Lost in Translation:

A Parallel Corpus-based Study of South Korean Government Translation

By Jinsil Choi (Keimyung University, South Korea)


This article introduces a systemic way to analyse translation products, based on a personally built sentence level tagged corpus, and it describes frequent translation change patterns, both qualitatively and quantitatively, with a case of spokesperson’s press briefings of the Ministry of Foreign Affairs, South Korea. Drawing on institutional translation as self-translation and Fairclough’s critical discourse analysis, this article explores to what extent translation changes such as additions and omissions impinge on the original messages and the way changes reify and reinforce the institutional ideology. The corpus consists of approximately 200,000 tokens of 92 press briefings of Korean source texts and English translations, delivered in 2012. This study reveals that ‘battles’ (Fairclough 2010: 424) between a journalist and a spokesperson and journalists’ negative comments about the South Korean government have frequently been left out of the translations. It argues that these changes are the informed decisions, indicative of the institutional ideology.

Keywords: institutional translation, Korea, parallel corpus, press briefing

©inTRAlinea & Jinsil Choi (2020).
"Lost in Translation: A Parallel Corpus-based Study of South Korean Government Translation", inTRAlinea Vol. 22.

This article can be freely reproduced under Creative Commons License.
Stable URL: https://www.intralinea.org/archive/article/2496

1. Introduction

On 16 August 2011, certain translation errors in the Free Trade Agreement (FTA) between South Korea and the US became a delicate political issue in South Korea. The agreement was first written in English and then translated into Korean, with each version having an equal legal effect. The South Korean government had announced in June 2011 that 250 Korean items had to be corrected. Then 야당공동정책협의회 [the Policy Meeting of the Opposition Parties] announced at a press conference, also on 16 August, that the ratification motion still contained 225 translation errors and deletions (Kwon 2011). The amendment issue was drawn to the attention of the court at the end of the same year, after 민주사회를위한변호사모임 [the Lawyers for a Democratic Society] had sued the government for not publishing the information. The court ruled that the information should have been made public, saying that publication was in the public interest and that the Ministry of Foreign Affairs (MOFA) had announced the amendments of the FTA between South Korea and the European Union (EU) previously (Na 2011). On 25 May, 2012, the government finally announced the amendments to 296 items, including translation errors, additions and omissions (Kim J 2012). 

This situation led to attention being drawn to translation practices and products, as evidenced in a Policy Discussion in April 2015, held at the Korean National Assembly, which proposed the establishment of an independent government translation organisation (Choi 2018)[1]. Scholars, such as Lee S (2014) and Cheong and Lim (2014), among many others, suggested models for Korean government translation in regard to the management of terminologies and practices. However, the extent to which changes in translation, which cannot be described merely as stylistic, impinge on the original message and the way translation changes reify and reinforce institutional ideology, have largely been ignored. Because translation reflects both the political and diplomatic voice of an institution, it requires considered choices to be made not only regarding the selection of the texts and the language or languages they are translated into, but their content and the translators and editors who have been chosen for the task. In such institutional settings, translation is de facto self-translation, with the institution being the author-translator. In this context, as Schäffner, Tcaciuc and Tesseur (2014: 495) put it, “the standardised ‘voice’ of the institution is the one to be heard” and the choice of content, both retained and omitted, reflect the institutional ideology (Gagnon 2006).

Previous research on institutional translation has largely been carried out with a qualitative focus. In Europe and Canada, for example, some text-centred analyses of national and supra-national institutional translations have been undertaken, but most of these have focused on their qualitative aspects (Mossop 2006; Gagnon 2006; Koskinen 2008; Schäffner and Bassnett 2010). This is attributed to the nature of a close comparative examination of both the source text (ST) and the target text (TT); however, qualitative research alone cannot capture a significant amount of cross-linguistic and translational variation, which is why qualitative methods should be complemented by quantitative methods to allow for a more holistic view (Malamatidou 2018: 9). In addition, translation changes may involve drastic changes in “content, quantity, focus and layout” (Schäffner and Bassnett 2010: 19). Schäffner and Bassnett (2010: 19–20) offered a case in point by identifying drastic differences in the newspapers Le Figaro, Der Spiegel and Times Online, regarding an interview with Vladimir Putin, in which drastic omissions and rearrangements of information at sentence level had been made in translation. Using a qualitative approach, they analysed these differences manually. Such drastic translation changes, however, cannot be examined quantitatively, or filtered automatically, unless a corpus is specially designed to do so.

Against this backdrop, this article introduces a preliminary parallel corpus-based methodology aimed particularly at investigating frequent sentence level translation changes in Korean government text productions, both qualitatively and quantitatively, by suggesting a way to suit the textual analyses of drastic translation changes. By using the proposed parallel corpus, the kind of which has “so many potential uses and applications […] in the field of translation” for the study on translation strategies (Mikhailov and Cooper 2016: 1), this article explores both the extent to which translation changes influence the original messages, and how an institutional ideology and translation policy are reflected in these changes. The data for this study, incorporating about 200,000 running words of 92 briefings of Korean STs and English TTs[2], delivered in 2012, the last year of Lee Myung-bak’s presidency, include press briefings (PBs) by a spokesperson of the MOFA. The last year of presidency has often been considered to be a “lame duck” syndrome, featuring a weakened leadership and declining public support (Lee 2015: 117). Such a perception may influence many aspects of a spokesperson’s briefings as they reflect frequent battles between participants; consequently, the briefings delivered in this period were chosen specially for this study.

This research contributes to Corpus-based Translation Studies, by introducing a method of filtering sentential additions and omissions that will allow for both quantitative and qualitative analyses of translation products. This case study is also a meaningful contribution to an understanding of institutional translation, which is still, as Schäffner, Tcaciuc and Tesseur (2014: 494) put it, “rather underexplored”, requiring “detailed case studies of different institutional contexts” (Koskinen 2011: 60), since different institutions in different countries may operate differently.                                                       

In what follows, therefore, I first discuss, as a theoretical framework, institutional translation as self-translation and Fairclough’s critical discourse analysis (CDA). The data and methodology are introduced, which is followed by an analysis and a discussion. Finally I present a conclusion.

