Earticle

The purpose of this study is to investigate different types of errors identified from the essays written by 68 Asian EFL learners at a Korean university. The learner corpus containing these essays consisted of 2498 errors; based on the schemes suggested by previous researchers, those errors were analyzed according to error classification, i.e., sentence fragments, run-ons, comma splices, word order, verb tense, verb form, subject verb agreement, word form, determiners, prepositions, and word choice. Of all the error types committed by college-level learners, word choice had the highest error frequency, accounting for around 33%. Word choice issues in EFL students’ writing could be due to a traditional way of English learning such as rote memorization of vocabulary and purely receptive reading skills. Thus, instructor suggestions on practical learning methods concerning word choice in writing can be certainly helpful and teachers would also need to pedagogically utilize EFL written word choice strategies in a more effective way.

본 연구는 베트남인 한국어 학습자의 발음 오류를 모국어 전이와 중간언어 발음의 관점에서 분석하였다. 국립국어원 학습자 말뭉치의 934건 오류 데이터를 기반으로 숙달도 및 학습 환경에 따른 분포를 검토한 결과, 중간언어 발음이 약 76%로 가장 높은 비중을 차지하였다. 카이제곱 검정 결과, 오류 유형 간 차이(χ²(1)=138.8, p<.001)와 숙달도와의 관계(χ²(5)=19.87, p<.01)는 유의미하였으나, 학습 환경에 따른 차이는 유의미하지 않았다. 이는 발음 오류가 중간언어 발달 과정의 체계적 산물임을 시사한다. 이에 본 연구는 단계 기반 발음 교육 모형과 AI 기반 학습 적용 가능성을 제안한다.

This study examines pronunciation errors of Vietnamese learners of Korean from the perspectives of L1 transfer and interlanguage pronunciation. Based on 934 error tokens from a learner corpus, the results show that interlanguage pronunciation accounts for approximately 76% of all errors. Chi-square tests reveal significant differences across error types (χ²(1)=138.8, p<.001) and proficiency levels (χ²(5)=19.87, p<.01), while no significant effect of learning environment is found. These findings suggest that pronunciation errors are systematic outcomes of interlanguage development. A stage-based instructional model and AI-assisted pronunciation learning are proposed.

본 연구는 DeepSeek가 한국인 중국어 학습자의 작문 오류를 수정하는 데 있어 어떠한 양상과 특징을 보이는지 분석하고, 이를 인간 교사와 비교한 연구이다. 이를 위해 글로벌 중국 어 중간언어 코퍼스에서 한국인 학습자의 오류 문장을 무작위로 추출하여 DeepSeek-R1 모 델과 인간 교사의 수정 결과를 분석하였다. 주요 연구 결과는 다음과 같다. DeepSeek는 재현율(Recall, 95.86%) 면에서 인간 교사 (84.83%)보다 우수하여 학습자의 오류를 거의 놓치지 않고 찾아내는 높은 민감도를 보였다. 그러나 정확도(Precision)는 49.47%로 인간 교사(85.42%)에 비해 현저히 낮았는데, 이는 오류가 없는 부분까지 수정하는 과도 교정 현상이 빈번했기 때문이다. DeepSeek는 오타 수정 및 어휘 오류 식별에 강점이 있으며, 문장을 더 문어체적이고 격식 있게 다듬는 경향이 있다. 반면, 인간 교사는 최소 수정 원칙을 준수하며 학습자의 본래 의도와 어조를 최대한 유지하는 신중함을 보였다. DeepSeek는 대규모 데이터를 빠르게 처리하고 수정하는 데 유용하며, 인간 교사는 문맥 에 맞는 자연스러운 표현과 교육적 관점에서의 정확한 피드백을 제공하는 데 탁월하다. 결론 적으로 인공지능과 인간 교사의 협력은 중국어 교육 현장에서 수정의 효율성과 정확성을 동 시에 높일 수 있는 실질적인 방안이 될 수 있다.

This study analyzes the patterns and characteristics of DeepSeek in correcting writing errors made by Korean learners of Chinese and provides a comparative analysis with human teachers. For this purpose, error-containing sentences from Korean learners were randomly extracted from the Global Chinese Interlanguage Corpus, and the correction results of the DeepSeek-R1 model were compared against those of human teachers. The key research findings are as follows. DeepSeek demonstrated superior performance in terms of recall (95.86%) compared to human teachers (84.83%), showing high sensitivity in identifying almost all learner errors. However, its precision was significantly lower at 49.47%, compared to 85.42% for human teachers, due to frequent over-correction—modifying parts of the sentence that were originally error-free. DeepSeek showed strengths in correcting typos and identifying lexical errors, with a tendency to refine sentences into a more formal and written style. In contrast, human teachers adhered to the principle of minimum revision, exercising caution to preserve the learner's original intent and tone as much as possible. While DeepSeek is highly effective for rapidly processing and correcting large-scale data, human teachers excel at providing natural expressions suited to the context and accurate feedback from an educational perspective. In conclusion, the collaboration between AI and human teachers can serve as a practical approach to simultaneously enhance the efficiency and accuracy of error correction in Chinese language education.

The study explores the common types of grammatical errors in the written interlanguage of intermediate-level first-year high school EFL students. Employing the error classifications of Dulay et al. (1982) and Ferris (2002), the analysis identified verb form errors as the most prevalent (26%), followed by errors in word form (17%), article/determiners (16%), prepositions (15%), and sentence structures (10%). The findings indicate that interlingual transfer is the primary source of errors across all examined categories, while intralingual transfer significantly contributes to errors in tense usage. These results underscore the need for targeted pedagogical interventions focusing on verb forms, word forms, articles, prepositions, and sentence structures to enhance the grammatical accuracy of EFL learns. The study offers practical implications for EFL educators aiming to improve students’ writing proficiency and suggests areas for further research in error analysis and interlanguage development.

