Table 1. Studies analyzing grammatical morphemes in obligatory
        occasions
____________________________________________________________________________

I.   Child L1 acquisition
     A.    Individual cross-sectional: none.
     B.    Individual longitudinal: Brown, 1973 (Adam, Eve, Sarah); 
           Curtiss, Fromkin, and Krashen, 1978 (Genie).
     C.    Grouped cross-sectional: de Villiers and de Villiers, 1973; 
           Kessler, 1975.
     D.    Grouped longitudinal: Brown, 1973

II.  Child L2 acquisition
     A.    Individual cross-sectional: Kessler and Idar, 1977 (Than, 
           three cross-sections); Rosansky, 1976 (Jorge, ten cross-sections, 
           Marta, Cheo, Juan).
     B.    Individual longitudinal: Rosansky, 1976 (Jorge); Hakuta, 1974 
           (Uguisu); Kessler and Idar, 1977 (Than).
     C.    Grouped cross-sectional: Dulay and Burt, 1973; Dulay and Burt, 
           1974a.

III. Adult L2 acquisition
     A.    Individual cross-sectional: Birnbaum, 1976 (Hector); 
           Strei, 1976 (Gerardo); Holdich, 1976 (Holdich); 
           Rosansky, 1976 (Alberto, Dolores).
     B.    Individual longitudinal: none
     C.    Grouped cross-sectional: Bailey, Madden, and Krashen, 1974; 
           Larsen-Freeman, 1975; Krashen, Houck, Giunchi, Bode, Birnbaum, 
           and Strei, 1977; Andersen 1976; Birnbaum, Butler, and Krashen, 
           1977 (also in Krashen, Butler, Birnbaum, and Robertson, 1978).

IV.  Agrammatics
     A.    Individual cross-sectional: de Villiers, 1974 (A3, 50, 14, 5, 
           24, 6, 43 (two cross-sections)).
     B.    Individual longitudinal: none.
     C.    Grouped cross-sectional: de Villiers, 1974.
____________________________________________________________________________

variation; just one additional correct response, for example, could change a subject's score on a morpheme by 10 percentage points. Nevertheless, I found an amazing amount of uniformity across all studies that used Monitor-free instruments, that is, in all studies where language was used for communication. (Rosansky (1976), it should be noted, allowed items to be analyzed that appeared in less than ten obligatory occasions for an individual cross-section. This probably accounts for our different conclusions, which are based on some of the same data.)

Table 2 lists the particular relations discovered and describes the criteria for selection. The counterexamples to these generalizations (Table 3) are rarely "serious". As Table 3 indicates, they often fall within 10 percentage points, are items represented by less than twenty obligatory occasions, or are within one or two ranks where percentages

57