There are two interesting questions here. The first is an ontological one. Is a token on screen something different from the pattern of light? If it's the same as the pattern of light, then there is at most one token, there being at most one relevant pattern of light (perhaps none, if our ontology doesn't include patterns of light), though this token is a token of pee, and a token of rho and a token of er. If a token is not identical with a pattern of light, then we might as well keep on multiplying entities, and say that there is a pattern of light and three tokens, of pee, rho and er, respectively, with the first entity constituting the latter three.
The second one is a philosophy of language one. What determines whether or not the pattern of light is or constitutes a token of, say, rho? Is it my intentions? If so, then indeed we have tokens of pee, rho and er, as making these was my intention, but we do not have a token of the Coptic letter ro or a token of the letter qof in 15th century Italian Hebrew cursive, since I didn't think of these when I was doing the drawing. Is it the linguistic context? But then it's not a token of any letter, since a displayed png file in an analytic philosophy post is not a the kind of linguistic context that determines a token.
Or is it that the pattern of light is or constitutes tokens of all the letters it geometrically matches, whether or not it was intended as such? If so, then we also have a letter dee (just turn your screen). But now suppose a new alphabet is created, and it contains a letter that looks just like the drawing. It would be odd to say that if a new language were created on another planet this instantly would multiply the entities on earth (at the speed of light? faster?). So it seems that on this view, we should say that the pattern of light is or constitutes tokens of all the letters in all the alphabets that will ever exist. But future actions shouldn't affect how many things there now are. So on this view, we should be even more pluralistic: the pattern of light is or constitutes tokens of all the letters in all possible alphabets.
We thus have two questions: one about ontology and one about what is being tokened. Both questions have parsimonious and profligate answers. The parsimonious answer to the ontology question is that there is one thing, which can be a token of multiple things. The profligate one is that we have many tokens. The parsimonious answers to the language question are that intentions and/or context determines what's been tokened. The profligate answer has an infinite amount of tokening.
We probably shouldn't combine the two profligate answers. For then on your screen there are infinitely many physical things, all co-located (and some perhaps even with the same modal profile). That's too much.
That still leaves three combinations. I think there is reason to reject the combination of ontological profligacy with parsimony on the philosophy of language side. The reason is that tokens get repurposed. Consider a Russian who has a Scrabble set and loses an er tile. She then buys a replacement pee tile, as it looks pretty much the same (I looked at online pictures--both have value 1 and look the same). Then it seems that a new entity, a token of er, comes into existence if we have ontological profligacy and linguistic parsimony. Does a mere intention to use the tile for an er what magically creates a new physical object, a token? That seems not very plausible.
That leaves two combinations:
- ontological and linguistic parsimony
- ontological parsimony and linguistic profligacy.