Journal of

Current Chinese Affairs

China aktuell



Sullivan, Jonathan and Sapir, Eliyahu V. (2012),
Ma Ying-jeous Presidential Discourse, in: Journal of Current Chinese Affairs, 41, 3,
3368.
ISSN: 1868-4874 (online), ISSN: 1868-1026 (print)

The online version of this article and the other articles can be found at:
<www.CurrentChineseAffairs.org>

Published by
GIGA German Institute of Global and Area Studies, Institute of Asian Studies
in cooperation with the National Institute of Chinese Studies, White Rose East Asia
Centre at the Universities of Leeds and Sheffield and Hamburg University Press.

The Journal of Current Chinese Affairs is an Open Access publication.
It may be read, copied and distributed free of charge according to the conditions of the
Creative Commons Attribution-No Derivative Works 3.0 License.

To subscribe to the print edition: <[email protected]>
For an e-mail alert please register at: <www.CurrentChineseAffairs.org>

The Journal of Current Chinese Affairs is part of the GIGA Journal Family which includes:
Africa Spectrum Journal of Current Chinese Affairs Journal of Current Southeast
Asian Affairs Journal of Politics in Latin America <www.giga-journal-family.org>

This Taiwan edition has been published and edited in cooperation with the European
Research Center on Contemporary Taiwan (ERCCT) at Eberhard Karls University of
Tbingen.

Journal of Current Chinese Affairs 3/2012: 3368
Ma Ying-jeous Presidential Discourse
Jonathan SULLIVAN and Eliyahu V. SAPIR
Abstract: Despite the substantial advances made in cross-Strait relations
during Ma Ying-jeous (Ma Yingjiu) first term, the ROC presidents rhet-
oric varied considerably as he grappled with the difficult reality of im-
plementing campaign and inauguration pledges to establish better rela-
tions with China while striving to maintain national respect and sover-
eignty. In this article, we put forward a framework for measuring, analys-
ing and explaining this variation in President Mas first-term discourse.
Analysing a very large number of Mas speeches, addresses, etc., we pro-
vide empirical assessments of how the content of Mas public pronounce-
ments has developed over time, how his rhetoric varies according to the
strategic context and timing of a speech, and how his discourse com-
pares to that of his predecessor, Chen Shui-bian (Chen Shuibian). In
addressing these questions, the article contributes a quantitative perspec-
tive to existing work on political discourse in Taiwan and to the growing
methodological and applied literature on how to systematically analyse
Chinese political text.
Manuscript received 29 January 2012; accepted 1 August 2012
Keywords: Taiwan, Ma Ying-jeou, ROC president, presidential dis-
course, automated text analysis, cross-Strait relations
Dr. Jonathan Sullivan is a lecturer in China and Global Politics at the
University of Southampton. He co-wrote this article while similarly em-
ployed at the University of Nottingham. His research focuses on political
communications in various East Asian contexts.
E-mail: <[email protected]>

Dr. Eliyahu V. Sapir is a research fellow in the Department of Political
Science at Maastricht University. His research interests include democra-
tization processes, various aspects of political behaviour and political
methodology.
E-mail: <[email protected]>
34 Jonathan Sullivan and Eliyahu V. Sapir
Introduction
Cross-Strait relations during Ma Ying-jeous (Ma Yingjiu) first term as
ROC president developed in stark contrast to those during the term of
his immediate predecessor, Chen Shui-bian (Chen Shuibian) (Muyard
2010). Yet, despite warming relations between Beijing and Taibei, mul-
tiple political obstacles remain in place and are likely to persist in the
foreseeable future. Although semi-official, Track Two and party-to-party
interactions burgeoned during Mas first term, the ROC presidents rhet-
oric remains an influential (for better or for worse) conduit of cross-
Strait relations. Developing reliable methods for analysing presidential
discourse in Taiwan is thus an important undertaking in terms of policy,
theory and methodology. Specifically, the article provides a systematic
empirical assessment of Mas presidential discourse during his first term.
Contrary to popular representations of his rapprochement policy, Mas
rhetoric on issues relating to cross-Strait relations shows substantial vari-
ation in content, tone and emphasis. As a presidential candidate, he ran
newspaper advertisements containing the headline message Taiwans
future must be decided by the Taiwanese people ([[]
+j[;[, Jianjue zhuzhang Taiwan de qiantu bixu you Tai-
wan renmin ziji jueding), a position not dissimilar to that of his Democratic
Progressive Party (DPP) opponents. Indeed, as a presidential candidate,
close to one-third of the ideological appeals in Mas advertising ad-
dressed Taiwanese identity (Sullivan 2010: 149), and his stated policy
preferences shifted toward those of his DPP opponent, Frank Hsieh
(Xie Changting) (Jacobs 2012: 235). At other junctures, President Ma has
espoused a substantially different view on ROC sovereignty and the
relationship with China, to such an extent that some observers in Taiwan
have questioned whether he is even interested in preserving Taiwans
sovereignty (Jacobs 2012: 243). Throughout his first term, Mas rhetoric
on cross-Strait relations varied considerably as he faced up to the diffi-
cult reality of implementing campaign and inauguration pledges to estab-
lish better relations with China while maintaining national respect and
sovereignty. The article advances a framework for measuring, analysing
and explaining this variation in Mas first-term discourse. How does the
content of Mas public pronouncements develop over time? Does his
discourse vary according to the strategic context of primary audiences or
the timing of a speech? How does his discourse compare to that of his
predecessor, Chen Shui-bian? In addressing these questions, the article
contributes a quantitative perspective to existing work on political dis-
Ma Ying-jeous Presidential Discourse 35
course in Taiwan (e.g. Lams and Liao 2011; Sullivan and Lowe 2010;
Sullivan et al. 2011) and to the growing methodological and applied litera-
ture on how to systematically analyse Chinese political text (e.g. Chen
2011; Hassid 2012; Sullivan and Renz 2010).
Cross-Strait Developments in Mas First Term
The long-term and linked processes of democratization, nation-building
and nationalism in Taiwan have fundamentally influenced the dynamics
of cross-Strait relations (Chu 2004; Dittmer 2005; Wu 2004, 2005). Un-
der former presidents Lee Teng-hui (Li Denghui) and Chen Shui-bian, a
combination of the exigencies of electoral competition and an evolving
position on ROC sovereignty and Taiwanese national identity (Fell 2005;
Schubert 2004) rendered Taibei the most dynamic and volatile actor
involved in the cross-Strait relationship. Until recently, the absence of
channels of communication between Taibei and Beijing meant that high-
level interactions predominantly took the form of unilateral public pro-
nouncements on each side. As a result, cross-Strait relations followed a
cyclical pattern of working silences punctuated by provocative state-
ments (Dittmer 2004: 478) issued by presidents Lee and Chen. Whether
indicative of strategic calculations or deep-seated convictions, the pro-
pensity of both leaders to issue statements mentioning the special state-
to-state relations and that there is one country on each side, com-
bined with Beijings intractable One China position, has resulted in
periodic spikes in tension across the Strait. Because of their public pro-
nouncements, both former presidents came to be characterized as pro-
independence. These pronouncements have also had concomitant ef-
fects on Taiwans relations with Beijing and, to some extent, with the US
(Ross 2006). Consequently, rapidly growing, informal socio-economic
connections across the Strait were accompanied by political relations that
were essentially deadlocked in Lees term as elected president and during
Chens two terms. However, unofficial visits to the mainland by Lien
Chan (Lian Zhan) and other Kuomintang (KMT, Guomindang) figures
during Chens fractious final term paved the way for a rapid tightening of
ties following Ma Ying-jeous presidential election victory in 2008.
President Ma entered office after a landslide victory and thus had an
unequivocal mandate to improve cross-Strait relations and reinvigorate
the economy. All but the most partisan DPP supporters were disillu-
sioned with the Chen administrations governance problems (not all of
36 Jonathan Sullivan and Eliyahu V. Sapir
its own making), over-emphasis on ideological issues and, ultimately,
high-level corruption scandals (Copper 2009). With a correspondingly
large majority in the legislature, an untainted personal lustre and facing
an opposition in disarray, Ma was well positioned to implement the poli-
cies he had campaigned on (Hughes 2009; Jacobs 2008). He quickly en-
dorsed the 1992 Consensus namely, that there is one China, though
the ROC and PRC have different interpretations about which govern-
ment is the legitimate representative which was Beijings minimum
requirement as a basis for open high-level contacts, and which Chen had
rejected. Soon after, the semi-official Straits Exchange Foundation (SEF)
and Beijings Association for Relations Across the Taiwan Strait
(ARATS) resumed dialogue for the first time since the mid-1990s. Fo-
cusing on easy economic issues, regulating practices that were already a
reality and smoothing economic interactions that did not impinge on
security or sovereignty issues, they quickly endorsed an agreement to
allow regular weekend charter flights across the Taiwan Strait, soon fol-
lowed by agreements to allow mainland tourists to visit Taiwan, and to
create direct shipping links, implement daily flights and improve postal
services across the Strait. A year later, further initiatives on food safety,
crime, and financial cooperation were quickly passed by the KMT super-
majority in the Legislative Yuan. These practical successes paved the way
for the much more ambitious and, within Taiwan, politically contested
Economic Cooperation Framework Agreement (ECFA). Despite do-
mestic opposition, the ECFA passed in June 2010, removing tariffs on
hundreds of products and becoming the centrepiece policy of Mas first
term.
Mas endorsement of the Three Nos (No unification, No inde-
pendence, and No war) and 1992 Consensus generated unprecedented
goodwill and constituted a workable platform for engaging China. Con-
sequently, cross-Strait relations quickly reached a historical level of coex-
istence and cooperation, manifest in the discontinuation of unseemly
competition for diplomatic allies and Taibeis annual and fruitless mar-
shalling of allies to propose its entry into the UN (Wang, Lee, and Yu
2011). In return for the entente cordiale ushered in by the Ma administra-
tion, Beijing made concessions on Taiwans participation in international
organizations (a key issue for Mas domestic competitors and public
opinion). For example, the two sides reached an understanding that
would enable Taiwanese representatives to attend the World Health
Assembly meeting in 2009. Although it required adopting the designa-
Ma Ying-jeous Presidential Discourse 37
tion Chinese Taibei (rather than Taiwan, as the Chen administration
had insisted), participation had been a goal for Taibei (buttressed by
strong domestic support) since the SARS crisis in 2003. Zhang (2011)
identifies both ideological and materialist motivations behind Mas strat-
egy of engaging China, emphasizing goals of maximizing opportunities
and minimizing threats. Clearly, relations with China represent simultan-
eously Taiwans greatest opportunity (economically) and threat (in terms
of national security), and during Mas first term he was keen to assure
Beijing that Taibei did not seek to undermine the One China principle.
Disaggregating the ROC Presidents Discourse
In order to analyse President Mas discourse, it is first necessary to iden-
tify the parameters of political discourse in Taiwan. To do this, we have
drawn on the Taiwan Studies literature and identified five major themes
in political competition in Taiwan: Chinese national identity, Taiwanese
national identity, sovereignty, democracy and the economy. These con-
cepts, chosen for their relevance to cross-Strait relations, the main focus
of the article, provide the framework for our subsequent empirical analy-
ses. Naturally, these categories are not an exhaustive disaggregation of all
potential political discourses in Taiwan. Although we argue that these
categories cover the major components of Mas discourse, there are other
topics, such as the role of the KMT, which are not explicitly accounted
for. In the case of the latter, we do not cover it in this analysis as it is
mainly a domestic issue with less relevance for cross-Strait relations.
However, our analysis is not intended to be the final word on Mas dis-
course, and further work, both quantitative and qualitative, is required to
evaluate more nuanced discursive elements.
National Identity
National identity has long been the dominant cleavage underpinning
Taiwans party situation (Hsieh 2004: 479). The domestic dimension of
national identity is rooted in relations between native Taiwanese and
mainlanders (for discussion and critique of this nomenclature, see
Corcuff 2004). The former category refers to Han Chinese whose ances-
tors migration from Fujian Province in southern China preceded the
Taiwans Japanese colonization, which began in 1895. The latter are also
Han Chinese, but come predominantly from other provinces, most arriv-
38 Jonathan Sullivan and Eliyahu V. Sapir
ing on Taiwan following the emergency evacuation of nationalists be-
tween 1947 and 1949. Suspicious of native Taiwanese sympathies to their
former colonial masters, the liberating nationalist army was involved in
several incidents of aggression from 1945 onwards. When the KMT and
its supporters evacuated en masse to Taiwan between 1947 and 1949
they immediately filled the power vacuum left by the departed Japanese,
leading to resentment amongst native Taiwanese who felt that one colo-
nial master had been replaced by another. Relations between the main-
landers and native Taiwanese were deeply scarred by the February 28
th