2. Theoretical framework

2.1. Institutional translation as self-translation

Considering the activity of translation as a social institution, the most abstract and broadest definition would be the tautology that “all translation is institutional”. However, for my research purpose, a narrower definition of institution is needed. As Gagnon (2010: 254) points out, there are cases in which institutions are also producers of the original texts, “giving them even tighter control of what and how texts are translated”. In this case, Gagnon uses the term, “self-translation” (ibid.).

The term, self-translation, was originally used in situations where the author of the original translates his or her own book into another language (Sorvari 2018). Likewise, when the institution produces both STs and TTs, it is a case of self-translation because the author of the originals and the translations is identical, the institution. According to Koller (1979/1992: 197), the difference between a general translation and self-translation or “autotranslation” can be related to the different degree of authority. Whilst an “ordinary” translator may find it difficult or may “hesitate” to change the original content, “the author-translator will feel justified in introducing changes into the text” (Koller 1979/1992: 197, cited from Montini 2010: 306). In the same vein, Perry (1981: 181) notes that the author-translator “can allow himself bold shifts from the source text which, had it been done by another translator, probably would not have passed as an adequate translation” (cited from Wanner 2018: 122, emphasis mine).

As Koskinen (2008: 21) points out, although translations are hardly ever performed outside any institutional settings, the degree of control exercised by institutions differs. Compared to the degree of regulation in the translations of official government documents, it is likely that the degree of regulation of literary translation, for example, is very different. This may largely result from the significance of the content to be translated. For the case of translations of official government documents, for example, what is to be translated has a considerable importance in itself strategically and diplomatically (Gagnon 2006; Mingxing 2012). Furthermore, without exception, the government institutions are always producers or initiators of the originals or translations while this is not the case for literary translations. When producers of the originals and distributors of the translated texts are the same or in the same institution, or when the initiator of a translation is within the originals’ institution, the degree of mediation in translation by the people concerned with the originals and of communication between producers of the STs and those of the TTs will be different from the opposite case in which none of the people concerned with the ST production is involved in any part of the TT production. In the former situations, “the voice that is to be heard is that of the translating institution” (Koskinen 2008: 22). In light of this, if changes are made in translations, especially when the producer of the originals and the translations is the same, the changes are “autotranslation” (Koskinen 2011: 57) by a given institution. As in the case of self-translation or autotranslation in literary genre, the institutional self-translation embeds the intention of the original, such as regimes, policies, and norms of the institution. In this setting, if changes are made, these are de facto reflections of the intention of the author-translator, the institution. 

2.2. Fairclough’s critical discourse analysis

For the analysis of translation changes in an institutional setting, this study adopts Fairclough’s CDA, which, as he states (2010: 131), is “suitable for” researching changes in discursive practices. Hatim and Mason (1997: 143) also state that a CDA approach is helpful to analyse “the way ideology shapes discourse and the way discourse practices help to maintain, reinforce or challenge ideologies” when combined with a corpus based approach. Shortcomings of CDA such as the selection of short texts, or text fragments (Stubbs 1997; Widdowson 1998) and its lack of robust methodology can be complemented by using corpus, or corpus supporting tools. Consequently, many studies on discourse and translation, such as Baker et al. (2008) and Kim (2017), have combined CDA and corpus based approaches in order “to criticise, connections between properties of texts and social processes and relations (ideologies, power relations)” to date (Fairclough 2010: 131). In Translation Studies, a CDA approach is particularly useful for investigating the impact of translation (Kim 2017: 35), for instance what is chosen for translation, who are involved in it, what their relationship to the material is, and what kind of discourse is intended for the target culture.

Based on Fairclough’s approaches, this study analyses how spokespersons and journalists employ certain discursive patterns to reinforce, or challenge, issues or institutional ideologies, how these aspects appear in translation as changes (Fairclough 2010), how participants negotiate their positions in order to control or manipulate the institutional identity and ideology, and how their positions are established by socially and institutionally situated translators. The study also demonstrates how language is used to control, or manipulate, institutional ideology through repeated discursive patterns, and conversely, how repeatedly omitted patterns in translation may have cumulatively reified institutional ideology; also, how such translation omissions can either manipulate or change the impact of translation, by what might have been written in translation (Mossop 2010: 95).

3. Data and Methodology

3.1. Data

In order to develop a corpus, the data for this study were chosen particularly in accordance with the following criteria: (i) to ease the corpus building process and copyright acquisition, the data available on institution’s websites were chosen, (ii) the translation direction from Korean into English was chosen because of its dominantly high ratio, 89.3 per cent of total Korean to X language combinations (Lee et al. 2001: 67–8), and (iii) the texts and their translations had to be of a genre that is regularly used in a particular context.

The MOFA delivers a variety of genres of news, from news of the Ministry to PBs[3]. Of the various PBs, its spokesperson’s is the most regularly delivered and translated; as such, it represents the institution. Usually it is delivered three to four times a week, which amounts to 63.3 per cent of all PBs produced by the MOFA; it also shows the highest translation rate at 98.1 per cent. A briefing is generally composed of two sections, the announcements of a speaker and the subsequent ‘questions and answers’ session with the press.

Several factors may inherently influence changes in PB translations: (i) the medium –while the original is oral, the translation is written, (ii) spontaneity – a PB is largely spontaneous, except for the spokesperson’s introductory statement, and (iii) the institution’s translation purpose, which is for reference[4]. While (i) and (ii) may involve stylistic changes in translation, such as the correction of false starts or slips of the tongue, which are typical features of oral discourse (Schäffner and Bassnett 2010), an alleged translation purpose may imply, or rationalise, errors or drastic changes in content and focus. This purpose of translation for reference could also be understood as a preference of productivity to “intrinsic quality of the translation output” (Svoboda 2018: 22). Considering that a PB translation mostly appears on the same day of the briefing, management may favour productivity to quality of translation. A weight on productivity by a managerial level is also identified in studies on institutional translation such as Svoboda’s research of the Czech Republic government institutions. Therefore, the spokesperson’s briefing translation receives due attention in this study, in order to investigate drastic changes, such as sentential additions and omissions in Korean government translation.

