This study was designed to investigate children's ability to use semantic syntax in connection with two types of meanings of polysemantic words. One hundred fourth graders ranked the familiarity of multiple definitions of 60 words given in a semantic survey. Those 34 words found to have common meanings or obscure meanings were used to construct a Words in Context Test. A second version of the test was developed using a low-association value trigram instead of the target real word. The two versions were administered to 64 subjects from three fourth-grade classes randomly assigned to experimental groups. There were consistently more errors on obscure meanings than on common meanings. The mean difference was greater with real words (-4.81) than with trigrams (-.75). Fewer errors were made with real words on common meanings, but more errors were made with real words on obscure meanings. Children's problems with obscure meanings represented by the frequent use of multiple-meaning words in sentences seem to be due to interference by better-known meanings of familiar words diverting them from the semantic syntax. (The appendices contain the semantic survey, the Words in Context Tests, and several Statistical Tables.) (Author/DD)