Translating Poetry with AI: A Comparative Study

 Hello Everyone! 

This blog is based on a thinking activity assigned by Prof. Dilip Barad as part of the Translation Studies Activity Worksheet: Using Gen AI for Translating Poems. The purpose of this activity is to critically examine how Gen AI tools such as ChatGPT and Gemini translate poetry across languages and cultures. And I done this task by help of AI ChatGPT. By engaging with poems from Urdu, English, Gujarati, and Hindi literary traditions, the blog reflects on translation not merely as a linguistic transfer but as a complex cultural, emotional, and semiotic process. Through step-by-step translation and comparative analysis, the activity encourages a deeper understanding of theoretical frameworks in translation studies and highlights the possibilities as well as limitations of AI in capturing the essence of poetry. And for more information on this activity visit Barad Sir's worksheet. And this second one on Guidelines for Using Generative AI in Translation Studies. 

Poem: “Mujh Se Pehli Si Mohabbat Mere Mehboob Na Maang”

Poet: Faiz Ahmed Faiz

Original Language: Urdu

میں نے سمجھا تھا کہ تو ہے تو درخشاں ہے حیات

تیرا غم ہے تو غم دہر کا جھگڑا کیا ہے

تیری صورت سے ہے عالم میں بہاروں کو ثبات

تیری آنکھوں کے سوا دنیا میں رکھا کیا ہے

تو جو مل جائے تو تقدیر نگوں ہو جائے

یوں نہ تھا میں نے فقط چاہا تھا یوں ہو جائے

اور بھی دکھ ہیں زمانے میں محبت کے سوا

راحتیں اور بھی ہیں وصل کی راحت کے سوا

ان گنت صدیوں کے تاریک بہیمانہ طلسم

ریشم و اطلس و کمخاب میں بنوائے ہوئے

جا بہ جا بکتے ہوئے کوچہ و بازار میں جسم

خاک میں لتھڑے ہوئے خون میں نہلائے ہوئے

جسم نکلے ہوئے امراض کے تنوروں سے

پیپ بہتی ہوئی گلتے ہوئے ناسوروں سے

لوٹ جاتی ہے ادھر کو بھی نظر کیا کیجے

اب بھی دل کش ہے ترا حسن مگر کیا کیجے

اور بھی دکھ ہیں زمانے میں محبت کے سوا

راحتیں اور بھی ہیں وصل کی راحت کے سوا

مجھ

🔷 Translation :

(A) Hindi Translation

मुझसे पहली सी मोहब्बत मेरे महबूब न माँग, मैंने समझा था कि तू है तो दरख़्शाँ है हयात, तेरा ग़म है तो ग़म-ए-दहर का झगड़ा क्या है, तेरी सूरत से है आलम में बहारों को सबात, तेरी आँखों के सिवा दुनिया में रक्खा क्या है।

तू जो मिल जाए तो तक़दीर निगूँ हो जाए, यूँ न था मैंने फ़क़त चाहा था यूँ हो जाए, और भी दुख हैं ज़माने में मोहब्बत के सिवा, राहतें और भी हैं वस्ल की राहत के सिवा।

अनगिनत सदियों के ये तारीख़ और बे-रहम तिलिस्म, रेशम-ओ-अतलस-ओ-कमख़्वाब में बुनवाए हुए, जा-ब-जा बिकते हुए कूचा-ओ-बाज़ार में जिस्म, ख़ाक में लुथड़े हुए, ख़ून में नहलाए हुए।

लौट जाती है उधर को भी नज़र क्या कीजे, अब भी दिलकश है तेरा हुस्न मगर क्या कीजे, और भी दुख हैं ज़माने में मोहब्बत के सिवा, राहतें और भी हैं वस्ल की राहत के सिवा। मुझसे पहली सी मोहब्बत मेरे महबूब न माँग।

(B) English Translation

Do not ask of me, my love, that love I once had for you. I used to think that life was bright because you were here; Why worry about the world's pain when I have your grief to share? Your face gave the spring its permanence in my sight; What else was there in the universe but your eyes' light?

Had I found you, my destiny would have been humbled so, It wasn't quite like this, I only wished it would be though. There are other sorrows in this world than the sorrow of love, There are other comforts than the solace of a lover’s cove.

Dark, brutal spells of centuries woven in silk and gold, Bodies sold in every street, covered in grime and cold, Flesh smeared in dust and bathed in blood I see, My gaze still turns that way, what choice is there for me?

You are still beautiful, but I am helpless, it is true; There are other sorrows in this world than the sorrow for you. Do not ask of me, my love, that love I once had for you.

🔷 Comparative Analysis:

“Mujh Se Pehli Si Mohabbat Mere Mehboob Na Maang”

Poet: Faiz Ahmed Faiz

Source Language: Urdu

Target Languages: Hindi & English

AI Tools Compared: Gemini AI vs ChatGPT

👉🏻 This poem was written by Faiz Ahmed Faiz in the 1940s, during the period of colonial rule, political unrest, and rising socialist consciousness in the Indian subcontinent. Although it appears to be a romantic love poem, its origin lies in progressive ideology and anti-colonial humanist thought. Faiz initially adopts the traditional Urdu ghazal idiom of romantic devotion, where the beloved is the centre of existence. However, as the poem progresses, its origin reveals a radical shift—the poet moves from personal love to collective suffering, exposing social injustice, poverty, exploitation, and historical violence.

The poem originates from Faiz’s belief that private love cannot remain untouched by public suffering. The beloved becomes a symbol of beauty and comfort, but the poet ultimately refuses to isolate love from reality. This ideological background explains why the poem is often described as a “revolutionary love poem”, where romance is subordinated to social responsibility. Its origin is thus deeply rooted in Progressive Writers’ Movement, Marxist humanism, and post-war disillusionment.

1. Challenges in Translating the Poem

a) Nature of Difficulty

This poem is difficult to translate because it fuses:

• Romantic love

• Political consciousness

• Historical suffering

• Marxist-humanist ideology

Faiz transforms personal love into a critique of social injustice. Translating this layered shift is the core challenge.

Gemini:

• Struggles with ideological layering

• Treats romantic and political elements separately

• Political undertones feel flattened

ChatGPT:

• Better recognizes the ideological transition

• Retains the dialectical movement from personal love → collective suffering

ChatGPT handles thematic complexity more effectively. 

2. Syntax (Sentence Structure)

Original Urdu:

• Long, flowing, emotionally layered sentences

• Frequent enjambment

• Persianized syntactic rhythm

Gemini:

• Preserves long sentence structure mechanically

• Sometimes results in awkward or heavy English syntax

• Over-literal sentence alignment

ChatGPT:

• Reconstructs syntax to suit English idiom

• Breaks long Urdu sentences into meaningful English units

• Maintains emotional flow without syntactic burdenh

ChatGPT prioritizes functional syntax over rigid equivalence. 

3. Metre and Sound Rhythm

Original:

• Strong internal rhyme

• Musical cadence

• Ghazal-like flow

Gemini:

• Attempts to mirror line length

• Sacrifices natural rhythm

• English version feels prosaic

ChatGPT:

• Accepts that metre cannot be replicated

• Preserves semantic rhythm instead of sonic imitation

• Emotional pacing feels closer to original intent

ChatGPT follows Jakobson’s principle: meaning over sound replication. 

4. Lexicon and Word Choice

Key Urdu Words:

• Mohabbat, Gham, Hayat, Rahat, Dahr

• Heavy Indo-Persian emotional load

Gemini:

• Chooses neutral or dictionary equivalents

• Emotional density is reduced

• Some metaphors become abstract

ChatGPT:

• Retains emotive equivalents

• Uses contextually rich English phrases

• Maintains poetic seriousness

ChatGPT preserves emotive lexicon better. 

5. Cultural Connotations

Original:

• Deep Indo-Persian literary culture

• Love intertwined with history, labour, poverty

Gemini:

• Cultural references remain but feel underexplained

• Relies on literal translation

ChatGPT:

• Uses interpretive translation

• Cultural meanings are suggested through tone and phrasing

• Better adaptation for non-Urdu readers

ChatGPT succeeds in cultural mediation. 

6. Tone and Emotional Fidelity

Original Tone:

• Melancholic

• Disillusioned

• Ethically serious

Gemini:

• Tone becomes descriptive

• Emotional intensity weakens

• Feels distant

ChatGPT:

• Retains emotional restraint

• Preserves moral seriousness

• Emotional arc remains intact

ChatGPT maintains emotional authenticity. 

7. Semiotics (Symbols & Imagery)

Key Symbols:

• Light → illusion

• Love → awakening

• Suffering → historical reality

Gemini:

• Translates symbols literally

• Semiotic depth reduces

• Images feel static

ChatGPT:

• Interprets symbols contextually

• Maintains metaphorical tension

• Symbolic progression is clearer

ChatGPT performs stronger semiotic transfer. 

8. Literal Accuracy vs Poetic Fluency

Gemini:

• High literal accuracy

• Lower poetic resonance

• Translation feels instructional

ChatGPT:

• Slightly less literal

• Higher poetic fluency

• Reads like a poem, not a paraphrase

ChatGPT balances meaning and music more successfully. 

9. Handling of Ideological Shift (Romance → Politics)

Core of Faiz’s Poem:

Love must yield to social responsibility.

Gemini:

• Treats ideological shift as thematic change

• Emotional rupture not fully conveyed

ChatGPT:

• Captures ethical transformation

• Love is consciously sacrificed, not simply replaced

ChatGPT understands Faiz’s Marxist-humanist vision. 

10. Theoretical Alignment

Gemini aligns with:

• Catford’s Formal Correspondence

• Structural fidelity

• Linguistic replacement

ChatGPT aligns with:

• Jakobson’s Intersemiotic Translation

• Ramanujan’s Contextual Recreation

• Devy’s Cultural Negotiation

ChatGPT is theoretically richer

Overall Evaluation:

Aspect	Gemini	ChatGPT
Syntax	Rigid	Adaptive
Emotion	Reduced	Preserved
Culture	Flattened	Mediated
Symbolism	Literal	Interpretive
Poetic Flow	Weak	Strong
Ideological Depth	Partial	Clear

Final Critical Insight

For Faiz Ahmed Faiz’s poem, ChatGPT clearly outperforms Gemini in:

• Emotional fidelity

• Cultural depth

• Semiotic interpretation

• Ideological awareness

This confirms that poetry translation is not linguistic substitution but cultural and ethical re-creation, and AI tools must be critically evaluated rather than accepted unconditionally.

Poem: “The Second Coming” – W. B. Yeats

Source Language: English

Target Languages: Hindi and Gujarati

🔹 Translation:

A). Hindi Translation

सब कुछ बिखर रहा है; केंद्र थामे नहीं रहता,

दुनिया पर केवल अराजकता छा गई है।

रक्तमय ज्वार उमड़ पड़ा है हर ओर,

और मासूमियत का संस्कार डूब गया है।

श्रेष्ठतम लोगों में विश्वास नहीं बचा,

और निकृष्टतम लोग उन्मादी उत्साह से भरे हैं।

निश्चय ही कोई प्रकाशन निकट है;

निश्चय ही दूसरा आगमन समीप है।

B). Gujarati Translation

બધું જ વિખેરાઈ રહ્યું છે; કેન્દ્ર ટકી શકતું નથી,

વિશ્વ પર અराजકતા છવાઈ ગઈ છે.

લોહીની ભરમાર લહેર છૂટી પડી છે,

અને નિર્દોષતાની વિધિ ડૂબી ગઈ છે.

શ્રેષ્ઠ લોકોમાં વિશ્વાસનો અભાવ છે,

જ્યારે નિકૃષ્ટ લોકો ઉન્મત્ત ઉત્સાહથી ભરેલા છે.

ચોક્કસ કોઈ પ્રકાશન નજીક છે,

ચોક્કસ બીજું આગમન આવતું જ છે.

🔹 Comparative Analysis:

👉🏻 “The Second Coming” was written in 1919, immediately after World War I, a time marked by mass destruction, political chaos, and moral collapse in Europe. The poem originates from Yeats’s profound sense that Western civilization was disintegrating. Influenced by his own philosophical system outlined in A Vision, Yeats believed history moved in cyclical patterns called “gyres.” According to this worldview, one historical era collapses before another begins.

The poem’s origin lies in post-war anxiety, the fall of old values, and fear of an unknown future. Christian imagery such as the “Second Coming” is used ironically, as the poem predicts not salvation but the birth of a monstrous new order. The “rough beast” symbolizes violent political ideologies and moral degeneration emerging in the modern world. Thus, the poem originates from a blend of historical trauma, mystical philosophy, and apocalyptic imagination, making it one of the defining modernist poems of the 20th century.

1. Syntax (Sentence Structure)

Original Poem

• Uses fragmented, prophetic syntax

• Frequent use of semicolon and enjambment

• Creates a sense of collapse and disorder

• Example:
  “Things fall apart; the centre cannot hold;”

Gemini AI

• Tends to retain original sentence divisions

• Syntax remains closer to English structure

• Sometimes sounds stiff or overly literal

• Maintains formal declarative tone

ChatGPT

• Reshapes syntax slightly for fluency in Hindi/Gujarati

• Breaks long sentences into clearer units

• Improves readability and poetic flow

• Slight loss of original fragmentation

Comparison Insight

• Gemini → Formal equivalence

• ChatGPT → Functional clarity

• Gemini preserves structure; ChatGPT adapts it

2. Metre

Original Poem

• Written in blank verse (unrhymed iambic pentameter)

• Yeats intentionally disrupts rhythm

• Creates tension and instability

Gemini AI

• Makes minimal attempt to recreate rhythm

• Focuses more on semantic accuracy

• Metre largely flattened into prose-like lines

ChatGPT

• Attempts approximate rhythmic balance

• Line breaks adjusted to retain poetic cadence

• Still cannot replicate iambic stress patterns fully

Comparison Insight

• Both sacrifice metre

• ChatGPT makes a greater stylistic effort

• Gemini treats metre as secondary

3. Sound & Rhythm

Original Poem

• Heavy use of harsh consonants (t, k, d, p)

• Creates cacophony

• Biblical, chant-like resonance

Gemini AI

• Retains semantic meaning

• Sound patterns largely lost

• Gujarati version retains some phonetic weight

ChatGPT

• Uses phonetic sensitivity in Hindi/Gujarati

• Tries to maintain solemn rhythm

• Still loses original English sound density

Comparison Insight

• Neither AI can reproduce English soundscape

• ChatGPT slightly better at emotional rhythm

• Gemini remains neutral and flat

4. Lexicon & Grammar

Original Poem

• High-register vocabulary:

  • Anarchy

  • Innocence

  • Revelation

• Dense symbolic diction

Gemini AI

• Chooses direct lexical equivalents

• Hindi: Sanskritized diction

• Gujarati: Formal literary vocabulary

• Grammatically accurate but less expressive

ChatGPT

• Uses context-sensitive vocabulary

• Lexicon slightly more interpretive

• Maintains grammatical clarity + tone

Comparison Insight

• Gemini → Literal lexical accuracy

• ChatGPT → Contextual lexical choice

5. Cultural Connotations

Original Poem

• Deeply rooted in:

  • Christian eschatology

  • Western apocalypse myths

• Symbols like:

  • Second Coming

  • Revelation

Gemini AI

• Keeps metaphors unchanged

• Strong foreignization strategy

• Assumes reader familiarity

ChatGPT

• Also retains metaphors

• Slightly softens theological rigidity

• Makes tone more accessible

Comparison Insight

• Both avoid domestication

• Gemini is culturally rigid

• ChatGPT slightly more adaptive

6. Tone & Emotion

Original Poem

• Apocalyptic

• Anxious

• Prophetic

• Foreboding

Gemini AI

• Tone becomes solemn and detached

• Emotional intensity slightly reduced

• More intellectual than visceral

ChatGPT

• Preserves urgency and dread

• Emotional continuity better maintained

• Reader impact is stronger

Comparison Insight

• ChatGPT retains emotional force better

• Gemini prioritizes accuracy over affect

7. Semiotics (Symbols & Imagery)

Original Symbols

• Centre → Moral order

• Blood-dimmed tide → Violence

• Second Coming → Radical transformation

Gemini AI

• Transfers symbols literally

• Leaves interpretation to reader

• Preserves ambiguity

ChatGPT

• Transfers symbols but with interpretive clarity

• Symbolic depth slightly simplified

Comparison Insight

• Gemini → Symbolic fidelity

• ChatGPT → Symbolic accessibility

8. Literal Accuracy vs Poetic Fluency

Aspect	Gemini	ChatGPT
Literal accuracy	High	Moderate–High
Poetic fluency	Moderate	High
Reader engagement	Lower	Higher

9. Overall AI Tool Evaluation 

Gemini AI

• Strengths:

  • Structural fidelity

  • Literal accuracy

  • Symbol preservation

• Limitations:

  • Flat tone

  • Reduced poetic impact

ChatGPT

• Strengths:

  • Emotional resonance

  • Readability

  • Poetic flow

• Limitations:

  • Slight interpretive liberties

  • Reduced symbolic density

Final Critical Insight 

This comparison proves that translating “The Second Coming” is not merely a linguistic task but a semiotic and cultural negotiation.

• Gemini aligns with Catford’s formal correspondence

• ChatGPT aligns with Jakobson’s interpretive equivalence

Conclusion:

For Yeats’s apocalyptic poetry, ChatGPT performs better in preserving tone and reader impact, while Gemini excels in formal and symbolic fidelity.

Gujarati Poem

“દિવસો જુદાઈના જાય છે” — ગની દહીંવાલા

Original Gujarati Poem:

દિવસો જુદાઈના જાય છે, એ જશે જરૂર મિલન સુધી, મારો હાથ ઝાલીને લઈ જશે, મુજ શત્રુઓ જ સ્વજન સુધી.

ન ધરા સુધી, ન ગગન સુધી, નહિ ઉન્નતિ, ન પતન સુધી, અહીં આપણે તો જવું હતું, ફક્ત એકમેકના મન સુધી.

હજી પાથરી ન શક્યું સુમન પરિમલ જગતના ચમન સુધી, ન ધરાની હોય જો સંમતિ, મને લૈ જશો ન ગગન સુધી.

છે અજબ પ્રકારની જિંદગી, કહો એને પ્યારની જિંદગી, ન રહી શકાય જીવ્યા વિના, ન ટકી શકાય જીવન સુધી.

તમે રાંકનાં છો રતન સમાં, ન મળો હે અશ્રુઓ ધૂળમાં, જો અરજ કબૂલ હોય આટલી, તો હૃદયથી જાઓ નયન સુધી.

તમે રાજરાણીનાં ચીર સમ, અમે રંક નારની ચૂંદડી! તમે બે ઘડી રહો અંગ પર, અમે સાથ દઈએ કફન સુધી.

જો હૃદયની આગ વધી 'ગની', તો ખુદ ઈશ્વરે જ કૃપા કરી, કોઈ શ્વાસ બંધ કરી ગયું, કે પવન ન જાય અગન સુધી.

🔹Translation of the Poem

A). Hindi Translation: 

जुदाई के दिन बीत रहे हैं,

वे मिलन तक अवश्य पहुँचेंगे,

मेरा हाथ थाम कर ले जाएँगे,

मेरे शत्रु भी स्वजन बन जाएँगे।

न धरती तक, न आकाश तक,

न उत्थान तक, न पतन तक,

हमें तो यहाँ जाना था केवल,

एक-दूसरे के मन तक।

अभी पुष्पों पर सुगंध बिखर न सकी,

जगत के उपवन तक,

यदि धरती की सहमति न हो,

तो मुझे आकाश तक मत ले जाना।

यह जीवन अजीब प्रकार का है,

इसे प्रेम का जीवन कहो,

बिना जिए जिया नहीं जाता,

न जीवन तक टिक पाया जाता।

तुम कंगालों के लिए रत्न समान हो,

अश्रुओं की धूल में मत मिलो,

यदि इतनी विनती स्वीकार हो,

तो हृदय से आँखों तक मत जाओ।

तुम रानी के वस्त्र समान हो,

हम निर्धन नारी की ओढ़नी,

तुम दो घड़ी अंग से लगे रहो,

हम साथ निभाएँ कफन तक।

यदि हृदय की आग बढ़ गई, ‘गनी’,

तो स्वयं ईश्वर ने कृपा की,

किसी ने श्वास को रोक दिया,

कि पवन अग्नि तक न पहुँचे।

B). English Translation:

The days of separation are passing,

They will surely reach reunion,

Holding my hand they will lead me,

Even my enemies will turn into kin.

Not up to the earth, nor the sky,

Neither rise nor fall we sought,

Here, we only had to travel

Into each other’s hearts.

The fragrance has not yet spread

Across the garden of the world,

If the earth does not consent,

Do not take me to the heavens.

Life is of a strange kind,

Call it a life of love,

One cannot live without living,

Nor endure all the way to life.

You are a jewel among the poor,

Do not mix with the dust of tears,

If this one plea be granted,

Do not journey from heart to eyes.

You are like a queen’s silk attire,

We, a pauper woman’s shawl,

You rest upon the body briefly,

We accompany till the shroud.

If the fire of the heart grew fierce, ‘Gani’,

God Himself showed mercy,

Someone stopped a breath midway,

So the wind would not reach the flame.

🔹 Comparative Analysis:

👉🏻 This Gujarati poem originates from the Gujarati lyrical–ghazal tradition, which blends romantic devotion, spiritual intimacy, and emotional sacrifice. Written in a cultural milieu where poetry is closely tied to oral recitation and musical rhythm, the poem reflects the lived emotional experiences of separation (જુદાઈ), longing, humility, and selfless love.

The origin of this poem lies in regional cultural values, where love is seen not merely as desire but as endurance, patience, and ethical commitment. Metaphors such as chundadi, kafan, rajrani, and agni arise from everyday Gujarati social and emotional life, making the poem deeply rooted in its cultural soil. The poet’s voice reflects the Bhakti-inflected sensibility of surrender and devotion, where personal emotion merges with spiritual acceptance. Hence, the poem originates from a localized emotional universe, shaped by language, tradition, and cultural memory. 

1. Sentence Structure (Syntax)

Original Gujarati:

• Uses long, flowing, emotionally layered sentences

• Syntax follows bhavarth-pradhan (emotion-centred) structure

• Frequent use of parallel negatives:
  “ન ધરા સુધી, ન ગગન સુધી”

ChatGPT Translation:

• Preserves original sentence flow and emotional continuity

• Maintains parallel structures and repetitions

• Syntax remains poetic and reflective, even in English

Gemini Translation:

• Slightly simplifies sentence complexity

• Breaks longer lines into clearer, shorter clauses

• Syntax becomes more explanatory than evocative

Comparison Insight:

• ChatGPT retains poetic sentence rhythm more faithfully

• Gemini prioritizes syntactic clarity over emotional layering

2. Metre and Sound (Rhythm & Musicality)

Original Gujarati:

• Strong end-rhyme: “મિલન / સ્વજન / મન / કફન”

• Musicality rooted in oral tradition

• Repetition creates a soft lament-like rhythm

ChatGPT:

• Attempts to retain rhythmic cadence, especially in Hindi

• In English, preserves semantic rhythm even if rhyme is lost

• Emotional musicality remains perceptible

Gemini:

• Metre largely sacrificed, especially in English

• Focuses on meaning transfer rather than sound

• Rhythm becomes flatter and more prose-like

Comparison Insight:

• ChatGPT performs better in retaining poetic rhythm

• Gemini treats metre as secondary to semantic accuracy

3. Literal Accuracy vs Poetic Fluency

Original Gujarati:

• Uses metaphor-heavy, emotionally compressed language
  Example:
  “તમે બે ઘડી રહો અંગ પર, અમે સાથ દઈએ કફન સુધી”

ChatGPT:

• Takes creative liberties to retain emotional depth

• Sometimes expands imagery for clarity but keeps tone intact

• Strong balance between meaning + poetry

Gemini:

• More literal and restrained

• Meaning is accurate, but poetic intensity is reduced

• Emotional compression gets diluted

Comparison Insight:

• ChatGPT leans toward poetic fluency

• Gemini leans toward literal fidelity

4. Lexicon and Grammar

Original Gujarati:

• Culturally dense words: ચૂંદડી, કફન, રાજરાણી

• Emotional lexicon tied to Indian social imagery

ChatGPT:

• Retains culturally marked words longer (especially in Hindi)

• In English, translates metaphorically but sensitively

• Grammar supports emotional nuance

Gemini:

• Uses neutralized vocabulary in English

• Lexical choices are correct but emotionally muted

• Grammar is technically strong but stylistically plain

Comparison Insight:

• ChatGPT preserves emotive diction better

• Gemini ensures grammatical precision over affect

5. Cultural Connotations

Original Gujarati:

• ચૂંદડી vs રાજરાણીનાં ચીર → class contrast

• કફન → ultimate sacrifice

• શ્વાસ / પવન / અગન → life, fate, divine mercy

ChatGPT:

• Retains these as living cultural symbols

• Avoids over-explanation

• Allows reader to experience cultural depth

Gemini:

• Cultural meanings are often implicitly explained

• Risk of turning metaphor into abstraction

Comparison Insight:

• ChatGPT better preserves cultural immediacy

• Gemini tends toward cultural generalization

6. Tone and Emotion

Original Gujarati:

• Tone: tender, resigned, sacrificial

• Emotion flows quietly, without melodrama

ChatGPT:

• Successfully retains emotional gravity

• Especially strong in final stanza about divine mercy

Gemini:

• Emotional tone is present but softened

• Becomes reflective rather than visceral

Comparison Insight:

• ChatGPT captures emotional depth more fully

• Gemini offers emotional clarity but less intensity

7. Semiotics (Symbols & Imagery)

Key Symbols:

• Journey → emotional union

• Fire & wind → desire and fate

• Breath → life being paused by grace

ChatGPT:

• Treats symbols as experiential

• Preserves ambiguity and interpretive openness

Gemini:

• Treats symbols as conceptual

• Meaning is clear but less layered

Comparison Insight:

• ChatGPT excels in semiotic transfer

• Gemini excels in semantic explanation

Overall Evaluation:

Aspect	ChatGPT	Gemini
Syntax	Poetic & flowing	Clear & simplified
Metre & Sound	Partially retained	Mostly sacrificed
Cultural Depth	Strong	Moderate
Emotional Intensity	High	Medium
Semiotic Richness	Preserved	Reduced
Overall Strength	Poetic recreation	Literal accuracy

Concluding Insight:

In translating “દિવસો જુદાઈના જાય છે”, ChatGPT performs better as a poetic translator, preserving emotion, rhythm, and cultural symbolism. Gemini functions more as an explanatory translator, ensuring clarity and correctness but at the cost of lyrical depth.

This comparison clearly demonstrates that poetry translation is a cultural and semiotic act, not merely a linguistic one—especially when handled by AI tools.

Poem (Hindi – Original)

Harivansh Rai Bachchan: “कोशिश करने वालों की हार नहीं होती”

लहरों से डरकर नौका पार नहीं होती,

कोशिश करने वालों की हार नहीं होती।

नन्हीं चींटी जब दाना लेकर चलती है,

चढ़ती दीवारों पर, सौ बार फिसलती है।

मन का विश्वास रगों में साहस भरता है,

चढ़कर गिरना, गिरकर चढ़ना न अखरता है।

आख़िर उसकी मेहनत बेकार नहीं होती,

कोशिश करने वालों की हार नहीं होती।

डुबकियाँ सिंधु में गोताखोर लगाता है,

जा जा कर खाली हाथ लौट आता है।

मिलते नहीं सहज ही मोती गहरे पानी में,

बढ़ता दुगुना उत्साह इसी हैरानी में।

मुट्ठी उसकी खाली हर बार नहीं होती,

कोशिश करने वालों की हार नहीं होती।

असफलता एक चुनौती है, स्वीकार करो,

क्या कमी रह गई देखो और सुधार करो।

जब तक न सफल हो, नींद-चैन को त्यागो तुम,

संघर्ष का मैदान छोड़ मत भागो तुम।

कुछ किये बिना ही जय-जयकार नहीं होती,

कोशिश करने वालों की हार नहीं होती।

A). English Translation:

Frightened by waves, no boat ever crosses the shore;

Those who keep trying are never truly defeated.

A tiny ant carrying a grain moves ahead,

Climbs the wall, slips again and again instead.

Faith in the heart fills the veins with might,

Falling and rising do not weaken the fight.

In the end, its labour is never in vain;

Those who keep trying do not fail again.

The diver plunges deep into the sea,

Returns empty-handed repeatedly.

Pearls are not found easily in ocean deep,

Yet wonder doubles the will to leap.

His fist is not empty every time;

Those who keep trying always climb.

Failure is a challenge—accept it true,

Find what went wrong and correct it anew.

Until success comes, forsake rest and ease,

Do not flee from struggle’s seas.

Victory comes not without action or fight;

Those who keep trying are never denied.

B). Gujarati Translation:

લહેરોથી ડરીને નાવ પાર નથી થતી,

પ્રયત્ન કરનારાની હાર નથી થતી।

નાની ચીંટી દાણો લઈને આગળ વધે,

દીવાલ ચઢે, સો વાર લપસે અને ખસે।

મનનો વિશ્વાસ નસોમાં સાહસ ભરે,

પડીને ઊઠવું તેને ક્યારેય ન ખરે।

અંતે એની મહેનત વ્યર્થ નથી થતી,

પ્રયત્ન કરનારાની હાર નથી થતી।

ગોતાખોર દરિયામાં ઊંડે ઊતર જાય,

ખાલી હાથે વારંવાર પાછો આવે।

મોતી સહેજે ઊંડા પાણીમાં નથી મળે,

અચંબાથી જ ઉત્સાહ બમણો ફૂલે।

એની મુઠ્ઠી હંમેશાં ખાલી નથી થતી,

પ્રયત્ન કરનારાની હાર નથી થતી।

નિષ્ફળતા એક પડકાર છે, સ્વીકાર કર,

શું ખોટ રહી ગઈ એ જોઈ સુધાર કર।

સફળતા સુધી આરામ-ઊંઘ ત્યાગ કર,

સંઘર્ષના મેદાનથી ક્યારેય ભાગ ન કર।

કર્યા વિના વિજયનો શોર નથી થતો,

પ્રયત્ન કરનારાની હાર નથી થતી।

🔹Comparative Analysis:

👉🏻 This poem originates from modern Hindi motivational poetry, written by Harivansh Rai Bachchan, one of the most influential poets of the Chhayavad and post-Chhayavad era. The poem emerged in a period when Hindi literature was increasingly addressing common people, emphasizing perseverance, moral strength, and self-belief.

Its origin is closely linked to Indian philosophical values such as कर्म (karma), श्रम (effort), and धैर्य (patience). Bachchan uses simple language, repetitive structure, and familiar imagery (ant, ocean, pearl) so that the poem could reach students, workers, and ordinary readers. Unlike elite romantic or symbolic poetry, this poem’s origin lies in didactic humanism, aiming to inspire resilience rather than aesthetic complexity. It became popular through recitations, textbooks, and public speeches, reinforcing its roots in collective moral education rather than individual introspection

1. Sentence Structure (Syntax)

Original Hindi

• Uses simple, declarative, parallel sentence structures

• Repetition reinforces moral certainty

• Syntax mirrors didactic oral tradition

ChatGPT Translation

• Preserves linear sentence flow

• Slight restructuring for English clarity

• Maintains cause–effect logic
   Example:

“Frightened by waves, no boat ever crosses the shore.”

→ Keeps original syntax but smooths inversion

Gemini Translation

• Retains closer sentence-to-sentence correspondence

• More literal rendering

• Slightly rigid English phrasing at places

Comparison

• Gemini = structurally faithful

• ChatGPT = syntactically adaptive and fluent
  Better for poetry: ChatGPT (natural readability)

2. Metre and Sound Rhythm

Original Hindi

• Strong refrain-based rhythm

• Oral chant quality

• Repetition of:

  “कोशिश करने वालों की हार नहीं होती”

ChatGPT

• Attempts end rhyme and internal rhythm

• Sound is approximated, not replicated

• Focuses on motivational cadence

Gemini

• Prioritizes meaning over musicality

• Rhythm becomes prose-like

• Less attention to sonic repetition

Comparison

• ChatGPT preserves motivational rhythm better

• Gemini sacrifices sound for semantic clarity

Stronger musical adaptation: ChatGPT

3. Literal Accuracy vs Poetic Fluency

Original

• Literal simplicity + philosophical depth

ChatGPT

• Takes minor liberties to preserve emotion

• Uses idiomatic English expressions
  
  

• Example:

“Those who keep trying are never truly defeated.”

Gemini

• Very close semantic equivalence

• Sometimes sounds instructional rather than poetic

Comparison

• Gemini = literal accuracy

• ChatGPT = poetic fluency

Better poetic balance: ChatGPT

4. Lexicon and Grammar

Original Lexicon

• Everyday Hindi

• Universally accessible imagery (ant, boat, ocean)

ChatGPT

• Uses emotionally charged but simple vocabulary:
  faith, struggle, effort, victory

Gemini

• Slightly formal and explanatory diction

• Less emotional immediacy

Comparison

• ChatGPT aligns better with Bachchan’s simplicity

• Gemini risks over-neutralization

5. Cultural Connotations

Original

• Deeply Indian moral philosophy:

  • मेहनत (effort)

  • संघर्ष (struggle)

  • धैर्य (patience)

ChatGPT

• Translates values into universal human ethics

• Cultural flavor slightly reduced but spirit retained

Gemini

• Neutralizes cultural tone further

• Treats poem as moral instruction

Comparison

• ChatGPT = cultural adaptation

• Gemini = cultural flattening

Better ethical resonance: ChatGPT

6. Tone and Emotion

Original

• Encouraging, hopeful, resolute

ChatGPT

• Retains inspirational warmth

• Reader feels motivated

Gemini

• Emotionally correct but less stirring

Comparison

• ChatGPT maintains emotional force

• Gemini maintains informational tone

7. Semiotics (Symbols & Metaphors)

Key Symbols

• Boat → human life

• Waves → obstacles

• Ant → perseverance

• Pearl → success after struggle

ChatGPT

• Transfers symbols experientially

• Reader feels the struggle

Gemini

• Transfers symbols conceptually

• Reader understands but does not fully feel

Comparison

• ChatGPT = semiotic depth

• Gemini = semantic transfer

Overall Comparative Evaluation:

Aspect	ChatGPT	Gemini
Syntax	Adaptive & fluent	Literal
Rhythm	Better preserved	Mostly lost
Lexicon	Emotionally resonant	Neutral
Culture	Adapted ethically	Flattened
Tone	Motivational	Didactic
Semiotics	Experiential	Conceptual

Critical Insight

This comparison clearly shows that:

• ChatGPT performs better as a poetic translator, prioritizing tone, rhythm, and emotional impact

• Gemini performs better as a literal translator, prioritizing structural and semantic accuracy

👉 In Bachchan’s poem—where emotion, repetition, and moral force are central—ChatGPT’s approach aligns more closely with the poet’s intent.

Conclusion:

The translation of “कोशिश करने वालों की हार नहीं होती” demonstrates that poetry translation is not merely linguistic transfer but an act of ethical, emotional, and semiotic recreation. While Gemini ensures formal equivalence, ChatGPT succeeds in preserving the spirit of perseverance that defines Harivansh Rai Bachchan’s poetic philosophy.

👉🏻 The comparative analysis of the translations of Harivansh Rai Bachchan’s poem into English and Gujarati reveals significant insights into how AI negotiates poetic meaning across languages. In terms of syntax, the Gujarati translation largely retains the original Hindi sentence structure due to linguistic proximity, preserving the flow and emotional cadence of the source text. The English translation, however, restructures sentences to suit grammatical norms, often prioritizing clarity over poetic inversion, which slightly alters the original rhythmic tension.

Regarding metre and sound, the Gujarati version performs more effectively, approximating the rhyme and musicality inherent in Bachchan’s lyrical style. English translation struggles to retain the original rhythm, as Hindi’s syllabic flow and rhyme patterns do not easily transfer into English; thus, musicality is often sacrificed for semantic accuracy.

In lexicon and grammar, both translations are largely accurate, but the English version occasionally opts for neutral or abstract vocabulary, reducing the emotional intensity found in Bachchan’s culturally rooted diction. Gujarati, sharing cultural and poetic sensibilities with Hindi, preserves idiomatic warmth and emotional immediacy more effectively.

Cultural connotations pose the greatest challenge in the English translation. Symbols tied to Indian ethos—such as suffering, resilience, and philosophical acceptance—require explanatory phrasing, which dilutes poetic compactness. Gujarati translation adapts these connotations seamlessly, as the cultural codes remain largely intact.

The tone and emotion of Bachchan’s poem—marked by existential reflection and restrained hope—remain more faithful in Gujarati, while English conveys the message but with a softened emotional resonance. Finally, in terms of semiotics, metaphors and symbolic imagery travel more successfully into Gujarati, whereas English translations often translate symbols conceptually rather than experientially.

Overall, this comparative analysis demonstrates that poetry translation is not merely linguistic substitution but a cultural and semiotic act, where closeness of language and culture significantly enhances fidelity. The exercise reinforces the idea that AI-generated translations must be critically evaluated, as meaning in poetry emerges from the complex interaction of sound, structure, culture, and emotion—not from words alone.

Comparative Analysis of AI-Generated Translations (All Four Poems)

1. Challenges in Translation

Across all four poems—Faiz Ahmed Faiz’s Mujh Se Pehli Si Mohabbat, Yeats’s The Second Coming, the Gujarati poem Divaso Judaina Jaye Chhe by Gani Dahiwala, and the Hindi poem by Harivansh Rai Bachchan—the most challenging aspects of translation were metaphor, emotional intensity, and culturally embedded expressions.

• Faiz’s poem posed difficulties due to its layered political and romantic metaphors, where love merges with social suffering.

• The Second Coming challenged translators because of its dense symbolism, biblical imagery, and apocalyptic tone.

• The Gujarati poem relied heavily on emotional intimacy, idiomatic phrases, and culturally rooted imagery such as kinship, sacrifice, and destiny.

• Bachchan’s poem presented challenges in rhythm, philosophical depth, and lyrical simplicity, which are deeply tied to Hindi poetic tradition.

These challenges show that poetry translation goes beyond word substitution and enters the realm of cultural and emotional negotiation.

2. Cultural Connotations and Collocations

Cultural connotations were handled differently by AI tools:

• In Faiz and the Gujarati poem, words related to love, sacrifice, and separation carried cultural weight that could not be fully transferred without explanation or adaptation.

• Yeats’s references to Christian eschatology and Western myth were often translated literally, sometimes losing their symbolic resonance for Indian-language readers.

• Bachchan’s culturally familiar imagery (life, struggle, endurance) translated more smoothly but still lost some musical familiarity in English.

AI tools attempted to resolve these issues by:

• Using neutral equivalents

• Paraphrasing culturally specific terms

• Occasionally flattening symbolic depth to preserve clarity. 

3. Untranslatable Words and Phrases

Certain words and expressions remained partially untranslatable:

• Faiz’s “rahat-e-didar”, Gujarati emotional idioms, and Bachchan’s rhythmic phrasing resist exact equivalents.

• These were resolved by AI through semantic approximation, prioritizing meaning over form.

• This confirms Roman Jakobson’s idea that poetry is the most resistant to complete translation.

4. Syntax, Metre, and Sound

• Syntax was often reshaped to suit the grammatical norms of the target language, especially in English.

• Metre and rhyme were frequently sacrificed, particularly in English translations, while Hindi and Gujarati versions retained rhythm more successfully.

• In Yeats and Bachchan, preserving rhyme often led to minor shifts in meaning.

• In Faiz and the Gujarati poem, emotional flow was prioritized over strict metrical accuracy.

This shows that AI translators tend to favor semantic clarity over sonic fidelity.

5. Tone, Emotion, and Semiotics

• The tone and emotion were largely preserved in all translations, though with reduced intensity.

• Symbolic elements (the “rough beast” in Yeats, fire and breath in the Gujarati poem, love and loss in Faiz, endurance in Bachchan) were translated literally rather than semiotically, sometimes weakening interpretive richness.

• Emotional continuity remained intact, but symbolic density was diluted, especially in English translations.

Comparison: ChatGPT vs Gemini

• ChatGPT performed better in preserving emotional tone, coherence, and readability, often choosing interpretive clarity.

• Gemini leaned toward literal accuracy, sometimes maintaining structure but losing poetic flow.

• Overall, ChatGPT was more effective in retaining the poetic spirit, while Gemini was stronger in formal equivalence.

Theoretical Framework: Translation as Linguistic, Cultural, and Semiotic Negotiation

This translation activity is grounded in a plural theoretical framework drawn from key thinkers in Translation Studies—Roman Jakobson, J. C. Catford, Ganesh Devy, and A. K. Ramanujan. Together, these theorists help explain why poetry translation—especially when mediated by Gen AI tools—is not a simple act of word substitution but a complex negotiation among language, culture, sound, and meaning.

1. Roman Jakobson: Translation as Semiotic Interpretation

Roman Jakobson’s theory of translation is central to understanding the challenges faced in translating poetry. Jakobson famously argues that “poetry by definition is untranslatable”, meaning that poetic meaning is inseparable from sound, rhythm, and symbolism.

Jakobson distinguishes between:

• Intralingual translation (within the same language),

• Interlingual translation (between languages),

• Intersemiotic translation (between sign systems).

In this activity, the poems move across languages and cultures, making translation an intersemiotic act where symbols, metaphors, and emotions must be re-encoded rather than replicated. For example:

• Images like falcon, gyre, wine, love, separation, or fire do not carry identical meanings across Urdu, Hindi, Gujarati, and English.

• AI tools often preserve semantic meaning but struggle with symbolic density, confirming Jakobson’s claim that poetic equivalence is never total.

Thus, Jakobson helps frame translation here as an interpretive act, not a mechanical process—an insight crucial for evaluating AI-generated translations.

2. J. C. Catford: Linguistic Shifts and Losses

J. C. Catford’s A Linguistic Theory of Translation provides tools to analyze structural and grammatical shifts during translation. Catford defines translation as the replacement of textual material in one language with equivalent textual material in another language—but he also acknowledges inevitable translation loss.

In the AI translations examined:

• Syntax often shifts to suit target-language grammar.

