The Application of Multimodal Generative AI in College English Creative Writing Instruction

The Application of Multimodal Generative AI in College English Creative Writing Instruction | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article The Application of Multimodal Generative AI in College English Creative Writing Instruction Liang Cheng This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7743384/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract With the rapid development of generative artificial intelligence (Generative AI), its application in education has become increasingly prominent, particularly in the fields of language instruction and creative writing. Grounded in Vygotsky’s Sociocultural Theory, Multimodal Representation Theory, and Distributed Cognition Theory, this study constructs a university English creative writing teaching model supported by multimodal generative AI. Through teaching experiments, analysis of student writing samples, and questionnaire surveys, the research explores the impact of AI-assisted creative writing on students’ language proficiency, creativity, and learning motivation. The findings demonstrate that the integration of AI significantly enhances students’ creative expression, multimodal language transformation, and overall writing quality. Furthermore, the use of AI tools improves students’ intercultural awareness, autonomous learning ability, and confidence in writing. This paper offers both theoretical insights and practical strategies for integrating AI into higher education, with considerable value for educational innovation and pedagogical reform. Generative AI College English Creative Writing Multimodal Teaching Language Competence Creativity Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Project This paper is a phased achievement of the 2024 Jilin Province Higher Education Research Project titled “Attempts and Explorations of ChatGPT in College English Translation” (JGJX24D1074); and a phased achievement of the 2025 China Tao Xingzhi Research Association’s 14th Five-Year Plan Special Project on “Reading and Teacher Development” titled “The Application of Multimodal Generative AI in College English Creative Writing Instruction” (202513164JN). 1. Introduction 1.1 Research Background and Problem Statement Against the backdrop of the ongoing global digital transformation in education, the teaching objectives of college English courses have gradually shifted from the traditional focus on language knowledge transmission toward a comprehensive cultivation of “language competence + cognitive qualities + intercultural communication ability.” Creative writing, as an essential component of language learning, not only enhances students’ comprehensive language application skills but also stimulates their desire for expression, strengthens their creative thinking, and promotes their personal development and cultural understanding (Hyland, 2016). However, in current teaching practice, several challenges still commonly hinder the effective implementation of creative writing. First, there is a prominent lack of creativity among students. Due to traditional teaching models’ overemphasis on grammar, vocabulary, and structural norms, students often lack space for autonomous expression and imaginative creation in writing. They tend to be confined to formulaic writing or imitative compositions, struggling to develop distinctive individuality and independent thinking. Moreover, a lack of awareness of global cultural diversity makes it difficult for students to accurately contextualize open-ended or cross-cultural writing topics, leading to superficial content and shallow expression. Second, teachers face delayed feedback and limited resources in teaching creative writing. Traditional pen-and-paper writing or post-class correction modes result in long feedback cycles, providing students with few opportunities for immediate reflection and revision, thereby weakening the dynamic and interactive nature of the writing process. Furthermore, constrained by teaching time, energy, and grading workload, teachers find it difficult to offer each student targeted, diversified, and well-structured feedback, which affects teaching quality. Third, the integration of digital literacy and artificial intelligence tools in college English courses remains low. Although the Ministry of Education has repeatedly emphasized the importance of “digital-empowered education” in recent years, generative AI technology has yet to be widely incorporated into writing courses. Some teachers’ understanding of AI technology remains superficial, lacking systematic integration plans, while students’ use of AI writing tools suffers from over-reliance, weak critical discernment, and insufficient ethical awareness (Zou et al., 2023). In this context, the rapid development of generative artificial intelligence (Generative AI) technology presents new opportunities for creative writing instruction. AI systems based on large language models (such as ChatGPT, Claude, Wenxin Yiyan, etc.) possess multimodal functions including language generation, image creation, and speech output. They can assist students’ writing in multiple dimensions such as content inspiration, structural organization, language refinement, and visual extension. These functions not only broaden students’ creative thinking space but also construct an interactive chain of “immediate feedback — continuous optimization — multimodal presentation” during teaching, thus alleviating issues of creativity deficiency and feedback delay common in traditional teaching. Meanwhile, Chinese strategies promoting “AI + Education,” the implementation of the “Digital China Construction Overall Layout Plan,” and the “14th Five-Year Plan for Education Informatization” provide technical foundations and policy support for higher education curriculum reform. Higher education especially needs to leverage intelligent technology to transform teaching paradigms, thereby better achieving educational goals such as “personalized instruction,” “integration of teaching and assessment,” and “intercultural integration.” Under this background, how to build a new teaching model for college English creative writing that involves generative AI — one that enhances students’ language competence and writing literacy while fostering their creative thinking development — has become an urgent and important research topic. 1.2 Research Objectives This study focuses on “The Impact of Generative AI-Assisted Creative Writing on College Students’ Language Ability and Creativity,” aiming to construct a college English creative writing teaching system integrating multimodal generative AI, and systematically explore feasible paths for deeply embedding AI technology into language teaching. The research is conducted from three dimensions: instructional design, instructional implementation, and instructional evaluation, with four specific objectives as follows: First, theoretical construction dimension: Based on Vygotsky’s sociocultural theory, multimodal representation theory, and distributed cognition theory, construct a “human–AI collaboration” cognitive writing model. This model systematically explains the role positioning and functional mechanisms of generative AI in the writing process, including its functions as a cultural mediator, cognitive scaffold, and creative partner, thereby providing theoretical support for technological intervention in language education. Second, teaching practice dimension: Design and conduct a cyclical AI-assisted college English creative writing teaching experiment. Combining task-driven, project-based learning, and multimodal expression strategies, guide students to engage in writing practice under the support of generative AI, exploring the teaching effectiveness in stimulating students’ language expressiveness and creative thinking abilities. Third, empirical evaluation dimension: Through a combination of quantitative analysis and qualitative research, systematically evaluate changes in students’ language accuracy, discourse organization ability, creative thinking, and writing motivation in an AI-assisted writing environment. Attention is also given to the adaptability and usage feedback of AI tools by students with different language proficiency levels, revealing teaching adaptability and individual difference effects. Fourth, innovative exploration dimension: Based on teaching experiments and empirical studies, summarize and refine replicable and scalable AI-assisted creative writing teaching models and evaluation mechanisms. Provide practical paradigms for the digital transformation of college English curricula and theoretical and operational support for improving teachers’ AI literacy and innovating teaching models. In summary, this study attempts to connect the research chain of “theoretical construction — instructional design — data-based empirical study — mechanism modeling,” promoting the deep integration of generative AI and language teaching. It aims to build a new writing teaching ecosystem centered on “human–machine collaborative creation,” ultimately achieving overall improvement in students’ language competence, creativity, and intercultural literacy, and providing new ideas and momentum for college English teaching reform in the new era. 2. Theoretical Foundations 2.1. Vygotsky’s Sociocultural Theory Vygotsky ( 1978 ), in his sociocultural theory, pointed out that the fundamental driving force of cognitive development stems from social interaction, especially collaborative activities mediated by language, tools, and other forms within specific cultural contexts. Higher psychological functions are not innate but gradually internalized through social symbols and cultural tools in the external world. In educational settings, teachers, peers, and technological media (such as AI) can all serve as critical scaffolds within the “Zone of Proximal Development” (ZPD), helping learners develop higher-level language abilities and creativity beyond their independent capacity. In this study, multimodal generative artificial intelligence (Multimodal Generative AI) is viewed as a “higher mental tool” playing the role of a “symbolic mediator” in sociocultural contexts (Vygotsky, 1986). AI not only provides immediate feedback, language samples, and grammar correction suggestions but also generates stimulating resources such as images and sounds that inspire students’ writing creativity. Through interaction with AI, students can extend their cognitive boundaries based on their original language levels, achieving breakthroughs within their ZPD. This human–machine interaction is essentially a social interaction process that promotes conceptual formation and language output in linguistic practice. Moreover, AI can serve as “intelligent scaffolding,” offering dynamic support at various stages of writing. For example, guiding idea generation during the brainstorming phase, optimizing language structure during drafting, and providing grammar correction and style suggestions during revision—aligned with Vygotsky’s principle of “dynamic assessment” (Lantolf & Thorne, 2006). Therefore, AI’s involvement in writing instruction is not merely technological intervention but a deep embedding into the cultural mediation system. 2.2. Multimodal Representation Theory Kress and van Leeuwen ( 2001 ) proposed that multimodal representation is a fundamental mechanism for human information cognition and communication. Individuals rely on multiple semiotic channels (such as visual, auditory, and linguistic) to construct meaning during language comprehension and creation. Especially in digital environments, the integration of images, sounds, and text significantly affects learners’ breadth of thinking and depth of expression. Learners no longer express themselves only through linear text but dynamically generate concepts via cross-modal combinations. The AI writing platforms used in this study (e.g., text-to-image AI, speech synthesis AI, ChatGPT) inherently possess multimodal transformation capabilities, able to convert students’ creative imagery into images and then feed back into textual expression. For instance, in the teaching case “The Poetic Voice: From Image to Verse”, students upload hand-drawn sketches, after which AI generates accompanying images and guides students to transfer expression between “visual imagery and linguistic imagery”. This reciprocal process between “image–language” and “language–image” not only helps students establish rich expressive pathways but also deepens their understanding of text structure and cultural connotations (Jewitt, 2008). With multimodal AI-assisted creation, students realize a free-flowing code-switching between “concept–text–image–sound”, forming a cognitively deep processing cycle. This model not only enhances accuracy and aesthetic quality in language expression but also greatly expands students’ aesthetic vision and cross-media expressive ability. 