Short sample from Summarize Evidence
The summary of
evidence, for which Wordcorr may put out pages and pages like the
example, proceeds in conventional phonetic order, with phonetically
similar but contrasting protosegments like b and p
(prefixed with an asterisk *) shown in proximity. The environment
for each cluster is given, followed by a number of reconstructed
forms (also asterisked) as evidence for the cluster.
has a gloss and a measure of its plausibility. The
measure is positive [1.0] when every correspondence set in the
reconstructed form is backed by a lot of identical correspondence
sets in other forms, and is close to [-1.0] when few or none of the
correspondence sets are attested in any other forms. Then the
actual raw data follow, each preceded by the short names of
one or more speech varieties that attest the datum.
In the top example,
all the varieties back up *buta 'stone'. Eleven of them have
initial *b in bunga as their word for 'flower', and one has a reduplicated
form. Eight have initial *b in *buwa 'fruit' attested
as buwa, one as buwa, and one as bua.
The other two have words that cannot be traced back to *bua,
so they contribute no evidence for the protosegment *b.
could trigger complementary correspondence sets, but 'flower,
fruit, mountain' make it likely that 'moon' has a different source
than a simple *b. The more data we run through Wordcorr, the
clearer the major historical patterns become. For patterns that
don't show consistency, genetic development is probably not the
full explanation of how they got that way, and we must look for
borrowings and internal analogies.
here to see the other display, Output
The Web software is not as sophisticated as Wordcorr's, so on this
Web page we use temporary fixes like writing ng for the
velar nasal and ? for glottal stop.
|Again, Nathan Davis put together the behind
the scenes components that make this possible. What you see here
will be available in Release 2.1, still in preparation.