Media Political Leanings: Polarised Depictions of Hong Kong Migration in Taiwan

Abstract

This chapter examines the influences of the Taiwanese media’s political leaning on their coverage of Hong Kong migration by analysing news articles from the five major pro-Blue/Green printed media outlets between 2014 and 2022. Using structural topic modelling, this chapter demonstrates a politicalised trend relating to Hong Kong migration, involving two themes: migration safety and Political Security. Even though the migration safety theme has gained much more attention than the Political Security theme since the 2019 Protests in Hong Kong, the polarisation of the depiction of Hong Kong migration becomes particularly evident when issues related to the People's Republic of China (PRC) are factored in through topic comparison. In other words, compared to the pro-Blue media, which is inclined to portray the Hong Kong migration from a human rights perspective, the pro-Green media, which favours a nativist viewpoint, offer a more security-based narrative,  such as cultural Sinicisation and the possibility of Chinese spies entering through Hong Kong migration. The results of sentiment analysis further highlight both media camps’ differences by detecting a positive or negative tendency.

Introduction

Immigrant and political asylum issues in Taiwan have become more prominent since the protests in Hong Kong as growing numbers of Hong Kong’s residents seek life opportunities or protection in Taiwan. According to the data from Taiwan National Immigration Agency, the number of Hong Kong people in residency and settlement with citizenship in Taiwan increased from 8203 in 2014 to 10,240 in 2022 (National Immigration Agency, n.d.). There were a total of 76,245 successful cases through various investment, dependent, and asylum channels.

The growing trend of Hong Kong migration has sparked contentious debates within Taiwanese society because of the China factor (the PRC recovered Hong Kong in 1997 and Taiwan is the next in line for Chinese unification). The two major political camps in Taiwan exhibit contrasting viewpoints, despite both initially claiming to support Hong Kong residents based on their humanitarian stances. The Green Camp is inclined towards Taiwan’s independence, Taiwan’s nativist values, and resisting the influence of the PRC. The leading party in this camp is the Democratic Progressive Party (DPP, the ruling party). The Blue Camp is inclined towards unification with the mainland China and making mitigating gestures to the PRC. The leading party in the Blue Camp is the Kuomintang (KMT). Due to the current DPP government’s view towards China, the Hong Kong immigration issue has become intertwined with Taiwan’s national security. Thus, Article 12 (for residency applicants) and Article 18 (for political asylum seekers) of Regulations Governing the Relations with Hong Kong and Macao (hereafter, the Hong Kong & Macao Regulations) are stringent and are implemented with caution. This circumstance is further reflected in a petition case to Control Yuan (a constitutionally mandated authority with a supervisory and auditory role) over Hong Kong residency in Taiwan.

As part of this case, the authorities, including the Immigration Agency, were criticised not only for the opaque, time-consuming, case-by-case review process but also for frequently rejecting applications for reasons of national security (by Article 22 (1) of Hong Kong and Macao Residents Entering Taiwan and Residence and Settlement Permit Measures). In addition, the petition blamed the government for drastically and unexpectedly raising the threshold of the applications for residency and settlement with citizenship. This places petitioners in a dilemma: whether to return to Hong Kong or wait without a definite date for their case to be settled in Taiwan. The Control Yuan’s press release about this petition case stated that the matter of petitioners involves the protection of human rights because it stems from the 2019 Anti-Extradition Law Amendment Bill Movement (the 2019 Protests) (Chi et al., 2022). Arguably, this press release shows the tension between human rights when they are related to migration (the human rights of Hong Kong immigrants seeking life prospects and protection in Taiwan), and national security (such as alleged Chinese spies entering through immigration), in Taiwan. What, however, exactly are the specific security interpretations of Hong Kong immigrants in both the Blue and Green Camps?

Media outlets frequently play an active role in framing migration coverage, using a particular political perspective (Caviedes, 2017). This coverage includes editorials, letters, and columns, which look at the security and human rights aspects of migration from different points of view. Some studies imply that all media outlets are complicit in the securitisation of migration (referring to a process of threat identification that is declaratory in nature, in which referent objects became securitised, which in turn legitimises urgent measures to tackle that perceived threat), and that the media’s political leanings, such as those of conservative newspapers, tend towards greater securitisation (Benson, 2013).

This is the case in Taiwan because the political environment of Taiwan’s media is mainly made up of two political camps, the “Green Camp” and the “Blue Camp” (Sullivan et al., 2018). Most studies on Taiwanese media representation, with regard to migration, focus on migrants’ identities (Chung, 2000), media framing (Liu, 2021), and discrimination (Chu, 2007), while factors relating to the media’s partisan leanings on migration are collectively an issue which has not received as much attention.

To fill this research gap, this chapter argues that Taiwanese media’s partisan leaning towards the Blue Camp or the Green Camp plays a main role in shaping the portrayal of the migration issue. To address the questions, this chapter examines the perspectives of the Green Camp and the Blue Camp on the subject of Hong Kong migration in Taiwan by analysing news articles from the major pro-Blue/Green printed media. The rest of the chapter is structured as follows. The second section presents the context of Taiwanese media’s partisan political leanings, which encourage media politicisation on the subject of Hong Kong migration. The third section presents the data and methods. The fourth section illuminates the results of the structural topic modelling (STM) (Roberts et al., 2016), including STM-identified topics, the trends, the proportions, conceptual categorisations, and the factorial effects on topic proportion. In the fifth section, the results of the sentiment analysis are highlighted. Concluding remarks and implications for Hong Kong migration are provided in the concluding section.

Partisan Leaning in Taiwanese Media

Since political liberalisation in the 1980s lifted the ban on opening media such as newspapers in Taiwan, independent newspapers have been allowed to open, shaping two major partisan media camps (Chen, 1998). One typical example of independent newspapers is the Liberty Times. The Liberty Times was founded in 1987. Its political stance is anti-KMT and anti-authoritarian rule, it emphasises Taiwan’s nativist value and affirms Taiwan’s independence. These are all compatible with the DPP’s political position on Taiwan. For example, the Liberty Times arranges the China news in the international section, as opposed to the pro-Blue newspapers, who put such news in the national section. In contrast, the pro-Blue newspapers, the United Daily News and the China Times were both founded in 1950. The launch year implies that both newspapers met the KMT’s requirement, and, in actual fact, for example, the founder of the United Daily News was a mainland Chinese man and also a core member of the KMT Central Committee (Guo, 2012, p. 4). Therefore, the pro-Blue newspapers generally empathise with a revival of Chinese culture and reunification with mainland China. As a result of Guo’s study (2012), it was found that the news narrative in Taiwan exhibited ideological bias. In this regard, it can be said that party-leanings have always fuelled Taiwan’s media polarisation, particularly when China-related issues arise.

One typical case of this polarisation is embodied in the issue of migration from mainland China. With the increase in transnational marriage (National Security Council, 2006), the issue of marriage migrants became a talking point in Taiwan. The media outlets in the pro-Blue Camp media tended to view mainland spouses through a lens of human rights, such as the right to residency in Taiwan. In contrast, the pro-Green Camp media tended to report on the spouses negatively, for example, suggesting that they were United Front conspirators (referring to those who assist the CCP in politically infiltrating the target countries by establishing friendly networks of groups and key individuals, Chao, 2004). For instance, one editorial in the Liberty Times (2008) stated that the increase in the number of mainland China spouses was equivalent to increasing the risk of the PRC’s United Front in Taiwan. The United News Daily (Lin, 2007) criticised the then-DDP government’s strict quota of mainland spouses, which drew criticism from human rights groups.

As mass media can shape the linkage between immigration and security, a comparison of the two main media camps’ depictions of Hong Kong migration can unpack the security concerns surrounding Hong Kong migration, as detailed in the next section.

