Notice that the experimenter’s prompt, “What do you call someone who eats rats/mice?,”
contains a plural form of the noun, and can be expected to bias the child very strongly towards
using a plural in his or her response, whenever this is possible.
The experiment used the same five irregular nouns mentioned above (mouse, man, tooth,
foot, goose), five matched regular nouns (rat, baby, bead, hand, duck), four pluralia tantum
(clothes, pants, (sun)glasses, scissors) and another four matched regular nouns (toy, shirt, shoe,
knife). Pluralia tantum, like irregular plurals, are acceptable as non-heads inside compounds.
The children at all ages made a clear distinction between the regular plurals on the one
hand, and the irregular plurals and pluralia tantum on the other. Out of 297 opportunities (33
children × 9 regular items), the children produced in total only six regular plurals inside
compounds (2.02 percent of the opportunities). Interestingly, five of these six were produced by
5-year-olds, 1 was produced by a 4-year-old, and none were produced by the 3-year-olds.
Gordon speculates that the older children’s metalinguistic awareness may have played a role.
In contrast, out of 165 opportunities (33 children × 5 irregular items), the children
produced a total of 36 irregular plurals inside compounds (21.8 percent of the opportunities).
With pluralia tantum, out of 132 opportunities (33 children × 4 pluralia tantum items) the
children produced a total of 68 pluralia tantum inside compounds (51.5 percent of the
opportunities). In most other responses to pluralia tantum items, and especially in responses to
glasses and scissors, the childen used a morphologically singular form, as in scissor-eater, which
is also possible for adult speakers. Thus, the findings provide powerful evidence that very young
children are already using the same grammar of synthetic compounding as adults.
The contrasts between regulars versus irregulars and pluralia tantum were found to be
statistically reliable using chi-square tests, where each subject was classified according to his or