The author challenges the idea that linguistic resemblances between languages are too complex to be random, citing human nature’s tendency to find patterns and trust them. Linguists caution that many resemblance lists lack linguistic expertise and overlook borrowings and native compounding. The author proposes a simple statistical model for estimating the probability of chance resemblances and adjusts it based on specifics of language comparisons. The model considers factors like phonological and semantic leeway to determine expected matches. A computer program generates and compares random languages to demonstrate expected matches and highlight the importance of comparison methodologies in assessing linguistic connections.
https://www.zompist.com/chance.htm