View Chapters

Book: Machine-Aided Linguistic Discovery

Chapter: Inferring Simplest Laws/Patterns: MINTYP and the Problem of Describing a Typology

DOI: 10.1558/equinox.25591

Blurb:

Given a dataset, a common problem in scientific knowledge discovery is to summarize this set by a collection of rules (laws, patterns, etc.) such that the resultant description is the simplest, or most economic. In linguistics, the problem occurs e.g. in attempts to describe a linguistic typology in terms of the smallest set of implicational universals that allow all actually attested, and none of the unattested, language types. In this chapter, I introduce the MINTYP (Minimum TYPological description) program, which handles this problem, illustrating it on the typologies in Greenberg’s Appendix II (Greenberg 1966a) and Hawkins’ Expanded Sample (Hawkins 1983).

Chapter Contributors