In the last few months I’ve just about finalized the extended version of the tone sandhi paper. Here is the current version of the abstract:

This study documents speaker variation and lexical variation in the phonological tone change patterns of Jinhua Wu Chinese, one dialect in a region known for particularly complicated tone sandhi systems and regional variation. Previous reports of Jinhua tone sandhi have agreed that there is some lexical variation, but they have disagreed considerably about the contour shapes as well as about mergers between tone categories. It was not clear whether these were differences in methodology, stimuli selection, speaker idiolect, or systematic regional or generational differences. I recorded a word list spoken by 15 speakers of Jinhua Wu from two locations and two generations, and carried out an acoustic analysis of the pitch tracks. The analysis proposes new methods of contour parameterization and tone categorization. The results indicate that there are systematic regional and generational differences in Jinhua sandhi, as well as lexical variation. Also, the space of tone contours is highly constrained, apparently by a combination of coarticulation and the obligatory contour principle (OCP). The generational differences show active tonal evolution that is predicted by theories of sandhi arising from tonal coarticulation. The complexity of the data and the role of phonetic constraints in the phonological patterns require an explanation that appeals to historical change as much as to synchronic grammar.

And here is a new picture of the distances among speakers’ contour systems:

The older speakers (blue border) are on the left and the younger speakers (red border) are on the right. The people from the village (yellow fill) are together in the lower half of the figure.