Television viewers are not normally conscious of the formal features of television commercials such as camerawork, editing, and sound-tracks (see, for instance, Messaris, 1994, p. 158). Commercials, like the television programs which exist to support them, still tend to follow the classic Hollywood tradition of ‘invisible editing’– it is the represented action which is meant to be foregrounded rather than the formal conventions involved in the process of representation. Our own current concern is with gender differentiation in the use of formal features in children’s commercials. Existing academic research into the production features of television advertisements for children has concentrated on their role in attracting and maintaining attention or on their interpretative importance for children (see, for instance, Bryant & Anderson, 1983; Chandler, 1997; Meyer, 1983; Salomon, 1981). Whilst the content of television ads has been widely studied in relation to gender issues, there is relatively little published research on differences in the ‘formal features’ of children’s commercials in relation to the gender of the primary audience targeted. Yet at least to some extent ‘the medium is the message’: even more than in other televisual genres, the form or style of an advertisement is richly meaningful, and ad-makers routinely link this to gender connotations (Messaris, 1997, xv).

We have not been able to find any published replications of the study by Welch et al. (1979), although some of the same research team have explored related developmental issues (Huston et al., 1984; Wright & Huston, 1983). Although the original study was only small-scale, it has subsequently been widely cited. If particular formal features of commercials are ‘gendered’ in the ways which the authors suggest then such findings are important. Related studies by the CRITC research group have shown that at the same time as children are learning to ‘read’ the semiotic codes of television they are learning that such codes are gender-differentiated. Young viewers use the formal features of the medium (as well as content cues) to determine whether they are designed for them or not (Wright & Huston, 1983). Children as young as 6-years-old can distinguish ads targeted at males from those aimed at females by their distinctive formats and visual styles (Huston et al., 1984). The gender-differentiated use of formal features is often associated with stereotypically gendered content in commercials– and this is likely to be how children initially learn the gender connotations of such features. However, there is empirical evidence that even when content is ‘neutral’ such production features can generate these gender connotations relatively autonomously (ibid.).

Alongside many other socialization factors, regular exposure in childhood to commercials which are sex-typed in style as well as in content may help to establish gendered preferences for particular stylistic traits such as a stereotypical ‘masculine’ taste for a rapid cutting style. Along with content cues, rapid cutting and the shorter duration of shots may also support a masculine self-image which is more action-oriented, whilst the salience of dissolves and longer shot lengths may tend to encourage an acceptance of the stereotypical association of women with passivity. The stylistic modes of address employed in advertisements may thus be a contributory factor in the gendering of tastes in televisual material, the cultural framing of ‘activity’ versus ‘passivity’ and perhaps also particular styles of viewing (Fiske, 1987; Morley, 1986). Findings to date cannot, of course, be taken as evidence for such speculations, but rather generate such questions for further research.

In addition to these situational differences from the original US study, regulatory differences between the US and UK might lead us to anticipate the possibility of differing findings from studies of commercials, although UK regulation does not extend to the formal features in which we are interested (ITC, 1998). In Britain, the content, timing, and total amount of television commercials is strictly regulated (Hart, 1990; ITC, 1998; Jefkins, 1992): for instance, there is a clear break between programs and ads and sales messages cannot be included within programs. In the UK, programs on ‘commercial’ channels are provided by regional contractors, not by advertisers– although since 1991 there has been regulated sponsorship of some programs (Jefkins, 1992, p. 169). In both countries the standard duration of a commercial is 30-seconds, although ads are occasionally longer (up to 60 seconds), and it is notable that shorter ads are being increasingly used (Condry, 1989, pp. 180ff; Myers, 1999, p. 124). The most obvious difference is that in Britain commercials appear less frequently than in the US– ads can be shown only in three specified time-slots per hour, the duration of each of these slots being three or four minutes. An American-born researcher living in Britain notes that ‘the ads in the US Superbowl alone would use up the allowance for 24 hours’ (Myers, 1999, p. 117).

 Methods
The purpose of the current study was to investigate toy commercials broadcast recently on UK television to check for gender-differentiation in the use of specific formal features. We drew up a series of investigable predictions based on a review of the literature. Those explored in this paper are as follows:

• More shots are likely to be used within each advertisement aimed at boys and consequently the average duration of each shot is shorter.
• More dissolves are likely to be used in advertisements aimed at girls.
• More male than female voice-overs are likely to be used overall, even in advertisements for girls.

The next stage was to select a number of advertisements that could be classed as being intended specifically for children. Toy commercials were selected because relatively little attention has been paid to them in the past, despite the fact that play is an integral part of childhood and the development of the self. Recordings were made of the advertisements broadcast on the HTV Wales channel (a regional version of ITV) between 7.00 a.m. and 9.30 a.m. on a number of Saturdays from the beginning of November to the middle of December 1996. Saturday mornings were targeted both for reasons of convenience and because there tends to be a greater number of advertisements aimed at children at this time (Condry, 1989, p. 203). Toys are well-known to be advertised most heavily in the months leading up to Christmas (Barcus, 1977, p. 120; Condry, 1989, pp. 188-9), and other researchers have chosen to study toy commercials during this season (e.g. Feldstein & Feldstein, 1982). We make no claim for the typicality of this seasonal choice, but variation seems likely to be more a matter of density than presentational style– and it was style which was our primary concern. The applicability of our findings to other seasons, days of the week and times of day, would, of course, need to be empirically investigated for the current findings to be generalized.

