Parents’ Expressed Emotions and, Children’s Executive Functions

Abstract

The ways in which people talk about their family members may say as much as the words themselves. For example, emotions expressed by family members of mentally ill patients during unscripted speech samples relate to the risk of relapse and the prognosis of psychiatric disorders; however, the idea of Expressed Emotions as a construct has previously been limited to parents of children and adolescents between the ages of four and 18 years who suffer from severe emotional or behavior dysregulation. Here, we apply an Expressed Emotions coding paradigm to speech samples obtained from mothers and fathers of 104 typically developing children when children were 14 months of age. This is the first study applying the Expressed Emotions coding paradigm at this age. Parents were prompted to give thoughts, attitudes, and feelings about their child; speech samples were coded for critical comments (e.g., “She is very whiny”), emotional over-involvement (e.g., “I was so worried, I couldn’t sleep”), and quality of relationship (e.g., “We get along great”). During the same home visit, children completed three executive functions tasks that measured children’s inhibition, working memory, and cognitive flexibility. We found negative associations between father’s criticism and child’s inhibition and between father’s emotional over-involvement and child’s working memory. In contrast, we found positive associations between mother’s expressed quality of relationship and her child’s working memory. This approach to analyzing parents’ speech samples may allow for unique insights into the thoughts, attitudes, and feelings of new parents and how that might guide children’s development.

Parents—as the most frequent and proximal interaction partners for many infants and toddlers—have an outsized influence on children’s early development (Parke et al., 2006). Parents shape children’s environments and provide them with a template for how to interact with the world through a process of co-regulation and provision of experiences (e.g., Bronfenbrenner, 1986; Sameroff, 2010; Vygotsky, 1978). Among many aspects of parenting (e.g., interactions, guidance, coercive behaviors, sensitivity, responsiveness; Morawska et al., 2019), individual differences in the quality of parent-child relationships has in particular been shown to relate to early development (Fay-Stammbach et al., 2014). Relationship quality—a complex and difficult-to-measure construct that often involves observation, extended semi-structured tasks, and/or burdensome questionnaires—is related to several child characteristics, including behavior and psychosocial wellbeing (e.g., Branje et al., 2010; Thompson, 2008) and self-regulatory skills (e.g., Fay-Stammbach et al., 2014), and can act as a protective factor to buffer against potential negative outcomes of stressful events (e.g., Essler et al., 2021). In the case of self-regulatory skills specifically, children who have higher-quality relationships with their parents tend to have better self-regulatory skills (e.g., Bernier et al., 2012; Schroeder & Kelley, 2009), whereas those with lower-quality relationships are more predisposed to develop behavior problems (Stubbe et al., 1993). As such, finding ways to reliably and easily measure aspects of relationship quality might be beneficial for understanding child development.

One potential proxy for and/or correlate of parent–child relationship quality is the emotional climate of the home. Family emotional climate has been linked to various long-term outcomes in a child’s life. For instance, aspects of negative family emotional climate in early schooling are associated with lower long-term academic achievement (i.e., math and reading) of children in the 5th grade (Bodovski & Youn, 2010). Family climate has also been found to be associated with the frequency of deliberate self-harm (via a pathway of adaptive emotion regulation skills) in adolescent girls (Sim et al., 2009). Relatedly, negative family climate (low maternal warmth and low positive emotion expressiveness) has been shown to predict depression in 7th-9th grade adolescents (Luebbe & Bell, 2014).

To measure the inherently nebulous construct of “emotional climate,” some researchers (primarily in the field of adolescent psychopathology) turn to a measure of expressed emotions (EEs; Khafi et al., 2019. EEs are described as an “emotional temperature” of the home (Vaughn, 1989) and are easily measured, leveraging a commonly used data collection practice within developmental psychology wherein parents talk freely about their child (Magaña et al., 1986). Much like mind-mindedness—a construct that broadly measures parents’ recognition of their child’s mental states and is thought to reflect their conceptual understanding of and interaction with their child as an intentional being rather than one who simply has needs that must be met (e.g., Meins et al., 2003)—EEs are extracted from speech samples; however, unlike mind-mindedness, EEs measure parents’ attitudes and feelings toward their child in a way that is thought to reflect the emotional valence with which they approach interactions. Furthermore, the EE coding scheme allows for an extension beyond the traditional bounds of coding for mind-mindedness in that it captures parents’ consideration of emotional states rather than simply mental states. EEs are categorized into three general categories: Critical comments (e.g., “She is exhausting”), emotional over-involvement (EOI) (e.g., “I was so worried, I couldn’t sleep”), and quality of relationship (e.g., “We get along great”) (Magaña et al., 1986). Critical comments, EOI, and negative quality of relationship are considered negative aspects of EEs and are negatively associated with outcomes (e.g., emotionality, emotional regulation, and psychological difficulties; Hastings et al., 2008; Khafi et al., 2015; Narayan et al., 2011). One study of three-year-old children found that high levels of parental criticism are associated with emotional dysregulation in early childhood (Hastings et al., 2008). Furthermore, children (2–12 years) who have higher-quality relationships with their parents tend to have better self-regulatory skills (e.g., Bernier et al., 2012; Schroeder & Kelley, 2009), whereas those with lower-quality relationships are more predisposed to develop behavior problems (Patterson, Cohn, & Kao, 1989; Stubbe, Zahner, Goldstein, & Leckman, 1993). In contrast, positive remarks and quality of relationship are considered positive aspects of EEs and are related to positive outcomes, such as in child’s emotional development and trust judgment (Woolfolk, 2019; Tang et al., 2018). Much of the research on EEs has examined extremes of social and emotional development, specifically looking at the relations between negative aspects of EEs and child psychopathology (e.g., Mullins et al., 2021; Musser et al., 2016; Rienecke et al., 2017). More research is needed to better understand relations between parents’ EEs and a variety of child outcomes in non-clinical populations and for a wider range of ages.