Most of the translations of general documents within the MOFA are done by in-house professional translators of a given department, although it is to a certain extent different depending on matters and documents. Four stages are involved in the translation process: drafting, revision, confirmation, and release. From the production of the original documents to the release of the translations, all processes of press briefings are controlled by in-house employees, Korean native speakers. Thus, it is a case of self-translation. Both translations and revisions are done by Koreans, most of whom had completed four years of regular education in one of the major English-speaking countries, and who are competent in the both languages. The translation process is hierarchical and linear. Not only are the drafting and revision stages of translation but also the people concerned in each stage of the originals and the translations clearly distinguished and separated, and translators are anonymous, according to my inquiry on 17 December 2013. Such translation process is certainly different from European Commission (Koskinen 2008) and the Language Services Division of the German Foreign Office (Schäffner, Tcaciuc and Tesseur 2014), in which cross-checks and communications between writers of the originals/clients/requesters and translators are active, and “translators are very much ‘visible’” (Schäffner, Tcaciuc and Tesseur 2014: 497).

3.2. Methodology

The process of corpus building consisted of three main stages, (i) data collection, (ii) categorisation and computerisation, and (iii) corpus construction (see Table 1).

Table 1. The process of corpus building

The decision whether to incorporate text samples or whole texts, and corpus size were important as they are in most studies employing electronic corpora (Mikhailov and Cooper 2016: 2). It was decided to use whole texts for this study, because the intention is not to describe general tendencies in terms of mean values of a certain linguistic item in several text samples, but to examine changes in translations, for which the scrutiny of the whole body of texts is pivotal.

The originals and their translations were collected from the institution’s websites. All raw data were converted to XML files and then sentence split and mark-up followed. The sentence alignment for the corpus was performed manually, because although it would have been possible to use an automatic aligner, such as the Translation Corpus Aligner 2 as in the Norwegian Spanish Parallel Corpus (Hareide and Hofland 2012: 90–4) or Hunalign[5], the pre-requisite process of listing anchor words requires considerable time and effort, because, in this study, each type of data exhibits different word usages. Hence, to make automatic alignment possible would be a big job, requiring the creation of a bilingual dictionary for word alignments and sets of rules and categories, as Lee (2010: 115) also points out. This is probably why no studies, in the Korean context, have suggested a systemic and automatic means of word alignment for corpus building, with the exception of Korean-Chinese word-pair extraction by No (2008).

The final stage consisted of annotation, the development of a supporting tool, and building the corpus engine. In order to categorise all data and make them sortable, it was necessary to attach unique titles and ids. Specifications of annotations are illustrated in Figure 1 and an annotation example of sentential additions and omissions are illustrated in Figure 2.

Figure 1. An example of annotations

Figure 2. An annotation example of sentential omissions

Figure 1 shows an example of a PB text, which is divided into two broad categories, according to who is speaking: a spokesperson or a deputy spokesperson. From January to December 2012, two spokespersons (Cho Byung-jae until 31 July and Cho Tai-young afterwards) and one deputy spokesperson (Han Hye-jin) delivered PBs. Along with document information, title information was added in order to make the title visible in the corpus engine. Figure 2 shows an annotation of sentential omissions in translations. When there is no corresponding TT, one sentence in the ST composed one pair; in the same way, sentential additions were annotated.

Next, a corpus supporting tool was developed, which makes it possible to open, save and revise raw and annotated data, and to control and supervise data input processes consistently. For the purposes of this study, I decided to design both a free search engine, allowing users to input their own search words, and a guided search engine, listing all categories to enable users to select one of them, so that filtering of data according to genre, subgenre, document, and mapping information can be made, which will be necessary when analysing translation changes. Using Microsoft Visual Studio 2010 allowed for the interface of the corpus to be designed with choices of corpus (genre, subgenre, etc.) and a free search function. When a PB genre is selected, the corpus produces responses, such as those shown in Figure 3. Figure 3 shows a corpus view of sentential omissions when the “Only Korean” function was activated. A user can save the result as a txt file, as in Figure 4.

Figure 3. A snap shot of the corpus

Figure 4. A snap shot of omissions in the SB translation

Figure 4 shows the sentential omissions in the spokesperson’s briefing (SB) translation and it offers subgenre, document id, pair id, and language information[6].

The corpus also produces a basic statistics such as the number of words[7] and pairs by genre, subgenre, document, and speaker, as in Figure 5.

Figure 5. A snap shot of corpus statistics

4. Analysis

For analyses, the corpus mapping function is used to identify sentential additions and omissions in translations. From the point where such changes occur, stretches of text in context are scrutinised for a qualitative analysis. In addition, a keyword analysis is carried out by using WordSmith Tool 7.0 in order to examine the importance of repeatedly omitted or added discursive patterns in translation and the extent to which such cumulative changes reflect institutional aims, objectives, and ideologies.

4.1. Sentential omissions and additions in translations

The corpus exhibited the frequency and ratio of sentential omissions and additions in briefing translations, as shown in Table 2.

Table 2. The frequency and ratio of sentential omissions and additions in translations

Of SB, 11.18 per cent of pairs were omitted while 0.46 per cent was added, and the deputy spokesperson’s briefings (DSB) showed a rate of 9.42 per cent was omitted, and 0.46 per cent was added. In terms of tokens, 7.34 per cent of Korean words were left out of the translation, while 0.13 per cent of English words were added to the SB translation; the DSB showed 5.30 per cent omissions and 0.10 per cent additions. These results indicate that the anonymous translators or revisers in the MOFA exercise greater discretion with omissions than with additions. As evidenced in Baker (1997: 121), when translating someone in a position of high authority, a translator’s discretion may be severely restricted when it comes to an addition of what has not been said and done.

From a total of 92 briefings, 21 SBs and three DSBs contained sentential additions in the translations. While most additions consisted of greetings and opening and closing statements, which could be interpreted as changes of text formatting, some repetitive content tended to be added in the translation, particularly in the context of South Korean-Japanese relations and Chinese government related matters, as appear in Figure 6 below.

Figure 6. A snap shot of additions in the translation

The highlighted parts of document id 42 and 54 concern South Korean-Japanese relations while part of 51 concerns the detention of a Korean human rights activist, Kim Young-hwan, in China (for detailed information regarding document id, see Appendix). Given that the numbers of additions in the translation were limited (see Table 2) and that not all inter-national relations were chosen to be repeatedly articulated and added in the translation, this result arguably represents the MOFA’s interests, aims and objectives, which largely takes care of the South Korean government’s relations with Japan and China.