2.3. Distributed Cognition Theory Distributed cognition theory, proposed by Hutchins (1995), emphasizes that cognitive processes are not confined to an individual’s brain but are “distributed” through interactions among people, tools, and environments. Cognition extends into cognitive ecosystems formed by artificial tools (e.g., computers, AI systems), social interactions, and physical environments. In AI-assisted creative writing instruction, students and AI platforms form a “human–machine collaborative cognition” model: AI provides language suggestions, image expansions, and grammar corrections, while students lead topic setting, value judgments, and discourse structuring. For example, in the “City Lights” project, AI generates poetic imagery texts based on students’ input of city impressions; students then perform aesthetic judgment and personalized revision, achieving a multi-stage interaction of “meaning–symbol–evaluation”, reflecting multi-agent distributed cognition and situational embedding (Zhang & Patel, 2006). More importantly, the distributed cognition perspective shifts instructional design from “teaching knowledge” to “designing environments”. Teachers need to create cognitive scenarios conducive to human–machine interaction, activating students’ sensitivity and integration skills regarding AI feedback. For instance, assigning tasks where students compare multiple rounds of AI suggestions promotes critical thinking habits, enabling the transition from tool dependence to cognitive autonomy. 3. Research Design This study adopts a mixed methods approach, combining quantitative data analysis and qualitative case studies to systematically investigate the practical impact of multimodal generative AI technology on college English creative writing instruction. It focuses on how AI enhances students’ language expression, creativity development, and learning motivation. To ensure scientific rigor and operability, the research design includes five main components: research questions and hypotheses, research subjects and sampling, instructional intervention, data collection tools and indicators, and data analysis methods. 3.1. Research Questions and Hypotheses The core research questions are: 1. Does AI-assisted creative writing help improve students’ language abilities? 2. What impact does multimodal generative AI have on students’ creativity development in creative writing? 3. Can AI as a collaborative partner enhance students’ writing interest and learning motivation? 4. Does AI participation change students’ language error patterns and performance? 5. What are teachers’ and students’ acceptance levels and attitudes toward AI-assisted creative writing? Based on these questions, the hypotheses are: • H1: AI-assisted creative writing significantly improves students’ language expression, especially in vocabulary richness, syntactic complexity, and discourse coherence. • H2: AI participation significantly enhances students’ creativity (e.g., imagination, expressiveness, artistry). • H3: Multimodal AI creation significantly increases students’ writing interest and classroom engagement. • H4: AI feedback effectively corrects grammar, spelling, logic, and other language errors. • H5: Teachers and students hold generally positive attitudes toward AI-assisted teaching but may have concerns regarding dependency and creativity anxiety. 3.2. Research Subjects and Sampling The study targets 60 first-year non-English major university students (divided into two natural classes). Using a quasi-experimental design, students are randomly assigned to an experimental group and a control group (30 each). The experimental group integrates multimodal generative AI systems (including ChatGPT, Midjourney, DALL·E) into writing instruction, guided by teachers to complete idea generation, content creation, language refinement, and visual expression with AI assistance. The control group receives traditional teaching, with writing tasks assigned and manually corrected by teachers, focusing on independent idea development and language organization. Two university English teachers participate as instructors and interviewees, with one AI technical staff supporting platform setup and maintenance to enhance ecological validity. 3.3. Instructional Intervention Design The intervention lasts eight weeks (16 sessions, two per week), structured around integrative task clusters of “AI + English Creative Writing”, with four phases: 3.3.1. Initiation Phase (Weeks 1–2) Training experimental students on AI tools (ChatGPT prompt design, AI image generation operation, multimodal editing platforms). The control group receives conventional writing skills and genre teaching. Pre-tests include English writing ability assessment, creativity scale (Torrance Test of Creative Thinking, TTCT), and writing motivation questionnaires. 3.3.2. Practice Phase (Weeks 3–6) Three rounds of AI-assisted creative writing tasks: • Idea generation: Experimental group uses AI for keywords and context setting; control group brainstorms in groups. • Drafting: Experimental group generates and optimizes drafts with AI; control group writes independently, receiving brief teacher feedback. • Multimodal expansion: Experimental group extends content into mixed media works (e.g., illustrated poems, AI-generated micro-story illustrations). 3.3.3. Presentation and Peer Review Phase (Week 7) All students submit final works (text + multimodal forms), conduct class presentations and peer reviews. Teachers organize grading and summarization. Experimental group submits an “AI usage reflection report” documenting experiences, challenges, and gains. 3.3.4. Evaluation Phase (Week 8) Post-tests on language ability, creativity, and motivation are conducted. Questionnaires collect feedback on AI-assisted teaching acceptance. Semi-structured interviews with some students and teachers supplement qualitative data. 3.4. Data Collection Tools and Indicators To ensure comprehensive and scientific data, various tools are used (see Table 1) Type Tool Main Indicators Quantitative English Writing Scoring Rubric (self-developed) Fluency, syntactic complexity, lexical diversity Quantitative Creativity Scale(TTCT) Fluency, flexibility, originality, elaboration Quantitative Writing Motivation Questionnaire (Likert 5-point) Interest motivation, self-efficacy, achievement goals Quantitative Error Analysis Coding System Grammar, vocabulary, spelling, logic errors Qualitative AI Collaboration Logs Frequency, method, satisfaction of AI use Qualitative Teacher/Student Interviews AI’s advantages, difficulties, improvement suggestions Writing task evaluations involve dual teacher scoring cross-validated with AI scoring tools (e.g., Grammarly, ChatGPT modules) to improve objectivity and reliability. 3.5. Data Analysis Methods SPSS 26.0 and NVivo 12 are the main tools. Methods include descriptive statistics for means and standard deviations on language ability, creativity, motivation pre- and post-tests; paired-sample t-tests to analyze changes before and after AI collaboration; ANOVA to compare experimental and control groups; Pearson correlation to explore relationships between AI use frequency and performance/satisfaction; thematic coding (open, axial, selective) on reflection reports and interviews to extract core themes; and visualization such as error distribution charts, satisfaction histograms, and AI collaboration frequency line graphs to support interpretation. 4. Instructional Implementation and Case Analysis 4.1. Overview To verify the effectiveness of multimodal generative AI in college English creative writing, the experimental group embedded ChatGPT, DALL·E, and other AI systems into traditional instruction for a blended approach; the control group followed a traditional “teacher lecture + student writing + teacher correction” model. The project adopted a “process-driven + project-oriented” method centered on thematic modules such as “cultural narratives”, “ecological writing”, and “identity imagination”, completed via a five-step creative process (inspiration, triggering—writing, planning—AI-assisted, generation—human–machine, collaborative editing—multimodal presentation). Emphasis was placed on AI’s “generative support” and “collaborative regulation” to enhance language organization, text-image matching, and emotional expression. 4.2. Instructional Design and AI Integration 4.2.1. Inspiration Triggering: AI-Guided Analogies and Metaphor Generation At the start, teachers guided thematic brainstorming combined with ChatGPT-generated metaphors, similes, and imagery. For instance, under the “Spring Emotions” theme, AI suggested poetic English phrases such as “light bleeds through the cherry blossoms” and “the wind carries unspoken verses,” inspiring students’ sensuous construction of spring. Students could input keywords like “spring + loneliness” to receive AI-generated verses or paragraphs for emotional orientation and language accumulation. 4.2.2. Writing Planning: AI-Assisted Structural and Logical Organization AI helped students build story frameworks, plot lines, and logical paragraphs before writing. When conceiving themes like “Loss and Rebirth”, ChatGPT provided typical narrative structures (e.g., three-act structure) and conflict resolution paths. Some students noted AI’s logical planning was more “immediate” and “comprehensive” than teacher guidance, able to reconstruct outlines quickly based on revisions, improving writing efficiency. 4.2.3. Text Generation and Iteration: AI–Student Collaborative Writing Model The core of AI-assisted writing is the “generate–evaluate–revise” cycle. Students input draft segments or semantic intentions; AI outputs initial text; students then refine language and style. For example, a student writing a poem on “Rainy Night Bookstore” had AI generate “her mind, a quiet shelf of unread storms,” which was creatively rephrased to “her thoughts shelved like unopened rain.” The process included an “AI role-play” mechanism, assigning ChatGPT personas such as “British Romantic poet” or “21st-century environmentalist”, enriching output styles and inspiring student imitation and revision. 4.2.4. Text-Image Interactive Creation: Multimodal AI Support Multimodal writing incorporated image generation tools like DALL·E. Students input keywords or paragraphs post-text completion to generate images for cover design, contextual expansion, or thematic symbolism. For example, a fantasy prose piece titled “Steam Forest” generated AI images depicting mechanical vines entwined on bookshelves, visually reinforcing the theme of “knowledge and alienation”. This process enhanced interaction between text and image and raised students’ awareness of cross-modal expression. 4.2.5. Human–Machine Collaborative Revision and Sharing: Writing Community Construction During revision, teachers organized “human–AI–human” collaborative peer review activities. Students uploaded texts to ChatGPT for error diagnosis and style suggestions, followed by peer review of creativity and structure. Students increasingly recognized AI as an assistant rather than a judge. Logs showed over 72% of experimental students believed “AI stimulated their self-editing awareness and motivation.” Final works, including text, images, and voice recordings, were uploaded to the teaching platform to form digital portfolios for display and review. 4.3. Case Studies and Text Analysis Case 1 “Garden Under the Night Sky” – AI-Driven Poetic Language Construction This poem focused on “loneliness and memory”, with three rounds of interaction with ChatGPT producing lines like “my shadow sleeps inside a sunflower,” rewritten by the student as “my solitude lives beneath moonlit petals.” The poetic language showed significant refinement and profound imagery, demonstrating AI’s positive role in language polishing and aesthetic guidance. The text established strong links between image and linguistic imagery; AI-generated night scenes (deep blue tones, star distribution) extended the poem’s ending: “I vanish with the stars, replanting silence in the dark.” Case 2 “Digital Dream of Legacy” – Identity Construction Supported by Multimodality This prose narrated a fictional AI robot’s quest for “human memory” in the digital era. Students used AI writing scripts for narration and incorporated descriptions like “echoes of synthetic sorrow,” inspired by DALL·E images of “lost city + machine ruins”. The style exhibited cyberpunk characteristics, reflecting creative leaps in style imitation and image-driven writing. 