I would define a Pawn as a piece of which you start with many, and has a value so low that they act merely as 'change' when other pieces of unequal type are traded.
Irreversibility or promotability IMO are not requirements. E.g. when a chess variant fills the second rank of the setup with pieces that move and capture like an inverted Y (fWbF), I think most people would consider those 'Pawns'.
Large Shogi variants tend to have multiple piece types of very low value. (But the Asian Pawn is significantly weaker than a Shatranj Pawn too.) E.g. a Shogi Knight (fN), Stone General (fF), or Go-Between (vW). People don't consider those Pawns, though, (even though they of course all promote in Shogi, and the first two are irreversible), because you only start with a pair of them.
That's a very good try at a definition, H.G.
I'd thought of something similar, in that I gave great weight to 'pawns' (in the context of a given CV) being most numerous in the setup of that given CV. Then I recalled Napoleonic Chess, which used a small number of Steward Pawns and a (only relatively) small number of Berolina Pawns, on top of the more numerous type of pawns, for each side's army.
I'd note that I haven't seen Steward Pawns as the most numerous piece type in any CV I've seen so far, unlike for Berolina Pawns.
Napoleonic Chess is the reason why I tried using a tentative definition of a 'pawn' being either numerous AND/OR promotable. Promotability being included as a criteria would allow for Napoleonic Chess, except I then thought, what if in Napoleonic Chess the Steward Pawns were replaced by Ultima Pawns (also known as Pincer Pawn), which are not promotable (on top of that there is shogi, e.g. with its small number of promotable knights)? So, I'm afraid we both might still be at square one when it comes to defining what a CV 'pawn' is. Unless, in the context of Napoleonic Chess alone, you wouldn't consider Steward or Berolina types as 'pawns' for that particular CV.
That's a very good try at a definition, H.G.
I'd thought of something similar, in that I gave great weight to 'pawns' (in the context of a given CV) being most numerous in the setup of that given CV. Then I recalled Napoleonic Chess, which used a small number of Steward Pawns and a (only relatively) small number of Berolina Pawns, on top of the more numerous type of pawns, for each side's army.
I'd note that I haven't seen Steward Pawns as the most numerous piece type in any CV I've seen so far, unlike for Berolina Pawns.
Napoleonic Chess is the reason why I tried using a tentative definition of a 'pawn' being either numerous AND/OR promotable. Promotability being included as a criteria would allow for Napoleonic Chess, except I then thought, what if in Napoleonic Chess the Steward Pawns were replaced by Ultima Pawns (also known as Pincer Pawn), which are not promotable (on top of that there is shogi, e.g. with its small number of promotable knights)? So, I'm afraid we both might still be at square one when it comes to defining what a CV 'pawn' is. Unless, in the context of Napoleonic Chess alone, you wouldn't consider Steward or Berolina types as 'pawns' for that particular CV.
https://www.chessvariants.com/large.dir/contest/napoleonic.html
https://www.chessvariants.com/play/pbm/play.php?game%3DNapoleonic+Chess%26settings%3DNormal