Background: Twins tend to lag behind singletons in their language development, but the causes were unknown.
Method: Ninety-six twin pairs from the Avon Longitudinal Study of Parents and Children (ALSPAC), for whom birth was after at least 33 weeks of gestation, were compared with 98 pairs of singletons, no more than 30 months apart in age. Parental qualities and family interaction were assessed through standardised questionnaires and interviews and both structured and unstructured observations in the home at 20 months and 36 months. The possible causal role of postnatal family influences was assessed through five criteria: i) the feature had to differ between twins and singletons; ii) individual differences in that feature had to relate to individual differences in language level within the sample of singletons and of twins; iii) the feature as measured at 20 months had to predict language as assessed at 36 months; iv) that had to apply after controlling for language level at 20 months; and v) introduction of the predictive feature into an overall model had to obliterate the twin–singleton difference in language level.
Results: Patterns of parent–child interaction and communication met these five criteria. The maternal factors all concerned aspects of interaction that were broadly concerned with communication: encouraging the child to speak, providing elaborating comments, engaging in reading to the child and talking about the story and its illustrations. The HOME inventory findings provided similar findings with respect to responsiveness, involvement and level of experiences involved. Family features that might have been influential, but which were not, included parental depression, breastfeeding, family size, and style of sibling interaction.
Conclusions: Patterns of parent–child interaction and communication within the normal range have environmentally mediated effects on language and account for twin–singleton differences in language developmently. The results indicate the value of a natural experiment in testing competing causal hypotheses, and show the role of environmental factors as influences on language variations within the normal range, for both twins and singletons.