[Corpora-List] Comparing Word/Pattern Association Strength Across Sub-Corpora

[Brian Schanding]

Hello,

Sorry for what may be a novice-level CL/stats question:

I'm interested in getting the association strength between particular words
and specific patterns in three sub-corpora (e.g., NOUNS that occur in the
pattern *the *N *of*).

Attempting to follow methods in publications, I wanted to do this by
applying a Fischer Exact test within each sub-corpus. I can then observe
how strongly these words associate with the patterns in rank order within
each corpus.

But what can I do next to find out if the word/pattern associations are
stronger or weaker in one sub-corpus compared to the other sub-corpora? Is
it just a matter of visually observing differences in the p log values in
each corpus or is there a statistical test I could do to show the degree to
which the corpora differ in word/pattern strength?

Thanks in advance for your thoughts!
Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/corpora/attachments/20131210/85a9302c/attachment.htm>
-------------- next part --------------
_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora