Can we say "language representation that is not a transcription of a spoken one"? A sequence of symbols could have meaning without mapping to sound, and taken at face value the separation on the map would seem to imply that any spoken language could have evolved dramatically out of sync with any physical representation. I can't think of any reason to think that - if you assume a phonetic interpretation - a symbol shared between North America and southern Africa would be pronounced at all similarly in both locations when the marks were made. The distances alone argue against a phonetic interpretation to me.
Do hobo signs count as a language? That seems intuitively much closer to what this might be. How much structure do you need?
"Can we say 'language representation that is not a transcription of a spoken one'?" Of course: sign languages. But sign languages have even more "phonemes" than spoken languages do, so this doesn't help the hypothesis.
And yes, "A sequence of symbols could have meaning without mapping to sound": that's what hieroglyphic writing systems were, more or less. Again, more distinct symbols, not fewer.
"Do hobo signs count as a language?" Depends on how you define "language". As most linguists would define language, though, the answer is no. All normal spoken or signed languages have oodles of structure (grammar). Pidgin languages probably do not, but that's just the first generation or so, after which they gain structure and are technically creoles. (Some creoles have the word "pidgin" in their names, like Tok Pisin of New Guinea, and Hawaiian Pidgin, but they're complex enough to justify the term "creole.")