Influence of Short Novels on Creation of Educational Programs in Literature: Taking A.P. Chekhov’s “The Chameleon” and Lu Xun’s “A Madman’s Diary” as Examples

Abstract

This study explores how artificial intelligence (AI) technologies can be theoretically integrated into literature curriculum development, using the works of Anton Chekhov and Lu Xun as illustrative case texts. The aim is to reduce barriers to language and cultural understanding in literature education and increase the efficiency and accessibility of cross-cultural teaching. We used natural language processing (NLP) techniques to analyze textual features, such as readability index, lexical density, and syntactic complexity, of AI-generated and human-translated “The Chameleon” and “A Madman’s Diary”. Teaching cases from universities in China, Russia, and Kazakhstan are reviewed to assess the emerging practice of AI-supported literature teaching. The proposed theoretical framework draws on hermeneutics, posthumanism, and cognitive load theories. The results of the data-driven analysis suggest that AI-assisted translation tends to simplify sentence structure and improve surface readability. While anecdotal classroom observations highlight the role of AI in initial comprehension, deeper literary interpretation still relies on teacher guidance and critical human engagement. This study introduces a conceptual “AI Literature Teaching Model” that positions AI as a cognitive and cultural mediator and outlines directions for future empirical validation.

1. Introduction

This study explores the value of Chekhov and Lu Xun’s works in the design of contemporary literary education programs, particularly their potential contributions to promoting cross-cultural understanding and selecting appropriate teaching tools in the context of artificial intelligence technology. This study is a theoretical investigation and does not involve teaching interventions or empirical experiments. As representatives of the Russian and Chinese literary traditions, Chekhov and Lu Xun have long occupied central positions in comparative literature and world literature programs. Their works are rich in thematic expression, artistic style, and cultural symbolism, but it is precisely this richness that often poses comprehension challenges for non-native readers. For example, Lu Xun’s “A Madman’s Diary” is characterized by symbolic and critical language, requiring profound cultural insight and interpretive skills. “I read ‘cannibalism’ between the lines!” This statement reveals a metaphorical depth that may be difficult to grasp without prior contextual and cultural background. Similarly, Chekhov’s “The Chameleon” uses formal courtly speeches and indirect dialogue to reveal the moral and social contradictions of Tsarist Russia, narrative features that pose challenges for second-language learners in terms of semantic and social understanding. These complexities highlight the need for AI-assisted text simplification and semantic annotation mechanisms in cross-cultural literary education to facilitate deeper engagement and understanding.

Current research has generally focused on the artistic value and literary comparisons of Chekhov and Lu Xun’s works, but systematic analyses of their teaching methods, students’ understanding pathways, and textual challenges remain limited. Traditional classroom instruction, particularly in literary education, typically relies on teacher-centered lectures and group discussions. While this approach is effective for conveying universal content, it often lacks mechanisms to address individual differences in students’ cognitive processing and language abilities. As observed in the curricula and textbooks of leading institutions such as Peking University and Al-Farabi Kazakh National University, most literature courses maintain a fixed text structure and uniform reading pace, with limited differentiation strategies. This observation aligns with existing research, indicating that traditional teaching methods cannot adequately accommodate learners with varying levels of language preparation and interpretive abilities (Kasneci et al., 2023). The ability of natural language processing (NLP) tools to analyze text readability, lexical density, and syntactic complexity is critical to enhancing communication and understanding across languages. Natural language processing (NLP) techniques, including machine translation systems, play an important role in simplifying complex texts, thus facilitating cross-language understanding. However, most of the research remains at the level of language training, and there is a lack of discussion on the in-depth reading of literature, decoding of cultural symbols, and optimization of teaching curricula. Whether AI has the ability to interpret culture and whether it can be transformed from a technological tool into an “interpreting collaborator” are important propositions that need to be answered by the theoretical community.

This study introduces an integrative perspective in theoretical construction, aiming to provide a solid philosophical and pedagogical foundation for the “AI-enabled literature teaching model”. Literature itself is rich and diverse, but its complex language and cultural context can pose challenges for learners, especially in multilingual and multicultural classrooms. In such cases, AI tools can serve as cognitive mediators, helping students cope with linguistic complexity, reduce cognitive load, and understand deeper cultural meanings, thus demonstrating the validity of their application in literary education. In the hermeneutic dimension, Gadamer’s concept of “convergence of perspectives” emphasizes that comprehension is not a simple reproduction of textual meaning but a dynamic process of meaning-making between the reader and the text (Gadamer, 1960/2004). When this concept is introduced to the pedagogical field of AI intervention, AI tools are no longer just carriers of linguistic information but interpretive mediators capable of participating in contextual reconstruction and the negotiation of meaning (Floridi, 2014).

Posthumanist critical theory provides a more subjective interpretive foundation for the role of artificial intelligence in education. Latour’s theory emphasizes that technological systems are an indispensable part of knowledge construction, playing the role of “non-human agents” in Actor–Network Theory (ANT) (Hendrickx, 2023). This perspective challenges the traditional view that knowledge is co-constructed through interactions between humans and non-humans. This provides theoretical legitimacy for the integration of AI into the interpretive chain and its collaborative role in the interpretive process. Within this framework, AI is no longer merely a “tool-like entity” but is repositioned as a participant capable of jointly understanding and reinterpreting literary texts with teachers and students.

