An AI-powered English teaching system that integrates translation and intercultural communication

By introducing a cross-language bilingual alignment mechanism and a quantitative classification of cross-cultural communication risks, the problem of English intelligent teaching systems being unable to distinguish between grammatical errors and cross-cultural pragmatic failures has been solved. This enables refined analysis of learners' language errors and quantitative assessment of their cross-cultural communication competence, outputting personalized teaching feedback and enhancing the intelligence of English teaching and the cultivation of cross-cultural communication competence.

CN123070141APending Publication Date: 2026-11-10ZAOZHUANG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611164976.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-03
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing intelligent English teaching systems cannot effectively distinguish between grammatical errors and cross-cultural pragmatic errors caused by literal translation from the mother tongue. They lack cross-language alignment analysis methods, cannot quantitatively assess cultural communication biases, and translation technology is disconnected from cross-cultural communication teaching, making it difficult to achieve deep integration.

Method used

A cross-linguistic bilingual alignment mechanism is introduced, and grammatical errors and cross-cultural pragmatic failures are distinguished through translation transfer source identification units. A cross-cultural communication risk quantification and grading system is constructed to generate differentiated text feedback and dynamically update learner profiles.

Benefits of technology

It enables refined analysis of learners' language errors, calculable assessment of intercultural communication risks, and output of personalized teaching feedback, thereby improving the intelligence level of English teaching and the effectiveness of cultivating intercultural communication skills.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123070141A_ABST
    Figure CN123070141A_ABST
Patent Text Reader

Abstract

This application discloses an AI-powered English intelligent teaching system integrating translation and intercultural communication, comprising a client-side interaction module, a data storage module, and a server-side computing module. The client-side interaction module collects English text input by learners and outputs teaching feedback results to them. The data storage module stores a pre-set bilingual alignment knowledge base and an intercultural communication norms knowledge base. The server-side computing module includes a translation migration source identification unit, an intercultural communication risk quantification and grading unit, and a risk-adaptive communication equivalence generation unit, all linked sequentially. The translation migration source identification unit performs fine-grained bilingual alignment on the collected learner's English text, extracts literal translation matching deviation features from the text, distinguishes between ordinary grammatical errors and intercultural pragmatic errors caused by literal translation transfer from the learner's native language, and outputs corresponding error source identification feature information. The intercultural communication risk quantification and grading unit...
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the technical field, and in particular to an AI-powered intelligent English teaching system that integrates translation and intercultural communication. Background Technology

[0002] With the deep application of artificial intelligence technology in language education, various AI-powered English teaching systems have become widely popular. Most existing intelligent teaching systems focus on detecting surface-level grammatical errors in English texts, such as spelling, tense, and syntax, and can perform basic functions such as writing correction and oral accuracy assessment.

[0003] Some teaching platforms integrate general machine translation modules to enable bilingual word and sentence lookups. However, these translation tools are used only as standalone supplementary functions and do not incorporate translation transfer mechanisms into the analysis of learners' language error causes. Learners commonly rely on literal translation from their native language to construct English expressions, which easily leads to grammatically correct but non-compliant cross-cultural pragmatic errors. Traditional intelligent assessment methods rely solely on monolingual text models for analysis, failing to distinguish between ordinary grammatical errors and pragmatic biases caused by native language translation transfer, and making it difficult to pinpoint the underlying causes of errors.

[0004] On the other hand, current intercultural communication teaching primarily relies on static text materials, lacking automated discrimination capabilities. Existing technologies struggle to quantify and assess cultural communication deviations in learners' output, failing to differentiate between varying degrees of pragmatic problems such as awkward expression, communicative indiscretions, and cultural taboo conflicts, and generally employing a uniform text correction scheme. This homogenized feedback strategy cannot dynamically adjust its effectiveness based on the severity of cultural conflicts, failing to address the needs of language correction, intercultural guidance, and differentiated instruction.

[0005] Existing intelligent English teaching systems rely solely on monolingual information for error identification, lacking cross-language alignment analysis methods. They cannot trace learners' literal translation transfer behavior and struggle to distinguish between grammatical errors and cross-cultural pragmatic errors induced by translation transfer. Furthermore, they lack standardized quantitative evaluation mechanisms, failing to automatically classify the cross-cultural communication risks arising from pragmatic deviations. Their text correction strategies are fixed and singular, unable to adaptively adjust and optimize solutions based on cultural risk levels. Translation technology and cross-cultural communication teaching are disconnected, hindering deep integration.

