This paper offers a model for dealing with individuals as well as groups, variability as well as systematicity in L2 research. This model consists of a revised version of the Ordering‐Theoretic Method (Bart and Krus 1973; Dulay and Burt 1974b) in conjunction with implicational analysis as used in sociolinguistics. After a brief introduction to implicational analysis, I describe and illustrate the model presented here with data on the use of 13 grammatical morphemes in English by 89 Spanish‐speaking learners. In the analysis of the data I follow Krashen and his colleagues in separating morphemes into V and NP; free and bound morphemes, as well as DeCamp's (1973:144) view that general implicational scales are ‘…surface composites resulting from simultaneous operation of simple grammatically‐relevant underlying scales.’ The conclusions of this study support Krashen's (1977)‘Natural Order’for the acquisition of grammatical morphemes, Larsen‐Freeman's (1976) work on frequency as an explanation for morpheme orders, and Rosansky's (1976) work on frequency as an explanation for morpheme orders, and Rosansky's (1976a,b) criticism of cross‐sectional methodology for failing to deal with individual variation in the data, but rejects Rosansky's claim that her cross‐sectional analysis of her data does not agree with the longitudinal analysis of one of her subjects and thus invalidates her strongest argument against cross‐sectional methodology.
Roger W. Andersen (Fri,) studied this question.