Man vs Machine: A Comparison of Linguistic, Cultural, and Stylistic Levels in Literary Translation

Received: November 21, 2020 Accepted: February 14, 2021 Volume: 4 Issue: 2 DOI: 10.32996/ijllt.2021.4.2.10 This study aims to remark the differences between human translation (HT) and machine translation (MT) on linguistic, cultural, and stylistic levels when translating English literary texts into Arabic. To accomplish the goal of this study, a comparison between the Arabic HT and MT of Saki’s (1914) short story ‘The Open Window’ is conducted. The study focuses on comparing the two translations (HT and MT) on linguistic, cultural, and stylistic levels to identify the differences between HT and MT in translating literary texts. Throughout this comparison, it is found out that both HT and MT have their advantages and disadvantages on different levels. It has also been found out that MT is unable to identify cultural items and consequently mistranslate them. It is, therefore, concluded that MT can work proficiently on certain levels besides the intervention of the human mind. The findings of this study provide translators using MT with a clear vision on the points of strength and weaknesses in translating literary texts. KEYWORDS

attempts to answer the following question: what are the differences (if any) between HT and MT when translating English literary texts into Arabic? In addition, two other questions are also dealt with in this study, namely: How effective MT is compared to HT? And to what extent can MT replace HT?
To answer the above-mentioned questions, a contrastive analysis between the Arabic HT and MT of Saki's (1914) short story 'The Open Window' is conducted. This selection is based on the assumption that prose translation in general and short story translation in particular received less interest of studies compared to poetry translation. Bassnett (2014, p. 119) states that "Although a large body of work debating the issues that surround the translation of poetry, far less time has been spent studying specific problems of translating literary prose". Therefore, this study has been conducted to study some problems of translating literary prose (short story). The study comprises the analysis of both HT and MT of the selected corpus one three main areas namely; linguistics, culture, and style. These three areas are briefly discussed below:

Linguistics
Linguistics is the science of language that mainly deals with all the aspects of the language. Catford (1965: 1) defined translation as "an operation performed on languages". Based on this definition, translation can never be separated from linguistics as they are two interrelated disciplines. Translation theories and practices are set based on the theory of how languages work, which is basically what linguistics is about (Kolawole, 2013).
There are many rules in language that show you how to write specific words or put words together to make sense. Sometimes there is more than one meaning of a word, so both HT and MT have to decide which one is appropriate to be used in that context. By that, both of them will not lose the intended meaning of the ST. It is, therefore, assumed that MT is not as capable as HT when choosing words based on their specific contexts.
purpose and meaning; otherwise, they might miss the ST writer's intention. In this case, it would be possible to see how different HT and MT are when producing the stylistic features in the TL.

