Skip to content

Latest commit

 

History

History
13 lines (8 loc) · 1.11 KB

README.md

File metadata and controls

13 lines (8 loc) · 1.11 KB

BUNDLE

BUNDLE finds isogloss bundles (recurring feature distributions) in a dialect continuum. Given a binary feature matrix (N features by L languages) it will construct clusters of features that have similar distributions over the L languages. This is the opposite of the typical clustering model which clusters objects by their features. Other notable properties of BUNDLE are:

  • It will automatically infer the number of clusters in the data.
  • It incorporates a simple attestation model that assumes that a certain fraction of the features in each language will go unattested. This fraction varies from one language to the next.

BUNDLE does not plot the results, but here is what a typical result would look like if plotted. This shows how 422 Polynesian etyma, grouped into twelve clusters, are distributed over 35 Central Pacific languages. Each cluster consists of etyma with similar distributions.

The distribution of Polynesian words