At 8:58 PM +0000 3/22/05, Arwel Parry wrote:
Stan Shebs wrote:
David Gerard wrote:
Gerard Meijssen wrote:
The idea is that you cannot reliably pronounce a word as it it should be pronounced just by seeing the characters when the word originated from another language. By making these resources available, it is clear how they should be pronounced in the original language. Having pronunciations available is important because they help people study a language and, the wikipedia articles are a great resource to learn a language; they are short, cover a subject well and many of the related words related to a subject can be found in the article.
This could be nightmarish in English - accents are widely variant and in Britain are used as markers of social status to a ridiculous degree. This is of course highly politicised. I confidently predict ten or more sound files per word.
It used to be said that an experts in English could get to within 50 miles of a person's birthplace just by listening to the dialect. Dunno if that's still true in this more mobile age. In any case, supplying the life background of the speaker is critical, just like identifying the location of a picture, and I hope everybody is doing that for uploaded pronunciations.
50 miles? Even in these more mobile days that's nothing. For some communities, real experts could locate a persons' accent down to a few streets.
FWIW
In England, within the first 30 seconds of someone opening their mouth, its still possible to place a person to within half a county or less purely from pronouciation.
With longer listening that picks up key marker vowels its also possible identify what class of person they spend their time with - via either community, work or travel. That is particularly true of immigrants who learn their English from different classes of contact.
e.g. an Australian who lived in Canada and then moved to S. London bears the footprints in their accent.
Peter