Carl W. Conrad was discussing hapax legomena on the B-Greek discussion list the other day and he made a brief aside about the odd situation of discussing hapax(c?)es in the plural since by definition a hapax is unique and singular.
I have been thinking about the issue of definiteness and individuation lately, and plurals are an interesting case because it is difficult to express a plural with specific referents in mind. By nature, once you begin grouping things together you have abstracted a step away from a specific referent. Ancient Sumerian actually had a different marker for an individuated plural, ḪI.A instead of MEŠ. For instance you could say there are kings (LUGAL.MEŠ) in Mesopotamia without any specific kings in mind, or you could say there are kings (LUGAL.ḪI.A) in Mesopotamia: Bob, Joe, and Larry. How often would you really need to specify an individuated plural though? Indeed, the “individuated” force seems to weaken as it is brought into Akkadian. I don’t know if any other language has a similar plural marker.