Perhaps a better way to think of the split is that some of the audio content needs to be processed deliberately and sequentially—it has grammatical structure, words that need to be decoded, semantics, etc. It requires processing from the part of our brain that needs to deliberately pay attention to things. It's like reading or doing math.
Meanwhile, some of it can be processed more intuitively. It's where we pick up tone and emotional intent. We will absorb that even if we don't deliberately focus on it—you can tell that someone is angry when they speak even if you don't know their language.
Of course, understanding both of those requires integrating them together. The lexical content of their speech might tell you why they feel a certain way or what they feel that way about.
But one way to look at music is that it's a sound artifact designed to deliberately convey "I'm almost all non-grammatical, it's OK to not focus on me."
Again, of course, this is all pure conjecture.
That can't possibly be true for actual information (in the Shannon sense).
When you say to someone "go to the grocery store and buy a soda", these words can't be just 7% of the total information. Or put another way, what exactly do you communicate in the other 93% (13 times more information)
In a sense, there is information to be had, whether or not you want or need to have it is totally up to you and perhaps the situation. Low stakes situations it probably matters less, negotiating with something big on the line would be a different story.
I think if humans always told the truth or could always express themselves fully in words, then yes, verbal communication would be a bigger chunk. But the truth is that humans can hide behind words or not know quite the right way to express themselves, but it’s much harder to do with tone and body movements.
Is the person smiling? Yelling? Pointing angrily? Making a motion like they are thirsty? Making puppy eyes? Laughing like it is an inside joke?
It can change from a mere friendly request to a demand possibly backed up by violence.
Surely way less than in the actual sentence, since you have millions of different posibilities for the words: go/store/soda
The speaker's motivation for that statement, the speaker's attitude towards the soda getter, the speaker's attitude towards soda, the implied outcome/motivation for the soda getter, the state of beverage availability and thirst in the immediate vicinity, the proximity of the store...etc
Every time someone says something we have to model what different interpretations imply about the mental state of the speaker, compare that with the recent conversational history, the tone and body language of the speaker (and other listeners), our knowledge of the speaker and the broader context. At the same time, each new statement can cause us to refine our past mental models of the speaker and reinterpret the interpretations based on those models.
Imagine a conversation with zero non-words information (a transcript).
Now one with zero words (mime game) .
Which one conveys more information?
Also, I don't see a way to actively separate information and emotion while someone is communicating to you while still being able to fully understand...that would just mean you are not paying attention to at least one of the signals...but maybe that IS what is happening when people pay more attention to body language and tone over the actual content.