Linked Cognates in the Data

[LinguList]

In order to make useful cognate judgments on the data, it is essential to consider cases of roots shared across concepts. These should ideally be listed and then one should work by considering them all at once.

Examples include:

• numerals (one recurs in eleven, two in twelve, etc)
• kinship and extended kinship terms (woman, female, man, husband, all contain roots for "woman/female" or "man/male")
• phonologically interesting items that show deep preservation, as they have a similar phonology but are not related, like "I / five / tail", or "heart / noon / old", which colexify in most language varieties in the sample
• pronouns (you formal vs. you plural)

Ideally, these cases are systematically collected in a document.

[LinguList]

I will put them into etc/cognates.tsv for now.