The sql-repeat syntax with group-beginning? and group-ending? makes sophisticated use of result sets much more accessible than looking behind/ahead a row manually.
Hoisting a function definition you quickly made within a page out to a sitewide definition is fast and easy, so doing something quick-and-dirty doesn't lock you into dirt forever.
And don't underestimate those square brackets. They totally optimize the human "inner loop" of web development. They beat the tar out of <%, %>, and the mess of quotes, backslashes and dots that constantly trip up newbies.
If Scheme is not popular among web programmers, there must be a reason behind it, more than one reason, maybe a hundred reasons. But it's a fact. So I think if you are going to bring Scheme/Lisp into the web development market, you (I mean, not you personally, but in general) have to bring 2-3 new killer features along with HTML embedding and the language itself.
I'll try to understand BRL better, but from what you say above, those things don't seem to be really killer features. Even square brackets thing isn't a killer, I think.
...The best way to think of BRL is as a Scheme interpreter with an alternative syntax for strings
is where you should look for reasons of BRL's unpopularity, I think.
Can you tell us please, what is really special about BRL other than it being "Scheme with square brackets"?