Exploring Linguistic Productivity of ChatGPT: A Comparison to L2 Learners in English Blending
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This study examined the linguistic productivity of ChatGPT via two English blending experiments with L2 learners and analyzed the size of the blend and the source word order. The results support previous research by demonstrating that the length of the longer source word and the second source word influence the size of blends. The results also indicate that the relative lengths of the source words influence the order of source words in blends, particularly when the second source word is longer than the first. However, the patterns of blending created by humans and ChatGPT were not identical. The responses of human participants also demonstrate the effect of the presence of onsets in the first syllable of source words on the order of source words in blends, whereas those of ChatGPT do not. These results highlight the differences between human and artificial language production and call for additional research to improve LLMs such as ChatGPT.
Abstract
This study examined the linguistic productivity of ChatGPT via two English blending experiments with L2 learners and analyzed the size of the blend and the source word order. The results support previous research by demonstrating that the length of the longer source word and the second source word influence the size of blends. The results also indicate that the relative lengths of the source words influence the order of source words in blends, particularly when the second source word is longer than the first. However, the patterns of blending created by humans and ChatGPT were not identical. The responses of human participants also demonstrate the effect of the presence of onsets in the first syllable of source words on the order of source words in blends, whereas those of ChatGPT do not. These results highlight the differences between human and artificial language production and call for additional research to improve LLMs such as ChatGPT.
참고문헌 (18)
- 송상헌, "챗GPT, 언어학 시험을 보다: 협동연구 보고", 한국코퍼스언어학회 -한국영어학회 2023년 여름 공동 학술대회 (2023)
- 오영림, "A Comprehensive Study of English Blends with Noisy Harmonic Grammar", 박사, pp.2020
- Algeo, J., "Blends, a structural and systemic view", American Speech, 52, pp.47~64 (1977)
- Arndt-Lappe, S., "The role of prosodic structure in the formation of English blends", English Language and Linguistics, 17(3), pp.357~563 (2013)
- Bauer, L., "English word-formation" (1983), Cambridge University Press
- Begus, G., "Large Linguistic Models: Analyzing theoretical linguistic abilities of LLMs", arXiv: 2305.00948, ~ (2023)
- "Blending in morphology" (2019)[URL]
- Berg, T., "Linguistic structure and change: An explanation from language processing" (1998), Oxford University Press
- Cai, Z. G., "Does ChatGPT resemble humans in language use?", arXiv: 2303.08014, ~ (2023)
- Cannon, G., "Blends in English word formation", Linguistics, 24(4), pp.725~753 (1986)
- Choi, J. H., "ChatGPT Goes to Law School", Journal of Legal Education (Forthcoming) (2023), SSRN
- Elisa, M., "A corpus-based analysis of new English blends", Lexis. Journal in English Lexicology, 14 (2019)
- Gilardi, F., "Chatgpt outperforms crowd-workers for text-annotation tasks", Proceedings of the National Academy of Sciences, 120(30), pp.e2305016120~ (2023)
- Gries, S. T., "Cross-disciplinary perspectives on lexical blending, Berlin", pp.145~167 (2012), De Gruyter Mouton
- Kelly, M. H., "To ‘brunch’ or to ‘brench’: Some aspects of blend structure", Linguistics, 36, pp.579~590 (1998)
- Kubozono, H., "Phonological constraints on blending in English as a case for phonology-morphology interface", Yearbook of Morphology, 3, pp.1~20 (1990)
- Kung, T. H., "Performance of ChatGPT on USMLE: Potential for AI-assisted medical education using large language models", PLOS Digit Health, 2(2), pp.e0000198~ (2023)
- Shaw, K. E., "Emergent faithfulness to morphological and semantic heads in lexical blends", Proceedings of Phonology 2013 (2013), The Linguistic Society of America