Abstract
Reading experience is viewed as ‘interactive and negotiable’ for different reading positions are created in readers’ responses to the same news report. To understand the differences between ‘preferred reading’ and actual readings, this article, drawing on the context models and the Appraisal framework, analyzes 785 readers’ comments attached to 23 hard news stories sourced from the China Daily mobile application (APP) and the People’s Daily Online website. The study combines corpus semantic tagging analysis for readers’ choices of evaluative lexis with a manual analysis of Appraisal resources for reading position construction based on a whole text perspective. The results show that readers are likely to use Appraisal resources strategically to defend their own opinions against other commenters or affiliate with large numbers of potential readers without commenting. Furthermore, comments were found to be directed either at the text or at other commenters, and through this a number of different reading positions were revealed. The readers’ context models are the cognitive link between discourse properties and social strategies in the process of news understanding.
Introduction
The people formerly known as the audience are those who were on the receiving end of a media system that ran one way, in a broadcasting pattern, with high entry fees and a few firms competing to speak very loudly while the rest of the population listened in isolation from one another – and who today are not in a situation like that at all.
The 21st century has witnessed a technologically afforded transformation of text–reader relations in mass media communication from a traditionally one-way pattern to online interactivity (Das, 2017), with ‘participatory features for audiences, such as reader comments sections’ (Bruce, 2018: 54) now available in many online environments. Before Web 2.0, ‘Isolation from the audience [was] a characteristic of mass communicators’ (Bell, 1991: 88). Audiences were not only isolated from the news media, but they were also separated from each other. Now, via the Internet, shared virtual communities (Rheingold, 1993) are formed for the public to express opinions in a way that they can interact with both news texts and other readers. This leads to more reading positions than ‘preferred reading’ (Hall, 1977: 344, italics in original) not only because ‘it is always possible to […] decode an event within more than one “mapping”’ (Hall, 1973: 13), but also because when readers comment, especially as a form of feedback, these could be ‘aimed at other viewers’(Matthewson, 1989, quoted in Bell, 1991: 88). This means that readings can comply with, resist, negotiate with and even deviate from a subject position established in the news and comments. Online reader comments sections, following news reports via news websites or news mobile APPs, create a virtual forum for the public to express and discuss their understandings or opinions on the associated news. This has the potential to result in the creation of different reading communities, or ambient affiliation (Zappavigna, 2011; Zappavigna and Martin, 2017), where the notion of ‘ambient’ means that ‘the users may not have interacted directly and likely do not know each other, and may not interact again’ (Zappavigna, 2011: 801). Thus, online readers’ comments may share some of the characteristics of imagined readership. How readers position themselves both in relation to the news report itself, as it is designed for a target audience, and in relation to other readers are the focus of this article, and to address this question I draw on the context models and the Appraisal framework to examine the evaluative resources put to use in forming these communities in the process of news understanding.
The actual products of mass media communication in the form of newspapers, broadcasts and TV programs have been the central concern of Critical Discourse Analysis (Fairclough, 1995; Van Dijk, 1988), Sociolinguistics (Bell, 1991), Social Semiotics (Kress and Van Leeuwen, 1996), Conversation Analysis (Greatbatch, 1998) and Discursive News Values Analysis (Bednarek and Caple, 2012, 2017). However, less attention has been paid to the production and reception of mass media communication. Focused on the reception end, this article draws on the Appraisal framework to examine how attitudinal resources are put to use in forming reading communities, in relation to the kinds of values the news organization wants readers to share around events and participants. Nevertheless, reader interpretation is a sociocognitive process and is based on a whole of text perspective (van Dijk, personal communication 5 April 2019). Through the mental interface of context models, readers adapt their comments to the social and communicative situations relevant for their news understanding. Appraisal resources in the comments, at the lexical-semantic level, are evaluative expressions of personal opinions, which are ‘internal, […] and mental, information’ (Van Dijk, 2014a: 51, italics in original) represented in the context models. This article adopts an integral approach of linguistic analysis and sociocognitive elaboration based on van Dijk’s theory of context in terms of the construction of reading position. Moreover, an investigation of readers’ comments on a given news text can provide new insights into the ways in which reading positions are established through the negotiation of values between the text and comments as well as among comments. As Bell and Garrett (1998) point out, ‘their (audiences’) evaluative reactions to discourse styles, their preferences for other variables, the situations in which they receive media texts, are crucial to our approach to media discourse’ (p. 18). This article uses a small data set of 785 readers’ comments on hard news stories in China Daily and People’s Daily (English language versions) to examine potential patterns of Appraisal linguistic items and reading position construction in the actual reading process.
