Table 3. Counterexamples to ordering relations among 
        grammatical morphemes
____________________________________________________________________________

Total instances consistent with relations given in Table 3 = 881
Total instances inconsistent with relations given in Table 3 = 98 (10%)

Analysis of counterexamples:
(a) within 10 percentage points                                   = 49 (50%)
(b) within 20 percentage points                                   =  9
(c) where percentages not given: one rank difference              = 16
(d) where percentages not given: two ranks difference             =  9
(e) studies producing "true" counterexamples (not (a)-(d) above):
    1. Uguisu (Hakuta, 1974)                                      =  5
    2. Larsen-Freeman (Imitation I)                               =  2
    3. Jorge 11 (Rosansky, 1976)                                  =  1
    4. Dolores (Rosansky, 1976)                                   =  2
    5. Andersen, 1976a                                            =  1
    6. agrammatic individual subjects (de Villiers, 1974)b        =  4
____________________________________________________________________________
a: Number of obligatory occasions not known.
b: Obligatory occasions known to be more than ten but may be less 
   than twenty.

consistently positive but often statistically insignificant correlation one sees between L1 and L2 scores when rank order correlations are used.

Figure 1 gives a proposed "natural order" for second language and for Broca's aphasia, and was constructed from the relations given in Table 2. It is extremely interesting to note that all "Monitor-free"

Fig. 1. Proposed "natural order" for second language acquisition and agrammatics

studies using adult subjects (except Imitation I) show a healthy positive correlation with this order¹ (Table 4).

The data presented here strongly confirm the reality of a "natural order", a reliably occurring order in longitudinal and cross-sectional,

59