Table 4 (cont.)

_________________________________________________________________________

III. Studies with seven countable grammatical morphemes:
                                                         Rank order
                            Study                        correlation
    Cheo                                                    0.712
    Juan                                                    0.368
    Jorge 9                                                 0.769
    Jorge 13                                                0.862
    Jorge 15                                                0.726
    Holdich                                                 0.726
    Alberto                                                 0.730
    Birnbaum et al., Free I                                 0.726
    Birnbaum et al., Edit. I                                0.802
    Birnbaum et al., Free II                                0.557
    Birnbaum et al., Edit. II                               0.712
    de Villiers, 1974-agrammatics combined                  0.880
    Agrammatics:
      A3                                                    0.955
      A50                                                   0.599
      A14                                                   0.962
      A43 time I                                            0.749
      A43 time II                                           0.755
      A5                                                    0.637
      A24                                                   0.768
      A6                                                    0.805

    for n = 7, significance at 0.05 level requires rho = 0.714 or larger.
               significance at 0.01 level requires rho = 0.893 or larger.
                (one-tail)
_________________________________________________________________________

a:  All individual cross-sections listed, with the exception of Uguisu 
    (Hakuta, 1974) and Holdich (Holdich, 1976), and the agrammatics, are
    from Rosansky (1976).
b:  All correlations have been corrected for ties.

A final objection that has been raised is that merely dealing with morphemes in obligatory occasions may fail to reveal at least some aspects of language acquisition, the overgeneralizations, and the transitional forms that acquirers go through. This is, I think, perfectly true, but does not detract at all from the validity of the results of the morpheme studies. The observed morpheme order is the result of the interplay of the underlying process of acquisition, and they only show the product, the surface order of acquisition. They do not directly reveal the pathway the acquirer took in arriving there. Nevertheless, there is no reason to assume the obtained order is invalid; it has been shown to be highly reliable, and occurs for the adult, in predictable

61