However, as Koskinen (2008: 136) notes, “the more numerous omissions are more dubious.”All briefings contained sentential omissions. While some omissions included changes for text formatting, such as the omission of “모두발언” [announcement] and confirmations of previous questions, many omitted briefings tended to involve North Korea’s nuclear experiment and missile issues (document id 7, 9, 13, 15, 16, 52, 79), China’s detention of Kim Young-hwan (47, 48, 51), Japanese government related issues (6, 18, 42, 48, 54), and MOFA and Korean government related issues (24, 26, 27, 28, 29, 35, 40). This phenomenon warrants further explanation.

The contextual analysis revealed that frequently and drastically omissions related to, in  Fairclough’s (2010: 424) words, “battles” between spokespersons and journalists, in relation to the content and timings of announcements[8]. These kinds of battles tended to appear in diplomatically sensitive issues relating to North Korea and China.

Extract 1

Source: 24 May 2012

<질문> 지난주와 달라진 게 없는 거군요? 그 사이에./

<답변> 예, 현재까지로는 특별히 보고 드릴만큼, 설명드릴만큼의 진전은 없습니다./

그리고 제가 외교부 외신담당을 통해서 북한이 기술적으로 핵실험 준비가 됐다는 보도에 대해서 확인을 부탁드렸을 때 아시는 바가 없다고 말씀하셨습니다./

오늘은 확인 가능한 내용이 있는지 여쭤보고 싶습니다. /


[journalist] <Q> Nothing is different since last week? Between?/

[spokesperson]<A> No, not much has been progressed for further explanation. /


[journalist] And when I asked you through the department of foreign press a confirmation about a report that North Korea is technically ready to conduct a nuclear experiment, you said that you know nothing. / I would like to ask if you can confirm any today. /

Extract 2

Source: 8 May 2012

어디에서 그런 것을 들으셨습니까? /출처를 확인해주십시오. /

<질문> 취재원 보호차원에서 말씀드리기 어렵습니다.


[spokesperson] Where did you hear that? /Please tell me the source. /

[journalist] <Q> It is difficult to tell you in order to protect the source./

Extract 3

Source: 13 December 2012

(동아일보 이정은 기자) 제가 이 질문을 드리는 이유는, 사실은 지금까지 과거에 문제가 되고, 핵실험을 하고 이랬을 때마다 나왔던 자료로 알고 있는데, 요청을 하면 말씀주시겠다고 하셨지만, 벌써 며칠 전부터 요청이 들어가 있던 사안입니다./ 그럼에도 불구하고, 가장 기본적인 이런 팩트나 수치조차 우리가 지금 확인을 못하고 있는 상황이라면, 기사를 쓰는데 좀 지장이 있다고 생각합니다./ 이것은 외교부 차원에서 시정을 해주셨으면 좋겠습니다./

<답변> 무슨 말씀인지 알겠습니다. /

<질문>(채널A 김정안 기자) 그러면 그 시점에 대한, 소위 말하는 우리들이 쓰는 마감시간에 대한 어느 정도의 염두는 두고 계시는지요./ 아니면 지금 협의가 언제까지 진행됐는지 조차도 모르는 상황이신 것인지요?/

<답변> 마감시간은 오늘 마감시간을 말씀하시는 것입니까?/


<Q> (Lee Jung-eun from Dongailbo)[9] The reason I asked this question is that [the material] was a problem in the past and has been publicised whenever [North Korea] has a nuclear experiment. [You] said that [you] would provide us [the material] and [[I]] requested it several days ago./ Nevertheless, [[I]] am still unable to check basic facts and numbers and [[I]] think that impedes [my] writing. /[[I]] want it to be rectified by the Ministry./

<A> [spokesperson] I understand what you are talking about./

<Q> (Kim Jung-eun from ChannelA) Then, about the time, do you consider the so-called deadline for our writing?/ Or, [you] don’t know until when the consultation was proceeded?/

<A> [spokesperson] By deadline, do you mean today’s deadline?/

In Extract 1, 2 and 3, the struggles between participants as to “what it is possible to say and do, what the various identities, social relations and forms of authority are, and the possibility and legitimacy of drawing conclusions or making decisions” (Fairclough 2010: 424) have been totally omitted from the translation, together with when it is possible to say and do. Extract 1 involves a question about any progress in consultation with the Chinese government, regarding the detention of Kim Young-hwan in China. Extract 1 also concerns the MOFA’s position on the National Defense Ministry’s statement that “North Korea appears technically ready to conduct a nuclear test at any time, citing commercial satellite images of the country’s nuclear sites” (official translation), and urges a confirmation by the MOFA regarding this matter. It can be inferred from the original briefing that there is disharmony between the Defense Ministry and the MOFA in regard to North Korea’s readiness for a nuclear test. The journalist appears to have asked the MOFA, before the Defense Ministry’s announcement, to check the matter. In fact, the press had started to report signs of North Korea’s imminent nuclear experiment a month before this briefing. On April 22, Kim S (2012), especially, reproached the President Office’s “미지근한” [tepid] response to these imminent signs not to announce potent sanctions, but seemingly to simply watch the situation, and to condone the Defense Ministry’s silence, as opposed to the active responses of the US and China. Considering the charged atmosphere then, the journalist may have urged a confirmation, or, at least, an announcement about the MOFA’s position, in addition to the Defense Ministry’s. In Extract 2, the struggle between the participants can be identified as the source of alleged military discussions with China, in addition to the MOFA’s announcement that “the discussions with Japan have been ongoing” (official translation). In the previous stretch of text, the journalist mentioned that (s)he was informed that the military discussion with China was also ongoing, and asked this to the spokesperson. Similarly, Extract 3 features the battle between the spokesperson and the journalists in regard to the timing of the announcement about North Korea’s nuclear experiments that appears, up to that point, to have been delayed. All these tensions and battles in the ST were entirely omitted in the TT.

