1. Introduction

In the current era when big data technology has deeply permeated, the functional boundaries of artificial intelligence (AI) continue to expand, and its application scenarios have covered almost all industries. The translation field has also rapidly developed due to the empowerment of AI technology. Machine translation, with its efficiency and convenience, has significantly improved cross-language information transmission. However, it has also impacted the survival of traditional human translation and the depth of cultural dissemination, sparking extensive discussions in academic circles about the future of the translation industry. Ancient Chinese poetry, as a key part of China’s traditional culture, embodies millennia of literary and cultural heritage. It is not only the origin of literary creation but also a key medium for transmitting the spirit of the nation's culture. Compared with ordinary texts, ancient poetry places higher demands on readers' acceptance and interpretation abilities, which also poses higher requirements for translations. When addressing readers from different cultural backgrounds, translations must not only accurately convey the literal meaning but also reproduce the poetic rhythm and aesthetic appeal as well as the deep emotions. Translations build a bridge of emotional resonance between the author and the reader. Exploring the differences between AI translation and human translation in the handling of ancient poetry is of great significance. In the field of ancient Chinese poetry translation research, there have been many previous studies. Some scholars believe that poetry translation should be based on an aesthetic perspective, reproducing the beauty of the poetry [1], some scholars, from a deconstructionist viewpoint, propose that the translation is not merely a replication of the original text but a re-shaping of the literary work by the translator [2], and some scholars hold that the metaphor in poetry is the key point of poetry translation, and it requires the translator to conduct in-depth study and understanding of it in order to accurately convey it to the readers [3]. The previous studies have mostly focused on the degree of restoration of the original text in the translation, and rarely considered from the perspective of the readers whether the translation can be adapted to the readers' acceptance. The value of literary works lies in the interpretation and perception of the readers [4], so this paper, based on the Reception Theory, takes Li Bai’s classic poem Qiangjinjiu as an example, conducts research on the influence of different versions of poetry translations on readers' acceptance, selects different versions of word and sentence translations, and conducts a comparative analysis from the three dimensions of language, culture and emotion to reveal the aesthetic and emotional expression of the translation from the perspective of readers' acceptance. This research aims to explore the interaction between poetry translation and readers' acceptance, draw attention to the unique value and current predicament of human translation in cultural dissemination in society, and prompt translators to pay more attention to readers' acceptance in translation, providing ideas and theoretical support for classical poetry translation practice and overseas dissemination in the AI era.

2. Introduction of Reception Theory

Reception Theory, also called Reception Aesthetics, is a theory about literature and cultural research. This theory originated in the late 1960s and was founded by Jauss and Iser. It places more emphasis on readers and believes that the meaning of literary works is generated in the acceptance and interpretation of them, not only decided by the author. A work, although all the contents remain unchanged, may constantly have its meaning altered and updated as time goes by. This suggests that the meaning of the work is bestowed by the reader, and it can only be generated during the reading process. Building on this idea, this theory identifies three important points: Horizons of Expectation, Appeal Texture and Implied Reader [5].

2.1. Horizons of Expectation

Horizons of Expectation is the main content of Reception Theory proposed by Jauss. By this, it means that when readers begin to read a work or during the process of reading a work, they have some expectations in their hearts while these psychological expectations from the readers are what we call the Horizons of Expectation. These expectations vary from person to person, influenced by the readers’ life experiences and cultural background, and they also continue to evolve along with the changes of the times [6].

2.2. Appeal Texture

In Iser’s Reception Theory, he proposed the core concept of the Appeal Texture. Iser pointed out that literary works contain unexpressed meaning spaces and ambiguity, and these gaps require the participation of the readers. Appeal Texture emphasizes that when authors create, they should leave some unspecified parts within the range of readers’ receptive capacity, allowing the readers to complete the supplementation of these parts based on their own expectations and understanding [6].