[0006] Therefore, we propose an AI-powered intelligent English teaching system that integrates translation and intercultural communication. Summary of the Invention

[0007] This application aims to at least partially solve one of the technical problems in the aforementioned technologies.

[0008] To achieve the above objectives, the first aspect of this application proposes an AI-powered English intelligent teaching system that integrates translation and intercultural communication, including a client-side interaction module, a data storage module, and a server-side computing module;

[0009] The client interaction module is used to collect the English text input by the learner and output teaching feedback results to the learner;

[0010] The data storage module is used to store a preset bilingual alignment knowledge base and a cross-cultural communication norms knowledge base. The server computing module includes a translation migration tracing and identification unit, a cross-cultural communication risk quantification and grading unit, and a risk-adaptive communication equivalence generation unit that are linked in sequence.

[0011] The translation transfer source identification unit is used to perform cross-language fine-grained bilingual alignment on the collected learner English text, extract the literal translation matching deviation features in the text, distinguish between ordinary grammatical errors and cross-cultural pragmatic errors caused by literal translation transfer from the mother tongue, and output the corresponding error source identification feature information.

[0012] The cross-cultural communication risk quantification and grading unit is connected to the translation migration source identification unit. It is used to receive the error source identification feature information and perform weighted calculations based on three types of quantitative features: scenario formality, pragmatic fit, and cultural conflict. The translation migration pragmatic errors are divided into three risk levels: mild fit deviation, moderate pragmatic inappropriateness, and severe cultural conflict.

[0013] The risk-adaptive communicative equivalence generation unit is communicatively connected to the cross-cultural communication risk quantification and grading unit. It is used to call the corresponding differentiated text generation constraint strategy according to the determined risk level, generate optimized text with communicative equivalence, match the corresponding cross-cultural communication analysis content, and output integrated teaching feedback information through the client interaction module.

[0014] In addition, the AI-powered English intelligent teaching system integrating translation and intercultural communication proposed in this application may also have the following additional technical features:

[0015] As a further description of the above technical solution:

[0016] The translation transfer tracing and identification unit constructs cross-language semantic mapping relationships through a phrase-level bilingual alignment model, and combines sentence transfer features and semantic unit misalignment features to comprehensively determine whether learners' English texts have native language translation transfer problems.

[0017] As a further description of the above technical solution:

[0018] The cross-cultural communication risk quantification and grading unit is based on the cross-cultural pragmatic politeness principle and the rules of difference between Chinese and English speech acts. It constructs a quantitative feature system of various dimensions to complete the graded judgment of pragmatic error risk.

[0019] As a further description of the above technical solution:

[0020] The differentiated text generation constraint strategies of the risk-adaptive communicative equivalence generation unit include: for mild adaptation bias, a semantically preserved lightweight text polishing strategy is adopted; for moderate pragmatic inappropriateness, communicative paradigm structure replacement optimization is performed; and for severe cultural conflict, mandatory cultural adaptation correction is implemented.

[0021] As a further description of the above technical solution:

[0022] The risk-adaptive communicative equivalence generation unit simultaneously matches the learner's existing language level and constrains the sentence difficulty of the optimized text during the generation of optimized text, ensuring that the output content is adapted to the learner's learning situation.

[0023] As a further description of the above technical solution:

[0024] The server computing module is also equipped with a learner profile update unit; the learner profile update unit is used to collect error source data and cross-cultural risk classification data of learners in previous text detections, dynamically update learners' literal translation behavior preference tags and cross-cultural communication ability evaluation tags, and construct learners' personalized ability profiles.

[0025] As a further description of the above technical solution:

[0026] The integrated teaching feedback information includes at least pragmatic error location markers, analysis of the causes of translation migration, optimized texts for communicative equivalence, and explanations of differences in Chinese-English communicative norms.

[0027] As a further description of the above technical solution:

[0028] The client interaction module supports receiving written English text input by learners, receiving spoken English audio from learners, and converting the audio into standard English text through speech-to-text technology, which is then sent to the translation migration tracing and identification unit for detection and analysis.

