British Journal of Research Open Access

Enhancing Translation Accuracy: Human and AI Approaches to Albanian-English Literature

Lorena Kapxhiu^* and Eglantina Gisht Department of English, Tirana University, Tirana, Albania
^*Correspondence: Lorena Kapxhiu, Department of English, Tirana University, Tirana, Albania, Email:

Received: 21-Aug-2024, Manuscript No. IPBJR-24-21214; Editor assigned: 23-Aug-2024, Pre QC No. IPBJR-24-21214 (PQ); , QC No. IPBJR-24-21214; Revised: 14-Jan-2024, Manuscript No. IPBJR-24-21214 (R); , DOI: 10.35841/2394-3718-12.01.127

Introduction

The book "Chronicle in Stone" by Ismail Kadare poses significant translation challenges, especially from Albanian to English, due to its intricate linguistic structure and cultural nuances. Human translators have historically navigated these challenges through their deep understanding of cultural subtleties and idiomatic expressions. However, the rise of AI translation tools such as DeepL and Google Translate has introduced new possibilities and complexities in translation [1].

Recent advancements in computer science and linguistics, particularly in machine learning and big data technologies, have revolutionised translation. Neural network-powered machine translation systems can analyse vast amounts of data to identify translation patterns, making AI translation an area of intense research interest. AI's ability to process large volumes of information quickly is advantageous in today's fast-paced information era.

Despite these technological strides, AI translation presents challenges such as the need for extensive datasets for training machine learning models. Translators must nowadapt to utilizing translation technology effectively and understand the limitations of AI translations to remain competitive. Complexities in translation, like ambiguous terms such as "bari" in Albanian, pose difficulties for AI systems lacking contextual awareness, whereas human translators excel in interpreting such nuances [2].

This study compares the human and AI translation processes, highlighting their strengths and weaknesses. While AI offers speed and efficiency, human translation retains the ability to effectively convey emotional depth and contextual nuances effectively. The future of translation lies in the collaboration between human expertise and AI capabilities, which requires ongoing adaptation to technological advancements [3].

Literature Review

This study involves carrying out a comparative comparison between human translations and artificial intelligence translations of selected sections extracted from Kadare's Chronicle in Stone. Multiple AI translation systems, including Lingvanex, iTranslate, Google Translate, and others, were employed to produce translations of specific excerpts. Subsequently, these translations were juxtaposed with human translations conducted by proficient translators such as Arshi Pipa and David Bellos.

The process of human translation goes beyond just converting words from one language to another; it entails exploring the cultural subtleties and fundamental meaning of the text. Translators utilise methods like paraphrase, localisation, and transcreation to accurately communicate intricate concepts and idiomatic idioms. Ethical considerations, such as ensuring the preservation of the author's original intention and being culturally sensitive, are of utmost importance in the field of human translation.

The AI translation process utilises neural networks to analyse extensive data sets in order to detect translation patterns. Translating from Albanian to English presents distinct difficulties, mostly because of the characteristics of the Albanian language and its divergences from other languages. The difficulties are increased by the restricted accessibility of translation tools specifically tailored for this combination of languages.

Research Questions

• What are the advantages and disadvantages of human translation as opposed to AI translation when it comes to translating intricate literary materials such as Chronicle in Stone?
• How do AI translation technologies address the cultural and linguistic complexities seen in Kadare's work?
• Can AI translation techniques accurately convey the emotional nuances and contextual intricacies of the original text?
• Is it more advantageous to utilise AI translation only or is it preferable to employ it as an aid to expedite human translation?

Discussions

Human Translation: Methodology and Benefits

Human translation process: Human translation is a nuanced and intricate process that transcends mere linguistic conversion, delving into the cultural nuances and essence of a text [4]. It encompasses various techniques, such as sight translation, dubbing, sequential, and simultaneous interpreting, each tailored to specific linguistic and situational demands. Translators employ methods such as paraphrasing, localization, and transcreation to convey complex concepts and idiomatic expressions effectively.

