On concord.

metal56 · Post by **metal56** » Wed Mar 02, 2005 8:18 am

Stephen Jones · Post by **Stephen Jones** » Wed Mar 02, 2005 11:12 pm

Fabulous! Now I don't have to suffer the "you can find anything on Googe" arguments in favour of arbitrary prescriptivism.

metal56 · Post by **metal56** » Wed Mar 02, 2005 11:23 pm

Stephen Jones wrote:Fabulous! Now I don't have to suffer the "you can find anything on Googe" arguments in favour of arbitrary prescriptivism.

Here's an extract from an e-mail I got from Costa Gabrielatos, linguist:

1. Some of the free online concordancers suggested by posters use
the web as a corpus. Granted, the web is much larger than the
largest corpus. To get an idea, consider that Google searches 8
billion web pages, which translates into trillions of words,
whereas the largest corpus (the Bank of English) has just above
half a billion words. However, the web is not representative of
any given language variety, medium or genre - in fact, it
contains both NS and NNS English.

Have fun at the link.

fluffyhamster · Post by **fluffyhamster** » Wed Mar 02, 2005 11:42 pm

metal56 wrote:However, the web is not representative of
any given language variety, medium or genre - in fact, it
contains both NS and NNS English.

Really? You learn something new every day!

Stephen Jones · Post by **Stephen Jones** » Thu Mar 03, 2005 12:04 am

YOu can normally get a fairly good idea of the register of a certain collocation by sifting through the first couple of pages of links.

The byu allows you to refine the search more. Also you can see if the proprotions are the same.

Stephen Jones · Post by **Stephen Jones** » Thu Mar 03, 2005 12:09 am

The proportion of web pages written by North Americans dwarfs that of the other native speaers, and, as long as you exclude the autotranslations, comepletey overshadows the number produced by Non-Native speakers.

Now, where you are likely to get into problems are when you simply use the statistics of the concordancer, wothout direct reference to the links.