Quality of translation via google translate in comedy texts

A translation process of considerable quality should possess the ability to effectively communicate a clear and definite meaning from the source language to the target language. While Google Translate serves as a useful tool for comprehending translated textual content in a general sense, it is important to note that automated machines still possess inherent weaknesses and limitations. The translation of jokes within comedic texts, particularly in relation to language and cultural differences, proves to be a challenging task. This research was undertaken utilizing a qualitative-descriptive approach, with the bilingual comedy book titled "Jokes in English-Book 1" serving as the subject of study. The primary objective of this research is to assess the level of translation quality achieved by machine translation in terms of accuracy, acceptability


Introduction
A translation process that has a good quality level should be able to convey a definite meaning from the source language to the target language (Smith et al., 2022).The essence of the intended translation is the equivalence of words or phrases with the same meaning between the source language and the destination language to be translated (Liu, 2018).Equivalence in determining the quality of a translation enters a challenging realm where translators are faced with the absence of similar items in the culture where the target language is located and unknown to the majority of the recipients of the translation results (Siukstaite, 2022).According to functionalists, translation is not simple as merely isolating the meaning of the text statically in a linguistic problem which produces the result in a different language (Daryl et al., 2011).The translator has a decision in selecting the text to be translated based on the knowledge possessed and can have an impact on the translation of the text into the target language where the selection of the text has two criteria that are embedded in the translator such as how the translator views the source text and the other is whether the impact is desired by the translator to created (Elnemr, 2023).
A crucial role has been played modernly in the realm of translation with regard to the dissemination of academic texts.Popularity in the translation category is related to research for non-native English circles when local journals are required to translate abstracts of manuscripts that have been composed into their mother tongue (Al Zumor, 2021).Translation has grown to play an important role in providing information to the public.The text of the translated data that is presented comes from literary works, newspapers, TV programs, and social networks.This has given rise to an area of interest in the development of more sophisticated translation methods such as the translation quality of machine translation programs (Ryabko & Savina, 2022).The significant advantage of using machine translation is in the form of google translate such as free, instant, language variations in source and target, voice recognition, and can translate entire web pages and uploaded files.However, according to Rashid's rationale, one of the shortcomings of Google Translate is that it cannot translate long texts correctly (Medvedev, 2016).
Google Translate is a very popular machine translation.The advantage of google translate is the way it is used, which is not only by typing text to be translated, it can even be in the form of photos of text to be translated into the various available languages.Google Translate can help to understand translated content in text in general, but automatic machines still have weaknesses and limitations.Google Translate can continue to develop all the time in improving the quality of translations because the algorithm is improved based on feedback from users and also uses an artificial intelligence algorithm which is expected that the common language will perform differently compared to languages with a small number of speakers.Thus, it still requires testing regarding the accuracy of the translation (Taira et al., 2021;Nasution, 2022).