4.4. Teaching Feedback and Student Reactions Post-teaching surveys and interviews showed: 92% of students felt AI-assisted writing improved their language expressiveness, especially vocabulary diversity and sentence variation; 85% said the integration of text and images enhanced overall creativity and multimodal comprehension; 68% felt AI feedback was more timely and targeted than teacher feedback, enabling faster revisions; 45% expressed concerns about “overreliance on AI” and difficulty distinguishing original from collaborative parts. Reflection logs frequently used keywords like “inspiration”, “co-creation”, and “digital collaborative writing”, indicating generative AI played a facilitative rather than substitutive role. 5. Results Analysis and Discussion This chapter systematically analyzes the effects of AI-assisted creative writing instruction, combining quantitative data, qualitative observations, questionnaires, pre-post writing comparisons, and interviews, focusing on AI’s impact on language ability, creativity, multimodal expression, and writing confidence. 5.1. Language Ability Improvement Analysis The research team used a pretest-posttest design with identical writing themes. Quantitative scoring involved grammar accuracy, lexical diversity, and syntactic complexity based on the “College English Writing Evaluation Standards” (2020). 5.1.1. Writing Score Improvements The table below shows the changes in students’ writing performance (full score: 10 points) (Table 2 ) Table 2 Changes in Writing Performance Dimension Pretest Mean Posttest Mean Improvement Improvement Rate Grammar Accuracy 6.1 7.5 + 1.4 + 22.9% Syntactic Complexity 5.8 7.2 + 1.4 + 24.1% Syntactic Complexity 6.2 7.4 + 1.2 + 19.4% Overall Average 6.03 7.37 + 1.34 + 22.2% After multiple AI-interactive writing practices, students improved overall by more than 22%. They used more diverse language structures, notably increased frequency of adjective phrases, complex sentences, and non-finite verbs, indicating AI’s precise exemplification helped develop language intuition and expression skills. 5.1.2. Error Type Comparison The error distribution (Fig. 1 ) showed the greatest reductions in article usage errors and subject-verb agreement, decreasing by 51% and 47% respectively after intervention. This trend confirms AI’s grammar prompt and correction mechanisms effectively improve accuracy and help non-native learners self-correct during writing (Swain, 2005). 5.2. Creativity and Multimodal Expression Improvement 5.2.1. Creativity Score Trends Using Torrance’s creativity framework (fluency, flexibility, originality, elaboration), AI-assisted texts were scored (see Fig. 2 , Table 4 ). Table 4 Chart of Changes in AI Collaboration Proportion Item Pretest Mean Posttest Mean Improvement Rate Originality 5.3 6.9 + 30.2% Flexibility 5.5 7.1 + 29.1% Elaboration 5.2 6.8 + 30.8% Fluency 6.0 7.3 + 21.6% Composite 5.5 7.0 + 27.3% AI collaboration frequency line chart aligned with AI collaboration intensity (Table 4 ), showing AI’s role evolved from“grammar assistance”to higher-level creative support such as “inspiration generation”and “context building”. 5.2.2. Multimodal Generation Capability Demonstration Student works displayed AI-driven multimodal features. For example, a poem excerpt: “ In the library built of glass and light , Thoughts bloom like flowers at midnight. ” By comparing Figs. 3 and 4 , it can be observed that the phrase “glass and light” in the student’s original sentence is depicted only as square windows in the image, whereas the AI-generated image presents a rich, multi-layered visual decoding: at the physical level, it shows the dispersion effect of a domed prism (diffracting the seven-color spectrum); at the symbolic level, light forms a river of knowledge (echoing Plato’s cave metaphor); at the cultural level, it incorporates Gothic flying buttresses (implying European university architectural heritage). This contrast helps students realize that linguistic signs have cross-cultural interpretive space, and it also confirms Kress’s ( 2001 ) theory of design learning—that modal transformation is essentially the recreation of meaning. Regarding the images generated by students and AI, the student-produced images have an element completeness of 50% and an abstraction level of 85%, including imagery elements such as 30% square buildings, 15% flowers, and 5% representations of light; the AI-generated images show a high degree of conceptual transformation reaching 92%, featuring glass domes refracting starlight and moonlight, books floating as glowing bouquets, and ground reflections forming a forest of knowledge, with three layers of spatial depth. As shown in Table 5 (Indicators of Multimodal Expression Improvement, based on analysis of 30 works) and Fig. 5 (Indicators of Multimodal Expression Improvement), AI image generation achieves triple cognitive enhancement. Across various dimensions, the average scores for student-generated images versus AI-generated images and their increments are: imagery complexity 2.1 vs. 4.7 (+ 124%); spatial layers 1.3 vs. 3.5 (+ 169%); metaphor visibility 38% vs. 92% (+ 142%); cultural symbol density 0.8 per image vs. 3.4 per image (+ 325%). Multimodal empowerment not only effectively stimulates students’ ability to associate images with text but also enhances their integrated perception of “imagery expression—language construction—visual communication,” transforming language learning from being text-bound to a form of concrete cognitive training. This resonates with the theory of visual language structure by Kress and van Leeuwen ( 2001 ). Table 5 Indicators of Multimodal Expression Improvement (N = 30 works analyzed) Dimension Student Image Average AI Image Average Increase Imagery Complexity 2.1 4.7 + 124% Spatial Layers 1.3 3.5 + 169% Metaphor Visibility 38% 92% + 142% Cultural Symbol Density 0.8per image 3.4per image + 325% 5.3 Learning Motivation and Affective Feedback Through questionnaire surveys and group interviews focusing on three core dimensions—learning motivation, satisfaction, and concerns—this study explored students’ emotional acceptance of the AI writing support system. The detailed feedback is shown in Fig. 6 below. Most students reported that the AI writing platform’s features such as inspiration prompts, language substitution, and real-time feedback made writing more enjoyable and fulfilling. Especially, the instant visual responses when generating poems and story segments greatly enhanced their willingness to express and explore. Student A stated: “AI feels like a writing partner who’s always there to brainstorm with me, helping me come up with sentences I couldn’t think of before.” Meanwhile, 45% of students expressed concerns about “originality anxiety” and “technical dependence” regarding AI-generated content. Some worried that “overreliance might weaken my own ideation ability,” and some noted that AI outputs were sometimes not idiomatic or lacked cultural expression. This suggests that when promoting AI-assisted writing, it is necessary to strengthen the “human-AI collaboration” educational concept, clarify that AI is a tool—not the creator—and guide students to use AI reasonably and maintain autonomous control over their writing process, building a “human-centered + technology-driven” dual-spiral learning model. 5.4 Teacher Role and Teaching Design Reconstruction After AI intervention in the teaching process, teachers are no longer primarily knowledge transmitters but become designers, facilitators, and evaluators. Teachers need to master digital teaching literacy such as prompt design, AI output evaluation, and multimodal integration to flexibly incorporate multi-source generated content into classroom instruction. Teacher B remarked: “AI’s creative outputs raise higher demands for my teaching; I have to learn to understand the technology faster than students, reorganize teaching processes and evaluation mechanisms.” Therefore, restructuring teacher training mechanisms, localizing AI teaching tools, and interdisciplinary collaborative curriculum design will be key issues in future educational ecosystems. 6. Conclusions and Recommendations 6.1 Research Conclusions This study, focusing on “AI-assisted college English creative writing,” combined Vygotsky’s Sociocultural Theory, Multimodal Representation Theory, and Distributed Cognition Theory to design and implement a multimodal generative AI-supported creative writing teaching experiment. Results show that AI intervention significantly enhanced students’ performance across multiple dimensions including language expression, writing creativity, cross-modal comprehension, and learning motivation: Through AI-assisted grammar correction and style optimization, students showed significant improvement in grammar accuracy, vocabulary diversity, and sentence complexity compared to the control group, with an average writing score increase of 12.6%, indicating noticeable language ability enhancement. AI-generated multimodal inputs (e.g., images, titles, inspiration words) promoted students’ idea generation and imagination; creativity scores rose from an average of 3.4 to 4.6 (out of 5), reflecting AI’s full activation of creativity in “creative guidance.” 85% of students reported that AI writing assistants increased their interest in English writing; 78% were satisfied with the AI-assisted writing model. However, 45% expressed concerns about over dependence on AI, highlighting the need to strengthen AI ethics education and autonomous learning awareness, reflecting improved motivation and self-efficacy. The five-step creative process designed in this study (inspiration stimulation—writing planning—AI-assisted generation—human-AI collaborative editing—multimodal presentation), combined with personalized AI-generated content and teacher intervention, effectively shifted teaching focus from “correction-centered” to “generation + construction-centered,” demonstrating the effectiveness of innovative teaching practices. This study not only verifies the positive effects of AI technology in college English creative writing teaching but also provides a paradigm sample for foreign language teaching reform in the era of intelligent education. 6.2 Teaching Recommendations Based on the above findings, the following teaching suggestions are proposed: Scientifically embed AI technology, emphasizing human-AI collaborative instructional design, avoiding mere tool stacking, but strengthening AI’s roles in “assistive generation,” “inspiration activation,” and “instant feedback,” forming a dynamic integration mechanism led by teachers, supported by AI, and centered on students. Strengthen AI ethics education and critical thinking cultivation; teachers should guide students to recognize AI output limitations and biases, encourage critical use of AI tools, and prevent mechanical acceptance. Improve teaching assessment systems, balancing language and creativity dimensions; incorporate language norms, discourse structure, cultural content, and innovative expression into evaluation frameworks to establish new writing assessment models suited for AI environments. Enhance teacher AI literacy training; universities are recommended to regularly hold AI teaching workshops to enable teachers to proficiently use AI tools and flexibly transform them into teaching resources, realizing paradigm upgrades in intelligent teaching design. 6.3 Limitations and Future Outlook Although preliminary results were achieved, some limitations exist: the sample size was limited to two classes at one university; future research could expand the sample for broader applicability. The teaching period was relatively short, so long-term AI effects on writing skills remain unobserved. Additionally, individual differences among students were not deeply analyzed; future studies may conduct cross-sectional research on gender, language background, digital literacy, etc. Future research could also focus more on AI writing integration with literary aesthetic education, multilingual AI writing teaching practices, and AI-assisted cultural schema construction in cross-cultural expression to further promote deep integration of AI and language education. References Bereiter C, Scardamalia M (1987) The psychology of written composition. Lawrence Erlbaum Associates Chen XR (2024) Research on second language writing instruction reform driven by AI. Theory Pract Foreign Lang Teach, (4), 16–29 Clark A (2008) Supersizing the mind: Embodiment, action, and cognitive extension. Oxford University Press Dang Y, Zhang D (2023) Designing AI-enhanced writing environments: A user-centered perspective. Br J Edu Technol 54(1):85–102 Flower L, Hayes JR (1981) A cognitive process theory of writing. Coll Composition Communication 32(4):365–387 Gao YN, Sun FF (2023) Application of ChatGPT in enhancing associative memory strategies in English teaching. Educational Explor, (5), 62–66 Gee JP (2003) What video games have to teach us about learning and literacy. Palgrave Macmillan Hu YJ (2023) Analysis of the enabling mechanism of generative AI in English writing instruction. China Educational Technol, (12), 112–119 Kim Y, Reeves TC (2007) Reframing research on learning with technology: In search of the meaning of cognitive tools. Instr Sci 35(3):207–256 Kress G, van Leeuwen T (2001) Multimodal discourse: The modes and media of contemporary communication. Arnold Levy M, Stockwell G (2006) CALL dimensions: Options and issues in computer-assisted language learning. Routledge Liu Y, Zhang H (2022) Exploring the affordances of ChatGPT in EFL creative writing: A case study in Chinese universities. Comput Assist Lang Learn 35(7):1284–1303 Wang XR (2022) A path analysis of college English writing instruction reform under the background of artificial intelligence. Foreign Language World, pp 102–110. 