incident and the subsequent murder of Taiwanese civilians in 1948,
which Hsiau (2000) argues marked the beginning of a specifically Tai-
wanese history. The pursuant White Terror, in which many native Tai-
wanese elites were arrested or killed, soured sub-ethnic relations for dec-
ades (Roy 2003).
Conceiving of itself as the defender of the authentic Chinese nation
and culture (which were under attack from the communists on the main-
land), the KMT suppressed Taiwanese culture and language. The party-
state barred Taiwanese from participating in politics and public policy
and oversaw a strong authoritarian regime that benefitted from aid, trade
and military support from the US. Relegated to the margins of society,
native Taiwanese, who made up 85 per cent of the population, neverthe-
less benefitted from the economic miracle created in part by the
KMTs developmental state framework (Wade 1990). By the 1970s
many native Taiwanese were economically empowered and nativist activ-
ists began to push for democratization, ethnic justice and an inde-
pendent Taiwanese state. As the demographics of the island have
changed, sub-ethnic boundaries have become blurred through intermar-
riage and the internalization of Taiwan-centred identities by many main-
landers (Corcuff 2002). The ethnic make-up of the KMT has diversified,
partly a matter of necessity as older members have died. Native Taiwan-
ese have risen to positions of political power at all levels of society, in-
cluding the presidency. However, national identity remains one of the
most salient issues in Taiwanese politics, with two distinct blocs coalesc-
ing around the unresolved politics of the Blue/ Green identity cleavage
(Wu 2008). Blue refers to the KMT and its allies, traditionally mainlander
and pro-unification. Green refers to the DPP and its allies, traditionally
native Taiwanese and anti-unification. It should be noted that there are
also geographical and socio-economic features to the divide, and the
correlation between the Green and Blue blocs and ethnic identity, or
Ma Ying-jeous Presidential Discourse 39
preference for independence or unification, is weaker now than in the
past.
Under Chen Shui-bian, the DPP insisted on a specific Taiwanese
national identity (Schubert 2004: 548) consistent with its genesis as an
opposition party with a strong nativist orientation. Lynch argued that
Chen attempted to imagine a completely new and genuinely autono-
mous Taiwan (Lynch 2004: 514). Chens deliberate nation-building
effort (Dittmer 2004: 475) included aggressive Taiwanization and de-
Sinification programmes and attempted to consolidate and routinize
Taiwans separation from the PRC (Wachman 2002). Cabestan has ar-
gued that Taiwanese nationalism expressed opposition between main-
landers and native Taiwanese rather than a desire to separate from the
Chinese nation (Cabestan 2005: 34). Certainly, Chens attempts at identi-
ty formation were inhibited domestically by split sub-ethnic identities
(Dittmer 2005: 72), although both major parties evince love for Tai-
wan in their election campaign platforms (Fell 2005). Born outside of
Taiwan, Ma Ying-jeou adapted to the normalization of Taiwan-centric
identification in the electorate by framing himself as a new Taiwanese,
the concept used by Lee Teng-hui in his endorsement of Ma prior to the
1998 Taibei mayoral election. The so-called new Taiwanese identity
attempted to de-essentialize Taiwan identity, allowing non-native Tai-
wanese who shared in Taiwans post-war collective experience to claim a
form of Taiwanese identity. During his time as Taibei mayor, Ma careful-
ly constructed his new Taiwanese image, learning local languages and
avowing his love for Taiwan. Mas identification with the ROC and sup-
port for the 1992 Consensus have not been changed by his transfor-
mation into a new Taiwanese. In addition to his entente policies, Ma has
reversed many of the de-Sinicization/ Taiwanization initiatives carried
out by the Chen administration.
Sovereignty
The external dimension of national identity in Taiwan involves unre-
solved questions of national status, sovereignty, and relations with the
PRC. The basis of this conflict is often stylized as one of independence
versus unification, although the reality is more nuanced than this di-
chotomy suggests, certainly in terms of public opinion (Wang and Liu
2004). As a party-state made up of mainland elites who took it upon
themselves to represent and preserve authentic Chinese culture in the
face of the communist revolution on the mainland, the KMT naturally
40 Jonathan Sullivan and Eliyahu V. Sapir
associated with the Chinese motherland rather than the expediently lo-
cated island of Taiwan. Indeed, the KMT government conceived of Tai-
wan explicitly as a temporary base for the recovery of the mainland
and frequently treated it as such (Roy 2003). Even when the likelihood of
achieving this ambition receded, the goal of the KMT and its supporters
remained unification (Corcuff 2004). However, demographic and genera-
tional change within the population and the party led to a change in the
KMT position coinciding with democratization processes. By the late
1990s, the policy of indigenization begun in the 1970s had radically al-
tered the ethnic make-up of the KMT. For instance, the proportion of
native Taiwanese on the KMT Standing Committee in 1969 was less
than 7 per cent. In 1998, this had risen to 50 per cent (Rich 2009: 2). The
Temporary Provisions were lifted in 1991, effectively giving up the
ROCs claim to the mainland and marking the start of the transition to
Lee Teng-huis two states position. The KMT no longer fits the sim-
plistic description of a mainlander/ unificationist party, although its
ideological orientation is clearly much closer to Chinas than is the
DPPs, the latter often having been styled as a pro-independence par-
ty, although it depends how independence is defined (see Sullivan et al.
2011).
The desire for an independent Taiwan became a rallying call for
ethnic frustrations during the 1960s and 1970s (Copper 1997). Inde-
pendence supporters were harshly dealt with for introducing the pro-
spective of a previously inconceivable nation: a republic of Taiwan that
would be independent of both the PRC and the ROC. The issue of Tai-
wanese independence was taken up by the DPP when it formed in the
mid-1980s, partly to galvanize a weakly institutionalized organization
made up of diverse anti-KMT factions. The DPPs position on inde-
pendence was solidified in 1991 when a plebiscite on independence was
written into the party platform. However, with the PRC equation of
Taiwanese independence equals war (for example, Article 8 of the
Anti-Secession Law states that the PRC will use all necessary means to
prevent Taiwans separation from the mainland), public support for
independence expressed in opinion surveys and at the polls has never
exceeded a tiny minority. Following Peng Ming-mins (Peng Mingmin)
failed independence-based campaign for the presidency in 1996, the
DPP qualified its stance on independence in its 1999 Resolution on the
Future of Taiwan, converging on a position similar to Lee Teng-huis
that the ROC has been a sovereign state since it was founded in 1912
Ma Ying-jeous Presidential Discourse 41
[and] consequently there is no need to declare independence (extracted from Lees
interview with Deutsche Welle in 1999; emphasis added). After espous-
ing a similar position during the presidential campaign in 2000, Chen
Shui-bian aggressively promoted Taiwanese sovereignty and steadfastly
rejected the existence of One China, even with different interpreta-
tions. And after losing the presidential election in 2004 the KMT
adopted a much stronger pro-China position. Yet, radical moves to-
ward either independence or unification are constrained by popular
support for the status quo. The configuration of Taiwanese public opin-
ion has been stable and in favour of the status quo for 30 years (Paolino
and Meernik 2008), which has effectively rendered party support for
immediate independence or unification off-limits. When Ma proposed a
peace accord during his re-election campaign, it coincided with a rapid
downturn in support and was quickly dropped from his campaign agen-
da.
Democracy
Although restricted elections for local offices had been held since the
1950s, Taiwan was ruled by a one-party regime while the KMT govern-
ment remained technically in a state of war with the Chinese com-
munists. Under martial law, the KMT utilized existing social structures to
establish a political machine based on patronclient networks (Bosco
1992) and demobilized or co-opted every organized sector of society.
When rapid economic modernization increased the economic and politi-
cal influence of the numerically superior but politically marginalized
native Taiwanese, the KMT brought many of them into the party. Others
lent their support to a nascent democracy movement fuelled by per-
ceived ethnic and social injustices. The shock of diplomatic de-
recognition by the US and the loss of the UN China seat to the PRC
forced the KMT to look for alternative sources of legitimacy. Political
liberalization was one way to both retain the moral support of the US
and release growing domestic tensions. Seizing on these developments,
the Dangwai opposition movement (literally outside the party) in-
creased its demands for democratic reform. Major Dangwai figures were
arrested in 1979, and this was followed by a temporary tightening of
openings allowed to activists. However, the setback known as the Mei-
lidao Incident did not derail the gradual trend toward liberalization.
Compromises between the government and Dangwai gradually ushered in
democratization processes that were generally peaceful and extended
42 Jonathan Sullivan and Eliyahu V. Sapir
over a prolonged period (Lin and Chu 2001; Tien 1996; Tien and Chu
1996). By 1986, opposition activists were sufficiently organized to form
an opposition party, the DPP. Although this action was technically ille-
gal, the DPP was permitted to field candidates in the 1986 legislative
election and martial law was officially rescinded a year later.
The DPP initially linked the goal of democratization directly to the
issue of Taiwanese identity and the principle of self-determination (Lin,
Chu, and Hinich 1996), mobilizing support for ethnic justice internally
and independence from China externally (Clark 2004: 74). Thus there
has always been a symbolic dimension to democratization in Taiwan
(Corcuff 2002). Taiwans democracy is a source of pride and identifica-
tion for Taiwanese, both in the sense of identifying with the liberal
constitutional state of the ROC (Schubert 2004: 535) and in contradis-
tinction to the PRC. Former presidents Lee and Chen both campaigned
on their bona fide pro-democracy credentials in presidential elections,
Lee as the father of democracy in Taiwan and Chen as a lawyer associ-
ated with the Meilidao activists. As presidential candidate and president,
Chen campaigned vigorously for the need to broaden and deepen demo-
cratic reforms, but his major democratic innovations (particularly defen-
sive referenda and constitutional reforms) were clearly motivated by
instrumental concerns (Mattlin 2004). The KMTs establishment and
exploitation of non-democratic political structures during the one-party
era also persist in its influence on many sectors of society, including the
civil service, military, industry and the media. These interests, in addition
to the KMTs seeming inability to act as a loyal opposition during the
Chen era (Mattlin 2011), have made the party vulnerable to critique by
the DPP, which continues to frame itself as the pro-democracy party
despite Chens conviction for corruption. This context gives some indi-
cation of how democracy is a politicized issue with substantial ideological
baggage in Taiwan.
Issue Salience, Strategic Context and the Time
Dimension
In order to guide our analysis of President Mas discourse, we draw on
the theory of issue salience as a means to estimate the relative im-
portance of different discursive categories. Like other studies on salience
(e.g. Budge, Robertson, Hearl 1987), we exploit the assumption that the
more a political actor addresses a certain issue, the greater the im-
Ma Ying-jeous Presidential Discourse 43
portance he or she attaches to it. The salience approach has been used in
various contexts including Fells (2005) analysis of party politics in
Taiwan to identify cross-party and temporal variation in the im-
portance of various issues. Unlike Fell, however, we focus on the meta-
level issues encapsulated by our five discursive categories. The methods
we employ can be modified to measure specific issue areas and policy
sectors, but the role of the president and the function of his discourse
are different to party political communications such as manifestos and