Quality of translation
In the history of discussions about the quality of translation, which began in the Roman era, important criteria were "accuracy" and "precision" which in the Medieval Ages were similar to faithfulness" and "loyalty" in modern equivalence-based translation theories (Karoubi, 2018).Quality of translation (TQ) has received phenomenal attention since translation activities have entered the realm of criteria in translation quality assessment (Jiang & Tao, 2018).The quality of translation in accordance with Drugan's historic approaches can be in the form of questions about what features can complete a translation to be assessed for quality and what criteria related to linguistics, context, and objectives can be measured (Taibi & Ozolins, 2022).The quality of translation in literature is a concept that is not easy to understand.Various reference criteria in terms of translation quality such as stylistic quality, accessibility, or readability.In terms of producing a quality translation, a professional translator must have competence and indepth knowledge of the source text, source language, source culture, and even experts in knowledge related to works created by the relevant authors (Vanderschelden, 2000).
Language analysis from the translated text compared to the source language text can bring up epistemological positions and assumptions.Interpretation of the text is carried out in translation activities to be able to convey a meaning that is not merely a word-to-word translation due to overlapping (Aloudah, 2022).The obstacle that often arises in translation activities is the challenge of dealing with culturally charged translation texts.This is because the translator does not only focus on linguistic differences but also deals with the cultural gap between the source language and the target language.Particular attention is needed in handling the translation of cultural texts because it maintains the class ability of local language wisdom in translating source texts to target texts (Muchtar & Kembaren, 2018).Translation has a significant goal in the form of cross-cultural bilingual communication that connects all of humanity which began to develop at first due to international trade relations, the increasing spread of immigrants, the entry of the era of globalization, and the development of technology and information.Therefore, translators have a stake in translating concepts in various texts regarding the realm of bilingual or multi-lingual, cross-cultural transmitters faithfully and accurately (Ibrahim & Mansor, 2019).
Literary works in various categories such as reading books have been scattered including local works, foreign works, and translation works.Translation of literary works can be in the form of comedy books.Connoisseurs of literary works strive to be satisfied so the market has tried to provide a variety of reading materials that have been translated from foreign languages.Connoisseurs of translation books are not only adults but also children.In general, most translations of literary works for children have experienced additions, didactic remarks, trivialization, lecturing, and deletions (Herianto et al., 2018).In translation, meaning has a more important influence than style.The message contained in the translated text must have an equivalent and be close to the original meaning of the source text.Translation products produced by professionals need to be evaluated through translation quality assessment because not all of these translations fall into the category of good quality and contain many errors when compared with original works in the source language.Translation quality assessment is an attempt to assess translation products in terms of translation quality based on accuracy, acceptability, and readability (Agriani et al., 2018).
Discussions regarding the quality of translation must review three determining aspects such as accuracy, acceptability, and readability.Accuracy is an aspect that refers to equivalence or is not related to the translation results compared to the source language when the message in the source text is similar to the target text.Acceptability is an aspect that expresses suitability between the translated language and the source language in terms of two or more cultures, rules, and norms so that the target reader can accept the translation naturally, easily understand, and not cause problems.Readability is an aspect that combines the ease of reading that involves the source language and the target language.A high level of readability occurs when the translated text can be easily understood by the target readers (Nababan et al., 2012).

