Corps de l’article

I. Introduction

A translator draws on all kinds of knowledge while translating, and such knowledge falls broadly under two types: linguistic and extralinguistic[1]. In addition, s/he must be equipped with translation methodology, i.e., knowledge and skills involving problem identification, problem-solving, decision-making, subject research skills, etc. A translator’s competence might also involve self-concept, aptitude and personality, attitudes and affective factors (Laukkanen 1996), translation effort, understanding of translation, etc. Moreover, since these components of translator competence appear to complement each other, a translator who is lacking in one component or another can resort to other component(s) to compensate for that lack.

The overall competence level of a translator correlates with the overall quality of translation. However, it does not mean that each component contributes equally to improving the final output. There might be a hierarchy among them, meaning that some components might be more important than others in determining the quality. Given the fact that most translation work involves more or less specialized texts, it is likely that extralinguistic knowledge, rather than linguistic competence itself, plays a major role in the success of translation. For example, a scientist can translate a scholarly scientific article better than a professional translator who has little or no scientific knowledge. In this case, scientific knowledge, rather than the languages involved, becomes a major factor influencing the product. Lehmann (1986: 140) echoes this point:

One necessary prerequisite for successful comprehension is knowledge. For textual comprehension the type of knowledge available and how it is structured is more important than what language the subject speaks. Therefore, a professor of literature reading a translation of “À la recherche du temps perdu” might understand Proust better than a French wine merchant reading the original text.

Indeed, previous studies found that successful translation is characterized by the extensive use of extralinguistic knowledge together with linguistic knowledge (Tirkkonen-Condit 1988, 1989, 1992; Jääskeläinen 1990 cited from Tirkkonen-Condit 1992). On the other hand, unsuccessful translation tends to be linguistically oriented. Research findings support the predominance of extralinguistic knowledge over linguistic knowledge in the translation process. However, no close examination has been made on this subject.

Questions can arise as to how important extralinguistic knowledge is and how it is used in translation processes, or in what way it is important. To illustrate, a translator cannot translate successfully without understanding a source text (hereafter, ST). And s/he is very likely to come across comprehension problems (of various sorts and in varying degrees) at one point or another. There seem to be two major strategies used in solving comprehension problems: one is inferencing and the other is the use of reference books (Krings 1986: 270). Let’s say that a translator does not know an ST item (lack of linguistic competence). S/He can infer meaning or refer to a dictionary (use of translation methodology). Here, one cannot say for sure whether inferencing as a comprehension strategy is superior to the use of dictionaries. However, studies suggest that excessive and unguided use of dictionaries, especially bilingual dictionaries, is not desirable (Tirkkonen-Condit 1989; Fraser 1993). Which strategy to use, or which to use first, depends largely on whether or not the relevant knowledge is available.

Knowledge used in comprehension can be classified into linguistic knowledge and extralinguistic knowledge. The type of knowledge used seems to determine the level of comprehension that can be reached, which ultimately affects the product quality. Dancette (1997, p. 94) posits three levels at which comprehension operates in translation: linguistic level, textual level, and notional level. His study shows that a translator who can conceptualize ‘contextual’ meaning at the notional level based on linguistic and extralinguistic knowledge can translate more successfully and creatively. On the other hand, a translator who remains at the linguistic and textual levels and fails to fully access the notional level (perhaps, with no extralinguistic knowledge) tends at best to find the ‘literal’ meaning and ends up with less successful translation. This study indicates that extralinguistic knowledge is important in the construction of comprehension because it enables a translator to comprehend a text at deeper, notional levels.[2]

The above discussion suggests that the availability and the use of extralinguistic knowledge is a major factor in determining the quality of the translation product. Specifically, extralinguistic knowledge seems to precede linguistic knowledge in its contribution to translation: it makes it possible for a translator to infer meaning at cognitive levels, leading to in-depth comprehension and thus successful translation. Studies on extralinguistic knowledge are too limited to draw any conclusions, however. Further studies are in order, and the present study attempts to investigate the status and the role of extralinguistic knowledge in translation. Research questions are posed as the following:

  1. How important is extralinguistic knowledge relative to other components or elements of the translator competence?

  2. How is extralinguistic knowledge used or in what way is it important in understanding an ST?

Research focuses on comprehension processes, and research design has been tailored to the questions posed: the translation is from L2 into L1 and the ST was rather difficult, with some specialized terms on economics and a few metaphorical expressions. The research uses three types of data from a translation, questionnaire, and think-aloud study. Translator competence can be measured by testing subjects. Alternatively, information can be gathered through a questionnaire by asking them relevant questions. How extralinguistic knowledge is used in the translation process can be investigated via a think-aloud method. The ultimate goal of the present study is to identify strategies and/or behaviors that are likely to lead to successful translation, and can be recommended for later use. In corollary, it is also to avoid those strategies and/or behaviors that could lead to unsuccessful translation. Thus, it reports on whatever is noteworthy in the two types of translation.

II. Research Design

2.1. Subjects

Participants in the study included five professional translators (the Pro group), eleven translation students (the TS group), and twenty-one language learners (the LL group), all native speakers of Korean. The professional translators had four to thirteen years of translating experience at the time of the task. The translation students were studying translation at the Graduate School of Translation and Interpretation, Hankuk University of Foreign Studies (HUFS) in Korea. (Six of them had some translation experience, ranging from six months to five years, and could be termed semi-professionals.) The language learners were college seniors who had taken the course “Introduction to Interpretation and Translation” and had volunteered for the task.

Among the subjects, only three subjects took part in the think-aloud study, each representing the three groups. (hereafter, termed Pro, TS, LL for the groups they represent) Pro had eleven years of experience, TS had two years, and LL had none. Pro’s English level was both advanced and sophisticated, TS had a fairly good command of English, and LL was far behind the other subjects. Pro specialized in the translation of scholarly papers, advertisements, and culture-related texts. TS had worked as a full-time translator at a financial institute for about a year, and s/he was quite familiar with texts such as the ST used in the study (as will be seen later). Pro had an advantage in relation to translation experience, whereas TS had the upper hand in relation to the subject matter (the ST was about the euro). LL, on the other hand, had much less knowledge in both English and the subject matter.

2.2. Text

The English ST selected was an article from England’s main financial newspaper Financial Times issued on June 3, 1999 (see Appendix 1). The entire text was given to the subjects, but only the shaded part was to be translated. The text, whose topic was the weakness of the euro, was chosen because it posed potential semantic translation problems. The text was fairly difficult, and understanding it required specialized knowledge in economics; in particular, terms such as imported disinflation. A few metaphorical expressions also contributed to increasing the difficulty level of the text: they included concentrate minds, brush aside, power ahead, and spluttering. It was predicted, therefore, that the subjects would face mostly comprehension problems rather than production problems.

2.3. Method

The subjects translated the English text into Korean, their mother tongue. They were allowed to take the text home and use whatever resources were available and there was no time limit for the task. In addition to the final products, other forms of data collected were a questionnaire[3] (see Appendix 2) and information recorded as per the think-aloud method. All the participants completed the questionnaire, while three subjects participated in the think-aloud study. The task descriptions for the translation and a questionnaire were given to the subjects in written form, and instructions for the think-aloud task were given orally.

The questionnaire was used in two ways: first, to measure the subjects’ overall translator competence (see I in Appendix 2) and second, to gather information about their problems and problem-solving processes or strategies (see II and III in Appendix 2). The subjects were instructed to complete the questionnaire after doing the translation. It might be that the amount of time between the translation work and the questionnaire varied considerably from one individual to another. Some might have done it immediately after the translation and others hours or even days after the translation. Consequently, the data collected via questionnaire may be ‘retrospective’ in some cases, while in other cases it may have relied upon the subjects’ relatively long-term recall of the translation process. The think-aloud method was used because it offered a concurrent account of translation processing, and provided a reliable way to investigate extralinguistic knowledge as it was actually being thought through.

III. Results

3.1. Results of the Questionnaire

3.1.1. Group performance

The subjects’ translations have been evaluated separately on a five-point scale by the researcher for 1) the sense conveyed and 2) the form in the target language, but the sense-based evaluation is used here for comparison. The sole use of semantic accuracy is quite appropriate given that the study focuses on comprehension, and difficulties involving comprehension rather than production have been anticipated. (Table 1 provides evaluation criteria.) It turned out that translation error was most often seen in three problematic sentences (second, fifth and seventh sentences).