Data and Methods

The primary textual data in this chapter was derived from news articles published in the five major Chinese newspapers in Taiwan (Renli Aikman Company, 2022): the Liberty Times (LT, Ziyou shibao); the United Daily News (UDN, Lianhe bao); the China Times (CT, Zhongguo shibao); the Economic Daily News (EDN, Jingji ri bao); and the Commercial Times (CTEE, Gongshang shibao). This study chose these five newspapers because they are issued throughout Taiwan and have a clear political leaning towards the Blue and Green Camps. As such, they are able to situate the national political debate on Hong Kong migration. Data collection was accomplished using the digital news media database of the Parliamentary Library, Legislative Yuan (https://nplnews.ly.gov.tw/). The search was conducted using the terms “Hong Kong” (Xianggang), “migrants”, (yimin) and “Taiwan” from 1 January 2014 to 5 December 2022. This period covered the political turbulence from the 2014 Hong Kong Occupy Central Movement (OCM), and included the 2019 Protests, and the implementation of Hong Kong National Security Law (NSL) in 2020. There were 259 pieces of news articles left after duplicates were removed, composing the corpus used in this study. Figure 7.1 shows the yearly number of news stories after the OCM, which peaked following the 2019 Protests. The news articles reached 31.6% (82 pieces) in 2020. Figure 7.3 illustrates the distribution of the news articles of our dataset. The LT represents 45% (117), the UDN 21% (55), the CT 14% (37), the EDN 11% (29), and the CTEE 8% (21). The LT represents the Green Camp with 45% (117 pieces), while the other four newspapers represent the Blue Camp with 55% (142 pieces), according to the classification of Sullivan et al. (2018, p. 114).

Fig. 7.1

Trend of news articles in our dataset

Data analyses were performed in STM. STM, which assumes that every document (including every news article) is a mixture of topics, is a word-clustering technique to identify latent topics. The benefit of STM is that it is able to extract a discursive structure comprised of exclusive and semantically coherent clusters of topics from unstructured textual data from within a period of investigation. For the objectives of this study, the term “topics” refers to particular discursive frames (Van Atteveldt et al., 2014). The topics include information on the topic’s proportion (which is the summed prevalence of each topic selected in this study), a list of FREX terms (referring to terms or phrases in a topic that appeared frequently and with exclusivity relative to other terms in other topics), and the top-ranking news of each identified topic. Moreover, STM can include metadata, document-level variables such as the date of publication and author (one variable in this study is pro-Blue/Green) for topic analysis (Grimmer et al., 2022, p. 153). With these variables for analysis, STM can present an identifiable trend, an examination of event effects on a topic’s patterns, and can form the basis of a comparison of pro-Blue and pro-Green media content. All these advantages are not only helpful for finding evidence of the selected media’s political leanings on the Hong Kong migration issue, but also present the general trends relating to this issue.

This study settled on a 12-topic model after processing the aforementioned corpus in STM. This study further selected and labelled the seven identified topics related to our research tasks for interpretation by carefully reading the top ten news articles about each topic (a total of 70 articles; for details of the topic model selection, see Appendix A, and see Grimmer & Stewart, 2013). The last step is to engage in topic interpretation, as elaborated in the following section.

The Results of STM

Table 7.1 shows the seven topic names used in this analysis and their proportions. By thoroughly examining the FREX terms and the top news stories, this study categorised the selected topics into two primary categories: “Migration Safety” and “Political Security”. Three topics (38.47%) fell in the category of Migration Safety, made of Political Migration (Topic 2, hereafter, T represents Topic), Insincere Assistance (T9), Conditions Imposed on Migration (T1). The three topics in this category relate to human security, with protection of life opportunities (Persaud, 2022, p. 141) being a central requirement of Hong Kong migrants. The topic with the largest proportion in the category of Migration Safety is Political Migration (T2), devoted to describing the lives of Hong Kong migrants in Taiwan and their reasons for migration. The four topics in the Political Security category (26.35%) include Empty Promise of OCTS (T4), Soft Penetration (T3), Blur Jurisdiction (T8), and Population Replacement (T12). Here this study adopts a wide definition of Political Security that refers to threats to both political stability and collective identity (Buzan et al., 1998, p. 141).

Table 7.1 Categories and topics

The Empty Promise of OCTS topic (OCTS means One Country, Two Systems, which was a unification model posed by the PRC for Taiwan that was used in Hong Kong first, and in which also promised that Hong Kongers can enjoy a high degree of autonomy that will remain unchanged for 50 years). It refers to the Hong Kong immigration issue and signifies the demise of OCTS in reality, implying that Taiwan has definitely rejected this political arrangement. The Blur Jurisdiction topic represented the conflict between Taiwan and Hong Kong over the question of jurisdiction or juridical sovereignty related to a Taiwan murder case that involved Hong Kongers. Soft Penetration and Population Replacement concerned the notion that the abrupt increase in Hong Kong immigrants posed a threat to Taiwan’s political stability and social resource allocation. The topic with the largest proportion is Empty Promise of OCTS, which presents the political root of Hong Kong migration in Taiwan: Hong Kong's freedom is dramatically dwindling.

