Ika Oktavianti / Journal of Economic Education 1 (2) (2012) ELT FORUM 7 (1) (2018) Journal of English Language Teaching http://journal.unnes.ac.id/sju/index.php/elt ANALYSIS OF GOOGLE TRANSLATE’S QUALITY IN EMPLOYING TRANSLATION TECHNIQUES Dewi Kartika, Arif Suryo Priyatmojo  Putri Anggraeni, et al / Journal of English Language Teaching 6 (1) (2017) Dewi Kartika, Arif Suryo Priyatmojo / Journal of English Language Teaching 7 (1) (2018) English Department, Faculty of Languages and Arts, Universitas Negeri Semarang, Indonesia 40 49 Article Info ________________ Article History: Received in April 2017 Approved in May 2017 Published in July 2017 ________________ Keywords : translation, google translate, jakarta post, translation techniques, accuracy, acceptability, readability. ____________________ Abstract ___________________________________________________________________ This research focused on the analysis of translation techniques and translation quality of Jakarta Post’s articles in terms of accuracy, acceptability, and readability. The objectives of this research were to find out the translation techniques applied by Google Translate in translating the articles of Jakarta Post and to know the quality of translation by Google Translate. This research applied a descriptive qualitative method. The data were obtained by using content analysis and questionnaire. The data in this research were ten articles of Jakarta Post from opinion section. The other data were the results of the translation quality assessments done by the respondents. The research findings showed that Google Translate applied 9 translation techniques. They were reduction, established equivalent, literal, calque, amplification, transposition, linguistic amplification, linguistic compression, and borrowing technique. The results of the questionnaires showed that the percentage of accuracy was 57.1% with average score 2.5, the percentage of acceptability was 58.6% with the average score 2.5, and the percentage of readability was 63.7% with the average score 2.5. Based on the results obtained, the quality of translation results by Google Translate was less accurate, less acceptable, and less readable. According to the analysis, it can be concluded that calque technique had the best translation quality considering this technique got the highest score in every aspect. While reduction technique was the least because this technique got the lowest score in every aspect. The researcher suggested the users to translate simple sentence instead of a complex one to gain a better quality of translation using Google Translate. © 2018 Universitas Negeri Semarang  Correspondent Address: ISSN 2252-6706 B3 Building FBS Unnes Sekaran, Gunungpati, Semarang, 50229 E-mail: dewikadewi19@gmail.com INTRODUCTION The analysis of translation processes become the focus of recent psycholinguist interest. Each culture has its own way to translate the language. When translating a source language to the target language, a translator has to understand the background knowledge of both source language and target language so that the cultural traditions can be transferred into the existing culture of target language. As an intermediary of two different languages, a translator has to know the differences between these two languages, so the translator could deliver the meaning and messages from the source language to the target language correctly. Almost all translators will find difficulties when translating a language to another language. Based on Oxford English Dictionary, “translation is written or spoken expression of the meaning of a word, speech, book, etc. in another language.” There are many definition given by linguists. Newmark (1998:5) explains that it is rendering the meaning of a text into other language in the way that author intended the text. The temptation on is to transfer as many SL (Source Language) words to the TL (Target Language) as possible. Translation is a phenomenon that has a huge effect on everyday life. This can range from the translation of a key international treaty to the following multilingual poster that welcomes costumers to a small restaurant near to the home of one of the authors (Hatim and Munday, 2004:3) According to Catford (1965: 20), translation is the replacement of textual material in one language (SL) by equivalent textual material in another language (TL). It means that translator should be able to replace the information in source text (SL) correspond with target text (TL). Catford (1965: 21) stated that the central problem of translation practice is that to find the target language tranlsation equivalence. Brislin in Suryawinata (1989: 1-2) stated that translation is the general term referring to the transfer of thoughts and ideas from one language (source) to another (target), whether the languages are written or oral form; whether the languages have established orthographics or do not have such standarization or whether one or both languages are based on signs, as with sign of the deaf. So the translation not only in the form of written