122
become so aware of these extremist ideas if they had not been active users of social media. Those who support SinhaLe and the BBS believe they have access to a ‘truth’ of which others are ignorant. The nationalist material that they are exposed to plays into their identity formation and worldviews (Ivarsson 2018:12).
Further, Ivarsson’s study is informative in two other aspects. First, Ivarsson correctly notes the incongruity of ideas expressed by ordinary people on social media. On social media, people express contradictory views on out-groups in different contexts, as they do in real life situations. Secondly, many of the young people ‘like’ and ‘share’ posts made or shared by their friends due to peer pressure.
While this, to a certain extent, explains their identity, the contents of the post are not always the critical factor that motivates them to re-post or ‘like’ (Ivarsson 2018:12). Thus, like any other source of data, social media also provides a partial view of the world, although social media data expose the researcher to the greater public than any other traditional data sources.
123
cognitive anthropology, and ethnomethodology (Ball & Smith 1992:3). This study uses content analysis as the tool of analyzing images based on the following justifications.
3.4.1 Content Analysis
Content analysis, one of the widely cited systematic and empirical tools that have been developed to analyze documentary data (Ball and Smith 20: 1992), is chosen over other methods due to its inherent merits that facilitate the purpose of the present study. First, content analysis is “an empirical (observational) and objective procedure for quantifying recorded ‘audio-visual’ (including verbal) representation using reliable, explicitly defined categories (values on ‘independent variables’) (Bell 2001:13). Thus, content analysis has been widely accepted as an apt method to analyze visual data in particular.
Second, it allows “quantification of samples of observable content classified into distinct categories” (Bell 2001:14). According to Berelson (1952), who is one of the pioneers of the modern version of content analysis, content analysis is a highly quantitative tool (Berelson 1952:18) although there are emerging non-quantitative methods of content analysis (see Schreier 2012). Since the present study intends to analyze data statistically, the pro-quantitative venture of content analysis is an advantage.
Third, content analysis can be conducted with a relatively high degree of objectivity. After selecting relevant documentary sources, the next step in the
124
process of content analysis is to devise the analytical categories43 and formulate a set of coding rules. Categories and the attached coding rules for each category are the very objective basis of content analysis. The set of categories are the central part of the analytical process－they should be representative of and sensitive to the research questions; should be mutually exclusive and collectively exhaustive as well. Coding rules, on the other hand, are devised to deal with ambiguities of any given content. If a given item is ambiguously appearing to fit into two or more categories, on such occasions coding rules provide proper instructions to treat the problematic item. Thus, coding rules mostly consist of decision rules. These categories and coding rules minimize the subjectivity of the coder (the person who is involved in coding) affecting the final results. It avoids opportunities of interpreting the contents, and coding is done in a more or less unobtrusive/mechanical manner that increases the objectivity. “In principle, different analysts using the same categories and rules would obtain identical results…of any given body of data” (Ball & Smith 1992:21), which increase the reliability of the analysis.
Also, one of the most important aspects of content analysis in this manner is that only the manifest content (that is obvious, palpable and self-evident content) is considered and coded, but not the latent content. This has been criticized by scholars using other methods such as Symbolism and Structuralist approaches to
43 Fragmentation of a given content/message into pieces has been criticized on a scholarly basis that although the categories could be designed in accordance with a particular theoretical approach, the same set of
categories might not reflect the reality. In other words, “the category system may or may not correspond to the categories that members of the society employ to understand the communicative message” (Ball and Smith 1992:27). In order to avoid this criticism, this study at the very outset of the data analysis took an inductive approach or rather utilized the knowledge on Grounded Theory Method, and categorized data upon the natural themes emerging through the data without employing any theoretically informed categories. Also, other approaches to visual data analysis such as Symbolism and Structuralism does not prefer fragmentation of the message, instead “[a]n appreciation of communication content in its totality” (Ball & Smith 1992:28) is preferred.
125
content analysis. “The essence of symbolism lies in the recognition of one thing as standing for (re-presenting) another, the relation between them normally being that of concrete to abstract, particular to general” (Firth 1973:15) and structuralists’
standpoint is more or less the same, that “a sign consists of two elements, the signifier (the material object, word, or picture) and the signified (the meaning ascribed to the material object, word or picture). Hence the structuralist slogan: A sign is always thing-plus-meaning” (Ball and Smith 1992:32, 46). For instance, a beard may symbolize masculinity in a specific social context while something else in another society. The other alternative for the objective and quantitative content analysis of visual representations is the cognitive anthropological and ethnomethodological approach. Treating human beliefs, concerns, and practices as ethnographic data that is related to its context is one common feature of both ethnographic and anthropological approaches. The main idea behind both these approaches is that “people’s experience of the seen world is culturally shaped and socially constituted and mediated” (Ball and Smith 1992:55). Typical content analysis lacks this context-specific meaning construction of a given set of data and also it does not consider the symbolism beneath the surface.
In what ways, were the manifest content (that is obvious, palpable and self-evident content) identified and coded in the present study? Below, the author provides practical examples of the way coding was conducted. One of the major convenient factors was that, the majority of the images consist of a direct message along with a short textual content. This can be considered as an internal caption (i.e.
an image picturing one or more soldiers comes with written text saying that there is a conspiracy against Sri Lankan war heroes; another image portraying some of the ruins of ancient hydraulic civilizations of Sri Lanka states that ancient Sinhalese
126
were highly skillful; another image depicting cow slaughter conducted by Muslims, consist a textual content that cow slaughter is an uncivilized practice). This sort of image that has internal textual contents can be easily coded without controversy.
However, there were several specific occasions that the author had to use cultural knowledge and context-specific knowledge to make meanings. For instance, in the sample, the lion image reappeared frequently. The lion image has been a historically used symbol to refer to the Sinhalese community, and their bravery (see Gunawardana 1990). The Lion image, in the Sri Lankan context and culture, is hardly used to refer to other non-Sinhalese communities. Thus, ‘Lion stands for Sinhalese’ is a clear symbolic fact and in the coding process it was considered as an obvious, palpable and self-evident content (several other examples of practical issues emerged in the coding process, will be discussed in the forthcoming sections when necessary).
In summary, the present study codes only manifest and obvious messages rendered from the images, since coding latent contents creates controversies. Thus, content analysis is primarily limited to what is expressly communicated by some document rather than the motives animating the construction of the document or the responses that persons make to it (Berelson 1952:16; Holsti 1969:12-14; Ball and Smith 1992:21). Specifically, in a study, that deals with sensitive aspects of humans such as identity, group or religious affiliations, and also when the researcher is also affiliated to the socio-cultural world under investigation, images could be highly polysemantic and deciding the meaning of the image become controversial. To avoid such complexities and also to maintain the objectivity of the data analysis, categorization upon coding rules is highly helpful. Finally,
127
content analysis permits processing large amounts of data spread over a long period, which is another merit of it (Ball & Smith 1992:25).
Having determined the research problem, and also having decided the documentary sources (data) to be analyzed (as indicated in the previous chapters), the next step is devising the categories and the coding rules. On what basis have the categories and related coding rules been determined? Being informed by the theoretical and conceptual discussion in the previous chapter and similar studies, the sections below explicate the central analytical categories designed for the present study and also justify on what basis those categories have been created.