Lecture | CHiLL series
A Computational Approach to the Segmental and Tonal Classification of Yue Dialects
- Date
- Wednesday 10 December 2025
- Time
- Serie
- Chinese Linguistics in Leiden (ChiLL)
- Address
-
Lipsius
Cleveringaplaats 1
2311 BD Leiden - Room
- 1.31
Abstract
Traditionally, dialectologists would identify bundles of isoglosses in order to classify dialects into different regions (Chambers & Trudgill 1998: 96). However, these maps tend to show feature(s) which are considered to be significant. Feature selection based on a dialectologist’s experience is highly subjective, as dialectologists actively look for isogloss bundles which are considered to be the boundaries. This methodology can easily introduce selection biases to the analysis. Active feature selection might be the cause of scholars showing discrepancies in their classifications. It is necessary to find a more objective way to address these issues.
The problems stated above can also be found in Yue-Pinghua dialectology. In order to reduce the selection biases, this presentation will introduce two novel classifications of Yue-Pinghua using dialectometric approaches. Dialectometry is the measurement of dialect similarities or distances. Through calculating dialect distances, we are able to aggregate a substantial amount of data (opposed to choosing certain features) and gain a better idea of the relationships between dialects by analysing these distances using other computational methods such as cluster analysis and multidimensional scaling.
In this presentation, we will be looking at Yue-Pinghua dialect variation on two linguistic levels, namely segments and tones. Dialectometry has existing toolkits for analysing segmental variation, but not so much for tonal variation. Using a recently developed technique called dialect tonometry, we are able to tap into tonal variation within the dialectometric workflow, which then allows us to compare the segmental vs. tonal variations in the Yue-Pinghua dialect landscape.