Corpus L. Question: What counts as a word in BNC spoken?

[Sivodsi]

Hi there,
I'm working with a list of words taken from transcripts, and want to compare them to the frequencies found in the BNC spoken, the list found at ucrel.lancs.ac.uk/bncfreq/lists/2_2_spokenvwritten.txt.

However I'm having a hard time trying to find the exact rules that were followed for defining 'word' in this corpus. For example, how did the BNC count multiword lexical items?

From scrutinizing the list you can find multiwords like 'brand new', 'even when', 'by now' with their own frequency count, and yet you find 'new' listed as "NoP~" with the diacritic mark indicating that its part of a noun like 'New York'... (I think). It seems inconsistent to me.

So if anybody can find BNC's guidelines for their spoken corpus online I'd be very grateful.

[5jj]

I am not sure if this helps, but some of the links in it may.

Multiwords and associated tags in BNC2