Figure 7.2 illustrates the thematic trend in Taiwanese mainstream media coverage that follows the two categories, Migrant Safety (the solid line) and Political Security (the dashed line), through the investigation period. It can be observed that the Political Security trendline decreased from a high point in 2014 (7.29%), shortly after the 2014 OCM, to a low point in 2015 (7.29%). Similarly, the Migrant Safety line went downward from the year of OCM (18.22%) to its lowest point in 2016 (12.80%). This implies that the media outlets’ handling of the Hong Kong immigration issue became less intense after the OCM. Since then, however, the Political Security and the Migrant Safety trendlines rebounded, with fluctuations, respectively. The upward trends were maintained in both trendlines until the 2019 Protests. Since then, the Political Security trendline has slumped, but the migrant security trendline has continued to boom to its peak (43.84%) in 2020, the year of the NSL legislation, and the gap between the two lines has been increasing. It shows that the NSL was probably the main driver of the re-politicisation of the issue of Hong Kong migrants.

Fig. 7.2

Changing thematic proportion trends

Factors Affecting Topic Proportions

This section identifies the differences in news portrayals with a baseline of factors, including the implementation of NSL and Blue/Green as covariates, in order to observe any effects on the topic patterns. The NSL factor was established on the date the NSL went into implementation, 1 June 2020. This is when the trendline for migrant security attained its peak in 2020 (see Fig. 7.2). Blue/Green party affiliation was also considered in order to investigate the media’s inclinations between Migrant Safety and Political Security.

The Effect of NSL on Topical Proportion

The result of the NSL effects on the topical patterns is shown in Fig. 7.3. The left plot and the right plot of the figure represent before and after the NSL’s execution, respectively. This figure has two highlights. One is Conditions Imposed on Migration (T1), which ranked third on the left plot, but climbed to the top in the right plot after the NSL’s implementation. Another, Soft Penetration (T12), ranked in fourth place in the right plot, a change from last place in the left plot. All in all, most of the discussion was focused on Migration Safety topics rather than Political Security. But look in depth: after the NSL’s implementation, the focal point subtlety shifted to Conditions Imposed on Migration, while the Political Security topic, Soft Penetration, also gained attention.

Fig. 7.3

The effect of NSL on topic proportionFootnote

The number below the plot is the mean score of topic prevalence. Here, the total of the topic prevalence is 1.

The Effect of “Blue/Green” on Topic Proportion

Figure 7.4 illustrates the effect that the “Blue/Green” divide has on the topic pattern. In the pro-Blue Camp news outlets, on the left plot, Political Migration (T2), Insincere Assistance (T9), Conditions Imposed on Applicants (T1) occupy the top three spots. In contrast, the right plot, which represents the pro-Green Camp media, shows that the Political Security topics moved up. One is Soft Penetration (T3), which occupies the second position but is ranked last in the pro-Blue Camp plot. Similarly, another one, Population Replacement (T12), advances one ranks to the fifth position in the pro-Green Camp plot, compared to the one in the pro-Blue Camp plot.

Fig. 7.4

The effect of “Blue/Green” on topic proportion

Topic Content

This section closely examines the content of the high-ranking topics between pro-Green and pro-Blue media coverage. Both outlooks strongly engaged with the Migration Safety topics, that is, Political Migration, Insincere Assistance, and Conditions Imposed on Migration. We will look at these topics first.

Political Migration

First, regarding the topic of Political Migration, the two camps had no differences in their news report stances. Both emphasised that the political situation in Hong Kong in the aftermath of the OCM was the primary reason for Hong Kongers’ emigration. Both camps’ news outlets believed that Hong Kong immigrants desired to live in a liberal democracy, particularly after the NSL’s implementation in July 2020. Moreover, the news articles on this topic reported on Hong Kong immigrants’ living circumstances and exposed the risks of migrating to Taiwan, including low salaries, homesickness, and the sincerity of their loyalty to Taiwan being doubted by local Taiwanese. In the interviews with Hong Kong immigrants, Chen, X.-Y. (2020) stated :

Following the passage of the NSL, Alan and Joys [respondents] decided to speed up their actions [referring to emigrating to Taiwan]. Joys stated that they are looking forward to receiving their national ID cards soon in order to vote for the [Taiwan] president in 2024, and that ‘this is a democratic experience that Hong Kong has never had before’. They also expressed concern about the situation in Hong Kong.

