Chapter 9. What counts in the retention of numeral classifiers in Japanese and Chinese?

This study compares L2 acquisition and attrition sequences of the syntax and semantics of numeral classifier systems in light of considerations of markedness, frequency, and the regression hypothesis. In classifier data elicited from adult learners and attriters of two unrelated languages, Japanese and Chinese, both exhibit in the attrition of syntax and semantics, a regression of the acquisition sequence. An implicational semantic scale, the Numeral Classifer Accessibility Hierarchy, coinciding closely with the relative frequencies of the classifiers in input, is examined as a path of least resistance for the learning and loss of the semantic systems.