Reading position
Contemporary studies of mass media audience date back to Hall’s (1973) Encoding and Decoding model, which is applied in Morley’s (1980) empirical exploration of two Nationwide program broadcasts. In his model, Hall identifies four kinds of viewer’s positions in television discourse: dominant code (which is an ideal-typical meaning encoding), professional code (taken by professional news broadcasters), negotiated code (which is a mixture of adaptive and oppositional meaning encoding), and oppositional code. Different from Hall’s ideological effects model, Richardson and Corner (1986), emphasizing attributions of the audience’s perception, propose an interpretative scheme of three-layer readings: transparency, mediation, and manipulative displaced readings, in order to understand whether viewers perceive directly, are mediated through other sources, or hold the complex reading position by manipulating other resources when they ‘make sense’ of a BBC documentary program. Bell (1991) recognizes the multiple roles and layers of audience for media language and describes the media audience as Addressee, Auditor, Overhearer and Eavesdropper based on whether they are known, ratified or addressed by the speaker, for example, auditors refer to those listeners who are known and ratified, but not addressed by the speaker in a speech. Richardson and Corner and Bell’s perspectives contribute to understanding how readers as participants are engaged in the reception end of the news production process. With regard to the meaning-making aspects of text–readers interaction, closely related to Hall’s model is Martin and White’s (2005) interpersonal topology of three reading positions, which is critical to understanding readers’ affiliation and their sharing of values through couplings of evaluative resources and ideational experiences (Knight, 2010) when they comment on news. A compliant reading, similar to the dominant code, refers to readers’ agreement on meanings in the text. In parallel with Kitis and Milapides’s (1997) definition of ‘the obliging reader’ and Eco’s (1984) term of ‘the Model Reader’, these four subcategories, whether it is named dominant code, Compliant reading, the Model Reader or the obliging reader, are oriented at those ideal readers in a ‘preferred reading’ position who read as designed. A Resistant reading, on the contrary, indicates that readers disagree on meanings in the text. This is what Hall defines as the oppositional code. A Tactical reading means ‘a typically partial and interested reading, which aims to deploy a text for social purposes other than those it has naturalised’ (Martin and White, 2005: 62), which is similar to Hall’s professional code. For example, doing discourse analysis of a news report is a Tactical reading by a professional, and it serves the purpose of linguistic research. Notwithstanding different perspectives of reading capacities and practices, all these reading position classifications listed in Table 1 are working under the same premises as Richardson argues in his reception analysis, in that ‘meaning is negotiated between a text and its readers’ (Richardson, 1998: 221). It is this ‘meaning negotiation’ not only between a text and readers but also between readers and readers that provides space for variable reading positions from different social and cultural situations. The manual analysis in the “Results and Discussions” section will put into application Hall’s and Martin and White’s models in comments on hard news stories and further explore more reading possibilities grounded in the Appraisal framework.
Reading position classifications.
In more recent audience studies, Smith and Adendorff (2014) combine corpus linguistics and Appraisal to investigate how different reading communities bond around the value of education through letters from tabloid readers and mainstream newspaper readers in South Africa. Kyriakidou (2015) proposes a framework for categorizing audience engagement with media stories as four types of (emotional) ‘media witnessing’ and examines this in an empirical study of Greek audience’s responses to human distant suffering. In his qualitative research into multilingual audience on Facebook, Androutsopoulos (2014) argues that ‘language choice, becomes a key resource by which to bring together or separate various parts of the networked audience’ (p. 71). Badarneh and Migdadi (2018) explore how readers use three discursive strategies to accomplish self and other positioning through 500 online news comments on Jordanian news websites. Using keyword analysis of online news articles and reader comments, Bruce’s (2018) findings show that news agents in Britain provide online environments for sharing stereotypical ideologies about minority religions through news commenting.
However, not much attention is paid to the difference in reading positions between the designed, or target readers and actual readers in real life. As Fairclough (1989) notes,
And since all discourse producers must produce with some interpreters in mind, what media producers do is address an ideal subject, be it viewer, or listener, or reader. Media discourse has built into it a subject position for an ideal subject, and actual viewers or listeners or readers have to negotiate a relationship with the ideal subject. (p. 49, italics in original)
Based on Bell’s speech accommodation theory, the audience accommodate when they agree, disagree or argue about text meanings designed for them. There are basically two ways for us to find out what positions the unknown and actual readers take up when they are reading: through audience interviews (Radway, 1984; Richardson and Corner, 1986) and examining readers’ comments (Bruce, 2018; Catalano and Fielder, 2018; Van Brussel, 2018). Readers’ online comments following the news report may be considered a form of direct feedback, and a reasonable source for investigating reading positions. In their investigation of dynamic reader positioning in a news text, grounded in Appraisal analysis, Coffin and O’Halloran (2005) propose that ‘the academic ideal would involve empirical research with actual target readers and how they interact with’ the text (p. 161). This article, inspired by these pioneering contributions to audience research in news discourse and drawing on the Appraisal framework and the sociocognitive slant, takes on the challenge of exploring actual reading position construction, text–reader and reader–reader interactions in hard news stories about China.
