What will it take for MIDI to actually produce words, so that you hear someone singing lyrics in a MIDI file?

MIDI is not designed to support lyrics in any fashion. It is a tool designed to be used by musicians. MIDI supports communications with instruments in a manner similar to that in which they are used in performances, which is not at all the same as constructing lyrical passages. MIDI supports 128 notes (numbers 0 to 127) per voice or tone, each of which maps to a specific note or to a standard percussive sound (e.g., low tom 1, high tom 1, kick drum 1, etc.).