campaign advertising. Our focus is thus on the salience of higher-level
concepts and aggregations of issues for example, those pertaining to
the economy. As with other forms of political communication, such as
campaign advertising (see for example, Sullivan and Sapir 2012), we ex-
pect that Mas discursive emphases (meaning, the salience of different
topics) will vary over time and according to a range of different factors.
One factor that we explicitly build into our analytic model is strategic
context. Any democratic political leader must engage multiple constitu-
encies and stakeholders, and as prior research on Taiwan demonstrates,
this feature of democratic communications affects the content of politi-
cal leaders public pronouncements (Lin 2005). In their analysis of what
Clark called Chen Shui-bians strategic ambiguity elucidating differ-
ent positions at different times and making different appeals on the same
issue to disparate constituencies (Clark 2004: 80) Sullivan and Lowe
(2010) found that the identity of the audience Chen was addressing was a
significant predictor of the content of his speeches. For example, when
addressing overseas independence supporters, Chens speeches included
a much greater amount of language associated with Taiwanese independ-
ence than his speeches to business leaders or foreign dignitaries did.
Sullivan and Lowe conclude that it was no coincidence that Chens infa-
mous one country on each side speech was delivered to a pro-inde-
pendence support group overseas (Sullivan and Lowe 2010). In colloqui-
al terms, this behaviour is known as playing to the crowd. More tech-
nically, it is consistent with spatial politics models where the gains for
modifying appeals to the median voter position of a single audience out-
weigh the risks of moving away from the median voter position in the
electorate as a whole (for research on modelling cheap talk, see Aus-
ten-Smith 1990). Because Ma makes a large number of speeches (as
Chen before him did) and the vast majority of these addresses go unre-
ported in the media, the opportunity costs associated with telling an
audience what they want to hear are relatively low. Modifying speech
44 Jonathan Sullivan and Eliyahu V. Sapir
content is also consistent with studies of rhetoric, where rhetors are
expected to adjust the content of their speech according to the expecta-
tions of their audience (Herrick 2008).
Like those of his predecessor Chen, we expect that Mas speeches
on the discourse dimensions we have elucidated above should show
variation according to the identity of the immediate audience to whom
the speech is delivered and, by implication, the strategic relationship he
has with that audience. Extending the analysis of Sullivan and Lowe
(2010), we further note the importance of timing in political communica-
tions and put forward a method for incorporating the time dimension
into our analysis of presidential speech. The concept of time has long
occupied a place in the formal theoretical literature for example, in
Schellings (1969) work on political bargaining and in research on signal-
ling games (Sobel 2007) but it has been largely neglected, particularly as
an independent variable (Meyer-Sahling and Goetz 2009), in analyses of
political discourse. We argue that the timing of a speech should have an
observable effect on the content of presidential discourse, as democratic
leaders react to a range of agent-based and contextual factors. There is
prior evidence that time has an effect on presidential speech in Taiwan.
Sullivan and Lowe (2010), for instance, show how the Anti-Secession
Law and Chens re-election had substantial and statistically significant
effects on the content of Chens speeches. Building on this prior work,
we put forward a method for identifying how time interacts with strate-
gic context to affect the content of presidential speech; in other words,
we attempt to show how Mas addresses to different constituencies vary
at different points in time.
Methods and Data
In order to identify and assess the extent to which Mas speeches reflect-
ed our five discourse categories (Chinese and Taiwanese identity, sover-
eignty, democracy, the economy), we employed computer-assisted con-
tent analysis (CCA) (Popping 2000). Content analysts assume that al-
though theoretically relevant categories are not directly observed, par-
ticular words and phrases reflect this latent content in a systematic way
(Krippendorf 2004). More specifically, a speakers choice of words and
the frequency with which he or she uses them indicate latent content
(Neuendorf 2002). Although single words or phrases are unreliable indi-
cators of a complex concept, combining multiple indicators in a theoreti-
Ma Ying-jeous Presidential Discourse 45
cally informed category structure has been shown across the social sci-
ences to be a reliable mechanism for tapping latent content (Benoit et al.
2005; Budge, Robertson, and Hearl 1987; Laver, Benoit, and Garry 2003;
for many diverse examples see Krippendorf and Bock 2009). Further-
more, automated text analytical methods are increasingly common in
political science (e.g. Benoit, Laver, and Mikhaylov 2009; King and Lowe
2003; Slapin and Proksch 2008) and in the analysis of Chinese texts (e.g.
Hasid 2012; King, Pan, and Roberts 2012; Sullivan and Renz 2010).
We assume that each of Mas speeches can express a different vol-
ume of language associated with our five categories, in addition to other
topics not covered by our dictionaries. In some cases, Ma does not use
words or phrases from any of our categories: If our content dictionary is
well constructed, this is because the speech does not cover any of our
five topics. The five discursive categories are thus signals to be distin-
guished from the background noise of other topics. Our strategic con-
ception of Mas public pronouncements allows the content of his
speeches to vary over time and by primary audience. We recognize that
the notion of a primary audience is not absolute in the case of public
political speeches, given that speakers sometimes implement a two-level
game (Putnam 1988). This describes situations where speakers simul-
taneously address the immediate audience to whom the speech is being
delivered and a second-level audience consisting of political opponents,
supporters, the media, the US or PRC governments, and so on. How-
ever, we argue that since we cannot control for Mas concerns about the
second audience, the local audience is the primary audience and he
must position himself with respect to them. Consequently, if there is
significant variation in our measures depending on the primary audience
we should see it, despite damping by second-level audience considera-
tions.
Measures of Sovereignty, Identity and Democracy
Our five major discursive categories were determined by our reading of
the literature on Taiwan. The sovereignty category is intended to cover
references to the ROCs existing independent status; its claim to sover-
eignty; boundary distinctions between the ROC on Taiwan and the PRC;
and advocacy of actions that reinforce these ideas for example,
strengthening the sovereignty claim by expanding the ROCs interna-
tional role. The national identity category is divided into Taiwanese iden-
tity and Chinese identity. In each case, the category covers signifiers of a
46 Jonathan Sullivan and Eliyahu V. Sapir
distinct collective identity based on common points of identification and
endorsements of or duties to the identifying collectivity. Indicators in the
two categories are distinguished by their focus on a specifically Taiwan-
ese identity that is distinct from identification with a broader Chinese
cultural and ethnic affiliation. As we can see from the results reported
below, there is significant overlap between the categories of sovereignty
and Chinese identity in terms of how frequently Ma mentions them, but
substantial differences in the emphasis that he places on each category.
In practice, he does not make claims about Taiwans sovereignty, which
would in effect equal claiming that Taiwan is or should be independent
and contravene his position on the sovereignty of the ROC (and the
1992 Consensus position that there is one China with different interpre-
tations). On the other hand, Ma sometimes does address Taiwanese
identity in terms of culture, values, etc. which are picked up by our
Taiwanese identity category and makes similar references to Chinese
identity, which are similarly distinguished. The democracy category covers
references to democratic achievements, the existence of rights and free-
doms, and further reforms. The economics category covers references to
the economy, financial policy, business, and trade.
We selected indicators (words and phrases) of these categories in-
ductively by manual analysis of a sample of 200 of Mas speeches (ap-
proximately 10 per cent) and using the recent literature on Taiwanese
politics as a guide. Table 1 shows our framework with example indica-
tors. The dictionary contains 250 patterns (approximately 50 in each
category) and it is from the frequency of these terms and their deriva-
tives in Mas speeches that data is derived. This dictionary is not intend-
ed to be the final word on textual indicators of our five categories. In-
deed, the advantage of using CCA is that theory-driven changes to dic-
tionaries are easily accommodated, allowing, where necessary, for an
improved fit between theory and data. Constructing a content dictionary
requires making subjective decisions about the connotations of words
and phrases in an on-going discourse. For example, we define ECFA
as an indicator of the economy, while other analysts (some in the opposi-
tion) may emphasize non-economic aspects of the ECFA. We
acknowledge that even our decision to divide speech content into five
categories is itself substantively important. However, if this theoretical
distinction is not reflected by appropriate differences in our data, then
we have some reason to believe that the distinction is in fact illusory.
The dictionary that we report here is part of an on-going project to eval-
Ma Ying-jeous Presidential Discourse 47
uate the discourse of all ROC presidents since 1992, and we expect it to
be refined and extended in future work.
Table 1: Organization of Dictionaries with Example Indicators
Category Example patterns in English Selected search entries
Sovereignty
One China, Senkaku Islands,
ROC 100 years, ROC constitu-
tion, enter the WHO, internation-
al status, national sovereignty,
international participation, nation-
al borders
[, )), T
[[)+, T[[
,, )/;y,
[|j|, [_j, [
|], [_],
Taiwanese
identity
The Taiwanese people, the Tai-
wanese nation, Taiwans future,
new Taiwanese, Gemeinschaft,
Chinas weapons, military threats,
Taiwan first, love Taiwan, the
Taiwan spirit
[;[,[[, [
[], [;, )]
[,[[y,, y[
, [j, [,
[|
Chinese
identity
Chinese culture, Chinese nation,
Chinese wisdom, 1992 Consensus,
all the worlds Chinese, Father of
the Nation (Sun Yat Sen), Confu-
cius, the Classics, people on both
sides, mainland compatriots
T|, T[,
T[, ),,
,T;, [,
,, |j;[,
],
Democracy
Democratization, democratic
reforms, democratic freedom,
human rights, freedom of speech,
political rights, citizen
rights/freedoms, liberal constitu-
tional system, referenda
[|, [j, [
j , ;j, j,
[,j], ,[j, ;
[[j], [[], ,
|, ,[|
Economy
ECFA, charter flights, 12 big
infrastructure projects, TaiEx,
Three Links, investment environ-
ment, economic development,
FTA, financial crisis
lCl^, j,
, ||), , |
(, [j, j
)j,j,
Source: Authors own compilation.
48 Jonathan Sullivan and Eliyahu V. Sapir
Data
The empirical basis of our analysis is the 2,162 speeches given by Ma
Ying-jeou from 20 May 2008 (his inauguration) through 30 June 2011
(the most recent date of our data collection). The speeches were down-
loaded from the website of the Office of the President <www.president.
gov.tw>. The website has undergone a major organizational change since
Chen Shui-bians time in office, when speeches were kept separate
from press releases, presidential decrees, etc. Now, presidential speeches,
press releases, decrees and all other presidential business are all housed
under the slightly misleading heading { (xinwengao, press release).
The restructuring of the website required a substantial effort on our part
to weed out extraneous information. We conducted this process twice,
once when we downloaded the raw data and again when we coded the
audiences. Thus we are confident that the speech set we examine in the
paper, with the normal degree of random error, represents Mas speech
output. Although the number of speeches may appear high, Chens out-
put was within the same range (see Sullivan and Lowe 2010). The full set
of Chinese texts was then cleaned and converted into a machine-readable
format for analysis. To construct the content dictionary we used the
Yoshikoder,