[0029] Advantages of this invention:

[0030] The AI-powered English teaching system integrating translation and intercultural communication, as described in this application, achieves precise tracing of learners' literal translation transfer behavior by introducing a cross-language bilingual alignment mechanism. This overcomes the limitations of traditional English intelligent teaching systems that rely solely on monolingual text analysis and can only identify surface-level grammatical errors. It effectively distinguishes between ordinary grammatical errors and intercultural pragmatic errors caused by literal translation transfer from the learner's native language, enabling a two-way differentiation between the appearance and underlying causes of learners' language errors. This significantly improves the refinement and professionalism of English output error analysis. Furthermore, by constructing a multi-dimensional intercultural communication risk quantification and grading system, it transforms the originally qualitative and vague cultural communication deviation problem into a calculable, gradable, and automatically identifiable standardized technical indicator. This solves the industry pain points of existing teaching systems, such as the inability to quantitatively assess the severity of intercultural communication errors and the difficulty in distinguishing between minor awkward expressions, moderate communication lapses, and severe cultural conflicts. Based on this, and relying on a risk-level-matched differentiated communicative equivalent text generation strategy, it abandons the traditional... The homogenized and fixed text correction mode of traditional teaching systems can adaptively adjust the intensity of text polishing, sentence structure optimization, and cultural correction according to different levels of cultural deviation. While ensuring semantic fidelity and conformity to English social norms and cultural scenarios, it outputs optimized content adapted to the learner's own language level and simultaneously outputs standardized cross-cultural communication analysis content. It realizes integrated intelligent teaching output of error location, cause tracing, intelligent error correction, and cultural popularization, effectively breaking down the technical barriers between machine translation technology and cross-cultural communication teaching, and achieving deep integration of the two. At the same time, it dynamically updates personalized ability profiles based on learners' historical error data, which can accurately capture learners' common literal translation errors and cross-cultural communication shortcomings, providing data support for differentiated teaching and targeted ability training. It significantly improves the intelligence, scenario adaptability, and cross-cultural education effect of AI English teaching, and effectively makes up for the technical defects of existing intelligent English teaching systems that focus on language form error correction and neglect cross-cultural communication ability cultivation.

[0031] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0032] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0033] Figure 1 This is an overall architecture block diagram of an AI-powered English intelligent teaching system that integrates translation and cross-cultural communication according to an embodiment of this application;

[0034] Figure 2 This is a block diagram showing the connection of the core units inside the server computing module of an AI English intelligent teaching system that integrates translation and cross-cultural communication according to an embodiment of this application;

[0035] Figure 3 This is a flowchart illustrating the internal differentiation generation strategy scheduling of a risk-adaptive communicative equivalence generation unit in an AI-powered English intelligent teaching system that integrates translation and cross-cultural communication, according to an embodiment of this application. Detailed Implementation

[0036] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0037] The following description, in conjunction with the accompanying drawings, describes an AI-powered English intelligent teaching system that integrates translation and cross-cultural communication, representing an embodiment of this application.

[0038] like Figure 1-3 As shown in Embodiment 1 of this application, the AI-powered English intelligent teaching system integrating translation and intercultural communication has an overall system architecture consisting of three core components: a client interaction module, a data storage module, and a server computing module. The client interaction module serves as the front-end human-computer interaction carrier, used to collect English text input by learners and visually push the integrated teaching feedback results generated by the back-end computing to the learners. The data storage module, as the underlying data support carrier, adopts an encrypted distributed storage architecture specifically for long-term storage of a pre-set bilingual alignment knowledge base, an intercultural communication norms knowledge base, and personalized learning profile data for all learners. The server computing module is the core intelligent computing carrier of the system, containing a translation migration tracing and identification unit, an intercultural communication risk quantification and grading unit, and a risk-adaptive communication equivalence generation unit that sequentially complete data linkage processing. The server computing module also features an independently deployed learner profile update unit, enabling real-time collection, analysis, and iterative updating of learner learning data and profiles. The overall hardware and software deployment mode of this system is adapted to various application scenarios such as mobile learning terminals, classroom computer teaching terminals, and online web teaching platforms. The underlying natural language processing operation logic is lightweight, eliminating the need for full-scale fine-tuning of large models. It can complete the entire intelligent operation process by relying on phrase alignment models, feature weighted scoring algorithms, and constrained text generation strategies, and has extremely low deployment costs and strong scenario adaptability.

