Perl can be close to this, and Larry Wall was fond of this aspect of the language, but it's also one of the most sigil-infested languages that isn't APL. APL itself stands proudly at the other end of the spectrum, for pure mathematicians.

Lisp has the problem that brackets aren't readily pronouncable and deeply-nested clauses aren't idiomatic English. Although I have noticed that programmers are far more likely to speak English with nested clauses than non-programmers.

Forth relies too heavily on the stack for my taste; ideally for me languages specify the variables they operate on by name, although an English-like language could clearly use "it" in the way that Perl uses "$_".

Rebol, like LOGO, is clearly Lisp with square brackets. Tcl is another candidate in this area that I have a soft spot for, but its reliance on string operations means it needs escape characters a lot.

Ironically I think the closest may be COBOL and its predecessor, Grace Hopper's FLOW-MATIC. These languages "correctly" end statements with a full stop "." rather than the unnatural semicolon ";" or the invisible newline. The problem is that COBOL pre-dates structured programming and object orientation by a long way. I don't think attempting to adapt them directly would be a good idea, but building something with some of the same principles might work.