How Did the China Daily and the New York Times Frame the 2016 G20 Hangzhou Summit?

CHAPTER 1: INTRODUCTION

1.1 Motives

Most people understand unobtrusive issues which have little to do with the general public “less through direct experience or past education than through the filter of journalistic language and imagery” (Walgrave and Aelst, 2006; Nelkin, 1987, p2), so news articles “are textual and visual structures built around a central axis of thought, from a certain perspective” (Ardèvol-Abreu, 2015, p424). Along this line, news could be regarded as a social and cultural product, which may embody the ideological beliefs and news-making process rather than an objective reflection of reality (Fairclough, 1995; Shoemake and Reese, 1996). Fowler (1991) also argues that news does not release the message objectively and everything in newspapers about a specific issue would be introduced from a particular ideological position. Although journalists claim to be objective in selecting and presenting the news, it is quite normal that many stories would possibly be shown differently in different countries, especially from particular angles or under the influence of a host of political, economic and ideological factors (Herman and Chomsky, 1994; Volkmer, 2012; Akhavan-Majid and Ramaprasad, 1998). Therefore, the term framing is proposed to be a proper way to depict the structure and power of news coverage (Tong, 2006).

The same media may frame an issue or event differently (Entman, 1993), so it is natural that different news outlets may employ different news frames to an issue. Especially when one Chinese mainstream medium-the China Daily and one U.S. important medium-the New York Times cover an event, the framing results may be quite different due to huge differences between them. In fact, previous studies have found many events referring to China are reported differently by the Chinese media and the U.S. media. For example, Chinese journalists presented the Tiananmen Square Incident as “a counterrevolutionary turmoil masterminded by a bunch of evil-minded people”, while the U.S. media framed the incident as “a democratic movement” (Chang, 2013, p254). Moreover, events taking place outside China could be framed differently. For example, frames of negotiation and compromise were applied by the Chinese media to report North Korea’s nuclear test, while Associated Press used the frame of war on terror (Dai and Hyun, 2010). Therefore, as an important world-class economic summit, the 2016 G20 Hangzhou Summit (hereinafter “the Summit”) could be a further indication of how the biggest developing country and the biggest developed country attracted worldwide attention and how the Chinese media and the U.S. media framed the issue.

1.2 Research questions and values

This study aims to compare how the China Daily and the New York Times framed the Summit, to identify similarities and differences in news frames and dominant frames and then to explore and analyze potential factors influencing the differences in the frame-building process of dominant frames of both news outlets. The following questions would be answered.

RQ1: how did the China Daily and the New York Times frame the Summit?

RQ2: What were dominant frames used by the China Daily and the New York Times?

RQ3: Were the prominence and distribution of dominant frames applied by the China Daily and the New York Times similar or different? If there were significantly differences, which factors have influenced the frame-building process of dominant frames of both news outlets?

In order to answer those questions mentioned above, framing related theories and potential factors influencing news frame-building such as ideology, media system, news sources and stereotyping, will be studied. The method of a combination of the quantitative content analysis and qualitative framing analysis will be adopted to examine the prominence and distribution of news frames and dominant frames and then to analyze the potential factors influencing the frame-building process.

Compared with agenda-setting and priming, the term framing has become more popular in communication study articles in the past decade with about 165 articles from 2001 to 2005 (Weaver, 2007). Therefore, framing theory is quite suitable to analyze communication in the mass media (Vliegenthart, 2012). However, based on the descriptive study of frames, researchers mainly explore the importance of framing effects rather than the causes of framing (Vliegenthart, 2012). In a similar vein, De Vreese also argues that frame-building ‘‘is part of the framing process that is often said to be important but is rarely studied’’ (2012, p366). Because frame building research has much room to further, this study, at its modest attempt, would endeavor to develop the theory in this aspect.

On the other side, up to now there are quite limited framing studies on the G20 Summit (e.g., Douai, 2014; Kutz-Flamenbaum, Staggenborg and Duncan, 2012). Most of these studies focus on the outside story of the summit in Toronto and Pittsburgh – the relations between police and protestors rather than the meeting itself, which mainly present how media framed the conflicts between them. What’s more, it seems that there is no relevant study by now on the framing analysis of media coverage of this event in China, who hosted the Summit in 2016. Therefore, this research also tries to fill this literature gap and to help readers better understand how the Chinese media and the U.S. media frame such global issues as the Summit.

1.3 Research structure

This study is composed of 5 chapters. Chapter 1 offers a general introduction to the study. Chapter 2 consists of the literature review of framing theories, which could be a theoretical foundation for the study. Chapter 3 depicts the research method which would be adopted to analyze how newspapers in China and the U.S. framed the Summit. Chapter 4 mainly deals with the findings and related discussions. Chapter 5 includes the conclusion and limitations of the study, and then gives some recommendations for the future research.

CHAPTER 2: LITERATURE REVIEW

This chapter reviews the literature relevant to the research topic and questions proposed in Chapter 1, which provides the background and serves as the theoretical foundation for the thesis. Three subsections are included in this Chapter. The first subsection is a review of literature on framing theory and news frames. The second subsection introduces the typology of news frames. The third subsection mainly discusses potential factors influencing the frame-building process, including ideology, media system, news sources and stereotyping.

2.1 Framing

The concept framing provides good theoretical foundation to analyze “the mechanism that construct news messages and subsequently influence what people learn from the media” (Tong, 2006, p4). Against this backdrop, this study mainly applies framing and relevant theories as the theoretical foundation for exploring news coverage of the Summit. Therefore, it is necessary and useful to review relevant theories in this field.

