ph.D.


This paper illustrates an application of computer content analysis in sexology. It compares a set of fictional fetishistic narratives published on a web site for rubber boot fetishists (n = 27) with a set of samples taken from general romance and love stories (n = 29).  Using Martindale's Regressive Imagery Dictionary, it is shown that the fetishistic narratives contain a significantly higher proportion of primary process content and a significantly lower proportion of secondary process content than the romance and love stories.  The subcategory of Icarian imagery is the main contributor to this effect.  These findings appear to support previous theoretical views of fetishism as a regressive state and a "destruction of reality".  Further content analysis studies of a wider range of fetishes may facilitate a typological categorization of fetishism.

1. Introduction

Computer content analysis has been a mainstream research technique in the social sciences since around the mid-1960s [1].  Reduced to its bare essentials, it is a methodology that is based on counting the frequencies of entities (normally words and phrases) within texts. However, content analysis is not just another name for word frequency analysis (cf. Leech et al 2001).  Rather than counting individual word occurrences, content analysis normally aggregates these into categories of some kind.  In one variant of content analysis (correlational content analysis), the frequencies of words across a number of texts or text units are subjected to a multivariate analysis (e.g. factor analysis or cluster analysis) in order to extract natural thematic categories from the text (Muskens 1985).  In the other main variant of content analysis (categorial content analysis), the categories are categories of related meaning (rather like the headings in a reference work such as Roget's Thesaurus) into which words are classified.  For example, there might be a category called Clothing into which words such as shoes, skirt, blouse, coat, boots, scarf and trousers are classified; each occurrence of one of these words in the text would increment the Clothing category count by one.  In this variant of content analysis, multivariate analysis is often not employed (although it may be), since the categorization of text words according to shared meanings is already assumed to have extracted the more general themes from the text.

The categories in a categorial content analysis may be either general linguistic categories (such as Clothing, Time, or Color) or they may seek to operationalize a psychological or sociological theory.  Examples of the latter type of category are the categories used in the Regressive Imagery Dictionary (Martindale 1975), which seeks to operationalize theories of regressive cognition as proposed by Freud and others; those used in the Dresden Anxiety Dictionary (Dresdner Angstwörterbuch -- Berth 1998), which is used to count the occurrence of anxiety types such as "death anxiety" or "mutilation anxiety"; and those used in the DOTA dictionary, which  operationalizes a theory of dogmatic thinking (Ertel 1972).

Computer content analysis has found a particularly strong following in psychology and psychiatry, where it has been used, amongst other things, to study psychological states (Viney 1983) and to distinguish between different psychiatric diagnoses (Oxman et al 1988; Gottschalk 1995).  In order to illustrate one of the many potential applications of computer content analysis in sexology, this paper will present a case study of a set of fictional fetishistic narratives.

2. The Case Study: Primary Process Content in Rubber Boot Fetish Stories

2.1 Theoretical Background and Hypotheses

According to Freud (1900), there exist two different modes of thought, known respectively as primary process and secondary process. Primary process thought is sensation-oriented, concrete, and ignorant of time, space, and social institutions. Secondary process thought, in contrast, is logical and oriented to time, space, and society. Similar cognitive distinctions have also been drawn by scholars such as Werner (1948), Piaget (1954) and Goldstein (1939).

In psychoanalytic theory, primary process thought is most closely associated with early childhood, whilst secondary process is the normal conscious mode of cognition for adults: indeed, a decrease in primary process has been demonstrated experimentally in stories written by older versus younger children (West et al 1985). In experimental studies, primary process has also been strongly correlated with altered states of consciousness, such as drug-induced states (Martindale and Fischer 1977; West et al 1983), and it has been shown to be strongly present in the folk tales of more primitive societies (Martindale 1976).