6 Wang Y, Chen N-S (2020) An empirical study of how the inclusion of an AI writing assistant influences EFL students’ writing performance and perceptions. Interact Learn Environ 28(6):761–776 Warschauer M, Healey D (1998) Computers and language learning: An overview. Lang Teach 31(2):57–71 Wei J (2023) The AI-embedded model of English writing instruction in universities from a multimodal discourse perspective. Mod Distance Educ, (3), 88–94 Zhang L, Liu J (2024) Pedagogical potentials and challenges of ChatGPT in college English classrooms. Computer-Assisted Foreign Lang Teach, (2), 55–64 Vygotsky LS (1978) Mind in society: The development of higher psychological processes. Harvard University Press Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7743384","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":522351003,"identity":"6435c14f-c395-4b19-897f-b3934992f07e","order_by":0,"name":"Liang Cheng","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzUlEQVRIiWNgGAWjYJCCAw8YJIAk84EDH34QqyUBrIUt8eDMHmKtSQBr5DE+zMFGhGr52c0PDyTUWOTx3cj5cJiBh0GeX+wAfi0Gd44ZHEg4JlEseSN3w+ECCwbDmbMTCGiRSDA4kNggkbgBpGUGD0OCwW0CWuRnpH+Aasl5cJiHjQgtDDdyYLbkMBCnxeBGTgHIL4kzzzwzAAayBGG/AB22+cOHmrrEvuPJjz98+GEjzy9NyGFwIABWKUGschDgP0CK6lEwCkbBKBhJAAD7gFBCoL493wAAAABJRU5ErkJggg==","orcid":"","institution":"Jilin international studies university","correspondingAuthor":true,"prefix":"","firstName":"Liang","middleName":"","lastName":"Cheng","suffix":""}],"badges":[],"createdAt":"2025-09-29 15:02:37","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":true,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":true},"doi":"10.21203/rs.3.rs-7743384/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7743384/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":92661747,"identity":"036fedf0-c33f-4dba-afac-65b33d7cf124","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"doc","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2354176,"visible":true,"origin":"","legend":"","description":"","filename":"anonymous.doc","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/4468304a4d7a5fbb720a6cea.doc"},{"id":92662935,"identity":"a28375e8-7bfd-4ba1-b5fa-90ca74a9d076","added_by":"auto","created_at":"2025-10-02 15:24:12","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342,"visible":true,"origin":"","legend":"","description":"","filename":"rs7743384.json","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/7f32caac0bfd8a7490a031c1.json"},{"id":92661755,"identity":"45284b32-5e3b-4135-90a3-e39cd6090c9d","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":78669,"visible":true,"origin":"","legend":"","description":"","filename":"rs77433840enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/643d5e935994b9a723b7eb6a.xml"},{"id":92662309,"identity":"8eec8e61-1a3b-4fa3-ae1f-73023af30764","added_by":"auto","created_at":"2025-10-02 15:16:12","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14742,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/fb295cd29b4ca25629971c8f.png"},{"id":92661752,"identity":"364e5cf1-3327-47fd-a6cd-e008780e4aab","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13339,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/6d5637c03eb84581a0546ee5.png"},{"id":92661760,"identity":"0dd75df9-58ac-4aff-baaf-7da8c5200b29","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":208576,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/4c4574d4a3d0d512ba9235b9.png"},{"id":92661757,"identity":"c38fa4ef-828e-4807-bee7-3685df0761ca","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":404378,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/53f705cb2f3e8950e87b9843.png"},{"id":92661758,"identity":"c85ca72d-0033-4966-bd59-5f9b6c97ffb8","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27213,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/012f5449c134d004fb1ee3e8.png"},{"id":92661756,"identity":"b1d13d80-d08b-4ed2-a599-267bcecb448a","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":25046,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/8ed7747b1ae8ed84bee4a8d5.png"},{"id":92661765,"identity":"6226cc1d-0163-4aa2-87c3-51ae8483b2fa","added_by":"auto","created_at":"2025-10-02 15:08:13","extension":"xml","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":75628,"visible":true,"origin":"","legend":"","description":"","filename":"rs77433840structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/1f1fcd1bfa61b04ed76e742e.xml"},{"id":92661766,"identity":"bae84d97-ec5c-443d-acf1-f653b6bc3c2f","added_by":"auto","created_at":"2025-10-02 15:08:13","extension":"html","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":83331,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/a9af724a5a165fd0d0ca4237.html"},{"id":92661746,"identity":"1802722f-e393-4b22-b1cb-d2e2033014ee","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":64462,"visible":true,"origin":"","legend":"\u003cp\u003eError Distribution of Pre-test and Post-test\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/9d2ea4c7385e743a11be9bd3.png"},{"id":92661751,"identity":"032979de-c74a-4ba3-ae5b-19b8b8dc96e3","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":96807,"visible":true,"origin":"","legend":"\u003cp\u003eAI collaboration frequency line chart\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/232c1a77a6eaaa89b135c6ae.jpeg"},{"id":92662307,"identity":"2282a6f5-8d7e-4687-a649-c21a7191d5f4","added_by":"auto","created_at":"2025-10-02 15:16:12","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":653412,"visible":true,"origin":"","legend":"\u003cp\u003eStudent-Generated Thought Images (4 Representatives)\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/cd75c56f44171aeb8465b60f.jpeg"},{"id":92661750,"identity":"0678c444-613e-4f33-965b-0ff44c4a77b3","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":438663,"visible":true,"origin":"","legend":"\u003cp\u003eAI-Generated Thought Image (1 Example)\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/cb4e7691eaa841daea456112.jpeg"},{"id":92662310,"identity":"b2664fac-77c9-4b3a-a438-d270e9307d87","added_by":"auto","created_at":"2025-10-02 15:16:12","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":166443,"visible":true,"origin":"","legend":"\u003cp\u003eIndicators of Multimodal Expression Improvement\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/a39b5a0881fd4ec5aa0794c8.jpeg"},{"id":92661754,"identity":"c35904eb-46dd-49a3-a7d4-54e0c6096e05","added_by":"auto","created_at":"2025-10-02 15:08:12","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":158219,"visible":true,"origin":"","legend":"\u003cp\u003eStudent Questionnaire Feedback Statistics\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/4e18e93027d14689fc8365d0.jpeg"},{"id":92663131,"identity":"8490bf53-1573-4fad-9858-934462e2438c","added_by":"auto","created_at":"2025-10-02 15:32:13","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2731686,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7743384/v1/bdf18f55-b7b1-4f8a-92fa-1e14b43ce80b.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eThe Application of Multimodal Generative AI in College English Creative Writing Instruction\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Project","content":"\u003cp\u003eThis paper is a phased achievement of the 2024 Jilin Province Higher Education Research Project titled \u0026ldquo;Attempts and Explorations of ChatGPT in College English Translation\u0026rdquo; (JGJX24D1074); and a phased achievement of the 2025 China Tao Xingzhi Research Association\u0026rsquo;s 14th Five-Year Plan Special Project on \u0026ldquo;Reading and Teacher Development\u0026rdquo; titled \u0026ldquo;The Application of Multimodal Generative AI in College English Creative Writing Instruction\u0026rdquo; (202513164JN).\u003c/p\u003e\u003c/p\u003e"},{"header":"1. Introduction","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e\u003ch2\u003e1.1 Research Background and Problem Statement\u003c/h2\u003e\u003cp\u003eAgainst the backdrop of the ongoing global digital transformation in education, the teaching objectives of college English courses have gradually shifted from the traditional focus on language knowledge transmission toward a comprehensive cultivation of \u0026ldquo;language competence\u0026thinsp;+\u0026thinsp;cognitive qualities\u0026thinsp;+\u0026thinsp;intercultural communication ability.\u0026rdquo; Creative writing, as an essential component of language learning, not only enhances students\u0026rsquo; comprehensive language application skills but also stimulates their desire for expression, strengthens their creative thinking, and promotes their personal development and cultural understanding (Hyland, 2016). However, in current teaching practice, several challenges still commonly hinder the effective implementation of creative writing.\u003c/p\u003e\u003cp\u003eFirst, there is a prominent lack of creativity among students. Due to traditional teaching models\u0026rsquo; overemphasis on grammar, vocabulary, and structural norms, students often lack space for autonomous expression and imaginative creation in writing. They tend to be confined to formulaic writing or imitative compositions, struggling to develop distinctive individuality and independent thinking. Moreover, a lack of awareness of global cultural diversity makes it difficult for students to accurately contextualize open-ended or cross-cultural writing topics, leading to superficial content and shallow expression.\u003c/p\u003e\u003cp\u003eSecond, teachers face delayed feedback and limited resources in teaching creative writing. Traditional pen-and-paper writing or post-class correction modes result in long feedback cycles, providing students with few opportunities for immediate reflection and revision, thereby weakening the dynamic and interactive nature of the writing process. Furthermore, constrained by teaching time, energy, and grading workload, teachers find it difficult to offer each student targeted, diversified, and well-structured feedback, which affects teaching quality.\u003c/p\u003e\u003cp\u003eThird, the integration of digital literacy and artificial intelligence tools in college English courses remains low. Although the Ministry of Education has repeatedly emphasized the importance of \u0026ldquo;digital-empowered education\u0026rdquo; in recent years, generative AI technology has yet to be widely incorporated into writing courses. Some teachers\u0026rsquo; understanding of AI technology remains superficial, lacking systematic integration plans, while students\u0026rsquo; use of AI writing tools suffers from over-reliance, weak critical discernment, and insufficient ethical awareness (Zou et al., 2023).\u003c/p\u003e\u003cp\u003eIn this context, the rapid development of generative artificial intelligence (Generative AI) technology presents new opportunities for creative writing instruction. AI systems based on large language models (such as ChatGPT, Claude, Wenxin Yiyan, etc.) possess multimodal functions including language generation, image creation, and speech output. They can assist students\u0026rsquo; writing in multiple dimensions such as content inspiration, structural organization, language refinement, and visual extension. These functions not only broaden students\u0026rsquo; creative thinking space but also construct an interactive chain of \u0026ldquo;immediate feedback \u0026mdash; continuous optimization \u0026mdash; multimodal presentation\u0026rdquo; during teaching, thus alleviating issues of creativity deficiency and feedback delay common in traditional teaching.\u003c/p\u003e\u003cp\u003eMeanwhile, Chinese strategies promoting \u0026ldquo;AI\u0026thinsp;+\u0026thinsp;Education,\u0026rdquo; the implementation of the \u0026ldquo;Digital China Construction Overall Layout Plan,\u0026rdquo; and the \u0026ldquo;14th Five-Year Plan for Education Informatization\u0026rdquo; provide technical foundations and policy support for higher education curriculum reform. Higher education especially needs to leverage intelligent technology to transform teaching paradigms, thereby better achieving educational goals such as \u0026ldquo;personalized instruction,\u0026rdquo; \u0026ldquo;integration of teaching and assessment,\u0026rdquo; and \u0026ldquo;intercultural integration.