Talking and reading science: Longitudinal data on sex differences in mother–child conversations in low-income families.

Overall, mothers of sons talked more about science process talk and social pretend play than mothers of daughters. Mothers also used more science process talk when their children were older than when they were younger. Maternal talk predicted children's later academic performance only at 9 years, where science process talk predicted later understanding of science text.

4.1. Sex differences in maternal talk to girls and boys

The low-income mothers in this sample talked more to their sons than their daughters. This pattern is the reverse of what is typically found in middle-income samples. However, the masculine-stereotyped nature of the magnet toys may have contributed to mothers' greater talkativeness with sons than with daughters. Similar to middle-class mothers, the urban, low-income mothers in the present sample tended to use more high-level talk (i.e., social pretend play and science process) with sons than with daughters, even after controlling for whether children initiated the first instance of science process talk. This finding was surprising, but others, e.g., McGillicuddy-De Lisi (1988), have found that the particular setting influenced the impact of gender on parental behavior; for example, boys received more high-level demands than did girls during a paper-folding task but not during a story task. Perhaps mothers viewed magnets as more interesting or appropriate for boys than girls. Without maternal interviews, however, we cannot know how the mothers interpreted the play activity. An examination of these types of talk in varying contexts is needed to tease apart the effects of context and gender.

An alternative explanation for mothers' greater talk to boys is that mothers responded to children's prior interests, preferences, or proclivities. For example, research on science museum exhibit preferences and in science fair participation suggests that boys are more interested in physical science than are girls. Additionally, the boys may have had higher scientific ability than girls. By middle school, boys slightly outperform girls on tests of physical science. This alternative cannot be tested without measures assessing children's early science proclivities and interests. In any case, our findings suggest that mothers' behaviors are related to children's learning outcomes, and that access to facilitative mother–child interactions may differ for boys and girls.

4.2. Relations between maternal talk and children's reading comprehension

The only effect of type of maternal speech was that process talk at 9 years predicts later science text comprehension. Although MLU predicted reading comprehension for science text, maternal talk at 9 years old predicted children's reading comprehension for a science text beyond MLU, child sex, children's own science process talk when they were 5 and 9 years old, and maternal education. In addition, although children's science process speech was selected as a control because their interest might have predicted their reading comprehension, children's own science process speech at ages 5 and 9 did not predict later reading comprehension scores. The particular text chosen required children to be familiar with scientific vocabulary and tested their ability to read expository writing. One question required inferential reasoning of precisely the type modeled by maternal science process talk. For this reason, it was expected that mothers' science process talk and children's reading comprehension for the science text would be related. Comprehension of scientific discourse is typically more dependent on specialized background knowledge than is comprehension of narrative text. Being able to read science texts is crucial for academic success in the middle and secondary grades. That mothers' speech predicted children's reading comprehension for the science text 2 years later gives support to the importance of parent–child interactions in fostering children's cognitive development, and to the notion that enrichment of children's early science-learning environments may contribute to their academic success.

Given the correlational nature of our results, we cannot rule out the possibility that more advanced children may elicit more science process talk from their mothers. However, we did control for advanced speech and for the quantity of child science process talk. However, children clearly contribute to their own socialization. Thus, children may have influenced the type of talk that mothers addressed to them. For example, whether children asked questions that elicited science process or even social pretend play talk was not addressed in the present study. Perhaps other child factors, such as science interest, may influence how mothers talk to children about magnets. Future research is needed to address how children's interests and their speech influence how mothers speak to them.

The finding of a relation between mothers' talk when children were 9 and later text comprehension outcomes, but no relation with mothers' talk at age 5, is not surprising. Children may need to be more cognitively advanced themselves within the domain of science to be able to benefit from parental input. Samarapungavan (1992) found that within the domains of astronomy and chemistry, third graders were more advanced than first graders in understanding the empirical and logical consistency of scientific theories. Additionally, children in this sample would not have been exposed to formal science instruction until they reached elementary school age, and the effects of parental input might be stronger when children have a more developed scientific framework. In addition, it is more likely to find a stronger prediction over the 2-year time span between reading outcomes at age 11 and maternal talk at 9 years old than over the 4-year time span from age 5.

4.3. Changes across time in mothers' and children's speech

Mothers used significantly more science process and social pretend play talk with children when they were older than when they were younger. Past studies examining parents' science talk to children have not found differences based on age. However, these studies have relied on cross-sectional designs, which may have been less powerful in detecting subtle changes in parent–child conversations. During middle childhood, children's reasoning skills and factual knowledge increase, and mothers may have responded to these changes by engaging children in more cognitively demanding speech.

4.4. Sex differences?

Boys heard more of the types of talk that are related to increased skills both in science learning and in literacy. Perhaps, as these children are followed beyond the sixth grade, the boys will begin to outperform girls in science literacy, as has been found in many studies. Moreover, from being involved in conversations with more science process talk, boys may become more interested in and self-confident about science than girls. In this sample, there was already a nonsignificant trend favoring the boys over the girls in the science reading comprehension measure, but not the narrative measure. Science and math self-confidence and interest are predictors of continued math and science course selection.