APPRAISING TWITTER USERS’ ATTITUDE ON “WRITING LAST LINE STORY”: SFL APPROACH ON CLOSING STATEMENT IN CYBER LITERATURE

: Twitter, an online platform that is said to have a more constructive and meaningful response to the issues, is also where people are creative writers of cyber literature. Lately, there has been a trend of Twitter users to write fiction lines that have a premise of “what is your last line if you in the ending of a relationship.” This research intended to seek the attitude of Indonesian writing fiction in English. This paper analyses a short written online discourse called ‘tweet’ using Systemic Functional Linguistics (SFL). This analysis focuses on appraising attitudes that are feeling on reacting to emotion, putting the judgment, and evaluating things . The data is taken from a Twitter account, namely @literarybase, by shorting 100 popular tweets taken to seek the trend and 20 tweets among them to be classified on its appraisal criterion. The results show that a relationship's trend ends with a sad ending rather than a happy ending. However, positive affection holds the most significant shares of lexical choice, followed by judgment and appreciation the least. That means the users mostly used ironic writing that contrasts expectation with reality. Linguistic research also implied that tragedy plot is favorable in online discourse. The study has identified the practical way of assessing attitudes in short-writing fiction on online platforms and the tendency of a community to see value in a relationship and how they express themselves in a foreign language.


INTRODUCTION
Social media has become a platform that cannot be parted from daily human interaction, which means language plays an important role. There is a trend that everyone has a desire to do things online (Smith, 2012). Due to that fact, it cannot be ignored that people also shift their linguistic PJEE Afriliani
The research on Twitter users likely analyzes social opinion about issues, which contains hundreds of experiments on it (Cortis & Davis, 2021). Given the example of the previous study, Falk (2017) analyzed the dialogic strategy with a qualitative method, finding that around 80% of users on Twitter update on news and others, creating a personal image of gaining power.
Of course, only some things evolve; social media is good. Yoo et al. (2014) stated that social media could be a disaster for society; they surveyed adult Twitter users to get their social image and identity using a quantitative method. The result is that acceptability and conformity on Twitter are seen as a burden that puts pressure on users since external factors are out of hand.
These, directly and indirectly, influence the users' writing to fit society's values. It shows how much like, retweet, quote, retweet, comment, and event interaction outside the main thread that forms norm and value, called sentiment (Khoo et al., 2012).
Exclusively on Twitter, human interaction that reflects on language use is openly accessed by everyone. Twitter is social media that lets users type relatively short discourse in every microblog/tweet. The Twitter algorithm usually presents the older tweet as the latest or the most popular to less famous reply. It is more organic, spreading mouth-to-mouth and immediate frank information (Bagley, 2012). Generally, researchers try to find a better way to assess social media behavior for each medium to differentiate consumers, platforms, and motivation for using it (Yap & Ng, 2018). Unlike other media platforms that can be set on exclusively for a group of people, Twitter users tend to make their profile public since users' anonymity is high. Anonymity secures people free of speech and confidence (Crow & Wiles, 2008).
By the polarity of 'like' and 'dislike' or 'agree' and 'disagree,' this research will likely dig into how to appraise precisely literature work on Twitter. Appraising user writing has been done in several ways, primarily in non-fiction writing, opinion, fact, and rarely in fiction. It was appraising trends like security on critical discourse analysis aiming to search media point of view on an issue (Afriliani, Yuwono & Kushartanti, 2018). A study by Neviarouskaya et al. (2010) assessed social opinion by data based on quantitatively categorizing words by SFL criterion.
Meanwhile, on Twitter, users' point of view is usually more organic and natural in lexicality on
The challenge in appraising from social media, Twitter, is difficulty finding the background story of the text and assessing context because of the limited character (Falk, 2017).
The previous study used the affect analysis model that mapped lexical choice (Neviarouskaya et al., 2011). When doing this research, the belief of lexical alone cannot classify the attitude, so we need context even in the text's minimal background story.
The context in analyzing discourse varies depending on its social function of it. In Systemic functional linguistics, context is divided into episodic, progression, and exchange patterns (Halliday & Matthiessen, 2013). It means that recognizing the contexts for discourse analysis are numerous and various. For example, context on educational, social, literary, political, legal, clinical, and so on (Cortis & Davis, 2021); in all these, the text may be analyzed as on its variety or register (Garrod & Terras, 2000).
Given the focus within systemic functional linguistics (SFL) on language as sociallyoriented semiotic, the study of context and lexis are worlds apart. However, they are not (Fontaine, 2017). We tend to think, metaphorically, that a word is a small unit and language is a much bigger unit, but this may not be the case. As in SFL, analyzing a word is an effort to find potential meaning, identify a gap, and represent a dimension (Halliday & Matthiessen, 2013).
Referring to the previous studies, the analysis of tweets by SFL in this discussion will cover lexical choice and consider the context involved. This research focuses on qualitative research on some sample tweets of writing short fiction. Using SFL, this research was arranged based on the assumption that lexical choice and context are not separated items (Fontaine, 2017;Glanzberg, 2002). The study implication is a starting point of SFL research on Twitter users' writing on appraisal broadly and systematically on the ideational level of analysis on fiction. This paper's goals of investigation are (1) to find writing trends on Twitter about the topic matter and (2) to find the result of language attitude on those writing. The result presents on description, grouping, and table to visualize and describe a whole finding. Thus, the research

Design
Research is qualitative and descriptive, with the steps of collecting, categorizing, and analyzing (Winda & Dewi, 2016). To be more precise, as qualitative, this research belongs to discourse analysis that aims to evaluate how the author expresses and manipulates feelings and emotions in fiction (Simmons, 2018). To sum up, the design aims to find collective data on the trend of writing and attitude.

