29 Aug 2011
Julius Pokorny reconstructed the Proto-Indo-European (PIE) etymon *dʰǵʰū- (which was later updated to *dʰǵʰuh₁-). This was to explain apparent cognates in Baltic (Lithuanian žuvìs, Lettish zivs) and Greco-Armenian (Greek ἰχθύς ikhthús; Old Armenian ձուկն jukn) languages.
Yet given the geographical restriction of the cognate set and given the strangely similar Ugaritic word, *dagu [dg] 'fish', why should we not instead ponder a more recent source by way of a Mediterranean loanword?
 Please note that the Greek reflex contains an irregular prothetic i- while a far more widely evidenced root, *dʰǵʰōm 'earth', has become khthōn. If we wish as linguists to obey our own sound laws, this 'fish' root must be abandoned in favour a more recent origin.