Corpus-based Low Saxon dialectometry

In this corpus-based study, we explore how the similarity of Low Saxon dialects
among each other and to the state languages Dutch and German has changed from
the 19th century to today. In particular, we want to investigate if the traditional
classification into an eastern and a western group is visible in the data and if the
Low Saxon dialects can be found to diverge at the Dutch-German border.

We apply principal component analysis and hierarchical clustering to n-grams of
characters, part of speech tags and morphological features and observe divergent
developments at the separate levels. As a reflection of different orthographic
traditions, a noticeable distance between Dutch Low Saxon and German Low Saxon can
be attested at the character level. At the PoS and morphological level, we however
find a particular closeness between Dutch Low Saxon and the northern dialects
from Germany, while we see German Westphalian in an outlier position. A shift
towards the state languages can be observed at the PoS level, but the overall distance
between Dutch Low Saxon and German Low Saxon does not seem to markedly
increase at the three levels we studied.