\u0026rdquo; Under this background, how to build a new teaching model for college English creative writing that involves generative AI \u0026mdash; one that enhances students\u0026rsquo; language competence and writing literacy while fostering their creative thinking development \u0026mdash; has become an urgent and important research topic.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e1.2 Research Objectives\u003c/h2\u003e\u003cp\u003eThis study focuses on \u0026ldquo;The Impact of Generative AI-Assisted Creative Writing on College Students\u0026rsquo; Language Ability and Creativity,\u0026rdquo; aiming to construct a college English creative writing teaching system integrating multimodal generative AI, and systematically explore feasible paths for deeply embedding AI technology into language teaching. The research is conducted from three dimensions: instructional design, instructional implementation, and instructional evaluation, with four specific objectives as follows:\u003c/p\u003e\u003cp\u003eFirst, theoretical construction dimension: Based on Vygotsky\u0026rsquo;s sociocultural theory, multimodal representation theory, and distributed cognition theory, construct a \u0026ldquo;human\u0026ndash;AI collaboration\u0026rdquo; cognitive writing model. This model systematically explains the role positioning and functional mechanisms of generative AI in the writing process, including its functions as a cultural mediator, cognitive scaffold, and creative partner, thereby providing theoretical support for technological intervention in language education.\u003c/p\u003e\u003cp\u003eSecond, teaching practice dimension: Design and conduct a cyclical AI-assisted college English creative writing teaching experiment. Combining task-driven, project-based learning, and multimodal expression strategies, guide students to engage in writing practice under the support of generative AI, exploring the teaching effectiveness in stimulating students\u0026rsquo; language expressiveness and creative thinking abilities.\u003c/p\u003e\u003cp\u003eThird, empirical evaluation dimension: Through a combination of quantitative analysis and qualitative research, systematically evaluate changes in students\u0026rsquo; language accuracy, discourse organization ability, creative thinking, and writing motivation in an AI-assisted writing environment. Attention is also given to the adaptability and usage feedback of AI tools by students with different language proficiency levels, revealing teaching adaptability and individual difference effects.\u003c/p\u003e\u003cp\u003eFourth, innovative exploration dimension: Based on teaching experiments and empirical studies, summarize and refine replicable and scalable AI-assisted creative writing teaching models and evaluation mechanisms. Provide practical paradigms for the digital transformation of college English curricula and theoretical and operational support for improving teachers\u0026rsquo; AI literacy and innovating teaching models.\u003c/p\u003e\u003cp\u003eIn summary, this study attempts to connect the research chain of \u0026ldquo;theoretical construction \u0026mdash; instructional design \u0026mdash; data-based empirical study \u0026mdash; mechanism modeling,\u0026rdquo; promoting the deep integration of generative AI and language teaching. It aims to build a new writing teaching ecosystem centered on \u0026ldquo;human\u0026ndash;machine collaborative creation,\u0026rdquo; ultimately achieving overall improvement in students\u0026rsquo; language competence, creativity, and intercultural literacy, and providing new ideas and momentum for college English teaching reform in the new era.\u003c/p\u003e\u003c/div\u003e"},{"header":"2. Theoretical Foundations","content":"\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.1. Vygotsky\u0026rsquo;s Sociocultural Theory\u003c/h2\u003e\u003cp\u003eVygotsky (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e1978\u003c/span\u003e), in his sociocultural theory, pointed out that the fundamental driving force of cognitive development stems from social interaction, especially collaborative activities mediated by language, tools, and other forms within specific cultural contexts. Higher psychological functions are not innate but gradually internalized through social symbols and cultural tools in the external world. In educational settings, teachers, peers, and technological media (such as AI) can all serve as critical scaffolds within the \u0026ldquo;Zone of Proximal Development\u0026rdquo; (ZPD), helping learners develop higher-level language abilities and creativity beyond their independent capacity.\u003c/p\u003e\u003cp\u003eIn this study, multimodal generative artificial intelligence (Multimodal Generative AI) is viewed as a \u0026ldquo;higher mental tool\u0026rdquo; playing the role of a \u0026ldquo;symbolic mediator\u0026rdquo; in sociocultural contexts (Vygotsky, 1986). AI not only provides immediate feedback, language samples, and grammar correction suggestions but also generates stimulating resources such as images and sounds that inspire students\u0026rsquo; writing creativity. Through interaction with AI, students can extend their cognitive boundaries based on their original language levels, achieving breakthroughs within their ZPD. This human\u0026ndash;machine interaction is essentially a social interaction process that promotes conceptual formation and language output in linguistic practice.\u003c/p\u003e\u003cp\u003eMoreover, AI can serve as \u0026ldquo;intelligent scaffolding,\u0026rdquo; offering dynamic support at various stages of writing. For example, guiding idea generation during the brainstorming phase, optimizing language structure during drafting, and providing grammar correction and style suggestions during revision\u0026mdash;aligned with Vygotsky\u0026rsquo;s principle of \u0026ldquo;dynamic assessment\u0026rdquo; (Lantolf \u0026amp; Thorne, 2006). Therefore, AI\u0026rsquo;s involvement in writing instruction is not merely technological intervention but a deep embedding into the cultural mediation system.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.2. Multimodal Representation Theory\u003c/h2\u003e\u003cp\u003eKress and van Leeuwen (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2001\u003c/span\u003e) proposed that multimodal representation is a fundamental mechanism for human information cognition and communication. Individuals rely on multiple semiotic channels (such as visual, auditory, and linguistic) to construct meaning during language comprehension and creation. Especially in digital environments, the integration of images, sounds, and text significantly affects learners\u0026rsquo; breadth of thinking and depth of expression. Learners no longer express themselves only through linear text but dynamically generate concepts via cross-modal combinations.\u003c/p\u003e\u003cp\u003eThe AI writing platforms used in this study (e.g., text-to-image AI, speech synthesis AI, ChatGPT) inherently possess multimodal transformation capabilities, able to convert students\u0026rsquo; creative imagery into images and then feed back into textual expression. For instance, in the teaching case \u0026ldquo;The Poetic Voice: From Image to Verse\u0026rdquo;, students upload hand-drawn sketches, after which AI generates accompanying images and guides students to transfer expression between \u0026ldquo;visual imagery and linguistic imagery\u0026rdquo;. This reciprocal process between \u0026ldquo;image\u0026ndash;language\u0026rdquo; and \u0026ldquo;language\u0026ndash;image\u0026rdquo; not only helps students establish rich expressive pathways but also deepens their understanding of text structure and cultural connotations (Jewitt, 2008).\u003c/p\u003e\u003cp\u003eWith multimodal AI-assisted creation, students realize a free-flowing code-switching between \u0026ldquo;concept\u0026ndash;text\u0026ndash;image\u0026ndash;sound\u0026rdquo;, forming a cognitively deep processing cycle. This model not only enhances accuracy and aesthetic quality in language expression but also greatly expands students\u0026rsquo; aesthetic vision and cross-media expressive ability.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.3. Distributed Cognition Theory\u003c/h2\u003e\u003cp\u003eDistributed cognition theory, proposed by Hutchins (1995), emphasizes that cognitive processes are not confined to an individual\u0026rsquo;s brain but are \u0026ldquo;distributed\u0026rdquo; through interactions among people, tools, and environments. Cognition extends into cognitive ecosystems formed by artificial tools (e.g., computers, AI systems), social interactions, and physical environments.\u003c/p\u003e\u003cp\u003eIn AI-assisted creative writing instruction, students and AI platforms form a \u0026ldquo;human\u0026ndash;machine collaborative cognition\u0026rdquo; model: AI provides language suggestions, image expansions, and grammar corrections, while students lead topic setting, value judgments, and discourse structuring. For example, in the \u0026ldquo;City Lights\u0026rdquo; project, AI generates poetic imagery texts based on students\u0026rsquo; input of city impressions; students then perform aesthetic judgment and personalized revision, achieving a multi-stage interaction of \u0026ldquo;meaning\u0026ndash;symbol\u0026ndash;evaluation\u0026rdquo;, reflecting multi-agent distributed cognition and situational embedding (Zhang \u0026amp; Patel, 2006).\u003c/p\u003e\u003cp\u003eMore importantly, the distributed cognition perspective shifts instructional design from \u0026ldquo;teaching knowledge\u0026rdquo; to \u0026ldquo;designing environments\u0026rdquo;. Teachers need to create cognitive scenarios conducive to human\u0026ndash;machine interaction, activating students\u0026rsquo; sensitivity and integration skills regarding AI feedback. For instance, assigning tasks where students compare multiple rounds of AI suggestions promotes critical thinking habits, enabling the transition from tool dependence to cognitive autonomy.\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Research Design","content":"\u003cp\u003eThis study adopts a mixed methods approach, combining quantitative data analysis and qualitative case studies to systematically investigate the practical impact of multimodal generative AI technology on college English creative writing instruction. It focuses on how AI enhances students\u0026rsquo; language expression, creativity development, and learning motivation. To ensure scientific rigor and operability, the research design includes five main components: research questions and hypotheses, research subjects and sampling, instructional intervention, data collection tools and indicators, and data analysis methods.\u003c/p\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e3.1. Research Questions and Hypotheses\u003c/h2\u003e\u003cp\u003eThe core research questions are:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e1. Does AI-assisted creative writing help improve students\u0026rsquo; language abilities?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e2. What impact does multimodal generative AI have on students\u0026rsquo; creativity development in creative writing?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e3. Can AI as a collaborative partner enhance students\u0026rsquo; writing interest and learning motivation?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e4. Does AI participation change students\u0026rsquo; language error patterns and performance?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e5. What are teachers\u0026rsquo; and students\u0026rsquo; acceptance levels and attitudes toward AI-assisted creative writing?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003eBased on these questions, the hypotheses are:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u0026bull; H1: AI-assisted creative writing significantly improves students\u0026rsquo; language expression, especially in vocabulary richness, syntactic complexity, and discourse coherence.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u0026bull; H2: AI participation significantly enhances students\u0026rsquo; creativity (e.g., imagination, expressiveness, artistry).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u0026bull; H3: Multimodal AI creation significantly increases students\u0026rsquo; writing interest and classroom engagement.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u0026bull; H4: AI feedback effectively corrects grammar, spelling, logic, and other language errors.