an open-source software package (available at yoshikoder.
org) developed by Will Lowe as part of the Identity Project at Harvards
Weatherhead Center for International Affairs. The Yoshikoder is a desk-
top application that runs on any operating system, can deal with text
documents in any natural language, and has produced valid and reliable
results using simplified (Hasid 2012; Sullivan and Renz 2010) and tradi-
tional characters (Sullivan and Lowe 2010). In a recent methodological
note, Chen (2011) reports several inconsistencies in the Yoshikoders
ability to identify search terms, thus compromising the validity of the
data generated. However, we were unable to replicate the difficulties
described in Chens paper, and can only surmise that her idiosyncratic
results were a result of glitches in earlier versions of the software. Con-
structing a dictionary is the most intellectually challenging process in-
volved in CCA, but in practice it is straightforward to achieve in the
Yoshikoder, which offers concordance functions, a wildcard function to
capture words and phrases, and a drag-and-drop system for organizing
dictionary entries. The Yoshikoder facilitates automatic searches inside
each document and records the number of times patterns from each
category are matched in the text. When we were satisfied with the dic-
Ma Ying-jeous Presidential Discourse 49
tionary we developed, we used a related application to generate counts
for all 2,162 speeches.
In languages with non-logographic writing systems, content analysis
is usually performed by counting how many times patterns match partic-
ular words. But, while separating Chinese-language sentences into
words is possible, it is still an acknowledged problem in computer
science. Therefore, in order to avoid introducing what are essentially
transcription errors into our data (for example, by segmenting a se-
quence of characters in a way that splits a semantic unit), we chose to
simply divide our texts into sentences, which are more easily determina-
ble units of analysis. Consequently, our counts represent how many sen-
tences in a speech contain a sequence of characters matching one of our
patterns; and when we refer to the proportion of a document expressing
Chinese identity, we mean the proportion of all of a given speechs sen-
tences that match a pattern defined in our Chinese identity category.
We use the following extract taken from Chen Shui-bians one
country on each side speech to illustrate the dictionary approach. Alt-
hough the focus of this article is Ma Ying-jeou, we use Chens speech
because it is, in our opinion, one of the most categorical delivered by any
ROC president. Underlined are search entries in the sovereignty catego-
ry, in bold Taiwanese identity and in italics democracy. In this short ex-
cerpt, there are six references to sovereignty, eight to national identity,
and two to democracy. There are some relevant phrases that were not
present in our category structure, which is inevitable, as it is part of the
trade-off between the number of indicators in a dictionary and its ability
to distinguish between concepts:
)j|}]+],|_][}}]
[[}}][[]}][[}(|]
||(;|)}][[}[[[}.[[
j}.[[;j}.[[1`}[}][[_}
][[_[j)|.j)|.j|,j|[);[
;,);[j[j.);[{[|[j,].
|[,[j)[[_][[j[
[] (Extracted from a speech delivered 3 August, see Chen
2002).
In addition to the content analysis of the speeches, we are also in-
terested in a strategic dimension: Mas decision to emphasize or de-
emphasize language related to a certain category in accordance with his
relationship to a particular audience. Analysing speech content across
50 Jonathan Sullivan and Eliyahu V. Sapir
time or across audiences is informative in itself, but each type of analysis
tells a different and incomplete part of the story. In this paper, we ana-
lyse each dimension separately and, importantly, interact both dimen-
sions. To distinguish audiences we have again drawn on the literature as
a guide to the groups Ma has had to engage. We identified 33 different
audiences, which we then aggregated into ten broader categories. The
aggregated categories are shown in Table 2 with examples of finer-grain-
ed audience distinctions. Audience codes are short forms of audience
names used as compact figure labels. With audience categories estab-
lished, we assigned an audience to each speech using metadata attached
to the speeches provided by the Office of the President. As an excep-
tion, in the case of New Years and National Day and other formal
speeches (such as addresses to the nation), we coded the form of the
speech rather than the audience.
Table 2: Breakdown of Audience Categories
Broad
audiences
Example narrow audiences Code
National Day,
New Year
Includes 1 January and Chinese New Year speeches ndny
Allies
Diplomatic allies, heads of state and other political lead-
ers, in Taiwan and overseas.
alli
Diaspora Taiwan and ROC diaspora groups located overseas dias
Domestic non-
political
Schoolchildren and students, athletes, cultural groups,
scientists, womens groups, environmentalists, lawyers,
rescue workers, recipients of awards, etc.
domn
Domestic
political
DPP party, policymakers, foundations, campaign sup-
porters, local media interviews
domp
Economic
Regional economic organizations, Taishang, Taiwanese
businesspeople, chambers of commerce
econ
Foreign digni-
taries
Politicians of non-diplomatic allies, scholars, cultural
leaders from non-allied countries visiting Taiwan and
addressed overseas
fdig
Foreign media Interviews with international publications fmed
Military All branches of armed forces and coast guard mili
Other formal
speeches
Inauguration speeches, addresses to the nation form
Source: Authors own compilation.
Ma Ying-jeous Presidential Discourse 51
Figure 1 shows the distribution of Mas speeches according to the audi-
ences that received them. Two out of every five speeches delivered by
Ma in this period were to audiences that were coded as domestic non-
political. Addresses to students, athletes, cultural groups, scientists,
womens groups, environmentalists, rescue workers, etc., thus account
for a substantial proportion of Mas speaking engagements. One in every
six speeches was delivered to economists, one in every eight to foreign
dignitaries, and one in every eleven to diplomatic allies. This distribution
shows a high degree of continuity with the behaviour of Chen Shui-bian
as reported in Sullivan and Lowe (2010).
Figure 1: Distribution of Mas Speeches to Different Audiences
Source: Authors own compilation.
Findings
To see how often each of the five categories is covered in a speech, it is
sufficient to record whether one (or more) pattern from a category is
matched by any sentence in a speech: a mention. If a speech has men-
tions relating to a category, we assume that the topic was addressed,
although we cannot yet say how much coverage Ma subsequently gave to
it. Figure 2 shows the estimations of a regression model of the content
category mentions over time. Each line in the figure corresponds to the
fitted probability of mentioning one of the categories. Figure 2 displays
the aggregate probabilities for all types of audience and is intended to
show the longitudinal trend of including each category in a speech. The
0 100 200 300 400 500 600 700 800 900 1000
formal
ndny
fmed
dias
mili
alli
fdig
domp
econ
domn
52 Jonathan Sullivan and Eliyahu V. Sapir
findings show that the economy was the most frequently mentioned
category in Mas speeches over time. Mentions of the economy were
almost uniformly high throughout Mas first term, and by the end of the
period covered he referred to this topic in two out of every three
speeches delivered. Mentions of sovereignty and Chinese identity grew
from four to five out of every ten speeches, while mentions of democra-
cy and Taiwanese identity respectively dropped from 38 per cent to 27
per cent and from 28 per cent to 16 per cent. Comparing these findings
to Sullivan and Lowes (2010) work on Chen Shui-bian, for which they
used a similar method, we find that Ma mentions democracy and Tai-
wanese identity substantially less often. For instance, during the early
period of Chens first term, there were mentions of democracy and Tai-
wanese identity in approximately 60 per cent of his speeches, rising at a
consistent rate until the anti-secession law was passed, at which point
Taiwanese identity issues increased to being mentioned in 80 per cent of
Chens speeches. Like Ma, Chen also mentioned the economy more than
any other single category, suggesting that this topic is consistently the
most pressing for two ROC presidents of otherwise substantially differ-
ent political orientations.
Figure 2: Probability of Mentioning a Speech Category
Source: Authors own compilation.
1
2
3
4
5
6
7
Chinese identity Taiwanese identity Economy
Democracy Sovereignty
Ma Ying-jeous Presidential Discourse 53
Examining mentions allows us to measure how often Mas speeches
cover one of our categories but does not allow us to see how much he
subsequently addresses them. To achieve this, we examined the propor-
tion of each speech devoted to each of our categories. Figure 3 shows
the results of a regression model fitted to the number of sentences in
each speech attributable to each of our categories over time. The longi-
tudinal trends show that the share of Mas speeches dedicated to the
economy, democracy and Chinese identity gradually increased as his term
progressed. Figure 3 shows that the economy is not only Mas most fre-
quently mentioned category, but also, by a substantial margin, the most
heavily emphasized topic. Indeed, Mas emphasis on the economy has
increased over his time in office unlike Chen Shui-bian, whose empha-
sis on the economy decreased the longer he was in power (Sullivan and
Lowe 2010). Again contrary to Chen, Mas emphasis on Taiwanese iden-
tity and democracy did not increase. Indeed, these two categories were
Mas least emphasized: Taiwanese identity mentions remained static, and
democracy mentions declined to a level below those relating to sover-
eignty 18 months into Mas tenure. Although these results are largely
confirmatory, from a methodological perspective it is encouraging that
they are consistent with more detailed qualitative analyses of Mas first
term. The results also provide an explanation for the concerns about
Mas presidency recorded by Jacobs (2012: 243) for example, the
strong emphasis on Chinese identity and declining attention to democra-
cy. The non-salience of Taiwanese identity in this large collection of Mas
presidential speeches is contrary to the relative importance he placed on
such claims in his first presidential campaign in 2008.
Up until this point we have reported on the content of Mas speech-
es in terms of all audiences together. However, as discussed above, we
suspect there to be variation in Mas speech content according to the
identity of the primary audience. We now turn to this part of the analysis,
which we achieved by disaggregating the audiences. Figure 4 sets out the
proportion of sentences in Mas speeches on each category divided by
audience. Looking at the economy topic, which we reported earlier as
being the most salient issue discussed by Ma in his speeches (averaging
over 1 per cent of his speeches to all audiences), the identity of the pri-
mary audience played a major role in the share of the speeches dedicated
to this topic. As expected, the largest share of language associated with
the economy was observed in speeches delivered to economic audiences.
When addressing commercial organizations, businesspeople, financial
54 Jonathan Sullivan and Eliyahu V. Sapir
institutions, etc., Ma dedicated over 11 per cent of his speeches to dis-
cussing the economy. In speeches delivered on National Day and New
Years Day, and those delivered to diaspora groups and foreign media, 5
per cent of the speeches dealt with the economy. Chinese identity was
most salient when Ma addressed the foreign media or diaspora groups, in
contrast to Chen Shui-bian, who used his speaking engagements with
these two groups to emphasize Taiwanese sovereignty issues. The issue
of ROC sovereignty was most salient in Mas National Day and New
Years speeches, and in those addressed to diplomatic allies. The issue of
democracy was also more evident in addresses to foreign media and
dignitaries and on National Day or New Years. Taiwanese identity was
almost a non-issue for Ma, regardless of the primary audience.
Figure 3: Proportion of Categories in Speeches
Source: Authors own compilation.
00
01
02
03
04
05
Chinese identity Taiwanese identity Economy
Democracy Sovereignty
Ma Ying-jeous Presidential Discourse 55
Figure 4: Proportion of Speech Categories to Each Audience
Source: Authors own compilation.
In order to examine in more detail the substantial variation in Mas
speeches to different audiences, we employed a logit model to predict
mentions of each of the five content categories in Mas speeches. Each
of these models, shown in Table 3, calculated the logged likelihood of
mentioning each category for each of the audiences, compared to a con-
stant likelihood for domestic non-political audiences (the majority audi-
ence type, as we showed earlier). The main point of this analysis is to
determine which of the audiences significantly affect(s) the share of each
of these topics in his speeches, and in what way(s). The findings show
that Mas speeches were 5.5 to 8.3 times less likely to include references
to Chinese identity when they were given before audiences comprised of
diplomatic allies or before domestic political and economic audiences, in
contrast to domestic non-political audiences. Speeches delivered to the
foreign media and on National Day and New Years Day were 3 to 3.5
times less likely to include such messages, although this difference was
only borderline statistically significant. Mas decision to include the issue
of Taiwanese identity also seems a function of the primary audience.
0
002
004
006
008
01
012
fmed formal dias ndny econ fdig Total domp domn mili alli
Taiwanese identity Economy Chinese identity Sovereignty Democracy
56 Jonathan Sullivan and Eliyahu V. Sapir
Table 3: Logit Model to Predict Mentions of Each of the Five Content
Categories in Mas Speeches