2.3. Implied Reader

The concept of Implied Reader does not refer to a specific actual reader, but rather refers to some hypothetical conceptions within the work. Through the introduction of these conceptions, Iser emphasizes the significance of the reader in the interpretation of the text. When the author is creating, they not only need to consider from their own perspective, but also need to take into account the reading experience and comprehension ability of potential readers. This point highlights the dynamic interaction between the creator and the recipient, and promotes the mutual influence between the two [6].

In the field of literary translation, Reception Theory offers a new perspective emphasizing that translators should take into account the reading requirements and aesthetic expectations of the target readers during the translation process. While meeting the expectation of the target language readers, it is also necessary to pay attention to the guiding role of the original text for the readers and the idealized reader image implicitly present in the text. This has considerable guiding significance for the translation of literary works [6].

3. Comparative analysis between AI translation and human translation of Qiangjinjiu

In recent years, AI translation has developed rapidly. With its strong semantic understanding and other capabilities, it has driven the leapfrog development of translation technology, and its application scenarios have been continuously expanded, covering a wide range of fields. AI translation is a translation method that utilizes artificial intelligence technology to simulate human translation thinking and achieve automatic conversion between different languages. It is the product of the development of machine translation at a certain stage. On the other hand, human translation is the activity of language conversion carried out by translators with professional language abilities [7]. Qiangjinjiu is a representative work of Li Bai. Its passionate and romantic atmosphere, along with the unique cultural imprint, vividly reflects the grandeur of the Tang Dynasty. Taking this as the research object for translation, it is possible to explore how to convey the artistic tension of the original work during the language conversion process, and provide a new research perspective for achieving more accurate international dissemination of Chinese classical poetry. Currently, this poem not only has classic version of artificial made by translators such as Xu Yuanchong, but also various AI translation systems have provided translations. Based on the Reception Theory mentioned above, the comparative analysis will respectively consider from the aspects of language, culture and emotion how to accurately convey the language elements while also taking into account the cognitive expectations and acceptance capabilities of the target readers. Building a bridge for communication between Chinese poetry authors and other native speakers through translation.

3.1. Linguistic aspect

From a linguistic perspective, the core concepts such as “Horizons of Expectation” and “Appeal Texture” in the Reception Theory hold important guiding significance in poetry translation. The translator needs to pay attention to the language cognition of the target language readers, that is, "Horizons of Expectation" of the readers. In word choice, one should avoid using obscure expressions that exceed the readers’ comprehension range. Secondly, in terms of rhythm, it should conform to the rhythm habits of the target language poetry. Finally, when converting cultural-specific images at the linguistic level, there should be space for readers’ imagination. This ensures that the translation not only achieves the accuracy of poetry translation but also enables readers to fill in the “blanks” of the text with their own language experience and generate resonance, thereby achieving the effective acceptance of poetic language art in cross-cultural communication. The following are several representative examples.

Example 1 Qiangjinjiu

Xu’s version: Invitation to Wine

AI’s version: Bringing in the Wine

In the connotation expressed by the original poem, it can be inferred that the character Qiang here means “please” or “wish”. The core of the poem is that the poet urges and invites the drinker with a heroic tone while the translation by AI focuses on expressing the act of carrying the wine in, without reflecting the bold atmosphere and emotional tension of the original poem. From the perspective of the “Horizons of Expectation” and “Implied Reader” concepts of the Reception Theory, the translated version of human translation meets the expectations of the target readers, making it easier for readers to accept literary works in a foreign language, and providing readers with a better reading experience. However, the translation produced by AI does not largely align with these two concepts.

Example 2 天生我材必有用

Xu’s version: Heaven has made us talents, we’re not made in vain.

AI’s version: Heaven has endowed me with talents for eventual use.

