Aside from the generall "well, duh" factor (see e.g. Groxx's comment), the usual complaint applies here: there are some really hard-to-control-for factors. For instance, I bet "talking seriously with children" is very strongly correlated with parents' intelligence, which in turn is strongly correlated with children's intelligence, which in turn is strongly correlated with language proficiency. (For any reasonable definition of "intelligence", slippery though that term is.)
The relevant publication here seems to be the one available in pieces at http://dare.uva.nl/record/334829 and, indeed, it doesn't look as if the researcher has tried to control for parents' intelligence, education, etc. For that matter, the sample size is only 25 so it would be difficult to do so and still get any results worth reporting.
The relevant publication here seems to be the one available in pieces at http://dare.uva.nl/record/334829 and, indeed, it doesn't look as if the researcher has tried to control for parents' intelligence, education, etc. For that matter, the sample size is only 25 so it would be difficult to do so and still get any results worth reporting.
I'm underwhelmed by this.