text, but also can be oral form. Newmark (1988:5) stated that translation is rendering the meaning of a text into another language in the way that the author intended the text. The second meaning concerns on the author intended the text that source language (SL). From here, we can know that the translator can use what language or style they want. Steiner and Meschonnic in Hewson (1991:9) consider that translation as a particulary consistent example of human consciousness in the process of understanding through hermeneutic exploration. Hermeneutic translation is the experience of the contradiction between two cultural words which at the same time cause the translator to question his own perceptions and to assimilate those which he finds to be foreign to himself. Hartono (2011:6), otherwise, defines “translation is reading the author’s will and purpose in the form of message which contains both denotative and connotative meanings that exist in the source text that must be reproduced by translators into the target’s language. This process runs in a simultaneous cycle. In this globalization era where most people use social media to communicate, it is no longer difficult to communicate with people from different countries even if we do not understand their language because there is google translate which can help us translate. Google translate is a free multilingual machine translation service developed by Google, to translate text, speech, images, sites, or real-time video from one language into another. There are 103 supported languages in Google Translate. In November 2016, Google announced that Google Translate would switch to a neural machine translation engine - Google Neural Machine Translation (GNMT) - which translates whole sentences at a time, rather than just piece by piece. It uses this broader context to help it figure out the most relevant translation, which it then rearranges and adjusts to be more like a human speaking with proper grammar. It means that Google Translate’s quality is improved. The writer observed when doing teaching internship in SMP N 24 Semarang last year, where English teachers there let students use Google Translate. It makes the writer wonder whether Google Translate is appropriate for students to use or not. However, Google Translate is a machine translation service and of course it has weaknesses. In this study, the writer tried to find out the quality of Google Translate in employing translation techniques. In translation proccess there are so many translation techniques we can use. The discrepancies in grammar and vocabulary between languages makes it difficult to literally translate every text word for word. In order to account for this, translators use a wide variety of translation techniques in order to accurately translate any given text. Molina and Albir (2002:507) propose a definition of technique of translation which is based on two premises: 1) the need to distinguish between method, strategy and technique; 2) the need for an dynamic and functional concept of translation techniques. Albir in Molina and Albir (2002:507) states that translation method, strategies, and techniques are essentially different categories. Each expert has a separate term in determining a translation technique, so it tends to overlap between the techniques of an expert from one another. The technique is the same but has a different term. In terms of diversity of course this is a good thing, but on the other hand, related research will lead to difficulties in determining the terms of a particular technique. Therefore, in this research, the writer uses 18 translation techniques proposed by Molina and Albir to identify translation techniques used in translation by Google Translate. The techniques presented by Molina and Albir has gone through a complex study with reference and compared with translation techniques from the previous translation experts. In establishing the acceptable results, firstly the source text should be well written. In this case, ten articles of Jakarta Post are taken as samples of well written text. The articles are opinion piece written by the readers or the experts that are sent to the newspaper for the editorial staff to sort and publish as an opinion article. It is a well edited text which is ensured the content is appropriate to publish. The Jakarta Post itself is a daily English language newspaper in Indonesia which has been published since April 25, 1983. Accordingly, articles of Jakarta Post newspaper can be an appropriate source text to get analyzed the accuracy of google translate. The objectives of this study are to find out what translation techniques applied in translation of Jakarta Post’s articles by Google Translate and to find out the quality of translation by Google Translate. In analyzing the quality of Google Translate, a translator needs to know criteria of good quality of the translation. According to Nababan (2003:86), there are three reasons to evaluate the translation products which are to see accuracy, acceptability, and readibility of a