Quick terminology question: "cognates"

[alice]

Strictly speaking, are two words "cognate" if they ultimately derive from the same root, or just if they derive from a common parent somewhere? Is it correct, for example, to say that all of "cognate", "know", and "can" are all cognates?

[Soap]

Im pretty sure theyd be considered cognates. Either that or a lot of people, including dictionaries, uses it wrong. Im pretty sure Ive seen things like "cognate to KNOW" (the word being compared is often in small caps) in etymologies.

[Niedokonany]

I've always been thinking we should have more terms in this area, not just one.

Some time ago when I was wondering myself it turned out various sources use different definitions. I've been always using a definition according to which borrowing precludes cognacy, so English international and French international aren't cognates, because they aren't inherited from a common proto-language or independently derived from inherited stems. It seems to lead to weird or dubious cases like Latinate borrowings in Romance not being cognate to the respective etymon in Latin. However, I do find it very useful to sift out loanwords in many contexts. But apparently some other people use a definition according to which loanwords from a common source, loanwords and their etymons etc. are cognates.

I also think that derivation doesn't preclude cognateness, otherwise IE languages with their plethora of thematic extensions with an obscure meaning and origin would have few cognate words.

[Ser]

Quick terminology question: "cognates"

Quick except you posted it in L&L, not at Quickies.

I've always been thinking we should have more terms in this area, not just one.

Seconded.

[MisterBernie]

As for number of terms, there is doublet for cognates occuring within one language, although afaik that doesn't distinguish between loans and words deriving from the same root in the same language, either. So frog and spring are doublets, but so are cow and beef.

So, I propose we start a movement for a distinction of cognate types:

• multilingual cognates (cognates in different languages from the same parent)
• monoglossal doublets (cognates within one language, no loanwords)
• polyglossal doublets (cognates within one language, including loanwords)

[Thomas Winwood]

So, I propose we start a movement for a distinction of cognate types:

• multilingual cognates (cognates in different languages from the same parent)
• monoglossal doublets (cognates within one language, no loanwords)
• polyglossal doublets (cognates within one language, including loanwords)

I propose "cognate" for pairs like English five and German fünf and "loan cognate" for pairs like shirt and skirt, or Greek κίνημα "movement" and σινεμά "movie theatre".