• Metre and rhyme are frequently sacrificed for clarity.

• Certain collocations or idioms lack formal equivalents.

Catford’s framework allows us to identify:

• Level shifts (e.g., poetic rhythm becoming prose-like meaning),

• Category shifts (changes in sentence structure or grammatical class).

AI tools tend to prioritize functional equivalence over formal equivalence, aligning with Catford’s model but also exposing the limitations of rule-based or probabilistic systems when handling poetic nuance.

3. Ganesh Devy: Translation as Cultural Memory and Survival

Ganesh Devy’s perspective moves translation beyond linguistics into the realm of cultural ethics and power relations. Devy views translation as a form of cultural survival, especially in multilingual and postcolonial contexts like India.

From Devy’s viewpoint:

• Translation is not neutral—it reshapes literary history.

• Cultural metaphors, emotional registers, and social hierarchies embedded in poems must be handled with sensitivity.

• AI tools risk flattening cultural specificity, especially when translating regional or non-Western poetry.

In poems by Faiz or Bachchan, words like mehboob, madira, virah, or rajrani carry historical and emotional weight. AI translations often approximate meaning but may dilute cultural resonance, validating Devy’s concern that translation involves cultural negotiation, not mere equivalence.

4. A. K. Ramanujan: Context, Creativity, and Multiple Truths

A. K. Ramanujan argues that there is no single “correct” translation; instead, translations are shaped by context, audience, and purpose. He emphasizes the need to balance:

• Fidelity to meaning

• Fidelity to poetic experience

Ramanujan’s idea of translation as “context-sensitive recreation” is especially useful for evaluating AI outputs:

• ChatGPT often opts for smoother, emotionally resonant lines.

• Gemini sometimes retains closer syntactic resemblance.

Neither approach is absolutely superior; each reflects different translation priorities, supporting Ramanujan’s belief that translation is inherently plural and interpretive.

Synthesis: Applying the Framework to AI-Based Translation

Taken together, these theorists reveal that:

• Jakobson explains why poetic meaning resists perfect translation.

• Catford helps identify linguistic shifts and structural compromises.

• Devy foregrounds cultural depth and ethical responsibility.

• Ramanujan legitimizes creative freedom and contextual adaptation.

When applied to Gen AI translations, this framework shows that AI performs best at semantic transfer but struggles with cultural nuance, rhythm, and symbolic density. Therefore, AI translation must be seen as a collaborative tool, requiring human critical judgment to evaluate, revise, and contextualize outputs.

Concluding Insight

This theoretical framework demonstrates that poetry translation—whether human or AI-generated—is not merely linguistic but deeply cultural and semiotic. The activity thus reinforces a key insight of Translation Studies:

👉 Translation is not about sameness, but about meaningful difference.

Conclusion:

This translation exercise demonstrates that poetry translation is not merely linguistic. It involves cultural negotiation, emotional calibration, and semiotic interpretation. While syntax and metre shift, the core spirit and symbolic meaning survive, proving that AI-assisted translation can support—but not replace—human literary sensitivity.

Learning Outcomes of This Translation & Comparative Analysis Activity:

Developed critical awareness of poetry translation by understanding how meaning, rhythm, emotion, and culture shift across languages rather than transferring mechanically.

Gained insight into AI translation tools (ChatGPT vs Gemini) by identifying their strengths and limitations in handling syntax, metre, symbolism, and cultural nuance.

Applied translation theories (Jakobson, Catford, Devy, Ramanujan) practically, moving beyond theory to real textual evaluation.

Recognized translation as a cultural and semiotic act, where language, history, emotion, and symbolism interact to produce meaning, especially in poetic texts.

References:

   Admin, Admin, & Admin. (2021, September 26). Hindi Poem: लहरों से डर कर नौका पार नहीं होती. FundaTimes. Retrieved December 20, 2025, from https://fundatimes.com/lahron-se-dar-kar/

 Anjum, R. Y. (2016). Cross cultural translation and translatability of poetry. Advances in Social Sciences Research Journal, 3(5). https://doi.org/10.14738/assrj.35.1909

 Barad, Dilip. (2024). Guidelines for Using Generative AI in Translation Studies. 10.13140/RG.2.2.29351.25766. 

 Barad, Dilip. (2024). Translation Studies Activity Worksheet Using Gen AI Tools for Translating Poems. 10.13140/RG.2.2.32287.27041. 

Catford, J. C. A Linguistic Theory of Translation. Oxford UP, 1965.

Devy, Ganesh. In Another Tongue: Essays on Indian English Literature. Macmillan, 1993.

 Gujarati Lyrics. (2018, January 26). Divaso Judai na Jaay Chhe Gujarati Lyrics. Retrieved December 20, 2025, from https://gujaratilyrics.com/divaso-judai-na-jaay-chhe-gujarati-lyrics/

 Jakobson, Roman. "On Linguistic Aspects of Translation". On Translation, edited by Reuben Arthur Brower, Cambridge, MA and London, England: Harvard University Press, 1959, pp. 232-239. https://doi.org/10.4159/harvard.9780674731615.c18

 Jahanzeb, K., Jahan, J., & Shahzadi, K. (2023). An analysis of English translation of Faiz’s poetry on Textual and Extra-Textual Level. Zenodo (CERN European Organization for Nuclear Research). https://doi.org/10.5281/zenodo.7773538

 Nikolaou, P., & Rossi, C. (2022). Translating poetry. In Cambridge University Press eBooks (pp. 480–498). https://doi.org/10.1017/9781108616119.025

 فیض احمد فیض - نظم. (n.d.). Rekhta. Retrieved December 20, 2025, from https://www.rekhta.org/nazms/mujh-se-pahlii-sii-mohabbat-mirii-mahbuub-na-maang-mujh-se-pahlii-sii-mohabbat-mirii-mahbuub-na-maang-faiz-ahmad-faiz-nazms?lang=ur

Ramanujan, A. K. “Three Hundred Ramayanas: Five Examples and Three Thoughts on Translation.” Cultural Dynamics, vol. 1, no. 1, 1987.

The Second coming | the poetry foundation. (n.d.). https://www.poetryfoundation.org/poems/43290/the-second-coming

Thank you!