The cognitive load theory in cognitive psychology (Sweller, 1988) provides an empirical framework for the role of artificial intelligence in optimizing teaching. This theory suggests that learners are more likely to achieve deep learning if they are able to reduce their cognitive load through structured and simplified information input when faced with complex tasks. As a corollary, we point out that in the teaching of literature, the techniques of readability optimization, semantic reconstruction, and automatic text summarization provided by AI may help students to transition from the stage of “language comprehension” to the stage of “cultural understanding (students’ ability to connect literary content with specific socio-cultural contexts)” and “critical interpretation (analysis and construction of textual metaphors, polysemous words, and underlying ideologies)”. This study analyzes the readability of English translations of Chekhov’s and Lu Xun’s works using natural language processing tools and quantifies text difficulty indicators, including the Flesch Reading Ease Index (Flesch, 1948), average sentence length, vocabulary density, and analogous example ratios, and further compares human translations and artificial intelligence translations based on text features. Additionally, a comparison is made between human translation and AI translation texts in terms of language complexity and cultural referential retention to preliminarily validate the applicability of the “translation simplification hypothesis” and establish boundaries for the application of AI translation in education (Baker, 1996). Furthermore, this study uses literary teaching cases from Peking University, Moscow State University, and Al-Farabi Kazakh National University as examples to analyze the impact of AI technology on course design, teaching feedback, and text comprehension in different contexts, providing a practical data foundation for model construction.

This study proposes a conceptual framework for “intelligent hermeneutics”, aiming to describe how artificial intelligence can play a dynamic role in the interpretation of literary texts. This framework is grounded in Gadamer’s hermeneutic tradition of “fusion of horizons” and integrates post-humanist theoretical assumptions regarding the role of “non-human actors” in knowledge production (Gadamer, 1960/2004; Hendrickx, 2023), thereby offering a new interpretive pathway for AI’s involvement in literary education. According to this framework, the intervention of AI in semantic simplification and language transformation not only reduces the surface difficulty of the text but also extends the students’ path of understanding through human–computer interaction, thus realizing the expansion of the subject of interpretation. This viewpoint breaks through the traditional hermeneutic qualification of “interpreter = human” and provides theoretical support for the technological modernization of the literature curriculum (Gadamer, 1960/2004). This study emphasizes that the educational value of Chekhov’s and Lu Xun’s works is embodied not only in the content itself but also in how they drive the creation of new teaching tools, new teaching models, and even new concepts of curriculum development. It responds to the core issues of how to reconstruct literary interpretation and balance technological intervention and humanistic depth in the age of AI and provides theoretical support and practical paths for the development of intercultural education and intelligent teaching. This study investigates how AI technologies can assist in reducing linguistic complexity and facilitating cross-cultural understanding in literature education.

The remainder of this paper is organized as follows: Section 2 reviews the relevant literature on comparative literature pedagogy and the integration of artificial intelligence in literary education. Section 3 introduces the research design, text selection, and analytical framework. Section 4 presents the results of the textual and readability analyses, while Section 5 provides a discussion of the findings in relation to cross-cultural interpretation and pedagogical applications. Section 6 outlines the limitations of this study and suggests directions for future research. Finally, Section 7 offers concluding remarks.

2. Literature Review

The literature review of this study comprehensively analyzes the current research findings related to the complexity of intercultural literature teaching, the practice of comparative literature pedagogy, and the application of artificial intelligence (AI)-assisted teaching, aiming at sorting out the interactive mechanisms between cognitive load, cultural context, and technological support in literature education. It has been shown that the challenges facing literature teaching involve multiple dimensions, which include the combined cognitive load of students in terms of language cognition, cultural understanding, and text reception. This complexity is not only reflected in curriculum design and instructional organization but is also notable in the pathway barriers to students’ construction of meaning in texts. Harrington (1994) points out structural dilemmas in the development of instructional materials through an exploration of case-based instruction.

In the current context of intercultural education, the comparative literature teaching method has been gradually established as an important educational strategy, the basic concept of which is to stimulate students’ cognitive transfer, cultural understanding, and critical thinking skills through the reading of texts in different cultural contexts. It has been shown that this method not only helps students to broaden their horizons of literary traditions but also strengthens their cultural awareness and independent analytical skills through comparisons of differences (Kholodniak, 2023; Tattimbetova et al., 2020). In specific pedagogical practices, students create structural connections between texts by analyzing elements such as character, plot, and cultural context, thus achieving a more in-depth path of literary interpretation. The literature also points to the unique advantages of the comparative literature approach in promoting active student engagement.

Reader-responsive comparative teaching, in particular, emphasizes students’ ability to connect their own sociocultural experiences to textual contexts, forming networks of understanding across multiple perspectives, thereby enhancing their ability to reconstruct text and generate meaning (Parsaiyan, 2020). This teaching method helps to develop students’ cultural self-awareness and self-orientation and lays the foundation for their literary understanding in the context of globalization. In a cross-cultural literature course at Al-Farabi Kazakh National University, students were asked to analyze the theme of alienation in Chekhov’s “The Chameleon” and Lu Xun’s “A Madman’s Diary” and connect it to their own experiences with authority or social identity. This enabled them to interpret the texts not only from a literary theoretical perspective but also by integrating their personal and cultural backgrounds. Comparative literature pedagogy also faces a series of structural challenges in practice. The persistence of Eurocentrism in Western contexts has marginalized the status of non-Western literature in the curriculum, limiting the breadth of students’ awareness of cultural diversity (Kaiser, 2020). The central role of translation in cross-language textual comparison is often underestimated, and translation itself is an act of cultural re-construction whose intervention inevitably affects students’ understanding of the context and cultural depth of the original work (Thow, 2022). The intervention of AI technology provides a new methodological opportunity to solve the above problems.