The context models and discourse comprehension
With reference to van Dijk’s sociocognitive approach, ‘contexts’ are defined as subjective constructs of participants, instead of verbal ‘co-text’ or objective situations (Van Dijk, 2009). Context models are socially based mental models representing the events and situations the discourse is about (Van Dijk, 2008, 2014b). They are the cognitive interface between discourse processing and social interactions, personal experiences and social structures. Linguistics ‘has largely ignored the complex underlying cognitive framework’ (Van Dijk, 2014a: 328), whereas this missing link between discourse and society is vital for illuminating discursive variations in similar situations. The main function of context models is to control discourse comprehension and production in an appropriate way for language users to adapt their texts to the communicative situation, as in the case for readers understanding of and commenting on the associated news reports. Other mental models, situation models, account for the semantics of discourse and explain the ways in which readers understand ‘the events texts are about’ (van Dijk, 2008: 58, italics in original). In line with the context models, scholars explore the practices of multidisciplinary discourse analysis in news translation (Chen, 2011), policy defense speech (Guillem, 2013), parliamentary debate (Cuenca, 2014), political discourse and communication (Ghachem, 2015; Kupolati and Boluwaduro, 2018; Trimithiotis, 2018). Little empirical research is related to the process of discourse comprehension and production in the field of news understanding. Comments are an explicit form of feedback in written language generated by readers to present their understanding of news texts, and, as such, carry both social and cognitive functions. As Van Dijk and Kintsch (1983) point out, ‘there is still a theoretical gap between a linguistic theory of language and discourse, on the one hand, and a theory of social interaction on the other, our cognitive model provides a potential link between these two theories’ (p. 19). The way that the context models influence readers is to construe situation relevant models for their understanding reflected in the comments of this case. At the same time, the semantic analysis can provide reflective insights into the formulation of readers’ mental models. The construction of reading positions in the 785 comments in this study is examined both in relation to the Appraisal resources and from a cognitive perspective within the theory of the context models.
Appraisal framework
The analytic tool for linguistic interpretation in this article is Appraisal, the discourse semantic interpersonal system developed by Martin and White (2005), through which we communicate feelings or emotions about people, things and activities (Martin and White, 2005). Appraisal is regionalized as three interacting domains: Attitude, Engagement and Graduation. Attitude is about emotional Affect, Judgment of behaviors and Appreciation of things; Engagement is about sourcing Attitudes; and Graduation is about grading Attitudes.
1
Attitudes may be inscribed by evaluative lexis explicitly, or invoked by experiential meaning that invites evaluation implicitly (Martin and White, 2005: 35–37). In terms of the relation between reading position and evaluation, ‘[…] a text naturalises a reading position we mean as far as evaluation is concerned that it will be fairly directive in the kinds of
This article is mainly focused on Attitude, or specifically what value positions readers bond with when they employ the evaluative resources of Affect, Judgment or Appreciation coupled with ideational experiences in their comments on a news story. To give a brief example, in the following excerpt, the reader comments on the news ‘China vows to take “countermeasures” against U.S. punitive tariff’:
Finally and officially, China is going to respond
The use of positive Judgment resources appropriately and correct demonstrates the reader’s support of the position being put in the text: that China should take countermeasure action against the United States, while the negative Judgment protectionism and Attitude hurts show the reader’s opposite position toward U.S. punitive tariff. At the same time, the reader is able to affiliate with other readers who share his or her value position and defend against those who may be on the side of the United States, thus form different imagined (Smith and Adendorff, 2014) or ambient (Zappavigna, 2011) online reading communities.