[0039] The client-side interaction module serves as the data acquisition entry point and result output terminal for this system. It integrates two parallel data acquisition channels: a written text acquisition channel and a voice / audio acquisition channel. This covers all mainstream English cross-cultural learning scenarios, including English writing practice, oral dialogues, short sentence translation training, and daily communication simulations. The client-side interaction module can be deployed on common smart hardware such as smartphones, tablets, and desktop computers. The voice acquisition branch incorporates a noise-canceling microphone array hardware unit with integrated multi-channel audio components. This effectively filters out various interfering audio signals such as classroom ambient noise, outdoor ambient noise, breathing noise, and equipment current noise, preserving the original voice waveform of the learner's English oral practice to the greatest extent possible. The module is pre-installed with an end-to-end Conformer speech-to-text model. It performs silence segmentation, noise cancellation, and speech feature extraction on the acquired English oral audio segments, converting the continuous speech waveform into standardized English text. During the transcription process, it automatically removes invalid text information such as colloquial filler words, repetitive interjections, and meaningless pauses, ultimately outputting standardized English text with regular sentence structure and no redundant interference. The written data collection branch supports learners manually inputting various types of English text, such as long essays, short sentence exercises, and situational communication Q&A. It also features a text paste and import function, allowing direct import of classroom assignments and online practice texts, reducing the input cost for learners. After the client-side interaction module completes the standardized English text collection, it encrypts and encapsulates the text data, generating a transmission message with a unique learner identifier. This message is then fully pushed to the backend server's processing module via an HTTPS encrypted transmission channel, officially initiating the backend's intelligent analysis, grading, and generation process.

[0040] The data storage module adopts a cloud-based distributed sharded storage architecture, which is divided into a static knowledge base storage partition and a dynamic learner data storage partition according to the purpose of the data. The two types of partitions are equipped with independent read and write permission isolation mechanisms. Only the server computing module has the permission to retrieve, write, and modify partition data. External clients cannot directly access the underlying knowledge base and user learning data, thus ensuring learner learning privacy and the security of the system's core knowledge base data. The knowledge base has two core structured knowledge bases that are permanently stored in static partitions: a bilingual alignment knowledge base and a cross-cultural communication norms knowledge base. The bilingual alignment knowledge base uses Chinese-English communicative semantic phrases as the smallest basic storage unit, systematically including Chinese-English bilingual semantic mapping entries for various high-frequency speech behaviors such as requests, praise, refusal, small talk, invitations, apologies, and persuasion. Each mapping entry is simultaneously labeled with the phrase's usage scenario, sentence structure differences, and direct translation deviation risk tags. Unlike traditional word-level bilingual dictionaries, all entries in this knowledge base are built around cross-cultural communication scenarios, focusing on communicative expressions that are highly prone to native language translation transfer deviations. The cross-cultural communication norms knowledge base divides its entries according to the communication scenario dimension, comprehensively storing structured knowledge such as British and American politeness rules, differences in Chinese and English speech behaviors, cultural taboos, privacy boundary norms, and euphemisms in four major scenarios: daily leisure communication, campus teacher-student communication, workplace business communication, and unfamiliar social communication. Each knowledge entry is accompanied by a typical Chinese direct translation error case, providing standardized material support for cultural analysis content in subsequent teaching feedback. The learner dynamic data storage partition is used to persistently store each learner's original records of text detection, translation transfer feature parameters, cross-cultural risk assessment results, personalized ability profile tags, and other time-series learning data. Within the partition, an independent storage folder is created according to the learner's unique ID to achieve isolation of individual learning data. At the same time, historical practice records that exceed the storage period are automatically compressed to balance storage usage and data integrity.