2.1.1 Framing theory and news frames

“The general assumption of the framing theory is that context informs our action, behavior and understanding” (Bryant and Miron, 2004, p693). Bateson (1972/1987) firstly uses the term frame in a psychological essay. He argues that frame is a psychological concept, but “gives the receiver instructions or aids in his attempt to understand the message” (Bateson, 1972/1987, p193-p194). However, the theoretical foundations that perfect framing theory lie in interpretive sociology, “which considers that people’s interpretation of reality and everyday life depends fundamentally on interaction and the definition of situation” (Ardèvol-Abreu, 2015, p427). Goffman (1974) introduces the concept and theories of framing to the sociological realm and holds that people organized their experiences by framing. He defines the frames as “the principles of organization which govern events-at least social ones-and our subjective involvement in them” (Goffman, 1974, p10-p11).

This expansion of framing enables scholars to use the theory to study news coverage, “when it is considered that the media have a great capacity to generate and modify the social framework of interpretation, by intervening in the creation of a shared social discourse” (Ardèvol-Abreu, 2015, p428). Tuchman has been regarded as the first scholar who used framing theory to study news coverage (Akhavan-Majid and Ramaprasad, 1998). She (1978) regards news as a window to the world whose frames such as size, shape and position can influence people’s perception of reality. Since then, a great number of scholars have been inspired to develop the theory within the mass communication field, which has been generally divided into three paradigms: cognitive, constructive and critical perspectives (D’Angelo, 2002). In the cognitive paradigm, scholars prefer to study the relationships between news frames and individuals’ experiences, values and knowledge (Hardin and Whiteside, 2010). The study made by Tversky and Kahneman (1981) is the most notable cognitive research, which focuses on the effects of presenting a story in different ways. From the cognitive perspective, Rhee refers to the news frame as “a combination of the textual features operating at the initial level of news interpretation where the textual features set limits on the use of knowledge” (1997, p28). This means that the cognitive study mainly uses “most often experiments to look at impact framing has on individuals” (Vliegenthart, 2012, p938). Constructionists argue that media discourse could be viewed as “a set of interpretive packages that give meaning to an issue” (Gamson and Modigliani, 1989, p3). Within the package, the frame has been termed as “a central organizing idea or story line that provides meaning to an unfolding strip of events, weaving a connection among them” (Gamson and Modigliani, 1987, p143). The constructivist approach offers “the most elaborate account of what a frame actually is and how different frames arise” (Vliegenthart, 2012, p937). Researchers working in critical paradigm argue that news frames are outcomes affected by interaction among political, economic and cultural power (Hardin and Whiteside, 2010). In other words, this approach “focuses mainly on the questions of variation (or lack thereof) of framing” (Vliegenthart, 2012, p937). After researching on U.S. news coverage of the antiwar movement, Gitlin falls in critical paradigm and defines news frames as “persistent patterns of cognition, interpretation, and presentation of selection emphasis, and exclusion, by which symbol handlers routinely organize discourse, whether verbal or visual” (1980, p7). Based on his extensive work in mass communication, Entman goes further and argues that framing is a process “to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described” (1993, p52). Entman proposes the cascading activation model which “explains how interpretive frames activate and spread from the top level of a stratified system (the White House) to the network of nonadministration elites” (2003, p415) and argues that “framing entails selecting and highlighting some facets of events or issues, and making connections among them so as to promote a particular interpretation, evaluation and/or solution” (2003, p417). Therefore, Entman’s framing approach is “‘top-down’ and clearly stems from critical paradigm” where “frames are the outcome of news-gathering routines by which journalists convey information about issues and events from the perspective of values held by political and economic elites” (Samoilenko, 2017, p83; D’Angelo, 2002, p876). In addition, in recent years frame variation has become a popular topic among studies based on the critical approach and researchers have explored “differences in framing, both over-time, as well as across outlets and countries” (Vliegenthart, 2012, p941).

This study focuses on the frame variation in different news outlets in different countries, which is situated in the critical paradigm. Therefore, Entman’s definition of frame will be applied to this study, which has been not only the most widely cited in current studies of mass communication (David et al., 2011), but also verified repeatedly by media scholars (e.g., Akhavan-Majid and Ramaprasad, 1998; Tong, 2006).

2.1.2 Difference between framing theory and other relevant theories

With the consideration of origins, process and effects, framing theory has close relations with agenda setting theory and representation theory (Scheufele, 2000; Avraham and First, 2010). Therefore, in order to better conduct this study, it is important to distinguish between all of these three theories.

Agenda setting theory refers to the “ability to influence the salience of topics on the public agenda” (McCombs and Reynolds, 2009, p1), which was first proposed and proved by McCombs and Shaw’s articleThe agenda-setting function of mass media in 1972. In other words, the mass media can emphasize the importance of the issue and set its agenda (Baran and Davis, 2009). Therefore, it can be easy to identify similarities between agenda setting theory and framing theory. Both theories focus on how issues or objects are reported in the media, emphasize the most remarkable aspects or descriptions of relevant objects, and underline the way of thinking (Weaver, 2007). McCombs even uses the term of second-level agenda-setting to include framing theory and regards framing as an extension of agenda setting as it is “is the selection of a restricted number of thematically related attributes for inclusion on the media agenda when a particular object is discussed” (1997, p37). McCombs and his followers propose that second-level agenda-setting and framing “share common concerns for attribute agendas (frames), the dynamics of the agenda-setting process (framing process), and agenda-setting influence (framing effects)” (Baran and Davis, 2009, p282), which means they can be used interchangeably. However, this extension view has not been accepted by many scholars. For example, Kim, Scheufele and Shanahan (2002) argue that the simple combination of framing and agenda-setting into a single theory confuses the distinctions between both concepts. Firstly, the theoretical premise of framing is attribution while agenda-setting focuses on salience (Scheufele, 2000).