Machine translation
The dream of using machine translation emerged long ago during its golden years from 1954 to 1960 in the world of translation.The positive side of using machine translation for translators is that relying on input data for translation into translation memory software such as search engines or online dictionaries and encyclopedias has made it easier and more effective.Meanwhile, the negative side for translators with the arrival of the era of machine translation is the low cost in return demanded by clients (Taivalkoski-Shilov, 2019).Machine translation has three different thematic codes between positive comments and negative comments such as productivity, tools or interfaces, and target-text quality which have an effect on aspects such as translation costs, translation process, translation deadlines, and translation identification in influencing the quality of the translation so that if there is an error it found from the results of the translation, the translator can easily blame the work of the machine.However, there are still translators who have a positive attitude that focuses on the output provided by machine translation in facilitating the post-editing process (Vieira & Alonso, 2020).
After the 2nd world war, the use of machine translation in translation tasks began to develop, starting with the help of a bilingual electronic dictionary combined with manual lexical rules.Machine translation is a language application known as NLP (natural language processing).At the beginning of the development of machine translation, the US government formed ALPAC or Automatic Language Processing Advisory Committee where translation is more expensive than human translation but is not perfect with regards to accuracy.There are three classifications of machine translation according to the attachment to human interaction, such as MAHT or machine-aided human translation, HAMT or human-aided machine translation, and FAMT or fully automatic machine translation (Chopra et al., 2018).
The quality of translation has undergone changes that have evolved from static to dynamic which is influenced by factors such as the emergence of the phenomenon of free and open access translation technology, a range of crowdsourcing and volunteer translation initiatives.The differences in the level of translation quality that are often encountered relate to the type of content and how the translation is done professionally through community or group sources, translating using machine translation and coediting by many people (Jiménez-Crespo, 2017).The quality of translation by humans is measured in various ways which are different but complementary when compared to methods for measuring the quality of translation by machines or computers depending on whether quality is judged by performance by hand, automatically or semi-automatically (Rojo, 2018).Business practices in translation since the appearance of machine translation have had an impact on the market and the economy so that it is predicted that the influence of this automated machine will replace the job of the translator (Vieira, 2020).
One of the advances in the use of machine translation is an application known as Google Translate, which is conveniently in the palm of your hand via a smartphone.Language translation is done automatically on text or audio orally (Nunez-Marcos et al., 2023).Google translate is indeed an important translation tool, but that doesn't mean that the machine can translate all languages consistently.Google translate has weaknesses such as not having proofread tools, output and structure that are not necessarily accurate, so human corrections are still needed to help machine translations (Winiharti, et al., 2021).Translation on a machine has quality or not because it is easier and simpler for several language pairs to be translated and the greater availability of parallel corpora in machine translation for several language pairs (de Vries et al, 2018).At present, the system on machine translation has input millions of sentences that have been translated by humans and systems on machines are continuously being developed to improve the quality of translation to make it more efficient.The effectiveness of using machine translation related to the quality of the output depends on the property of the same language family (Munkova et al., 2021).
There are reasons why the text translated from the machine loses its meaning, such as translations in the form of words that are produced not in accordance with gold standard documents or human translations, which are related to different term-document matrices (TDMs) and words in the corpus contained in machine translation are different when compared with words in the corpus with the same topic in the gold standard corpus (de Vries et al., 2018).Prospective translators related to the use of machine translation need to learn to use translation methods and techniques and edit text after translating via Google Translate.Errors that occur when using google translate are weaknesses in terms of the meaning of the source text.Overall the quality of the translation via google translate is top notch so a skilled mastery of the use of machine translation is an important factor for training future translators (Borodina et al., 2021).The limitations of learning translation via Google Translate are that it is only a learning strategy that cannot replace the teacher's role and the type of technology that cannot be separated from e-learning readiness, dependence on internet access and the use of applications between hardware and software (Bahri & Mahadi, 2016).
Basically, translation problems with Google translate that often occur are found at the lexical level.Limitations on the lexical stage, translation using Google translate does not match the format according to the language background, deviations from the context of lexical forms in general, definitional problems and errors in the field of idiomatic expressions.Meanwhile, Tytler's law of translation states that a translation must contain a perfect transcript of ideas from the source language, the writing style of the translated text must match the characters in the source language and the translation must have the composition of all conveniences derived from the source language (Vidhayasai at al., 2015).The quality of translation products via Google Translate still requires substantial post-editing so that the text function can be fulfilled (van Rensburg et al., 2012).
Machine translation is famous for its programs that manage statistical calculations on a sentence from the source language to be translated into a different language.A popular statistical machine translation is Google Translate using a multilingual machine translation cloud service under the auspices of Google Inc.The results of the translation via Google Translate still require revision based on the accuracy, acceptability and readability aspects where the programming on the machine should have included knowledge between the source language and the target language as well as knowledge related to the two cultures of the different languages (Fayruza et al., 2020).