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u0026bull; H5: Teachers and students hold generally positive attitudes toward AI-assisted teaching but may have concerns regarding dependency and creativity anxiety.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.2. Research Subjects and Sampling\u003c/h2\u003e\u003cp\u003eThe study targets 60 first-year non-English major university students (divided into two natural classes). Using a quasi-experimental design, students are randomly assigned to an experimental group and a control group (30 each). The experimental group integrates multimodal generative AI systems (including ChatGPT, Midjourney, DALL\u0026middot;E) into writing instruction, guided by teachers to complete idea generation, content creation, language refinement, and visual expression with AI assistance. The control group receives traditional teaching, with writing tasks assigned and manually corrected by teachers, focusing on independent idea development and language organization. Two university English teachers participate as instructors and interviewees, with one AI technical staff supporting platform setup and maintenance to enhance ecological validity.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.3. Instructional Intervention Design\u003c/h2\u003e\u003cp\u003eThe intervention lasts eight weeks (16 sessions, two per week), structured around integrative task clusters of \u0026ldquo;AI\u0026thinsp;+\u0026thinsp;English Creative Writing\u0026rdquo;, with four phases:\u003c/p\u003e\u003cdiv id=\"Sec12\" class=\"Section3\"\u003e\u003ch2\u003e3.3.1. Initiation Phase (Weeks 1\u0026ndash;2)\u003c/h2\u003e\u003cp\u003eTraining experimental students on AI tools (ChatGPT prompt design, AI image generation operation, multimodal editing platforms). The control group receives conventional writing skills and genre teaching. Pre-tests include English writing ability assessment, creativity scale (Torrance Test of Creative Thinking, TTCT), and writing motivation questionnaires.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section3\"\u003e\u003ch2\u003e3.3.2. Practice Phase (Weeks 3\u0026ndash;6)\u003c/h2\u003e\u003cp\u003eThree rounds of AI-assisted creative writing tasks:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u0026bull; Idea generation: Experimental group uses AI for keywords and context setting; control group brainstorms in groups.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u0026bull; Drafting: Experimental group generates and optimizes drafts with AI; control group writes independently, receiving brief teacher feedback.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u0026bull; Multimodal expansion: Experimental group extends content into mixed media works (e.g., illustrated poems, AI-generated micro-story illustrations).\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section3\"\u003e\u003ch2\u003e3.3.3. Presentation and Peer Review Phase (Week 7)\u003c/h2\u003e\u003cp\u003eAll students submit final works (text\u0026thinsp;+\u0026thinsp;multimodal forms), conduct class presentations and peer reviews. Teachers organize grading and summarization. Experimental group submits an \u0026ldquo;AI usage reflection report\u0026rdquo; documenting experiences, challenges, and gains.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section3\"\u003e\u003ch2\u003e3.3.4. Evaluation Phase (Week 8)\u003c/h2\u003e\u003cp\u003ePost-tests on language ability, creativity, and motivation are conducted. Questionnaires collect feedback on AI-assisted teaching acceptance. Semi-structured interviews with some students and teachers supplement qualitative data.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003e3.4. Data Collection Tools and Indicators\u003c/h2\u003e\u003cp\u003eTo ensure comprehensive and scientific data, various tools are used (see Table\u0026nbsp;1)\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eType\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTool\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMain Indicators\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eQuantitative\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eEnglish Writing Scoring Rubric (self-developed)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFluency, syntactic complexity, lexical diversity\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eQuantitative\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCreativity Scale(TTCT)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFluency, flexibility, originality, elaboration\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eQuantitative\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eWriting Motivation Questionnaire (Likert 5-point)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eInterest motivation, self-efficacy, achievement goals\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eQuantitative\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eError Analysis Coding System\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGrammar, vocabulary, spelling, logic errors\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eQualitative\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAI Collaboration Logs\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFrequency, method, satisfaction of AI use\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eQualitative\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTeacher/Student Interviews\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAI\u0026rsquo;s advantages, difficulties, improvement suggestions\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWriting task evaluations involve dual teacher scoring cross-validated with AI scoring tools (e.g., Grammarly, ChatGPT modules) to improve objectivity and reliability.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003e3.5. Data Analysis Methods\u003c/h2\u003e\u003cp\u003eSPSS 26.0 and NVivo 12 are the main tools. Methods include descriptive statistics for means and standard deviations on language ability, creativity, motivation pre- and post-tests; paired-sample t-tests to analyze changes before and after AI collaboration; ANOVA to compare experimental and control groups; Pearson correlation to explore relationships between AI use frequency and performance/satisfaction; thematic coding (open, axial, selective) on reflection reports and interviews to extract core themes; and visualization such as error distribution charts, satisfaction histograms, and AI collaboration frequency line graphs to support interpretation.\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Instructional Implementation and Case Analysis","content":"\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\u003ch2\u003e4.1. Overview\u003c/h2\u003e\u003cp\u003eTo verify the effectiveness of multimodal generative AI in college English creative writing, the experimental group embedded ChatGPT, DALL\u0026middot;E, and other AI systems into traditional instruction for a blended approach; the control group followed a traditional \u0026ldquo;teacher lecture\u0026thinsp;+\u0026thinsp;student writing\u0026thinsp;+\u0026thinsp;teacher correction\u0026rdquo; model. The project adopted a \u0026ldquo;process-driven\u0026thinsp;+\u0026thinsp;project-oriented\u0026rdquo; method centered on thematic modules such as \u0026ldquo;cultural narratives\u0026rdquo;, \u0026ldquo;ecological writing\u0026rdquo;, and \u0026ldquo;identity imagination\u0026rdquo;, completed via a five-step creative process (inspiration, triggering\u0026mdash;writing, planning\u0026mdash;AI-assisted, generation\u0026mdash;human\u0026ndash;machine, collaborative editing\u0026mdash;multimodal presentation). Emphasis was placed on AI\u0026rsquo;s \u0026ldquo;generative support\u0026rdquo; and \u0026ldquo;collaborative regulation\u0026rdquo; to enhance language organization, text-image matching, and emotional expression.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\u003ch2\u003e4.2. Instructional Design and AI Integration\u003c/h2\u003e\u003cdiv id=\"Sec21\" class=\"Section3\"\u003e\u003ch2\u003e4.2.1. Inspiration Triggering: AI-Guided Analogies and Metaphor Generation\u003c/h2\u003e\u003cp\u003eAt the start, teachers guided thematic brainstorming combined with ChatGPT-generated metaphors, similes, and imagery. For instance, under the \u0026ldquo;Spring Emotions\u0026rdquo; theme, AI suggested poetic English phrases such as \u0026ldquo;light bleeds through the cherry blossoms\u0026rdquo; and \u0026ldquo;the wind carries unspoken verses,\u0026rdquo; inspiring students\u0026rsquo; sensuous construction of spring. Students could input keywords like \u0026ldquo;spring\u0026thinsp;+\u0026thinsp;loneliness\u0026rdquo; to receive AI-generated verses or paragraphs for emotional orientation and language accumulation.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec22\" class=\"Section3\"\u003e\u003ch2\u003e4.2.2. Writing Planning: AI-Assisted Structural and Logical Organization\u003c/h2\u003e\u003cp\u003eAI helped students build story frameworks, plot lines, and logical paragraphs before writing. When conceiving themes like \u0026ldquo;Loss and Rebirth\u0026rdquo;, ChatGPT provided typical narrative structures (e.g., three-act structure) and conflict resolution paths. Some students noted AI\u0026rsquo;s logical planning was more \u0026ldquo;immediate\u0026rdquo; and \u0026ldquo;comprehensive\u0026rdquo; than teacher guidance, able to reconstruct outlines quickly based on revisions, improving writing efficiency.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\u003ch2\u003e4.2.3. Text Generation and Iteration: AI\u0026ndash;Student Collaborative Writing Model\u003c/h2\u003e\u003cp\u003eThe core of AI-assisted writing is the \u0026ldquo;generate\u0026ndash;evaluate\u0026ndash;revise\u0026rdquo; cycle. Students input draft segments or semantic intentions; AI outputs initial text; students then refine language and style. For example, a student writing a poem on \u0026ldquo;Rainy Night Bookstore\u0026rdquo; had AI generate \u0026ldquo;her mind, a quiet shelf of unread storms,\u0026rdquo; which was creatively rephrased to \u0026ldquo;her thoughts shelved like unopened rain.\u0026rdquo; The process included an \u0026ldquo;AI role-play\u0026rdquo; mechanism, assigning ChatGPT personas such as \u0026ldquo;British Romantic poet\u0026rdquo; or \u0026ldquo;21st-century environmentalist\u0026rdquo;, enriching output styles and inspiring student imitation and revision.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec24\" class=\"Section3\"\u003e\u003ch2\u003e4.2.4. Text-Image Interactive Creation: Multimodal AI Support\u003c/h2\u003e\u003cp\u003eMultimodal writing incorporated image generation tools like DALL\u0026middot;E. Students input keywords or paragraphs post-text completion to generate images for cover design, contextual expansion, or thematic symbolism. For example, a fantasy prose piece titled \u0026ldquo;Steam Forest\u0026rdquo; generated AI images depicting mechanical vines entwined on bookshelves, visually reinforcing the theme of \u0026ldquo;knowledge and alienation\u0026rdquo;. This process enhanced interaction between text and image and raised students\u0026rsquo; awareness of cross-modal expression.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\u003ch2\u003e4.2.5. Human\u0026ndash;Machine Collaborative Revision and Sharing: Writing Community Construction\u003c/h2\u003e\u003cp\u003eDuring revision, teachers organized \u0026ldquo;human\u0026ndash;AI\u0026ndash;human\u0026rdquo; collaborative peer review activities. Students uploaded texts to ChatGPT for error diagnosis and style suggestions, followed by peer review of creativity and structure. Students increasingly recognized AI as an assistant rather than a judge. Logs showed over 72% of experimental students believed \u0026ldquo;AI stimulated their self-editing awareness and motivation.\u0026rdquo; Final works, including text, images, and voice recordings, were uploaded to the teaching platform to form digital portfolios for display and review.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec26\" class=\"Section2\"\u003e\u003ch2\u003e4.3. Case Studies and Text Analysis\u003c/h2\u003e\u003cp\u003e\u003cstrong\u003eCase 1\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Garden Under the Night Sky\u0026rdquo; \u0026ndash; AI-Driven Poetic Language Construction\u003c/p\u003e\u003c/p\u003e\u003cp\u003eThis poem focused on \u0026ldquo;loneliness and memory\u0026rdquo;, with three rounds of interaction with ChatGPT producing lines like \u0026ldquo;my shadow sleeps inside a sunflower,\u0026rdquo; rewritten by the student as \u0026ldquo;my solitude lives beneath moonlit petals.\u0026rdquo; The poetic language showed significant refinement and profound imagery, demonstrating AI\u0026rsquo;s positive role in language polishing and aesthetic guidance. The text established strong links between image and linguistic imagery; AI-generated night scenes (deep blue tones, star distribution) extended the poem\u0026rsquo;s ending: \u0026ldquo;I vanish with the stars, replanting silence in the dark.