Data and methodology
Collected from the China Daily app and People’s Daily Online website, 785 readers’ comments, related to 23 news reports from 01 January to 30 June 2018, are composed of verbiage and image, including pictures, videos, poem, map, advertisements, emoji and links. All 23 news stories in this data set are about international conflicts, national issues and accidents (see Appendix 1), and are categorized as ‘hard news’ (Bell, 1991). Initially, 34 news stories were collected for their divergent values as expressed in the verbal and visual resources deployed in the texts, and subjected to discursive news value analysis, which is the work of another article and will not be discussed any further here. However, these texts do either construct the news event as positive or negative; thus it is possible to assess the main position being stated in the news story and the extent to which readers align with or against this position. In compiling this small data set, 23 stories were discovered to have comments associated with them, hidden behind a fire wall and out of reach for the readers (including the author) in mainland China. 3 Realizing their existence, the author decided to collect the comments associated with these texts to see in what ways positions are established toward China by readers outside of mainland China. 4 The 23 hard news texts from China Daily (http://www.chinadaily.com.cn/) and People’s Daily (http://en.people.cn/) are characteristic of divergent couplings of ideation and value, which are (1) positive image coupled with negative story, (2) negative image coupled with positive story or (3) positive image coupled with discursively positive story but foregrounded by negative attitudinal lexis. As such, these stories present what Bednarek and Caple (2010) call an ‘evaluative clash’ (p. 17). Thus, I also aim to test whether news reports with divergent values open up possibilities for more reading positions to be negotiated.
In discourse analytic research, Van Dijk (2014a) advocates a multidisciplinary approach and highlights that media linguistic analysis needs to ‘account for the various levels or dimensions of whole “texts”’ (Van Dijk, 1985: 1). From the cognitive perspective, the context models of readers control their understanding of news texts and subsequently adapt their comments to the particular communicative situations. As he indicates that ‘ideological positions […] can also be “signaled” through such apparently “context-free” language characteristics’ (Van Dijk, 1985: 4), reading positions, as ‘ideological positions’ influenced by the context models, can be signalled linguistically through the emotions and attitudes expressed in reader comments that are attached to specific social and communicative situations, on one hand. On the other hand, the mental models of readers are not only built upon the writer’s context models, shared sociocultural knowledge and the communicative situations, but also linked with their own experiences and specific goals, which monitor the appropriate representations of personal opinions in the comments. Referring to the discourse–cognition–society triangle (Van Dijk, 2014a, 2014b), the construction of reading position in this article starts from linguistic analysis of Appraisal resources for expressing personal opinions, and then relates the cognitive bridge of readers’ context models to the social strategies of aligning or dis-aligning with different reading communities.
Because Appraisal resources are largely context-dependent, manual analysis of attitudinal identification and categorization is assisted by corpus-based linguistic analysis using the semantic tagging tool, Wmatrix (Rayson, 2003), developed by Lancaster University (http://ucrel.lancs.ac.uk/wmatrix/). This software automatically annotates lexical words with semantic tags within the framework of UCREL Semantic Analysis System (USAS). There are 21 semantic fields with subcategories in the tagset, but this article is mainly focused on the categories of ‘Evaluation’ (A5), ‘Emotion’ (E), ‘Crime law and order’ (G2), ‘Warfare, defense and the army; Weapons’ (G3) and ‘Social Actions, States and Processes’ (S1), which are related to Appraisal lexis in the data. This semantic tagging method has been applied in a 36-million word corpus of news reports on Hurricane Katrina to investigate news value construction and provides useful insights into the whole picture of newsworthiness in large data (Potts et al., 2015). However, as Bednarek (2016a) points out, because of ‘the multiple meanings and uses of words, and the broad nature of semantic fields’, it is ‘necessary to combine quantitative with qualitative analysis’ (p. 253).
Results and discussions
Corpus analysis
In the corpus analysis of semantic tagging for the 25,634-word data set of readers’ comments, there are 13,895 tokens in total frequency overall of semantic tags. Images, links, abnormal or wrong spellings, and Chinese characters are identified as unknown words and thus excluded from the semantic tagging process by the USAS system. Attention is paid to the patterns of Appraisal choices in the frequency list and the appraised target in the concordance lines. The major findings are as follows:
First, negative Emotion (E–) tokens account for 1.12% of the total semantic number, which are twice as frequent as positive ones (0.54%) in reader comments. This result is similar to studies in Coe et al. (2014) that found ‘hard news’ topics invite from newspaper websites comments more incivility (impolite language), which ‘is associated with emotion’ (p. 674). Since the hard news data set studied here is characterized with divergent couplings of value and ideation, this may generate evaluative tensions between online readers holding different stances both with the news articles and with other readers. These occurrences of incivility and tension are semantically reflected in the Appraisal choices of readers’ comments. The semantic tags for negative Emotion category that tops the frequency list include threat, fear, retaliate, scared, concern, threats, bashing suffered, afraid and hate, and are used by readers to negotiate their reading positions with respect to the specific events or participants reported on in the news. The concordance lines of threat (see Figure 1), for example, with 18 occurrences, show that 12 of them are explicitly directed at China, which is a major participant in 10 news reports on international conflicts involving China, the United States and Australia. By extending the context in the Concordance lines, it is found that there are more instances where readers comment in a supportive way about China through ‘unconfirmed’ or ‘nonsensical’ attitudes toward the China threat (theory) than those who take the opposite view. Similarly, while the token like has the highest frequency (55 occurrences) in the list of positive Emotion (E+) in the data, 49 of these are grammatical prepositions functioning as introducing examples or similarity between participants and only 6 are used by readers to express their preferences or opinions in a negative way when collocated with the negation ‘not’, for instance, ‘I’m sure all would not like to be treated as some kind of mistrusted foreigners’ and ‘Even if you don’t like it’. This shows that reading positions realized through Appraisal lexis are highly context-sensitive and need to be followed up with a manual check of the data.
