It's amazing, but the Direct Method never worked. It was fabricated out of whole cloth by the U.S. Army (believe it or not) during World War II. The original purpose was to train commandos to speak all kinds of behind-the-line Tagalog so they could complete their ranger missions in the South Pacific (sounds like I'm making this stuff up doesn't it? but the academic references are available, if you want them).

After the war, it was re-dubbed 'The Audio-Lingual Approach,' and has since become the basis for most ESL curriculums. Its dialogue segments, pair work, imaginary situations, and all the rest of its very characteristic pedagogy, never had any underlying linguistic theories to call upon for support, and have all consistently been adjudged by academic surveys to yield extremely poor results. These results are seen, by many, to go below zero: that is to say, studying language in this curriculum probably creates learning blocks.

Anyway, its rules, like "No Speaking Your Native Language In Class," "No Dictionaries," and particularly "Listening 50% of the Time -- Speaking 50% of the Time," have always been bogus; and have been known to be so for at least 40 years.

Going back to that 2-year old kid: If you demand she speaks as much as she listens, what results do you predict you'll get?