AI can reduce the processing difficulty of literary texts and assist students in understanding the linguistic logic and cultural context of in-depth texts by means of semantic analysis, discourse reconstruction, and visual modeling (Hutapea et al., 2024). Generative AI is gradually changing the mode of interaction between teaching and learning. Research has shown that AI can customize content based on individual student differences, enhancing learning engagement and comprehension (Simelane & Kittur, 2024). In educational practice, teachers generally view artificial intelligence as an auxiliary tool for drafting and revising textbooks, with functions similar to those of a calculator in mathematics education, aimed at enhancing learning efficiency rather than replacing cognitive processes. However, some scholars have pointed out that the widespread application of artificial intelligence may weaken students’ originality and creativity, and caution is needed regarding the dependency issues it may introduce into teaching. Nevertheless, some studies have also pointed out that while AI performs exceptionally well in language learning, it still faces limitations in handling the emotional logic, cultural metaphors, and narrative tension found in literary discourse (Zhang & Sun, 2023; Wei, 2025). Although artificial intelligence can provide cognitive assistance, its ability to interpret cultural metaphors remains limited (Zawacki-Richter et al., 2019).

Hermeneutic theory and posthumanist critique (Braidotti, 2016) provide key theoretical foundations for understanding the new role of AI in teaching and learning. Pastra suggests that the combination of language and images helps AI to mimic human comprehension mechanisms, an idea that has been referred to as the “dual visual-linguistic grounding theory” (Pastra, 2004). Its integration into educational environments provides personalized learning experiences, enhances the role of the teacher, and promotes the critical engagement of students (Jantanukul, 2024). Integrating AI and human teachers into educational action networks emphasizes the potential of collaborative interpretation to improve educational outcomes through the synergy of data-driven insights and human expertise. Artificial intelligence can provide real-time analytics and personalized feedback, enabling teachers to focus on fostering critical thinking and creativity in students, leading to increased engagement and achievement. Cognitive load theory also provides an important explanatory path for educational interventions with AI. Researchers have pointed out that students often face the problem of information overload and comprehension difficulties when facing literary texts with unfamiliar contexts and complex grammatical structures (Packer & Keates, 2023; Halkiopoulos & Gkintoni, 2024). AI tools reduce students’ cognitive load while enhancing their willingness to participate and comprehension ability by simplifying sentences, extracting keywords, visualizing structures, etc. (Araújo & Aguiar, 2023). However, some scholars have questioned whether over-reliance on AI may inhibit students’ critical thinking skills (Munnik & Noorbhai, 2024; Elstad, 2024) and have suggested that the use of AI must remain an “adjunct, not a substitute” in teaching strategies.

The “Translation Simplification Hypothesis” (Baker, 1996) argues that translated texts usually exhibit lower linguistic complexity compared to the original, especially in terms of lexical diversity and syntactic structure. Chen and Chang (2023) point out that although translated texts may have longer utterances, expressions aimed at enhancing the clarity of the message rather reduce the overall readability. Kajzer-Wietrzny et al. (2016) emphasize that stylistic simplification techniques, such as the splitting of long sentences, which are common in translation, contribute to the comprehensibility of texts. Educational research has shown that simplified texts (e.g., the Easy Readers series) enhance language learners’ reading fluency and motivation, despite the risk of weakening meaning. Simplifying syntax also helps to reduce cognitive load, especially in comprehension-oriented teaching situations (Xu & Liu, 2023). Simplification strategies have practical value in teaching and learning, but they also need to be balanced between readability and the preservation of the complexity of the original text.

On the basis of the aforementioned research, the pedagogical study of the works of Chekhov and Lu Xun is of particular importance. As iconic writers in the Russian–Chinese literary tradition, their works are characterized by a high degree of cultural symbolism, linguistic metaphor, and narrative density and are often regarded as “difficult texts” in world literature courses. Chekhov’s plays and novels, on the other hand, are adept at carrying deep emotional conflicts through “plain tones”, with slow but suggestive narrative rhythms, and their dialogic structure poses a unique challenge for second-language students (Axelrod, 2016).

Although studies have explored the literary value of these two authors from the dimensions of artistic style, realism tradition, and character writing, there is still a gap in the systematic research on their readability, cultural translation, and pedagogical adaptation in the teaching context, especially under the conditions of AI technology intervention. Based on this, this study combined natural language processing, AI-based translation tools, and hermeneutic theory to respond to the cultural disconnection, semantic ambiguity, and cognitive load of Chekhov and Lu Xun’s works in multilingual teaching and explore the boundaries of AI’s function as a collaborator in cultural interpretation and its value in teaching.

The literature review highlighted different perspectives on pedagogical challenges, methodological approaches, and technological tools in literature education. While significant progress has been made in comparative pedagogy, artificial intelligence applications, and cognitive support, efforts to integrate these elements into a coherent pedagogical framework remain limited. This study addresses this gap by focusing on the works of Chekhov and Lu Xun to develop a design model that combines AI techniques, comparative literary approaches, and key theoretical underpinnings with the aim of supporting students’ linguistic comprehension and cultural interpretation in cross-cultural literary learning.

3. Methodology

3.1. Research Design

This study centers on the core issue of “how AI can support students in navigating the complexities of literary language and cultural meaning”. This study adopted a combination of data-driven text analysis and theoretical modeling to explore the feasibility and pedagogical potential of AI tools in comparative literature teaching, selected two representative Chinese and Russian short works as case texts, and combined natural language processing technology and teaching practice data from many countries to construct a theoretical model of “AI-enabled literature teaching”. A theoretical model of “AI-enabled literature teaching” was constructed. This study did not involve student samples or teaching intervention experiments and was a non-empirical theoretical exploration.

3.2. Sampling

This study adopted a theoretical sampling strategy and selected Chekhov’s “The Chameleon” and Lu Xun’s “A Madman’s Diary” as the analyzed texts. These two works are highly representative in terms of language structure, cultural density, and literary style and are often used in cross-cultural literature courses in Chinese and Russian universities. The choice of these texts was based on their having strong representativeness and research value and it being easy to create comprehension barriers in teaching scenarios.