Ethical considerations play a pivotal role in human translation, requiring translators to uphold the author's intent, maintain confidentiality, and adhere to professional standards of accuracy and neutrality [5]. This ethical aspect extends to cultural sensitivity, political correctness, and inclusivity in translation practice.

The translation of Ismail Kadare's "Chronicle in Stone" serves as a case study highlighting the challenges and intricacies of translating a culturally rich and deeply embedded work. Different translation approaches, exemplified by Arshi Pipa's "A Chronicle in Stone" and David Bellos's "Chronicle in Stone," illustrate the complexities faced in capturing the essence of Albanian culture within the translated text (TRANSLATORIUM).

Bellos's translation, known for its multilayered interpretations, explores the depth of Kadare's narrative, acknowledging the nuances of vocabulary and references that enrich the story (TRANSLATORIUM). However, this endeavor requires a comprehensive understanding of Albanian society, history, and language to accurately convey cultural subtleties and maintain narrative authenticity [6].

The challenges of human translation are compounded by the need to preserve literary techniques such as allegory, symbolism, and metaphorical language inherent in Kadare's storytelling style [7]. While AI translation technologies offer speed and efficiency, they often struggle to capture the nuanced cultural references and literary intricacies found in works like "Chronicle in Stone". Thus, a collaborative approach integrating AI tools with human expertise can enhance translation efficiency, while ensuring the integrity and emotional impact of the original text across diverse audiences and languages.

Benefits of Human Translation

Human translation has several benefits compared to AI translation owing to the complex interaction of language, cultural, and environmental factors. Human translation offers a significant advantage in capturing the sensual, linguistic, and cultural nuances of a text. In Ismail Kadare's "Chronicle in Stone," a skilled translator can comprehend the intricate details of Albanian culture, historical allusions, and language peculiarities that enhance the tale. This degree of comprehension encompasses the ability to capture the distinctive style and subject matter of human cognition accurately, resulting in a more refined and precise translation that effectively conveys the original text's profoundness and intricacy [8].

In addition, human translation relies on intuition, a cognitive ability that enables translators to quickly detect the alignment or divergence of concepts within a text. Translators can use their intuitive understanding to provide translations that are grammatically precise, idiomatically eloquent, and smooth. In "Chronicle in Stone," having this intuition is essential for understanding Kadare's allegorical and metaphorical language, maintaining the poetic nature of the text, and assuring clarity and consistency in the translated version.

An essential element of human translation involves considering the target reader's traits and emotional connection to the original material. When translating "Chronicle in Stone" into English, a human translator would consider the cultural context and preferences of Englishspeaking readers. They modified the translation to appeal to their sensibilities while remaining faithful to Kadare's original objectives. Engaging and resonating with varied audiences requires significant cultural and interpretive alteration, which is crucial for overcoming language and cultural barriers in translations [9].

Additionally, experienced human translators possess a vast amount of knowledge and skills that they apply to the translation process. They are proficient at meeting deadlines, adept at handling a wide range of subject subjects and writing styles, and possess extensive knowledge of rhetorical traditions, literary forms, and linguistic subtleties. A proficient human translator would adeptly handle the amalgamation of historical realism, allegory, and folklore in "Chronicle in Stone," guaranteeing the preservation and successful transmission of these aspects in the target language. Human translators possess the knowledge and skills necessary to remove uncertainties, preserve the original author's writing style, and produce translations of higher quality than translations made by artificial intelligence.

However, AI translation is deficient in the qualities of empathy, intuition, and cultural comprehension that are naturally present in human translations. Although AI systems may generate translations that comply with grammatical rules, they frequently struggle to convey the true meaning, writing style, and emotional nuances of an original text. This constraint is especially apparent in literary works such as "Chronicle in Stone," where cultural intricacies, historical allusions, and language nuances have a substantial impact on communicating the author's message and narrative purpose. Human translators are essential for maintaining the integrity and depth of literary writing, guaranteeing that they genuinely connect with readers of other languages and cultures.

