Tuesday, August 31, 2010

•nd•rst•nd•ng t•xt w•th••t v•w•ls

In written language, consonants are more informative than vowels simply because there are more of them and each one occurs less often. You c•n •nd•rst•nd t•xt q••t• w•ll w•th••t v•w•ls, because the consonants carry enough information to determine the words.

But •a• •ou u••e•••a•• •e•• •ui•e •e•• •i••ou• •o••o•a••s? No you can’t! (That was : can you understand text quite well without consonants?)
(Ancient Hebrew was, I believe, written with consonants only, which leads to occasional ambiguities, in instances where a word could plausibly contain more than one variant vowel, yielding different meanings.  This has given rise to difficulties in interpreting early manuscripts of the Hebrew bible, though I don’t have an example to hand.)
But in spoken language, it’s the other way round. In contrast to text, studies of people listening to speech show a ‘vowel superiority effect’; that is, vowels seem to carry more information, because listeners perform more accurately when understanding sentences with missing consonants than with missing vowels.
The categories of vowels and consonants, however, are problematic as units of information. Researchers are looking for new categories such as using phoneme transitions, onsets and offsets, and other more fine-grained components. 
Did I understand that last sentence? No! I haven't a clue!  But the first bit about vowels and consonants was cool.   
It’s from an article in the 12th August edition of Nature, headed Information Theory, by one Michael S. Lewicki.   I visit the Boole Library at University College Cork once a week on average and I always make it my business to look through recent copies of Nature.  Often an article will catch my eye where I can only glean a few nuggets of information before having to abandon the enterprise entirely, but even so, it’s worth it!

No comments:

Post a Comment