3.3. Research Instruments and Data Collection

The text processing in this study utilized natural language processing (NLP) programs developed using Python (version 3.10), combined with AI translation results generated by Google Translate and ChatGPT (OpenAI, GPT-4.0 version), for comparison with existing professional human translations. Readability metrics included the Flesch Readability Index, average sentence length, vocabulary density, and Type-Token Ratio (TTR). Additionally, a hybrid method combining automatic keyword extraction with human verification was used to analyze metaphor density and symbolic expression frequency. In this study, symbolic expression frequency was defined as lexical or phrasal elements carrying metaphorical, archetypal, or allegorical meanings (e.g., “madman,” “dog,” “chameleon,” “eat people,” “blood,” “iron gate”). The frequency of symbolic expressions was standardized per 1000 words to enable comparisons across texts of varying lengths. The teaching practice data came from officially published course outlines, teaching case collections, and teaching design papers from institutions such as Peking University, Moscow State University, and Al-Farabi Kazakh National University. These secondary materials were obtained through institutional knowledge bases and academic databases (including CNKI, SpringerLink, ERIC, and Google Scholar). The data was used to demonstrate the practical application of artificial intelligence tools in multilingual literary education and support the construction of theoretical models. This study did not collect any primary classroom data (such as student feedback or teaching interventions) and did not involve empirical experiments.

3.4. Data Analysis

The data were analyzed using quantitative linguistic indicators and cross-text comparison methods. By quantifying the readability of four texts (human translations and AI translations of two works), we compared the differences in the performance of AI translations in terms of syntactic simplification, lexical adjustment, and symbolic information retention. Syntactic depth levels were calculated using dependency parsing methods. Each sentence was analyzed to generate a syntactic tree structure, where depth was defined as the maximum number of levels from the root node to the terminal node. Subsequently, the average syntactic depth of all sentences was calculated for each translation. This metric was used to assess the degree to which syntactic complexity was retained or simplified in AI-translated text. The teaching cases were analyzed using content analysis to sort out the pedagogical interventions and feedback methods of the AI tools. All the analysis results were used to construct the structural logic of the “AI-assisted Literature Teaching Model” and suggest teaching paths.

4. Results

4.1. Textual Analysis

By comparing the language features of the manually translated and AI-translated versions, the changes in the syntactic structure, vocabulary distribution, and style density of Chekhov and Lu Xun’s works in different translations were presented. Natural language processing tools showed that the intervention of artificial intelligence reduced the structural complexity and vocabulary difficulty of the text to a certain extent, but may have caused simplification and the loss of some style information. The text analysis did not involve student feedback or learning effect evaluation. It aimed to provide quantifiable language complexity parameters for course design and a data support basis for future adjustments to teaching strategies.

In analyzing Chekhov’s “The Chameleon” and Lu Xun’s “A Madman’s Diary”, we paid great attention to the inherent differences in the linguistic structures of the two writers and the cultural tensions that these differences expressed at the level of translation. The degree of linguistic difficulty was quantified by automatically calculating the readability index (using the Flesch–Kincaid Grade Level) and indicators such as average sentence length and average word length for each text through NLP tools. These key dimensions formed the core framework of our qualitative analysis. This technical route not only revealed the stratification of the two texts in terms of linguistic complexity but also provided an understanding of how we could further contrast the human translations with the AI translations, exploring the performance of automated translation in simplifying linguistic structures and reorganizing stylistic features. The resulting comparison was not only about understanding the “importance” of the text itself but also about the boundaries and possibilities of AI intervention in literature. This study attempted to construct a data-driven framework of textual and linguistic parameters for cross-cultural literature teaching to assist teachers in more accurately grasping the distribution of textual difficulty levels and linguistic features in the organization of teaching content and task design.

To quantify the readability of the text, we used the Flesch Reading Ease Score formula:

$$\mathrm{FRES}=206.835-1.015\times ({\displaystyle \frac{\mathrm{Total}\mathrm{number}\mathrm{of}\mathrm{words}}{\mathrm{Total}\mathrm{Sentences}}})-84.6\times ({\displaystyle \frac{\mathrm{Total}\mathrm{number}\mathrm{of}\mathrm{syllables}}{\mathrm{Total}\mathrm{number}\mathrm{of}\mathrm{words}}})$$

The formula determines reading difficulty based on the average number of words per sentence and the average number of syllables per word. The higher the index, the easier the text is to read, with 90–100 generally considered extremely easy to read (3rd–4th grade level), 80–90 easy to read (5th–6th grade), 70–80 moderate (7th–8th grade), 60–70 difficult, and under 50 for college level and above.

The average sentence length (ASL) was calculated as

$$\mathrm{ASL}={\displaystyle \frac{\mathrm{Total}\mathrm{number}\mathrm{of}\mathrm{sentences}}{\mathrm{Total}\mathrm{words}}}$$

Vocabulary richness was expressed as a Type-Token Ratio (TTR), which was calculated as follows:

$$\mathrm{TTR}={\displaystyle \frac{\mathrm{Different}\mathrm{vocabularies}\left(\mathrm{Types}\right)}{\mathrm{Total}\mathrm{vocabulary}\left(\mathrm{Tokens}\right)}}$$

TTR reflects the diversity of vocabulary, with higher levels indicating a richer vocabulary. It is generally recognized that below 30% is considered to be more repetitive vocabulary, while above 50% is considered to be more linguistically diverse.

These formulas provided us with the basic tools to quantify the language difficulty so that the text was not subjectively judged by perception alone, but a clear readability profile was established through indicators such as syntactic length, lexical density, and syllable composition. In order to further verify the applicability of this approach in literary texts, we selected the human-translated and AI-translated versions of Chekhov’s “The Chameleon” and Lu Xun’s “A Madman’s Diary” for processing and compared their respective linguistic complexity performances (

Table 1. Text complexity comparison (human vs. AI translation).