“Agenda setting looks on story selection as a determinant of public perceptions of issue importance and, indirectly through priming, evaluations of political leaders. Framing focuses not on which topics or issues are selected for coverage by the news media, but instead on the particular ways those issues are presented” (Price and Tewksbury, 1997, p184).

Secondly, Schefuele (2000) argues that framing and agenda-setting affect people’s mind through distinct cognitive processes and result in different outcomes, because framing has a wider scope of cognitive processes including “causal interpretation”, “moral evaluation”, and “treatment recommendation” than does second-level agenda setting (Weaver, 2007). Thirdly, as to media effects, framing is concerned with applicability, whereas agenda-setting emphasizes repetition and accessibility (Ardèvol-Abreu, 2015). Applicability refers to the ability to produce interpretive schemas that can be used to many different messages (Weaver, 2007), while repetition and accessibility imply that the more frequently and prominently the news media report an issue, the more that the issue becomes accessible to the audience’s mind (Baran and Davis, 2009). In a nutshell, framing emphasizes the presentation of issues, whereas agenda-setting theory is concerned with salience of issues (De Vreese, 2005).

On the other hand, representation is defined as “using language to say something meaningful about, or to represent, the world meaningfully … Representation is an essential part of a process by which meaning is produced and exchanged between members of a culture” (Hall, 1997a, p15). Avraham and First (2010) argue that the representation approach and framing theory are complementary theories. For example, both theories stem from “the phenomenological approach”; both theories are affected by “social-political reality, as well as symbolic reality and the interaction between the two”; both theories are essential part of a process meaning which is “produced and exchanged between members of a culture in these two realities” (Avraham and First, 2010, p482). However, there are clear differences between them. The concept representation is mainly applied to cultural study, which links meaning and language to the culture and results in fictions and mistakes (Hall, 1997a; Derrida, 1973). Moreover, representation includes “the use of language, of signs and images which stand for or represent things” (Hall, 1997a, p15), while framing theory mainly focuses on the language usage. What’s more, the representation research is primarily concerned with minorities and marginal groups in news coverage, whereas the framing study centres on organizational levels (Avraham, 2003; Wolfsfeld, 1997).

According to the literature, because this study mainly explores and analyzes how a world-wide economic event was presented by two news outlets in different countries, framing theory is the most appropriate theoretical framework to the study among these three theories mentioned above.

2.2 Typology of new frames

News producers apply different frames in their news coverage, but “this abundance in choice in how to tell and construct stories can be captured in analyses as certain distinctive characteristics” (De Vreese, 2005, p54).

Because the frame can be used to present and comprehend news stories, media frame and individual frame can be categorized (Scheufele, 1999). As mentioned above, news frames are attributes of the news, built to identify the problem, causal reasoning, moral judgment and appropriate treatment (Entman, 1993), while the individual frames can be viewed as “mentally stored clusters of ideas that guide individuals’ processing of information” (Entman, 1993, p53). This means that “media frames are attributes of the news themselves, while individual frames are information and cognitive schemas” (Ardèvol-Abreu, 2015, p431). Since this study mainly focuses on frame variation of news articles rather than readers’ interpretations, this study belongs to the category of media frame.

With consideration of persuasive power, Chong and Druckman (2007) argue that frames could be divided into strong frames and weak frames. This means that frames are not all equal in strength, which can be influenced by their frequency, accessibility and relevance (Chong and Druckman, 2007). Frequency indicates the number of times and news-makers in which a frame is repeated; accessibility refers to the possibility that a previously stored thought in the memory is used to make an assessment; relevance emphasizes the importance of the issue: core or peripheral (Chong and Druckman, 2007). However, Chong and Druckman (2007) confess that it is difficult to measure the frame’s strength. Therefore, this typology could not be adopted in this study.

With the consideration of nature and contents of news frames, De Vreese (2005), based on previous research, proposes a more general typology: issue-specific frames and generic frames. The issue-specific frame can be employed to a specific event or topic, and the generic frame can be used to different topics without any thematic limitation, sometimes even exceeding different time and cultural contexts (De Vreese, 2005). As to generic frames, Neuman, Just and Crigler (1992) find that “human impact”, “powerlessness”, “economies”, “moral values” and “conflict” are the most common generic frames adopted both by the media and the public. Semetko and Valkenburg (2000) develop the research and identify 5 new generic frames including “conflict”, “human interest”, “attribution of responsibility”, “morality” and “economic consequences”. The conflict frame means confrontation or conflict between individuals, groups or relevant stakeholders; the human interest frame refers to a human face, an emotional angel or a personal story adopted to present an event or issue; the attribution of responsibility frame focuses on an issue or event in a way of attributing responsibility for cause or solution to individuals, groups or relevant stakeholders; the morality frame indicates the interpretation of an issue or event by use of religious doctrines or moral directions; the frame of economic consequences emphasizes the economic impact of an issue or event on individuals, groups or relevant stakeholders (Semetko and Valkenburg, 2000; De Vreese, 2005). On the other hand, although these generic frames are quite normally adopted in news coverage, it is acceptable that presentation of them may rely on the topic nature or the medium type and all of them do not need to be used in every single story (Neuman, Just and Crigler, 1992).

Generic frames not only improve the comparison results of different research in different places but also can be applied to cross-national comparative study on media frames (Ardèvol-Abreu, 2015; De Vreese, Peter and Semetko, 2001), so in this study generic frames are employed to analyze media coverage of the Summit. In addition, due to the special feature of the Summit, issue-specific frames are also employed in the study.

2.3 Potential factors influencing the frame-building of news coverage

According to Edelman (1993), the framing of an issue or event is an outcome of intentional consideration. In view of this, De Vreese (2015) argues that internal and external factors influence frame-building. Internal factors refer to editorial policies and news values and external factors mean elite influences and social movements. Along these lines, Scheufele (1999) proposes three potential factors to influence frame-building, including journalist-centred influences (e.g., ideology), the type or political orientation of the medium (e.g., organizational routine) and external sources of influence (e.g. elites and social norms and values).