Stella [another respondent] stated that the low salary is the main barrier to living in Taiwan, and that she must work multiple part-time jobs and invest in order to maintain her income levels.

In other interviews, a Hong Kong illustrator expressed anxiety about the Taiwan identity issue by asking “if a Hong Kong person falls in love with or marries a Taiwanese after coming to Taiwan, will it be considered that they are trying to save the six million dollars immigration fees?” (Lu, 2022).

Insincere Assistance

This topic mainly reflected the pro-Blue news media criticisms of the DPP’s governmental assistance for Hong Kong migrants and political asylum seekers. News articles by the critics of the government’s Hong Kong Humanitarian Assistance Action and Caring Project (hereafter, Assistance Project) referred to it as “old wine in new bottles”, having no new or helpful mechanism for political asylum seekers, rather just intending to “handle them ambiguously”. In a view presented in a news article in the China Times (Ji et al., 2020), the KMT Member of the Legislative Yuan, Chen Yu-Jen, stated that “the whole Project does not appear to differ from current regulations [targeting political asylum seekers]” (because the current regulations had been available through general immigration channels such as investment immigration). In the same news article, another KMT legislator, Chen I-Hsin, also questioned whether the Project only targeted legal immigrants:

Chen Ming-Tong, Chairman of the Mainland Affairs Council, stated that this Project is not for rescue (received political asylum seekers) but rather assistance provided to those in need after entering the country. The question is, how can we help if there is no rescue plan (referring to taking in political asylum seekers from Hong Kong who entered Taiwan illegally first)? Furthermore, the targets of applicable assistance are confined to legal immigrants; could it be that we deport or detain Hong Kong and Macao residents who are under threat due to political factors and have entered Taiwan illegally? (Ji et al., 2020)

Professor Chen Yi-Xin, in an interview by the China Times (Cai, 2020), also stated that the Assistance Project was intended to handle the Hong Kong immigration wave in an ambiguous manner (using a case-by-case review) and thus did not detail the rescue targets due to political reasons. The reasons for this included an attempt to avoid angering mainland China, so Taiwan could not dramatically widen the definition of refugee for Hong Kong protesters as that could cause the over-acceptance of a large number of sensitive political activists. Professor Chen explained that the Taiwanese do not want “criminals” to enter, as the radical activists could undermine social order in Taiwan as they did in Hong Kong and thus would not lead to the government’s public support. He concluded that the “ambiguous” strategy could somewhat please all parties, but this “narrowly-defined rescue” (termed by Professor Chen) is actually far away from the DPP’s claim—that is, to “stand with Hong Kong”.

Conditions Imposed on Migration

This topic contained news articles about the conditions imposed on immigration applicants as a result of the record number of applications received during the 2019 Protests. Coverage from both camps agreed that applications for residency in Taiwan had reached a record high, as shown in the news headlines, Hong Kong people’s residence in Taiwan hits a new high, 11,173 last year, in LT, and, The number of Hong Kong people who came to live in Taiwan increased by 1.5 times in the first four months, in EDN (Chen, 2022a; Cheng, 2020). Nevertheless, there was a subtle distinction between the two media camps on this topic. The pro-Blue news outlets were likely to criticise the DPP administration’s lack of sincerity in supporting Hong Kong, and its use of national security concerns as an for refusing Hong Kong immigrants’ settlement in Taiwan. A UDN news story (Chen, 2022) raised inquiries regarding the “secret” imposition of a quota on Hong Kong applicants in immigration by the DPP government. The news story suggested that such a measure may have been implemented due to apprehensions that the immigration of individuals with affiliations to Chinese party organisations, government, and military, who had previously served in Hong Kong, may be utilised for the purposes of the “United Front” in Taiwan.

The LT does portray apprehension about migration issues. The phenomenon of “fake investment, true migration” (using investment in Taiwan as a pathway to residency without actually making the investment) was highlighted (Chen, 2022c). A similar concern also appeared with reference to applicants for professional migration, as exemplified in a news title, People from Hong Kong and Macau who apply for professional migration should engage in business as opposed to retiring and enjoying medical insurance (Chen, 2021). Furthermore, a news story on migration’s national security implications prompted the DPP administration to reach a compromise (Chen, 2022b). One plan is issuing a permanent resident identification card to those Hong Kong immigrants in lieu of granting them citizenship. In the news article, an official explanation of this compromise measure was that it could lower the risks to Taiwan’s security because the applicants would not have citizenship.