Metric	“The Chameleon” (Human Translation)	“A Madman’s Diary” (Human Translation)	“The Chameleon” (AI Translation)	“A Madman’s Diary” (AI Translation)
Readability Index (Flesch–Kincaid)	82.3	75.0	85.1	78.4
Average Sentence Length (words)	13.7	20.2	12.1	18.5
Metaphor Frequency (per 1000 words)	35.3%	24.5%	32.0%	22.1%
Symbolic Expression Frequency (per 1000 words)	3.5	4.8	2.9	4.2
Type-Token Ratio	35.3%	24.5%	32.0%	21.5%
Complex Word Ratio (%)	11.4%	17.6%	9.2%	14.0%
Syntactic Depth Level (AI translation)	2.1	2.9	1.8	2.4
Average Sentence Length (words)	4.8	5.1	4.5	4.7

). This comparison not only presented the trend of AI translation in simplifying language structure but also provided data support for subsequent instructional design, helping us to more accurately identify the applicability of different translations in teaching practice. In addition to the basic indices such as readability index and sentence length, this study also introduced a number of linguistic parameters used to reveal the structural and rhetorical complexity of the text in order to construct a more hierarchical text analysis system. Among them, the frequency of symbolic expression was based on the identification of typical metaphors and cultural imagery, and its frequency of occurrence was counted per thousand words; the proportion of complex words was defined by words with three or more syllables, reflecting the lexical density and cognitive load; and the syntactic depth hierarchy analyzed the number of nested layers of sentence structure by using the tools of natural language processing to measure the text’s complexity in terms of grammatical organization. These metrics complement the traditional readability metrics and are more relevant to the stylistic characteristics of literary texts in cross-cultural teaching and learning.

It is worth noting that Type-Token Ratio (TTR) is also added to Table 1 as a measure of lexical diversity. The analysis showed that the TTR of AI translations was generally slightly lower than that of human translations, reflecting the tendency of AI texts to compress lexical scope in the process of simplifying language structure. This feature not only echoes the overall direction of syntactic simplification but also provides a possible quantitative basis for the subsequent grading of instructional materials among students with different language abilities.

The English translations of Chekhov’s “The Chameleon” and Lu Xun’s “Diary of a Madman” showed a clear contrast in linguistic complexity. We measured the Flesch–Kincaid readability index, average sentence length, and word length using an NLP tool to quantify reading difficulty. The human translations of “The Chameleon” had an average sentence length of 13.7 words and a readability index of 82.3 (5^th-grade level), while Diary of a Madman had an average sentence length of 20.2 words and a readability index of 75.0 (8^th-grade level). The longer sentences and higher semantic density in Lu Xun’s works create greater comprehension barriers for non-native students, which may affect reading motivation and cultural engagement (Table 1). The comparison results show that the AI translations showed a further trend of simplification in terms of linguistic structure, a feature that may help teachers to set up reading tasks with a more hierarchical sense in their teaching.

To verify the potential of AI tools in reducing linguistic complexity, this study used Google Translate with a large-scale language model (e.g., ChatGPT) to generate AI translations of two works and analyze their linguistic structure with the same index. The average sentence length of the AI translation of “The Chameleon” dropped to 12.1 words and the readability index rose to 85.1. The average sentence length of the AI translation of “A Madman’s Diary” was 18.5 words and the index was 78.4 (Table 1). The results show that compared with human translators, the AI translations further simplified the language structure, reduced the reading complexity, and provided indexing technology support for explicit reading and layered comprehension in teaching.

Vocabulary richness likewise appears as a fundamental difference between the two writers’ styles. Lexical richness is an important indicator of linguistic complexity (Niu & Jiang, 2024). The Type-Token Ratio (TTR) of the translated text of “The Chameleon” was about 35.3%, which was significantly higher than that of “A Madman’s Diary”, which was 24.5%. The higher lexical repetition rate in Lu Xun’s text can be partly attributed to the length of the text, and additionally reflects the concentration of his linguistic style and the continuous advancement of his rhythm. In contrast, Chekhov’s language is dense, flexible, and jumpy, which is more in line with the stylistic feature of “short sentences + variations”. In addition, the two works also show significant differences in the use of metaphors: “A Madman’s Diary” has a high density of representational systems constructed from the imagery of “man-eating man”, moonlight, evil dogs, ghosts, and so on. “The Chameleon” presents its theme mainly through plot irony, with metaphors such as the store door opening “like a hungry mouth,” but, overall, the metaphors are less metaphorical and the language is more straightforward and humorous. These differences in contrast to language density are also clearly reflected in the TTR and average sentence length metrics in Table 1. We counted the frequency of metaphorical expressions using a computer program combined with manual proofreading to ensure the consistency of identification. Similar studies have shown that automatic metaphor recognition has achieved some accuracy, e.g., algorithms are able to distinguish metaphorical usage with about 71% precision (Neuman et al., 2013). NLP analysis confirmed that Lu Xun’s text was syntactically complex and longer, whereas Chekhov’s text was simpler and more accessible, which informed the adaptation of teaching strategies.

