Productivity in acquisition of grammar

[Elizabeth Bates]

In "From first words to grammar" we actually had a horse race of sorts
among different criteria for calculating productivity, to see how
they correlated with one another, and with other language measures
in our longitudinal data set.  The criteria were applied at 28 months.
Data were available at 10, 13 and 20 months as well.  Quantity-based
criteria that rely on numbers of cases of a construction were relatively
unreliable within this data set (we had 27 children, but speech samples
were relatively short), and did not show high correlations with other
measures.  More qualitative criteria (relying on appearance of
over-regularizations; or same verb/noun appears in more than one
morphological form) showed substantially larger correlations with
other variables, and seemed to line up with a more 'analytic' stream
of measures in factor analyses.  There is a chapter dedicated to these
analyses in the book (Cambridge University Press, 1988).  -liz bates