AI Translation: Methodology and Benefits

AI translation process: Translating from Albanian to English presents unique challenges because of the nature of the Albanian language and its differences from other languages. One significant limitation is the availability of translation apps that are specifically designed for this language pair. Unlike the more widely spoken languages with numerous translation tools available, Albanian lacks comprehensive support in many AI translation services. This limitation stems from the unique characteristics of the Albanian language, including its alphabet with special characters such as, ‘ë' ‘ë,' 'ç,' 'gj,', 'th' ‘th,' 'xh,' and 'sh,' each considered a single letter in Albanian but not as distinct in many other languages. These letters are crucial for accurate translation, as they carry specific meanings and nuances in Albania, which may not directly translate into English.

When examining literary translations, particularly in the case of Ismail Kadare's "Chronicle in Stone," these difficulties become much more evident. The work is characterised by its linguistic complexity as well as its abundance of lyrical phrases and Albanian colloquialisms. This poses a significant challenge for both human translators and Machine-Translation (MT) algorithms.

Human translators face the complex challenge of safeguarding the cultural and linguistic nuances of the text, a task that AI systems frequently find difficult because it depends on preestablished rules and patterns. Choosing appropriate algorithms and platforms for English-Albanian translations in the field of AI involves considerable challenges. Owing to the limited backing of Albanians in prominent Machine Translation (MT) systems, developers sometimes have to customise or construct their own translation services. This customization requires complex programming using languages such as Python and frameworks such as Flask to interact with exclusive MT APIs, such as AWS, Google Cloud, and Microsoft Azure.

Furthermore, assessing the efficacy of AI translations in intricate texts requires complex methodology. Experimenting with translation apps and employing repeated techniques is essential for enhancing the precision of AI-generated translations. This process involves rigorous assessment and comparative examination conducted by experienced professionals in both languages. Their expertise allows them to detect and rectify translation faults, inconsistencies, and cultural subtleties that automated systems may fail to recognize.

The feedback loop in AI translation evaluation, where users can provide feedback on the quality of translations, helps refine the algorithms and enhance the overall translation process. However, despite these advancements, the inherent complexities of literary translations from Albanian to English highlight the ongoing challenges faced by AI systems in capturing the richness and depth of human language and expressions.

Benefits of AI translation

Conversely, there are some advantages to machine translation. Machine Translation (MT) software is more affordable compared to Human Translation (HT) software. During time-constrained periods, such as the start of an academic semester or examination season, the speed of Machine Translation (MT) may make it a more favourable option. An additional benefit of Machine Translation (MT) is its high level of word precision, which is particularly important in specialised domains, such aslaw or medicine, where accurate word translations are crucial. The new Machine Translation (MT) program boasts remarkable lexical achievements that are absent in Human Translation (HT) applications because HT does not translate word-for-word. It rapidly translates and provides the user with an understanding of the meaning of a text. Therefore, Machine Translation (MT) can be beneficial throughout the first phase of the translation process. The citation refers to the publication by Santi et al. Thisstudy was conducted by Rezapour et al.

In this study, the novel "Chronicle in Stone" by Ismail Kadare was translated by a bilingual doctor proficient in both languages English and Albanian. These translations were subsequently compared with a human translation derived directly from the original Albanian text. Brutus was used for the artificial intelligence translation procedure. Based on this case study, we determined that human translation is not without value. However, it is important to recognize that human translation has its limitations, and ultimately, the AI translation system's final versions have the potential to supplement human translations. Machine translation can assist individuals in managing the substantial challenges encountered in the translation field, particularly in technical and general translations. However, the search for translation remains incomplete because of the enormous breadth of the literary field and the need for a more comprehensive comprehension of the entire text. Understanding and preserving the essence of a text is crucial for both human and AI translations. Research indicates that the most competent human translator is typically more proficient in literature and has extensive experience in literary works.

Comparative Analysis

Human translations: Arshi Pipa's translation from Albanian to English presents the city as a strange entity emerging abruptly from a valley, like a prehistoric creature clawing its way up the mountain side. This imagery conveys a sense of struggle and primal existence, capturing the essence of the original text. On the other hand, David Bellos’s revised translation in 2007 maintained the core elements while refining language and imagery. The city is depicted as strange, with old stone buildings resembling giant scales, and the tender flesh of life thrivesbeneath the powerful exterior.