This practice, however, was not limited to several lines, but was found across the entire corpus, often involving drastic changes in content i.e.the deletion of dozens of sentences in translation. Similar battles were identified in domestic issues, too, which mostly involve journalists’ negative comments about the government.The 10 January briefing, for example, featured, in 42 sentences, the battle over how the Ministry publicised the information, including a journalist’s critical comment, “브리핑을 하지 마시고 관보를 통해서 알려주세요” [[You] don’t do [press] briefings and make an announcement through the official gazette], and the spokesperson’s response, “저는 그런 코멘트는 도저히 수용을 못하겠습니다” [I cannot accept such comments], but this battle was completely omitted in the TT; hence, the target reader has no idea about the intense tension of the original exchange. Another drastic omission occurred in the 19 January briefing translation. In the ST, the announcement of the inspection result of CNK International, a diamond mining company, had involved a 31 sentence intensive battle. Kim Eun-suk, ambassador for Energy and Resources in the MOFA, was being prosecuted for selecting a “부실한” [weak] CNK International to receive governmental aid as part of “energy cooperation diplomacy” (Heo 2018). The head of CNK, Oh Deok-gyun, had been accused of allegedly “leading government officials to release false information about a diamond mine in Cameroon for which CNK International won the mining rights” and the acquisition overpriced CNK stocks (Lee H 2014). Although the issue was heavily debated in subsequent briefings on 26 and 31 January, none had been translated. Also in the 31 January briefing translation, the prosecution’s “전례없는” [unprecedented] search of the MOFA, and a journalist’s urge for the issue to be announced were entirely omitted. A policy of an institutionalised and collective voice results in such repetitive translation strategies, thereby ignoring both the battles between spokespersons and journalists, and any negative remarks about the MOFA and government.

Regarding Japan, what was salient was that the debates over individual opinions or their appropriateness tended to be omitted in the translations. In these cases, an emphasis on “the collective, institutionalised voice” featured not only in the TT but also in the ST.

Extract 4

Source: 29 May 2012

<답변> 제 개인의견을 물으시는 것입니까?/

<질문> 아닙니다./ 한국 국민들의 입장으로 그렇게 요청해도 된다고 생각하십니까?/                                                   


[spokesperson] <A> Are you asking my personal opinion?/

[journalist]<Q> No. / Do you think that the Korea nationals can demand that?/

Extract 5

Source : 29 March 2012

<질문> 대한민국의 주적이 어디입니까?/

<답변> 지금 무엇에 관한 질문을 하시는 것입니까?/

<질문> 북한이 우리의 주적입니까?/ 일본이 주적입니까?

<답변> 질문을...지난번에도 제가 말씀드렸습니다만, 이 질문을 한 분이 계속 이어가시고, 지정도 받지 않고 독점하시는 부분에 대해서는.../

<질문> 워낙 중요한 대외관계이고, 독도문제이기 때문에 계속 물어보는 것인데,/

<답변> 중요하고, 안 하고에 대한 것은 기자분이 일방적으로 판단하시는 것은 바람직하지 않다고 생각합니다./ 일단 조금 이따가 질문을 제가 나중에 다시 할 수 있는 기회를 드릴 테니까./

<질문> 한 마디만 얘기할게요./ 외교부가 지금 하고 있는 조치가 일본의 독도에 대한 침략 야욕에 대해서 제대로 정확히 대응하고 있느냐고 묻고 싶은 것이죠./


[journalist] <Q> Which country is South Korea’s main opponent?/

[spokesperson] <A> What is the question now?/

[journalist] <Q> Is North Korea our main opponent?/ Is Japan the main opponent?/

[spokesperson] <A> The question...as I mentioned before, about this matter that one person asks questions continuously and exclusively without interruption .../

[journalist] <Q> That is because [South Korea and Japan] is a very important international relation and this matter is related to Dokdo./

[spokesperson] <A> [[I]] think that it is not appropriate that a journalist decides whether the matter is important or not. / I will give you an opportunity to ask again. /

[journalist] <Q> Let me make one comment./ I wanted to ask if the Ministry is accurately taking action against Japan’s ambition to invade Dokdo. /

In Extract 4, the spokesperson and the journalist argue about what it is possible to report about a Korean Supreme Court judgement. A week earlier, the Court had included the statement that the right to demand compensation for illegal actions against humanity and the Japanese colonial administration was not included in the South Korea-Japan agreement on compensation signed in 1965 (Lee 2012). As a consequence, individuals who had suffered during the colonisation could now demand compensation in addition to the compensation received as a result of the 1965 agreement[10]. In the published translation, the definition used by the spokesperson regarding what “개인적으로” [individually] actually meant, and the journalist’s comments were deleted; thus the struggle between what the spokesperson was prepared to say in the name of institution or as his personal opinion, and what the journalist would like to have heard, was entirely omitted in the translation. Extract 5 features the similar battle, however in it, the spokesperson evades answering ‘which country South Korea’s main opponent is’ by changing the subject and criticising the journalist’s continued questioning. The issue here concerns Dokdo, described in Japanese textbooks, and also in a Japanese diplomatic white paper, as Japanese territory. The territorial dispute regarding Dokdo, known as Takeshima in Japan, but presently administered by South Korea, has been a long-simmering source of tension between the two countries. This is why discourses about Japanese related issues are particularly steeped in diplomatic overtones and are likely to spark debates regarding in whose names particular words are spoken. Choe (2012) states that this long history of tension “is rooted in good part in anger over Japan’s brutal dominance” of Korea some decades ago; however, Japan is now an ally of South Korea. In this context, the MOFA repeatedly asserts its aim is to develop its relations with Japan, while “squarely” facing up to the history (see Figure 6). The institutional objectives of MOFA, therefore, appear to be to distance the government in a wider sense from the critical perspectives of journalists, as in “일본의 독도에 대한 침략 야욕” [Japan’s ambition to invade Dokdo], which are omitted from the English translation. This is indeed indicative of the spokesperson’s briefing translation as “self-translation”, which encapsulates the intention and ideology of both the ST and the institution.

Bearing the foregoing in mind, there follows a keyword analysis in order to examine the extent to which frequently omitted patterns are considered key compared to a reference corpus, the entire corpus, since the numerous omissions appear to be dubious. Then, the parallel corpus will be used to scrutinise keywords in context together with their collocates.

