از هوش انسانی تا هوش مصنوعی: بررسی برگردان‌های فارسی فنون ادبی غزلوارۀ ش. 62 شکسپیر ازمنظر راهکارهای انسانی و ماشینی

ترجمه‌گری ژانر شعر به سایر زبان‌ها همیشه به چند دلیل مسئله‌ساز بوده است. نخست، ترجمه‌گری متون ادبی با دشواری‌های خاص همراه است؛ دوم، متون منظوم نیز به‌سبب کیفیاتی خاص به زبانی دیگر سخت ‌ترجمه‌پذیر جلوه می‌کنند؛ سوم، سبک و سیاق شاعر در ترجمه به «بدیلی» کم‌رنگ از اصل بدل می‌شود؛ چهارم، ترجمه‌گری شعر از رهگذر هوش مصنوعی نیز مسئله‌ساز است. این مقاله می‌کوشد به‌شیوه کیفی‌تبیینی و با کمک ابزارهای هوش انسانی و ماشینی برخی از این مسائل را در پرتو برگردان‌های فارسی فنون ادبی غزل 62 شکسپیر بررسی کند. هدف اصلی توجه به چالش‌های پیش‌رو در ترجمه فنون ادبی است. فرزاد ترجمه‌ای روان؛ موقر، ترجمه‌ای به‌نسبت آزادانه‌تر و شفا، ترجمه‌ای وفادار به مضمون و طبیب‌زاده، ترجمه‌ای بسنده و پذیرفته از غزل‌واره ارائه کرده‌اند. از میان ابزارهای هوش مصنوعی، گوگل ترجمه‌ای تقریباً نزدیک به متن اصلی و تحت‌اللفظی؛ بات تلگرامی، ترجمه‌ای کامل‌تر و روان‌تر و اوپن چت جی‌پی‌تی، ترجمه‌ای دقیق‌تر و  نزدیک‌تر به هوش انسانی از غزل‌واره ارائه کرده‌اند. به نظر می‌رسد ترجمه ادبی به‌ویژه شعر از رهگذر ترجمه ادبی نیاز به خلاقیت هوش انسانی دارد تا حضور «بدیل» خود را به «اصل» نزدیک‌تر کند.

from human to artificial intelligence:an examination of persian translations of shakespeare’s literary devices in sonnet 62 regarding human and machine strategies

translating poetry has always been a challenging issue for some reasons. firstly, it is accompanied by cultural challenges. secondly, it is considered untranslatable due to its unique qualities. thirdly, the poet’s style, if not fully captured in translation, becomes a pale substitute for the original. fourthly, issues related to artificial intelligence (ai) also arise. this article tried to analyze some of these issues in the light of the persian translations of the literary techniques of shakespeare’s sonnet 62 in a qualitative-explanatory way and with the help of human and machine intelligence tools. the main goal was to pay attention to the upcoming challenges in the translation of literary techniques. farzad provided a fluent translation, moghadam provided a relatively free translation, shafa provided a translation faithful to the content, and tabibzadeh provided an accurate and accepted translation of the sonnet. among the ai tools, google translate provided a literal translation close to the original text, telegram bot provided a more complete and fluent translation, and openai gpt provided a more accurate and human-like translation of the sonnet. it seemed that literary translation, especially poetry, required human intelligence creativity in order to bring its alternative presence closer to the original. keywords: human intelligence, artificial intelligence (ai), sonnet, accuracy, acceptability. introductionit has been over a hundred years since shakespeare’s works have been translated into persian. during this time, shakespeare’s sonnets have also been translated and published in periodicals. translators, such as masoud farzad, majid movaqqar, shoja al-din shafa, taghi tafazzoli, behnam moghadam, amrullah abjadiyan, and omid tabibzadeh, are among those, who have translated shakespeare’s sonnets into persian. besides using human intelligence, shakespeare’s sonnets can also be translated into persian through artificial intelligence (ai), which would be challenging and interesting in its own way. in this article, sonnet 62, which had been translated into persian by most of the aforementioned translators and ai tools, was examined from the perspectives of the proposed strategies for translating poetry. the discussion of translating the genre of poetry into other languages has always been raised alongside issues related to the original works themselves. in translating conventional texts generally and literary texts, especially poetry, we do not merely deal with translating linguistic items from one language to another. since literary texts are considered an inseparable part of the cultures of nations and carry specific semantic loads, translation of these texts also involves particular difficulties and problems due to cultural issues. in the process of translating an english sonnet into a persian ghazal, some issues arise that need attention. can a sonnet be translated into a sonnet while preserving its formal and structural features or undergoing slight modifications and can it be translated into the form of a persian ghazal? to what extent is it possible to transfer cultural elements in addition to poetic ones in the translation process? to what extent can ai succeed in conveying the poet’s most intimate states and feelings into another language, considering that ai, due to its structural nature, lacks the emotional intelligence that characterizes human intelligence? materials & methodsin the common and traditional categorization that is theme-based rather than form-based, shakespeare’s sonnets are divided into three general categories: sonnets 1-126 address the fair youth; sonnets 127-152 deal with the dark lady; and sonnets 153 and 154 are known as the anacreontic sonnets (anacreon 583 bc-485 bc), which have romantic themes (abjadiayn, 2008; tabibzadeh, 2017). sonnet 62, along with sonnets 59, 60, and 65, deals with the relationship between time and beauty (abjadiyan, 2008, p. 205). according to tabibzadeh, it is the first of the two sonnets about the poet’s old age (p. 184). in this sonnet, shakespeare first speaks of the sin of self-love, which has taken over his eyes and his entire being, rooting deeply in his heart (first quatrain). then, in the second quatrain, he talks about his beauty, which is unparalleled in the world. however, in the third quatrain, he says that he sees himself as a very worn-out, old, and decrepit man when he looks into a mirror. finally, in the couplet, he attributes all that splendor and beauty he has thought he has had to the fair youth and imagines his own old age to be beautiful in the light of the fair youth’s youthful beauty. this sonnet was first translated by masoud farzad. then, majid movaqqar translated it. the third translation was done by shafa. the fourth translation, a prose translation, was done by tafazzoli. then, behnam provided a rhythmic and rhymed translation of this sonnet. finally, tabibzadeh (2017) provided a sufficiently acceptable, though unrhymed and unmetered, translation of it. additionally, the author translated this same sonnet into persian using ai tools, such as google translate, and, more recently, chatgpt. research findingsthe translations of shakespeare’s sonnet 62 by various human translators and ai tools revealed interesting insights into the strengths and limitations of each approach. the human translators, including farzad, movaqqar, shafa, tafazzoli, moghadam, and tabibzadeh, had produced translations with varying degrees of fluency, faithfulness, and literary quality. farzad and movaqqar had prioritized fluency over fidelity, resulting in translations that had strayed from the original text. in contrast, shafa and tafazzoli had focused more on faithfulness, with tafazzoli’s prose translation staying very close to the original content. moghadam had offered a relatively free but rhythmic and rhymed translation, while tabibzadeh had presented a sufficiently accurate and acceptable unrhymed and unmetered version. when examining the translations produced by ai tools, some interesting patterns emerged. google translate, despite advancements in the field, generated a largely literal and close-to-the-original translation, focusing more on primary word meanings rather than secondary connotative meanings. on the other hand, the telegram bot and openai’s gpt produced more accurate and human-like translations. the telegram bot’s translation was more complete and fluent than google’s but less comprehensive and precise than openai’s. notably, openai’s translation was, in some cases, even more literary than human translations. still, the assumption that ai cannot effectively translate literary texts due to the presence of emotion and feeling could be challenged. openai’s more complex algorithms