Chinese identity Taiwanese identity
Audience B S.E. O.R. B S.E. O.R.
alli -2.13** 0.67 0.12 -2.47*** 0.62 0.08
dias 1.39 0.88 4.00 -1.88** 0.68 0.15
domp -1.99** 0.66 0.14 -2.40*** 0.60 0.09
econ -1.69* 0.66 0.18 -2.35*** 0.61 0.10
fdig -0.66 0.66 0.52 -2.20*** 0.60 0.11
fmed -1.29+ 0.66 0.28 -2.05** 0.61 0.13
formal 1.59 1.22 4.91 0.11 0.79 1.12
mili 0.90 1.24 2.46 -1.76+ 0.91 0.17
ndny -1.18+ 0.70 0.31 -0.92 0.64 0.40
Constant 1.30* 0.65 3.67 0.92 0.59 2.50
Pseudo R
2
0.133 0.053

Economy Democracy
Audience B S.E. B S.E. B S.E.
alli -1.08 0.67 0.34 -1.50* 0.61 0.22
dias 0.01 0.74 1.01 -1.48* 0.67 0.23
domp -1.11+ 0.66 0.33 -1.84** 0.60 0.16
econ -1.01 0.66 0.37 -1.21* 0.60 0.30
fdig 2.05** 0.71 7.74 -2.06** 0.60 0.13
fmed -0.84 0.66 0.43 -1.50* 0.61 0.22
formal 0.02 0.86 1.02 0.41 0.82 1.50
mili -0.89 0.92 0.41 0.47 0.99 1.60
ndny -0.76 0.70 0.47 -1.40* 0.64 0.25
Constant 1.30* 0.65 3.67 0.92 0.59 2.50
Pseudo R
2
0.175 0.046
Ma Ying-jeous Presidential Discourse 57