Table 1

Criteria for Rating Translations on a Five-Point Scale

Criteria for Rating Translations on a Five-Point Scale

-> Voir la liste des tableaux

The average evaluation scores are provided in the column labeled “Sense” in Table 2. No marked difference is evident between the Pro group and the TS group. On the other hand, the LL group shows a much lower average score than the other two groups. There seem to be some individual differences in the Pro group: two subjects received scores of 5, two received 4, and one received 3. Little variability was found among the subjects for the TS group: four received 5 and the rest, 4. Greater individual differences were found among the LL group: two received 5, two received 4 and two received 3. About two-thirds of the LL participants scored 1 or 2.

For the questionnaire, the subjects were asked to rate themselves on their competence (as reflected in the five categories in Question I in Appendix 2). Points have been added up and averaged out by group for each of the five categories. Results are also shown in Table 2. First, in terms of Current Affairs (CA), Subject Matter Knowledge (SM), and Vocabulary (Voc), there was little or no difference between the Pro and TS groups. The LL group, however, obtained much lower average points (lower than the midpoint ‘3’) than the two groups. This indicates that on the whole, the language learner subjects were not as interested in current affairs, were unfamiliar with the subject matter, and had little knowledge of terminology in the article.

Table 2

Evaluation of Translation and Self-Rated Competence of Subjects

Evaluation of Translation and Self-Rated Competence of Subjects

(CA: interest and knowledge in current affairs; SM: knowledge about subject matter; Voc: vocabulary knowledge; Eff: translation effort; UT: understanding of translation)

*The total score is 5 for each category.

-> Voir la liste des tableaux

As regards the other two categories, Translation Effort (Eff) and Understanding of Translation (UT), it is noteworthy that the Pro group rated themselves excellent in UT, scoring 4.6. It might be that they were able to understand the nature of translation as a result of years of experience and research. It can also be noted that the LL group rated themselves high (3.9) for Translation Effort. It might be that they put a lot of effort to the translation work in order to compensate for their overall lack of competence. This contrasts with the performance of the TS group, who didn’t seem to make much effort (as evidenced by their score, 2.5).

When viewed in relation to the final translation product, the results of the questionnaire indicate that deficiencies in knowledge in current affairs, subject matter, and vocabulary led to the lower translation quality. On the other hand, translation effort and understanding of translation did not markedly contribute to overall higher quality of translation. This, however, reflects group performance. It would be difficult to generalize the results across the subjects due to rather high levels of inter-subject variability. Therefore, close examination of individuals is needed here for the sake of higher validity. What follows is a report on a separate examination of those who achieved a perfect translation. What made them so successful is of major interest here.

3.1.2. Individual performance

The two professionals in the Pro group who achieved a score of 5 on their translations thought that they specialized in economics and finance in their translation. In the questionnaire, one of them marked 5 (a perfect score) in four of the five categories (except Translation Effort). A close examination showed that this subject had no difficulty understanding the ST, and there were no problematic segments whatsoever. Perhaps his ease in translating accounts for his perceived lack of effort, as indicated by giving himself a “3” for Translation Effort. The other subject seemed to have made extra effort in translating, even asking an expert about one problematic segment (imported disinflation). This is in sharp contrast with the performance shown by one professional in the same group: this subject marked 3 for Translation Effort and scored 3 (lowest in the group) for his translation. His idea of an experimental (as opposed to real-life) task and tight schedule might have affected his motivation to perform the task. As for those four subjects in the TS group, there was nothing remarkable about them compared to the rest in the same group.

