This study applies linguistic, psychological, and sociological approaches. The main topic discussed in this study is the construct underlying parents' behavior when they are rearing their children, particularly, their construct of thought or ideas about children's acquisition of language.The sample subjects in this study, selected on the basis of the purposive random sampling technique, are both mothers whose children are 4 to 5 years of age and the children themselves. The subjects come from two suburban districts and two urban districts in Semarang Municipality, Central Java. This study uses several kinds of variables, those are the independent variable (parents' ideas); intervening variables (rearing strategy, home dimensions); socio-demographical variables (education, domicile), control variable (vocational status); and the dependent variable (preschool children's ability in speaking Indonesian).This study uses cross-sectional comparison design which is non-experimental between-groups in nature, fashioned with interviews, observations, and measurements as the methods for collecting the data. The statistical analysis used to compute the data are one-way ANOVA, multiple regression discriminant, correlation matrix and crosstab/chisquare test.From the analysis, it is found that there are three groups of parents with different ideas concerning with children's acquisition of Indonesian. These are modern, traditional, and ambivalent ideas. It is also found that parents' ideas have only indirect influence on the ability of speaking Indonesian among preschool-children. The one which has direct influence is intervening variable, particularly home dimensions variable.