Comedy in translation
Comedy is an act of translation as a theme with universal and historical stock so that situations are equally appropriate to new listeners.The decisions made by comedians are no different from translators when choosing to adopt material so that a strong connection is established between translation learning and comedy learning (Kirk, 2011).Comedy translation or in other words, humorous translation cannot be done carelessly because it requires aspects of the complexity of the joke, unformulated and must be understood very intuitively.Cultural factors that influence comedy translation are divided into two classifications, namely intracultural and intercultural.Intracultural such as period style and strategic orientation.Meanwhile, intercultural factors are cultural-specific expressions, aesthetic differences, ethical influence and political interference.In science, comedy translation is too young in development with three universal comedy topic categories such as absurdity, degradation, and sex (Phimtan & Tapinta, 2011).
The self-translator in translating the case of comedy requires confirmation and validation with regard to the production of the 'same' text in different languages which is usually accompanied by a desire for freedom.Text written by self-translation is a resume of the process of text production which experiences fluidity.Conversely, verbal situations can also occur during translation as a support for written communication that accepts fluidity orally which is not a form of betrayal on the mission of the translation actor (Palmieri, 2017).
Jokes in comedy texts in relation to language and culture are not easy to translate because of language and cultural differences (Alnusairat & Jaganathan, 2022).Comedy text contains linguistic, cultural and universal elements.Comedy text is one of the challenges in the field of translation science.This type of text must be able to trigger the same sense of desire to laugh in the source language and target language.At first, the comedy elements in translation were found in subtitles and comics which became a start that the meaning of comedy could be translated into the target language.The function of comedy text is to convey a joke message, but if the translation produces the meaning of the joke cannot be conveyed to the target reader, the quality of the translation is not good.Artificially intelligent and deep learning relates to the use of google translate to translate comedy texts properly because this application is supported by the Neural Machine Translation system.
Although, in translating comedy texts, machine translation still requires postediting so that the jokes contained in the translation product are good (Ardi et al., 2022).The selection of this comedic text is motivated by the contemporary preference of students for engaging in extensive reading that incorporates a humorous element.Consequently, it is imperative to provide a valid rationale for investigating the aptitude of translation skills within this domain.This can be achieved through an examination of bilingual literature and its correlation with the widely utilized translation tool, Google Translate, which currently enjoys considerable popularity.

Research design
This research was conducted with a qualitative-descriptive approach.Research with a qualitative-descriptive approach is a procedure that is commonly used in many disciplines including in second language teaching and learning because it belongs to the part that recognizes complex motivation.The approach in question is widely used in manuscripts on second language teaching and learning because it contains naturalistic data without intervention and manipulation of variables.Sometimes, a holistic qualitative approach can also be analyzed quantitatively when relevant themes and ideas are converted to numerical form for comparison and evaluation.In general, descriptive is explaining the phenomenon and its accompanying characteristics (Nassaji, 2015).

Sample
The research object under study was taken from a bilingual comedy book in English-Indonesian with the title "Jokes in English-Book 1" which consisted of 38 schoolthemed jokes, 30 jokes about family and relatives, 19 jokes about office and job, 19 jokes about shopping, 18 jokes about vehicles and traffic, 18 jokes about travel and vocation, 10 jokes about crime and law, 21 jokes about animals, 12 jokes about death and funeral, 5 jokes about tales, and 8 jokes about good luck or bad luck?.The research sample was selected purposively with the title "Saved by a blind horse" which is a group of jokes on the theme of travel and vacation (Baehaqi, 2011).Purposive sampling or judgement sampling is a technique for choosing quality information by choosing it deliberately which is not random and does not require a basic theory or a set of informants but the researcher determines the information based on the valid quality of the knowledge and experience contained in the information as a representative (Tongco, 2007).Purposive sampling is the choice of technique in determining the objects in the study that are in accordance with the aims and objectives of the research so that the results from the data can increase trust and accuracy because of confirmability, transferability, dependability and credibility (Campbell et al., 2020).

Data collection method
In analyzing data about the quality of translation related to comedy or humour text, the instrument is in the form of a translation test (Rezqi & Ardi, 2022).The sample was examined using the test method via Google translate from joke sentences about speaking English to Indonesian.This is done to measure the level of translation quality of the machine translation in terms of accuracy, acceptable and readability aspects.Then, the results of the quality of the Indonesian translation via Google Translate will also be compared with the Indonesian language sentences contained in the bilingual comedy book under study in order to find out which translation quality is the best, which is the translation product via Google Translate or the translation product provided in the bilingual comedy book.Buck (1992) in teaching foreign languages about translation, it is common to use the test method.Hayati et al. (2020) the test method is a measurement tool regarding commands and questions in order to obtain a response according to the instructions.The activity of taking these measurements is the beginning of an evaluation in the assessment of learning outcomes.