From the perspective of the Reception Theory, firstly, the literal expression of “talents” and “not made in vain” in the manually translated version reduces cultural barriers and largely meets the readers' existing “expectations”. Secondly, the formation of logical pauses through sentence structures stimulates the readers' initiative to fill in the meaning blanks. Thirdly, by using “we”, the individual perspective is expanded into a group, forming a dynamic communication between the creator and the recipient. Whether it is the “Horizons of Expectation”, “Appeal Texture”, or “Implied Reader”, the manually translated version has taken into account the reader’s reading experience very well and is in line with the three core concepts of the Reception Theory.

3.2. Cultural aspect

Ancient Chinese poetry serves as a carrier of culture, embodying the history, customs and the ways of thinking of the Chinese nation. Readers from different cultural backgrounds have significant differences in their understanding of cultural images and historical allusions in poetry. Poetry translation is not merely a linguistic transformation but also a dialogue between cultures. Whether it is AI translation or human translation, it should not only consider the cultural acceptance ability of readers but also accurately convey the cultural connotation contained in the poetry. From the perspective of Reception Theory, readers’ personalized interpretation of the text is based on their accurate understanding of the meaning of the poetry. First, the translator needs to provide supplementary information to help readers understand what the poetry is about, and then readers can fill in the “blanks” left by the creator. Guided by Reception Theory and aiming to accurately convey cultural information, poetry can truly be understood and appreciated in different cultures. The following are several representative examples.

Example 1: 岑夫子,丹丘生

Xu’s version: Dear friends of mine

AI’s version: Master Cen, Master Danqiu

Mr. Xu’s translation achieves the adaptive transformation of culture through the phrase “Dear friends of mine”. It was translated from the perspective of interpreting Li Bai and his relationship with them. Even readers who have no knowledge of historical allusions could understand who Cenfuzi and Danqiusheng were, thereby reducing the difficulty of cross-cultural acceptance and better corresponding to readers’ horizon of expectations regarding character identity. “Master Cen, Master Danqiu” merely retains the exclusive titles without providing a clear explanation of the identities of Cenfuzi and Danqiusheng. From the perspective of Reception Theory, AI translation does not take into account the readers’ cultural cognitive level. Xu’s version emphasizes clarity, which facilitates readers’ comprehension and better reflects the core concepts of Reception Theory.

Example 2:五花马

Xu’s version: Flower-dappled horse

AI’s version: Five-flowered horse

From the perspective of Reception Theory, Xu’s translation met the two concepts of “Horizons of Expectation” and “Implied Reader”. It achieves a better balance between cultural connotation transmission and reader acceptance. By explaining and elaborating on the core image of Wuhuama, it retains the aesthetic beauty of the poem while allowing readers to actively imagine what a horse with flower-like patterns looks like. However, the AI translation was limited to literal information and lacked the corresponding cultural reference in English. This could easily confuse readers about the specific meaning of “five kinds of flowers” and did not fully take into account their receptivity, resulting in an inferior reading experience compared with Xu’s version.

3.3. Emotional aspect

The Reception Theory emphasizes that readers’ understanding and response to a text are the key to realizing the meaning of the text. As poetry is a carrier of highly concentrated emotions, one of the core tasks of its translation is accurately convey the emotions contained in the original poem to the target language readers, enabling them to have a similar emotional experience as the readers of the source language. Poetry translation should not only preserve the emotional characteristics of the original poem but also adapt to the emotional acceptance habits of the target readers. Only when the target readers can accurately capture the emotional information of the original poem within their own emotional cognitive framework can the poetry translation truly achieve the effective interaction between the text and the readers as pursued by the Reception Theory. The following is a representative example.

Example:人生得意须尽欢

Xu’s version: When hopes are won, oh! Drink your fill in high delight.

AI’s version: When life is going well, one must enjoy the moment to the fullest.

