Chapter 3

Methodologies

Abstract

From the outset, this investigation was intended as an interdisciplinary approach to a field that has previously only been considered in isolated units such as the consideration of media content without any reference to audience interpretation and vice-versa. In order to contribute to an understanding of how young children might interpret media texts, it was vital to approach the question of ‘negotiation’ or ‘meaning making’ from varying perspectives. In this way, the identifiable points of the theoretical triangle Text-Producer-Receiver might be consolidated to offer an empirically strong representation of the processes involved in building gender identities within the commercial marketplace. Therefore, a combination of distinct yet related methodologies were selected based on the two broad categories of ‘text’ and ‘receiver’ or ‘audience’, where the former utilised content and semiotic analysis (cf. Research Issues 2-3), and the latter included ethnography, discourse analysis and ‘art’ (cf. Research Issue 5).

At its most basic level, the framework of this investigation can be reduced to the two major components of ‘text’ and ‘audience’. It was therefore important to make these components the central focus when selecting the most suitable research methodologies. I found that the selection process was facilitated by constant reference to the research issues identified in Chapter One to ensure that the chosen methods would actually meet the demands of the investigation (cf. Goodwin & Whannel, 1990: 62-70). Hence, in terms of ‘text’, content and semiotic approaches were considered the most suitable forms of analysis, while ethnographic-style observation, discourse analysis and ‘art’ were selected as the best ways to study ‘receiver’ or ‘audience’. These approaches will each be considered in turn and related back to the overriding research issues.

3.1 Textual analysis

The deconstruction of a (media) text can be approached in many ways and the process can often begin with broad commentary before considering textual components in greater detail. The nature of the commentary might be based on a number of points of reference such as the modes of address, forms of presentation, points of identification, preferred meanings and so on (see Goodwin & Whannel, 1990: 62-70). Indeed, the analysis of a text can draw on any number of disciplines such as semiotics, linguistics, psychoanalysis and anthropology (Allen, 1992). Perhaps the most important point to note is that forms of textual analysis can reveal the underlying meanings within (media) texts that are unobtrusive and likely to be only subconsciously recognised by the ‘reader’ (Livingstone, 1998: 33). Effective textual analysis should therefore seek to draw out and make explicit these underlying patterns and structures. The basis for selecting content and semiotic analyses was two specific research issues concerning (textual) framework and possible (textual) meanings. That is to say, content analysis – a quantitative approach – was considered the most effective way to determine whether or not the construction of the (advertisement) texts in this investigation comprised distinct gendered patterns (Research Issue 2). Semiotic analysis – a qualitative approach – was considered appropriate in determining whether the construction of media images might have gender connotations (Research Issue 3).

3.1.1 Content analysis

In order to conduct a detailed study of a sample of toy advertisements aimed at children, I needed to find a way to investigate and quantify the structural subtleties that often exist within such texts. The main aim was to consider any patterns in terms of the production techniques and the overall style of an advertisement text. After considering many different methodologies, content analysis seemed the most appropriate approach. It is an analytical approach based on the coding and quantification of various elements in any kind of text (including advertisements) in any medium. It involves establishing categories and counting the number of instances when these categories are utilised within a text (Silverman, 1993: 59). At the most basic level, the main purpose of the method is to locate the nature of the relative patterns within and between sets of data. It was exactly this feature that made the method applicable to the identification of possible gendered patterns in the advertisement sample used in this investigation (cf. Research Issue 2). Weber (1985: 9) described the method as a means of utilising a set of procedures to make valid inferences from a text. He also suggested that content analysis could be used to describe trends in communication content including ‘sales communications’ in advertisements.

The underlying assumption of this method is that there is a relationship between the frequency with which certain items appear in a given text and the audience response (Hart, 1991: 108). In this investigation, therefore, there may be a correlation between the sales intentions of the advertisers (to sell) and the eventual responses from the target (child) audience (to buy). On a more cautionary note, the significance of textual elements is not always synonymous with frequency. Indeed, significance cannot always be reduced to manifest objective content. Yet it is arguable that one must understand the structure of a toy advertisement before one can begin to speculate about how the texts might be received and responded to by children. Hart (ibid.) further suggested that the ‘hidden meaning’ of a text, as intended by the producers, can be revealed by identifying and counting any significant textual features. It is for this reason that the method is generally confined to large-scale and systematic surveys of manifest content (Fiske & Hartley, 1978: 21). In essence, the counting of content items provides a sound framework or foundation for later studies of interpretation (see Leiss et al., 1990: 218), and in this study it is used as the starting point and main reference for the others methods of investigation. 

Weber (1985: 12) explained that content analysis is designed to take many different textual elements and classify them in as few categories as possible, so facilitating the development of a skeletal framework. It is also thought to reveal features that are not always obvious through impressionistic readings of texts (Hart, 1991: 111). When analysing the content of the toy advertisements in this sample, it was necessary to reduce the initial list of production and other features into more contained categories and sub-categories. This ‘funnel approach’ was not considered limiting but helped to organise the data into a strong frame of emergent patterns. It is arguable that the stronger the data, the more representative of a typical sample the emerging patterns become. Needless to say, a very detailed list would have been unworkable during an analysis requiring such attention to detail (see Leiss et al., 1990: 220).

