Abstract
Reber and Lewis (1977) exposed subjects to a subset of letter strings generated from a synthetic grammar, then asked them to reorder scrambled letters to generate new grammatical strings. The distribution of the frequency of the bigrams composing their solutions correlated better with the frequency of the bigrams composing the full set of strings generated by the grammar than with the frequency of the bigrams composing the subset of strings displayed in the study phase. In his recent overview on implicit learning, Reber (1989) develops this experimental result into one of the main supports for his contention that studying grammatical letter strings gives access to the abstract structure of the grammar.
However, this result can be accounted for by a set of biases inherent to the Reber and Lewis procedure. In the present experiment, a group of subjects learned from a list of the bigrams making up the study strings, a condition which precludes the abstraction of any high-level rules. The pattern of correlations outlined above also emerged in this condition, thus lending support to our re-interpretation.
