Machine Translation and Human Translation : An experimental study of Gujarati short story 'Jumo Bhisti' by Dhumketu.
Himanshi Parmar
P.G Student
Department of English
Maharaja Krishnakumarsinhji Bhavnagar University, Bhavnagar, Gujarat.
Email. Id. - himanshiparmar3004@gmail.com
Abstract :
Machine translation and human translation are two different approaches to the task of translating text or speech from one language to another.Machine translation is the use of software to automatically translate text from one language to another.Human translation, on the other hand, is the process of translating text or speech by a person who is fluent in both the source and target languages. Translation is a transfer of meaning from one language to another.It plays a vital role in literature of all the languages. This research paper includes an experimental study of Gujarati short story 'Jumo Bhisti' by Dhumketu. Machine translation has its limitations, and it may not always produce accurate translations. One of the main challenges is to understand the context and idiomatic expressions that are hard to translate. While human translation can be more time-consuming and expensive than machine translation, especially for large volumes of text. This research paper aims to research on advantages and Disadvantages of Machine translation. Also emphasizes the difficulties encountered while doing machine translation by translating a Gujarati short story of Dhumketu titled as 'Jumo Bhisti' in Google translator.
Key words - Jumo Bhisti, Dhumketu, Machine Translation, Usefulness, Difficulties, Llimitations.
History of Human Translation and Machine Translation
The history of translation dates back to ancient times when different cultures and civilizations came into contact and needed to communicate with each other. Translation has been essential for the dissemination of knowledge, the preservation of cultural heritage, and the promotion of cross-cultural understanding.
In the early days of translation, the most common method was oral translation, in which a speaker of one language would translate a text into another language for an audience. This method was used for religious texts, as well as for legal and commercial documents.With the invention of writing, written translation became more common. The ancient Greeks are credited with creating the first known written translation, a translation of the Iliad into Egyptian. During the Renaissance, the rise of humanism and the interest in classical literature led to an increase in written translations.
In the 20th century, machine translation emerged as a new field of study. The first machine translation systems were developed in the 1950s, using rule-based methods. However, these systems were limited in their ability to translate accurately and effectively. With the advent of deep learning and neural networks in the 21st century, machine translation has made significant progress and is now widely used for a variety of purposes, including language education, e-commerce, and communication. The history of translation reflects the evolution of human communication and the desire to share ideas and knowledge across cultures and languages. It is a field that continues to evolve and advance with technological advancements and the increasing need for cross-cultural understanding.
Problems while using MachineTranslation
1 - Inaccurate translations :
Despite advancements in machine translation, it still struggles to achieve human-level accuracy in translating complex texts.Machine translation can result in inaccurate translations due to the following reasons: 1) Lack of context: Machine translation lacks the ability to understand and retain context, which can result in misinterpretations of the source text. 2) Ambiguity in language: Natural language is often ambiguous and machines may not understand the intended meaning, resulting in incorrect translations. 3)Technical limitations: Machine translation algorithms may not be able to accurately translate idiomatic expressions, figurative language, or sarcasm. 4)Limited knowledge: Machine translation is only as accurate as the data it was trained on, and it may not be up-to-date with the whicht information and changes in language usage.5Machineral differences: Machine translation may not take into account cultural differences, resulting in translations that are not culturally appropriate. It is important to have professional human translators review and edit machine translations to ensure accuracy and cultural sensitivity.
In the story 'Jumo Bhisti' by Dhumketu, the fifth point is proved right. Culture of Gujarati is different from the English. The story includes many native words which Google translator failed to translate due to lack of cultural understanding such as 'પણે' means 'There' but Google translator cut that part. The another word is 'રણકીને' that was removed by Google translator.
2 - Lack of human-like nuance and understanding / Fail to capture feelings and emotions :
Machine translation does not possess the cultural understanding, emotional intelligence, and linguistic sophistication of human translators, leading to grammatically correct but semantically incorrect translations.Machine translation still has limitations in accurately capturing cultural, idiomatic, and linguistic nuances, leading to limitations in its ability to understand and replicate human language and communication. This results in awkward or incorrect translations. Advances in NLP and deeper language understanding are continuously being made to improve the accuracy and naturalness of machine translation.