Concordance lines for threat.
Second, in the semantic field of Evaluation (A5), these tokens occupy 1.56% of the total frequency number in the data. It is interesting to find that most of the positive words (good, truth, better, great, positive, and true) are used negatively by readers in specific contexts to comment on participants or processes in the news stories, apart from a few negative Evaluations such as fake, wrong and worst. In Figure 2, the concordance lines of good show that 14 occurrences are targeted at news participants (China and the United States), news related topics (economy and policy) or other topics covered in the comments (option, evidence and your analysis). However, nearly all the instances of good are either used with the negation word ‘not’ (Good people don’t last long; It’s not a good option) or collocated with something negative (smear the good name; the absence of good evidence), which cancels the positive value of good in the clause as a whole. This shows that readers interact more often in a negative way both with the news texts and with other readers from the comments section in an imagined relationship. Again, manual analysis is necessarily complementary to corpus tagging methods given the context-sensitive nature of Appraisal lexis.
Concordance lines for good.
Third, in the data, tokens of war(s), military and invade(d), with high frequency in the semantic fields of ‘Crime law and order’ (G2) and ‘Warfare, defense and the army; Weapons’(G3), are associated with the happenings reported on in the news stories. Used in the comments, these are meanings ‘activated in the media texts’ (Van Brussel, 2018: 383) and show the interactivity between readers and the text. These meanings invoke negative emotions and are directed at trade wars or potential wars between the conflicting countries mentioned in the news stories. As for interactions between readers, there are two very negative tokens in the semantic fields of Social actions, states and processes (S1) and Religion and supernatural (S9): stupid and troll(s), which are directed at other readers or their ideas in the comments section. By exploring Concordance lines for these two tokens, it is found that readers are likely to use negative emotions when faced with disagreement or different opinions from other commenters.
When doing computational analysis for semantic tagging, two circumstances need to be taken into consideration: one is that some semantic resources, such as invoked attitudes in the form of word groups or phrases cannot be automatically annotated by the corpus tool. In the following comments from the data, negative attitude is assigned for mud in the brain (being stupid) in example (1) and let us down (disappoint) in example (2). The other problem is that ‘a given lexical item will vary its attitudinal meaning according to that context’ (Martin and White, 2005: 52), and manual verification is needed for an appropriate understanding of a reading position in actual situations. For instance, in example (3) no problem is as a whole positively charged attitude and glorification in example (4) is judged as a negative behavior for its target is war crimes and they are evil.
(1) It is you who has (2) We only turned to the East when you people in the West (3) Many years ago I had (4)
In short, the corpus analysis of semantic tagging contributes to the bigger picture of linguistic patterns of readers’ comments (summarized in Table 2). Appraisal resources in the field of negative Emotion are usually directed at news participants and used more often than positive ones; linguistic lexis expressing positive Evaluation tend to be used negatively in specific situations and target topics or ideas associated with the news expressed in the comments; some negatively invoked meanings in other semantic fields aim at other commenters or their ideas. These lexical patterns in terms of positive and negative Appraisal meanings are signals of the ideological positions taken by readers to defend against opponents and align with supporters. The underlying ideologies are controlled by the context models of readers to make their comments appropriate in the particular social and communicative situations. A manual analysis is essential for the concrete construction of reading positions through negotiating values in readers’ comments in actual reading process, given the context-sensitive and open-ended characteristics of Appraisal system (Bednarek, 2008) as well as the unique context models of different readers.
Linguistic patterns and evaluative targets in the semantic taggers.