To summarise, the three topics about Migration Safety portrayed the risks for Hong Kongers of living in Taiwan (e.g., low salaries, doubt about the immigrants' loyalty to Taiwan), and also looked at the impact that the large flow of Hong Kong immigrants had on concerns such as national security, social welfare distribution, and administration resources, which caused or justified the DPP government’s limited assistance to the applicants.

Soft Penetration

This subsection examines the difference between pro-Green and pro-Blue news outlets with regard to Political Security topics, focusing particularly on the pro-Green media. The Soft Penetration topic raised the possibility of Chinese cultural penetration through immigration. Most contributors to this topic took the situation in Hong Kong as an example of how Beijing took control of the city through immigration from mainland China. The contributors believed that the Chinese immigrants were filled with authoritarian beliefs. An LT editorial exemplified this phenomenon with Australian opinion poll results and discovered that the proportion of respondents (that is, Australian Chinese migrants) who identified with a democratic system (36%) was lower than the proportion who identified with a non-democratic system (41%). In an LT interview (He, 2021), Lam Wing-Kee, a Hong Kong immigrant and a book seller in Taiwan who had been detained in mainland China for selling prohibited books in Hong Kong involving CCP General Secretary Xi Jin-Ping, said that the concept of the human rights was lacking due to Confucian culture (He, 2021). It is, he added in another interview (Yang, 2019), because Confucian ethics do not talk about human rights, being based on the idea that people are born unequally and have different social statuses.

Most articles on this topic suggested that Taiwan had to beware of Chinese penetration and thus should cut the cultural tie between Taiwan and China. One solution, as Lin proposed, was to transform the Confucian-centric perspective in Taiwanese society through education. Another was to prohibit any plan for mainland Chinese immigration. As a Hong Kong resident claimed in a letter to the LT’s editor (Chen, M.-Y., 2019), the reasons the notion of human rights should not apply to China’s “emigration penetration strategy” were because a huge influx of Chinese immigrants would “dilute” Taiwanese minds, and some of them could be spies, as happened in Hong Kong. The letter to the LT’s editor was not only one person’s view; it also projects society’s mindset in Taiwan regarding Hong Kong immigration, as discussed in the following section, about the topic Population Replacement.

Population Replacement (Topic 12)

Like the Soft Penetration topic, contributors in this topic examined the Hong Kong case; that is, the daily quota of 150 mainland Chinese immigrants entering Hong Kong (Yang, 2011, pp. 10–11), which had resulted in an estimated more than one million new immigrants entering Hong Kong from mainland China since the 1997 Hong Kong handover (Hong Kong Economic Times, 2019). This situation was discussed in a letter to the LT’s editor (Chen, Z.-G., 2019), stating that, “[the 1.5 million mainlanders] have crammed into a tiny place, Hong Kong, making it like a sardine can containing more than seven million individuals. Public facilities, medical resources, and educational resources are depleted. How can any hope for human rights exist?”.

The alert mindset of the contributors regarding the issue of “fake immigrant, true Chinese spy” (Chen, Y.-C., 2020) reflected a cautious stance towards Hong Kong asylum seekers. Lin Pao-Hua (2019), a senior LT columnist, took a conservative stance on the refugee law for two reasons, in the news article titled “The refugee law and the spies in the background”. First, he said it would incur a large-scale influx of Hong Kong refugees entering Taiwan, potentially exhausting Taiwan’s human resources and management costs. As Lin pointed out, the worst-case scenario for Taiwan would be to become like Hong Kong was in the 1980s with the Vietnam refugee issue (also called the Vietnamese boat people) or as happened with the pre-handover illegal immigrants from mainland China. Second, and even more significantly, Lin believed that China would use the refugee law to conduct spying. Someone pretending to be a pro-democracy supporter could intent to incite rebellion—a personal experience he shared. Lin suggested that Taiwan should handle Hong Kong immigrants in a different way, by enacting more relaxed regulations to allow Hong Kong youth to study or work. In addition, political asylum regulations should be improved, according to Lin, but taking national security into account to cover investment immigrants.

