Solving Galton’s problem: practical solutions for analysing language diversity and evolution

Abstract

Comparisons between languages can illuminate the processes of language change, by revealing meaningful associations between language features, or the influence of external factors on the patterns and rates of language change. But comparisons between languages raise statistical challenges, because close relatives will tend to be more similar to each other in many respects than they are to more distantly related languages, and languages from the same regions will be subject to many of the same influences. Therefore observations made on different languages will usually fail to meet the requirement of statistical independence inherent in all standard statistical tests. This fundamental challenge of cross-cultural analysis, known as Galton’s problem, is no cause for despair, because there are a range of solutions that fit comfortably within the scope of most projects, using only data that are easily available for all languages. This paper discusses a range of practical solutions, including phylogenetic analysis, sister pair comparisons, and spatially structured models, that can be applied to analyses of language variation and change

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 12 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.