Orthographic networks in the developing mental Lexicon. Insights from graph theory and implications for the study of language processing

Abstract

In this study, we examine the development of orthographic networks in the mental lexicon using graph theory. According to this view, words are represented by nodes in a network and connected as a function of their orthographic similarity. With a sampling approach based on a language corpus for German school children, we were able to simulate lexical development for children from Grade 1-8. By sampling different lexicon sizes from the corpus, we were able to analyze the content of the orthographic lexicon at different time points and examined network characteristics using graph theory. Results show that, similar to semantic and phonological networks, orthographic networks possess small-word characteristics defined by short average path lengths between nodes and strong local clustering. Moreover, the interconnectivity of the network decreases with growth. Implications for the study of the effect of network measures on language processing are discussed.