At various stages in the history of psychology, sexual fetishism has been associated with all three of these phenomena: childhood, altered states, and primitive society. The majority of pscyhoanalysts, for example, view fetishism as a regressive state: Chasseguet-Smirgel (1991) is a notable proponent of this view, but it has also been advanced by Abraham (1948) and is implicit in the equation of fetish objects with Winnicott's (1953) childhood transitional objects. Fetishism has sometimes also been associated with altered states, most notably in Mitchell et al's (1953) famous case, in which the fetishist entered a trance-like state when gazing on a safety-pin.  One of the earliest studies of sexual fetishism by Binet (1887) also likened it to a form of primitive religion, a comparison taken up by Comfort (1978), who proposed that fetishistic sexual rituals had much in common with magical and religious rites.

On the basis of  these comparisons, and particularly the widely held theory of regression, we might hypothesize that fetishistic texts are likely to contain a significantly higher proportion of primary process content than other, non-fetishistic texts. In view of Chasseguet-Smirgel's (1991) view of fetishism as a "destruction of reality", we might also hypothesize that a reduced amount of secondary process will be encountered. As primary and secondary process are typically negatively correlated, the measurement of secondary as well as primary process content has frequently been neglected (as, for example, in West and Martindale 1988), but a count of secondary process is arguably a valuable additional statistic: for instance, in a content analytic study of speech under hypnosis, Elter-Nodvin (2000) found no significant difference in primary process between the hypnotic and waking states, but she did find a significant reduction in secondary process in the hypnotic state.

2.2 Data

The data for this investigation comprised 27 English language stories from a main web site for rubber boot fetishists (mean length = 1420 words, maximum = 4353, minimum = 314). These made up the entire English language content of the story section of that web site on the date of collection, with the exception of multi-part stories, which were excluded. The data were collected in the summer of 2001.

As a comparison group, the study used the Romance and Love Story section of the Freiburg-LOB ("FLOB") Corpus of British English (Hundt, Sand & Siemund 1998). This was considered an appropriate comparison, since the majority of the rubber boot stories were relationship oriented: they focussed on social activities in which rubber boots featured strongly, rather than just on the boots themselves. The entire FLOB Corpus is a 1 million word collection of samples of written English that were published in the year 1991; the Romance and Love Story section contains 29 samples taken from novels or short stories (mean length = 2058 words, maximum = 2108, minimum = 2017).

2.3 Method

To identify primary and secondary process thinking, the stories and corpus samples were content analysed using a version of the Regressive Imagery Dictionary (or "RID" -- Martindale 1975) [2]. This is a well validated measure of primary and secondary process thought. It contains two main summary categories -- primary and secondary process -- as well as several further categories, which were not used in the present study. Each of the two summary categories is made up of several subcategories, and, in the case of primary process, these subcategories also contain further subcategories. Table 1 shows the make-up of the primary and secondary process categories.
Main category and primary subcategories Secondary subcategories Example words
Drive Oral breast, drink, lip
Anal sweat, rot, dirty
Sex lover, kiss, naked
Sensation General sensation fair, charm, beauty
Touch touch, thick, stroke
Taste sweet, taste, bitter
Odor breath, perfume, scent
Sound hear, voice, sound
Vision see, light, look
Cold cold, winter, snow
Hard rock, stone, hard
Soft soft, gentle, tender
Perceptual Disinhibition Passivity die, lie, bed
Voyage wander, desert, beyond
Random movement wave, roll, spread
Diffusion shade, shadow, cloud
Chaos wild, crowd, ruin
Regressive Cognition Unknown secret, strange, unknown
Timeless eternal, forever, immortal
Altered consciousness dream, sleep, wake
Brink passage road, wall, door
Narcissism eye, heart, hand
Concreteness at, where, over
Icarian Imagery Ascend rise, fly, throw
Height up, sky, high
Descend fall, drop, sink
Depth down, deep, beneath
Fire sun, fire, flame
Water sea, water, stream
Abstract Thought know, may, thought
Social Behavior say, tell, call
Instrumental Behavior make, find, work
Restraint must, stop, bind
Order simple, measure, array
Temporal Reference when, now, then
Moral Imperative should, right, virtue
Table 1: RID categories and example words (based on Martindale 1975)