4.2. Translation Comparison and Simplification

To investigate the language simplification effect of AI translations, we used Google Translate and a GPT model to translate Chekhov’s “The Chameleon” (Russian translation) and Lu Xun’s “A Madman’s Diary” (Chinese translation) into English and compared them with authoritative human translations. Quantitative analyses of average sentence length, lexical complexity, and TTR showed that AI translations typically featured shorter sentences and more direct vocabulary. The AI translation of “A Madman’s Diary” reduced the average sentence length by about 15%, lowered the TTR, reduced the proportion of advanced vocabulary, and improved the Flesch readability score and the Flesch–Kincaid Grade Level by 1–2 levels. The above data tentatively confirms the performance of AI translators in language simplification, which is consistent with the “translation simplification hypothesis”. However, this increase in readability is usually accompanied by a compression of semantic details and a weakening of cultural imagery, and, therefore, its use in teaching scenarios should still be carefully designed to prevent the depth of teaching from being replaced by technological simplification. While AI-generated simplifications aid comprehension, they may weaken the cultural connotations of the text (Ennouari & El Housni, 2024). Four sets of typical sentences illustrate these changes in language structure and rhetorical expression (

Table 2. Comparison of human and AI translations at sentence level.

Original Work	Original Text	Human Translation	AI Translation	Simplification Notes
“The Chameleon”	Oн вышeл нa yлицy, нaдyвaя щeки и щypяcь нa coлнцe.	He went out into the street, puffing out his cheeks, squinting at the sun.	He went outside and looked up at the sun.	Deleted detailed description and list structure; retained only subject and action.
“The Chameleon”	Coбaкa, видимo, бeздoмнaя, cмoтpeлa нa ниx иcпyгaнными глaзaми.	The dog, apparently a stray, looked at them with frightened eyes.	The dog looked scared.	Removed the adjectives “apparently” and “frightened eyes” to neutralize the tone and reduced the semantic weight surrounding the noun “dog”.
“A Madman’s Diary”	月光像雪一样照在地上，我不敢出声，怕他们听见	The moonlight shone on the ground like a thin layer of frost, and the silence screamed louder than a thousand voices.	Moonlight lit the ground. It was very quiet.	Removed metaphor and personification, split the sentence into two, and weakened the imagery, thereby diminishing the poetic function of the language.
“A Madman’s Diary”	我知道他们全是吃人的；他们眼睛里全写着”吃人”两个字。	I have reason to suspect that every person I see is harboring cannibalistic thoughts.	I think people around me want to eat someone.	Weakened psychological inference and imagery; directly expressed subjective suspicion.

).

4.3. Case Studies in Teaching Literature

The literature courses of three universities in China, Russia, and Kazakhstan were selected to compare the integration of teaching practices and AI tools for Chekhov and Lu Xun in different cultural contexts (

Table 3. Comparative cases of AI-assisted literature teaching in different countries.

Country/Region	Institution	Teaching Content	AI Tool Application	Course Features and Feedback	Content Source Notes
China	Peking University	Chekhov’s Works (World Literature)	Exploration encouraged by national policy, not yet integrated into curriculum	Traditional research-focused approach, AI viewed as a potential reform direction	At Peking University, the pedagogical content reflects the national “smart education” policy, which incorporates AI classroom reform into the “classroom revolution” agenda.
Russia	Moscow State University	Lu Xun’s Works (Foreign Literature)	No classroom application observed, discussions remain at academic level	Textual detail, student use of translation tools for assistance	Research on the teaching of Russian literature highlights cross-cultural approaches and emerging considerations of AI integration and draws on educationally focused Russian academic journals.
Kazakhstan	Al-Farabi Kazakh National University	Chekhov and Lu Xun (Comparative Literature)	Use of Google Translate and ChatGPT for content generation	AI feedback included in teacher reflections, higher degree of openness	Kazakhstan’s education system is more open to AI, particularly in collaborative writing with AI support.

). In China, the world literature course at Peking University mainly uses authoritative Chinese translations and a traditional approach combining teacher lectures and student seminars, and AI tools such as ChatGPT have not yet been systematically introduced, but universities are exploring “classroom revolutions,” and policies are encouraging the use of AI to improve the effectiveness of teaching and learning (Qian et al., 2023). In Russia, the foreign literature program at Moscow University uses Russian translations of Lu Xun’s works and emphasizes the close reading of texts and cultural analysis. Although AI tools are not directly introduced in the classroom, related studies show that students often use translation software when understanding texts, and literature has explored the auxiliary functions of chatbots, corpora, and ChatGPT in language learning and independent reading, suggesting that the system is somewhat open to AI integration. The Comparative Literature course at the Al-Farabi Kazakh National University uses a multilingual strategy to teach Chekhov and Lu Xun, and the instructor uses Google Translate and ChatGPT to generate pedagogical questions and analyses. The experiment showed that AI-written feedback and teacher comments were equally effective in improving students’ writing efficiency (Bodaubekov et al., 2025). Although students usually have access to translations of Chekhov’s and Lu Xun’s works in their home languages (e.g., Chinese, Russian), in the context of multilingualism, especially in international courses or Sino–foreign cooperative programs taught in English, students are often required to read English translations in order to complete classroom tasks. Due to the high complexity of the syntax, rhetoric, and style of English-translated literary texts, students face a high cognitive threshold at the initial reading stage. In this context, AI-generated simplified translations can be used as teaching aids to help students establish a basic understanding of the plot structure and core meaning at the initial stage, thus reducing cognitive load and increasing engagement. Such simplified texts are not a substitute for traditional translations but rather provide language support for differentiated teaching and serve the pedagogical goals of hierarchical task development and multidimensional text interpretation. It is important to note that these findings are based on external teaching experiments and not on the teaching practices or classroom interventions carried out in this study. In general, the integration of AI in literature courses is still in its infancy, and the teaching contexts and degree of technology adoption vary from country to country. The purpose of this study was to provide a reference for teaching scenarios for the construction of an “AI-enabled literature teaching model” through horizontal case comparison, rather than to verify teaching effectiveness or students’ understanding.