[0041] The first-level processing unit of the server computing module is the translation migration tracing and identification unit. This unit establishes bidirectional communication connections with the front-end client interaction module and the underlying data storage module. It is mainly used to receive the learner's standardized English text transmitted by the client and complete a series of pre-processing operations, including text preprocessing, fine-grained cross-language bilingual alignment, multi-dimensional deviation feature extraction, and hierarchical error type discrimination. It is the core foundation unit for this system to achieve deep integration of translation technology and English teaching, and completely breaks through the technical limitations of traditional English correction systems that rely solely on monolingual text analysis and cannot trace the underlying causes of errors. The translation transfer tracing and identification unit incorporates a lightweight phrase-level bilingual alignment model. This model is based on the XLM-R pre-trained model and is fine-tuned only for the Chinese-English communicative phrase alignment task, resulting in low computational resource consumption and fast response speed. After receiving English text input, the unit first performs three layers of text preprocessing operations: sentence segmentation, word segmentation, and syntactic dependency parsing. This breaks down the complete English text into independent semantic phrase units, marking the syntactic components, semantic functions, and speech act types corresponding to each phrase. After preprocessing, the unit automatically retrieves the bilingual alignment knowledge base in the data storage module and performs fine-grained matching and mapping between English semantic phrases and corresponding Chinese native language semantic units one by one. During the mapping and matching process, three types of core deviation feature data are extracted simultaneously: literal translation matching deviation features, sentence transfer features, and semantic unit misalignment features. The literal translation matching deviation feature is used to quantify the overlap between English phrases and native Chinese phrases in word-for-word matching. The higher the overlap, the more obvious the learner's literal translation transfer behavior. The sentence transfer feature is used to compare the degree of matching between the learner's English sentence structure and the sentence logic of native Chinese, and to identify English expressions that copy Chinese word order and Chinese connection logic. The semantic unit misalignment feature is used to locate phrase fragments where the semantic boundaries between Chinese and English do not match, and to identify expressions whose semantic scope has been expanded, narrowed, or shifted due to literal translation. The unit incorporates a multi-feature joint discrimination program, which automatically stratifies and distinguishes between two types of errors based on the comprehensive scores of three types of deviation features: The first type is ordinary grammatical errors, including spelling mistakes, tense misuse, subject-verb disagreement, fixed collocation errors, and misuse of articles and prepositions—superficial linguistic errors. These errors are caused solely by learners' insufficient vocabulary accumulation and weak grasp of grammatical knowledge, and are not related to native language translation transfer. The second type is cross-cultural pragmatic errors caused by native language translation transfer. These texts do not contain superficial grammatical errors such as spelling, tense, or collocation, but the overall sentence structure and semantic expression logic completely copy Chinese native language expression habits. In the corresponding communicative context, this violates the common social norms of native English speakers and is a latent communicative defect that traditional monolingual detection models cannot identify. After completing error classification and labeling and multi-dimensional deviation feature extraction, the translation transfer source tracing and identification unit integrates complete error source feature information, error type labels, and deviation phrase location coordinates, and synchronously transmits them to the downstream cross-cultural communication risk quantification and grading unit, providing accurate underlying feature support for the quantification and grading of communication risks.