translation. The accuracy, acceptability, and readability rating instrument in the form of questionnaires will be used. METHODOLOGY OF THE RESEARCH This final project concerned with texts as the data analysis. Therefore, the investigation approach employed this study was descriptive qualitative method. According to Flick (2009:45) qualitative research is centrally concerned with the production and analysis of texts, such as transcripts of interviews or field notes and other analytic materials. Meanwhile, according to Best (1981:156), qualitative studies are those in which the description of observation is not ordinarily expressed by quantitative terms. It is not suggested that numerical measures are never used, but that other means of description are emphasized. In this research, the qualitative method was used in identifying and analysing the accuracy of Google Translate in employing translation techniques. There are three roles of the researcher in conducting this research. Those roles are: first, as data collector; second, as data analyser; and third, as data reporter. As a data collector, the researcher used ten articles directly taken from Indonesian daily English language newspaper “Jakarta Post” from “Opinion” section. Then, as a data analyser, the researcher analyzed the data that have been collected, identified translation techniques used in the translation results and analysing the quality of translation by Google Translate using Nababan’s quality rating instrument. The last, as a data reporter. The researcher reported the results of the research after identifying and analysing the data. In this study, the writer collected the data from “Opinion” section of the Jakarta Post newspaper which is available at http://thejakartapost.com/academia/opinion. Those texts were then translated into Indonesian using Google Translate. Next, the output texts of the Google Translate were analyzed. In analyzing the data in this study, the writer conducted some steps. First step was reading the articles of Jakarta Post newspaper. Second, the writer translating the articles into Indonesian using Google Translate. Third, the writer identifying translation techniques used in the texts. Forth, the writer counting translation techniques used to know the frequency and percentage of each translation technique. Sixth, the writer made and distributed Nababan’s translation quality rating instrument including accuracy, acceptability, and readability in the form of questionnaires to measure the quality of translation techniques used in translation by Google Translate to respondents. Seventh, the writer counting the data from the respondents and then those data were classified based on the level of accuracy, acceptability, and readibility. Next, the writer counting which techniques had the best accuracy, acceptability, and readability score. The last step was drawing conclusion from the findings. RESULT AND DISCUSSION Among eighteen translation techniques provided by Molina and Albir, there were nine techniques used in translation of ten articles by Jakarta Post using Google Translate. Table 3.1. Translation Techniques Used in Translation of Jakarta Post’s Articles Using Google Translate No. Technique Frequency % 1 Borrowing 145 41% 2 Established Equivalent 65 18% 3 Literal 46 13% 4 Linguistic Compression 27 7% 5 Reduction 25 7% 6 Amplification 20 6% 7 Calque 18 5% 8 Transposition 8 2% 9 Linguistic Amplification 3 1% Total 357 100% The highest frequency of technique used by Google Translate was borrowing with the percentage of 41%. The others techniques were established equivalent (18%), literal translation (13%), reduction (7%), linguistic compression (7%), amplification (6%), calque (5%), transposition (2%), and linguistic amplification (1%). Table 4.1 showed the frequency of translation techniques in this study. Table 3.2. Accuracy of Jakarta Post’s articles Translation by Google Translation Indicator Number of Data Frequency Accurate 463 57.1% Less accurate 284 35.1% Not accurate 63 7.8% Total 810 100% In Table 2, it was shown that the accuracy level of the translation is divided into three level of accuracy. Those are accurate, less accurate and inaccurate. The writer found that 57.1% of the translation was translated accurately. Meanwhile, 35.1% was translated less accurately and 7.8% was translated inaccurately. Table 3.3. Acceptability of Jakarta Post’s articles Translation by Google Translation Indicator Number of Data Frequency Acceptable 475 58.6% Less acceptable 283 35% Not acceptable 52 6.4% Total 810 100% In Table 3, it was shown that the acceptability level of the translation is divided into three level of acceptability. Those are acceptable, less acceptable and not acceptable. The writer found that 58.6% of the translation was translated accurately. Meanwhile, 35% was translated less accurately and 6.4% was translated inaccurately. Table 3.4. Readability of Jakarta Post’s articles Translation by Google Translation Indicator Number of Data Frequency Readable 516 63.7% Less Readable 232 28.6% Not Readable 62 7.7% Total 810 100% In Table 4, it was shown that the readability level of the translation is divided into three level of readability. Those are readable, less readable and not readable. The writer found that 63.7% of the translation was translated readable. Meanwhile, 28.6% was translated less readay and 7.7% was translated inaccurately. 