One of the main problems with using content analysis is that of validity. It is vital that the classification procedure is reliable in the sense of being consistent. As McQuail (1994: 277) observed the usual practice of constructing a category system involves the risk of an investigator imposing his or her own meaning-system rather than ‘taking’ it from the content (cf. Weber, 1985: 15). In content analysis studies, it is often desirable for multiple coders to set about the task of negotiating categories and quantifying the features present within a given text (Leiss et al., 1990: 221). Unfortunately, since this particular investigation was largely unfunded I had to undertake the coding task myself. However, in an effort to mimic a multiple-coder scenario the sample advertisements were re-coded annually throughout the three-year study. Minor variations did emerge and were rectified. In addition, those counts that emerged as most significant were recounted on a fourth occasion to minimise the chance of error.

Multiple coders were, however, used to categorise the target audience for each advertisement. It is arguable that the perception of target audience would be more likely to vary between coders than counting the presence or absence of production features. Target audience might also be a stronger determinant of advertisement appeal than the integral content elements. Consequently, I hoped to achieve an acceptable level of validity in the toy advertisement analysis where any unconscious degree of personal bias would be counterbalanced by the views of the target-audience coders.

One should also be aware that content analysis is often criticised as a research methodology because its does not move beyond the confines of a given text. Content analysis is not, however, concerned with questions of quality, response or interpretation (Fiske & Hartley, 1978: 21/36). It is arguable that one should not expect such an analytic methodology to yield any qualitative data and it often seems a little absurd that this is sometimes expected. Content analysis is most appropriate for building theoretical frameworks but it should be stressed that other methodologies are more appropriate for the consideration of response and interpretation, such as ethnography or semiotic analysis. Here, the reader is encouraged to regard the content study as a preliminary to other analyses and discourses, and should avoid taking the findings as representative of ‘true meaning’ (cf. Hart, 1991: 111).

3.1.2 Semiotic analysis

While content analysis is effective in constructing a framework of textual ‘features’, no reference is made to the way they might be interpreted by audiences. Perhaps the notion of interpretation can preliminarily be outlined by analysing given media texts semiotically because such a method can demonstrate how the counted elements relate to one another within a narrative sequence and the ‘meanings’ they might connote to the audience. The findings of such an analysis might indeed identify some of the gender connotations within an advertisement text and can be used as a basis to suggest how the audience might interpret them (cf. Research Issue 3). Williamson (1978) produced one of the classic examples of the semiotic deconstruction of advertisements, describing the genre as ‘providing a structure which is capable of transforming the language of objects to that of people and vice-versa’ (ibid.: 12) carrying with it powerful symbolic and ideological ideas. Semiotics is burdened with heavy jargon so, for ease of reference, explanations of the terms used in this investigation are given before the actual analysis (cf. Chandler, 1994a; Thwaites et al., 1994). The intention here is to raise some of the general issues that one should consider when undertaking such an analysis.

Leiss et al. (1990: 198) argued that semiotics can be used to study many kinds of social phenomena and has been compared with structuralism (Hawkes, 1977: 124). It is traditionally defined as the ‘science of signs’ (Fiske & Hartley, 1978: 37), where ‘signs’ refer to anything that has ‘meaning’ or can communicate messages to people (Leiss et al., 1990: 200). The most important consideration of signs is how they are built together into organised structures of codes, relating to one another in certain perceptible ways (cf. Berger & Luckmann, 1966: 51). In an advertisement, for example, signs might exist and mean something within the confines of the textual structure but are also likely to relate to wider belief systems in society (Leiss et al., ibid.: 201). Before meaning is generated, the individual audience members will generally have to ‘work’ to understand the connotative values of the signs.

Bignall (1997: 141) argued that the semiotic analysis of any medium begins with the identification of the signs involved. Berger (1991: 26) posited that, when applying semiotics to television, focus should be placed on those aspects that function as ‘signs’. He referred specifically to camera techniques and argued that each production and post-production feature can function as a signifier such as a close-up signifying intimacy, tilt down signifying power and authority, and tilt up signifying smallness and weakness. Signifiers were also said to include lighting, colour and audio features (ibid.) (cf. Bignall, 1997: 142; Gunter & McAleer, 1997: 139; Hecker & Stewart, 1988: 225; Fiske & Hartley, 1978: 45; Winick et al., 1973: 37). Each of these features was identified during the content analysis of the sample of toy advertisements so one can begin to understand how they might interrelate to relay given messages to the audience. Berger (ibid.) argued that these interpretations can be defined as the ‘grammar’ of television (cf. Kress & van Leeuwen, 1996), and one could argue that children must learn to ‘read’ this ‘grammar’ in order to understand what they are seeing.

Fiske & Hartley (1978: 59) presented a simplified explanation of codes, defining them as ‘vertical’ sets of signs (paradigms) that can be combined according to certain ‘horizontal’ rules (syntagms). Codes must also be broadly agreed upon by users and are seen to continually evolve to meet changing needs (ibid.). Berger (1991: 23) defined codes as highly complex patterns of associations that are grounded in a given society, although the role they play in daily life is often imperceptible, where a formal semiotic analysis might reveal these underlying structures of meaning. The signs that one classifies as belonging to particular codes become more complicated when one considers that they can also belong in other codes that overlap and interrelate into what Fiske & Hartley (1978: 64) term a ‘network of signification’. Additionally, individual codes also operate on a number of different levels within a hierarchy (ibid.). Fiske & Hartley (ibid.) illustrate this notion by referring to voiceovers which are commonly associated with news and documentary programmes but are often used elsewhere (particularly in advertising) to create a sense of ‘reality’ and ‘believability’. This is arguably akin to the notion of intertextuality within and across television genres (cf. Chandler, 1994a).