Xu’s version conveys a strong sense of joy directly through “oh!” and “high delight”. The short sentence structure and the use of exclamation words simulate the colloquial and cheerful rhythm of the original poem, allowing English readers to quickly grasp the excitement of indulging themselves to the fullest upon the fulfillment of their desires. It closely approximates the emotional intensity of the original poem while evoking the readers' emotions, enabling a dynamic exchange between the readers and the creator through the translation. However, the AI version’s emotional expression leans towards rational admonition, with weaker emotional tension. It only completes the translation process but fails to complete the process of allowing readers to participate in the communication with the author. AI translation cannot mobilize readers’ emotions, nor can it evoke the same level of excitement as the author intended. To a large extent, this affects the readers’ reading experience and understanding.

4. The advantages and disadvantages of AI translation and human translation

4.1. AI translation

Through extensive data training, AI has not only mastered a vast amount of poetic information, but also grasped the corresponding conversion rules of vocabulary and sentences, enabling it to quickly provide poetry translations to meet the initial understanding needs of ordinary readers [8]. For example, as mentioned earlier, when AI translates specific cultural terms like Cenfuzi and Danqiusheng, it simplifies the explanation of historical allusions and uses “master” to conform to the language habits of English readers, leaving some space for imagination and creating “blank spaces” for readers. This allows readers to develop the desire to conduct further research on the poetry, which is the value of the poetry translation. Additionally, for readers with higher requirements for understanding poetry culture, AI can also generate personalized translations based on their needs and preferences, meeting the reading expectations of readers with different reading requirements.

Most ancient Chinese poems have a mixture of long and short lines, with a strong sense of rhythm. However, AI translation often disrupts the rhythm of the original poem while striving for accurate meaning [9]. Based on the research of Reception Theory in this article, the AI translation version of “Heaven has endowed me with talents for eventual use.” although it conveys the meaning of the original poem, loses the poetry's recitation nature and fails to meet readers' expectations for the rhythm and rhyme of the poem. Secondly, the direct translation phenomenon of AI translation largely affects readers' understanding of the poem and their reading experience. Cultural terms like Wuhuama lack necessary explanations and guidance in AI translation. Once readers have misunderstandings about the text, their interpretation thinking will be greatly affected, and the “blank spaces” reserved by the creator cannot be filled through readers' reading. More importantly, the training of AI mostly focuses on the corresponding conversion of words and sentences, and is difficult to capture the emotional needs of both the creator and the readers. For example, the translation of “When life is going well, one must enjoy the moment to the fullest.” in AI translation version is more like a rational suggestion, simplifying the boldness and freedom that the author intended to convey through the poem. The creator's emotions are weakened through AI translation, and when readers read the translation, it is difficult for them to evoke emotions and have an effective dynamic communication with the author.

4.2. Human translation

Manually translated interpreters will fully consider the rhythm and cadence of the poem during translating, fully reproducing the formal beauty of the original poem [10]. For example, in the translation of “Heaven has made us talents, we’re not made in vain.”, the translator splits the entire sentence into two separate lines, creating a logical pause, leaving space for readers to think and understand while maintaining the structural beauty of the original poem [11]. Secondly, manually translated interpreters will fully consider the cognitive level and acceptance habits of the target language readers, thereby making the created translation more acceptable. For instance, when translating Cenfuzi and Danqiusheng, Mr. Xu's translation version explains the relationship between the two and the poet Li Bai, avoiding the readers' confusion about the meaning of the poem while popularizing ancient literary knowledge to the readers. Most importantly, manually translated interpreters can deeply understand the cultural background, creative intention and emotional connotation of the poem, and accurately convey them on both the language and cultural levels to achieve emotional resonance between the creator and the reader. Manually translated versions can well balance the transmission of culture and the acceptance of readers, shorten the distance between readers and the original author, and better meet the requirements of the Reception Theory that centers on readers and emphasizes dynamic communication.