1. Borrowing Technique Borrowing technique was the most translation technique applied in translation results of Jakarta Post’s articles by Google Translate. There were 145 (41%) items found to be translated using borrowing technique. There were two types of borrowing techniques; pure borrowing and naturalized borrowing. Sample 1 ST : The plan to withdraw the free-visa facility for citizens of dozens of countries is the latest display of not only inconsistency but also lack of prior study on the government’s part when it comes to policy-making. TT : Rencana untuk menarik fasilitas visa gratis bagi warga negara dari puluhan negara adalah tampilan terbaru tidak hanya inkonsistensi namun juga kurangnya kajian sebelumnya mengenai pemerintah saat membuat kebijakan. In sample 1 (data 87) the word “inconsistency”is the fact of being inconsistent or things are not same. It is not translated into “tidak konsekuen”, but it is translated “inkonsistensi”. This is naturalized borrowing technique. Sample 3 ST : Now, 2030 SDGs is a new opportunity to start it over by prioritizing and incorporating tobacco control in every development agenda. TT : Sekarang, 2030 SDG adalah kesempatan baru untuk memulai dengan memprioritaskan dan menggabungkan pengendalian tembakau di setiap agenda pembangunan. In sample 3 (data 229) the word “agenda” is a plan of things to be done. It is not translated into “rencana yang harus dilaksanakan”, but Google Translate borrows it from English instead. This is pure borrowing technique. 2. Established Equivalent Technique There were 65 (18%) items found to be translated using established equivalent technique. Established equivalent technique used a term or expression recognized (by dictionaries or language in use) as an equivalent in the TL. Sample 5 ST : So be brave like my son-in-law: Despite his wife’s occasional criticism. TT : Jadi berani seperti menantu laki-laki saya. Meskipun kadang kadang kritik istrinya. In sample 5 (data 22) the term son-in-law was translated into menantu in Indonesian. The term menantu was an expression recognized by Indonesian dictionaries or language in use as an equivalent in the English term which meant that the technique used in this translation was established equivalent. 3. Literal Translation Technique There were 46 (13%) items found to be translated using literal translation technique. This technique translates a word or an expression word for word. Sample 8 ST : Just ask your spouse. TT : Tanya saja pasangan anda. In Sample 8 (data 44), the sentence just ask your spouse was translated into Indonesian as tanya saja pasangan anda. Ask in Indonesian was translated literally into tanya, just into saja, and your spouse into pasangan anda. The translation technique used in this translation was literal translation because every word in the SL was translated literally into the TL. 4. Linguistic Compression Technique There were 27 (7%) items found to be translated using linguistic compression technique. This technique synthesized linguistic elements in the TT. Sample 11 ST : “So what’s your name?” I said. TT : "Jadi siapa namamu?" Kataku. In sample 11 (data 34) the clause I said was translated into kataku instead of translating the ST into TT with the same number of words like aku bilang, which indicated that the linguistic elements in the ST was synthesized. That was why the translation technique used in this translation was linguistic compression. 5. Reduction Technique There were 25 (7%) items found to be translated using reduction technique. Reduction technique suppressed a ST information item in the TT. Sample 14 ST : Actually you hunks can really help boost the economy; if you handle the kias and more house chores, millions more women could join the workforce and earn income, inside or outside the home. TT : Sebenarnya Anda benar-benar dapat membantu meningkatkan ekonomi; Jika Anda menangani anak-anak dan lebih banyak pekerjaan rumah, jutaan lebih wanita bisa bergabung dengan angkatan kerja dan mendapatkan penghasilan, di dalam atau di luar rumah. In sample 14 (data 4), the word hunks which meant pria menarik was ommited. The translation technique used in translating the SL into TL was reduction since the term hunks was omitted. 6. Amplification Technique There were 20 (6%) items found to be translated using amplification technique. Amplification technique is the opposite of reduction technique. It added details that were not formulated in the ST: information, explicative paraphrasing. Sample 17 ST : But many overlook what men can do and what it takes to get males to contribute to a more prosperous society. TT : Tetapi, banyak orang mengabaikan apa yang dapat dilakukan pria dan apa yang diperlukan untuk membuat laki-laki berkontribusi pada masyarakat yang lebih sejahtera. In sample 17 (data 3), the term many found in the sentence but many overlook what men can do and what it takes to get males to contribute to a more prosperous society was translated into Indonesian as banyak orang. The term orang was added to make it clear that the term in the ST was the people. Because of that reason, this translation was called amplification technique. 