AI translations: When comparing these human translations with the AI-generated versions, nuances and details come into play. AI translation captures the core elements of human translation but introduces slight variations in phrasing and imagery. It maintains the concept of a strange city struggling up the mountain, with old stone structures resembling large scales.

Key differences and limitations: Despite the similarities, one notable difference across all AI translations is the omission or incorrect translation of the word "perteritej," which means "life survived and generations stand still" in English. This omission alters the nuance of the original text, as it adds a specific element of struggling, suffering, and survival at the end. This illustrates a limitation of AI translation in capturing nuanced language and imagery from literary texts.

In contrast, human translators such as Arshi Pipa and David Bellos are able to grasp these nuances and convey them effectively.

In their translations. Their expertise and understanding of the cultural and linguistic context allows for a more nuanced and accurate portrayal of the original text (Table 1).

Table 1: Key differences and limitations.

Explanation

Imagery: All translations, whether human or AI, capture the imagery of a prehistoric creature or being, struggling or crawling, and the city's old, stone structures resembling scales. The variations in phrasing do not significantly alter the core imagery.

Language: Human translations are nuanced and evocative, while AI translations vary slightly in phrasing but generally capture the essence of the original text.

Nuance: Human translations capture nuances effectively, while AI translations may lack some nuanced details but generally maintain the essence of the text.

Cultural references: Both human and AI translations do not explicitly grasp specific cultural references from the original text.

Accuracy: Both human and AI translations accurately depict the original text in terms of structure and content.

Word omission: AI translations omit the specific word "perteritej," which alters the nuance of the original text as it's not accurately translated or reflected in tone by any AI app.

This Table 1 summarises the comparison between human and AI translations, highlighting key differences in imagery, language, nuances, cultural references, accuracy, word omission, and limitations.

The original Albanian text from Ismail Kadare's "Chronicle in Stone" offers insight into a conversation between Xhexhua and her grandmother, enriched with cultural subtleties and idiomatic phrases. The word "Luaj vendit," which may be loosely translated as "You're joking," not only expresses astonishment but also has a profound cultural meaning of shock and disbelief, akin to the expression "God forbid that might be true."

Human translation: Arshi Pipa’s translation accurately reflects the fundamental nature and cultural subtleties of the original text. Pipa interprets the phrase "Luaj vendit as" "You're joking!" This accurately portrays grandmothers’ sense of incredulity. The translation preserves the context and essence of the original discourse, while ensuring comprehensibility to English-speaking readers. Pipa's interpretation of "A black beard, indeed, akin to that of a man" maintains the comical yet earnest atmosphere of the scenario, guaranteeing the preservation of the cultural context during the process of translation.

AI translation: The phrase "Luaj vendit" is a literal translation by different apps/websites such as "Play country" "Play the place" that does not accurately convey the colloquial connotation on the contrary it fails to capture the idiomatic meaning. Moreover, we notice that the statement "The girl that Ceco Kaili has lost his beard" is both syntactically and semantically erroneous, highlighting the AI's difficulty in maintaining gender and contextual consistency. Lastly, the phrase translated by AI "Beard and black, we're like men's" demonstrates syntactical and semantic misunderstanding, resulting in an inaccurate representation of the original text's intended meaning.

The profound comprehension of both the source and destination languages by the human translator allows for accurate preservation of the original tone and meaning of the text. Despite their rapidity and effectiveness, AI translations frequently generate clumsy and occasionally inaccurate translations that have the potential to misinform or perplex the reader.

This analogy emphasises the crucial significance of human translators in preserving the integrity and depth of literary writings. Although AI translation systems demonstrate remarkable efficiency in processing vast amounts of text rapidly, they still lack the ability to accurately capture the intricate and culturally significant elements of languages, such as Albanian. This case study highlights the essentiality of human supervision in translation assignments, particularly for literature intricately intertwined with cultural surroundings. The combination of AI technology with human knowledge is vital for improving translation accuracy and maintaining cultural integrity as the technology advances.