Manual analysis
The target audience for China Daily and People’s Daily Online websites essentially includes millions of readers around the world. For instance, China Daily is designed for ‘high-end readers as an insert in mainstream newspapers around the world’ (http://www.chinadaily.com.cn/e/static_e/about). As national English-language newspapers subordinated to the China Daily Group, both of them play the role of ‘A Voice of China on the Global Stage’ (http://www.chinadaily.com.cn/e/static_e/about). People’s Daily has ‘258 million readers per day across 210 countries and regions around the world’ (http://en.people.cn/n3/2018/0706/c90828-9478507.html). Those who comment on the news on the website are international readers who are expected to read in English and hear the voice of China talking to the world. In all, 120 comments associated with 4 news texts, sourced from People’s Daily Online in the data, are employed for a manual Appraisal analysis to investigate reading positions and text–comment relations. The context models of readers represent the relevant communicative situations and guide their news understanding; however, the formulation of context models is complicated. In terms of sources, they are built upon personal knowledge and experiences, shared sociocultural knowledge, inferences from the news as previous discourse, and the news writer’s (old) context models. In terms of categories, or parameters of communicative situations construed in the context models, they are organized by a schema of Setting (time and place), Participants (roles, identities and relationships), (Inter)Actions, Knowledge, Intentions and Goals (Van Dijk, 2014a, 2015). The analysis of reading position in this article is only concerned about relevant discourse properties of the communicative situations of news understanding in terms of evaluative expressions in the reader comments. Before the analysis, a brief summary is given to each of the 4 news stories and their preferred readings in Table 3, in order to relate the ways in which the readers interact with the text and other readers in the comments. Thus, reading position construction will be discussed from two dimensions of text–comment relations and comment–comment relations.
Information about the four sample news stories.
Text–reader comments
Compliant reading
The first reading position investigated in the data is Compliant reading, which suggests readers ‘subscribe to’ the meanings in a text (Martin and White, 2005: 62). In this text–comment relationship, readers fit into an ideal subject position the news text has built into it and comply with the preferred reading naturalized in the news. The instantiated readings are presented in the following examples:
(1) There are Australian citizens of Chinese decent from the goldfield years, from the Mainland, Taiwan, from Hong Kong and also from SE Asian nations. All have made Australia their home and have been (2) (3) Glorification of What about your own reaction to this glorification of
In comment (1) the reader strongly aligns with the preferred reading of this text, which positively judges the Chinese while condemning the Australian government for its negative appraisal of the Chinese. This attitude is mirrored in the reader’s comment. The inscribed Judgment law abiding is directed at Australian citizens of Chinese descent from the goldfield years. This positive reading complies with the positive values constructed toward China’s peaceful rise in the news story. The negative Affect materialized by not like, negative Judgment expressed by mistrusted and espionage are also aimed at those Chinese people, who are supposedly related to the China threat, and viewed as an overreaction in the preferred reading. The target of the negative Judgment disgrace is ASIO (Australian Security Intelligence Organisation), a department of the Australian government, which claims China as the extreme threat in the news story. This attitudinal lexis exhibits the reader’s disapproval of Australian overreaction toward China and is still in a Compliant position with the preferred reading in the text. Comment (2) is a strong denouncement of the splitting forces in China and their supporters. The negative Affect realized by No leniency, relentlessly and negative Judgment committing terrorism are all directed at those in Xinjiang or Tibet who try to split the country. These attitudinal expressions show a supportive position with the preferred reading of strict policies on those split forces in Text 13. In comment (3), the reader criticizes the glorification of the massacre in the Japanese publication. Both invoked and inscribed negative Judgment realized in butchery, killing and repulsive, and vilification are aimed at the discriminatory contents of the Japanese comics. These evaluative resources all echo with the ideal position of no Japanese discriminatory book being published in China. From the cognitive point of view, compliant readers are those preferred readers whom the writer assumes about in his or her (old) context models for writing the news. They recognize the writer’s intentions through the compliant expressions in the semantics of comments. The Appraisal representations are used by the readers to socially align with other readers who are likely to share the similar opinions.
Resistant reading
In this text–reader relation, readers take an opposite position with the preferred reading naturalized in the news story. They object to events or ideas in the text. When readers ‘oppose the reading position naturalised by the co-selection of meanings in a text’ (Martin and White, 2005: 62), they are resistant readers. Or, in van Brussel’s conceptualization, it is a logic of dis-identification, which means readers ‘radically reject the discourses activated in the media texts’ (Van Brussel, 2018: 392). This kind of reading is less frequent than Compliant reading in the data, but they do occur in the following examples:
(4) It is China that is a And Australia is 100% The sooner our Subs are built the better. (02_PD) (5)
In comments (4) and (5), the accusation against China and the support for Australia are both stressed. The reader uses negative Judgment Threat and positive Judgment Right to target China and Australia respectively, which completely opposes the positive evaluation constructed for China’s peaceful rise and negativity toward the Australian claim of China as the extreme threat in the news story. In the same way, negative Judgment, expressed by Nobody…trust, directed at China in comment (5), deviates from the subject position of positive China designed for the ideal readers. With respect to cognitive processes in discourse comprehension, resistant readers reject the writer’s intentions conveyed in the news and their comments are contradictory to his or her context models. Readers use evaluative presentations to keep aloof from those who agree with the writer and align with their own supporters.