Leiss et al. (1990: 215) identify gender as one of the most important forms of codified behaviour in all societies, where cultures have accepted ‘routine’ forms of communicating gender identities. One can therefore argue that if gender codes are identifiable within a sample of children’s toy advertisements then such ‘routine’ forms of identity are being subtly communicated to the target audience. By identifying these codes and considering the various connotations, one can arguably gain insight not only into how commercial and social worlds are constructed in terms of gender, but also how the audiences might ‘read’ such connotations.

In terms of methodological weaknesses within the study of advertisements, semiotics is criticised for being too heavily dependent upon the skills of the individual analyst (Leiss et al., 1990: 214). Hence, the conclusions drawn are dubious in terms of validity, consistency and reliability. Semiotic analysis cannot be quantified either because one can rarely generalise from a large sample of individually analysed texts (ibid.). In the context of this study, however, an extensive content analysis provided the quantifiable patterns, while semiotic analysis was applied in order to indicate some of the possible structures of meaning that the advertisement audiences might decipher when viewing a given text. A final criticism focuses on the fact that semiotics is not always easily applied to an analysis of all advertisement types (ibid.). In my own experience, I found the analysis of static magazine advertisements easier to formulate than an analysis of moving televisual images (cf. Griffiths, 1996).

One of the main strengths of semiotic analysis as a research methodology, however, is that it is able to dissect and examine cultural codes in a way that is sensitive to the many interpretational subtleties that exist within cultural systems (Leiss et al., 1990: 214). The main aims of semiotic analysis can be clarified when one considers the way in which Propp (1928/68) perceived the narrative structure of fairytales. Despite the many details and characters involved in the genre, Propp (1968: 20, in Silverman: 1993: 73) suggested that ‘the number of functions is extremely small’, within a basic framework of ‘surface’ and ‘underlying’ systems. This was certainly an aim in my own semiotic analysis in that I wished to reduce the overwhelming details identified in the structure of the advertisements to the fewest possible denominators. This would hopefully reveal the underlying meaning systems within the advertisements aimed at boys and girls respectively – a gendered formula reused by advertisers in accordance with the product being advertised.

3.2 Audience analysis

While content and semiotic approaches are effective ways to construct theories and frameworks about (media) texts, they are not designed as a way to identify how the audience might receive the texts. Despite the fact that semiotic analysis has interpretative possibilities it is still confined to the text and is often subjective. Consequently, before being able to offer observations concerning the children’s understandings of and responses to the sample advertisements (Research Issue 5), suitable methods of audience study needed to be selected. It is possible to divide audience studies into two sub-categories. Firstly, the researcher is often required to generate some kind of audience response and, in the context of this investigation, field observations and interviews were considered the best way to facilitate this process (cf. Fiske, 1987, in Nightingale, 1996: 23). Secondly, there is a need to analytically deconstruct the audience responses in a meaningful way, and discourse analysis was considered appropriate for this study. These two sub-categories will now be discussed.

3.2.1 Field observations and interviews

The method of field observations and interviews was considered appropriate in the context of this investigation because the researcher’s aim was to spend time with the children in a local primary school, noting their uses and negotiations of television in both formal and informal research contexts. To this end, it was useful to frame the audience research methods in terms of the main underlying principles of ethnography, and the eventual practices employed in this study can be loosely described as ethnographic-style.

Ethnography has been defined in a number of ways, many of which seem subtly contradictory. Hammersley & Atkinson (1983: 1) note that there is disagreement as to whether the methodology is designed to elicit cultural knowledge (Spradley, 1980: ibid.), investigate patterns of social interaction (Gumperz, 1981: ibid.) or analyse societies as a whole (Lutz, 1981: ibid.). It is sometimes perceived as storytelling (Walker, 1981: ibid.) or even as a means of developing and testing formal theories (Glaser & Strauss, 1967; Denzin, 1978: ibid.). The method has also been described as a way of understanding the organisation of society in terms of speech, actions and context (cf. Silverman, 1993: 60; Agar, 1986: 12) or ‘mediating frames of meaning’ (Giddens, 1976, in Agar, 1986: 19). Lull (1990: 31) also stressed that ethnography can be used as a means of understanding social groups (children), their patterns of interpersonal communication (friendships, rivalries, peer culture, family) and their uses of the mass media (television, advertising). These perspectives are perhaps indicative of the diversity and flexibility of the methodology to suit the intentions of the researcher, which is indeed an advantage for those who wish to tailor-make an approach that they deem most appropriate.

One of the advantages of the relative flexibility of the method is that no pre-testing of subjects is required and the direction of research can shift in accordance with the subjects’ responses (Hammersley & Aitkinson, 1983: 24). This relative freedom might once again promote theory development since the researcher can ‘go with the flow’, working economically and instinctively. Most researchers tend to favour a semi-structured reflexive approach when working with subjects, not asking any specific questions but referring to a set of ‘themes’ that they would like their subjects to consider (cf. Hammersley & Atkinson 1983: 112-3; Agar, 1986: 19). Stubbs (1983: 225) argued that a quest to find what he calls ‘pure, natural or authentic data’ is a chimera because we are all continually monitored by and monitoring other people, adjusting our use of language accordingly. This tendency is a fact of life, and not simply something we do when confronted by a clipboard-wielding investigator! Furthermore, there is acknowledgement that all interactions are a joint production and ‘structured’ by both the researcher and the subjects (cf. Silverman, 1993: 116-7; Hammersley & Atkinson 1983: 110-11; Agar, 1986: 18-19; Stubbs, 1983: 21). This was certainly something that I wished to achieve when working with the children, only imposing ‘control’ if they began to stray from the broad focus on media or if the atmosphere became a little volatile.