Conclusion

Machine translators are commonly assessed based on their correctness, punctuation, and other linguistically measurable features that can be calculated. Nevertheless, these measurements may not comprehensively capture the complexity inherent to human cognitive processes associated with translation. The current research has focused on using qualitative comparison studies to identify the most appropriate translation system for a given scenario. A cruci discovery is that modern machine translators fail to capture the fundamental linguistic features of the source text. Inaccurate word selection and phrasing frequently result in substantial misinterpretations, thus emphasizing the constraints of machine translation systems.

The investigation compares machine-generated and humangenerated translations of Ismail Kadare's "Chronicle in Stone" to highlight these concerns. Kadare's work, which is deeply connected with Albanian culture and history, has distinctive difficulties in translation because of its subtle language and contextual allusions. The focus is mostly on translating a brief dialogue from the novel, which challenges the limitations of machine-translation systems. The investigation revealed many issues with machine-generated translations, which resulted in translations that were, at best, insufficient and, at worst, deceptive and distorted the intended meaning, ambiance, and context of the work.

Machine translators frequently demonstrate a form of "enthusiasm" while translating conversation content, resulting in countless preventable errors. Excessive excitement leads to translations that do not adequately maintain the nuanced and precise nature of the original text, resulting in interpretation. For example, machine translation often fails to accurately translate culturally significant phrases and idiomatic expressions because of the absence of cultural and contextual comprehension possessed by human translators. This inadequacy is most noticeable in languages with extensive cultural backgrounds, such as Albanians.

The ultimate finding of our study is that translations produced by humans are far superior to those made by machines in correctly communicating the message and cultural context of "Chronicle in Stone." Automated translation tools such as Google Translate, Tradduka, iTranslate, and Lingvanex exhibit significant constraints when it comes to interpreting materials written in Albanian. Machine learning and neural networks have made significant progress, but they still face challenges when it comes to translating natural languages, especially for less widely spoken languages. This conflict underlines the persistent difficulties in attaining precise and contextually suitable translations through machine translation.

Despite almost 50 years of progress, natural-language translation continues to face substantial obstacles. Although there have been advancements in machine translation, the overall quality is still poor, especially when translating between European languages. The intricacies of human language, which are strongly influenced by cultural and contextual factors, remain beyond the scope of machine translators. This highlights the need for ongoing studies and advancements in this subject.

Nevertheless, it is imperative to recognize that AI may greatly benefit human translators. AI translation systems are highly efficient tools capable of rapidly translating extensive amounts of material. This enables human translators to concentrate on adapting and perfecting translations to suit specific situations. The adoption of this cooperative method significantly reduces time, allowing translators to enhance their efficiency and effectively manage larger workloads. AIgenerated translations cannot be released without human modification because they may include mistakes. Human input is necessary to ensure accuracy and cultural appropriateness. Furthermore, when human translators increasingly utilize AI, they enhance the AI's proficiency by introducing a fresh lexicon and contextual understanding, thereby enhancing its potential for future translation. This mutually beneficial partnership increases the entire translation process by utilising the advantages of both human experience and AI technology.

References

1. Forcada ML (2017) Making sense of neural machine translation. Transl Spaces. 6(2):291-305.

   [Crossref] [Google Scholar]

2. Garcia I (2015) Machine translation and professional translation. In: O’Hagan M (ed) 175-192.

   [Google Scholar]

3. Goodfellow I, Bengio Y, Courville A (2016) Deep Learning. MIT Press.

   [Google Scholar]

4. House J (2015) Translation Quality Assessment: Past and Present.

   [Crossref] [Google Scholar]

5. Koehn P (2020) Neural Machine Translation. Cambridge University Press.

   [Google Scholar]

6. Pym A (2013) Translation Skill-sets in a Machine-translation Age. Meta. 58(3):487-503.

   [Google Scholar]

7. Snell-Hornby M (2006) The Turns of Translation Studies: New Paradigms or Shifting Viewpoints? John Benjamins Publishing.

   [Google Scholar]

8. Venuti L (2017) The Translator's Invisibility: A History of Translation.

   [Crossref] [Google Scholar]

9. Forga-Berinson L (2019) The Craft of Translation. Routledge, New York.