[0042] The cross-cultural communication risk quantification and grading unit connects with the upstream translation migration tracing and identification unit to fully receive all error tracing feature information. Based on a pre-set three-dimensional quantitative evaluation system, it automates the risk level determination of translation migration-related pragmatic errors. This unit transforms cross-cultural communication problems, which were previously only qualitatively descriptive and could not be automatically assessed, into calculable, quantifiable, and standardized grading indicators, addressing the industry pain point that existing intelligent teaching systems cannot distinguish between different degrees of cultural communication deviations. The three-dimensional quantitative evaluation system constructed in this unit is strictly based on the Brown and Levenson cross-cultural pragmatic politeness principle and the rules of difference between Chinese and English speech acts as its underlying theoretical benchmarks. It divides the system into three independent quantitative dimensions: scenario formality, pragmatic suitability, and cultural conflict. Each dimension has a continuous quantitative score ranging from 0 to 100 and a differentiated weighting coefficient. The formality of the scenario is assigned a base score based on the learner's current practice scenario. The weighting coefficient is 0.7 for daily casual communication scenarios, 0.85 for campus teacher-student communication scenarios, 1.0 for workplace business communication scenarios, and 0.95 for unfamiliar social communication scenarios. The higher the formality of the scenario, the higher the comprehensive score of the communication risk corresponding to the same pragmatic deviation. The pragmatic fit dimension quantifies the difference between the learner's output expression and the standard English native language communication paradigm. The larger the difference, the more prominent the pragmatic defects such as lack of politeness level and insufficient euphemism. The cultural conflict dimension quantifies the degree of conflict between Chinese literal translation and British and American social taboos, personal privacy boundaries, and mainstream cultural cognitive habits. It focuses on identifying literal translations that are likely to cause offense, such as those involving age, income, marriage and love, religion, and appearance evaluation. The unit incorporates a multi-feature weighted scoring algorithm, which reads the quantified scores and corresponding weight coefficients of each of the three dimensions to perform a weighted summation operation. Based on the comprehensive score range, all translation transfer pragmatic errors are standardized and classified into three risk levels: mild adaptation deviation, moderate pragmatic inappropriateness, and severe cultural conflict. A comprehensive score of 0-30 indicates mild adaptation deviation, where the literal translation deviation only results in awkward English sentence structure and obvious Chinese influences, without affecting the recipient's complete understanding of the core semantics. There are no issues of social rudeness or cultural offense; only a slight defect in the expression's naturalness. A comprehensive score of 31-70 indicates moderate pragmatic inappropriateness, where the literal translation transfer results in a lack of politeness in the English expression, a stiff and direct tone, and does not conform to the standard social norms of the corresponding scenario. This can easily cause slight embarrassment in the communication process and a poor experience for the recipient, exhibiting a clearly perceptible pragmatic deviation. A comprehensive score of 71-100 indicates moderate pragmatic inappropriateness. This category is divided into severe cultural conflicts. Such literal translations directly touch upon taboos and personal privacy boundaries in British and American culture, and contain obviously offensive and transgressive expressions. They are very likely to cause serious cultural misunderstandings and social conflicts, resulting in irreversible cross-cultural communication barriers.After the cross-cultural communication risk quantification and grading unit completes the standardized grading of a single pragmatic error, it simultaneously pushes the risk level label, the original quantitative score of the three dimensions, and the deviation scenario identifier to the downstream risk adaptive communication equivalence generation unit, providing a grading basis for the scheduling of differentiated text optimization strategies.

[0043] The risk-adaptive communicative equivalence generation unit connects the upstream cross-cultural communication risk quantification and grading unit with the underlying data storage module via bidirectional communication. Its core function is to dynamically activate differentiated text generation constraint strategies based on the received risk level labels, generating communicative equivalence optimized texts that fit the communication scenario, adapt to English cultural norms, and match the learners' learning situation. At the same time, it retrieves cross-cultural communication norms knowledge base to match corresponding knowledge points on differences between Chinese and Western cultures, and integrates and generates integrated teaching feedback content with a complete structure and clear logic, completely abandoning the one-size-fits-all and homogenized text correction mode of traditional English teaching systems. The unit pre-sets three independent text generation constraint strategies, corresponding to three risk levels. The constraint weights, rewriting extent, and retention rules are clearly differentiated between these strategies: For the mild adaptation deviation risk level, a semantically preservative lightweight polishing strategy is employed. The core constraint of this strategy is to preserve the learner's original sentence structure, core vocabulary, and expression logic to the greatest extent possible, making only minor adjustments to phrases with traces of Chinese literal translation. This significantly reduces the rewriting extent, correcting awkward expressions while avoiding extensive rewriting that could undermine beginners' confidence, thus meeting the needs of basic English communicative practice. For the moderate pragmatic inappropriateness risk level, a communicative paradigm structure replacement strategy is employed. The constraint of this strategy is to fully preserve the learner's... The core semantic information is replaced entirely with sentence structures, greetings, and requests that do not conform to English communication etiquette norms. Simultaneously, poorly adapted Chinese-style communication phrases are replaced to output authentic, compliant, and polite standardized communication expressions, suitable for the needs of communication practice in regular formal scenarios such as campuses and workplaces. For severe cultural conflict risk levels, a mandatory cultural adaptation correction strategy is implemented. The strategy's constraints include completely reconstructing expressions that contain cultural offenses, social transgressions, or touch upon cultural taboos, thoroughly eliminating all literal translations that could easily trigger cross-cultural conflicts, strictly adhering to the underlying communication norms of the target language culture, eliminating communication risks at the root, and adding cultural warning labels to highlight learners' avoidance of similar literal translation errors. The entire text generation process is synchronized and linked with built-in learning-based adaptation constraints. Each unit retrieves and updates learner profiles in real-time, matching learners' current English vocabulary and sentence structure mastery levels. This strictly constrains the vocabulary difficulty, clause complexity, and proportion of long and difficult sentences in the optimized text, preventing the output of sentences beyond the learner's current learning stage and obscure vocabulary. This ensures the optimized content is easy for learners to understand, memorize, and reuse. After the communicative equivalence optimized text is generated, the unit accurately retrieves cross-cultural communication norms knowledge from the data storage module based on the speech acts and translation deviation types corresponding to the current pragmatic errors. It then retrieves matching analysis content on differences in Chinese and Western communication habits, ultimately integrating this information into standardized, unified teaching feedback. The feedback content consistently includes four modules: pragmatic error location markers, explanations of the causes of translation transfer, communicative equivalence optimized text, and explanations of differences in Chinese and English communication norms. The complete feedback data package is transmitted to the front-end client interaction module for visual display.

