[Corpora-List] Instances of coordination/pp-attachment ambiguity in the BNC

Mon Apr 28 16:55:22 UTC 2008

Dear corpora members

We are interested in:

How frequently coordination ambiguity of the form 'Adj Noun1 and Noun2' (e.g., old men and women) occurred in the BNC?

How frequently PP-attachment ambiguity of the form 'Verb NP PP' (e.g., buy books for children) occurred in the BNC?

How frequently PP-attachment ambiguity of the form 'NP1 and NP2 PP' (e.g., the men and the women in the garden) occurred in the BNC?

Using Gsearch with an appropriate English Grammar, I have found the following:

1.        Number of files searched = 2169

2.        Number of sentences in the BNC = 3043718 (~ 3.04 millions)

3.        Number of Noun Phrases (NPs) in the BNC = 2391489 (~ 2.3 millions)

4.        Number of occurrences of the NPs of the form 'Adj Noun1 and Noun2' = 60192 (~ 0.06 millions)= 2.52% of the NPs = 1.98% of the sentences

5.        Number of occurrences of PP-attachment ambiguity of the form 'Verb NP PP' = 60906 (~ 0.06 millions)

6.        Number of occurrences of PP-attachment ambiguity of the form 'NP1 and NP2 PP' = 19035 (~ 0.02 millions)

I know that the BNC contains ca. 100 million words, but Gserch output shows that approximately 62 million words are searched (as some of the files couldn't be filtered).

>>From this I can conclude, for example, that 'Adj Noun1 and Noun2' accounts for 2.52% of the NPs in the BNC.

Am I correct? Any suggestions.

Regards

Imtiaz H. Khan
PhD Research Student
Computing Science Department
University of Aberdeen
http://www.csd.abdn.ac.uk/~ikhan/ <https://mail.abdn.ac.uk/exchweb/bin/redir.asp?URL=https://mail.abdn.ac.uk/exchweb/bin/redir.asp?URL=http://www.csd.abdn.ac.uk/~ikhan/> 

The University of Aberdeen is a charity registered in Scotland, No SC013683.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/corpora/attachments/20080428/1845488a/attachment.htm>
-------------- next part --------------
_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora