Well, as I described here. Whatever semantic content we assign to it. I think I see your point - you mean the semantics of the script's creators? Yes, good poin

Well, as I described here. Whatever semantic content we assign to it. I think I see your point - you mean the semantics of the script's creators? Yes, good point, and no not that cos we don't know anything about what they meant, nor likely ever will. Any semantics we assign to it are a number (unicode value), possible collation (ordering), and the name (possibly other stuff, I don't know). So if we found more phaistos disks with punky-mohawk-guy on that, and we decided they were intended by the original creators to mean the same 'character' on all disks, all those glyphs would get the same unicode character. Scholars could then index them by the same 'character', look for relationsips between those characters they've established are the same. More concretely, what a letter is in english doesn't link much to how it looks. For example, the uppercase letters here https://en.wikipedia.org/wiki/File:Gebrochene_Schriften.png https://en.wikipedia.org/wiki/File:Gebrochene_Schriften.png are all english but to my eyes, each taken individually, are grotesques and some (like the capital 'H' in Textur font), are illegible to me. Literally I would not guess what letter it is. Yet they're all your common or garden alphabetical letters. So the semantics are, whatever the 'H' looks like, we know what it is, how to sort with it, where to find it in an (english) dictionary, etc. HTH (and IANAExpert)