The RID was applied to the texts using the PROTAN suite of programs for content analysis (Hogenraad et al 1995). PROTAN first divides the input file into the segments pre-marked by the analyst (in this case, the individual stories and corpus samples). The words in these segments are then reduced by another procedure to their basic, uninflected forms (e.g. loves, loving and loved all become instances of love). The reduced text is then matched against the entries in the RID. For each text segment, PROTAN produces a frequency count of each RID category, which shows how many word occurrences fell into that category. However, because text segment lengths vary, these raw counts are typically not the most appropriate measures to use. To take account of segment length, therefore, PROTAN also offers a frequency rate for each segment, based upon the frequency count. This rate is calculated as follows:

Frequency rate = SQRT [ ( frequency count / segment length ) * 1000 ]

Since textual data often fail to fit a normal distribution, the non-parametric Mann-Whitney U-Test was used to identify significant differences between the two groups. All statistical tests were carried out on PROTAN frequency rates using the procedures in SPSS for Windows version 10.0.7.

2.4 Results

All cited p values are two-tailed.

As predicted, the rubber boot stories showed significantly higher frequency rates of primary process words than did the control group of romance and love stories (U = 169, z = -3.648, p < 0.001); the latter, in turn, had significantly higher rates of secondary process words (U = 128, z = -4.321, p < 0.001). Within the summary category of primary process, only one subcategory -- Icarian Imagery -- was significantly more frequent in the rubber boot stories (U = 32, z = -5.895, p < 0.001).
A further subcategory of primary process -- Perceptual Disinhibition -- proved to be significantly commoner in the control texts, at a lower probability threshold (U = 250, z = -2.320, p < 0.05).

2.5 Discussion

In line with the predictions made in 2.1, this case study has demonstrated significant differences between the fetish-oriented stories written for and/or by rubber boot fetishists and the general sample of romance and love stories.  The rubber boot stories showed a much greater presence of primary process thinking, whilst the romance and love stories showed a greater tendency towards secondary process thinking.

The subcategory of primary process which most strongly discriminated between the two groups was Icarian Imagery.  According to Ogilvie's (1968) classic study of Icarian personalities, Icarians tend to be fixated at a regressive stage: they wish to remain children, are anxious and confused about genital development, and show an underlying fear of women.  That the rubber boot stories contained substantially more Icarian imagery than the romance and love stories might thus be seen to support both Chasseguet-Smirgel's (1991) regression hypothesis and Storr's (1991) suggestion that fetishists "feel themselves to be inadequate as men".

However, we should be careful not to generalize too widely: this finding is based on just one kind of text (fictional narrative) taken from just one variety of fetish (women in rubber boots).  As Chalkley and Powell (1983) showed, fetishism is an extremely varied phenomenon in terms of both the fetish object itself and the activities in which the fetishist seeks to include the object.  Further studies are thus called for, drawing both on different fetishes and on different text types: for example, they might involve interviews with fetishists themselves as well as analysing fictional narratives.  A well controlled series of content analysis studies might ultimately, perhaps, be able to detect typological differences between the different kinds of fetish.

3. General Conclusion

This case study has illustrated the application of one content analysis dictionary -- the Regressive Imagery Dictionary -- to one type of text -- narrative fiction -- within sexology.  The results of this study alone have suggested a more substantial research avenue which may be explored using the same technique.  However, computer content analysis has the potential to be applied much more widely in sexology, wherever there are relatively large quantities of text to be analysed: published material, therapy transcripts, interviews, open-ended questionnaires, etc.  In its various forms, it allows the analyst both to explore the contents of their data and to operationalize and test specific hypotheses about it.


[1] For a general overview of work in computer content analysis, see, for example, Weber (1990) and Roberts (1997).

[2] Version for PROTAN, revision of 11 March 1992 by Robert Hogenraad amd Dan Kaminski. UCSPO, Université catholique de Louvain, Belgium.


Abraham K 1948 Remarks on the psychoanalysis of a case of foot and corset fetishism. In Abraham K, Selected papers on psychoanalysis, Hogarth, London, pp. 125-136.

Berth H 1998  Das Dresdner Angstwörterbuch (DAW). Ein Versuch zur textinhaltsanalytischen Erfassung von Angst. Forschungsbericht Nr. 22 des Instituts für Pädagogische Psychologie und Entwicklungspsychologie der TU Dresden.

