cllg f mn pntd t t m vr smpl t vr ntrstng bsrvtn tht s wrth shrt pst. t’s bvs hmn lnggs r qt rdndnt, lrd knw wht th lst tw lttr f ths sntnc __. Hwvr, nt ll lttrs br sm mnt f nfrmtn – mst f t s n cnsnnts nd vwls r jst pddng fr mr cmfrtbl prnnctn. Tht s wh sll ndrstnd rgnl ccnts, nl vwls r trnsfrmd, th nfrmtn s kpt thgh.

A colleague of mine pointed out to me very simple yet very interesting observation that is worth a short post. It’s obvious human languages are quite redundant, you already know what the last two letter of this sentence a__. However, not all letters bear same amount of information – most of it is in consonants and vowels are just padding for more comfortable pronunciation. That is why you usually understand regional accents, only vowels are transformed, the information is kept though.

You can try it for yourself for instance with such a pipe

echo "Héllo world!" | iconv -f utf8 -t ascii//TRANSLIT | tr -d "aeiouyAEIOUY"

If you find the first paragraph unintelligible try the following one.

A oeaue o ie oie ou o e ey ie ye ey ieei oeaio a i o a o o. I oiou ua auae ae uie eua, you aeay o a e a o ee o i eee a__. oee, o a ee ea ae aou o ioaio – o o i i i ooa a oe ae u ai o oe ooae ouiaio. a i y you uuay uea eioa ae, oy oe ae aoe, e ioaio i e ou.