I understand that they must modify many of their words through context or other words, but it's not arbitrary. A Chinese person could tell you that he had a bad day yesterday, is having a good day today, and hopes to have a good day tomorrow. He could use words to substitute for verb tense. And there would be sensible and non-sensible ways to do that.
Our bonobos and other chips and gorillas don't do anything of the sort. The closest they come is making strings of associations. They could say "You" "Walk" "Here," perhaps (though I'm not completely sure they could do that- it's theoretically possible without language). They could potentially come up with many combinations like that, and even grammatically correct sentences could be learned by repetition and repeated back. They aren't going to use rules about word order, though they may use word orders that get the best results. They won't form new sentences that combine parts of speech in ways that they have not been taught, and be consistent in coming up with new sentences that follow the same rules. That's one of the ways they could demonstrate language.
I'm not close to an expert so I'm not confident taking that much further, but I would accept it's language if the ape could do something like telling you about her poop every day, where she pooped, and describing how it was different from other poops- and correcting errors by the trainer when the trainer says the poop is a different kind of poop. If the ape could do that every day, consistently ordering the words, without gibberish or making a lot of errors, and then start applying whatever sentence structure she chose to other activities, then I think I would have to accept that as language. Maybe another simpler example could be made by somebody who know better than me, but that's the sort of thing I have in mind.