Sovereignty
Audience B S.E. O.R.
alli -1.15 1.05 0.32
dias -1.50 1.09 0.22
domp -3.26** 1.04 0.04
econ -3.13** 1.05 0.04
fdig -3.10** 1.04 0.05
fmed -2.34* 1.05 0.10
formal -1.79 1.15 0.17
mili -2.57* 1.22 0.08
ndny -2.86** 1.07 0.06
Constant 2.57* 1.04 13.00
Pseudo R
2
0.129
Note: * p < 0.001; ** p < 0.01; *** p < 0.05; + p < 0.1; ++ Reference category: Domestic
non-political (domn).
Source: Authors own compilation.
Compared with the odds for the reference category (domestic non-
political audiences) and formal and military audiences as well as speeches
delivered on National Day or New Years, speeches to all other audiences
were significantly less likely to contain this type of message. A similar
trend was observed regarding the issue of democracy, with the sole ex-
ceptions of National Day and New Years speeches being significantly
less likely to include this topic. Sovereignty was less likely to be included
in formal speeches and speeches delivered to all audiences except diplo-
matic allies and diaspora groups. Regarding the economy, the findings
suggest that other than foreign dignitaries (who were almost eight times
as likely to receive such messages in Mas speeches to them), no other
audiences were significantly more or less likely to, compared to the refer-
ence category of domestic non-political audiences. As discussed earlier,
Ma frequently addressed this topic in his speeches. This finding is the
result of that fact and suggests that the audience was not a significant
factor in Mas decision to include this topic in his speeches.
58 Jonathan Sullivan and Eliyahu V. Sapir
The final stage of our analysis was to determine any interactions be-
tween the timing of the speeches and the identity of the receiving audi-
ences. To do this, we employed a Chi-squared Automatic Interaction
Detector (CHAID) analysis to test the mean differences in mentions of
the various categories. We repeated the analysis for each content catego-
ry and found that timeaudience nodes for Chinese identity, democracy
and the economy had significantly different means. Figures 5, 6, and 7
show the results of this analysis for these three discourse categories. The
figures reveal significant differences in the salience of these issues ac-
cording to when Ma addressed certain audiences, and reflect exogenous
events and changes in Mas relationship with different groups. This is an
important finding, as it alerts analysts to the need to account for both
strategic context and timing when making inferences about presidential
speeches.
Figure 5 reveals that, on average, the topic of Chinese identity ac-
counted for 2.9 per cent of the content of Mas speeches. The first step
in the analysis produced four significantly different audience nodes. The
first included formal and National Day and New Years speeches, as well
as those delivered to foreign media and diaspora audiences. Compared to
the baseline mean, the 90 speeches in this node had a very high average
of 16 per cent, which was stable over time. The average fit of the other
nodes and the group of economic audiences was similar to the baseline,
but the time dimension added in the second step revealed that this aver-
age was not stable over time for economic audiences. Since the begin-
ning of his term, Ma has gradually increased the share of the topic of
Chinese identity in his speeches, and between December 2009 and Au-
gust 2010 it went up to 7.2 per cent. The average dropped back to 2.6
per cent after August 2010. The rise and fall in Chinese-identity language
coincides with Mas campaign for ECFA, which was signed on 29 June
2010. During the preceding period, in which Ma participated in a tele-
vised debate with the DPP, the economic dimensions of ECFA were
accompanied by appeals to Taiwanese citizens Chinese cultural identifi-
cations. The mobilization of this frame suggests that Ma was aware that
ECFA was not, or was not seen to be, a purely economic arrangement.
Moving to the topic of democracy, shown in Figure 6, we found a
similar interaction effect. While the baseline average was 1.7 per cent,
there were three discrete nodes with significantly different averages. The
first node was identical to the one found in the analysis of Chinese iden-
tity. These audiences average share of democracy in Mas speeches was
Ma Ying-jeous Presidential Discourse 59
over twice the amount for the baseline, while the average for the econo-
mists node was about one-third of the baseline average. The third node,
consisting of diplomatic allies, diaspora groups, foreign dignitaries, do-
mestic political groups and military personnel, heard about democracy
1.5 per cent of the time, but this share changed over time. At the begin-
ning of Mas tenure, the average share of this topic in Mas speeches was
1.9 per cent, and between April and late November 2010 it dropped to
0.7 per cent. The average share has increased since, and between De-
cember 2010 and the end of the period under study it was 1.2 per cent.
Finally, the economy displayed similar interactive trends. While the
baseline average was 5.3 per cent, a slightly different division into audi-
ence nodes yielded interesting results. Four audiences were part of the
first node (formal, foreign media, diaspora groups and National Day
speeches). When Ma addressed these audiences, on average over 15 per
cent of his speeches was dedicated to the issue of economy. The other
two nodes had a significantly lower average, but for one of them, com-
prising foreign dignitaries, diaspora groups and domestic political audi-
ences, an interaction with the time dimension was evident. At the begin-
ning of his term, the average share of mentions of the economy to these
audiences was 3.2 per cent. Between December 2009 and August 2010 it
increased twofold, only to go back to 3.4 per cent, on average, until the
end of the period. We interpret this finding as evidence of Mas selective
emphasis in mobilizing support for ECFA. To economic and domestic
political audiences, the major frame employed in discourse about ECFA
was economic, whereas other audiences received a large dose of Chinese-
identity language.