For the two subjects in the LL group who distinguished themselves by scoring 5 on their translations, they didn’t know a number of the terms (as shown by 2 and 3 for Voc) and had little knowledge about the subject matter (as shown by 1 and 2 for SM). However, they exerted a great deal of effort in their translation tasks (they marked 4 and 5 for Translation Effort). They also said that they had read their translation ten times or more and had made corrections (ten for one subject and “countless” corrections for the other subject) before submitting the final products. In total, they spent three and five hours, respectively, doing the task. In addition, a detailed examination of their problem-solving behaviors/strategies showed that they were deeply involved with the task. They consulted every resource available including dictionaries, encyclopedia, news articles (in English and Korean) dealing with the subject matter, parallel Korean texts, acquaintances with expert knowledge, etc. It is quite surprising that they delivered the message accurately in spite of relatively lower English proficiency and little specialized knowledge in the subject matter.

The results above suggest that knowledge about the subject matter (a type of extralinguistic knowledge) and translation effort are the two most important factors that affect the final translation product. To be specific, one can come up with a good translation without much time and effort if s/he is familiar with the subject matter of an ST. Also, one can anticipate a good translation without familiarity with the subject matter if s/he devotes a great deal of time and effort to translating.

3.1.3. Translation problems

In the questionnaire, the subjects were asked to check the problematic segments (see III in Appendix 2).[4] It turned out that the average number of potentially problematic segments was 3.8. 3.1, and 5.1 for the Pro, TS, and LL groups, respectively. This means that the LL group has the most difficulty translating the text, followed by the Pro group and the TS group. The segments found to be problematic across groups included imported disinflation, export its way out of trouble, brush aside, powers ahead, concentrate minds, and should. These problems can be classified into three types: the first concerning specialized knowledge (imported disinflation, export its way out of trouble), the second metaphors (brush aside, powers ahead, concentrate minds), and the third grammar (should). The three groups coincided in the most problematic segment, imported disinflation, which required knowledge of economics.

As problematic segments do not necessarily lead to translation error, translations of the problematic segments were checked for accuracy and adequacy. Most translation errors were detected in the segment export its way out of trouble across groups. It can be noted that the segment which has been identified as the most problematic one (imported disinflation) and the segment which has actually turned out to be the most difficult one (export its way out of trouble) had something in common: for interpretation, they both required some specialized knowledge of economics. The difficulty of the translation seemed to be associated with the inaccessibility of specialized knowledge or background knowledge about the subject matter. This supports the observation made earlier that knowledge about the subject matter is a major element in determining the quality of the product.

The grammatical problem should is ambiguous as it could mean either “had better” (necessity or obligation) or “probably” (certainty). Those who have such grammatical knowledge should have identified the segment as potentially problematic. Only one subject in the Pro group and about half the subjects in each of the other two groups did so. The final translation shows that three in the Pro group, three in the TS group, and all in the LL group got it right by translating it as meaning “necessity”.[5] The ability to identify potential translation problems and the capacity to produce accurate translation should be two separate elements. At any rate, the problem of should cannot be solved within the boundary of the text to be translated. The meaning can only be disambiguated using information from the rest of the text. Clues can be found in these two sentences: It is now time for the politicians, not the bankers … and At today’s summit, EU leaders should…

3.2. Results of the Think-Aloud Study

A comparison of the three subjects has been made in terms of general translation behaviors, problem-solving strategies, and the use of dictionaries. The subjects differed along a few dimensions, which are reported below.

3.2.1. General translation behaviors

The subjects began their tasks somewhat differently. Pro started out by taking into consideration the source and publication date of the ST and the type of the target text. She then attempted to translate the title and to read the text silently. TS said that she would read the whole text first and started reading silently, which took about one minute. LL sat in front of her computer with three sites open on the monitor (an English-Korean dictionary, a Korean-English dictionary, and a search engine) and started translating the title right away (by looking up the word sagging in the dictionary) before reading the text aloud. Pro mentioned while reading the text that there were some parts she could not understand, and after reading she ventured a guess on the main point of the text. TB commented that the text would not be so difficult to translate, since the subject matter was somewhat familiar to her from her previous reading of similar texts. In contrast, LL said that it would be quite difficult to translate the text.

Pro proceeded through the text translating sentence by sentence rather spontaneously except the fifth sentence beginning with “While the U.S…” Since her translation was rather automatic and she did not verbalize her thoughts, not much information could be obtained about the internal processing. For the problematic sentence, however, she spent a fair amount of time with it, trying to extract meaning from the text. She didn’t seem to care much about her target renderings throughout. Either she was preoccupied with the problematic sentence, or she was simply not verbalizing it due to automatic processing.

