Important to note, however, that spoken language information density (per syllable) and syllabic rate are strongly inversely correlated [1]; languages such as Mandarin with lots of information contained in each syllable tend to be spoken slower than languages like Spanish, which uses many more syllables to convey the same amount of information, but in a similar amount of time.
So, the compactness really is more just on the writing side.
It's not the DIV-soup that is an accessibility problem (though that's what often takes the blame), it's the lack of adding accessibility attributes (`aria-*, tabindex, title`, etc.). Lots of builtins, especially form elements, give you many accessibility traits "for free", but there are a wide variety of ways to tag a bowl of DIV soup for accessibility.
Then there's "semantic HTML(5)" (`<article>, <section>`, etc.), which still exists, but is not _particularly_ worthwhile for accessibility: Too much of the web lacks these tags, so most common consumers of this nature (scrapers, screen readers, content summarizers) use other, often AI-based, strategies to tag content semantics.
So, the compactness really is more just on the writing side.
[1] "A cross-language perspective on speech information rate" https://pdfs.semanticscholar.org/16e6/43a9e39674c58acd7cdf56...