Data analysis
The translation results via Google Translate were assessed using the translation quality of Nababan et al. (2012) based on instruments related to accuracy, acceptability and readability on a sentence by sentence basis.Instruments related to the accuracy parameter are divided into three categories, namely accurate because the meaning of the sentence from the source language is translated accurately into the target language and there is no distortion of meaning.Therefore, a translation is considered accurate if the message contained in the translated text must be the same as the message in the source text so that activities to reduce or add to the content and message of the source text must be avoided in the target text.Less accurate because the majority of the meaning in the source language sentence is transferred to the target language accurately but there are meaning distortions, double meaning translations, omitted meanings or incomplete message meanings.Inaccurate because the meaning of the sentence in the source language is translated inaccurately or deleted into the target language.
Instruments related to parameters regarding acceptability are classified as acceptability because the translation is felt natural, technical terms are familiar to the reader, and the sentences used are in accordance with the rules; less acceptability because in the majority it already feels natural but there are still problems with technical terms or grammatical errors; and not acceptability because the translation is not natural, the translation feels like a work of translation, the technical terms are not familiar to the reader, and the sentences used are not in accordance with Indonesian language rules.Instruments related to parameters regarding readability are divided into three with explanations such as high readability because sentences in the translated text can be understood easily by readers; moderate readability because in general it can be understood by readers but there are certain parts that must be read repeatedly more than once to understand the translation; and low readability because the translation is difficult to understand by readers.The weighting of scores in the classification of translation quality that is accurate, acceptability and high readability is 3. Less accurate, less acceptability and moderate readability have a score of 2. Meanwhile, the criteria for translation quality that are inaccurate, not acceptability and low readability have a score of 1.Then, the results of the data obtained will be presented through three technical descriptions in terms of data reduction, data presentation, and conclusion drawing or verification (Miles & Huberman, 2005;Bania & Imran, 2020).

Findings and discussion
The research object was taken from English sentences in the bilingual book "Jokes in English" from book 1 version on page 126 with the title "Saved by a blind horse" which has 8 sentences to be translated into Indonesian via google translate to determine the level of accuracy, acceptability and readability of the machine translation results.

Table 1
Quality of translation via Google lens through the parameters of Nababan et al. (2012).The results of the eight sentences translated via Google Lens from English to Indonesian obtained 5 sentences or 62.5% translated with high accuracy because the score obtained was 3 in sentences numbered 1,2,5,7 and 8.The five translation products are accurate.Then, 3 translations or 37.5% were obtained with a score of 2 due to distortion of meaning.In sentences number 3 and 4, there is a translation of the horse's name from Buddy to Budi so that the completeness of the translation in the sentence becomes imperfect.Meanwhile, in the translation in sentence number 6, Buddy is translated as Sobat in Indonesian.In translating names from text characters, it is not necessary to translate because it can make the translation results less accurate.Even though sentence number 2 is accurate where the horse's name, namely Buddy, is not translated or it is still written Buddy on google translate so that the translation results in this sentence are highly accurate.Accuracy is an aspect that includes the accurate understanding of the source language into the translation in the target language as much as possible as a reference for the quality of the translation with a high standard.With regard to the quality of translation by machine translation, the accuracy obtained in translations via Google Translate is highly variable where usually translations between European languages are good but translations related to Asian languages are relatively poor (Aiken & Balan, 2011;Aresta et al., 2018).