The examination of topic contents in this section demonstrates that pro-Blue media tended to report on the situation of Hong Kong immigrants in Taiwan and to criticise the DPP government’s assistance, whereas the pro-Green media tended to highlight potential national security and management cost issues related to immigrants.

The Results of the Sentiment Analysis

Sentiment analysis is a dictionary-based text-mining technique for extracting opinions (Lei & Liu, 2021), which has been used to grasp the public sentiment towards refugees (Backfried & Shalunts, 2016; Öztürk & Ayvaz, 2018) or migration (Heidenreich et al., 2020). This chapter uses sentiment analysis to identify tendencies in the media’s partisan leanings and to look at the portrayals related to them.

The technique involves using a sentiment dictionary to count the words in the news articles that match those in the established dictionary. To improve the accuracy of word identification, the sentiment dictionary used in this study is a combination of two sentiment dictionaries created by National Taiwan University and the Dalian University of Technology from mainland China (translating from simplified Chinese into traditional Chinese when used). The calculation is straightforward. One word (e.g., “assistance”, yuanzhu) with a positive meaning in the dictionary is counted 1, whereas one word (e.g., “illegal”, feifa) with a negative meaning in the dictionary is counted -1. When the sum of the positive and the negative words in a news article is a positive number (i.e., 1 or above), the document is considered to contain positive sentiment. A negative sum (i.e., −1 or below) indicates that the document is considered to contain negative sentiment. Words are not counted if they are not listed in the sentiment dictionaries used in the chapter.

To conduct sentiment analysis, the same dataset used in this study was divided into two subsets: the pro-Blue Camp dataset (composed of UDN, CT, EDN and CTEE) and the pro-Green Camp dataset (represented by LT). By doing so, we returned 117 and 142 hits, respectively. The results of pro-Green and pro-Blue news outlets are depicted in Fig. 7.5. For the group of pro-Green news outlets, the number of news articles (41 of 117, or 35.04%) expressing negative sentiments regarding Hong Kong immigration was greater than the number of news articles (76 of 117, or 64.96%) expressing positive sentiments regarding Hong Kong immigration. For the pro-Blue group, only 31 out of 142 (21.83%) expressed negative sentiments, while 111 out of 142 (78.17%) expressed positive sentiments. This pattern indicates that the proportion of pro-Green news outlets having negative sentiments covering Hong Kong migration is greater (13.21%) than that of the pro-Blue news outlets.

Fig. 7.5

Sentiment analysis of the pro-Green and pro-Blue news outlets

The most positive and negative news articles from each camp were then selected based on sentiment scores, with the top ten news articles on each topic being included in the selection. For the pro-Blue media, the most positive (39 marks) news report (in the Insincere Assistance topic) was titled The Hong Kong Assistance Project has been announced; Chen Ming-Tong: Protesters seeking asylum must enter legally, and the law allows for assistance for Hong Kong students studying in Taiwan (Lu et al., 2020). This news story, from UDN, positively affirmed that the DPP government established a “Taiwan-Hong Kong Office for Exchanges and Services to Provide Necessary Assistance to Hong Kong Citizens”. Phrases used in the news report included, for example, “affirm the government’s specific approach” and “the information provided to protesters entering Taiwan has become more specific”.

The most negative news story was a CT editorial within the topic of Empty Promise of OCTS titled Hsu Guo-Yong “self-slaps” Hsu Guo-Yong (minus 42 points; Hsu was Minister of the Interior and in charge of immigration at the time). This editorial criticises the DPP government for “abandoning juridical sovereignty” in the “Chen Tong-Jia” case (a Hong Kong citizen suspected of murdering his Hong Kong girlfriend in Taiwan and then fleeing back to Hong Kong; both Taiwan and Hong Kong had no agreement for fugitives) in order to win the 2020 ROC presidential elections. The reason of this criticism is that this case would strengthen the case for the highly controversial Extradition Law, which was questioned for its violation of human rights and which sparked unprecedented protests in Hong Kong. If the party’s candidate had claimed support of the Extradition Law bill, it could have harmed the DPP’s chances of winning the presidency. Therefore, as the editorial added, the DPP government would not accept Chen’s surrender to Taiwan’s police.