Review of Literature
For the last two decades with the vast development of technology and artificial intelligence, MT has drawn the attention of translators and translation technology experts. Therefore, many studies on the usefulness of MT have been conducted to compare between HT and MT. For instance, a study comparing HT with MT has been conducted by Ahrenberg (2017). The study was highly concerned with investigating how useful MT tools are in producing an adequate translation. It also assessed the difference between HT and MT. The study was conducted by taking an English opinion article entitled 'Why I Left my Liberal London Tribe', written by David Goodhart and published by the Financial Times on March 17, 2017 and its two Swedish translations (HT and MT). The HT, which was published in the June issue of Axess, Swedish magazine, did not involve any use of tools other than dictionaries. The MT was produced by using 'Google Translate' in June 2017. It turned out that HT was longer than MT in terms of the number of words and length of sentences. MT had some changes in words, but the majority was correctly rendered with some mishaps in word order. In contrast, HT was not only concerned about the grammatical errors but the style of the source text, too. The final conclusion of the study was that MT can produce a text that is more similar to the ST than HT. However, it does not mean that it is on a par with the proficiency level of HT.
Another study about MT has been conducted by Zheng (2015) to assess the efficiency and feasibility of MT. Basically, a Chinese song entitled '我的根在草原' (My Root's in the Grassland), and sung by Dedema, a popular Mangolian singer was chosen as a case study. The lyrics were translated into English using 12 translation tools, such as Google Translate, Lingoes Translator, Baidu, etc. Zheng assumed that MT of the lyric song was not fully clear because of the excessive occurrence of translation mistakes. Therefore, the translation needed to be refined and improved. Furthermore, Zheng noticed that MT missed the style and aesthetic features of the literary language that a song might carry adding that MT might be applicable in translating non-literary texts. Eventually, it turned out that MT was not highly appreciated to be used in translating that song because of the extensive use of expressive words and figurative language. Zheng highly recommended that the outcome MT needs to be improved by human mind for better results.
A study on MT was conducted by Stiegelbauer (2012). Throughout her study, Stiegelbauer compared between HT and MT discussing the limitations of computer or MT. She argued that computers do not translate as efficiently as humans. In order to support her argument, she selected several English texts taken from Dan Brown's (Da Vinci Code) and Romanian texts from Marin Sorescu's writings, a Romanian writer of prose and poetry. She chose to translate these texts from English into Romanian and vice versa using Google Translate to check the proficiency of MT and see if it needs human intervention or not. She realised that the MT of the prose resulted in various mistakes under linguistic and stylistic levels. The order of the words and the use of verb tenses did not logically make sense. Moreover, some words were missed in the translations because MT did not recognise them in the original texts. As for the poems, MT did not successfully deliver the poetic style of poetry. Poems include aesthetic qualities and sound patterns of rhythms and rhymes, but MT ignored these features and kept them away. Similarly, Google Translate did not detect the meaning of words and verb tenses as what happened in the previous case when it came to translating prose. Stiegelbauer concluded that MT has some limitations and works usefully to a certain extent. The poems and prose samples she worked on helped her decide that MT is far from being consistent and accurate. She added that MT can save time and process fast compared to HT, but it can never work efficiently without any assistance from the human mind.
Another study comparing HT to MT was conducted by Jebbar (2008). In his study, Jebbar focused on the translation concept and how translators were significant throughout the past years. He also highlighted the origin and evolution of MT. Jabbar, then, selected a text for comparison. The English source text was taken from 'My Son the Fanatic' by Hanif Kureishi and translated into Arabic by both HT and MT. The study aimed to focus on both semantic and pragmatic differences in both translations. It was stated that a translator must have the knowledge and understanding of the intended meaning before translating any text. MT is said to have emerged in the 1950s, more specifically during the Cold War between the US and the Soviet Union. By this way, it would save time and be less costly to translate large amounts of papers. Jebbar (2008, p. 5) stated that "replacing HT with MT would more than likely end in failure". One of his reasons was that it is sort of difficult for machines to keep the same effect of the ST in the TL. Human translators are the only ones who understand cultural, linguistic, and semantic differences to produce an equivalent TT. Jebbar (2008) added that human translators are more efficient when it comes to reading words with different vowels or diacritic marks. Sometimes, MT would not be able to provide an equivalent term, so it is left as it is (transliterated) in the TT.
The studies listed above have mainly dealt with general comparisons between HT and MT. The way the comparisons were

Findings
The findings are divided into the three levels: linguistic, cultural, and stylistic as follows:

Linguistic Level
1) MT adopted literal translation in translating the story. As a result, some words made no sense and led to vague sentences. Overall, HT reads more natural Arabic in terms of the smoothness of the language. That is due to the translator's full understanding of the ST and ability to produce smooth meaningful sentences, in addition to the translation procedures and techniques he opted for in translating the story. 2) MT picked words randomly, disregarding the original context of the ST. On the contrary, HT seems natural because of the suitable choice of words that fit into their contexts. 3) Various errors have been discovered in MT under the grammatical level, whereas HT has fewer grammatical mistakes.

Cultural Level
1) HT is closer to the target language (Arabic) than MT, as there are some cultural words and expressions that were localized in the TT. 2) HT was more concerned with cultural items translating them freely, while MT was based essentially on the literal translation. Therefore, some cultural items appeared in the ST have been neglected, translated literally, and/or transliterated so their meanings were lost in the TT.

Stylistic Level
1) MT followed the Arabic storytelling style, while HT did not use the narrative voice, but kept it as it was in the ST.
2) HT reads more literary because of the stylistic devices used, such as metaphors and literary words.

‫أخت‬ ‫ابنة‬
3) MT was not consistent in translating certain words, i.e. some words appeared several times in the ST were translated differently by MT each time. 4) Sentences in MT were not connected cohesively as in HT, which sounded weird in Arabic.

Linguistic level
In some cases, MT produced was meaningless and unclear in terms of word selection. As a result of this, many sentences lost their meanings in the TT. However, HT was produced in a smooth language where the translator translated the text freely using literary devices, so the reader feels as it is an original text. If we take "She broke off with a little shudder" as an example, MT mistranslated it into ‫االرتجاف‬ ‫من‬ ‫بقليل‬ ‫,انفصلت‬ while HT translated it correctly into ‫خفيفة‬ ‫برجفة‬ ‫الكالم‬ ‫عن‬ ‫.وتوقفت‬ According to Macmillan Dictionary (2020), the phrasal verb 'break off' means "to stop doing something, especially in speaking". Therefore, HT here succeeded in delivering the same meaning of the ST, while MT's suggestion is not suitable at all, and instead, it refers to a totally different thing. The language of MT in the same example is not natural and makes no sense as it is literally translated. Another example of mistranslation is when MT translated "a fine mess" into ‫جيدة‬ ‫ى‬ ‫.فوض‬ In that context, the writer meant an enormous mess by saying "a fine mess".

