research methodology

[Arnold Zwicky]

i taught a seminar this spring quarter on syntactic variation,
in which almost all the students used standard corpora or web
searches to find examples of particular usages.  this makes a
great first step, but after that the students began to piece out
pictures of clusters of idiolects, differing in many different
details, and for this purpose, corpora alone aren't particularly
useful, since you need to know what *isn't* natural for specific
speakers; at some point, you have to collect judgments from
individuals.

some of the students are pursuing these projects further, hoping
to do further work with specific people from the web searches
(or from adventitious collection of examples on the fly).  has
anyone explored this two-step methodology (finding potential
informants in the first step of data collection, which nets people
who have some feature, then exploring the limits of that feature
through further data collection, interviews, judgment tasks, etc.)?

(the amount of systematic and structured variation among individuals
on the details of specific constructions is quite impressive, by
the way.)

arnold (zwicky at csli.stanford.edu)