For the pro-Green media, the most positive (23 marks) news report (in the Conditions Imposed on Emigration topic) was titled Hong Kong people’s residency cases in Taiwan hit a new high of 11,173 last year. This news story affirmed the DDP government’s effort to assist Hong Kong people settling down in Taiwan. As stated, “the government is revising immigration regulations for Hong Kong and Macau residents who wish to reside and settle in Taiwan, as well as enhancing the policy for assisting Hong Kong”. As to the most negative news story, it was also mentioned in Population Replacement, that is, titled the refugee law and the spies in the background (minus 29 points).

Conclusion

In contrast to studies either of traditional migration border security or of progressive–conservative media, this chapter's text-mining findings contribute to an understanding of how subjective perceptions, that is, the Blue/Green partisan leanings of Taiwanese media, influence the portrayal of the migration issue. It can especially be seen when the migration issue involves the PRC, which intensified the politicisation of the Hong Kong migration issue and the divergent portrayal of Hong Kong migration. It is evident that the pro-Green media outlets offer more of a security narrative (such as cultural Sinicization and the possibility of Chinese infiltration through Hong Kong migration), which serves to justify the DDP’s restrictive migration policy.

This chapter contends that while the media may aid in justifying such government policy actions, it also exacerbates the contradictions between human rights and national security. In other words, this tension calls the political legitimacy of the DPP and Taiwan into question, because they purportedly uphold the value of human rights while disadvantaging Hong Kong residents (especially those who are without full citizenship) who wish to reside and settle in Taiwan with ease. The side effect of the restrictive policies on Hong Kongers will impact social coherence in the long term. This chapter suggests that the matter of Hong Kong migration ought to be de-securitised and efforts by the Control Yuan should be redirected towards policy implementation concerns, such as administrative efficacy or the authories’ violations of the Hong Kong & Macao Regulations.

Last but not least, this chapter highlights that fear of the PRC motivates the pro-Green media’s negative portrayal of Hong Kong migration. However, it must be noted that this chapter has not delved comprehensively into the political roots of what structures the securitisation issue with regard to the Hong Kong migration. This includes an examination of Green Camp’s normative framing and political identity such as Taiwan’s subjectivity. These need further study but have already exceeded the scope of this chapter.

Appendix: A discussion of the topic model selection

We used a package stm (version 1.3.6) operating in R, specifically designed to produce the structural topic model used in this chapter. We worked in R. There are several steps in the two main stages: the pre-processing stage (Stage 1), and the topic model-building stage (Stage 2).

In Stage 1, we pre-processed the news data. First, we retrieved and downloaded news articles as textual data and converted them into a machine-readable UTF-8 format compatible with the RStudio platform. Next, we stored the data in a document database, MongoDB, using Studio 3T, a graphical user interface, to tidy the data according to the news page, date, title, main text, and so forth. We then underwent two text processing stages, word segmentation and stop-word removal. We used the R package jibea (a popular dictionary-based Chinese word segment module, version 0.11) to segment the news articles into a corpus. We then removed the stop-words (i.e., common words such as “others” [qi ta]). Once the corpus had been established, we went to the next stage.

In Stage 2, we first used the R package STM to convert the corpus into a document-term matrix. Our matrix contained 259 documents. We then had to determine a “correct” number of topics (number of word clusters) because there is no overarching gold standard. To find the best and most substantive model, we referred to the methods used by Quinn et al. (2010, pp. 215–216) and Lucas et al. (2015, p. 264), first narrowing down the searching scope. We thus set the number of topics, ranging from five topics to 40 topics in 5-topic increments.

After referencing the model’s diagnostic scores for “exclusivity” (the top words in a topic that significantly differ from others) and “semantic coherence” (the top words in a topic that are most likely to co-occur in a passage) (Fig. 7.6), the author independently close read the top news articles and top words in each candidate's topic model to validate the candidate models.

Fig. 7.6

Diagnostic values by number of topics in the candidate models

We chose the 12-topic model for our research task. In the topic selection step, we chose relevant topics and discarded uninterpretable or useless topics to avoid overwriting the selected model (e.g., fragmented, over-low prevalence). For example, Topic 6, Hong Kong Culture, was not closely related to our research tasks, and Topic 5 was an over-fragmented topic mixing economic strategy, conflict among generations and tourism, so the topics dropped. This step resulted in the seven topics used in our study for further interpretation. The author assigned corresponding labels in accordance with the core meanings of the top ten news articles for each topic.