Network science in a dynamic usage-based approach

Abstract

In this study, we test the ability of network science to capture linguistic development over time in individuals from a dynamic usage-based perspective, a combination of a complex dynamic systems theory (CDST) approach and usage-based (UB) linguistics. Network science is designed to quantitatively analyze entire systems and captures complex interrelationships between components of those systems. We select network science measures that have potential to represent theoretically predicted individual language learning processes, specifically focusing on the development from prototype to schematized constructions at the macro- (network), micro- (word), and meso‑level (syntactic structure). To test this approach, we traced the beginning L2 Chinese development of two English speakers. The results are generally aligned to dynamic usage-based predictions in terms of prototypicality, schematization variability, and variation. Additionally, the network science approach allows us to identify a core set of words that emerges at the early stage of learning and is highly connected, seeming to drive development; it also suggests that variability plays different roles at different levels of analysis. Although it is preliminary to make strong conclusions about this set of words, it suggests that, as with other areas of inquiry, network science can reveal previously unidentified information about linguistic development and should be further explored.