As for TB, very little was recorded as she did not verbalize much of the time, perhaps as a result of automatic processing. Only a small portion of the translation processes or behaviors manifested themselves in the recording of this subject. The recorded parts allowed the researcher to observe and infer a few things about the process. It seems that the subject verbalized only when she encountered items she could not translate immediately. As she herself had expected at the outset, she didn’t seem to have much difficulty understanding the ST. Her comprehension problems were restricted to a few items. She solved the problems through inferring and reasoning, using prior knowledge and information from the text. She seemed to pay a lot more attention to the way her translation was rendered. She actually said that her translation should flow naturally and that she needed to use “fancy” expressions since the translation was going to be published in a Korean current affairs magazine. In a word, she was conscious about coherence and stylistic and text-type adequacy as was observed in professional translators (Lörscher 1993).

LL actually struggled with the text, as evidenced by the sheer amount of time spent translating and recording (about one hour and forty minutes compared to forty minutes for Pro and thirty minutes for TS). Her performance was a typical example of word-for-word translation, and it was quite evident that translating the ST, with all the difficult terms and expressions, was simply beyond her capability. Since she didn’t seem to understand much of the text, it was understandable that some parts of her translation were incomprehensible and totally unacceptable.

3.2.2. Problem-solving strategies

Two problematic segments have been selected here to compare the ways in which the three subjects tackled the problems. One is the segment imported disinflation, which has been identified as the most difficult to translate by the subjects. The other is the segment should (concentrate minds): about half the subjects identified it as problematic. They have been chosen because they present different kinds of translation problem. (Appendix 3 provides the protocols for the processing of these items by the subjects.)

Imported disinflation

The three subjects produced the following translations for this item. (The back translations are given in the parentheses.). It can be seen that Pro’s translation is similar to the ST segment in form, but it deviates a little from the original sense with its nuance of “the U.S. government taking positive measures”. (This was not counted as a semantic error, however.) On the other hand, TS’s translation is far from the segment on the surface, and yet it delivers the message quite clearly. LL’s translation is incomprehensible.

Pro: 이 들 지역을 활용하는 인플레 해소책

(anti-inflationary measures which use these regions)

TS: 외국으로부터의 값싼 수입품으로 인해 물가상승 압력이 낮아져

(with lower inflationary pressure thanks to cheap imports from foreign countries)

Foreign countries from cheap imported products owing to inflation lower

LL: 국내 시장의 통화팽창 압력을 저지할 수 있는 디스인플레이션을 포함한

(including disinflation which can contain inflationary pressure of the local market)

The subjects differed in some respects. Problem-solving strategies used by Pro and TS consisted of inferencing followed by dictionary search. LL’s strategy, however, was limited to dictionary search. (As will be seen later, Pro and TS exhibited differences in terms of knowledge they used in making inferences.)

Pro didn’t seem to know exactly what the segment meant. She dealt with imported and disinflation separately. She immediately came up with the interpretation 이들 지역에서 수입해온 (back translation: imported from these countries) for imported. Later, she checked this interpretation through parsing as shown in her protocol imported disinflation from A, B, C (see Appendix 3). For disinflation, she seemed to have inferred meaning using linguistic clues: she might have guessed the term as meaning the opposite of inflation by morpheme analysis (dis+inflation). She then attempted to derive meaning in context by extending the scope of analysis to the sentence containing the item, and then the adjoining sentences, and then the rest of the text. After reading the whole text, she said it didn’t help at all. Finally, she referred to a dictionary to check if the meaning inferred was right, and she decided to go ahead with her first interpretation. Here the basis for her inference was mainly her knowledge in English, although she sought clues from the text (albeit, in vain). It can be clearly seen that all along she was not quite sure of the adequacy of her interpretation. Unfortunately, she herself was not satisfied with her translation (as assessed by herself in the questionnaire) after all the lengthy processing (see Appendix 3).

