The present study is a qualitative with a sociolinguistic approach. In the analysis, I combined content analysis with computer-mediated discourse analysis. In combining the mentioned approaches I aimed to achieve a thorough understanding of the questions at hand and employ the best qualities of the different methods. In the next sections I will be discussing computer-mediated discourse analysis and content analysis and present their individual merits and qualities and further justify why both methods of analysis were used in the present study.
6.3.1 Computer-mediated discourse analysis
Owing to its novelty and fast growing popularity, computer-mediated discourse has received a considerable amount of attention from researchers of different fields during the last few decades. Considering just the linguistic aspects, computer-mediated discourse has been approached from pragmatic, conversation and discourse analytic,
sociolinguistic, genre analytic, and ethnographic perspectives and as such, methods and key concepts have been borrowed from various research traditions (Androutsopoulos and Beißwenger 2008: 1). Therefore critical reflection on the challenges of applying research methods to new settings is partly lacking; however, new frameworks for research are already appearing, such as Herring’s (2004) approach to computer-mediated discourse analysis which will be in use during the present study.
According to Herring (2004: 339) computer-mediated discourse analysis (henceforth CMDA) is an approach that “applies methods adapted from language-focused disciplines such as linguistics, communication, and rhetoric to the analysis of computer-mediated communication” and it is informed by a linguistic perspective as online behaviour is viewed from the point of view of language and language use.
CMDA is not a single method that can be applied to any study, but rather it provides a set of methods with which to make observations and interpretations based on empirical analysis (Herring 2004: 342).
6.3.2 Content analysis
Content analysis is used to systematically and objectively analyse the content of written data (Kyngäs and Vanhanen (1999), cited in Tuomi and Sarajärvi 2009: 103) but can also be applied to spoken, signed or multisemiotic data. This method aims to provide a condensed and general account of the phenomenon and it can be used even with unstructured data (Tuomi and Sarajärvi 2009: 103). This makes content analysis ideal for the purposes of the present study, because the data is quite unstructured and the focus is on arriving to a general conclusion instead of a detailed analysis of every aspect of the data. Compared to discourse analysis, content analysis aims to find the meanings in the data, whereas discourse analysis exposes how meaning is created in the data (Tuomi and Sarajärvi 2009: 104) consequently making it particularly suitable for the present study, as it is already known how meaning is created: by code-switching.
Tuomi and Sarajärvi (2009: 95-98) present three different analytical approaches that can be taken with content analysis: data-bound, theory-bound and theory-guided analyses, of which theory-guided approach is selected as the approach of the present study. The distinction between the three can be seen in the ways how the theory describing the phenomenon guides the collection and analysis of the data, and the reporting of the results (Tuomi and Sarajärvi 2009: 98). In the data-bound approach, the whole process of analysis is not influenced by the theory at all and all conclusions are purely gathered from the data. On the contrary, theory-bound approach is frame worked by an existing theory and the study is aimed at testing the theory in a new context. The theory-guided approach can be seen as being somewhere between the other two. It is connected to earlier theories, but it is not intended to testing the theories and the items that are analysed can be freely selected from the data, keeping in mind the theoretical framework. For the purposes of the present study, the theory-guided approach to content analysis the most suitable, because of its flexibility and openness for interpretations. Tuomi and Sarajärvi (2009: 97) discuss theory-guided analysis as being abductive, meaning that the analysis connects the theory with the observations, mixing together the deductive and inductive approaches and coming to new, inspired conclusions.
7 ANALYSIS
The detailed qualitative analysis of the findings is presented in this chapter. The data can be divided into two parts: the profile texts and the actual tweets. In this chapter I will first look at the profile texts somewhat briefly, and then focus more on the actual tweets.
The data consisted of 478 tweets that I divided into three categories according to the presence of English in the tweets. The first, and by far largest category included all tweets that had no English elements in them, 276 in total. This accounted for 57.9 percent of all tweets. Second category consisted of tweets that had some English elements mixed in with other languages and included 85 tweets altogether, which was 17.6 percent of all tweets. 117 tweets that had only English in them made up for the final category which accounted for the remaining 24.5 percent of all tweets. For clarity, these categories will be referred to as No English, Some English and Only English throughout the rest of the present study.
There were only a few cases where languages other than Finnish or English were used.
Swedish was used in eight tweets, so in about 1.7 percent of all the tweets. Swedish is the first language of 5.3 percent of Finns (Statistics Finland 2014), so the amount of Swedish on Finnish Twitter is clearly not representative of the number of Swedish-speakers in Finland. Additionally there were two tweets that used Russian and Sami.
In one tweet about a Finnish TV-show’s episode about Russia, a Russian word was added in the end of the tweet. Another tweet about an article on a Lapland based newspaper contained a headline in Sami. As the focus of the present study is on the use of English by Finns on Twitter, I will not go into any more detail about the occurrences of other languages in the data.
In the analysis I will be providing relevant examples of the data to illustrate the findings. I will present the whole tweet as an example with no changes made in the orthography or punctuation. For brevity, I will omit the links and pictures so often included in the tweets by marking [link] or [picture] in their respective place. A
description of the link or the picture will be provided in cases where knowing the omitted content is relevant to the understanding of the example. Emoticons and emojis are also used frequently on Twitter, so they also appeared in the data. Emoticons are made from symbols on one’s keyboard, such as using a colon and a parenthesis to form a smiley face, whereas emojis are small cartoon pictures that can be added to messages on many platforms, including Twitter. In the examples, I have included any emoticons that were part of the original example, but emojis are replaced by [emoji]
for clarity. A translation of the tweet when appropriate is provided under the example in parenthesis. The name of the user who has written the tweet will be presented at the beginning of the example accompanied by @ sign. Since all the tweets collected for the data of the present analysis were originally posted publicly, there is no need to censor the identity of the writer.