[CentOS] recursively count the words occurrence in the text files

I just can't google for it:

I'm searching for a "bash" "one liner" (awk, perl, or anything) for this: 

there are text files, in several directories: 

mkdir one
mkdir two
mkdir three

echo "word1 word2 word3" > one/asf.txt
echo "word2 word4, word5" > one/asfcxv saf.txt
echo "word1. word2" > one/dsgsdg.txt

echo "word6, word3!" > two/sdgsd dsf.txt
echo "word6" > two/ergd.txt

echo "asdf, word2" > three/werdf.txt
echo "word7, word8 word9 word10" > three/qwerb erfsdgdsg.txt
echo "word4 word3" > three/web erg as.txt

so it does the magic* "recursively":

$ SOMEMAGIC > output.txt
cat output.txt
asdf 1
word1 2
word2 4
word3 3
word4 2
word5 1
word6 2
word7 1
word8 1
word9 1
word10 1
$



*recursively count the words occurrence in the text files like: "word1 2"
can anyone point to a howto/link? [re: i just can't google for it :\]