Interactionism is generally favoured by those who reject the restriction of standardised interviews which often involve such closed questions that the interviewee is likely to have to compromise his/her real point of view or feel uncomfortably ‘cornered’ when responding. Standardised interview techniques are useful if the aim is to generate easily quantifiable data, but the intention was to generate qualitative material during this investiagtion.  Denzin (1970: 125, in Silverman, 1993: 95) suggested three good reasons for employing the open-ended, interactionist style of interview. Firstly, it allows respondents to use their ‘unique ways of defining the world’. Secondly, it assumes that no fixed sequence of questions is suitable for all respondents, and finally, it allows respondents to ‘raise important issues not contained in the schedule’. Like other research methodologies, interactionism also has its drawbacks (cf. Silverman, 1993: 95) but I do not regard any of these negatives as serious obstacles. Ultimately, my aim was to ‘tentatively-suggest’ rather than ‘imply-as-universal’ the opinions expressed by the children in my sample.

There are many recommended methods of recording fieldwork data but audio-recordings and field-notes were used in this investigation (cf. Agar, 1996). Had a video camera been used, the static positioning of the tripod may have failed to capture all the ‘action’ and would have been far more intrusive that a tape recorder. Typically, the field-notes were no more than jottings taken down during the course of the workshop exercises but proved helpful when the audio-recordings were transcribed because they facilitated reconstruction of events, detailing non-verbal behaviours. An ethnographic-style approach also allows for greater use of multiple data sources (Hammersley & Aitkinson, 1983: 24) which, in the context of working with the children in my sample, included oral responses to advertisements, basic questionnaires and ‘art’ to encourage the expression of ideas.

In terms of actually transcribing conversation data, there are a number of techniques that one can employ depending on one’s research intentions. The intention here was to study expressions of opinion about a particular topic within a social context so a simple system of symbols was sufficient. I decided to adopt the same conventions as those used by Buckingham (1993a: x), marking simple aspects of the conversation such as pauses, interruptions and simultaneous talk. Talk was transcribed exactly as it occurred and the temptation of ‘tidying-up’ any ‘errors’ was avoided (Silverman, 1993: 117). In this way, the conversations would be transcribed as ‘authentically’ as possible, avoiding what Abercrombie (1954, in Stubbs, 1983: 227) termed the ‘initial classification and even theorising about the raw material’. One of the main identifiable problems for researchers who wish to adopt an ethnographic-style approach is that of access to the desired data source. Fortunately, the school visited during this study allowed unlimited access throughout the three-year period of major investigations, so the same sample of children was maintained without problem.

Being able to maintain the same sample throughout a field study seems quite unusual but, as Lull (1990: 2) emphasised, it is vital to know the subjects you intend to analyse and this requires sustained contact. Lindlof & Meyer (in Lindlof, 1987: 1) suggested that investigators might be deterred by the time and effort needed to achieve rigour, but I was fortunate to have the luxury of a three-year period. This is something that may have proved impossible for other researchers working with children (e.g. Buckingham, 1993a) due to the impracticability of prolonged contact with the same group and other time constraints.

Despite ensuring a degree of acceptance amongst the subjects, it is also emphasised that a researcher should attempt to maintain a degree of detachment (Hammersley & Atkinson 1983: 100). Indeed, many practitioners in the various forms of ethnography have defined their positions in a series of binary oppositions such as ‘insider-outsider’ (Lofland, 1972: ibid.), ‘stranger-friend’ (Powdermaker, 1966; Everhart, 1977: ibid.) or ‘marginal-native’ (Freilich, 1970: ibid.) as a way of producing ‘creative insight’. Yet, the presence of the researcher may still have a significant effect on the interaction, even when they are not directly involved. If awareness can be shown about potential influences, however, then insightful comments and observations are still possible (Hammersley & Atkinson, 1983: 111-2).

More negatively, the flexible nature of ethnographic-style research is often deemed too subjective (Hammersley & Atkinson, 1983: 2) revealing little more than the research intentions of the investigator. Another criticism highlights the fact that the subjects being studied belong to only a small sample and may not consequently be representative (Hammersley & Atkinson, 1983: 44). However, if the investigator avoids reification and consistently grounds any comments within the context of the sample then validity can arguably be maintained. Indeed, as Hammersley & Atkinson (1983: ibid.) stressed, the interpretational possibilities in any qualitative investigation are infinite. Once interview data are elicited, however, one must consider how best to approach a ‘deconstruction’ in order to make it insightful in the context of the study.

3.2.2 Discourse analysis

Having generated and transcribed a series of conversations amongst children, it is often a foreboding challenge to deconstruct and comment upon aspects of the discourse. It is therefore wise to consider some established methods of analysing talk and there are two main methods to choose from, subtly different in their focus. Conversation analysis (CA), on the one hand, concentrates mainly on so-called ‘informed’ conversation between equals such as a telephone conversation, with recent work branching-out to include institutional types of discourse of which media discourse is one example (Fairclough, 1995: 22). Discourse analysis (DA), on the other hand, seeks to analyse the activities present in talk, within a number of academic disciplines, including linguistics, sociology, psychology and anthropology. DA has three features that makes it more suited to my own research aims. Firstly, unlike CA, it is concerned with a broader range of activities often related to more conventional social science concerns, such as gender relations. Secondly, it does not always use analysis of ordinary conversation as a baseline for understanding talk in institutional settings, and thirdly, it works with less precise transcripts than CA (see Silverman, 1993: 121).

