Graph structures in comparison and directionality of change: A case study in the domains of perception & cognition

This paper presents a methodological approach that integrates crosslinguistic data with language-specific historical data. It focuses on graph structures, specificaly comparing semantic maps and colexification networks, and demonstrates how diachronic data, such as diachronic colexifications (i.e., meaning extensions within a single language), can be incorporated into both types of graphs. The case study draws on the history of Greek and explores the semantic fields of PERCEPTION and COGNITION, using Vanhove’s (2008) study on crosslinguistic semantic associations in these domains as a starting point. Vanhove examines associations between vision, hearing, and mental perception across 25 languages, involving 46 lexical items coexpressing at least two of the folowing meanings: SEE, HEAR, HEED, UNDERSTAND, KNOW, LEARN, THINK, OBEY, REMEMBER. Linguistic typology has a long-standing tradition of using diagrams to represent such data. The most prevalent representation, the semantic map, was introduced in the early 1980s (Anderson, 1982) and developed further in the folowing decades (e.g., Haspelmath, 1997; van der Auwera & Plungian, 1998; Croft, 2001). More recently, colexification networks have also gained prominence (e.g., Jackson et al., 2019). This paper will examine the key differences between these two graph structures, particularly in how they incorporate diachronic data. Despite these differences, both share several similarities, such as nodes representing concepts and edges representing colexification patterns, and both can integrate information like frequency of attestation. Examples of weighted (incorporating this information) and unweighted graph structures (without frequency data) are presented in Figures 1-4, al based on Vanhove’s dataset. For the diachronic analysis of Greek, which covers three stages - Homeric Greek, Classical Greek, and Helenistic Greek - we wil adapt a protocol developed by AUTHORS, involving three steps:
1) Onomasiological step: Begin with the concept and identify the different lexemes with which the concept can be named (see Geeraerts, 1997) using dictionaries and resources
providing information about the relevant lexemes such as Montanari (2015) and Buck (1949). Examples include the verbs: horáō, dérkomai, leússō, akoúō, klúō.
2) Semasiological step: Chart the different meanings of these lexemes across different
stages, focusing on those present in Vanhove’s synchronic network.