To codify and quantify relevant data we used content analysis (which may be applied to the form as well as the manifest content of texts). There were two key coding tasks. One concerned the identification of the target audiences for each of the advertisements; the other concerned the classification of various formal features of the advertisement. There are various ways in which one could classify the target audience for ads. We quickly found that both toy manufacturers and advertising agencies tend to be reluctant to disclose to outsiders details of the target markets for products or for the commercials promoting them. In any case, what the promoters may see as their target audience seems less critical to our own concerns than viewers’ perceptions: a separate study by one of the authors [MG] investigates children’s own perceptions; in this study we chose to rely on how parents classified the target audiences of specific ads. In practice, our parental classification of the ads did not differ very much from classification by the presence or absence of boys or girls on screen– the indicator used by Welch et al. which is also known to be used by children as a content marker of target gender (Wright & Huston, 1981, p. 85). Of the ads in our sample which featured people, all but two of those judged to be for boys had only males on screen; all but two of those judged to be for girls had only females on screen; all of those judged as mixed had both males and females on screen (see Figures 1, 2 and 3).

After the advertisements for toys had been selected from the recordings, ten individuals (five male and five female) were recruited to code each advertisement according to their perception of its primary target audience. The coders were selected through non-probabilistic sampling of parents known to one of the authors (MG), and they ranged in age from 21 to 55 years. Each was interviewed alone in their home and shown a 35-minute edited video containing each of the 117 different advertisements in the sample. The coders were asked to decide whether each advertisement was aimed primarily at boys, primarily at girls, or at a mixed audience and to offer reasons for their judgements. We elected to read the data in terms of the majority response to an advertisement, where six or more coders agreed on the intended target audience. In this instance, the audiences for 115 of the 117 advertisements in the sample were agreed, resulting in an inter-coder reliability level of 98.3%; a strong level of consensus. Different coders were thus in broad agreement about the audience categories to which particular ads were assigned (see Figures 1, 2 and 3). Furthermore, not one of the advertisements judged by the majority as being aimed primarily at boys was judged by any coder as being aimed primarily at girls, and not one of those judged by the majority as being aimed primarily at girls was judged by any coder as being aimed primarily at boys. Where there was any disagreement, the dissenters classified such ads as ‘mixed’. The consistency of judgement on target gender for these ads was thus startling.

We noted the various reasons given by the coders for their choice of target audience. The majority of the advertisements judged to be for girls showed toys that were pink, soft, and cuddly, designed to be played with indoors, and marketed to encourage mother-baby role-play, household activities and the pursuit of female beauty. In stark contrast, advertisements judged to be for boys showed acts of aggression, dark colors, and an emphasis on competitive behavior accompanied by noise and rapid activity. Our coders’ explicit judgements focused on content rather than the formal features in which we were interested so circularity was not a problem. In any case, there is clearly little dispute that advertisements for boys differ from those for girls– we sought to check some of the less immediately obvious ways in which they differ.

After establishing the apparent target sex for each product, the various ‘technical’ features of each advertisement were considered, focusing in particular on those which had already been identified by Welch et al. (1979), but also including key camerawork features. After we decided which features would be recorded, these were identified by working shot-by-shot through each advertisement in the sample– a shot being defined as being a filmed sequence uninterrupted by a transition (a cut or dissolve). Our use of the relevant technical terms throughout this paper follows professional usage on both sides of the Atlantic (Jones, 1974; Millerson, 1985; Watts, 1984; Zettl, 1992). The formal features considered in this paper are:

• editing and other post-production features: transitions (e.g. cuts and dissolves), duration of shots, and voice-overs;

• camerawork features: shot sizes, angle of shots, and camera and lens movement.

The codings of the various features were re-checked on a subsequent occasion and some minor differences in counts were rectified. Critical counts were re-checked for a third time. However, we had no funds for multiple coding.

To analyze the statistical significance of noticeable differences observed in the raw counts of features, we used the Chi-Square test of independence. The use of this non-parametric test was appropriate because it was designed for comparisons of two or more independent samples (in our case, boys’ ads, girls’ ads, and mixed ads) where the data counts are also in discrete categories (e.g. long shots, mid-shots, and close-ups) (Wimmer & Dominick, 1991, pp. 234ff). We did not employ multi-variate statistics since this was a small-scale project, the purpose of which was to investigate specific features which had already been identified in existing studies rather than an exhaustive check for features correlating positively with the gender differentiation of target audiences.

 Results
In summarizing the classifications of the sample in terms of the target gender, it is useful to adopt the distinctions made by linguists and semioticians between ‘tokens’ and ‘types’ (Carroll et al., 1971; Chandler, 1994). In the case of this sample, the count of tokens is a count of the total number of advertisements, regardless of whether any samples are repeated; a count of types, in this case, is a count of the number of different advertisements shown. For the purpose of the content analysis, we are only concerned with the types in this sample. Overall, the sample included a total of 117 different advertisements. After incorporating the results of the inter-coder reliability study, 43 advertisements were classed as being aimed at boys, 43 as being aimed at girls, and 31 designed for a mixed audience. A summary of the counts of tokens and types is provided in Table 1.