‫افتقد‬ ‫هنا‬ ‫صوت‬ ‫بصورة‬ ‫متلعثما‬ ‫وغدا‬ ‫الواثقة‬ ‫ته‬ ‫ر‬
‫نث‬ ‫الطفل‬ ‫.أدمية‬ Moreover, MT contained many grammatical errors. For instance, "Frampton grabbed wildly at his stick and hat" was rendered into ‫وقبعته‬ ‫عصاه‬ ‫ي‬ ‫ى‬ ‫ف‬ ‫بعنف‬ ‫امتون‬ ‫فر‬ ‫.أمسك‬ MT indeed followed the verbal style that reads more Arabic, but the grammatical mistake represented in adding the preposition ‫ي‬ ‫ى‬ ‫ف‬ affected the entire meaning. The HT ٍ ‫بعنف‬ ‫امتون‬ ‫فر‬ ‫خطف‬ ‫بإهتياج‬ ‫وقبعته‬ ‫عصاه‬ seemed more acceptable and grammatically correct. Sometimes a change in a letter or a word can affect and alter the complete meaning.

Cultural level
The examples presented previously under the cultural level are remarkable evidence of the poor translation quality MT can produce. The cultural terms in MT were mainly translated based on the word-for-word translation. This resulted in translating ST cultural terms into words that are not suitable for their specific context. In other words, the words chosen in the TT do not carry the same meaning of those in the ST. It is obvious that MT which adopted literal translation was incapable of understanding the context in which culture is represented. Although literal translation is an efficient procedure to use in translation, it sometimes does not work accurately, especially when it comes to translating literary works.
If we take "French Window" as an example to support the previous statement, HT translated it into ‫ي‬ ‫ى‬ ‫الخلف‬ ‫,الباب‬ while MT translated it literally as ‫فرنسية‬ ‫.نافذة‬ In this case, it is clear that MT did not consider the importance of logical and contextual meaning. Generally speaking, people do not enter or get out of their houses from a window, but a door. Moreover; the Cambridge Dictionary (2020) defines French window as "a pair of glass doors, usually opening from the back of a house into its garden". In this example, HT produced better translation than MT in terms of giving a logical and reliable translation that conveys the exact meaning. Accordingly, MT seemed unable to translate some cultural words, consider the word "mackintosh". It is transliterated by MT in Arabic as ‫.ماكنتوش‬ In contrast, HT looked for the appropriate meaning translating it into ‫مطر‬ ‫,معطف‬ which is a good selection as 'mackintosh' means "raincoat, especially one made from a particular kind of waterproof cloth" (Collins Dictionary).
Another example is "Spaniel" which was translated into ‫كلب‬ by the HT, whereas MT translated it into ‫.اللوح‬ In fact, 'Spaniel' is a type of dogs, thus the HT transferred it into ‫كلب‬ which is an accurate translation as it preserves the meaning and presents the right picture of "Spaniel". However, MT's suggestion makes no sense and does not suit the context. All in all, it can be noticed that HT is more reliable and capable of translating cultural items where the translator can consult others and search for the right equivalent revealing any ambiguity with such cultural terms and contexts by adopting appropriate translation techniques.