2.2.1 Ideology

Ideology can be seen as a useful and necessary concept in media analysis (Devereux, 2007). The word ideology was coined in 1796 by a French writer Destutt de Tracy for the sake of his ideas of science (Rudé, 1980). Since then, many famous scholars such as Marx, Gramsci and Adorno have studied the concept successively. It is widely accepted that there is no single definition of ideology (Eagleton, 1996), but definitions given by Becker (1984) and Hall (1986) may serve the aim of this study. Becker argues that an ideology “is an integrated set of frames of reference through which each of us sees the world and to which all of us adjust our actions” (1984, p69). Further on, Hall (1986) regards ideology as “the mental frameworks-the languages, the concepts, categories, imagery of thought, and the systems of representation-which different classes and social groups deploy in order to make sense of , define, figure out, and render intelligible the way society works” (1986, p29). This means that both ideology and frames may offer certain people a framework to interpret issues, explain problems, identify causes and obtain remedies (Akhavan-Majid and Ramaprasad, 1998).

Based on the previous studies on ideology, Akhavan-Majid and Ramaprasad (1998) identify three types of ideology: dominant ideology, elite ideology and journalistic ideology. Among them, journalistic ideology plays an important role in the selection of news frames (Ardèvol-Abreu, 2015). On the hand, journalist ideology tends to strengthen elite ideology in the U.S. and advocate dominant ideology in China; on the other hand, elite ideology is normally consistent with dominant ideology (Akhavan-Majid and Ramaprasad, 1998). Therefore, it is necessary and feasible to identify the dominant ideology in China and the U.S. in this study.

According to Akhavan-Majid and Ramaprasad, dominant ideology means “views and ideas shared by the majority of people in a given society” (1998, p134). Communism has been adopted as the dominant ideology by the Chinese society with the governance of Communist Party of China (CPC) since 1949 when the People’s Republic of China (PRC) was establishment (Schurmann, 1968). However, with China’s reform and opening-up from the 1980s, China endured the weakness of communist ideology and the rising nationalism (Cao, 2005). Mitter (2000) argues that due to the slow economic growth and the breakdown of the Soviet Union in the 1990s, there is an ideological gap that emerged. Therefore, it is believed by many political scholars that the CPC actively resorts to a new ideology of nationalism as the dominant ideology in China (e.g., Frank, 1996; Friedman, 1997; Mitter, 2000). Friedman explicitly indicates that “an extraordinarily strong nationalism infuses elite political circles in Beijing at the end of the twentieth century” (Friedman, 1997, p5). In China, the term nationalism asks for the public to identify themselves with the Party, which means the CPC can be regarded as “the embodiment of the nation’s will” and discuss nationalism, socialism, and communism together (Zhao, 1998, p291). Especially, after coming to power, President Xi Jinping, based on nationalism, proposes the concept of “the China Dream”, a dream of “Wealth and power”, which has been interpreted as the rejuvenation of the Chinese nation (Veg, 2014). This concept is viewed as the “political manifesto and signature ideology of Xi’s administration” and the use of the word rejuvenation emphasizes that the Chinese believe they will come back to the previous greatness or past glory rather than rising from nothing (Wang, 2016).

The anti-communist ideology has been regarded as a “religious faith” by western elites (Herman 2000, p109). In the U.S., capitalism and anticommunism can be viewed as the dominant ideology (Akhavan-Majid and Ramaprasad, 1998). However, anticommunism “play out in different ways at different times, contingent upon specific time/place contexts, and is extremely broad” (Klaehn, 2009, 45). After 9/11 attacks, Mullen (2009) argues that free market ideology, antiterrorism and the ‘war on terror’ in current American society may make anti-communism redundant. This means that they have become “a strong co-replacement for anticommunism and the basis for the new world order of neoliberalism now in some disarray but without an ideological rival resting on any kind of power base” (Mullen, 2009, p15). However, it is noted that anti-communism is “not dead and is still used when needed” (ibid).

2.2.2 Media system

To some extent, the media system can reflect the major political divisions of society with distinctive political orientations (Morlino, Berg-Schlosser and Badie, 2017). According to Siebert, Peterson, and Schramm (1956), there are four press models including “authoritarian”, “libertarian”, “communist” and “social responsibility”. Hallin and Mancini (2004), by conducting a seminal study of comparing media systems of 18 Western democracies, further the theory of four models and propose three models of western media including “the Mediterranean or polarized pluralist model”, “the north/central European or Democratic corporatist model” and “the North Atlantic or liberal model”.

Due to different dominant ideologies, media systems of China and the U.S. are quite different. After the establishment of the PRC in 1949, the CPC referred to the “the Soviet communist media model” and required all media to follow “the line of the party leadership” and only to be used as propaganda (Yang, 2012; Sparks, 2000, p38; Shirk, 2011). The economic reforms from the 1970s have witnessed a huge change of the Chinese media function from propaganda tools to multifunctional entities (Shao, Lu and Hao, 2016). Firstly, the news publications emphasize information providing; secondly, they provide more and more entertainment programs; thirdly, media economic function has been widely accepted; finally, “watchdog journalism has taken root among media professionals” (Shao, Lu and Hao, 2016, p36). It is inevitable to result in “increasing concern to China’s news media as growth in number of available news outlets has made commercial survival imperative” (Luther and Zhou, 2005, p858). Winfield and Peng claim that “with a convolution of the Party line and the bottom line, a Chinese media system is moving from totalitarianism to market authoritarianism” (2005, p255). This means that on the one side the Chinese media have to fulfill their obligation of mouthpiece of the Party, on the other side, they have to make a profit to support themselves (Shao, Lu and Hao, 2016). Therefore, violence, pornography or even incorrect ideologies may be found in some Chinese media, which increases difficulties in government’s control over the media (Ma, 2000; Shao, Lu and Hao, 2016).