During the analysis of examples of Complaint reading and Resistant reading, attention needs to be paid to which specific participants, events or ideas in the news the readers comply with or resist. News stories may draw on and integrate multiple voices into the text, for example, politicians, professionals, quotes from other news sources or ordinary people, and so on (Bednarek, 2016b). This is particularly the case in the data sets studied here where divergent values, and more than one value position or standpoint in the news are instantiated. To make the situation even more complicated, readers partially identify and partially dis-identify with aspects of the preferred reading. In this case, they are likely to interpret the news stories in a negotiated way.
Negotiated reading
Hall (1973) describes this Negotiated position as ‘a mixture of adaptive and oppositional elements’ (p. 17). Targeted at particular news participants, events or topics in the news story, readers establish a polarized position and negotiate the pros and cons with the text. It is possible that the writer intends to leave some open space in the message where readers, as interactive participants in discourse comprehension, can voice their doubt or different opinions. They comply with part of the preferred reading while resisting other parts of it. Here are two examples:
(6) ‘ China should (7) Do
In comment (6), the reader seems to be discussing his or her opinions toward the actions of the Australian government and people in both countries. The negative Judgment wrong is directed at the description of Australia’s overreaction as hysteria or hysterical, showing that the reader opposes the preferred negative position of Australian overreaction toward China as hysteria. At the same time, this reader uses incompetence to target Australia’s hysterical exhibition, showing an adoptive position of Australia’s problematic foreign policy and overreaction. The reader further negotiates support for China and its relationship with Australia through positive Affect and Judgment resources of never fear, open and friendly. And clumsy is a negative lexis but used here positively in a self-mocking way toward Australians. In comment (7), the reader condemns the fabrication of foreign teachers’ diplomas, but the focus is put on the fabricators and judges instead of the teachers. On one hand, the reader endorses the ideal reading of foreign teachers’ qualifications through negative Judgment actualized by culprits, crime and fabricating, aimed at the two Chinese employees’ illegal activities. On the other hand, the reader uses positive Judgment not blame, directed at the Caucasians, or foreign teachers to reject the ideal position of suspicion over foreign teachers’ qualifications in the text. The reader further negotiates more negative Judgment of light sentence and light punishment toward the two fabricators. Different interpretations may influence the context models of readers, leading to deviations from the writer’s intentions. Controlled by their own context models, readers adapt personal opinions and negotiate with the text. In this way, more readers with different opinions are likely to be convened around specific values and positions.
Diffused reading
In their readings, at times, audiences do not just agree or disagree with the designed readership, they negotiate their reading positions beyond ‘an ideal subject’ (Fairclough, 1989: 49, italics in original) and bring in more readers to convene around other values/ideas. In this text–comment relation, other participants or events extended from the news per se are involved:
(8) English is a universal language, but I know so many that get along just (9) I
Comment (8) is not concerned with the diploma fabricating crime or suspicion over foreign teachers’ qualifications; instead, the reader focuses on speaking highly of English as a useful universal language. By using positive Judgment fine and Appreciation helpful, the reader presents a positive position toward English and those who use it, which are not directly related to the news topics in the text. In comment (9), the reader shows his or her emotions through repeating the Affect token of love which is directed at another commenter, the Chinese President, China and everybody. None of these are the target of the Japanese controversial novel with discriminatory contents in the preferred reading. Both of these two comments reach out to evaluations beyond the naturalized reading position designed for the target readers and diffuse other values or ideas into the imagined reading communities. These two might be considered diffused readings. The reader’s comprehension of the news and opinions in their comments has no direct connections to the intentions sent from the writer. Those diffused ideas or topics about ‘English language’ or ‘love’ in the examples lead other readers to socially shared attitudes or ideologies which are not presumed in the writer’s context models when writing the news.
Constructive reading
Apart from agreement or disagreement, opposition, and extension from the preferred reading for the news, reflective readers present their suggestion or solutions to issues reported on in the news:
(10) Why not the Chinese Education Department makes a bigger and (11) The
In comment (10), the reader does not express his or her standpoint on foreign teachers’ qualification or diploma fabricating behavior in China. Instead, the reading position is pointed to the Chinese Education Department and the reader uses positive Judgment expression of serious to propose that the Department should take the responsibility and deal with the issue. In comment (11), the reader does not comment directly on the controversial Japanese novel or its discriminatory content, but offers alternatively appropriate, better and profitable attitudes which express positive Appreciation and aim at his or her suggestion about responses to the discriminatory Japanese novel. At the same time, the reader points out that yours (probably the response mentioned in the news) and the government’s response are ridicule and bad, which are negatively judged. These reflection-on-the-news readings are not often found in the comments. Whether targeted at the news or other commenters, they propose a reading position toward problem-solving out of a constructive starting point. The readers take initiatives in their context models of news comprehension and represent their constructive ideas in the comments, in order to interact with the writer and other readers in the communication.
