Ainu & Gilyak, Japanese & Korean

[Richard M. Alderson III]

Johanna Nichols writes:
 
>what, in a nutshell, is the evidence that Japanese and Korean are related?
>I've read as much of the relevant literature as I could find, and the only
>support offered seems to be that if you assume they are related you can find
>sound correspondences and apply the comparative method.  But what is the
>evidence for assuming relatedness in the first place?
 
I'm a little confused by the question.  The assumption of relatedness is the
hypothesis tested by the application of the comparative method; what evidence
is needed to form such an hypothesis?  As historical linguists we may decide
heuristically that some such hypotheses are too difficult to pursue based on
the number of cognate sets we may expect to find, but there is never (_pace_
the Greenberg/Ruhlen camp) a requirement for any particular sort of evidence of
relationship prior to starting the comparison.
 
Put another way, hypotheses of relationship, like hypotheses in any scientific
discipline, are made first by the imagination of the researcher, and then put
to rigourous test.  The researcher's experience and training in the field, any
field at all, provides a first-cut filter on hypotheses in that field--which
the researcher is free to ignore if a particular insight seems to call for it.
 
What have I misunderstood?
 
                                                                Rich Alderson