Accuracy and Readability from Oscar Wilde’s Quotes on The Feature of Facebook Neural Machine Transla
Siti Munawaroh Lita Septiana Darsono
This research applies translation quality assessment approach to help the researcher masters the quality of quotes translation in Facebook neural machine translation. This research uses qualitative method in which the result tends to answer the problem statement that focus on the terms of accuracy and readability. The result of this research, it can be concluded that the translation of Oscar Wilde’s quotes on the feature of Facebook neural machine translation is quite accurate since this research describes that from 100 data there are 87% accurate, 12% are less accurate, and 1% is inaccurate. In the terms of readability, it can be concluded that the translation is quite readable since the research describes that from 100 data the researcher took five raters to scoring the readability level to each quotes translation and found 42,4% are readable, 35,5% are less readable and 22,3% are unreadable.
Keywords : Accuracy, Readability, Quotes, Facebook, Neural Machine Translation
Nowadays, human being can not be separated with technology. Moreover, technology is being a part of human activities with all the new brands that had offered by many companies include in social media. Many people from around the world can communicate each other and using social media as the mediator. Facebook, Twitter, Instagram, Skype, Line, Friendstalk, Messenger, Kakaotalk, are examples of social media. All the social media are mentioned above can not be separated with all our daily life. Those kinds of social media can link two or more people in getting in touch with their family or friends even if they separate thousand miles just in one click and a second waiting. In addition, it is easy to keep communicate each other by video call and other application that social media have given. One of the most popular social media here is Facebook, Facebook in this case become the most popular website in the world with over 600 million users (Ahmad, 2011).
Facebook first launched on February 4th, 2004, by Mark Zuckerberg along with fellow Harvard College students. Facebook may be accessed by a large ranges of desktop, laptops, tablet computers, and smartphones over the internet and mobile networks. After registering to use the site, users can create a user profile indicating their name, occupation, schools attended and so on. Users can add other users as “friends”, exchange messages, post status updates and digital photos, share digital videos and links, use various software applications, and receive notifications when others update their profiles or make posts. Additionally, users may join common interest user groups organized by workplace, school, hobbies or other topics, and categorize their friends into lists such as "People From Work" or "Close Friends". Additionally, users can complain about or block unpleasant people. Because of the large volume of data that users submit to the service, Facebook has come under scrutiny for its privacy policies.
Generally, people tend to use Facebook as a self promotional (Carpenter, 2012). Because when, the users are seeking a wider audience, they also predicted to accept friend requests from strangers because they would be seeking an audience rather than using Facebook to engage in social interaction with existing friends. They may also attempt to gain the attention of their audience by frequently offering new content.
Posting status updates, posting pictures of themselves, and changing their profile are all methods of using Facebook to focus attention on the self. This tendency of posting status on Facebook to catch other feeling and attention also influence the people to share their feeling on a personal status by citation the other person word as reflected as a quote. Based on the survey, the researchers has been searching for ten participants as a Facebook user, the researchers gave a questionnaire with several questions. Ten participants allowed to choosing whether they usually posting a status on Facebook using a quotes are to seek wider audience and catch another attention or share their feelings by updating some quotes related to their feelings. The answer is full of variation, but the higher answer is they posted a status by using quotes is to share their feelings related to the quotes they posted.
The researchers choose Facebook rather than other social media like Instagram or Twitter because Facebook and Instagram has one corporation and same tools to translate. In Instagram, we can not just update status, but we must also update a picture. Facebook more facilitate the researchers rather than Instagram. Twitter also use Bing translator to translate that just can translate direct translation or word to word and when we use Twitter, we can not update status more than 140 characters. It is why the researchers more choose Facebook rather than other social media.
Related to the survey, the researchers attempt to analyze the post translation based on the quotes.
The selected quote in this research is by the most quoted figure from Oxford Dictionary of Humorous Quotation (2008) and regarding to the statement of Gary Quinn (2013) that placed Oscar Wilde as the most quoted figure. Moreover, when users post status on Facebook, the user from other countries will understand what the status posted though we are using our native language. Facebook use machine translation to translate text in posts and comments automatically, in order to break language barriers and allow people around the world to communicate with each other.