However, in order to maintain the Party’s goals of stability and economic development, media are still deeply influenced by the CPC through ownership, sponsorship and censorship (Liebman, 2005; Luo, 2015). Especially Party’s newspapers such as People’s Daily and the China Daily “are closely watched” (Ma, 2000, p24). Although pursuing profits from advertisements, they try their best to avoid political mistakes, “sticking to the party’s propaganda principles on major issues and even catering to the needs of party leaders” (Shao, Lu and Hao, 2016, p38). In February 2016, President Xi in a symposium emphasized that all news media run by the Party “strictly follow the Party’s leadership and focus on ‘positive reporting’” and “must work to speak for the Party’s will and its propositions and protect the Party’s authority and unity” (Xinhua, 2016a).

The U.S. media belongs to the Anglo-American model, a specific example of a liberal system with a high degree of autonomy (Hallin and Mancini, 2004; Curran, 2012). They are regarded as “independent both of the state and social ‘subsystems’ comprised by political, economic or solitary groups” (Curran, 2012, p129-p130). Moreover, it is true that the First Amendment to the U.S. Constitution offers strong legal guarantee of media freedom (Freedom House, 2015). Against this backdrop, the U.S. journalists play “the role of exposing, condemning, or excluding from the public agenda those who violate or challenge the political consensus” (Hallin, 1989, p117). Therefore, the primary role of the media in liberal countries could be regarded as the watchdog, which surpasses all other functions of the media in the degree of importance (Curran, 2012). However, due to the development of the media since the early 21^st century, the watchdog role of the media has been questioned. Media system in the U.S. is “given over largely to entertainment” and “even many so-called news media allocated only a small part of their content to public affairs” (Curran, 2012, p219).

Although there is no distinct governmental department which issues directives and censor rules to media in the U.S. that is the purest case of the liberal model, “the role of the state cannot be ignored” (Chen, 2009; Hallin and Mancini, 2004, p228). On the one hand, the state is normally regarded as the regulator, funder and owner of the media; on the other hand, it can exert impact on the process of news coverage (Hallin and Mancini, 2004). It is clear that news-making is organized through information and interpretation offered by government officials (Hall et al. 1978). What’s more, the concept of national security interests has resulted in a substantial cultural premise that “journalists and government officials both in some sense represent a common public interest, and to the institutionalized relations of trust and mutual dependence that have developed between them” (Hallin and Mancini, 2004, p234).

2.2.3 News sources

As mentioned above, the external factors of frame-building include “political actors, authorities, interest groups, and other elites” (Scheufele, 1999, p115). This is similar to Entman’s (1993) argument that political elites can exert influence on the framing of events. De Vreese and Lecheler argue that elite influence “becomes apparent when journalists use parts of political speeches or ‘soundbites’ to illustrate an issue” (2012, p294).

In Liebler and Bendix’s study (1996), they propose that frames of television news have been influenced by different news sources. Cozma’s study (2006) shows that readers view issues with varied sources as more convincing and interesting. By the content analysis of the relationship between framing and sourcing during the golden age of foreign correspondence in the U.S., Cozma (2015) also finds that specific types of officials relied on by the news make positive contributions to the favourable mood to the U.S. engagement in the World War II. Scheufele also finds that journalists are most likely to adapt elite framing on “relatively new issues (i.e., issues for which no frames have yet to be established)” (1999, p116). What’s more, Abrajano and Singh (2009) through content analysis prove that different news sources may change audience attitudes towards the same event. In other words, news sources frame the news in a way that is consistent with their preferred framing (Hallahan, 1999).

It is noteworthy that official sources are always the dominant sources (Blumler and Gurevitch, 1981), which can exert great influence on the frame-building process. Official sources mean “those speaking for individuals or organizations” (Powers and Fico, 1994, p90), which can be classified as external factors of political elites. Based on previous studies (e.g., Pan, 2003; Sun, 2008), the news sources cited in China-related news coverage could be mainly categorized into six types: “the Chinese government (central or local)”; “foreign government or international authoritative institutions; journalists”; “Chinese citizens; foreign citizens”, “Chinese media agencies”; and “foreign media agencies”. Therefore, this study mainly focuses on the following 7 official sources: Chinese government officials, President Xi Jinping, U.S. government officials, President Barack Obama, other foreign government officials, international organizations officials, experts and business leaders.

2.2.4 Stereotyping

As a part of human society, stereotyping and stereotypes have been regarded as “a means of preserving those values, ideals, norms, and human behavioral and other identities that we see as unique to us and superior to others” (McFarlane, 2014, p142). This means that social norms and values have close relations with stereotyping that has been one of most prevalently cited concepts in framing studies (Stangor, 2015; Yang, 2015). Therefore, it is feasible for this study to analyze the frame-building process under the influence of stereotyping and stereotypes. Entman (1993) suggests that stereotyped images can be regarded as one of important factors to exam and identify news frames. This is similar to Edelman’s (1993) argument that besides ideology, prejudice, which is closely associated with the concept of stereotype, can exert influence on the choice of frames. In other words, stereotypes “provide a fast and easy way for mass communicators to shortcut the nuances in a story. We can rely on the stereotype to create an image or expectation about a certain group” (Neher and Sandin, 2015, p224).

