english.txt and welsh.txt are training files. I simplified them with help from
simplify.sh

englishwelsh.m is the demo, just pass the funciton some text and it will run an
incredibly naive classifier. Below are some examples:


Getting it right, but with unbelievable confidence:

	>> englishwelsh('this is english')
	"Probability" English is 1.00000
	 
	Log "probability" English is -0.00000
	or "probability" English is 1.00000


	>> englishwelsh('this')
	"Probability" English is 0.99873
	 
	Log "probability" English is -0.00127
	or "probability" English is 0.99873


Numerical underflow if you are not careful:

	>> englishwelsh('this is an english sentence')
	"Probability" English is   NaN
	 
	Log "probability" English is -0.00000
	or "probability" English is 1.00000


Getting it right on long text:

	>> englishwelsh('gwelais bren yn dechreu glasu ei ganghennau yn yr ardd ac yn dwedyd wrth yr adar wele daeth y gwanwyn hardd daeth aderyn')
	"Probability" English is   NaN
	 
	Log "probability" English is -102.02991
	or "probability" English is 0.00000


Getting it very wrong on short text:

	>> englishwelsh('gwelais bren')
	"Probability" English is 1.00000
	 
	Log "probability" English is -0.00000
	or "probability" English is 1.00000

