Mental State Terms in General: Concurrent Relations

Table 7.3 shows the studies to be reviewed in this section. As in earlier chapters, I have indicated the origin of the sample only when it is drawn from a population from other than one of the Western countries (United States, Canada, Great Britain, Germany) that have contributed most of the data on theory of mind. Studies that include special populations (e.g., deaf children, children with autism) have been deferred for coverage in a later section.

The most general question addressed by this research is whether variations in how parents talk about mental states relate to variations in children’s theory-of- mind development. This question is easy to answer: Virtually every study summarized in Table 7.3—and indeed, every study to be covered in the next section on longitudinal relations—has reported a relation between parental talk and theory of mind. The only exception is the study by Randell and Peterson (2009), and, as the authors acknowledge, this study suffered from restriction of range on both the parent and the child measures. The overall positive conclusion does not mean, of course, that every possible relation emerges as significant; often, some predictors turn out not to be predictive and some outcomes have no identifiable antecedents. It is also true that no one knows how many unpublished studies with null results may reside in file drawers somewhere. Nevertheless, the general conclusion is not in doubt: Parental talk does relate to theory of mind.

How strongly does it relate? The answer, of course, varies across studies. The relation, however, is often fairly substantial—for example, correlations in the .40 to .50 range. It is worth noting in this respect that the average sample size for the studies reviewed in this section is 52, and for several the value is in the 20s or 30s. It is understandable, of course, that the sample sizes are not larger, given the labor-intensive nature of such research. But the modest sample sizes ensure that small effects are not being inflated to statistical significance simply because of the sample size; significant effects are generally fairly large effects.

Because this research is correlational, an important question is whether the relations hold up when potentially important third factors are controlled. I have already indicated that the answer is positive. Not in every instance, of course— occasionally a significant relation falls to nonsignificance, and the magnitude of

Table 7.3 Studies Relating Parental Talk to General Aspects of Theory of Mind: Concurrent Relations

Table 7.3 Continued

(continued)

Table 7.3 Continued

effects may shrink some when controls are imposed. It is true as well that no study can control for every potentially important third factor. It is worth noting, in fact, that no study to date has controlled for shared genes as an explanation for parent- child relations, given the absence of adoption studies in this literature. But the effects are robust in the face of controls for some of the most obvious possible confounds—in particular, child language ability and parent educational level.

Fundamental though such information is, the existence and the magnitude of an effect is always just the starting point in the examination of any topic. What more do we need to know once we know that parental talk can affect children’s development? The answer, of course, is quite a few things. Here I will discuss five further questions that research to date has raised but not yet fully answered. Because the longitudinal studies also speak to these issues, I simply list the questions here and then return to them in the next section.

One question was touched on earlier, and that is possible effects of context. It has long been clear that aspects of parents’ speech to their children vary across different contexts (Haden & Fivush, 1996; Hoff-Ginsberg, 1991). Does the same hold true for talk about mental states? Although this is most immediately a question about methods of study, the methodological contrasts can also tell us something about the real-life contexts in which parental talk occurs.

A second question concerns the interpersonal context. Does it matter who it is who is producing the talk? Here, of course, the main question is whether talk from fathers has the same effect as does talk from mothers.

The remaining questions concern aspects of the talk itself. Again, one question was touched on earlier, and it concerns the target for the mental state reference.

Does it matter whether the talk is about the child’s own mental states, or do children learn equally well or perhaps even better from talk directed to the mental states of others? We can also ask about the specificity of links between talk and theory-of-mind learning. Is any sort of mental state talk helpful for the developments in question, or do children need to hear talk about desires to learn about desire and talk about beliefs to learn about belief? Finally, we can ask about the quality of the speech input. Beyond just the number or variety of mental state terms, what is it about some parental talk that makes it especially effective in nurturing theory-of-mind understanding?

Before leaving this section, I will note a few points of interest from several of the studies summarized in Table 7.1. First, one of the measurement options listed is the presentation of a standardized questionnaire to measure parents’ tendency to engage in mental state talk. The only entries to date in this category are a pair of studies reported by Peterson and Slaughter (2003). These investigators devised an instrument called the Maternal Mental State Input Inventory on which mothers chose among various response options when presented with a series of parenting vignettes. In each case one of the options emphasized discussion of mental states, and mothers’ preference for this option turned out to be associated with good theory-of-mind performance by their children in both studies. Although this approach clearly does not provide the direct measure of talk found with the other measurement options, it has some compensating strengths—in particular, a wider range of relevant situations than is likely to occur in most observational samples. The Maternal Mental State Input Inventory or similar measures could certainly serve as a valuable complement to the other ways of measuring parental talk; to date, however, its initial use remains its only use.

The study by Hutchins and colleagues (Hutchins, Bond, Silliman, & Bryant, 2009) prefigures the work to be discussed in Chapter 8. These researchers measured mental state talk in both mother and child and showed, as have numerous other studies, that the two were related. The main goal of their research, however, was to determine whether mothers’ use of mental state language was a reflection of their general belief system concerning knowledge and ways to acquire knowledge, a belief system that was measured through an instrument called the Ways of Knowing Interview. I consider this aspect of the work in Chapter 8.

Finally, Adrian et al. (2005) provide the only storybook study to supplement the laboratory assessment with a measure of how often the mothers in the study read storybooks to their children at home. Their data confirmed that storybook reading is indeed a familiar experience for many children; 32% of the mothers reported that they read daily to their child, and another 47% read between one and six times per week. Frequency of reading at home related positively to children’s theory-of-mind understanding, a finding that has emerged in other research as well (Mar et al., 2010). The mother’s use of mental state terms while reading the storybook in the laboratory also related to theory-of-mind performance.

Interestingly, however, there was no relation between the home and laboratory measures; that is, mothers who were frequent readers were no more likely than mothers in general to use mental state terms.

I will add that content analyses of storybooks for children make clear that such books can indeed be rich sources of mental state information. One analysis found that mental state words occurred on average every three sentences in books intended for 3- to 6-year-olds (Dyer, Shatz, & Wellman, 2000). Although there are some intriguing variations across countries and languages, mental state language has been found to be a common feature of storybooks in every country studied, which to date include the United States, China, Japan, Italy, Romania, and Turkey (Dyer-Seymour, Shatz, Wellman, & Saito, 2004; Shatz, Dyer, Marchetti, & Massaro, 2006; Tsai, Louie, Chen, & Uchida, 2007; Vander Wege et al., 2014). Perhaps not surprisingly, the mental content of storybooks fits children’s interests. Children prefer stories that focus on people over stories that focus on objects, and they prefer stories with some mental content over those that are described solely in terms of action (Barnes & Bloom, 2014).