The basic purpose of this study was to establish association values for nonsense words to be used in learning experiments with children from culturally-different backgrounds. Responses to 50 stimuli (44 nonsense and six real words) individually administered to 164 children from kindergarten, day care, and nursery school settings, representing two levels each sex, SES, and race (Black and Caucasian) and three age groups (4-, 5-, and 6-year-olds) were recorded. Association values for each word were calculated, providing a hierarchy with significant differences between the 10 high and 10 low terms, but little dependable difference between adjacent items. No significant difference in association value could be attributed to sex, SES, or race, but age-related differences were found. Data were also analyzed in terms of semantic, syntactic and phonological components. In the syntactic and phonological analyses, major differences were also age-related. Four-year-olds failed to respond significantly more frequently than 6-year-olds, and produced the lowest number of both verbs and abstract nouns. While advantaged children produced a significantly larger number of abstract nouns than disadvantaged children, there was no support for the Bernstein hypothesis that disadvantaged children demonstrate restricted use of adjectives and adverbs. (Author/WY)