Participant
The participants are online Twitter users that rely on the primary tweet. The chosen participants are being authenticated by visiting each personal page and ensuring they are personal accounts, not automated, Bot (robot), or scam accounts, as these types of participants are not legit to obtain. The participants are Indonesia twitter users that write their fiction in English.

Instruments and Types of Data
The instrument is a human feeling evaluation by the category of happy and sad. Then, further by appraising attitude using the SFL approach in the tweet or microblogging written text. Data are open-source tweets gathered from a Twitter community. The data are collective tweets that are accessibly open to everyone in terms of twitter privacy. The tweets are copied singly before being analyzed. The tweets divide into Indonesian and English, but this research only chooses English replies with fixed criteria. Since Indonesian users are using English as a foreign language, they may need help finding it. With that consideration, we ignore the grammatical error as they are not native, like many SEA users, yet more concerned with meaning (Song, 2010).

Data Collecting Technique
The tweets were collected and shorted by a specific criterion. The criteria are the popular tweets defined by reply and like, appearing forward on the reply section, and must be in English.
After categorization, the data must be evaluated and made into a tabulation by the result. Data

Data Analysis Technique
In qualitative, Assessing SFL is mainly about nominalization, grammatical metaphor, thematic structure, and qualitative data . Then after categorizing, selected tweets will be appraised with the SFL approach. The appraisal uses three ideational semantic discourse analysis groups: effect, judgment, and appreciation (J.R. Martin and P.R.R. White, 2005).
We took the 100 most popular tweets to find three options for the story's ending: "if you were to write a book of two people of fell out of love, what would be the last sentence?". The three endings we provide are happy, sad, and open end. Happy and sad are natural meanings in a movie (Fiorelli, 2016), while open is a condition that we cannot classify the feeling of both.
We justified them as happy endings while they end up together happily, as sad endings while they are broken-hearted, and as the open ending. At the same time, we cannot identify the polarity and leave it as many possibilities happened. Open-ending classification exists since the discourse lacks a background story, and we have a tiny context to decide.
From about 100 tweets (2900+ words), we eliminated duplicated tweets with low engagement that were not qualified for appraising. So, it left 89 qualified and measured tweets.
Then, about 20 tweets are taken to be analyzed on the SFL-ideational level. Ideational or shared experience analyzes subjective opinions on define them in a more measurable grouping. The appraisal is the evaluation system that focuses on mapping the possible meaning (Hood, 2012).
Appraising explores and classifies semantic resources; here, we take as sample data to respond to affect, weighting judgment, and present appreciation. (J.R. Martin and P.R.R. White, 2005) (Afriliani, Yuwono & Kushartanti, 2018). Then with attention to the 100 samples and tapering to 20 samples, this research compared the gap of evaluation on them and presented model analysis by three selected tweets. Further, the article's writing employs an automatic system using Mendeley Cite as applied in the previous study (Turmudi, 2020). There are two tables to help with the investigation or analysis. Table 1 Table 2 shows how to classify a sentence on appraisal. In this research, only three tweets take SFL appraisal classifying. They are only three because it just serves as an example of how categorizing steps goes on. In table 2, you may find sentences in some brackets and lines that separate each part to make it easy to assess. Mood comes before residue; meanwhile, the end of the story can sometimes be concluded among the last lines (Halliday & Matthiessen, 2013).

Categorizing the trend
Before explaining the finding of this research, several previous research about appraisal had been published. Scholars said that evaluating the lexicon can assess opinions, feelings, and different attitudes and tendencies rather than judge the text by its topic. The research also points out the helpfulness of having polarity and an SFL appraisal approach to evaluating word choices (Argamon et al., 2009). While they worked more into general and quantitative to categorize data as low, median, high, and max, in our research, we conducted more compact measurements on qualitative data to be clustered as happy, sad, and open end. (Argamon et al., 2009) The decision to label the data depend on the characteristic of discourse.   (62), and open endings (12). Here is an example of each end:

Appraising the tweets
Entering the appraisal analysis, we would like to highlight previous findings on several evaluations and its tendency. In security critics' text (Afriliani, Yuwono & Kushartanti, 2018), the result significantly weighed the judgment of discourse and very little appreciation.
Meanwhile (Khoo et al., 2012) distributed an appraisal analysis of news sentiment on USA policy, the president, the economy, and the Iraq war. (Khoo et al., 2012) The highest attitude is the appreciation of questioning the quality, about 16,9%. The two are projecting different results on different discourses, and by that previous study, we expect this research explains the different trend in its' discussion.
Appraisal begins by taking the top 20 tweets that received impressions, such as the most to receive likes and retweets. It assesses effect, judgment, and appreciation. From the data, the tweets are composed mainly of effect (47), followed by judgment (27), and appreciation (17) for the least. The positive traits often appear as desire, happiness, capacity, tenacity, and reaction. By these, it can be described that a relationship depends on how happiness is built.
Desire is longing, capacity is persisting, and tenacity supports each other. The norm of these is people feeling secure in the relationship when they feel happy enough, feel excited and desire 1. Happy ending: "thank you for choosing me to be one of the happiest people on this earth." 2. Sad ending: "meeting you was a nice accident, but this accident damaged me a lot."
This step tries to have language awareness of who, whose, or who is doing in discourse, especially in non-fiction. Take an example from work (Simmons, 2018). While identifying the lexicon is essential, acknowledging the people or things that got the things done is also a skill for the reader. For example, fiction by Shakespeare and J.K. Rowling is done to critic linguistic resources, in other words make the discourse convincing (Simmons, 2018).
In SFL, the appraiser gives the quality, and appraised is someone or something to be qualified. Table 4 has the three most popular tweets being evaluated. They are three appraisers: Authors A, B, and C. The authors write three pronouns: they, we, and he. The fact that they are plural more than only means the ending of a relationship depends on both sides. Metaphors are always invented on these data, such as color and inkless. Writing fiction is famous for accompanying metaphors and comparisons, proven here. Color is defined as positive happiness, while inkless refers to an adverse reaction. Even though the background story is limited, an appraisal helps the researcher find a comparative discourse context, such as in the color and inkless case.  Three examples are presented to enhance comparison on discourse analysis by using SFL appraisal. If you look closely at example one, the mood for the first and second clauses is contrasting. Example A is an example of how a sad ending does not happen suddenly; it may come by happy sequences of events but unfortunately end badly. It also happened in example

Irony and tragedy on the trend
The trend of the analyses had been done. Appraisal in many modes can explore a different side of the discourse. Here, short writing fiction on Twitter is a great example to investigate. Moreover, as it has been mentioned that Twitter is usually more about writing nonfiction than fiction, this data represents that writing fiction also excites the users. In writing fiction, everyone is positioning themselves in the center of the universe or looking into their interest and representing their identity (Simmons, 2018). It primarily influences the writer's egocentrism and ethnocentrism (Goudsblom, 2000). There is no guarantee that the majority of lexical choices of something positive will give the mood of discourse positive and vice versa. One lexical choice can make the plot that had been built different and make the ending sad. It can happen when a writer uses irony in his style of language. The irony is a disagreement between what is stated and what is meant (Tavadze, 2019). For example, in Table 6. Ironically, the accident associated with unfortunate things was valued as nice at first but then contradicted as a damaged one. Finally, to answer the research question, there are two answers to explain. First is the writing trend in this case, and the investigation found that Twitter users write sad endings rather than happy and open endings. Second is the appraisal attitude toward the discourse. A deeper analysis exposed that attitude-affect dominated the lexical choice of stories, and the choice of appraised is likely third person's point of view. It means a stereotype that relationship literacy about affection is proven true in this research. To sum up, the appraisal approach makes it easier for linguists to investigate the discourse in many layers of investigation and reflects the social tendency of language use.

Conclusion
Appraisal in the SFL approach provides tools in different modes to evaluate language.
In Twitter, this tool can be applied to fit into which part of language evaluation the researcher wants to study. This research found that analyzing discourse from big data that simplify to more minor data can always find something new to assess. The study identifies that the lexical choice of the writer tends to be positive, but the end of stories is a sad ending. The power of lexical choice at the end of discourse plays enormous momentum in deciding whether one is a happy ending or not. This paper also exposed that the trend of cyber literature about relationships and love nowadays seems pessimistic about happiness by using ironic language and producing tragedy plot. Further research suggests that the study explores meta-analysis on SFL that can be beneficial for more linguistic investigation layers on discourse and social aspects of cyber literature.

Implication
The Systemic Functional Linguistic (SFL) study in online literature implies that writers mostly choose affection lexically in writing fiction about relationships and prefer open endings with less contextual information. Moreover, online literature on fiction turns out to be an organic piece of work for Indonesian writers as the media to practice their talent and express their feeling. It also implied that Indonesian writers with a literacy Twitter base tend to write the irony type of sentences and tragic plot endings.

Limitation
This study was limited in using SFL on appraising attitude and plot. The Data are conducted on open-source Twitter access means that a private locked account will not measure. Grammatical error is ignored since the writers are Indonesian, that being assumed to have English as Foreign Language.

BIO-PROFILE:
Afriliani holds a bachelor's degree in English Language Education from Universitas Negeri Padang and got master's degree in Linguistics, concentration in Language and Culture at Universitas Indonesia. She is currently a lecturer at Universitas Terbuka, department of English Literature, translation study. Her research scale is discourse analysis, translation, culture, and English education.