Discourse, whether spoken or written, is social and also includes non-verbal types of communication such as body language and facial expression (Cook, 1992: 4). Volosinov, a Russian linguist writing in 1930 (cited in MacDonell, 1986: 1), recognised that the use of words and their subsequent meanings were dependent on where and against what the statement was made. Stubbs (1983: 1) explained that perhaps the fascination of discourse analysis comes from the realisation that ‘language, action and knowledge are inseparable’. He further stressed, however, that discourse analysis is a formidable task (ibid.: 3). One can only hope to show an awareness of the strata of discourses involved in any one situation, whilst striving to treat each discourse type as a kind of exploration into what Stubbs termed ‘some kind of theory of social action’ (ibid.).

In the context of this investigation, five main discourse types have been identified (although there may have been more) which seemed to clearly operate within each fieldwork activity – advertising discourse, media discourse, audience discourse, child discourse and school discourse. Within each of these discourse types, there were further identifiable sub-categories.

There are clearly multiple discourses at work within advertisements. These are summarised by Cook (1992: 6) who emphasised that advertisements have no single sender because they are created by many different specialists, from marketing experts and toy companies to advertising designers and camera operators (see Chapter Six). When this is added to the impact of the finished product on may thousands of individuals and ‘different’ minds there are an infinite number of possible interpretations of the commercial texts. It was assumed that the children in the sample would have a level of understanding about the medium of television, an understanding of how advertisements are constructed, and some awareness of the commercial intent of advertisements to ‘sell something’. Perhaps the discourses in the advertisements seen by the sample children could ‘colour’ the nature of their subsequent conversations.

Audience discourse can be divided into adults and children. The adult perspective in this investigation was sub-divided between my own perception of advertisements, as someone who had studied advertising theory, and the perceptions of others who may never have considered the content of advertisements before. Child discourse could also be divided in this investigation into the spontaneous talk that was often heard in the playground and through non-work talk in the classroom, as well as more formalised classroom talk in which questions were asked and answered. This overlapped with school discourse that included children’s status in the classroom as well as the role of the teacher who might deliver anything from lessons to advice and punishment.

It is important to briefly locate the idea of ‘audience’ within the broader context of discourse types, and the possible ‘effect’ that this may have on the media message. As Ang (1991, in Strinati, 1995: 252) noted ‘we cannot presume to be speaking with the authentic voice of the “real audience” because there is no such thing’. Receivers of advertising discourse are likely to interpret the ‘message’ in different ways depending on whether they are children or adults, women or men, rich or poor and so on (Cook, 1992: 65). The eventual process of ‘decoding’ may result in a ‘message’ that resembles the intention of the sender but is not identical to it. Eco (1986: 141, in Nightingale, 1996: 35) explained how the interpretation of a media message varies, in that ‘messages set out from the source and arrive in distinct sociological situations where different codes operate’.

In any analysis of mass media ‘texts’, it is impossible to escape the ideological ‘vicious circle’ of knowing where your own attitudes end and other people’s attitudes begin, or where such ideas originated. In modern day society, we are all bombarded with images, sounds and statements that are all supposed to mean something to each of us. Very often, media output is designed with a specific audience in mind, and the advertising industry is skilful in defining its audiences so as to be able to ‘sell’ to them. As Strinati (1995: 250) stressed, however, there is no guarantee that the audience will behave in the way that the advertisers predict and may even actively resist being ‘constructed’ in this way (cf. Research Issue 5).

3.2.2.1 Considerations when using conversation extracts

Selected extracts of conversation were used predominantly for illustrative purposes in this investigation, so their inclusion within the text is far from ‘innocent’. It could also be argued that the reader should be equally wary of the misconception that conversation data can ‘speak for itself’. Here, it is hoped that a satisfactory balance will be found between choosing extracts of transcript to illustrate points and simply observing ‘what the children said’. Therefore, the commentary will aim to be as reflexive as possible in terms of contextualising the comments made.

One should be aware that any number of factors may come in to play when attempting to comment on the nature of conversation that will often govern the flow of discussion as a whole. These factors might include the context of talk and the relationships of the participants to one another and to the researcher. Stubbs (1983: 2) explained that the language and the social situation are always interrelated, while Cook (1992: 1) emphasised the importance of context. In this investigation, for example, the patterns of discourse were likely to reflect the fact that talk was occurring within the classroom environment, where it is generally expected that children will answer questions and voice opinions while looking upon any adult as a ‘teacher’. The flow of the conversation may also be influenced by the research agenda, whether consciously in an interventionist sense or unconsciously as a result of internalised research rationale and overall aims.

While the interviewer’s role may be clear to both him/herself and to the interviewees, the position of the interviewee may be less certain. Indeed, there are many considerations when interviewing young children that may colour or ‘distort’ their responses particularly in the early stages of talk. Denzin (1970: 133-38, in Silverman, 1993: 97) listed a number of potential ‘problems’ that should always be considered when analysing transcripts. For example: the interviewer and interviewee adopt different interactional roles, while initial self-presentation may prove difficult. Added to this are problems of the so-called ‘fleeting relations’ of the topic under discussion to which the interviewees may have little commitment and consequently decide to fabricate their responses according to their mood at the time. There is also a difficulty in penetrating private worlds of experience, while the relative status of the interviewer and interviewee and the context of the talk will all have an influence on the pattern of conversation.