Facebook, which uses machine translation to translate text in posts and comments automatically, announced in a blog post on August 3, 2017 that it has completed transitioning to a neural machine translation (NMT) system. New features of Facebook neural machine translation facilitate the users to understand the language from other language, in this case people from different countries are free to choose the language into their native language, as an example the researchers’ native language is Bahasa Indonesia and the researchers choose the setting of the researchers’ Facebook account language is Bahasa Indonesia. When users send a post using English, other users with less comprehension on English can click the button “Translate” and Facebook will automatically translate the foreign language into the users’ native language. This kind of feature functions not only on the posted status but also in the comment. When users comment on the status with foreign language, the wall owner can check the translation by clicking ‘Translate’. Then, the translation will appear automatically.
The translation will appear on the figure below:
The purpose of giving a translation in posted status in other language on Facebook neural machine translation is for making easy for the users when they are reading the status of people in different languages. In rendering one language to another, Facebook websites certainly involves a translation process. According to Newmark (1988:7), translation is rendering the meaning of a text into another language in the way that the author intended the text.
Facebook, in this case with certain application as mentioned above translating the posted status in the system Machine Translation. Machine Translation is a translation produced by advanced technology, without the intervention of human translators as suggested by Rebecca and Stiegelbauer (2012) it also as a replacement of human’s work in translating text from source text into target text. The aim of analyzing quality assessment from Facebook neural machine translation in the terms of accuracy is to measure the accuracy and readability of each quote translations whether the translations are accurate / readable, less accurate / less readable, or inaccurate / unreadable.
Regarding to the new models, the researcher interest to analyze whether the new models of Facebook neural machine translation truly provide more accurate and fluent translations or not. Facebook’s status used as the source data of this research is from the quotation of Oscar Wilde. The researcher also interest to analyze and provide the Facebook neural machine translation is readable for the target reader or not. When Facebook translation has good accuracy but less readable, the target reader will not understand the meaning of the phrases or sentences. And if the translation has less accuracy but good readability, the meaning of the phrases or sentences will not deliver to the target reader.
In analyzing the quality assessment from each quotes translation, the researcher took five raters with the background of study beyond from English department or not correlated with English major. Five raters were asked to scoring the readability from each quotes translation just in the terms of readability. The score 3 is readable, 2 is less readable and 1 is unreadable. This research took 3 days to share the questionnaire and analyze each quotes translation.
The research of Facebook neural machine translation above is qualitative. This research aimed to analyze the accuracy and readability of each translation from Facebook neural machine translation updated as a status by using quotes from Oscar Wildes. The steps to analyze the accuracy used were first, collecting the data namely the quotes from Oscar Wilde by updating status via Facebook total one hundred quotes with every status update has a sentence. Second was collecting the target text from Facebook automatic translation. Every status updated will automatically translated by Facebook and the researchers collected the target text from the feature. Third, was analyzing the target text from Facebook neural machine translation of the quality assessment of each quote in the terms of accuracy.
The steps to analyze the readability used were first, collecting the data which translated by Facebook automatic translation.
Second was the researcher finding five raters whom have major non-English. Third, the researcher analyzes the quality assessment of translation in terms readability obtained from questionnaire distributed to five raters. Fourthly, the researcher calculates the finding data into the table of percentage. The researchers focus on how is the accuracy and readability of the quotes translation on Facebook neural machine translation. The researchers only analyze the translation on Facebook neural machine translation feature. The researcher focuses on translation quality assessment based on the theory of Nababan (2012).
The data that had been analyze, classify based on the total of accuracy of each quote translations. The data input into the table below:
Score: Accurate: 88/100 x 100% = 88%
Less Accurate: 12/100 x 100% = 11%
Inaccurate: 1/100 x 100% = 1%
The sentences are included in accurate translation if meaning words, technical terms, phrases, clauses sentences or text language sources accurately transferred into the target language; the same no distortion of meaning. To make the conclusion easier, the sentences also classify based on the type of the sentence.
The score of this translation is three, it means that the message from the source of text is accurate conveyed to the target text. The example above is an example of accurate translation from quote’s translation.