[0044] The server computing module is also independently configured with a learner profile update unit. This unit establishes real-time data communication links with the translation transfer source identification unit, the cross-cultural communication risk quantification and grading unit, and the risk-adaptive communication equivalence generation unit. After a single teaching feedback is pushed down, the unit automatically collects all computational time-series data of the learner's English text, including the original error source data, multi-dimensional feature parameters of literal translation transfer, cross-cultural risk grading records corresponding to all pragmatic errors, and the frequency of deviations under various speech acts, continuously iterating and updating the learner's personalized learning profile. The learner profile system adopts a structured tag storage mode, mainly constructing two types of core tag sets: the first type is the literal translation behavior preference tag, used to record learners' frequently occurring literal translation phrases, fixed Chinese-style transfer sentence patterns, and frequently occurring literal translation deviation scenarios, accurately capturing learners' normalized native language literal translation thinking errors; the second type is the cross-cultural communication ability level tag, which is divided into three levels of ability gradients according to four major communication scenarios, quantifying the learner's ability to avoid cultural pragmatic errors in different scenarios. All updated learner profile tags and time-series learning data are synchronously written to the learner dynamic data storage partition of the data storage module for long-term retention and retrieval for subsequent operations. When the system subsequently conducts specialized communication training pushes, tiered teaching difficulty scheduling, and targeted exercise recommendations, it uses learner profile data as the core scheduling basis to achieve full-process teaching empowerment from single text error correction feedback to long-term personalized cross-cultural communication ability improvement.

[0045] The entire system operates as a closed-loop, intelligent teaching processing chain, with interconnected data across the entire chain, unit-driven progress, and logically sequential, without any independent or isolated functional modules: The client interaction module collects and encrypts the learner's English text before transmitting it to the server processing module; the server processing module first completes fine-grained bilingual alignment and literal translation transfer feature extraction through the translation transfer tracing and identification unit, accurately distinguishing between ordinary surface grammatical errors and cross-cultural pragmatic errors induced by literal translation from the native language, addressing the shortcoming of traditional systems in being unable to locate the cause of errors; the tracing features output from the upstream unit flow into the cross-cultural communication risk quantification and grading unit, which uses a three-dimensional quantitative scoring system to complete the standardized grading of communication conflict risks, enabling calculable evaluation of cross-cultural communication deficiencies; the grading results are input into the risk-adaptive communication equivalence generation unit, which dynamically matches corresponding optimization strategies to generate communication equivalence texts that are suitable for the learner's learning situation and cultural norms, and simultaneously retrieves cultural knowledge bases to generate complete teaching feedback; the learner profile update unit collects full-process calculation data in real time to iterate personalized learning profiles, providing data support for long-term differentiated instruction. This entire technical system truly achieves a deep integration of machine translation analysis technology and intercultural communication teaching content. It's not a superficial integration solution like conventional English teaching platforms that simply pile up translation tools and cultural materials. It breaks down the three major technical barriers at the underlying computational logic level: translation mechanism analysis, intercultural risk assessment, and adaptive intelligent teaching. This systematically addresses several core technical shortcomings of existing intelligent English teaching systems, such as prioritizing surface-level language error correction, neglecting the cultivation of intercultural communication skills, failing to identify implicit literal translation transfer pragmatic errors, and exhibiting homogenized feedback strategies. The system's underlying computational solution is lightweight and has low barriers to entry. It can be deployed in standardized intelligent teaching equipment in universities and primary and secondary schools, or packaged as a standalone mobile learning software for individual learners. It possesses broad application scenarios and high industry promotion value, significantly improving the refinement, professionalism, and intelligence of AI English teaching, and systematically cultivating learners' comprehensive language abilities to avoid literal translation thinking in their native language and to conduct standardized intercultural English communication.

