But the pauses are semantic in a way. Those concepts cannot be well separated. Not a linguist, but I can tell pauses are used by speakers and listeners for a variety of reasons, linked with both semantics and delivery -- to separate parts, to give and demarcate "processing time" for speakers and listeners, to create and satisfy expectation, etc. The cadence of speech is semantic (though its semantics are not of a simple propositional logic type).
There was a long tradition in English and maybe also in other European languages of choosing punctuation depending on how long a pause a speaker would take at the point, with a comma denoting a shorter pause than a semicolon, which in turn denoted a shorter pause than a colon, which in turn denoted a shorter pause than a period -- instead of using punctuation to convey some semantic distinction as most writers do today. The comma splice might be a vestige of that tradition.