Example 2 :
The sentences are included in less accurate translation if the most of the meaning of words, technical terms, phrases, clauses, sentences or source language text has been transferred accurately into Bahasa Indonesia target, however there are still a distortion of meaning or no meaning or translation double meaning or no meaning eliminated, which disrupt the integrity of message. The example above is an example of less accurate translation from quote’s translation.
The score of this translation is one, it means that the message from the source of text is inaccurate conveyed to the target text. The example above is an example of inaccurate translation from quote’s translation.
The data that had been analyze, classify based on the total of readability of each quote translations. The researcher took five raters to scoring of each quotes translation. The data input into the table below:
Score: Readable: 212/500 x 100% = 42.4 %
Less readable: 193/500 x 100% = 35.3 %
Unreadable: 95/500 x 100% = 22.3 %
A translation text can be said readable if the target readers can understand it because it has a good diction and meaning. Therefore, the readers can receive well the message is transferred from the source language into target language. From the raters, the researcher found 212 data from 500 or 42.4% is readable. The examples are presented as table below:
The translation considered less readable when the target reader could understand the meaning, but there is a part is needed to read more than one time. The rater can not understand the meaning briefly 193 data or 35.3% data is less readable. The examples are presented as table below:
The translation considered unreadable if the target reader is difficult to be understood the meaning. The unreadable quote are also influenced by the techniques used by the writer and the diction used in captioning. The unfamiliar dictions used in the quote are also makes the translation difficult to be understood. The examples are presented as table below:
Notes: Mean for Readable: 2.4 – 3.0 point
Less Readable: 1.8 - 2.4 point
Unreadable: 1.0 - 1.6 point
To sum up, machine translation, in this case are useful to a certain extent, but they are not to be trusted, as they are not useful without the intervention of the human mind. Reading the translations above, one can easily notice that all of them are close from being accurate and readable. Moreover, Machine Translation technology is improving all the time. Many programs are running around the world right now and it constitutes an exciting area of translation research, especially when combined with the human touch. It is likely that, over time, this research will gradually extend the boundaries within which the Machine Translation can operate or within its evolution.
In this paper, the correlation between quality assessment from machine translation are quietly accurate and readable because Facebook developing their machine translation by the intervention of human also. The user from each or around the world can give a better translation of each status in other’s native language so that it can improve Facebook machine translation in translating the status. As the researchers see it for the result that had been analyzed, the result provide the Facebook neural machine translation has a good quality in translating a post from source text into the target text.
A, Nida, Eugene and R. Taber, Charless. 1974. The Theory and practice of Translation. United Bible Societies: E.J. Brill Leiden.
Ahmad, A. (2011). Social Network Sites and Its Popularity. International Journal of Research and Reviews in Computer Science, 2, 522–526.
Brislin, Richard. W. 1976. Translation: application and research. New York: Gardner Press Inc.
Catford, C.J. (1974). A Linguistic Theory of Translation. Oxford University Press. London
Gary Quinn (2013), Oscar Wilde named most quotable figure in the English Language. Available at : http://www.irishtimes.com/culture/books/oscar-wilde-named-most-quotable-figure-in-the-english-language-1.1563213 (Accessed on 16th November 2016)
Nababan, Nuraeni & Sumardiono. (2010). “Pengembangan Model Penilaian Kualitas Terjemahan” Laporan Penelitian Hibah Kompentensi. Surakarta: Universitas Sebelas Maret.
Newmark, Peter. 1988. A Textbook of Translation: Theoritical and Practical Implications. Philadelphia: Garant Publishers.
Nida, E. 1964. Toward a Science of Translating. Leiden: E. J. Brill
Rebeca, L. & Stiegelbauer, P. (2012). Automatic translation versus human translation in nowdays world. Published by Elsevier Ltd.
Rochayah, Machali. 2000. Pedoman Bagi Penerjemah. Jakarta: Grasindo.
Tripathi, S. & Sarkhel, J., K. 2010. Approach in Machine Translation. Annals of Library
Whitson Gordon (2011), Facebook Now Translates Posts and Comments Not In Your Native Language.
Available at http://www.jostrans.org/issue11/art_fiederer_obrien.php (Accessed on 16th November 2016)
www.brainyquotes.com/author/oscar_wilde (Accessed September 3rd, 2016)