In human translation, the translator's understanding of the work directly influences the direction of the translation. Therefore, the translator needs a considerable amount of time to study and interpret the original poem, thereby producing a translation that closely matches the original work. Thus, the efficiency of human translation is relatively low. Secondly, the translator's interpretation also largely affects the reader's understanding. For example, in the poem, Shengxian are translated by Xu Yuanchong as "sages", which mainly refers to a person full of wisdom. However, Shengxian in the original poem imply the cultural connotation of “those who cultivate their moral characters and put their family affairs in order” in Confucianism. This simplification reduces the difficulty of understanding but also narrows the interpretation space for the word Shengxian for English readers, limiting the extension of meaning during the acceptance process. This is not a denial of Mr. Xu Yuanchong's translation version, but rather a common limitation of human translation in cross-cultural communication. Human translation is inherently limited by the translator's cognition and the cultural differences between the target language and the source language. Different versions of translation often have different styles due to the individual translator's preferences. If the translator's interpretation or translation style differs greatly from the “Horizons of Expectation” or acceptance level of the readers, it will widen the gap between the two cultures and prevent the effective dissemination of culture.

4.3. The reasons for the differences between the two translation methods

The differences between human translation and AI translation stem from the differences in thinking logic. Human translation relies on cultural accumulation and subjective understanding, excelling in capturing the deep meanings and emotional connotations, and being able to predict the “Horizons of Expectation” of the target language readers. AI translation, on the other hand, is based on big data, focusing on the processing of surface information, but has difficulty handling cultural metaphors and is prone to fall into patterned expressions. Therefore, it is necessary to form a “human-led, AI-assisted” translation model [12]. AI handles the initial translation, while humans control the rhythm and emotional tone of the poetry, taking the reader's acceptance as the core, reducing cultural barriers. Different versions of translations are generated for different readers, with ordinary readers focusing on readability, supplemented by cultural annotations to lower the threshold, and professional readers focusing on creativity, leaving sufficient space for interpretation. The combination of the two balances cultural transmission and reader acceptance, achieving effective transmission of cross-cultural texts [12]. This study is based solely on Mr. Xu Yuanchong's translation versions, so further in-depth research is needed on the influence of different translators' translation styles on readers' understanding. In the future, resolving the subjective interpretation of poetry by translators and developing a human-machine collaboration translation model are key to promoting the international dissemination of poetry.

5. Conclusion

This study reveals that while AI translation tools demonstrate significant advantages in output efficiency, they still have functional deficiencies. AI translation focuses on efficiency as its core advantage, but generally lacks humanistic care, whereas human translation, although less efficient than machines, can accurately convey emotional connotations, taking into account the readers' acceptance level and ensuring the accuracy of understanding and expression. The main reason for the difference between the two is the difference in thinking logic. Human translation relies on subjective understanding, which can capture the deep emotional connotations and convey them, while AI translation relies on big data and pays more attention to the processing of surface information, and is difficult to understand cultural metaphors. Taking Qiangjinjiu as an example, after in-depth study and research on Reception Theory, this paper concludes that the core of translation creation lies in the in-depth understanding of cultural connotations, with efficiency being placed in a secondary position. This indicates the irreplaceability of human translation in the translation of cultural texts. In the translation industry, as various AI translation tools continue to develop, people's reliance on them is increasing day by day. However, the defects of AI translation may lead readers from different cultural backgrounds to overlook the true connotations behind literary works, which is not conducive to the complete transmission of cultural information and the realization of literary value. Therefore, the key to changing this phenomenon lies in awakening people's attention to the emotional transmission in translations and the readers' acceptance level, and clarifying the auxiliary tool positioning of AI translation. However, in the fast-paced modern society, there is an objective demand for translation efficiency. How to balance the emotional transmission in human translation and the efficiency and convenience of machine translation in practice, and achieve the complementary advantages of both, remains an important topic that the translation industry needs to deeply study in the future. This article only selected Xu Yuanchong’s translation version as the representative of human translation. However, the translation style of poetry varies from person to person, so this article still has limitations. Future research can select more human translation versions as samples to conduct a more in-depth discussion.