4.2. A keyword analysis of omissions in translation

A keyword analysis is useful for the examination of significantly more or less frequent words in a research corpus than in a reference corpus (Mikhailov and Cooper 2016: 133). Table 3 shows 12 keywords of omissions in the translation.

Table 3. Keywords of omissions in the briefing translation

While the recurrent omission of 모두발언 and demonstratives could be construed as stylistic changes, Table 3 reveals a particularly interesting result that ‘we’ and the ‘inspection result’ of ‘the Board of Audit and Inspection of Korea’ appear significantly more frequently than they do in the entire corpus, but are repeatedly omitted in the translation. Considering that in Korean, subjects are frequently omitted, the level of importance of 우리들이 [we] is high. The negative keywords in Table 3, such as 북한 [North Korea], 중국 [China], and 정부 [government], however, do not mean that the related issues are less important in the research corpus. The result could be attributed to its small size, compared to the reference corpus, as the related issues significantly appear in the previously discussed sentential omission analysis. This supports the need to integrate quantitative analyses with qualitative ones to allow for a more holistic view, as it does in this study.

The search word, 우리들이, in the corpus when the ‘only Korean’ function is activated, produces its collocates, as in Figure 7, such as 조치 [measures], 소통 [communication], 이 문제 [this issue], 연구하고 [studying], 시행하고 [reinforcing], 관보[official gazette], 조사 [investigation], CNK 보도자료 [CNK press release], 감사원 [the Board of Audit and Inspection of Korea], 대응[responding], and 감사 [inspection], all of which indicate the MOFA’s positions, measures, and policies regarding certain diplomatic and political issues.

Figure 7. A snap shot of 우리들이 in the omission

The collocate of 우리들이 with the other keywords, 감사원 and 감사결과 [inspection result] interestingly reveals the MOFA’s reiterated emphasis on its position regarding the CNK International stock-rigging scandal that was harshly criticised by journalists, whereas it was most frequently omitted in the translation, thus undermining the markedly critical original voice.

4.3. Discussion

The analyses revealed that diplomatic and political issues concerning North Korea, Japan, and China, together with domestic issues, particularly containing negative comments about the MOFA, were frequently omitted in the translations, while reiterations tended to be added to the translation involving Japan and China. In particular, the battles between spokespersons and journalists regarding the content, timing and method of announcement were frequently omitted, when sensitive or provocative issues with North Korea, China, and the MOFA were the subejct. Particularly striking, in the case of Japan, was the MOFA’s recurrent discursive pattern of distancing itself from journalistic criticism by changing the subject or criticising the journalists by contesting in whose name certain words had been spoken. Also, all such tensions were invariably omitted in translations. All these results are indicative of the MOFA’s ideology that appreciate highly of those relations, and informed decisions in order to promote and maintain its positive image by emphasising significant changes in translation, thereby facilitating the receiving audience to have, in Mingxing’s words, “positive identification thereof” (2012: 2).

The MOFA’s “institutionalised and collective voice” policy results in repetitive translation strategies that are revealed in sentential translation changes and keyword analyses, which suggest that its institutional voice is of greater importance. In this respect, Koskinen’s and Schäffner et al.’s considerations regarding an ‘institutional voice’ apply to the MOFA. Such recurrent translation practices of ignoring negative comments about the institution, diplomatically sensitive tensions, and critical perspectives, as advanced by journalists, support Koskinen’s argument that “all institutions constrain and regulate behaviour” (2008:18). Such an argument also includes local translation strategies.

However, the data in this study do not seem to be strictly standardised in terms of terminologies and stylistics (Choi 2018), in contrast to the cases of the European Central Bank (Schäffner, Tcaciuc and Tesseur 2014: 16) and the Czech translation department, which, according to Svoboda (2018: 28), uses a machine translation service. The MOFA responded to my enquiry on 17 December in 2013 that “the translators do not use any computer-assisted tools for translation”. These examples show the importance of examining in detail case studies set in different institutions.

5. Conclusion

This article introduced an efficient parallel corpus methodology as a means of examining both qualitative and quantitative changes in the translations of spokesperson’s briefings. Both the analysis and result confirmed the MOFA’s translations to be self-translations, in that the MOFA deliberately reconstructed and framed them in accordance with its institutional aims and ideologies, through the omissions of negative comments about the government and diplomatically sensitive and provocative issues pertaining to Japan, China, and North Korea, thereby downgrading the marked voice in the English translation.

Due to space restrictions, however, it has not been possible to incorporate an analysis of textual variations of different speakers in the ST and TT. Nevertheless, this aspect is worth examining in order to reveal to what extent individual speakers’ speech styles and idiolects contribute to the collective voice of an institution, how differently each speaker corresponds to journalists’ sensitive and provocative questions, and how differently, or similarly, the struggles between the participants are presented in translation as regards to the changes made. Because the current corpus incorporated briefings of only three speakers, for such an in-depth investigation, much more data would be required.

Another possible research area includes examining whose voices, such as journalists and their affiliations, are most frequently omitted in the translations, in order to shed important light on the MOFA’s institutional ideology. In this regard, the current study could be extended by incorporating data from different administrations that have differing political affiliations – such as by comparing President Park Geun-hye’s right-wing administration (2013– 17) with President Moon Jae-in’s left-wing administration (2017 to present), in the corpus, because many aspects of governmental changes, from administrative policy to translation policy, will be observable (Choi 2014).

Given the MOFA’s selective choices of texts and content for translations, the change patterns identified in this study reflect what it considers to be expedient to cover. In summary, this study supports previous studies by revealing that institutional translations inculcate the intentions and ideologies of those institutions.

Appendix: The list of data in the parallel corpus[11]

A1. The list of data and monthly keywords


This article has been developed based on the author’s unpublished PhD thesis, ‘A Corpus Based Genre Analysis of Institutional Translation in Korea’, completed at the University of Leicester in 2014. The author would like to thank anonymous reviewers who provided insightful ideas for this current article.


Baker, Mona (1997) “Non-cognitive Constraints and Interpreter Strategies in Political Interviews” in Translating Sensitive Texts: Linguistic Aspects, Karl Simms (ed), Amsterdam, Rodopi: 111–30.

