For a layman’s definition of “cognate”, it doesn’t really matter though it waters down the concept considerably.
In the field of linguistics, “cognate” absolutely precludes borrowing of vocabulary.
For the heck of it, I went and checked out the etymologies for dacha and borscht. Dacha is cognate of Latinate words like date and data (which, yes, are borrowings) – the sense of the associated Proto-Indo-European root was “something given”. Borscht is a cognate of German Borste and the first syllable of English bristle (the -le is originally a diminuitive ending) – the sense of the associated Proto-Indo-European root was “tip, point”.
When the topic is a linguistic one, one might consider using relevant terminology as used in linguistics, not as they might appear in a general purpose dictionary, in a secondary or tertiary definition with little detail. Especially if that usage blurs a significant distinction, and doubly so if that distinction is directly relevant to the conversation.
Yep, it’s better to use terms as used in the field of discussion than more general definitions. It’s not that the general definitions are wrong, but they’re less precise and less useful when in specialist discussions. Part of the charm of SDMB is that we often have people who can explain the technical terms.
For example take the Modern English words “flower” and “flour”. They are cognates, as they are both derived from Middle English “flour”, which is a loan word from Anglo-Norman French “flur”.
(This also brings up a lumper-splitter decision: are “flower” and “flour” two separate words in Modern English, or are they one word with variant spellings associated with particular connotations?)