The concept stereotype dates back to 1922 when Lippmann argued that stereotypes were selective, self-fulfilling and ethnocentric which constituted “a very partial and inadequate way of representing the world” (1922, p72). This means that people communicate with each other based on images existing in their minds rather than engagement and observations (Lippmann, 1922). Oakes, Haslam and Turner, argue define stereotyping as “a process of ascribing characteristics to people on the basis of their group membership” (1994, p1); while a stereotype refers to “a collection of association that links a target group to a set of descriptive characteristics” (Gaertner and Dovidio, 1986, p81). As Hall further argues,

“Stereotyping…is part of the maintenance of social and symbolic order. It sets up a symbolic frontier between the ‘normal’ and the ‘deviant’, the ‘normal’ and the ‘pathological’, the ‘acceptable’ and the ‘unacceptable’, what belongs and what does not or is ‘Other’, between ‘insiders and ‘outsiders’, us and Them” (1997b, p258).

Most stereotypes are prone to transmit negative feelings (McLeod, 2008). However, people can “make use of stereotypical ideas and images – both positive and negative– about their own group(s) as well as others” (Devlin, 2006, p12). Positive stereotypes “serve to justify existing intergroup inequality” and has been defined as “subjectively favourable beliefs about members of social groups that directly or indirectly connote or confer domain-specific advantage, favourability, or superiority based on category membership” (Czopp et al., 2015, p451), which not only is a way to build valued and unique identities, but also mitigates negative impressions of a certain group (Czopp et al., 2015). When people use positive stereotypes to describe themselves and reject negative stereotypes, self-stereotyping occurs, which can enhance the sense of connection between members of one’s own group and establish the sense of group cohesion and solidarity (Biernat, Vescio and Green, 1996; Huntsinger and Sinclair, 2010). Therefore, in order to explore the frame-building process, how the China Daily and the New York Times used stereotyping and stereotypes needs to be explored and analyzed in the study.

In sum, this study intends to use the frame defined by Entman (1993) to address how such factors as ideology, media system, news sources and stereotyping exert impact on the frame-building process of both news outlets.

CHAPTER 3: METHODOLOGY

This chapter outlines the methodology adopted in this study. The following subsections introduce the reason why the research method is chosen, how to identify news frames, how to make sampling and coding respectively.

3.1 Research method

Generally speaking, a methodology of the combination of quantitative content analysis and qualitative framing analysis is adopted in this study. In media and communication research, the quantitative and qualitative approaches are mostly adopted to analyze media text (Wimmer and Dominick, 2010). Quantitative research method refers to “numbers, magnitude, and measurement” (Berger, 2000, p13), while qualitative method refers to “the text’s properties, degree of excellence and distinguishing characteristics” (Berger, 2000, p13). In other words, quantitative research focuses on counting and measuring of things which produces reliable results and qualitative research emphasizes the meanings, characteristics, metaphors and descriptions of things which understands their connotations (Macnamara, 2005; Monfared and Derakhshan, 2015). However, it is accepted that both of them have in irresolvable weaknesses. Quantitative researchers are criticized for their narrowness due to their research only focusing what they study and neglecting other things and qualitative researchers are accused of their subjective interpretation which may exceed the meaning of things (Berger, 2000). Therefore, “a combination of the two seems to be the ideal approach” (Macnamara, 2005, p5). This means that researchers “collect or analyze not only numerical data, which is customary for quantitative research, but also narrative data, which is the norm for qualitative research in order to address the research question(s) defined for a particular research study” (Williams, 2007, p70).

Content analysis has been regarded as a quantitative method in nature (Franzosi, 2007) and “the fastest-growing technique over the past 20 years or so” (Neuendorf, 2002, p1). The concept dates back to 1927 when Lasswell (1927) applied it to the study of propaganda and classified it as a quantitative method. Since then, content analysis has gradually become a main research methodology in mass communication studies and social science (Prasad, 2008). Berelson defines it as a “research technique for the objective, systematic and quantitative description of the manifest content of communication” (1952, p18). It could be used to not only analyze all forms of communication including newspaper coverage, but also to exam a wide scope of data over a specific period to uncover media texts and their possible meaning (Zito, 1975; Macnamara, 2005). What’s more, content analysis is “most efficient when explicit hypotheses or research questions are posed” (Riffe, Lacy and Fico, 2014, p44). All of these indicate that this method seems to be suitable for this study of media coverage on the Summit. As one of the most well-known researchers of content analysis, Neuendorf (2002) emphasizes that the analysis needs to be done through scientific methods such as a priori design, reliability, validity and hypothesis testing. This means that there are several necessary steps for this study to follow. After research questions are proposed, the method needs to be conducted through sampling and unitizing, coding the data, testing the reliability and then interpreting the data to answer research question (Wimmer and Dominick, 1997).

However, content analysis could not “capture the context within which a media text becomes meaningful” (Van den Bulck, 2002, p84) and “is and should be enriched by the theoretical framework offered by other more qualitative approaches” (Hansen et al., 1998, p91). Wimmer and Dominick (2010) argue that as a kind of qualitative analysis, frame analysis could compare media content to the true world. This means that this qualitative method could be valuable to understand how people observe their communication content and present what is absent in the news coverage (Wood, 2004; Holstein, 2002). According to Connolly-Ahern and Broadway, qualitative framing analysis “involves repeated and extensive engagement with a text and looks holistically at the material to identify frames” (2008, p369). Therefore, it is appropriate to complement content analysis with framing analysis so as to better answer research questions.