Baker, Paul, Costas Gabrielatos, Majid Khosravinik, Tony McEnery and Ruth Wodak (2008) “A Usefull Methodological Synergy? Combining Critical Discourse Analysis and Corpus Linguistics to Examine Discourses of Refugees and Asylum Seekers in the UK Press”, Discourse Society 19, 273–306.

Cheong, Ho-jeong and Hyun-kyung Lim (2014) Public-sector Translation. Seoul: Hankookmunhwasa.

Choe, Sang-hun (2012) “Fight Over Rocky Islets Opens Old Wounds Between South Korea and Japan”, The New York Times, 4 October.

URL : https://www.nytimes.com/2012/10/05/world/asia/south-korea-and-japan-fight- over-rocky-islets.html (accessed 23 January 2019).

Choi, Jinsil (2014) “Changes in Press Briefings of the Ministry of Foreign Affairs and Trade in South Korea”, Journal of North-East Asian Cultures 40: 47–62.

_______(2018) “Introducing a Korean-English Parallel Corpus for the Standardisation of South Korean Government Documents: A Terminology Focus”, in Business and Institutional Translation: New Insights and Reflections, Eric Porier and Daniel Gallego (eds), Newcastle upon Tyne, Cambridge Scholars Publishing: 33–44.

Fairclough, Norman (2010) Critical Discourse Analysis: The Critical Study of Language, 2nd Edition. London, Routledge.

Gagnon, Chantal (2006) “Language Plurality as Power Struggle, or: Translating Politics in Canada”, Target 18, no.1: 69–90.

_______(2010) “Political translation”, in Handbook of Translation Studies Volume 1, Yves Gambier and Luc van Doorslaer (eds), Amsterdam, John Benjamnis: 252–56.

Hatim, Basil and Ian Mason (1997) The Translator as Communicator, London, Routledge.

Hareide, Lidun and Knut Hofland (2012) “Compiling a Norwegian-Spanish Parallel Corpus”, in Quantitative Methods in Corpus-based Translation Studies, Michael P. Oakes and Meng Ji (eds), Amsterdam, John Benjamins: 75–113.

Heo, Baek-yoon (2018) “CNK 주가조작 김은석 前 대사, 파기환송심 패소 강등 처분 정당”[CNK stock-rigging scandal Former Embassador Kim Eun-suk lost a case] 서울신문 [Seoulsinmun], 12 November.

URL:http://www.seoul.co.kr/news/newsView.php?id=20181112500138&wlog_tag3=naver (accessed 25 January 2019)

Kim, Ji-hwan (2012) “외교부의 뒤늦은 한미 FTA 번역오류 정오표 공개”[the MOFAT’s belated announcement of the ROK-US FTA translation error amendments], 경향신문 [Kyunghyangshinmun], 25 May.

URL:http://news.khan.co.kr/kh_news/khan_art_view.html?artid=201205251456331&code=920100 (accessed 8 May 2014).

Kim, Kyung-hye (2017) “Newsweek Discourses on China and their Korean Translations: A Corpus-based Approach”, Discourse, Context and Media 15: 34–44.

Kim, Sung-man (2012) “북한의 3차 핵실험 저지해야” [Must block North Korea’s third nuclear experiment], KONAS, 22 April.

URL: http://www.konas.net/article/article.asp?idx=28395 (accessed 23 May 2014).

Koller, Werner (1979/1992) Einführung in die Übersetzungswissenschaft, the 4th revised edition, Heidelberg, Quelle & Meyer.

Koskinen, Kaisa (2008) Translating Institutions: An Ethnographic Study of EU Translation, Manchester, St. Jerome.

______ (2011) “Institutional Translation”, in Handbook of Translation Studies 2, Yves Gambier and Luc van Doorslaer (eds), Amsterdam, John Benjamins: 54–60.

Kwon, Min-seok (2011) “한미 FTA 협정문 오류 225건 추가 발견” [225 South Korea-US FTA agreement errors additionally identified], YTN, 16 August.

URL:http://news.naver.com/main/read.nhn?mode=LPOD&mid=tvh&oid=052&aid=0000368446 (accessed 8 May 2014).

Lee, Hyung-jeong (2014) “Prosecutors to Resume CNK Stock-rigging Investigation”, The Korea Herald, 12 March. URL: http://khnews.kheraldm.com/view.php?ud=20140313001255&md=20170812134336_BL (accessed 28 January 2019).

Lee, Jung-gyu (2010) “A Sentence Alignment Tool for Building a Parallel Corpus”, Language Facts and Perspectives 25: 115–38.

Lee, Seung-jae (2014) “A Suggestive Model for Public Sector Translation: A Case Study of Foreign Language Translation Center in the Korea Tourism Organization”, Interpretation and Translation 16, no.3: 59–80.

Lee, Ok-geun (2015) “최고의사결정자인 대통령의 레임덕 현상에 관한 연구: 참여정부 노무현 대통령의 리더쉽을 중심으로” [A Study on President’s lame duck syndrome: Focusing on President Roh Moo-hyun’s Leadership], Journal of Governmental Studies 21, no.3: 95–120.

Lee, Seung-jae, Cho-lim Sung, Gye-Yon Lee, Hyang Lee, Young-jim Kim, Hyun-joo Jang, Sang-won Lee and Sang-eun Cho (2001) “The Present Situation of Translation in Korean Public Organizations”, The Journal of Translation 2, no.2: 57–107. 

Lee, Sang-sun (2012) “대법원, 강제징용 일 기업 배상 책임 인정”[The supreme court admits Japanese company’s responsibility to pay compensation]’, YTN, 24 May.

URL:http://search.ytn.co.kr/ytn/view.php?s_mcd=0103&key=201205241640448298&q=%ED%95%9C%EC%9D%BC%EC%B2%AD%EA%B5%AC%EA%B6%8C (accessed 22 May 2014).

Malamatidou, Sofia (2018) Corpus Triangulation: Combining Data and Methods in Corpus-Based Translation Studies, London, Routledge.

Mikhailov, Mikhail and Robert Cooper (2016) Corpus Linguistics for Translation and Contrastive Studies: A Guide for Research, London, Routledge.

