1. 서론 2. 이론적 배경 및 선행연구 검토 3. 분석 방법론 4. 추출 후 조립 아키텍처와 정량 검증 5. ‘미래파’ 논쟁 분석 결과
6. 결론을 대신하여
This study reconstructs a single literary debate, the 2000s Korean Miraepa ("Futurist") poetry controversy, using semi-automated Text Encoding Initiative encoding aided by generative artificial intelligence and corpus-assisted discourse studies. We propose an "extract-then-assemble" architecture in which a large language model only performs entity identification and semantic judgment, while deterministic code governs document structure, generation, and schema validation; its accuracy is initially tested against a gold standard drawn from outside the corpus. On this basis, 61 critical essays (2005-2011; approximately 156,000 content-word tokens) are divided into three periods and analyzed using collocation, keyness, naming dispersion, discourse prosody, and camp-based contrast, with each measure validated by close reading of concordances. Results reveal that Miraepa serves not as a stylistic predicate but as a metadiscursive term for naming and dispute. The discursive center shifts from "modernity/lyric" toward "poetry and politics," and the term itself corresponds to the critical rather than the advocating camp. The study's contribution is twofold: first, the findings themselves, and second, the reproducible workflow that rendered them observable. |
