| Previous research with preterm infants has found great variability in the linguistic outcomes for this population and that both environment and biological variables should be included in statistical analyses in order to adequately predict language development. The present investigation sought to expand upon the existing body of research with preterm infants by utilizing structural equation modeling to consider which of the two hypothesized models best fit the data in a longitudinal study. The first model, referred to as the single developmental domain model, had the environmental and biological variables (e.g., including (a) the HOME, (b) hearing thresholds, (c) the cumulative frequency of otitis media, (d) a factor score that represented early medical risks, and (e) a factor score that represented ongoing medical problems) directly influencing language measures. However in the second model, the environmental and biological variables were hypothesized to influence cognitive and motor variables, which in turn would predict language outcomes. As several domains of development were included, this model was referred to as the multiple developmental domain model. The single and multiple developmental domain models were evaluated at 12, 24, and 30 months with an at-risk preterm sample (;Future research should continue to investigate how the home environment contributes to language acquisition for preterm infants, specifically aspects of parent-child interaction such as joint attention or maternal responsiveness. In addition, as ongoing medical risks have been neglected in previous studies, the relationship between ongoing medical problems and language development deserves further attention.;In general across all time periods and for both the high-risk subjects and the entire sample, the single developmental domain model best fit the data. In the single models, the percent of variance that was accounted for by the exogenous variables ranged from 16% to 46% for receptive language and from 46% to 80% for expressive language. In certain models the cognitive and motor variables were significant predictors of receptive and/or expressive language, however, overall, they appeared not to function as mediator variables. For the high-risk subjects at 12 months, the HOME and the ongoing medical factor were significant direct predictors of receptive language and indirect predictors of expressive language. At 24 months, both the HOME and the ongoing medical factor continued to be significant predictors of language measures, and the hearing variable also emerged as a significant direct predictor of receptive language and an indirect predictor of expressive language at 24 months. In addition at 24 months, both the cognitive and motor variables had significant path coefficients to language measures, but the cognitive and motor variables were functioning as predictor, not as mediating variables. At 30 months, the HOME and the early medical factor were directly predictive of expressive language, although the path coefficient was in an unanticipated direction for the early medical factor. Thus across most of the developmental periods, the HOME and the ongoing medical factor were important predictors of language development. |