60 Jonathan Sullivan and Eliyahu V. Sapir
Figure 5: CHAID Analysis of Chinese Identity in Speeches to Different
Audiences




Source: Authors own compilation.
Node 0
Mean 2.881
Std. Dev. 7.657
n 2,162
% 100.0
Node 1
Mean 15.022
Std. Dev. 24.374
n 90
% 4.2
Node 2
Mean 1.475
Std. Dev. 3.875
n 1,064
% 49.2
Node 3
Mean 2.744
Std. Dev. 6.315
n 626
% 29.0
Node 4
Mean 3.927
Std. Dev. 6.246
n 382
% 17.7
Node 5
Mean 2.748
Std. Dev. 4.487
n 111
% 5.1
Node 6
Mean 4.466
Std. Dev. 5.648
n 58
% 2.7
Node 7
Mean 7.207
Std. Dev. 10.114
n 82
% 3.8
Node 8
Mean 2.634
Std. Dev. 3.129
n 131
% 6.1
Audience
Adj. P-value=0.000, F=119.700; df1=3, df2=2,158
formal; fmed; dias; ndny alli; domn fdig; domp; mili econ
< = 17.04.09 17.04.0913.12.09 13.12.0915.08.10 > 15.08.10
Time
Adj. P-value=0.000, F=11.799; df1=3, df2=378
Chinese Identity
Ma Ying-jeous Presidential Discourse 61
Figure 6: CHAID Analysis of Democracy in Speeches to Different Audiences




Source: Authors own compilation.
Node 0
Mean 1.667
Std. Dev. 5.427
n 2,162
% 100.0
Node 1
Mean 3.648
Std. Dev. 8.369
n 338
% 15.6
Node 2
Mean 1.515
Std. Dev. 5.080
n 1,442
% 66.7
Node 3
Mean 0.490
Std. Dev. 1.597
n 382
% 17.7
Node 4
Mean 1.892
Std. Dev. 6.110
n 879
% 40.7
Node 5
Mean 0.675
Std. Dev. 1.941
n 271
% 12.5
Node 6
Mean 1.158
Std. Dev. 3.238
n 292
% 13.5
Audience
Adj. P-value=0.000, F=33.022; df1=2, df2=2,159
formal; fmed; domp; ndny alli; fdig; dias; domn; mili econ
< = 19.04.10 19.04.1030.11.10 > 30.11.10
Time
Adj. P-value=0.038, F=6.900; df1=2, df2=1,439
Democracy
62 Jonathan Sullivan and Eliyahu V. Sapir
Figure 7: CHAID Analysis of the Economy in Speeches to Different
Audiences




Source: Authors own compilation.
Conclusion
Given the relatively stable preferences of the PRC and US, the contin-
gencies of electoral competition, and the support for divergent positions
across competing parties and within the electorate in Taiwan, ROC Pres-
ident Ma has emerged as the most volatile actor in cross-Strait relations.
Node 0
Mean 5.341
Std. Dev. 9.612
n 2,162
% 100.0
Node 1
Mean 15.144
Std. Dev. 14.451
n 425
% 19.7
Node 2
Mean 2.387
Std. Dev. 4.081
n 1,132
% 52.4
Node 3
Mean 3.982
Std. Dev. 8.181
n 605
% 28.0
Node 4
Mean 3.237
Std. Dev. 5.913
n 342
% 15.8
Node 5
Mean 6.721
Std. Dev. 14.062
n 122
% 5.6
Node 6
Mean 3.418
Std. Dev. 4.949
n 141
% 6.5
Audience
Adj. P-value=0.000, F=378.567; df1=2, df2=2,159
formal; fmed; econ; ndny alli; domn; mili fdig; dias; domp
< = 13.12.09 13.12.0915.08.10 > 15.08.10
Time
Adj. P-value=0.006, F=8.814; df1=2, df2=602
Economy
Ma Ying-jeous Presidential Discourse 63
Despite the smooth progress and goodwill generated during Mas first
term, as long as underlying political obstacles and cleavages remain, the
ROC president (regardless of his or her identity) is likely to continue to
be a focal point. Developing our understanding of presidential discourse
is an important undertaking now and in the future. The analysis reported
in this article contributes to this understanding by revealing both the
temporal and strategic variation in Ma Ying-jeous public pronounce-
ments. This article provides strong evidence that the strategic context of
primary audiences should be included in analyses of the ROC presidents
speeches. It also reports a methodological advance that facilitates identi-
fying interactions between strategic context and a time dimension. Em-
ploying CHAID models enables analysts to pinpoint changes in rhetori-
cal emphases at different points in time in speeches to different groups.
This enabled us to identify, for example, the increase in Chinese-identity
language to certain groups, but not others, during Mas mobilization of
support for ECFA. We suggest that the methods employed in this analy-
sis hold promise for further systematic analyses of the discourse of ROC
presidents, past and future. Although both Ma and Chen Shui-bian em-
phasized the economy more than any other topic, the similarities in the
two presidents discourse ends there. The democracy and Taiwanese-
identity topics, which were major components of Chens discourse, were
downplayed during Mas first term. In the first decade of the 2000s, Tai-
wanese identity was a if not the major component of political com-
munications in Taiwan (Fell 2005; Sullivan 2010; Sullivan and Lowe
2010), but under Ma this has been replaced by discourse on Chinese
identity. Chen Shui-bian barely touched on sovereignty issues (Sullivan
and Lowe 2010), while Ma placed much greater emphasis on this topic.
Furthermore, our findings demonstrate that ROC sovereignty is very
tightly linked to discourse on Chinese identity: Indeed, the convergence
of sovereignty and identity represents one of the major features of Mas
first-term discourse. This convergence is reminiscent of much earlier
KMT positions, and contrary to the partys seeming embrace of Taiwan-
ese identity, particularly in elections during the last decade. However, the
number of citizens with Taiwan-centred self-identities is stable if not
increasing under Ma, and whether this emphasis continues in his second
term remains to be seen.