In the story such words like જો! જો! (See! See!) Was translated as If! If! In the sentence 'તમે માબાપ, ટેકો દો તો પગ નીકળી જાય,..' (If you help, the legs will come off) the word 'માબાપ' indicates two persons whom jumo ask for help but Google translator translated it as a 'parents'. The word was used to request people not for parents. Also the word 'ભાઈ માણસ' (male) is wrongly translated as ' brother - man', word 'થોડુંઘણું' (few/little, some) translated as 'little more' which again wrong. The word 'ખંજવાળવું' (to pamper) is a quite positive word that translated as negative 'Scratched'.The word 'કેડિયું' (Gujarati male outfit) is translated as chain because lack of cultural understanding.
3. Limitations in handling idiomatic expressions and cultural references :
Machine translation can struggle to translate expressions that are specific to a particular culture or language, and can result in awkward or meaningless translations.Idiomatic expressions and cultural references are often difficult for machine translation to handle as they may not have a direct equivalent in the target language or may carry different connotations. This can result in lost meaning or misunderstandings in translations. Machine translation systems are improving in their ability to handle such expressions, but the challenge remains due to the complex and nuanced nature of human language and culture.
In each culture there are different words for different expressions. In the story 'Jumo Bhisti' expressions that used are 'હો!' Which one can correctly be translated as Okay! But here translation machine translated it as 'ho!' only.
4 ) Some words remain as it is :
While experimenting with the story, the result was that some words remained as it is. Machine translation made no changes in it. The words that remained as it is are mostly rural, native words. It is hard to do an English translation of those words and even some words don't have any specific words in English. The words that remained as it is are Lakshadhipati' (Rich), 'Seth' (owner), 'Paise' (Money/Rupees), 'Gadab' (A kind of Grass for animals), 'Netar' ( A kind of Wood), 'Parvardigar' (God), 'Ho' (Okay) etc.
5 ) Difficulty in dealing with homonyms and polysemy :
Words with multiple meanings can be challenging for machine translation, as it may select the incorrect meaning for a given context. Homonyms and polysemy (multiple meanings for a single word) present challenges for machine translation as a single word can have different translations depending on context. This can result in ambiguity and incorrect translations. Machine translation systems often use context and semantic analysis to disambiguate such words, but the task remains difficult due to the complexity of human language and the multitude of possible meanings a word can have. For example the word 'જો! જો!' has two meanings within. The first is 'to see' and the second is 'If'. In the story the word is used to mean 'see!' But Google translator used the word 'If' instead of 'see' which raised a different meaning of sentence and was unmatched translation with story.
6) Negative to positive - Positive to Negative :
While using machine translation, one of the difficulties is, it sometimes maken positive words negative and negative words positive. For example in chosen story of Dhumketu the word 'જાનવર', here Venu, the Gaur is translated as 'Beast' which is very negative term in reference to the story. While in the original Gujarati story the word is depicted in a very positive manner.
7) Spelling errors and Gender Changes :
Sometimes the translation machine also made spelling errors and misspelled some words. It especially happens in names, like name of place, people, things etc. In the story the name protagonist is 'Jumo' which sometimes translated as 'Juma', also the word 'Pado' which should translate as Gaur was remained as 'Pado' only and sometimes machine translator made mistake and translated it as 'Paddy' also. Google translator translated 'Gaur' as female by using 'her' for it, where 'It' is the proper word to show the gender of Gaur.
8) Inability to capture context and background knowledge :
Machine translation often lacks the ability to consider contextual information, background knowledge, and the intended meaning behind a text, leading to misunderstandings and errors .Machine translation often lacks the ability to capture context and background knowledge, which is important for accurate understanding and translation of a text. This can result in incorrect translations or loss of meaning. Machine translation systems are utilizing more advanced techniques, such as incorporating context and background knowledge, to improve the accuracy and naturalness of translations. However, the challenge remains as language and human communication are complex and multifaceted. In the short story such words like 'થીગડા' (Patch) translated as 'Frozen Rock', 'હુક્કો' (Hookah) translated as 'Hook', 'મસક' ( kind of a pot of water) translated as 'Mask', 'લીન થવું' (to sink into) translated as 'Absorbed'. All these words are wrongly translated and lose their meaning after translation, that is showing limitations of translation.