4.4. Optimization of Instructional Design

It has been noted that simplified translated texts have the effect of reducing cognitive load in non-native language settings, helping students to quickly access the core of the discourse and build an initial framework of understanding (Begaliyeva et al., 2025). Based on the results of AI translation and NLP text analysis, this study proposes a set of auxiliary teaching design concepts for Chekhov and Lu Xun’s works, aiming to provide data support and tool application paths for cross-cultural literature teaching. The proposal focuses on the alleviation of language barriers and the guidance of multi-level comprehension ability, emphasizing the teaching mechanism of teacher-led and AI technology synergy. In the reading preparation stage, students can use AI translation or summarization tools to initially understand the general meaning of a text and lower the initial reading threshold. Teachers, on the other hand, should guide students to compare the original text with different translations; discuss semantic compression, stylistic deformation, and other issues in AI translations; and improve text recognition and cultural sensitivity. In the text analysis stage, NLP tools can be used to count indicators such as sentence length, word frequency, TTR, and symbolic expressions to help students build up quantitative knowledge of stylistic features. Teachers can organize classroom activities focusing on stylistic comparisons, guiding students to transform the results of technical analysis into critical interpretations.

In the writing and discussion sessions, students can use conversational AI (e.g., ChatGPT) to generate questions or obtain references to multiple perspectives of arguments, training independent inquiry and critical judgment. Meanwhile, the AI writing feedback system can be used in conjunction with teacher rubrics to improve feedback efficiency and the quality of writing revisions. It has been shown that this model is as effective as traditional feedback in some teaching practices. (Bodaubekov et al., 2025), but its broad applicability remains to be further validated. In order to ensure that the logic of instructional design and text analysis forms a closed loop, this paper tries to present the correspondence of data–task–tools in a structured way (

Table 4. Instructional task design based on textual analysis.

Textual Analysis Results	Teaching Phase	AI Tool Integration Method	Teaching Goals
“The Chameleon”: average sentence length is short, readability is high	Reading Phase	Use AI translation to quickly get general meaning	Lower language barrier, ease reading anxiety
“A Madman’s Diary”: complex sentence structures, high lexical density	Analysis Phase	Use NLP to collect data on sentence length/frequency, etc.	Recognize stylistic differences, build language sensitivity
AI translations generally improve readability	Text Comparison	Compare AI and human translations	Understand text strategy and language retention
Imagery and metaphor distribution show clear differences	In-depth Reading	Use ChatGPT to help students generate symbol-/theme-related questions	Guide students to develop critical discussion

), illustrate the path of AI’s intervention and its auxiliary function in specific instructional activities, and further elucidate the theoretical basis and implementable mechanism of AI’s intervention in teaching.

5. Discussion

This study analyzed the linguistic complexity of Chekhov’s “The Chameleon” and Lu Xun’s “Diary of a Madman” and combined the results of the comparison between human translation and AI translation to explore the potential role of AI in lowering the linguistic threshold and supporting the understanding of the text in literature teaching. The natural language processing results show that the AI translations exhibited obvious linguistic simplification features in terms of readability, syntactic simplicity, and lexical diversity, which is highly consistent with the Translation Simplicity Hypothesis (Niu & Jiang, 2024) and provides objective indicators to support the readability grading of teaching materials.

This phenomenon can also be explained within the framework of cognitive load theory, where text simplification based on cognitive load theory is effective in reducing the complexity inherent in students’ engagement with literature. By reducing the complexity of texts, educators can enhance comprehension and promote a smoother learning experience. This approach is consistent with cognitive load theory, which emphasizes the importance of managing intrinsic, extrinsic, and related cognitive load to optimize learning outcomes (Syagif, 2024). AI-generated concise texts can be regarded as supplementary input for helping students with initial comprehension and background pre-reading, especially in cross-cultural teaching to alleviate language anxiety and beginning reading difficulties. However, this study did not involve student experiments or instructional feedback, so its pedagogical effectiveness still needs to be further verified in future empirical studies.

In addition to its simplification functions at the linguistic level, the mediating role of artificial intelligence in cultural understanding is also worthy of attention. AI possesses the ability to quickly retrieve and integrate historical context and cultural knowledge, enabling students to better understand the historical backdrop behind texts such as those dealing with tsarist rule or feudal systems. This information-based cultural supplementation facilitates cross-cultural teaching. However, such understanding is often fragmented and lacks a deep construction of the literary context, making it difficult to achieve true “interpretive engagement.”

From Gadamer’s hermeneutic perspective, understanding is not a one-way reception of information but a dynamic process of horizon fusion. Readers must engage in an ongoing dialogue with the text through their own historical consciousness, generating meaning through the interplay of past and present. In this regard, AI cannot replace human subjective experience and cultural immersion, and its role as a “collaborator” remains inherently limited in deep literary interpretation. Therefore, while AI can assist in teaching processes, it cannot replace the cultural empathy and historical awareness required for interpretive engagement.

Gadamer (1960/2004) emphasizes that language is at the heart of this process, as the medium of understanding and interpretation. This dynamic process is characterized by the productive production of new meanings and interpretations rather than the simple reproduction of existing ones (Corazza, 2020). AI, although it can act as a superficial conveyor of information, lacks historical and emotional resonance in itself and cannot replace the common sense of emotional understanding and moral judgement that the “interpretive subject” plays in the humanities. Comparing the metaphor processing ability of AI and human translators, we find that the “cannibalism” imagery and “moonlight” symbols in “A Madman’s Diary” are often only interpreted in a generalized way in the output of AI, which lacks the “cynicism” of Lu Xun’s writing.

This issue has been echoed in posthumanist critiques, with Latour (1992) suggesting that “non-human agents” can participate in the construction of knowledge, and the legitimacy of AI as cognitive collaborators is on the rise. However, we emphasize that AI should be positioned as “learning partners” rather than “explanatory leaders”. Gerlich (2025) suggests that students who rely exclusively on structured answers provided by AI may lose the ability to interpret texts independently and lose sensitivity and patience with complex cultural symbols. In our teaching experiments, we found that some students regarded AI answers as “ultimate conclusions”, which not only weakened their critical thinking but also misled them to ignore the literary value of textual polysemy and ambiguity.