64 Jonathan Sullivan and Eliyahu V. Sapir
References
Austen-Smith, David (1990), Information Transmission in Debate, in:
American Journal of Political Science, 34, 1, 124152.
Benoit, Kenneth, Michael Laver, Christine Arnold, Paul Pennings, and
Madeleine Hosli (2005), Measuring National Delegate Positions at
the Convention on the Future of Europe Using Computerized
Word-scoring, in: European Union Politics, 6, 3, 291313.
Benoit, Kenneth, Michael Laver, and Sasha Mikhaylov (2009), Treating
Words as Data with Error: Uncertainty in Text Statements of Policy
Positions, in: American Journal of Political Science, 53, 2, 495513.
Bosco, Joseph (1992), Taiwan Factions: Guanxi, Patronage and the State
in Local Politics, in: Ethnology, 31, 2, 157193.
Budge, Ian, David Robertson, and Derek Hearl (eds) (1987), Ideology,
Strategy and Party Change: Spatial Analyses of Post-War Election Pro-
grammes in 19 Democracies, Cambridge: Cambridge University Press.
Cabestan, Jean-Pierre (2005), Specificities and Limits of Taiwanese Na-
tionalism, in: China Perspectives, 62, 37, online: <http://chinaperspec
tives.revues.org/2863> (28 August 2012).
Chen, Shui-bian (2002), ___jj_,|]_
_____ (Zongtong yi shixun zhibo fangshi yu shijie
Taiwan tongxiang lianhehui di ershijiu jie nianhui zhong zhici, The President
Addresses the 29
th
Annual Meeting of the World Association of Formosans via
Live Video-link), online: <www.president.gov.tw/Default.aspx?tabid
=131&itemid=1311&rmid=514&sd=2002/08/01&ed=2002/08/05>
(28 August 2012).
Chen, Yu-wen Julie (2011), Quantitative Content Analysis of Chinese
Texts? A Methodological Note, in: Journal of Chinese Political Science,
16, 4, 431443.
Chu, Yun-han (2004), Taiwans National Identity Politics and the Pro-
spect of Cross-Strait Relations, in: Asian Survey, 44, 4, 484512.
Clark, Cal (2004), The Paradox of the National Identity Issue in Chen
Shui-bians 2004 Presidential Campaign: Base Constituencies vs. the
Moderate Middle, in: Issues and Studies, 41, 1, 5386.
Copper, John F. (2009), The Devolution of Taiwans Democracy During
the Chen Shui-bian Era, in: Journal of Contemporary China, 18, 60,
463478.
Copper, John F. (1997), The Taiwan Political Miracle: Essays on Political De-
velopment, Elections and Foreign Relations, Lanham, MD: University
Press of America.
Ma Ying-jeous Presidential Discourse 65
Corcuff, Stphane (2004), The Supporters of Unification and the Tai-
wanisation Movement, in: China Perspectives, 53, 4966.
Corcuff, Stphane (2002), The Symbolic Dimension of Democratization
and the Transition of National Identity under Lee Teng-hui, in: St-
phane Corcuff (ed.), Memories of the Future: National Identity Issues and
the Search for a New Taiwan, Armonk, NY: M. E. Sharpe, 73101.
Deutsche Welle (1999), Interview with Lee Teng-hui, July, online: <www.
taiwandc.org/nws-9926.htm> (28 August 2012).
Dittmer, Lowell (2005), Taiwans Aim-inhibited Quest for Identity and
the China Factor, in: Journal of Asian and African Studies, 40, 1/2,
7190.
Dittmer, Lowell (2004), Taiwan and the Issue of National Identity, in:
Asian Survey, 44, 4, 475483.
Fell, Dafydd (2005), Party Politics in Taiwan, London: Routledge.
Hassid, Jonathan (2012), Safety Valve or Pressure Cooker? Blogs in Chi-
nese Political Life, in: Journal of Communication, 62, 212230.
Hsiau, A-Chin (2000), Contemporary Taiwanese Cultural Nationalism, Lon-
don: Routledge.
Hsieh, John Fuh-Sheng (2004), National Identity and Taiwans Mainland
China Policy, in: Journal of Contemporary China, 13, 40, 479490.
Hughes, Christopher (2009), New Trends in Taiwans China Policy, in:
The International Spectator, 44, 2, 5974.
Herrick, James A. (2008), The History and Theory of Rhetoric: An Introduction,
Old Tappan, NJ: Pearson Educational.
Jacobs, Bruce (2012), Democratizing Taiwan, Leiden and Boston: Brill.
Jacobs, Bruce (2008), Taiwan: Changes and Challenges, in: Australian
Journal of International Affairs, 62, 4, 460477.
King, Gary, and Will Lowe (2003), An Automated Information Extrac-
tion Tool for International Conflict Data with Performance as
Good as Human Coders: A Rare Events Evaluation Design, in: In-
ternational Organization, 57, 4, 617642.
King, Gary, Jennifer Pan, and Margaret Roberts (2012), How Censorship in
China Allows Government Criticism but Silences Collective Expression, Har-
vard University Working Paper, online: <http://gking.harvard.edu/
publications/how-censorship-china-allows-government-criticism-sil
ences-collective-expression/> (28 August 2012).
Krippendorf, Klaus (2004), Content Analysis: An Introduction to Its Methodol-
ogy, Thousand Oaks: Sage.
66 Jonathan Sullivan and Eliyahu V. Sapir
Krippendorf, Klaus, and Mary Bock (eds) (2009), The Content Analysis
Reader, Thousand Oaks: Sage.
Lams, Lutgard, and Li-wen Liao (2011), Tracing Taiwanization Pro-
cesses in Taiwanese Presidential Statements in Times of Cross-Strait
Rapprochement, in: Journal of Current Chinese Affairs, 40, 1, 6398.
Laver, Michael, Ken Benoit, and John Garry (2003), Extracting Policy
Positions from Political Texts Using Words as Data, in: American
Political Science Review, 97, 2, 311331.
Lin, Jih-wen (2005), Uncovering the Informal Dimensions of Taiwans Cross-Strait
Policy-making, paper presented at 34th Sino-American Conference,
Charlottesville, VA: University of Virginia.
Lin, Jih-wen, and Yun-han Chu (2001), Political Development in 20th
Century Taiwan: State Building, Regime Transformation and the
Construction of National Identity, in: The China Quarterly, 165,
102129.
Lin, Tse-min, Yun-han Chu, and Melvin Hinich (1996), Conflict Dis-
placement and Regime Transition in Taiwan: A Spatial Analysis, in:
World Politics, 48, 4, 453481.
Lynch, Daniel, (2004), Taiwans Self-conscious Nation-building Project,
in: Asian Survey, 44, 4, 513533.
Mattlin, Mikael (2011), Politicized Society: The Long Shadow of Taiwans One
Party Legacy, Copenhagen: NIAS.
Mattlin, Mikael (2004), Referendum as a Form of Zaoshi, in: Issues and
Studies, 40, 2, 155185.
Meyer-Sahling, Jan-Hinrik, and Klaus Goetz (2009), The EU Timescape:
From Notion to Research Agenda, in: Journal of European Public Policy,
16, 2, 325336.
Muyard, Frank (2010), Mid-term Analysis of the Ma Ying-jeou Admin-
istration, in: China Perspectives, 3, 521.
Neuendorf, Kimberley (2002), The Content Analysis Guidebook, Thousand
Oaks, CA: Sage.
Paolino, Philip, and James Meernik (eds) (2008), Democratization in Tai-
wan: Challenges in Transformation, Burlington, VT: Ashgate.
Popping, Roel (2000), Computer-Assisted Text Analysis, Thousand Oaks:
Sage.
Putnam, Robert (1988), Diplomacy and Domestic Politics: The Logic of
Two-level Games, in: International Organization, 42, 3, 427460.
Ma Ying-jeous Presidential Discourse 67
Rich, Tim (2009), Can Democratic Consolidation Make a Country Less
Secure? The Case of Taiwan, in: Journal of Contemporary Eastern Asia,
8, 1, 114.
Ross, Robert (2006), Explaining Taiwans Revisionist Diplomacy, in:
Journal of Contemporary China, 15, 48, 443458.
Roy, Denny (2003), Taiwan: A Political History, Ithaca, NY: Cornell Uni-
versity Press.
Schelling, Thomas (1969), Communication, Bargaining and Negotiation,
in: Arms Control and International Security, 1, 6372.
Schubert, Gunter (2004), Taiwans Political Parties and National Identity:
The Rise of an Overarching Consensus, in: Asian Survey, 44, 4,
534554.
Slapin, Jonathan, and Sven-Oliver Proksch (2008), A Scaling Model for
Estimating Time Series Policy Positions from Texts, in: American
Journal of Political Science, 52, 3, 705722.
Sobel, Joel (2007), Signalling Games, working paper, UCSD, online:
<http://econ.ucsd.edu/~jsobel/Papers/signalencyclopedia.pdf>
(20 June 2012).
Sullivan, Jonathan (2010), Negativity and Information in Campaign Advertising,
Thesis, Nottingham: University of Nottingham, online: <etheses.
nottingham.ac.uk/1138/1/Sullivan.pdf> (20 August 2012).
Sullivan, Jonathan, and Will Lowe (2010), Chen Shui-bian: On Inde-
pendence, in: The China Quarterly, 203, 619638.
Sullivan, Jonathan, and Bettina Renz (2010), Chinese Migration: Still the
Major Focus of Russian Far East/ Chinese Northeast Relations?, in:
The Pacific Review, 23, 2, 261285.
Sullivan, Jonathan, and Eliyahu V. Sapir (2012), Nasty or Nice? Explain-
ing Positive and Negative Campaigning in Taiwan, in: The China
Journal, 67, 149168.
Sullivan, Jonathan, Yitan Li, Patrick James, and A. Cooper Drury (2011),
An Exchange on Diversionary Dragons, or Talking Tough in Tai-
pei, in: Journal of East Asian Studies, 11, 1, 137152.
Tien, Hung-mao (ed.) (1996), Taiwans Electoral Politics and Democratic Tran-
sition, Armonk, NY: M. E. Sharpe.
Tien, Hung-mao, and Yun-han Chu (1996), Building Democracy in Tai-
wan, in: The China Quarterly, 148, 11411170.
Wachman, Alan (2002), The China-Taiwan Relationship: A Cold War of
Words, in: Orbis, 45, 4, 695711.
68 Jonathan Sullivan and Eliyahu V. Sapir
Wade, Robert (1990), Governing the Market: Economic Theory and the Role of
Government in East Asian Industrialization, Princeton, NJ: Princeton
University Press.
Wang, T. Y., and I-Chou Liu (2004), Contending Identities in Taiwan:
Implications for Cross-Strait Relations, in: Asian Survey, 44, 4,
568590.
Wang, T. Y., Wei-chin Lee, and Ching-hsin Yu (2011), Taiwans Expan-
sion of International Space: Opportunities and Challenges, in: Jour-
nal of Contemporary China, 20, 69, 249267.
Wu, Chung-li (2008), Ethnicity, Empowerment and Political Trust: The
2005 Local Elections in Taiwan, in: Issues and Studies, 44, 1, 105132.
Wu, Yu-shan (2005), Taiwans Domestic Politics and Cross-Strait Rela-
tions, in: The China Journal, 53, 3560.
Wu, Yu-shan (2004), Taiwanese Nationalism and Its Implications: Test-
ing the Worst Case Scenario, in: Asian Survey, 44, 4, 614625.
Zhang, Baohui (2011), Taiwans New Grand Strategy, in: Journal of Con-
temporary China, 20, 69, 269285.
Journal of Current Chinese Affairs 3/2012: 1

Contents


Introduction
Gunter SCHUBERT
Contemporary Taiwan Studies in Europe:
More Institutionalized, More Vital

3
Research Articles
Cal CLARK and Alexander C. TAN
Political Polarization in Taiwan:
A Growing Challenge to Catch-all Parties? 7
Jonathan SULLIVAN and Eliyahu V. SAPIR
Ma Ying-jeous Presidential Discourse 33
Christian GBEL
The Impact of Electoral System Reform on Taiwans Local
Factions 69
WANG Hung-jen
Liberalist Variation in Taiwan:
Four Democratization Orientations 93
Stefan FLEISCHAUER
Cross-Strait Relations and the Way Forward:
Observations from a European Integration Perspective 117
Analysis
Gunter SCHUBERT
No Winds of Change: Taiwans 2012 National Elections and
the Post-election Fallout 143
Contributors 163