When applied to this investigation, I hoped to have taken all these issues into account while the benefits of prolonged contact with the same groups of children perhaps lessened the problems identified by Denzin (1970: 133-38, in Silverman, 1993: 97). Hopefully, my being a regular visitor to the school avoided the initial ‘need’ for the children to present a special ‘image’ of themselves. Many of the visits to the school did not involve any formal fieldwork exercises, but the children’s personalities and interrelationships were always observed and noted. The enthusiasm with which they greeted every fieldwork exercise was certainly indicative of the fact that they enjoyed the work, so one would anticipate that through their enjoyment they were at least thinking about the issues and trying to articulate their thoughts. This is not to say that my aim was to search for some absolute ‘truth’ about what children ‘really think’. Indeed, Baker (1982: 109, in Silverman, 1993: 90) explained that both the interviewer and the interviewees work together to construct a version of the world or, in my case, how advertisements for toys fit into the wider context of (the children’s) childhood.

Hodge & Tripp (1986, in Silverman, ibid.: 45) made some interesting points about the context of children’s talk. They argued that the most powerful determining factor in the course of the conversation was that of existing social relations between members of the group. Negotiation and re-definition of these inter-group relations throughout the course of the discussion may have significance for the overall ‘meanings’ intended by the speakers, as well as the judgements that the children appeared to make about television. Buckingham (1993a: 56) cited as an example the effect on the overall group if an ‘unpopular’ child expressed an opinion. It is likely that such a child would be marginalised at the first opportunity even if their opinions reflected the general group consensus. As a result, vital issues may be left out of the discussion because they happen to have been raised by the ‘wrong’ person. It would be both impossible and futile to try and ‘filter-out’ these social relations so one should accept them as part of the group composition. Ethnographic-style approaches at least allow the interviewer sufficient opportunity to acknowledge and understand such interrelations.

As previously noted, the whole notion of ‘discourse’ is complicated by the fact that it is multi-faceted. Any number of discourse types may operate simultaneously in any single situation, and one can only ever hope to show an awareness of such subtle, inter-linked ‘layers’. Indeed, as Cook (1992: 4) emphasised, discourse types are never mutually exclusive; they may be several types at once, interacting and shifting in accordance with the situation, topic, participants and other individual factors (cf. Chandler, 1997b). These considerations can be applied to this investigation and summarised as follows:

3.2.2.1.1 Situation

·         The formal classroom environment may have an influence on the language used by the children, while the possibility of any physical action is often restricted.

·         The composition of the groups being interviewed may have an effect on the flow of conversation. For example, dominant personalities may govern the pattern of answers given by all the children.

·         Particular mixes of boys and girls may work to either encourage or discourage talk.

·         Individual children may feel less confident about using one language over another (Welsh or English).

·         Unavoidable interruptions and disturbances when conducting fieldwork exercises may unsettle the flow of conversation.

3.2.2.1.2 Topic

·         The questions used to direct the flow of the conversation may well confuse some children and restrict others who have plenty of ‘unexpected’ views.

·         The children may remember other fieldwork exercises, and their recollections of these could colour or influence their responses in subsequent activities.

·         Since the children were likely to be aware that my main interest area was advertising and (to a less obvious extent) gender, the views they express may simply be what they feel I ‘want to hear’.

·         Some children may already know more about a certain topic than their peers, so the content and frequency of their contributions may be influenced accordingly.

3.2.2.1.3 Participants

·         The Welsh speakers may have felt disconcerted by the dominance of English speakers, and vice-versa.

·         The boys and the girls in the sample may have approached the workshop sessions and perceived things differently on the basis of gender.

·         Certain imbalances in the ratio of boys to girls, or girls to boys (depending on group composition) may influence the nature and extent of individual’s contributions to the conversation.

·         The inter-pupil relationships already present in the classroom may have a significant impact on the nature of the conversation. For example, his/her peers may oust an ‘unpopular’ child while a ‘popular’ child may prompt others to giving similar answers.

·         The inevitable considerations of how the children perceive the investigator.

3.2.2.1.4 Individual factors

·         Certain families may already have discussed with and ‘educated’ their children about the nature of advertisements, while other families may never have discussed such matters.

·         The mood of the discussion, whether light-hearted and humorous or hostile and serious, may influence the ‘openness’ of the talk and the type of responses made.

·         The rapport that I may or may not have with the children is central to their readiness to be constructive and participate.

·         The responses made by the children will also depend on whether they find the topic interesting or stimulating enough.

·         The individual personalities of the children may effect participation in the conversation, with the confident talkative children dominating the shy, nervous children.

·         ‘Personal agendas’ and the need for ‘one-up-manship’ may influence the interaction between certain children in an attempt to ‘point-score’.

·         Some individuals may simply be seeking attention without necessarily wanting to contribute to the flow of conversation. For example, interrupting, kicking or poking their classmates.

·         Linguistic competence (in either Welsh or English) may influence the degree or ‘quality’ of the children’s input.

By offering a very basic skeletal idea of the multiple interacting discourses that one needs to be continually aware of when evaluating transcribed conversations, it becomes clear that no interview is every free of constraint and influence. This is why no evaluation of conversation should imply that the statements made are universal or that the group of children one has chosen to interview are representative of all children, nor should the conclusions drawn be reductive in any way.

Since the ethnographic method of data collection tended to rely on oral communications, and to ensure that this investigation took account of all the possible shortfalls and criticisms, efforts were made to find an alternative means by which the young children in the sample could articulate their ideas. In the following section, the use of ‘art’ is considered as a potentially illuminating alternative to ‘talk’ and is used in this (media-based) investigation as an interesting (if rarely used) medium of expression.

3.3 ‘Art’

Buckingham (1993a: 44) argued that children possess ‘discursive repertoires’ that comprise various types of communication, from straightforward talk to more forceful arguments, and even singing. However, the problem of trying to penetrate the ‘private world’ of young children could be especially difficult in that they may not yet possess the necessary linguistic capabilities to articulate their thoughts.

