Me or my: two different patterns of pronoun case errors

This paper investigates why some young children prefer to replace I with me (me-children), whereas others prefer to replace I with my (my-children). The data come from 12 children, observed monthly, from 1;0 to 3;0. It was found that the percentage of errors in which me replaced I (the me-error rate) was positively correlated with the correct production of me as an objective pronoun (the me-total). The me for I and my for I errors were antagonistic, with one of the patterns almost always dominating over the other, resulting in a clear individual difference between me-children and my-children. It was also found that the me-total during the period in which my for I replacements first emerged prefigured whether a child would become a me-child or a my-child.