[0046] In the description of this specification, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0047] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0048] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. An AI-powered English teaching system integrating translation and intercultural communication, comprising a client-side interaction module, a data storage module, and a server-side computing module; The client interaction module is used to collect the English text input by the learner and output teaching feedback results to the learner; The data storage module is used to store a preset bilingual alignment knowledge base and a cross-cultural communication norms knowledge base, characterized in that... The server computing module includes a translation migration source tracing and identification unit that links data sequentially, a cross-cultural communication risk quantification and grading unit, and a risk-adaptive communication equivalence generation unit. The translation transfer source identification unit is used to perform cross-language fine-grained bilingual alignment on the collected learner English text, extract the literal translation matching deviation features in the text, distinguish between ordinary grammatical errors and cross-cultural pragmatic errors caused by literal translation transfer from the mother tongue, and output the corresponding error source identification feature information. The cross-cultural communication risk quantification and grading unit is connected to the translation migration source identification unit. It is used to receive the error source identification feature information and perform weighted calculations based on three types of quantitative features: scenario formality, pragmatic fit, and cultural conflict. The translation migration pragmatic errors are divided into three risk levels: mild fit deviation, moderate pragmatic inappropriateness, and severe cultural conflict. The risk-adaptive communicative equivalence generation unit is communicatively connected to the cross-cultural communication risk quantification and grading unit. It is used to call the corresponding differentiated text generation constraint strategy according to the determined risk level, generate optimized text with communicative equivalence, match the corresponding cross-cultural communication analysis content, and output integrated teaching feedback information through the client interaction module.

2. The AI-powered English intelligent teaching system integrating translation and intercultural communication as described in claim 1, characterized in that, The translation transfer tracing and identification unit constructs cross-language semantic mapping relationships through a phrase-level bilingual alignment model, and combines sentence transfer features and semantic unit misalignment features to comprehensively determine whether learners' English texts have native language translation transfer problems.

3. The AI-powered English intelligent teaching system integrating translation and intercultural communication as described in claim 1, characterized in that, The cross-cultural communication risk quantification and grading unit is based on the cross-cultural pragmatic politeness principle and the rules of difference between Chinese and English speech acts. It constructs a quantitative feature system of various dimensions to complete the graded judgment of pragmatic error risk.

4. The AI-powered English intelligent teaching system integrating translation and intercultural communication as described in claim 1, characterized in that, The differentiated text generation constraint strategies of the risk-adaptive communicative equivalence generation unit include: for mild adaptation bias, a semantically preserved lightweight text polishing strategy is adopted; for moderate pragmatic inappropriateness, a communicative paradigm structure replacement optimization is performed; and for severe cultural conflict, a mandatory cultural adaptation correction is implemented.

5. The AI-powered English intelligent teaching system integrating translation and intercultural communication as described in claim 4, characterized in that, The risk-adaptive communicative equivalence generation unit simultaneously matches the learner's existing language level and constrains the sentence difficulty of the optimized text during the generation of optimized text, ensuring that the output content is adapted to the learner's learning situation.

6. The AI-powered English intelligent teaching system integrating translation and intercultural communication as described in claim 1, characterized in that, The server computing module is also equipped with a learner profile update unit; the learner profile update unit is used to collect error source data and cross-cultural risk classification data of learners in previous text detections, dynamically update learners' literal translation behavior preference tags and cross-cultural communication ability evaluation tags, and construct learners' personalized ability profiles.

7. The AI-powered English intelligent teaching system integrating translation and intercultural communication as described in claim 1, characterized in that, The integrated teaching feedback information includes at least pragmatic error location markers, analysis of the causes of translation migration, optimized texts for communicative equivalence, and explanations of differences in Chinese-English communicative norms.

8. The AI-powered English intelligent teaching system integrating translation and intercultural communication as described in claim 1, characterized in that, The client interaction module supports receiving written English text input by learners, receiving spoken English audio from learners, and converting the audio into standard English text through speech-to-text technology, which is then sent to the translation migration tracing and identification unit for detection and analysis.