Generative Linguistics and Corpus Linguistics: Two Paradigms of Language Study and Their Applications to AI
Kyu-Hong Hwang
(Dong-A University/Professor)
This paper discusses generative linguistics and corpus linguistics from the perspective of their philosophical foundations, research methodologies, and basic notions. It focuses on how they make distinct but complementary contributions to the development of AI. It is shown that generative linguistics and corpus linguistics arrive at language from fundamentally different directions in that the former is rationalism, reasoning-oriented, and based on deduction and introspection while the latter is empiricism-oriented and resorts to induction and data. Moreover, generative linguistics is said to focus on generic features of language, mentalism, and linguistic competence, but corpus linguistics is claimed to stress the sociality of language and linguistic performance. It is also held that both generative linguistics and corpus linguistics have made contributions to AI development, especially in the field of NLP, machine translation, and information retrieval. In particular, generative linguistics can provide theoretical support for language-related aspects of AI, and corpus linguistics can help AI by providing tools to analyze various aspects of a language system based on powerful real data and by generalizing language patterns based on probabilistic information. The paper then suggests that generative linguistics and corpus linguistics can be used in a symbiotic methodology to boost AI study.