Stylistic level
The examples provided for the stylistic level were meant to show how competent HT is compared to MT regarding the stylistic aspects. Obviously, HT takes TT culture, audience and text type into consideration when translating any text, because that can help specify the text style. Taking style into account, it is noticed that MT followed the Arabic storytelling way more than HT with the use of the narrative voice. For example, "A most extraordinary man, a Mr. Nuttel, said Mrs. Sappleton" has been translated by HT into ‫سابلتون"‬ ‫السيدة‬ ‫قالت‬ ‫نتل،‬ ‫يدىع‬ ‫سيد‬ ‫ابة!‬ ‫الغر‬ ‫شديد‬ ‫,رجل‬ and by MT as ‫"أ‬ ‫سابلتون:‬ ‫السيدة‬ ‫قالت‬ ‫هو‬ ‫رجل‬ ‫عظم‬ ‫نوتيل"‬ ‫.السيد‬ As we can see, both translations are acceptable, but the MT stuck to the original way of Arabic storytelling style, i.e. fronting the verb and using a verbal sentence. HT, in contrast, followed the same structure of the ST, which is fine as long as the meaning is there. If we look at the cohesiveness of the sentences "an absence of mental excitement, and avoidance of anything in the nature of violent physical exercise" which was translated by HT into ‫العص‬ ‫اإلثارة‬ ‫عن‬ ‫الكامل‬ ‫واالبتعاد‬ ‫أي‬ ‫وتجنب‬ ‫بية‬ ‫العنيفة‬ ‫البدنية‬ ‫الرياضة‬ ‫طابع‬ ‫له‬ ‫ء‬ ‫ي‬ ‫ر‬ ‫,ش‬ and into ‫العنيفة‬ ‫البدنية‬ ‫التمارين‬ ‫طبيعة‬ ‫ي‬ ‫ى‬ ‫ف‬ ‫ء‬ ‫ي‬ ‫ر‬ ‫ش‬ ‫أي‬ ‫وتجنب‬ ‫العقلية‬ ‫اإلثارة‬ ‫وغياب‬ by MT. HT is more meaningful because of the way the translator put words together and structured the sentence using a pronoun for back reference in ‫.له‬ On the other hand, in the MT, the choice of words made the sentence less cohesive and little confusing for the readers to understand. In addition, MT went for the literal translation, which was not a good choice in translating the word 'appearance' into ‫.مظهرها‬ HT, in contrast, translated it into ‫مظهرها‬ ‫.هندمة‬ Both translations were acceptable, but the addition of ‫هندمة‬ in the HT added a stylistic flavour to the text making it more literary rather than literal. In translating "On the matter of diet they are not so much in agreement, he continued", the HT translated it into ‫"أما‬ ‫حديثه‬ ‫أكمل‬ ‫ثم‬ ‫يتعلق‬ ‫فيما‬ ‫ذاته"‬ ‫بالقدر‬ ‫يتفقوا‬ ‫لم‬ ‫فإنهم‬ ‫بالغذاء‬ , while the MT translated it into ‫ا"‬ ‫ر‬ ‫كثث‬ ‫يتفقون‬ ‫ال‬ ‫فإنهم‬ ، ‫ي‬ ‫الغذائ‬ ‫بالنظام‬ ‫يتعلق‬ ‫"فيما‬ ‫.وتابع‬ It is clear that MT chose a more literal approach to translate the sentence, while HT went with how it would be written in Arabic. That is not to say that this is right or wrong, but rather which one sounds more natural and accurate when reading it in the TL. Moreover, it has been noticed that most sentences produced by MT were verbal, which is the typical style of writing in Arabic, as Arabic sentences usually begin with a verb. However, HT neglected the verbal style producing more nominal sentences as a result of following the ST style. In fact, this is quite unusual for an Arabic literary text. There are also some problems with consistency of word choice in translating the word niece into the TL. In the HT, it was translated into ‫األخ‬ ‫بنت‬ for both times it appeared in the ST. However, in the MT, it was translated into ‫أخت‬ ‫ابنة‬ the first time and ‫األخ‬ ‫ابنة‬ in the second. Sometimes a change in the choice of word is fine as long as the meaning is the same, but in this case, the change in words changes the meaning and confuses the reader. That also affects the consistency and coherence of the language which impacts the general style of the text.

Conclusion
Having compared between HT ('Google Translate' in this study) and MT on linguistic, cultural, and stylistic levels, it was found out that both HT and MT were not perfect and each had its own advantages and disadvantages. MT followed the Arabic storytelling style, but had some issues with cohesion, coherence, and the structure of the sentences. In contrast, HT was more cohesive, coherent, and sentence structure was more natural apart from being affected by the ST style in some cases. HT payed more attention to the cultural items and consequently produced more culturally acceptable translation. In contrast, MT ignored the cultural aspects in translating some items. Moreover, when it comes to word choice, HT adopted a more literary approach using some stylistic devices, e.g. collocations. Therefore, it sounded smoother and more natural Arabic. In answering the research questions stated earlier, the findings of this study have shown the differences between HT and MT and how effective MT is compared to HT. They have also shown that MT cannot replace HT completely but both can work together, i.e. MT can be used but with the intervention of HT/human mind In conclusion, MT ('Google Translate' in this study) would never work professionally without the involvement of human mind. Both HT and MT are not flawless, yet this does not mean that either is bad or one is superior to the other. However, HT was more acceptable than MT in some cases although MT might work more efficiently with other types of texts than literary ones. Both HT and MT can be used together efficiently if done in the right way. If this research conducted again, it would be more efficient to choose various types of texts to see how different HT and MT are in translating various types of texts. This is because this study was limited to a literary text only. Perhaps the findings of HT and MT of other text types would be different than the findings here. It is, therefore, highly recommended that further investigations are needed particularly on the usefulness of MT in translating different types of texts from English into Arabic and/or vice versa.