For TB, comprehension of the ST item in question occurred rather easily. She treated the item imported disinflation as one chunk and inferred meaning based on her background knowledge. In a word, she had no difficulty understanding this item. Rather, she was concerned with how it would be rendered most adequately in the target language, Korean. Specifically, she hesitated between the literal target equivalent, 수입된 디스인플레이션, and a simpler explanation of the term. After doing a dictionary search, she decided upon the latter, saying that no equivalent Korean term was shown. This decision was guided by her thinking that the target readers would include the general public, who might not be able to understand the specialized target term 수입 디스인플레이션. It can be seen that, basically, her problem was of production (rather than comprehension) type as was commented by her in the questionnaire. The protocol shows that her comprehension processing was brief and concise (see Appendix 3): she used only her background knowledge related to the problematic item, and her translation was satisfactory (she said so in the questionnaire).

LL first tried to recall disinflation, saying that she had learned something like inflation and deflation before (see Glahn 1980 cited from Krings [1986: 270], for retrieval from learning situations). She then looked it up: first in a bilingual dictionary and a dictionary of economic terms and tried to understand the term (by reading the entries and the term description). Without understanding the item fully and without further attempting to understand it, she adopted the transliteration 디스인플레이션. As for imported, she tried to figure out meaning, giving 의미있는(meaningful) and 포함한(including) as candidate interpretations. She somehow associated it with the word with before the item (‘with imported disinflation) and decided to translate it into 포함한, which of course was an off-target translation. Throughout the process, she exhibited little concern for the exact meaning of the term or perhaps, was simply unable to capture the meaning. Her strategy was restricted to dictionary search and her processing remained at a linguistic level. Furthermore, the scope of her analysis was limited to the sentences containing and surrounding the item. Her final translation didn’t make sense.


The final translations of the segment should (concentrate minds) produced by the three subjects are provided together with the back translations. They all interpreted it as denoting “necessity” as it should be. According to the think-aloud protocols, Pro and LL didn’t recognize should as a problematic item at all. (In the questionnaire, Pro marked should as unproblematic, whereas LL marked it problematic.) Therefore, the analysis is restricted to TS.

Pro: …집중적으로 다루어져야 한다

(…should be dealt with intensively)

TS: … 중심의제로 다루어져야 한다

(…should be dealt with as a main agenda)

LL: … 주의를 기울여야만 하는 문제이다

(…is an issue that should be paid much attention to)

TS immediately identified should as problematic and decided to wait until the meaning clarified itself in context as the text proceeded. After the first translation, she came back to the problem. She seemed to lean toward “certainty” rather than “necessity,” and she provided three target variations. Her final translation of the item, however, was rendered as meaning “urging the politicians to do some action” (necessity). The questionnaire and the telephone conversation with her revealed that she changed her mind after reading the rest of the text closely. Thus, her decision-making was based on information from the text, that is, textual knowledge.

3.2.3. Use of reference material

The three subjects differed in their use of reference material in terms of frequency and nature. Pro used a bilingual dictionary and a monolingual dictionary four times: once for the unknown item splutter at the stage of comprehension and three times for double-checking the items disinflation, powers ahead, export its way out of trouble while monitoring, whose meanings she had inferred first. It should be noted that although she did not know the exact meaning of splutter, she was able to infer meaning correctly from the surrounding context. She didn’t refer to other materials except the dictionaries.

TS used the Internet for reference. First, she looked up the item brush aside in a bilingual dictionary, but did so only after she inferred the meaning by herself. She also searched for imported disinflation on the Internet, which was for an adequate target rendering and not for meaning establishment. Other than these two instances, no other use of resource materials was observed in her protocol. And yet, she said in the questionnaire that she also used the search engine Google on the Internet.

LL relied heavily on dictionaries and other reference materials both for understanding and reformulation. For the most part, she used a bilingual dictionary, and sometimes a monolingual dictionary as well as a dictionary of economic terms. She seemed to resort to dictionaries for almost every unknown lexical item, even looking up words she knew, probably to double-check on meanings retrieved from memory (not ‘inferred’ from context or other ways). Sometimes, she used a dictionary instead of intuition, while monitoring the correctness of target equivalents found. She also consulted parallel Korean texts to find appropriate target renderings. She even asked for help from her father and her boyfriend, who were not experts on the subject matter. LL’s behavior points to a lack of confidence on her part.