\u0026rdquo;\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eCase 2\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Digital Dream of Legacy\u0026rdquo; \u0026ndash; Identity Construction Supported by Multimodality\u003c/p\u003e\u003c/p\u003e\u003cp\u003eThis prose narrated a fictional AI robot\u0026rsquo;s quest for \u0026ldquo;human memory\u0026rdquo; in the digital era. Students used AI writing scripts for narration and incorporated descriptions like \u0026ldquo;echoes of synthetic sorrow,\u0026rdquo; inspired by DALL\u0026middot;E images of \u0026ldquo;lost city\u0026thinsp;+\u0026thinsp;machine ruins\u0026rdquo;. The style exhibited cyberpunk characteristics, reflecting creative leaps in style imitation and image-driven writing.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec27\" class=\"Section2\"\u003e\u003ch2\u003e4.4. Teaching Feedback and Student Reactions\u003c/h2\u003e\u003cp\u003ePost-teaching surveys and interviews showed: 92% of students felt AI-assisted writing improved their language expressiveness, especially vocabulary diversity and sentence variation; 85% said the integration of text and images enhanced overall creativity and multimodal comprehension; 68% felt AI feedback was more timely and targeted than teacher feedback, enabling faster revisions; 45% expressed concerns about \u0026ldquo;overreliance on AI\u0026rdquo; and difficulty distinguishing original from collaborative parts. Reflection logs frequently used keywords like \u0026ldquo;inspiration\u0026rdquo;, \u0026ldquo;co-creation\u0026rdquo;, and \u0026ldquo;digital collaborative writing\u0026rdquo;, indicating generative AI played a facilitative rather than substitutive role.\u003c/p\u003e\u003c/div\u003e"},{"header":"5. Results Analysis and Discussion","content":"\u003cp\u003eThis chapter systematically analyzes the effects of AI-assisted creative writing instruction, combining quantitative data, qualitative observations, questionnaires, pre-post writing comparisons, and interviews, focusing on AI\u0026rsquo;s impact on language ability, creativity, multimodal expression, and writing confidence.\u003c/p\u003e\u003cdiv id=\"Sec29\" class=\"Section2\"\u003e\u003ch2\u003e5.1. Language Ability Improvement Analysis\u003c/h2\u003e\u003cp\u003eThe research team used a pretest-posttest design with identical writing themes. Quantitative scoring involved grammar accuracy, lexical diversity, and syntactic complexity based on the \u0026ldquo;College English Writing Evaluation Standards\u0026rdquo; (2020).\u003c/p\u003e\u003cdiv id=\"Sec30\" class=\"Section3\"\u003e\u003ch2\u003e5.1.1. Writing Score Improvements\u003c/h2\u003e\u003cp\u003eThe table below shows the changes in students\u0026rsquo; writing performance (full score: 10 points) (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e)\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eChanges in Writing Performance\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDimension\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePretest Mean\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePosttest Mean\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eImprovement\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eImprovement Rate\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGrammar Accuracy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e7.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;1.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e+\u0026thinsp;22.9%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSyntactic Complexity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e5.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e7.2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;1.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e+\u0026thinsp;24.1%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSyntactic Complexity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e7.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;1.2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e+\u0026thinsp;19.4%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOverall Average\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003e6.03\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e7.37\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e+\u0026thinsp;1.34\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e+\u0026thinsp;22.2%\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eAfter multiple AI-interactive writing practices, students improved overall by more than 22%. They used more diverse language structures, notably increased frequency of adjective phrases, complex sentences, and non-finite verbs, indicating AI\u0026rsquo;s precise exemplification helped develop language intuition and expression skills.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec31\" class=\"Section3\"\u003e\u003ch2\u003e5.1.2. Error Type Comparison\u003c/h2\u003e\u003cp\u003eThe error distribution (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) showed the greatest reductions in article usage errors and subject-verb agreement, decreasing by 51% and 47% respectively after intervention. This trend confirms AI\u0026rsquo;s grammar prompt and correction mechanisms effectively improve accuracy and help non-native learners self-correct during writing (Swain, 2005).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec32\" class=\"Section2\"\u003e\u003ch2\u003e5.2. Creativity and Multimodal Expression Improvement\u003c/h2\u003e\u003cdiv id=\"Sec33\" class=\"Section3\"\u003e\u003ch2\u003e5.2.1. Creativity Score Trends\u003c/h2\u003e\u003cp\u003eUsing Torrance\u0026rsquo;s creativity framework (fluency, flexibility, originality, elaboration), AI-assisted texts were scored (see Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eChart of Changes in AI Collaboration Proportion\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eItem\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePretest Mean\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePosttest Mean\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eImprovement Rate\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOriginality\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e5.3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e6.9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;30.2%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFlexibility\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e5.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e7.1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;29.1%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eElaboration\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e5.2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e6.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;30.8%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFluency\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e7.3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;21.6%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eComposite\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003e5.5\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e7.0\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e+\u0026thinsp;27.3%\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eAI collaboration frequency line chart aligned with AI collaboration intensity (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e4\u003c/span\u003e), showing AI\u0026rsquo;s role evolved from\u0026ldquo;grammar assistance\u0026rdquo;to higher-level creative support such as \u0026ldquo;inspiration generation\u0026rdquo;and \u0026ldquo;context building\u0026rdquo;.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec34\" class=\"Section3\"\u003e\u003ch2\u003e5.2.2. Multimodal Generation Capability Demonstration\u003c/h2\u003e\u003cp\u003eStudent works displayed AI-driven multimodal features. For example, a poem excerpt:\u003c/p\u003e\u003cp\u003e\u0026ldquo;\u003cem\u003eIn the library built of glass and light\u003c/em\u003e,\u003c/p\u003e\u003cp\u003e\u003cem\u003eThoughts bloom like flowers at midnight.\u003c/em\u003e\u0026rdquo;\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eBy comparing Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, it can be observed that the phrase \u0026ldquo;glass and light\u0026rdquo; in the student\u0026rsquo;s original sentence is depicted only as square windows in the image, whereas the AI-generated image presents a rich, multi-layered visual decoding: at the physical level, it shows the dispersion effect of a domed prism (diffracting the seven-color spectrum); at the symbolic level, light forms a river of knowledge (echoing Plato\u0026rsquo;s cave metaphor); at the cultural level, it incorporates Gothic flying buttresses (implying European university architectural heritage). This contrast helps students realize that linguistic signs have cross-cultural interpretive space, and it also confirms Kress\u0026rsquo;s (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2001\u003c/span\u003e) theory of design learning\u0026mdash;that modal transformation is essentially the recreation of meaning.\u003c/p\u003e\u003cp\u003eRegarding the images generated by students and AI, the student-produced images have an element completeness of 50% and an abstraction level of 85%, including imagery elements such as 30% square buildings, 15% flowers, and 5% representations of light; the AI-generated images show a high degree of conceptual transformation reaching 92%, featuring glass domes refracting starlight and moonlight, books floating as glowing bouquets, and ground reflections forming a forest of knowledge, with three layers of spatial depth.\u003c/p\u003e\u003cp\u003eAs shown in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e5\u003c/span\u003e (Indicators of Multimodal Expression Improvement, based on analysis of 30 works) and Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e (Indicators of Multimodal Expression Improvement), AI image generation achieves triple cognitive enhancement. Across various dimensions, the average scores for student-generated images versus AI-generated images and their increments are: imagery complexity 2.1 vs. 4.7 (+\u0026thinsp;124%); spatial layers 1.3 vs. 3.5 (+\u0026thinsp;169%); metaphor visibility 38% vs. 92% (+\u0026thinsp;142%); cultural symbol density 0.8 per image vs. 3.4 per image (+\u0026thinsp;325%).\u003c/p\u003e\u003cp\u003eMultimodal empowerment not only effectively stimulates students\u0026rsquo; ability to associate images with text but also enhances their integrated perception of \u0026ldquo;imagery expression\u0026mdash;language construction\u0026mdash;visual communication,\u0026rdquo; transforming language learning from being text-bound to a form of concrete cognitive training. This resonates with the theory of visual language structure by Kress and van Leeuwen (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2001\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eIndicators of Multimodal Expression Improvement (N\u0026thinsp;=\u0026thinsp;30 works analyzed)\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDimension\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eStudent Image Average\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAI Image Average\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eIncrease\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eImagery Complexity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;124%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSpatial Layers\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;169%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMetaphor Visibility\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e38%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e92%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;142%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCultural Symbol Density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.8per image\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.4per image\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e+\u0026thinsp;325%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec35\" class=\"Section2\"\u003e\u003ch2\u003e5.3 Learning Motivation and Affective Feedback\u003c/h2\u003e\u003cp\u003eThrough questionnaire surveys and group interviews focusing on three core dimensions\u0026mdash;learning motivation, satisfaction, and concerns\u0026mdash;this study explored students\u0026rsquo; emotional acceptance of the AI writing support system. The detailed feedback is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e below.