영어와 한국어 음운이웃 네트워크의 정량적 분석

초록

Focusing on the phonological neighborhood defined by Turnbull and Peperkamp (2017: 83) as “two words are neighbors of each other if they differ by the deletion, addition or substitution of one and only one segment,” we analyzed the characteristics of phonological neighborhood networks (PNN) in English and Korean words in order to investigate the interrelation between words in these two languages. Here, a PNN is assumed to represent the mental lexicon. In the case of English, 33,329 high-frequency words were selected as target words from the Corpus of Contemporary American English (COCA) (Davies 2008), and in the case of Korean, 32,698 high-frequency words were selected as target words from SJ-RIKS (Kang and Kim 2009). Using R (R Core Team 2016) and Pajek (de Nooy et al. 2011), we formed the PNN matrices of both languages, and using these matrices, measured the key indicators of each language’s PNN, such as the number and size of sub-PNNs in each PNN, the proportion of the giant component size, average shortest path length, average clustering coefficient, and assortative mixing by degree (AMD). Through a quantitative analysis of these key indicators, it was shown that the network measurements reflect a language particularity between English and Korean, and at the same time the two PNNs shared the characteristics of a “small-world network” (Watts and Strogatz 1998) and a high value of AMD.

키워드