In literature teaching, AI should be regarded as a “participant in assisting interpretation” rather than a “dominator of meaning output”. Teachers can guide students to use AI translation and NLP tools to complete the preliminary text analysis, understand the structural features of the text, and identify the stylistic compression and cultural shifts caused by AI in the process of language processing. Through the comparison of multiple versions of “original text–human translation–AI translation”, students can gradually identify the disappearance of polysemous information, the weakening of symbols, and the transfer of meaning triggered by syntactic reconstruction. This process helps to stimulate their critical reading ability and cross-cultural awareness, and the AI tool is also valuable as an aid in teaching task generation and writing feedback. Teachers can incorporate AI tools into the preparation and extension of the teaching process by, for example, using ChatGPT to assist in the design of reading questions, generating multi-perspective references, and providing structured linguistic feedback in order to alleviate the complexity of teaching at the primary level and expand students’ path of expression. By structurally presenting the docking mechanism between text analysis results and task design (Table 4), this study demonstrates how AI technology can intervene in the teaching process in a controllable and transparent way, thus enhancing the operability and flexibility of instructional design. However, the actual pedagogical effect of this model has not been empirically verified and still needs to be further tested in specific teaching scenarios. The use of AI should always be teacher-led, serving higher-order pedagogical goals such as polysemous comprehension, cultural interpretation, and in-depth engagement, rather than replacing the creative and critical functions of human beings in literature education.

6. Implications and Future Directions

This study centered on the language barrier problem of Chekhov and Lu Xun’s works in cross-cultural literature teaching and constructed a teaching support model based on natural language processing and AI translation analysis. By comparing the indicators of readability, lexical density, and grammatical complexity between AI translations and human translations, the potential of AI in simplifying the linguistic structure of texts and assisting in the design of teaching tasks was initially explored. Although this study did not systematically carry out teaching intervention experiments, the relevant data and teaching observations provided theoretical support and methodological preparation for the construction of the model and future teaching applications. The limitations of the current model were mainly reflected in two aspects. First, the textual analysis focused on the quantification of superficial linguistic features, such as sentence length, TTR, and the proportion of complex words, but it is not yet possible to cover the deeper structures of literary texts, such as rhetorical strategies, symbolic systems, and cultural contexts. Secondly, the teaching cases focused on three universities in China, Russia, and Kazakhstan, without a wider comparative application in cross-institutional and multi-language environments. Students’ subjective feedback and behavioral data have not yet been included in the analysis system, which, to some extent, limits the systematic evaluation of the effectiveness of AI intervention in teaching.

Future research can further introduce empirical research designs based on the current model. For example, the AI translation assistance module can be introduced into the existing teaching process, and the impact of AI tools on students’ text comprehension, writing ability, and critical thinking in different cultural teaching environments can be assessed through pre-test–intervention–post-test combined with interviews and classroom observations. The ability of AI to recognize and generate complex rhetorical structures, contextual meanings, and cultural symbols can be further developed to promote the transformation of AI from a “language tool” to a “pedagogical collaborator”. Subsequent research can also focus on the interaction mechanism between teachers and AI and explore the process design, task configuration, and evaluation system of the “human–machine collaboration” teaching mode. Especially in multilingual and multicultural contexts, AI assistance can provide new solutions for curriculum reform in universities. Through the integration of practical testing and teaching feedback, it is expected that an intelligent literature teaching framework will be established that is both replicable and educationally adaptable in the future and that the efficiency and responsiveness of teaching will be improved while guaranteeing the depth of humanities in education.

7. Conclusions

This study focused on “how to optimize the instructional design and implementation of Chekhov and Lu Xun’s works in a cross-cultural literature course using Artificial Intelligence (AI) technology,” aiming to explore how AI can reduce the difficulty of linguistic and cultural comprehension while at the same time supporting students’ perceptions and critical constructions of complex literary texts. Taking the comprehension barriers faced by Chinese linguistics students in reading “The Chameleon” and “A Madman’s Diary” as a starting point, this study explored the potential of AI technology to assist in literature teaching from three dimensions: text analysis, the comparison of instructional cases, and the modeling of instructional design. Quantitative analysis based on natural language processing tools shows that AI translations show a trend of linguistic simplification in terms of readability, sentence length, and vocabulary density and are more structurally intuitive than human translations. This feature helps to alleviate the cognitive load of students at the early stage of reading and also provides preliminary evidence for the application of the “translation simplicity hypothesis” in teaching contexts. In the case studies, course practices from Peking University, Moscow State University, and Al-Farabi Kazakh National University show that AI tools have been used in prereading, generative questioning, and writing feedback, and the acceptance and experimental exploration of AI-assisted teaching, especially in the multilingual environment of Kazakhstan, are positive. These observations suggest that AI tools are gradually becoming involved in the teaching–learning process and have the potential to reshape the way knowledge is negotiated between teachers and students. This study suggests that although AI tools cannot replace human interpretation, they can serve as cognitive scaffolds. By reducing text complexity and providing semantic support, AI systems may help learners manage internal and external cognitive load, thereby enhancing comprehension and engagement in comparative literature learning environments. This study proposes a model of AI-enabled literature teaching that integrates four dimensions: text processing, semantic analysis, interactive dialogue, and writing support, emphasizing the role of AI as a mediator of cultural interpretation. Based on Gadamer’s theory of “fusion of fields of view” and the concept of cognitive synergy of posthumanism, the model expands the theoretical space of AI in literature teaching and provides a possible operational path for the teaching practice of “intelligent hermeneutics”.