No. Source Language
With regard to acceptability, 3 translation sentences or 37.5% are obtained with a score of 3 while the other 5 translated sentences obtain a score of 2 or 62.5%.In sentences number 2, 7 and 8, the results of the translation are in accordance with the culture, rules and norms of the target language.In sentence number 1 there is a translation of the part "drove his car into a ditch" to "mengendarai mobilnya ke sebuah selokan" which is not accepted because the message is conveyed that the driver deliberately put the car into a ditch even though the incident was unintentional.Sentences with numbers 3 and 4 are less acceptable because the translation of the horse's name from Buddy to Budi because Budi in Indonesian when translated into English literally means "Character", so this translation is very far from culture and rules.Sentences with numbers 3 and 4 are less acceptable because the translation of the horse's name from Buddy to Budi because Budi in Indonesian when translated into English literally means "Character", so this translation is very far from culture and rules that make sentences feel ambiguous.Likewise, in sentence number 6 when Buddy is translated as Sobat, even though the literal meaning in the dictionary is correct, this result undermines the integrity of the message in the sentence because the name of the horse should not be translated.If the animal name is translated, it will give the impression in Indonesian that the horse is a friend or relative.Acceptability of translation is related to the culture of the receptor and the reader of the translation so that the translation product must comply with the appropriate norms and expectations of the target readers (Gross, 2003;Castilho & O'Brien, 2017).
For the results of the translation related to the quality of the translation in the readibility section, 3 translated sentences in numbers 2, 7, and 8 or 37.5% have a score of 3 so that the readibility level is high because the translation can be read and understood easily without reading it repeatedly by the reader of the target text.Meanwhile, the remaining 62.5% of the translation has moderate readability as in sentence number 1 where readers of the translated text need to understand the text repeatedly because it is unusual that the message in the translation can be concluded as someone deliberately dropping his car into a ditch.In the problem of sentences number 3 and 4 where Buddy is translated into Budi and in the translation sentence number 6 where Buddy is translated into Indonesian to become Sobat has made the target text reader to read and understand repeatedly due to the inconsistency of the naming of the horse from the translated reading material even in Translation sentence number 2 which is correct does not translate the horse's name and remains with Buddy.Although, the problem with the naming of the horse that was translated from Buddy to Budi or Sobat, this has damaged the quality of the translation as a whole is not good.Finally, in sentence number 5 at the end of the sentence there is a translation of "Nothing" from the English source text with the word "tidak ada" but this translated sentence is still incomplete, it still feels like a translation and is not natural so it needs to be read and understood repeatedly where there should be deepening with regard to "nothing" or "tidak ada".
Readers do not understand what is "tidak ada" to be directed where the translation is to the actions of the horse, the condition of the driver, the condition of the car or something else.Readability is a requirement in the scope of comprehension and has a benchmark in terms of difficulty in vocabulary (Acar & İşisağ, 2017).Readability or accessibility is a reference that the text can be read and comprehend easily so that it can be seen that the quality of the text is easy or difficult which is influenced by factors such as the length of words or sentences, the interrelationships between reading content, interests and purposes of writing, the complexity or simplicity of the use of sentences and vocabulary abstraction (Guo, 2022).
The following is a comparison of the quality of the original translation adapted from bilingual books in Indonesian with the quality of translation via Google based on indicators of translation quality by the satisfaction of 100 respondents, as follows: The data obtained from the translation results were compared between the existing Indonesian translations found in the bilingual book and translations into Indonesian via Google Translate, it was found that the respondents were completely or 100% satisfied in the aspects of accuracy, acceptability and readability of the translations in the bilingual book.

Conclusion
The outcomes of the translation from English to Indonesian concerning comedic texts through the utilization of Google Translate demonstrate a commendable level of proficiency; nevertheless, there is room for enhancement and modification in order to achieve a more refined result.This ongoing development aims to bring the three fundamental facets of translation quality, namely precision, acceptability, and comprehensibility, closer to a state of perfection, particularly in relation to cultural connections, regulations, and norms between the source and target languages.Presently, the recipient audience manifests a greater degree of satisfaction towards translations that have not been entirely generated via machine translation.