Mingxing, Yang (2012) “The Principles and Tactics on Diplomatic Translation: A Chinese Perspective”, Babel 58, no. 2: 1–18.  

Montini, Chiara (2010) “Self-translation”, in Handbook of Translation Studies Volume 1, Yves Gambier and Luc van Doorslaer (eds), Amsterdam, John Benjamins: 306–8.

Mossop, Brian (2006) “From Culture to Business: Federal Government Translation in Canada”, The Translator 12, no.1: 1–27.

______(2010) “Translating What Might Have Been Written”, in Text and Context, Mona Baker, Maeve Olohan and María Calzada Pérez (eds), Manchester, St. Jerome: 95–113.

Na, Hwak-jin (2011) “법원: 한미 FTA 협정문 번역 오류 공개하라” [The court: announce South Korea-US FTA agreement translation errors]. 연합뉴스 [Yonhapnews], 2 December.

URL: [url=http://news.naver.com/main/read.nhn?mode=LSD&mid=sec&sid1=101&oid=001&             ]http://news.naver.com/main/read.nhn?mode=LSD&mid=sec&sid1=101&oid=001&             [/url]; aid=0005400711 (accessed 8 May 2017).

No, Yong-gyun (2008) “Word Pair Extraction from the Chinese and Korean Parallel Corpus: Process, Principle, and Lesson”, Proceedings of the Korean Society for Language and Information Conference, 147–65.

Perry, Menakhem (1981) “Thematic and Structural Shifts in Autotranslations by Bilingual Hebrew-Yiddish Writers: The Case of Mendele Mokher Sforim”, Poetics Today 2, no.4: 181–92.

Schäffner, Christina and Susan Bassnett (2010) “Politics, Media and Translation-Exploring Synergies”, in Political Discourse, Media and Translation, Christina Schäffner and Susan Bassnett (eds), Newcastle upon Tyne, Cambridge Scholars Publishing: 1–29.

_______, Luciana Sabina Tcaciuc and Wine Tesseur (2014) “Translation Practices in Political Institutions: A Comparison of National, Supranational, and Non-governmental Organisations”, Perspectives: Studies in Translation Theory and Practice 22, no.4: 493–510.

Sorvari, Marja (2018) “Altering Language, Transforming Literature: Translingualism and Literary Self-translation in Zinaida Lindén’s Fiction”, Translation Studies 11, no.2: 158–71.

Stubbs, Michael (1997) “Whorf’s Children: Critical Comments on Critical Discourse Analysis”, in Evolving Models of Language, Ann Ryan and Alison Wray (eds), Clevedon, Multilingual Matters: 100–16.

Svoboda, Tomáš (2018) “Institutional Translation in National Contexts: Quality Assurance in Governmental Institutions Across Europe”, in Business and Institutional Translation: New Insights and Reflections, Eric Porier and Daniel Gallego (eds), Newcastle upon Tyne, Cambridge Scholars Publishing: 15–32.

Wanner, Adrian (2018) “The Poetics of Displacement: Self-translation among Contemporary Russian-American Poets”, Translation Studies 11, no.2: 122–38.

Widdowson, Henry G. (1998) “The Theory and Practice of Critical Discourse Analysis”, Applied Linguistics 19, no.1: 136–51.


[1] However, at the time of writing October 2018, no independent government organisation for translation  has been established in South Korea.

[2] The current corpus size in total approximates to 450,000 tokens of three genres of Korean institutional translations, (i) PBs of the spokesperson at the MOFA, (ii) presidential speeches of Lee Myung-bak, and (iii) the web-magazine of the National Museum of Contemporary Art, under the Ministry of Culture, Sports and Tourism. Because of limited space, this article will focus on an analysis of PBs delivered in 2012.

[3] The MOFA produced 668 PBs between January 2008 and June 2012, of which 463 were translated into English; this does not include press releases, press conferences, and speeches.

[4] The focus of translating PBs depends rather on capturing the content for reference purposes, as I was informed when I enquired on 22 May 2012[4]; presumably, therefore, it is acceptable to make rather large-scale alterations.

[5] For Hunalign : http://mokk.bme.hu/resources/hunalign/ (accessed 09 January 2019).

[6] Korean is coded as 0, while English is coded as 1.

[7] The corpus recognises words by blanks.

[8] All Korean and their corresponding English sources were extracted from the MOFA’s website. For Korean PBs: [url=http://www.mofa.go.kr/www/brd/m_4078/list.do]http://www.mofa.go.kr/www/brd/m_4078/list.do[/url] (accessed 27 October 2018)

and English: [url=http://www.mofa.go.kr/eng/brd/m_5679/list.do]http://www.mofa.go.kr/eng/brd/m_5679/list.do[/url] (accessed 27 October 2018).

[9] From 15 November 2012, the MOFA reveals affiliations and names of journalists who ask in the briefings. Dongailbo is one of the right wing news outlets in South Korea, and Channel A belongs to the same company, DongA Media Group.

[10] The Agreement on the Settlement of Problem concerning Property and Claims and the Economic Cooperation between South Korea and Japan in 1965 was an agreement on compensation for the Japanese colonisation. However, the issues of comfort women and forced labor have long been controversial because the agreement was understood not to include the individual human right to demand compensation.

[11] CBJ : Cho Byung-jae, CTY : Cho Tai-young, HHJ : Han Hye-jin

About the author(s)

The author works as an assistant professor at Keimyung University, South Korea. Her research interests include a corpus-based translation discourse analysis, pre-modern Korea in translation, audiovisual translation, and English language teaching with translation. She has published the Korean translations of “Discourse and the Translator” by Basil Hatim and Ian Mason and “Linguistics and the Language of Translation” by Kirsten Malmkjær. Currently, she is working on a research monograph, “Government Translation in South Korea: A Corpus-based Study” (working title), to be published by Routledge in 2020.

Email: [please login or register to view author's email address]

©inTRAlinea & Jinsil Choi (2020).
"Lost in Translation: A Parallel Corpus-based Study of South Korean Government Translation", inTRAlinea Vol. 22.

This article can be freely reproduced under Creative Commons License.
Stable URL: https://www.intralinea.org/archive/article/2496

Go to top of page