In short, the use of reference material by Pro and TS was rather limited. Further, they first inferred meaning before using a dictionary, which was used mainly to double-check the inferred meanings. In contrast, LC used various reference materials excessively for both unknown and known items, and for both understanding and production.

IV. Discussion

The present study addressed two questions regarding the use of extralinguistic knowledge in translation, and the questions are repeated here: (1) How important is extralinguistic knowledge relative to other components or elements of translator competence?; (2) How is extralinguistic knowledge used or in what way is it important in understanding an ST?

With regard to the first question, data analysis revealed that expert or specialized knowledge, which is part of extralinguistic competence, was a more significant predictor of the success of translation compared to linguistic knowledge. Specialized background knowledge about the subject matter of an ST makes the translation process easy, concise, and efficient, leading to a higher quality translation. This is evidenced in the performance of two translators (one was observed in the questionnaire and the other in the think-aloud study), who were well versed in the subject matter in question and produced perfect translations without much difficulty and effort. On the other hand, a lack of this knowledge makes the process difficult, lengthy, and inefficient, and could lead to a lower-quality translation (as was shown by a professional translator in the think-aloud study).

Another important element found to contribute to producing good translation is “translation effort.” It can be recalled that two language learners conveyed the accurate message in their translations despite deficient vocabulary and subject-matter knowledge. This was the result of tremendous effort (including extensive subject research) focused in the right direction, which made up for the overall low competence. The concept of translation effort has not drawn much attention in translation studies. Kim (2001), however, found that a professional translator made great efforts, as exhibited by a variety of strategies and complex cognitive processes, to find the most appropriate translation equivalents. This was in marked contrast to a language learner in the study who had tried limited strategies such as a dictionary search and a structural analysis, or had sometimes even given up translating. Gerloff (1986) also attributed the success of one of his subjects with particular problem passages to his high level of personal involvement with the text (along with others).

It is understandable that the overall level of translator competence is in proportion to the overall level of translation quality. This does not seem to apply in every translation situation, however. Recall that the translations produced by a few professional translators were either not as good or about the same as those done by some language learners. This was a rather unexpected result considering that the former were much better at English, more familiar with the subject matter, and had more field experience as a translator. One possible reason for this is that there is a threshold level of competence in every element required to ensure a good translation. However competent in language and however experienced in translating, one cannot produce a good translation without expending a certain minimal level of translation effort. It is hoped that this finding adds to understanding the nature of translator competence.

In relation to the second research question, three important findings emerged from the think-aloud study. First, the strategy of inferencing was used in addition to the dictionary search in efficient and successful translation. Second, comprehension problems could best be solved by inferencing based on extralinguistic knowledge rather than linguistic knowledge. Third, the type of knowledge most relevant in making inferences varied in accordance with the type of problems.

Inferencing appears to be a major strategy used by competent translators in comprehending a text. In this study, when the professional translator (Pro) and the translation student (TS or the semi-professional translator) encountered a problematic ST item, they first tried to infer meaning instead of and/or before doing anything else. And, this was observed for almost every problematic item for both translators. On the other hand, the language learner (LL) had a strong tendency to use a dictionary whenever possible. She might have had no other choice, since her linguistic and extralinguistic knowledge were insufficient. A word of caution here: it is not certain whether a specific strategy is better, or whether one order of strategy use is superior to others. That is, no causal relationships can be established yet between the strategy use and the product. Rather, knowing when and how to use what strategy in a specific case seems to be a vital skill for the translator.

For the ST item imported disinflation, TS came up with accurate and adequate translation, while Pro’s translation diverged a bit from the item. (LL produced a nonsensical translation.) Pro and TS showed commonality in inferring meaning for this item, but they differed in one important respect: Pro did ‘linguistic’ inferencing and TS did ‘extralinguistic’ inferencing. To be specific, Pro’s inferencing was mainly based on linguistic analysis using her knowledge of the language (English), whereas TS’s inferencing was based on her prior background knowledge of the economic world. Furthermore, Pro was not sure of the inferred meaning and struggled for quite a while to find an adequate interpretation from the text. Unfortunately, her comprehension of the item was incomplete. Unlike Pro, TS grasped the meaning of the term thoroughly as being associated with the concept she is familiar with. She was sure of her interpretation and so proceeded with her processing in a concise and confident manner. This confidence might have allowed her to concentrate more on the target rendering. As could be expected, TB’s target rendering of the item was easy to understand and more to the point. (TS’s choice of the target translation based on the TT recipients shows a macro-level approach involving text awareness (Jonasson 1998) or control by translation-expectation structures (Kiraly 1995).