Reader–reader comments
While the first five types of reading, taken as the first layer of positioning, are situated in text–comment relations, the alternative reading to be accounted for here exists in comment–comment relations. In this case, readers do not comment directly on any participants or happenings within the news, but comment on other readers or their comments:
(12) (13) It looks like I must have touched one of your (14)
Comment (12) is an agreement with another commenter in the readers section. The reader at first, through positive Affect of Thank you, aligns with Mr Smith, who also comments under the same news, and then judges his comment as positive to show their agreement on the negative Judgment embodied by ambiguity and not…tolerant about China’s relationship with Islam. In this way, the reader poses a Compliant reading position toward Mr Smith and his comment. Comment (13) shows the ironic attitude of the reader. Through the negative Appreciation expressed in raw nerves, the reader establishes a Resistant reading position directed at another commenter. Comment (14) delivers a positive reaction to another commenter along with a suggestion as an English language user. The reader uses positive Affect item Thank you to establish a Compliant reading position toward another commenter, but then takes a Constructive position that his or her not good English needs to be improved.
In summary, these alternative readings can be viewed as a second layer of positioning. They take place between readers forming comment–comment relations, instead of text–comment relations. In this layer of comments, readers may form, as in the first layer, Compliant, Resistant, Negotiated, Diffused and Constructive reading positions toward other commenters and their comments.
Conclusion
One way to assess readership is to collect statistics about readers’ social status, educational background, income and gender, by showing the source of the target audience through demographic and ethnographic research. This article adopts an alternative approach: to construct readers’ positions through value negotiation with news stories in readers’ comments. It is impossible to know who are commenting behind those user names in the virtual communities of online news readers; however it is their evaluative stances established toward specific subject positions in the news that offer a lens on what reading communities the news stories are trying to affiliate with and in what ways readers respond to the bonding invitation from the news organization. As Bell (1991) highlights, audience kind is more important than audience size. Using a combination of corpus method and qualitative manual analysis, this study discovers some attitudinal patterns associated with the appraised targets in the comments and different types of reading positions constructed in text–comment and comment–comment relations.
The investigation of reading position in this article integrates linguistic analysis with sociocognitive perspective in the process of news comprehension. The theory of context models contributes to mediating between the discourse properties like Appraisal lexicons construed in readers’ mental models and the social strategies of defending oneself or aligning with others when interpreting readers’ understanding of news reports. The corpus analysis using the semantic tagger shows that readers employ negative attitudinal resources in the semantic fields of Emotions, Evaluations, Crime law and order, Warfare, defense and the army; Weapons and Social state, actions and processes to construe an opposite reading position directed at news participants responsible for the negative news story and other commenters holding different opinions, in order to bond with or distance from different reading communities. Whereas readers employ positive Appraisal choices in the field of Evaluation, negatively to defend against opponents or positively to affiliate with other commenters as well as other potential reader communities. Corpus software tools like Wmatrix are very useful and efficient technologies to deal with data processing and semantic annotation. However, in the linguistic analysis of Appraisal resources, the context-dependent meaning making nature of the discourse semantic system cannot be neglected. A manual analysis is indispensable for a more comprehensive picture. This article has revealed two layers of five reading positions in the 120 comments linked to 4 news reports in the data analysis: Compliant reading, Resistant reading, Negotiated reading, Diffused reading, Constructive reading in text–comment relations and in comment–comment relations. Interestingly, there are no comments on any of the visuals in the 23 news reports that made up this data set.
Technical affordances facilitate readers to interact with news stories and other readers in diverse ways. They can express their opinions on the news, discuss with other commenters and extend to more ideas and values. This article has explored news–reader relations from the perspective of the interpersonal dimension, to elucidate affiliation through different reading positions shared in the comments on news. The data in this article are quite small-sized and invoked Appraisal resources have not been analyzed in a systematic way, which is important for getting at a more complete picture. There may be more reading varieties to be found in future empirical research on news circulated via social media.
Footnotes
Appendix 1
Acknowledgements
I would like to express my special gratitude to Dr. Helen Caple, who kindly provided me the opportunity to work with her and, with her great insight and inspiration, mentored me in this article via the International Research Practicum at the University of New South Wales in Sydney from 17 July to 18 December 2018. I would also like to thank Prof. Teun van Dijk very much for his thought-provoking comments on the earlier version of this article.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship and/or publication of this article.