It might be instructive to collect a
cautionary list of pseudo-cognates, i.e. words so spectacularly similar in form
and meaning that anyone but a linguist (who can _prove_ that they are not
related) would take a connection for granted. Oft-quoted handbook examples
include:
Eng. day / Lat. die:s
'day'
Eng. bad / Farsi bad
'bad'
Gk. tHeos 'god' / Lat. deus
'god'
Mod.Gk. máti 'eye'/ Indonesian mata
'eye'
I'd add things like
Eng. much / Sp. mucho
Eng. freeze / Lat. fri:geo: (no,
<fridge> and <freezer> aren't cognates)
Any ideas? The condition is that the
pseudo-cognacy should be due to pure chance (which excludes onomatopoeia,
nursery words and the like).
Piotr