The majority of studies concerning children’s understanding of the mass media seem to have consistently depended on oral expressions of opinion. It has, however, already been stressed that linguistic competence is a very real issue for young children. One could argue that language issues are even more prevalent in this investigation because of the bilingual nature of the sample school and surrounding community, and (perhaps to a lesser extent) Welsh-language television programming. In order to alleviate the problems associated with a single mode of communication, art seemed to be the most viable option for providing the children with an alternative ‘outlet’ for the expression of ideas. The ‘expressive’ potential of art might also demonstrate whether young children have sufficient understanding of advertisement conventions to be able to reinterpret the techniques in the context of a different medium.

Many researchers offer powerful arguments as to the advantages of using art as a means of understanding the inner-workings of a child’s mind. Davis (1985, in Sitton & Light, 1992: 26) implied that drawings convey a great deal of information. Fury et al. (1997: 1154) suggested that the non-verbal nature of drawings allow children to express emotions and attitudes that would not normally be expressed. Lewis & Greene (1983: 13) argued that the symbolic nature of children’s drawings and pictures often reveal things that the child may be either unwilling or unable to put into words. 

Lewis & Greene (1983: 7) argued that drawings can reveal many things within the spectrum of the child’s emotional being such as ‘dreams and fears, hopes and despairs, conflicts and confusions…’ Other researchers have employed the techniques of ‘art psychology’ to understand other aspects of childhood. Fury et al. (1997: 1162), for example, suggested that drawings could be used to investigate what they referred to as ‘children’s representational modes of attachment’. That is to say, the way in which children compose family pictures can reveal their levels of self-concept, the way in which they perceive themselves, perceptions of other family members and notions of interrelationships.

Fury et al. (1997: 1162) also suggested that drawings are particularly useful during ‘early middle childhood’ where verbal skills are not yet fully developed but where expressive play is regarded as less appropriate. The (7- to 11-year-old) children in this investigation often regarded toys as being too childish, so it was hoped that art would provide a more ‘acceptable’ outlet for them to express their views about toys, the presentation of toys within advertisements and the notion of so-called ‘appropriate gender behaviour’. If children’s drawings do indeed reveal emotions and family relations, then it must surely be reasonable to suggest that the same pictures could also reveal children’s notions of gender (traits, roles, preferences and interrelationships) and, depending on the drawing task, their knowledge of technical (televisual) conventions.

3.3.1 Gender differentiation in children’s drawings

Perryman (1977, in Chen & Kanter, 1996: 44) argued that children rapidly begin to differentiate and distinguish between gender roles by the time they reach the age of four. In addition, Derman-Sparks (1989, in Chen & Kanter, ibid.) stressed that children are strongly influenced by the social values attached to so-called ‘gender appropriate behaviour’. While these perceptual and social changes are occurring, there is also a significant shift in children’s artistic ability (Herberholz & Hanson, 1985: ibid.). Many researchers have claimed that there are strong differences in the way that different genders draw the human figure (Harris, 1963; Willsdon, 1977; Majewski, 1978; Cox, 1979, all cited in Chen & Kanter, 1996: 44). Others have observed how the subject matter of children’s drawings can differ according to gender (Feinburg, 1979; Majewski, 1979; McNiff, 1982; Reeve & Boyette, 1983, all cited in Flannery & Watson, 1995: 114). These drawing differences have also been observed cross-culturally (Anastasi & Foley, 1936, in Flannery & Watson, ibid.).

Gender preferences were also seen to emerge in a number of other studies. For example, Jolles (in Harris, 1963, in Flannery & Watson, ibid.: 45), in a study of children between the ages of five- and eight-years, found that 80% of the children would always elect to draw their own sex first. Willsdon (1977) and Majewski (1978) (ibid.) also noted a similar tendency for children to prefer drawing their own sex or at least drawing their own sex more frequently than the opposite sex. Majewski (1979) (in Flannery & Watson, 1995: 115) further noted that girls tended to be more expressive in their drawings than boys were and were more likely to paint happy faces on people and animals.

Chen & Kanter (1996) examined a number of kindergarten (5- to 6-year-old) and third-grade (8- to 9-year-old) children’s drawings in terms of gender differentiation, looking at a number of variables that essentially concentrated on the form of the drawings. These variables included subject matter, gender selection, the shapes used, the colours selected and the level of artistic skill. Interestingly, they only found statistically significant differences in terms of drawing skills (girls being superior to boys) and the children’s preferences for drawing their own sex.

Flannery & Watson (1995: 115-7) considered sex and gender-role differences in terms of four drawing variables. Firstly, they considered what they termed ‘theme realism’, or the likelihood of the scene depicted in the drawing to occur in real-life. Secondly, they considered levels of aggression and the degree of violence depicted in the drawing. Thirdly, they looked at expressiveness and how much emotion was displayed in the drawing. Finally, they considered artistic skill and the degree of aesthetic competence seen in a particular drawing. The investigation also sought to ascertain whether there were any gender-role differences in children’s drawings aside from sex differences, based on the definitions of masculine and feminine personality traits proposed by Bem (1977, in Flannery & Watson, ibid.).

Again, Flannery & Watson’s (1995: 120-1) investigation seemed to reveal more similarities than differences in terms of the drawing styles exhibited by the two sexes. Their research found that the only real difference lay in the fact that male drawings were rated as more unrealistic and more aggressive when compared with female drawings. No sex differences were found in terms of expressiveness or artistic skill. One could also argue that the slight differences identified by the researchers would have little influence on the overall effect of the drawing, where boys and girls used ‘art’ in similar ways. Using ‘art’ as a formal research methodology would arguably therefore offer young children (irrespective of gender) the same levels of expressive opportunities, and seems well suited to this investigation.