3.2 Frame identification

It is widely accepted that there is no unified method to identify news frames in news coverage, but communication researchers usually adopt deductive and inductive methods (De Vreese, 2005). The deductive method mainly deals with predefined frames and test whether these frames exist in relevant news articles (Ardèvol-Abreu, 2015; De Vreese, 2005). Due to the replicability of the method, it is suitable to apply this method to comparative analyses (Ardèvol-Abreu, 2015). On the other hand, the inductive method requires an open approach applied to examine relevant news coverage in order to identify frames inside (Ardèvol-Abreu, 2015; De Vreese, 2005). This means that the deductive method is mainly used to identify generic frames and the inductive approach focuses on detecting issue-specific frames. It is undeniable that both of methods may have deficiencies. The first method is thought to solely depending on predefined frames and ignoring other frames due to the loss of relevant information; while the second one has been accused of relying on small samples and lack of objectivity and reliability (Ardèvol-Abreu, 2015; Hertog and Mcleod, 2001). Therefore, in order to make the framing analysis objective and comprehensive, both the deductive method and inductive method are used in this study.

As mentioned in Chapter 2, generic frames including “conflict”, “human interest”, “attribution of responsibility”, “morality” and “economic consequences” identified by Semetko and Valkenburg (2000) through the deductive method could be used in the analysis. As to issue-specific frames, they need to be abstract enough so that they could be applied in other situations (Ardèvol-Abreu, 2015). Van Gorp (2007) introduces how to indentify frames through inductive method, within which framing devices such as terms, metaphors, examples, descriptions, arguments, images and arguments could result in news frames. This study adopts Van Gorp’s inductive approach to identify issue-specific frames of the Summit. In addition, because this study deals with comparative analysis of news articles, photos, photos captions and logos published with the news coverage together are excluded here.

3.3 Sampling

As mentioned above, a sampling process is needed firstly so as to conduct the following analysis. In the process, sampling source, sampling period, sampling units and selected samples are included.

3.3.1 Sampling source

The aim of this study is to comparatively analyze the China Daily and the New York Times’ news coverage regarding the Summit. The main reason why these two newspapers are chosen is that both of them are well-known as representative and influential newspapers in their countries (Yang, 2003; Luther and Zhou, 2005).

The China Daily is not only the first and only national English newspaper in China, but also “one of the most-frequently quoted Chinese media around the world” (Tong, 2006; China Daily, nd). It was founded in 1981 to conform to China’s reform and opening up policy and serve high-end readers with a circulation of 900,000 copies (Chang, 1989; China Daily, nd). As one of Party’s newspapers mentioned above, the China Daily operates “under the same rules as the local language press and allows outsiders a glimpse of the country and its perspective on the world” (Stevenson, 1994, p156). This means it could offer a reliable and representative way to observe the overall reporting approach taken by most Chinese media toward one news event (Akhavan-Majid and Ramaprasad, 1998). What’s more, due to its editorial policy of meeting the needs of reading habits of foreigners, the China Daily could exhibit the process of news framing similar to the U.S. newspapers (Chang, 1989; Luther and Zhou, 2005).

The New York Times, founded in 1851, servers readers of policy makers, academic researchers, foreign affairs analysts with a daily and Sunday circulation of about 571,500 and 1.1 million copies respectively (Yang and Liu, 2012; New York Times, 2016). It is normally perceived as the most respected news medium in the U.S. and possesses huge agenda-setting power (Dearing and Rogers, 1996). Gitlin (1980) argues that it could be regarded as the newspaper of recording international news and influencing the content of other mass media globally. This means that when an event is newsworthy, other news agencies could take a cue from the New York Times (Dearing and Rogers 1996). Moreover, it serves as an elite capitalist newspaper (Bagdikian, 2000). In other words, it works with other institutional networks, to communicate capitalist ideology and other useful information so as to keep the connection among political business elites (Sweezy, 1962).

3.3.2 Sampling period

The sampling period could be decided in line with the research goal (Wimmer and Dominick, 2010). This study aims to compare news frames and dominant frames in both newspapers, so it could be comparable that there need enough relevant news articles in the same period. Originally, the start point of the period was set on March 8 2016 when Foreign Minister Wang Yi replied the question of the Summit (Ministry of Foreign Affairs of China, 2016), while, because two weeks are needed to reflect the changes in news coverage (Riffe, Aust and Lacy, 1993), the end point was on October 3 2016, one month after the end of the Summit, which meant that another two weeks were extended to comprehensively analyze the news coverage. However, the initial search showed that articles in the China Daily could be found from the start point to the end point, whereas there were no relevant articles in the New York Times in March and October. The first article directly associated with the Summit in the New York Times was published on August 30, 2016 and the last one was on September 23. Therefore, based on the initial search, the sampling period was shortened from August 29, 2016 (a week before the Summit) to September 26, 2016 (three weeks after the Summit), which meant a study with four consecutive weeks. This period not only included the Summit time (September 4 and 5), but also expanded to observe the following effects of the Summit to make the analysis more comprehensive and valid.

3.3.3 Sampling units

According to Wimmer and Dominick (2010) a word, a sentence, a paragraph or an entire news article could be used as the unit of media content analysis. However, a word or a sentence may not be a proper unit for framing analysis because all generic frames could not exist in a word or sentence (Dai, 2013). On the other hand, the unit in framing analysis is often “individual news articles which appeared in the selected newspaper during the selected study period” (Linström and Marais, 2012, p29). Therefore, in order to better find out generic frames and issue-specific frames and then identify dominant frames in the comparison, the entire news articles were chosen as the units of analysis. The news articles in the study include news, editorials, commentaries and opinion and other articles such as letters to editor and visual pictures were excluded.

