How does that strengthen his point? Koko stumbled around frequently, well after the learning period was advertised as done. Kids stumble around less than was claimed, as you mention.
You're asking an animal to internalize a system of communication quite literally designed from the ground up for different hardware. It's like asking why does my Oculus headset take so long to communicate over ZigBee. What would be magical is that it did it at all. Lamenting how long it took to bend the Oculus hardware into some contorted SDR would be stupid.
My point is you shouldn't fudge auxiliary data in making a point. I don't know if I can believe your actual argument due to that.
Consider, my dog can fumble around trying to make me happy with really good emotional understanding of my reactions and basic mimicry. I wouldn't claim that is a usage of language, though. From the counters here, there is an argument that if my dog was dextrous with his arms and hands, he could pass off similar results.
I didn't say it makes it a strong point. But deceitful numbers greatly weaken it.
That is, with how off the idea that 4 year old can't take language is, I didn't make it to the argument. And I am sympathetic to the idea that animals can pick up on language.
I think you are grossly missing the point.