3.3.2 Themes, ‘reality’ and influences

Other researchers have considered children’s drawings in terms of the themes they choose to depict, the ‘reality’ of the images on the page and the possible influences (cultural, social, mass media and so on) on the types of drawings produced. In a particularly interesting article, Wilson & Wilson (in Madeja, 1978: 91 ff.) considered children’s shifting notions of ‘reality’. They argued, in accordance with Powers (1973) and from the perspective of cognitive psychology, that the notion of ‘reality’ as we know it is actually ‘our mind’s model of reality’. Furthermore, this ‘model’ is said to continually shift in an endless process of reconstruction as new schemata are incorporated into the ‘conceptual gestalts’ (Neisser, 1976, in Madeja, ibid.) of individuals. When an individual encounters something, their (perception of) ‘reality’ will continue to grow.

Wilson & Wilson (in Madeja, 1978: 94 ff.) encouraged a group of primary school-age children to draw what they term ‘visual narratives’. That is to say, they analysed the way in which young children told stories through a series of pictures. They found that the narratives produced were particularly complex and drew on varied sources that they list as ‘myth, fiction, language, games, cinematography and art’ (ibid.). They identified a number of themes that seemed to run through the visual narratives such as good versus evil, justice, birth and rebirth, cycle, growth, metamorphosis and odyssey (ibid.: 96). A number of these themes were identified in my sample of televised toy advertisements (cf. Semiotic analysis) perhaps suggesting that they form a major part of the so-called ‘narrative of childhood’.

The most intriguing aspect of Wilson & Wilson’s study (in Madeja, 1978: 99-100) was how the children’s visual narratives were seen to draw explicitly on the ideas and conventions used in television genres, particularly cartoons and film. They referred specifically to a narrative sequence entitled ‘Rocket Villain’, drawn by a six-year-old boy. The narrative consisted of six frames showing a villain being chased by gun-toting police, before setting a rocket alight and being blown into little bits. The authors noted that the use of abrupt scene changes (cuts or ‘harsh’ transitions), long shots and overhead views in the sequence were reminiscent of comic books, television programmes and films. The authors also observed that time-lapse sequences, zoom shots and extreme close-ups appeared regularly in narratives drawn by other children. Added to the techniques of cinematography were those of cartoon animation such as action lines, curlicues, stars, radiating lumps and speech bubbles. One could argue, therefore, that young children are sufficiently aware of televisual conventions to spontaneously reproduce and reinterpret the techniques in their own drawings. This was the overriding assumption when the children in this sample were asked to design their own toy advertisements.

From a somewhat sweeping overview of the gender and advertising considerations that form the basis of this investigation and the various methodologies used to generate data, it is hoped that the subsequent chapters will prove illuminating in terms of identifying both patterns of advertisement composition and patterns of child-response. Selecting a collection of diverse yet complementary methodologies seemed the best way to work through the investigation and attempt to answer the key research issues.

Summary

The discussion on methodologies was divided into the two distinct categories of ‘text’ and ‘audience’, with the aim of selecting the most effective approaches to both the (toy) advertisements and the young children’s responses to them.

The first form of ‘textual analysis’ was that of content analysis. It was chosen as an effective means to ‘quantify’ the various elements that combine to make an advertisement text. The tool was particularly suited to this investigation because it facilitated the identification of integral and distinctive gendered patterns of commercial presentation (cf. Research Issue 2), as well as generating a solid framework of toy advertisement conventions. The main cautionary note about this methodology, however, is that it should not be taken as a measure of how the (advertisement) texts might be received and interpreted by the audience. 

The second ‘text’ approach was that of semiotic analysis. This method is arguably complementary to the quantitative approach of content analysis, in that it places greater emphasis on the qualitative interpretations of (media) texts. This method was chosen because it seemed particularly suited to a consideration of whether certain advertisement could be described as having gender connotations (cf. Research Issue 3), hence revealing the so-called structures of underlying meaning. Both content and semiotic analyses are effective ways to define and make explicit the so-called ‘grammar of television’, which is an important consideration when studying the media in relation to children. It is reasonable to assume that young children somehow have to learn the ‘grammatical rules’ before they can meaningfully ‘decode’ a text.

When considering how to approach the issue of ‘receiver’ or ‘audience’, ethnographic-style field observations and interviews were chosen as the best way to generate responses from children within the context of a fieldwork study. The flexibility possible within the confines of the methodology was also conducive to working with children in that it would allow them to digress from the major research concerns of the investigator whilst still maintaining a high level of relevance to the investigation as a whole.

A version of discourse analysis was then selected as the most meaningful way to deconstruct the children’s recorded conversations. Perhaps one of the most important acknowledgements when applying this analytical tool, however, is that discourse is often composed of multi-layered strata which may have an impact on the overall conclusions generated. Some key considerations during the process of analysis might include situation, topic, participants and individuals, and an awareness of the relationship between these factors can arguably result in a richer study.

A final method was used with the express purpose of being able to move away from the conventional (over-) reliance on oral communication, where the children were asked to draw their own advertisements. ‘Art’ was seen as a viable alternative form of ‘expression’ since it has been used as a means to measure such things as emotion and family attachment. Children’s drawings have also been used in the study of gender differences and as evidence of children’s abilities to transfer televisual conventions to the static medium of paper (cf. Research Issue 5).

▲