7. Calque Technique There were 18 (5%) items found to be translated using calque technique. Calque translated foreign word or phrase literally into Indonesian. Sample 20 ST : However, the policy raises a serius question over the Directorate General of Taxation’s ability to ensure there is no opportunity for corrupt tax officials to abuse their power amid a slew of graft cases implicating the tax office. TT : Namun, kebijakan tersebut menimbulkan pertanyaan serius mengenai kemampuan Direktorat Jenderal Pajak untuk memastikan tidak ada kesempatan bagi petugas pajak korup untuk menyalahgunakan kekuasaan mereka di tengah serangkaian kasus korupsi yang melibatkan kantor pajak. In sample 20 (data 111), the phrase Directorate General of Taxation in the sentence However, the policy raises a serious question over the Directorate General of Taxation’s ability to ensure there is no opportunity for corrupt tax officials to abuse their power amid a slew of graft cases implicating the tax office was translated into Direktorat Jenderal Pajak which belonged to calque translation technique. It was because the foreign phrase was translated literally into Indonesian. 8. Transposition Technique There were 8 (2%) items found to be translated using transposition technique. This technique changed a grammatical category. Sample 23 ST : To the delight of many we have yet another holiday — Pancasila Day, affirming June 1 as the birthday of the state ideology. TT : Untuk menyenangkan banyak kita memiliki liburan lain – Hari Pancasila, yang menegaskan 1 Juni sebagai hari lahir ideologi negara. In sample 23 (data 46), the term the delight in the ST was translated into menyenangkan in the TT. The delight is a noun whereas menyenangkan is a verb which made it clear that the grammatical category in the ST changed. This type of translation technique was called transposition. 9. Linguistic Amplification There were 3 (1%) items found to be translated using linguistic amplification technique. Linguistic amplification added linguistic elements. Sample 26 ST : It would be truly cynical if Jokowi, elected to office on a human Rights campaign, reverted to New Order-style brutal reaction to claimed attacks on state symbols, including Pancasila. TT : Akan benar-benar sinis jika Jokowi, yang terpilih untuk memimpin kampanye hak asasi manusia, beralih ke reaksi brutal bergaya Orde Baru terhadap serangan yang diklaim terhadap simbol negara, termasuk Pancasila. In sample 26 (data 62), the sentence It would be truly cynical if Jokowi, elected to office on a human rights campaign, reverted to New Order-style brutal reaction to claimed attacks on state symbols, including Pancasila was translated into Akan benar-benar sinis jika Jokowi, yang terpilih untuk memimpin kampanye hak asasi manusia, beralih ke reaksi brutal bergaya Orde Baru terhadap serangan yang diklaim terhadap simbol negara, termasuk Pancasila. In the TT, the linguistic element made the back translation become It would be truly cynical if Jokowi, who is elected to office on a human rights campaign, reverted to New Order-style brutal reaction to claimed attacks on state symbols, including Pancasila. That was why the technique employed in this translation was called linguistic amplification. CONCLUSIONS Finding and describing the translation techniques applied in ten articles of Jakarta Post is the purpose of this research. This research also had a purpose in explaining the quality of translation results by Google Translate which included accuracy, acceptability, and readability. Based on the data analysis in the previous chapter, it can be concluded that there are 9 techniques found in the data. They are reduction, established equivalent, literal, calque, transposition, linguistic amplification, amplification, linguistic compression, and borrowing. The total data are 257 sentences from ten articles of Jakarta Post. The total techniques of the data found are 357. The percentage of each techniques as follows: (1) Borrowing technique used 145 times and represents 41%, (2) Established equivalent technique used 65 times and represents 18%, (3) Literal technique used 46 times and represents 13%, (4) Linguistic compression used 27 times and represents 7%, (5) Reduction technique used 25 times and represents 7%, (6) Amplification technique used 20 times and represents 6%, (7) Calque technique used 18 times and represents 5%, (8) Transposition technique used 8 times and represents 2%, (9) Linguistic amplification technique used 3 times and represents 1%. The most dominant translation technique found is borrowing technique, it means that Google Translate uses more familiar terms in target language to make the readers easy