[1] Terminology here may be confusing.
Following common practice in recent literature, I use "representation" in a
fairly broad sense, which includes even linguistic representations. Goodman
does not -- for him, "representation" means somehing like "iconic
representation", or possibly even "pictorial representation", since he
distinguishes between the representational and the (merely) diagremmatic. He
does not, therefore, count languages as representational. What he claims,
strictly, is that similarity is irrelevant to all denotation, but this
is uncontroversial for what I call non-iconic or linguistic representations.
[2] He therefore claims that all symbol schemes which are at
all "representational" in his sense (see note 1) must be dense.
[3] Or a plagiarism!