ITU-R BS.1770-1 or -3 are measurement standards, they don't prescribe any limits. It gives a way of measuring the subjective "loudness" of a program based on a psychoacoustic model but it presumes total control over the speaker system (which TV doesn't provide), and it doesn't say "how loud."
EBU R128 gives a single standard, and you use it with ITU-R BS.1770. The problem is that it treats a dialogue-heavy program the same as a musical program; a musical program has a lot more signal, over a half hour average, than a dialogue one, so a musical performance will tend to sound quieter when put next to a dialogue heavy one, given they're mixed with the same level normalization scheme.
The CALM Act is actually based on Dolby Laboratories technical definitions and the dialnorm subcode metadata in an ATSC bitstream actually has to be decoded and properly enforced. It's not actually LAW but it's an adopted FCC federal regulation. Dolby's standard is to measure the average dialogue level in the program, and only the dialogue, and to use that to derive the normalization level -- EBU R128 uses the entire program mixed together, dialogue, music and sound effects. I think Dolby's approach is superior but more technically demanding, since it requires the person encoding the AC3 bitstream to have access to the dialogue mix-minus, but on professional productions this isn't an issue.