In your “sample 2” the child’s voice and the demonic derivative are synchronised , definitely not “half a second behind it”.
If you must have the demonic voice lagging behind, not more than 0.05 seconds would be my suggestion, otherwise the mixture will be incomprehensible.
If you want to play the original and demonic tracks together omit the step 2 above : the optional 5% slowdown.
“Reverse” is in the Audacity “Effects” Menu.,it flips the track back to front : like playing a vinyl record (ask your parents 
) backwards.
To create the reverse reverb effect, reverse the track, apply reverb, then reverse it back the right way round. There is no “unreverse” just reverse the reversed.
Gverb produces a horrible metallic sound and does not lend itself to the reverse reverb technique.
This ANWIDA reverb effect is very good and free.
I’ve found a way of improving the Demon voice. Make a copy of the Demon voice you’ve made using the above instructions to run simultaneously with the first. Then apply the equalization on this graph to the copy.
The resulting mixture then has plenty of high frequency content and sounds a lot better: clearer & scarier . A before after example is attached …
[Applying plenty of reverb only to the copy track with the high frequency content produces an ethereal effect]