9) Use of unwanted punctuation marks -
Machine translation many times use unwanted or unnecessary punctuation marks such as question mark, exclamation mark, full stop etc. In the story 'Jumo Bhisti' by Dhumketu also we find use of unnecessary question marks at two three places. It also sometimes misses out to use such punctuation marks where it actually needs to. Sometimes machine translation also simplifies complex sentences that cause changes in meaning of sentences and other problems.
Though machine translation has many Disadvantages and it makes lots of errors, it is also useful in many ways. Here are some.
Usefulness of Machine Translation
1] Speed :
Machine translation can translate large amounts of text quickly, making it an efficient tool for processing large volumes of information. This speed of translation can be particularly useful in industries such as journalism, e-commerce, and customer service, where quick turnaround times are necessary. However, the speed of machine translation does not always equate to high-quality translations, and post-editing or review by a human translator may still be necessary. But it saves time.
2] Availability :
Machine translation is available 24/7 and can be accessed from anywhere with an internet connection, making it highly accessible. This allows for quick and easy translation of text at any time, which can be particularly useful in industries such as customer service, where prompt response times are necessary. Additionally, machine translation eliminates the need to wait for a human translator to be available, further increasing its accessibility
3] Cost-effectiveness:
Machine translation is often more cost-effective than human translation, especially for large volumes of text. This can be attributed to the speed and automation of machine translation, which reduces the time and labor required for translation. Additionally, the cost of machine translation software and tools has decreased over time, making it an increasingly affordable option for businesses and individuals. While machine translation may not always produce high-quality translations, it can still provide a quick and cost-effective solution for translating basic or simple text.
4] Consistency:
Machine translation provides consistent translations as it follows set rules and guidelines. This consistency can be particularly useful for translating repetitive or standardized text, such as product descriptions or legal documents. Additionally, machine translation can ensure that terminology and branding are consistently translated, improving brand image and consistency across multilingual materials. However, consistency can also lead to limitations in the ability of machine translation to capture the nuances and complexity of human language and communication.
5] Accessibility:
Machine translation can break down language barriers and provide access to information in different languages. This can greatly enhance communication and information exchange, making information and resources more widely available. For example, machine translation can provide access to news, literature, and websites in different languages, increasing cultural understanding and global connectivity. Additionally, machine translation can make it easier for individuals who are not proficient in a particular language to communicate and access information in that.
Conclusion -
Thus to conclude we can say that machine translation has its own benefits and also it is very popular in present time, but it has its limitations too. The system is getting better day by day and the level of translation quality is also increasing but still it is behind in many ways. On one hand machine translation is fast, saves time, is available day and night, it is not very costly and it is biased free at one point but has a lack of emotional understanding that causes problems many times. Human translation is more trustworthy than machine translation but it is very costly and time consuming. So at last both Machine translation and Human translation has its own advantages and disadvantages.
Citation
Bassnett, S., & Lefevere, A. (Eds.). (2015). The Oxford Handbook of Translation Studies. Oxford University Press.
Dhumketu. (1926). Tankha Mandal 1 (Vol. One). Gurjar Grantharatna Karyalaya.
FreeCodeCamp (2021, January 13). A History of Machine Translation: From the Cold War to Deep Learning. FreeCodeCamp. https://www.freecodecamp.org/news/a-history-of-machine-translation-from-the-cold-war-to-deep-learning-f1d335ce8b5/
Lingual Consultancy (2021, June 7). Machine Translation: Advantages and Disadvantages. Lingual Consultancy. https://lingualconsultancy.com/blog/machine-translation-advantages-and-disadvantages
Malmkjær, K. (2015). Translation: An Essential Guide. Routledge.
Munday, J. (Ed.). (2012). Translation: A Multidisciplinary Approach. Palgrave Macmillan.
No comments:
Post a Comment