Binet A 1887 Fetishisme dans l'amour. Revue Philosophique 24: 143-167, 252-274.

Chalkley A, Powell G 1983  The clinical description of forty-eight cases of sexual fetishism. British Journal of Psychiatry 142: 292-295.

Chasseguet-Smirgel J 1991 Sadomasochism in the perversions: some thoughts on the destruction of reality. Journal of the American Psychoanalytic Association 39(2): 399-415.

Comfort A 1978 Sexual idiosyncrasies: deviation or magic? Journal of Psychiatry 9: 11-16.

Elter-Nodvin E 2000 Computerized content analysis: a comparison of the verbal productions of high hypnotizable, low hypnotizable and simulating subjects. Ph.D. dissertation, University of Tennessee, Knoxville.

Ertel S 1972 Erkenntnis und Dogmatismus. Psychologische Rundschau 23: 241-269.

Goldstein K 1939 The organism. Beacon, Boston.

Gottschalk L 1995 Content analysis of verbal behavior. New findings and clinical applications. Lawrence Erlbaum Associates, Hillsdale, NJ.

Hundt M, Sand A, Siemund R 1998 Manual of information to accompany the Freiburg - LOB Corpus of British English ('FLOB'). Englisches Seminar, Albert-Ludwigs-Universität Freiburg.

Freud S 1900 Die Traumdeutung. Franz Deuticke, Leipzig/Vienna.

Hogenraad R, Daubies C, Bestgen Y 1995 Une théorie et une méthode générale d'analyse textuelle assistée par ordinateur. Le système PROTAN (PROTocol ANalyzer). Version of 2 March 1995. Psychology Department, Catholic University of Louvain, Louvain-la-Neuve.

Leech, G, Rayson P, Wilson A 2001 Word frequencies in written and spoken English: based on the British National Corpus. Longman, London.

Martindale C 1975 Romantic progression: the psychology of literary history. Hemisphere, Washington, DC.

Martindale C 1976 Primitive mentality and the relationship between art and society. Scientific Aesthetics 1: 5-18.

Martindale C, Fischer R 1977 The effects of psilocybin on primary process content in language. Confinia Psychiatrica 20: 195-202.

Mitchell W, Falconer MA, Hill D 1954 Epilepsy with fetishism relieved by temporal lobe lobectomy. Lancet 2: 626-30.

Muskens G 1985  Mathematical analysis of content. Quality and Quantity 19(1): 99-103.

Ogilvie D 1968 The Icarus complex. Psychology Today 2(7): 30-34.

Oxman T, Rosenberg S, Schnurr P, Tucker G 1988  Diagnostic classification through content analysis of patients' speech. American Journal of Psychiatry 145: 464-468.

Piaget J 1954 The construction of reality in the child. Basic Books, New York.

Roberts C (ed) 1997 Text analysis for the social sciences. Methods for drawing statistical inferences from texts and transcripts. Lawrence Erlbaum Associates, Hillsdale, NJ.

Storr A 1991 The psychopathology of fetishism and transvestism. In Samuels A et al (eds), Psychopathology: contemporary Jungian perspectives, The Guilford Press, New York, pp 255-274.

Viney L 1983 Assessment of psychological states through content analysis of verbal communications. Psychological Bulletin 94: 542-563.

Weber R 1990 Basic content analysis. 2nd ed. Sage, Beverly Hills.

West A, Martindale C 1988 Primary process content in paranoid schizophrenic speech. Journal of Genetic Psychology 149: 547-53.

West A, Martindale C, Hines D, Roth W 1983 Marijuana-induced primary process content in the TAT. Journal of Personality Assessment 47: 466-67.

West A, Martindale C, Sutton-Smith B 1985 Age trends in the content of children's spontaneous fantasy narratives. Genetic, Social, and General Psychology Monographs 111: 391-405.

Winnicott D 1953 Transitional objects and transitional phenomena. International Journal of Psychiatry 34: 89-97.

Corrected on 10/16/02

Return to Front Page