3.3.4 Samples selected

After sampling period and unit were decided, the relevant news articles from both newspapers were collected. Firstly, Factiva Dow Jones database was applied to retrieve news articles from the China Daily and the New York Times in the designated period mentioned above, based on the key words “G20 Summit”, “Group of 20 economic summit meeting”, “G20 Huangzhou” and “G20 meeting”, which were not mutually exclusive and used in the full text. Because the database may “often provide a narrow sample of media content” (Macnamara, 2005, p14), a similar procedure was adopted by searching websites of these newspapers respectively. This crosscheck method might ensure the accuracy and reliability of the selected samples. After the initial searching result, 197 articles from the China Daily and 43 articles from the New York Times were retrieved and all of them were read. It is found that some articles were not directly associated with the Summit in both newspapers^[1] and some articles were not written by journalists of both newspapers. So, those news articles which were relevant to the Summit, written by journalists of these two newspapers were chosen first. After the preliminary selection, it is found that the words of all sample articles of the New York Times were over 500. In order to better conduct comparative analysis, sample articles of the China Daily with less than 500 words were given up. In addition, due to limited time and resource, articles that were only published in the websites, social media and other editions of these two newspapers were also excluded. Finally, 49 news articles from the China Daily and 20 news articles from the New York Times were retrieved.

3.4 Coding

According to Prasad (2008), after research questions are formulated and samples are selected, the next step is coding. Generally speaking, there are two coding methods in quantitative content analysis: human coding and computer coding (Macnamara, 2005). “The notion of the completely ‘automatic’ content analysis via computer is a chimera… The human contribution to content analysis is still paramount” (Neuendorf, 2002, p40), so human coding was adopted in this study.

In human content analysis coding, it is a protocol to set up a coding form and a codebook (Neuendorf, 2002). The coding form of this study (Appendix 1) was formed in light of research questions. This means that some general analytical categories such as case number, medium source (newspaper’s name), article title, date, location and length in most content analysis needed to be included (Altheide and Schneider, 2012), while some specific categories such as news frames, dominant frames, news sources, overall attitudes towards the Summit and the host were added into the form, in order to better examine the possible factors influencing the frame-building process. Based on the coding form, a code book (Appendix 2) for the analysis was drafted. According to Neuendorf, it is more common that the coding form seems to be like “an empty shell, merely a convenient repository for numeric information” and the codebook has more details (2002, p132).

As to frames, 5 generic frames including “conflict”, “human interest”, “attribution of responsibility”, “morality” and “economic consequences” mentioned above were added in the coding form first. To locate and count these 5 frames, three or four questions were designed in the codebook for each frame. Each question needs to give a “yes” or “no” answer to confine subjectivity and the identification of a frame could be decided by at least two affirmative answers. In this study, news frames were not mutually exclusive in news articles but there was only one dominant frame in each article, in order to help examine frame differences and influencing factors. According to Linström and Marais, the dominant frame means “the main theme of the news article” (2012, p30).

Piloting coding is “a crucial step before launching the full-scale content analysis” (Prasad, 2008, p14), which means that test coding of a small sample could help revise and perfect coding form and code book (Neuendorf, 2002). There are 10 randomly selected articles (5 articles from each newspaper) examined in the pilot study. The test result revealed that most news frames and dominant frames detected in the selected articles could fall into the 5 generic frames, except for two articles. One article from the China Daily mainly presented China’s special contribution to world economic recovery, which was quite different from those 5 generic frames. Based on inductive method proposed by Van Gorp (2007) mentioned above, an issue-specific frame called “China’s global economic governance” was identified and added into the coding form. On the other hand, one article from the New York Times focused on regional conflicts in Syria, another issue-specific frame “regional security” was also added. Three questions were proposed for these two frames and two “yes” answers could identify the frame. In addition, the Belt and Road initiative and the incident of Barack Obama’s staircase episode were repeatedly mentioned by the China Daily and the New York Times respectively in the test, so the frequencies of these two as new categories were also added into the coding form.

After the pilot test, the coding form and codebook were revised and applied to all selected samples. The coding was conducted by two coders, both of whom were Chinese postgraduate students studying international relations in the UK. The reasons why these two coders were selected were that on the one hand, they were the only persons who could be available to do the coding and others could not be available due to limited time; on the other hand, the selected coders were proficient in English and had profound understanding of China and the U.S. society due to their Chinese identification and MA research topics related to the U.S., which could make them better understand not only English news articles from both newspapers, but also latent connotations behind articles. In order to maintain the objectivity and validity of the coding, careful training of coders was needed before formal coding, which could make the result more reliable (Prasad, 2008). Therefore, the coding rule and coding form were fully explained to two coders respectively and then both of them were trained to follow the categories of the codebook to code 5 news articles from each newspaper independently without any discussion and collaboration in order to compensate for the possible bias. Based on the coder training and familiarization of the coding procedure, two coders coded all selected news articles.

Lacy and Riffe (1993) argue that content analysis reliability is a minimum requirement to assess the validity of the study (cited in Riffe, Lacy and Fico, 2014, p60). When the analysis is conducted by human coders, the reliability means inter coder reliability (Neuendorf, 2002). Therefore, it is necessary to check the inter coder reliability in this study. According to the formula[2] of proposed by Holsti (1969), which is “the recommended method” (Neuendorf, 2002, p149), the inter coder reliability of the coding was 95%, which indicated that the level of reliability was acceptable (Macnamara, 2005).

[1] For example, a news article retrieved from the China Daily named Setting ambitious long-term goals mainly mentioned Qingdao would hold Civil Society 20 China 2016 Conference before the G20 Summit in Hangzhou, thus it is discarded. The similar case was found in the New York Times article entitled Chinese company’s payment to lawmaker sets off furor in Australia mainly discussed an Australian civil affair with a statement from Barnaby Joyce, Deputy Prime Minister, who attended the Summit.

[2] The Holsti’s formula is PA₀=2A/N_A+N_B where PA₀ represents “proportion agreement”; A is the number of agreements on categories between two coders; N_A and N_B are the number of categories coded by both coders respectively. The result ranges from 0% (no agreement) to 100% (perfect agreement).