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eMost students reported that the AI writing platform\u0026rsquo;s features such as inspiration prompts, language substitution, and real-time feedback made writing more enjoyable and fulfilling. Especially, the instant visual responses when generating poems and story segments greatly enhanced their willingness to express and explore. Student A stated: \u0026ldquo;AI feels like a writing partner who\u0026rsquo;s always there to brainstorm with me, helping me come up with sentences I couldn\u0026rsquo;t think of before.\u0026rdquo;\u003c/p\u003e\u003cp\u003eMeanwhile, 45% of students expressed concerns about \u0026ldquo;originality anxiety\u0026rdquo; and \u0026ldquo;technical dependence\u0026rdquo; regarding AI-generated content. Some worried that \u0026ldquo;overreliance might weaken my own ideation ability,\u0026rdquo; and some noted that AI outputs were sometimes not idiomatic or lacked cultural expression. This suggests that when promoting AI-assisted writing, it is necessary to strengthen the \u0026ldquo;human-AI collaboration\u0026rdquo; educational concept, clarify that AI is a tool\u0026mdash;not the creator\u0026mdash;and guide students to use AI reasonably and maintain autonomous control over their writing process, building a \u0026ldquo;human-centered\u0026thinsp;+\u0026thinsp;technology-driven\u0026rdquo; dual-spiral learning model.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec36\" class=\"Section2\"\u003e\u003ch2\u003e5.4 Teacher Role and Teaching Design Reconstruction\u003c/h2\u003e\u003cp\u003eAfter AI intervention in the teaching process, teachers are no longer primarily knowledge transmitters but become designers, facilitators, and evaluators. Teachers need to master digital teaching literacy such as prompt design, AI output evaluation, and multimodal integration to flexibly incorporate multi-source generated content into classroom instruction. Teacher B remarked: \u0026ldquo;AI\u0026rsquo;s creative outputs raise higher demands for my teaching; I have to learn to understand the technology faster than students, reorganize teaching processes and evaluation mechanisms.\u0026rdquo; Therefore, restructuring teacher training mechanisms, localizing AI teaching tools, and interdisciplinary collaborative curriculum design will be key issues in future educational ecosystems.\u003c/p\u003e\u003c/div\u003e"},{"header":"6. Conclusions and Recommendations","content":"\u003cdiv id=\"Sec38\" class=\"Section2\"\u003e\u003ch2\u003e6.1 Research Conclusions\u003c/h2\u003e\u003cp\u003eThis study, focusing on \u0026ldquo;AI-assisted college English creative writing,\u0026rdquo; combined Vygotsky\u0026rsquo;s Sociocultural Theory, Multimodal Representation Theory, and Distributed Cognition Theory to design and implement a multimodal generative AI-supported creative writing teaching experiment. Results show that AI intervention significantly enhanced students\u0026rsquo; performance across multiple dimensions including language expression, writing creativity, cross-modal comprehension, and learning motivation: Through AI-assisted grammar correction and style optimization, students showed significant improvement in grammar accuracy, vocabulary diversity, and sentence complexity compared to the control group, with an average writing score increase of 12.6%, indicating noticeable language ability enhancement. AI-generated multimodal inputs (e.g., images, titles, inspiration words) promoted students\u0026rsquo; idea generation and imagination; creativity scores rose from an average of 3.4 to 4.6 (out of 5), reflecting AI\u0026rsquo;s full activation of creativity in \u0026ldquo;creative guidance.\u0026rdquo; 85% of students reported that AI writing assistants increased their interest in English writing; 78% were satisfied with the AI-assisted writing model. However, 45% expressed concerns about over dependence on AI, highlighting the need to strengthen AI ethics education and autonomous learning awareness, reflecting improved motivation and self-efficacy. The five-step creative process designed in this study (inspiration stimulation\u0026mdash;writing planning\u0026mdash;AI-assisted generation\u0026mdash;human-AI collaborative editing\u0026mdash;multimodal presentation), combined with personalized AI-generated content and teacher intervention, effectively shifted teaching focus from \u0026ldquo;correction-centered\u0026rdquo; to \u0026ldquo;generation\u0026thinsp;+\u0026thinsp;construction-centered,\u0026rdquo; demonstrating the effectiveness of innovative teaching practices. This study not only verifies the positive effects of AI technology in college English creative writing teaching but also provides a paradigm sample for foreign language teaching reform in the era of intelligent education.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec39\" class=\"Section2\"\u003e\u003ch2\u003e6.2 Teaching Recommendations\u003c/h2\u003e\u003cp\u003eBased on the above findings, the following teaching suggestions are proposed: Scientifically embed AI technology, emphasizing human-AI collaborative instructional design, avoiding mere tool stacking, but strengthening AI\u0026rsquo;s roles in \u0026ldquo;assistive generation,\u0026rdquo; \u0026ldquo;inspiration activation,\u0026rdquo; and \u0026ldquo;instant feedback,\u0026rdquo; forming a dynamic integration mechanism led by teachers, supported by AI, and centered on students. Strengthen AI ethics education and critical thinking cultivation; teachers should guide students to recognize AI output limitations and biases, encourage critical use of AI tools, and prevent mechanical acceptance. Improve teaching assessment systems, balancing language and creativity dimensions; incorporate language norms, discourse structure, cultural content, and innovative expression into evaluation frameworks to establish new writing assessment models suited for AI environments. Enhance teacher AI literacy training; universities are recommended to regularly hold AI teaching workshops to enable teachers to proficiently use AI tools and flexibly transform them into teaching resources, realizing paradigm upgrades in intelligent teaching design.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec40\" class=\"Section2\"\u003e\u003ch2\u003e6.3 Limitations and Future Outlook\u003c/h2\u003e\u003cp\u003eAlthough preliminary results were achieved, some limitations exist: the sample size was limited to two classes at one university; future research could expand the sample for broader applicability. The teaching period was relatively short, so long-term AI effects on writing skills remain unobserved. Additionally, individual differences among students were not deeply analyzed; future studies may conduct cross-sectional research on gender, language background, digital literacy, etc. Future research could also focus more on AI writing integration with literary aesthetic education, multilingual AI writing teaching practices, and AI-assisted cultural schema construction in cross-cultural expression to further promote deep integration of AI and language education.\u003c/p\u003e\u003c/div\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBereiter C, Scardamalia M (1987) The psychology of written composition. Lawrence Erlbaum Associates\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen XR (2024) Research on second language writing instruction reform driven by AI. Theory Pract Foreign Lang Teach, (4), 16\u0026ndash;29\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eClark A (2008) Supersizing the mind: Embodiment, action, and cognitive extension. Oxford University Press\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDang Y, Zhang D (2023) Designing AI-enhanced writing environments: A user-centered perspective. Br J Edu Technol 54(1):85\u0026ndash;102\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFlower L, Hayes JR (1981) A cognitive process theory of writing. Coll Composition Communication 32(4):365\u0026ndash;387\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGao YN, Sun FF (2023) Application of ChatGPT in enhancing associative memory strategies in English teaching. Educational Explor, (5), 62\u0026ndash;66\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGee JP (2003) What video games have to teach us about learning and literacy. Palgrave Macmillan\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu YJ (2023) Analysis of the enabling mechanism of generative AI in English writing instruction. China Educational Technol, (12), 112\u0026ndash;119\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim Y, Reeves TC (2007) Reframing research on learning with technology: In search of the meaning of cognitive tools. Instr Sci 35(3):207\u0026ndash;256\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKress G, van Leeuwen T (2001) Multimodal discourse: The modes and media of contemporary communication. Arnold\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLevy M, Stockwell G (2006) CALL dimensions: Options and issues in computer-assisted language learning. Routledge\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu Y, Zhang H (2022) Exploring the affordances of ChatGPT in EFL creative writing: A case study in Chinese universities. Comput Assist Lang Learn 35(7):1284\u0026ndash;1303\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang XR (2022) A path analysis of college English writing instruction reform under the background of artificial intelligence. Foreign Language World, pp 102\u0026ndash;110. 6\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang Y, Chen N-S (2020) An empirical study of how the inclusion of an AI writing assistant influences EFL students\u0026rsquo; writing performance and perceptions. Interact Learn Environ 28(6):761\u0026ndash;776\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWarschauer M, Healey D (1998) Computers and language learning: An overview. Lang Teach 31(2):57\u0026ndash;71\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWei J (2023) The AI-embedded model of English writing instruction in universities from a multimodal discourse perspective. Mod Distance Educ, (3), 88\u0026ndash;94\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang L, Liu J (2024) Pedagogical potentials and challenges of ChatGPT in college English classrooms. Computer-Assisted Foreign Lang Teach, (2), 55\u0026ndash;64\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVygotsky LS (1978) Mind in society: The development of higher psychological processes. Harvard University Press\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Jilin International Studies University","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Generative AI, College English, Creative Writing, Multimodal Teaching, Language Competence, Creativity","lastPublishedDoi":"10.21203/rs.3.rs-7743384/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7743384/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eWith the rapid development of generative artificial intelligence (Generative AI), its application in education has become increasingly prominent, particularly in the fields of language instruction and creative writing. Grounded in Vygotsky\u0026rsquo;s Sociocultural Theory, Multimodal Representation Theory, and Distributed Cognition Theory, this study constructs a university English creative writing teaching model supported by multimodal generative AI. Through teaching experiments, analysis of student writing samples, and questionnaire surveys, the research explores the impact of AI-assisted creative writing on students\u0026rsquo; language proficiency, creativity, and learning motivation. The findings demonstrate that the integration of AI significantly enhances students\u0026rsquo; creative expression, multimodal language transformation, and overall writing quality. Furthermore, the use of AI tools improves students\u0026rsquo; intercultural awareness, autonomous learning ability, and confidence in writing. This paper offers both theoretical insights and practical strategies for integrating AI into higher education, with considerable value for educational innovation and pedagogical reform.\u003c/p\u003e","manuscriptTitle":"The Application of Multimodal Generative AI in College English Creative Writing Instruction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-02 15:08:07","doi":"10.21203/rs.3.rs-7743384/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8d246c48-edaa-465e-bd4f-ec48409f1180","owner":[],"postedDate":"October 2nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-10-02T15:08:08+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-02 15:08:07","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7743384","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7743384","identity":"rs-7743384","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}