This finding suggests that background knowledge or extralinguistic knowledge supercedes linguistic knowledge in solving comprehension problems in translation. This is so because it enables a translator not only to do deeper-level comprehension processing but also to do such processing in an efficient, economical, and confident way, leaving much room for other types of processing.

As regards the third finding, the problem of imported disinflation was best solved by TS by using her specialized background knowledge related to it. So, inferencing based on “extratextual” knowledge was more efficient than language-based inferencing for this item. On the other hand, it was “textual” knowledge (i.e., information derived from the text) that was useful in solving the grammatical problem of should. The type of problem at a local level, or the type of text at a global level, seems to correlate with the type of knowledge to be utilized. If a translator has knowledge which is most relevant to solving a specific problem, using that knowledge should be the simplest and most efficient way of solving it (i.e., an optimal solution). In short, the importance of extralinguistic knowledge as an inference-based problem-solving resource in translation cannot be denied. What matters is not the knowledge per se but the relevancy of the knowledge accessed to solve a specific translation problem. It is the knowledge base a translator has and the type of knowledge a translator chooses that determines the quality of translation.

Besides the findings discussed above, one striking difference in the subjects’ behavior should also be mentioned. It concerns the use they made of dictionaries. Pro and TS consulted them only on a few occasions, whereas the language learner used them quite heavily. More importantly, the former used dictionaries “not for establishing meaning but as a stimulus to the process of refining meaning and selecting an appropriate rendering” (Fraser 1996: 72). In contrast, the main purpose for the language learner to use dictionaries was to search for meaning. This finding replicates the previous findings. A new finding of the study is that the translators tend to make inferences using all knowledge available before turning to dictionaries, and they used dictionaries as the last resort or an additional aid. On the contrary, dictionaries were the first and foremost choice for the language learner.

The lack of self-confidence seems to underlie the language learner’s excessive reliance on the bilingual dictionary (for comprehension of the source language) and the Korean parallel texts (for reformulation in the target language). Fraser (1995) points out that self-confidence brings with it the ability to tolerate ambiguity involved in successful translation. Teachers need to help students refrain from using dictionaries too much and move their focus toward developing strategies for extracting and assessing meaning and finding and selecting an appropriate target rendering. They also need to help them build their self-confidence. Confidence can come from years of experience, or it might be a matter of personality. If the latter is the case, it is all the more important that curricula are developed so that every student has one or two specialized fields in which s/he feels most confident. One such curriculum can be found at the École d’Interprètes Internationaux in Belgium, where students learn a range of specialized knowledge from the various university departments (Klein and Lovenberg 1999). In this way, one can kill two birds with one stone – acquisition of extralinguistic, specialized knowledge and self-confidence at the same time.

V. Conclusion

In conclusion, it should be remembered that better language proficiency does not always lead to higher translation quality. Also, more experience, thus, overall higher-level translator competence, does not necessarily correlate with better translation, either. Expert or specialized background knowledge is perhaps by far the most important element in the translator competence, as is indicated in the remark: “[…]the best translations are produced in cooperation between professional translators who are also linguists and experts in the subject matter of the text” (Tirrkonen-Condit 1992: 440). Equally important is deeper involvement with the task or translation effort, which can guarantee you a good (though not always perfect) translation. The results of the study are not yet conclusive due to the small amount of data and the limited scope. Future studies will decide whether they are applicable to other subjects, other texts types, other translation problems, other language directions and combinations. For now, the job of a translator is to maintain an interest and curiosity in things around him/her and make a continuous effort to be equipped with encyclopedic knowledge. Such knowledge will eventually come in handy in one way or another.