This study employs Error Analysis (EA) to examine Spanish essays written by intermediate Korean students enrolled in a university-level ELE course. The students composed these texts without the aid of automatic spell checkers or artificial intelligence (AI). The primary objective was to identify the types and frequency of genuine grammatical errors produced by learners whose native language (L1) is typologically distinct from Spanish. For this purpose, the students produced texts across various discourse modes, including exposition, description, narration, and argumentation, and analysis of these modes made it possible to identify specific areas requiring improvement. Based on the frequency of the classified errors, we conclude that three areas require detailed attention in ELE instruction: agreement, the use and omission of articles, and verb conjugation.

본 연구는 한국어를 모국어로 하는 학습자의 인도네시아어 쓰기 자료를 분석하여 중간언어(interlanguage)에 나타나는 문법, 통사, 어휘 오류의 유형과 원인을 규명하였다. 연구 대상은 사이버대학교 중급 수준의 학습자 67명이며, 동일 주제로 작성된 작문 과제를 기반으로 오류 분석을 실시하였다. 오류는 문법적 오류(시제 부사, 전치사, 대명사), 통사적 오류(주어·대명사 오용, 어순, 부정문 구성), 어휘적 오류(철자, 날짜 표현, 대소문자 표기)로 분류되었다. 오류의 주요 원인은 한국어 구조의 부정확한 전이, 목표어 규칙의 불완전한 내면화, 음운·표기 습관 등으로 나타났다. 특히 의미 충돌을 유발하는 시제 부사의 병치, 전치사 누락, 고유명사의 소문자 표기 등은 중간언어 발달 특성을 반영한다. 분석 결과는 반복적 피드백, 대조언어적 접근법, 자동화 기반 오류 진단 도구의 활용 가능성을 시사한다. 본 연구는 온라인 외국어 교육 환경에서 중간언어 분석이 가지는 이론적·실천적 가치에 주목하며, 향후 교수법 개선 및 학습자 맞춤형 교육 설계에 기여할 수 있다.

This study investigates grammatical, syntactic, and lexical errors found in the Indonesian writing of Korean native speakers, focusing on features of interlanguage. The data were collected from writing assignments by 67 intermediate learners at a cyber university in Korea. Errors were categorized into grammatical (tense adverbs, prepositions, pronouns), syntactic (subject/pronoun misuse, word order, negation), and lexical (spelling, date expressions, capitalization). Major causes included negative transfer from Korean, incomplete internalization of Indonesian grammar, and orthographic interference. These errors reflect developmental characteristics of interlanguage, rather than simple mistakes. The findings suggest that corrective feedback, contrastive analysis, and automated error detection tools can effectively support learners. This study emphasizes the theoretical and practical value of interlanguage analysis in online foreign language education and contributes to the development of more effective teaching strategies tailored to learners' needs.

Hypothetical conjunctions play a key role in expressing conditional relationships and logical reasoning, making them an important grammatical feature in international Chinese language education. For native Korean learners, due to semantic and syntactic differences between the Chinese hypothetical conjunction “如果” and the Korean equivalent “-(으)면,” errors frequently occur, and the difficulty of acquisition is relatively high. This study examined 615 instances of the “如果…就/那么…” structure produced by Korean native speakers in the Global Chinese Interlanguage Corpus, identifying 111 errors, which corresponds to an error rate of 18%. Based on this data, a comparative analysis of “如果” and “-(으)면” was conducted from semantic and syntactic perspectives. Semantically, differences were found in the types of conditions, the strength of logical inference, and the speaker’s attitude. Syntactically, differences emerged in sentence structure, word order, and combination constraints. As a result, learners tend to make cognitive errors caused by transfer based on the “formal equivalence” of the two expressions, which manifests as “cognitive modularization errors. This study combines grammatical analysis with real corpus data to systematically identify the types and causes of errors, providing deeper insight into the acquisition of hypothetical conjunctions by Korean learners of Chinese.

韩国学生在学习中文时，造成偏误现象的主要原因是将母语语法系统套用在中文语法中。为此，本文首先对韩国语中状语类型副词状语、体词与各类副词格词尾结合构成状语、谓词活用型与副词型语尾‘게’结合构成状语、谓词派生体词与词尾‘도록’及依存名词‘대로、듯’结合构成状语等，进行了逐一分析，列举了韩国学生在韩中翻译中容易出现偏误的知识点。其次，韩中翻译中多项状语语序偏误是最为常见的现象，本文选取范围类状语与其他状语共现、语气类状语与其他状语共现、否定类状语与其他状语共现、时频类状语与其他状语共现等，4种典型的状语语序偏误，对其产生原因进行了分析。最后，本文提出了关于状语翻译的一些教学建议，帮助学生掌握韩汉语言状语语法规则差异，提高中文翻译能力。

The main reason for errors among Korean students when learning Chinese is the application of their mother tongue grammar system to Chinese grammar. Therefore, this article first analyzes the types of single adverbs in Korean one by one, and lists the knowledge points that Korean students are prone to errors caused by negative transfer of their mother tongue in translating single adverbs into Korean. Secondly, multiple adverbial word order errors are the most common phenomenon in Korean Chinese translation. This article selects four typical adverbial word order errors and analyzes their causes to help students grasp the differences in grammar rules of Korean Chinese language adverbials and improve their Chinese translation ability.