to understand the translation. The application of translation techniques in terms translation has some influences to the translation quality. Based on the translation accuracy assessment, the result shows that the average score of accuracy is 2,5 with 57.1% of all data are translated accurately, 35.1% are less accurate, and 7.8% are inaccurate. The result of translation acceptability assessment shows that the result shows that the average score of acceptability is 2,5 with 58.6% are acceptable, 35% are less acceptable, and 6.4% are unacceptable. While the results of translation readability shows that the result shows that the average score of readability is 2,5 with 63.7% of all data translated readable, 28.6% are less readable, and 7.7% are unreadable. It can be inferred that most of terms translation are accurate, acceptable and readable. In accuracy, calque and borrowing technique got the highest level of accuracy with the percentage of accurate translation of 73%. The other translation techniques that got quite high level of accuracy with the percentage of accurate translation above 50 were linguistic amplification technique (66.7%), literal technique (65%), linguistic compression technique (65%), and amplification technique (60%). There were three translation techniques which got the level of accuracy below 50%. Those techniques were established equivalent technique (45%), transposition technique (42.5%), and reduction technique (29%). In acceptability, calque technique got the highest level of acceptability with the percentage of acceptable translation of 75%. The other translation techniques that got quite high level of accuracy with the percentage of accurate translation above 50 were borrowing technique (73%), linguistic amplification technique (70%), literal technique (64%), linguistic compression technique (62%), amplification technique (60%), and established equivalent technique (52%). There were two translation techniques which got the level of accuracy below 50%. Those techniques were transposition technique (47,5%) and reduction technique (30%). In readability, calque technique got the highest level of acceptability with the percentage of readable translation of 80%. The other translation techniques that got quite high level of accuracy with the percentage of accurate translation above 50 were literal technique (75%), linguistic amplification technique (73.3%), borrowing technique (72%), amplification technique (69%), linguistic compression technique (65%), transposition technique (55%), and established equivalent technique (52%). There was only one translation technique which got level of readability below 50%, it was reduction technique (37%). Based on data above, it can be concluded that calque technique had the best translation quality considering this technique got the highest score in every aspect. While reduction technique was the least because this technique got the lowest score in every aspect. Considering the result and significance of the study, there are several suggestions derived from those aspects. The researcher comes with some suggestions for the Google Translate users and other researchers who are interested in translation analysis. First for Google Translate users, Google Translate is a machine translation and it has limitations. While it can help a user to understand the general content of a foreign language text, it does not, consistently, deliver accurate translation. Google Translate has difficulties in translating complex sentences. It will be better if users translate simple sentence to gain a better quality of translation. Second suggestion is addressed to researchers especially English Department students, I suggest this research is expected to give more reference for those who studying translation field or translation machine for example Google Translate. REFERENCES Best, John W.(1981). Research in Education. New Jersey: Prentice-Hall. Catford, J.C. 1965. A Linguistic Theory of Translation. London: Oxford University Press. Flick, Uwe. 2009. An introduction to qualitative Research Forth Edition Sage. London: SAGE Publications Ltd. Hatim, B. and Munday, J. 2004. Translation – An Advanced Resource Book. New York: Routledge.Latif, Muhammad Abdul (2017). “Translation Techniques and Accuracy of The Indonesian Subtitle of The Movie Philomena”. Skripsi. FBS. Pend. Bahasa Inggris, Universitas Negeri Semarang.Nababan, M.R. 2003. Teori Menerjemah Bahasa Inggris. Yogyakarta: Pustaka Pelajar. Hartono, Rudi. 2011. Teori Penerjemahan (A Handbook for Translators). Semarang: Cipta Prima Nusantara Semarang. Hewson, L. and Jacky, M. 1991. Redefining Translation: The Variational Approach. Great Britain: Biddles of Guildford. Molina, L. and Albir A.H. 2002. Translation technique Revisited: A Dynamic and Functional Approach. Meta, 47 (4). Spain, Barcelona: Universitat Autònoma de Barcelona. Newmark, Peter. A textbook of Translation. 1988. United Kingdom: Prentice Hall International (UK) Ltd. Suryawinata, Zuchridin. 1989. Terjemahan: Pengantar Teori dan Praktek. Malang: Publisher IKIP Malang.