As a measure of the emotional climate of the home, most extant studies involving EE have focused on relations between parents’ use of EEs and children’s emotional development and regulation; however, it is possible that EEs are also associated with more cognitive aspects of self-regulation. While it is difficult to disentangle emotional and cognitive aspects of goal-oriented self-regulation at any age (see, for example, discussions of “hot” and “cool” executive function; Carlson, 2003; Hongwanishkul et al., 2005; Zelazo & Carlson, 2012), this is particularly difficult in infancy and toddlerhood when emotion regulation is a more active process (Carlson, 2003; Sroufe, 1997). Extensive research has highlighted the long-term implications of individual differences in self-regulation broadly: Children who can effectively control, regulate, and inhibit their emotions and thoughts to prepare and plan ahead are more likely to reach their goals and report higher earnings, education, and occupational prestige in later life than children who are less proficient at these regulatory skills (Moffitt et al., 2011). We see examples of this in day-to-day life: Children (and, indeed, adults) may find themselves needing to resist the urge to subsist solely on cookies and instead eat vegetables, to inhibit an impulse to shout out to be polite to an authority figure whom they believe has treated them unfairly, or to switch between rule sets to succeed in a game. Specifically, the cognitive mechanisms that help children self-regulate and inhibit impulses are termed executive functions (EFs). EFs are interrelated processes that aid in the daily challenges of planning, switching, and inhibiting cognitions, emotions, information, and actions in pursuit of goals (e.g., Chevalier et al., 2015). EFs undergo a period of rapid development in early childhood (Chevalier et al., 2014), and predict important life outcomes (e.g., academics, health, wealth). There are already substantial individual differences in children’s EFss in toddlerhood (e.g., Devine et al., 2019; Hughes et al., 2020). Thus, it is essential to further consider origins of differences in children’s cognitive self-regulatory skills to better understand their development. Extant studies have suggested parent–child relationship quality (e.g., Fay-Stammbach et al., 2014) and aspects of the household climate (e.g., Hughes & Ensor, 2009) are related to the development of EFs; we here test whether EEs can effectively capture some of the same underlying characteristics as measures of relationship quality and household climate to suggest there is predictive validity for its use in developmental psychology research.

Current Study

Despite suggestive findings on associations between parents’ EEs and children’s social and emotional development, EEs and their relations to the development of cognitive skills remain understudied. While characteristics of speech samples have often been analyzed for associations with children’s mind-mindedness (e.g., Arnott & Meins, 2007), the few studies which have explored associations between EEs and potential correlates in typically developing children have been limited to emotion and behavior regulation (e.g., Hastings et al., 2008; Musser et al., 2016) and have been limited to children aged four to 18 years. Although the study of EEs is not novel, relations between parents’ EEs and cognitive self-regulation (i.e., EFs) in young typically developing children have yet to be examined.

Here, we apply the Expressed Emotions Coding Paradigm (Gottschalk & Gleser, 1969; Magaña et al., 1986) to mothers’ and fathers’ five-minute speech samples when their children are 14 months of age. We test three hypotheses: First, in accordance with prior studies that show negative consequences of critical comments, we anticipate a negative association between the number of critical statements made by parents and children’s EFs. Second, we predict a negative association between statements indicating EOI on children’s EFs. Third, we expect to observe a positive association between the quality of the relationship between parents and child and children’s EFs. The differences between mothers’ and fathers’ EEs, and particular facets of children’s EFs will be explored. While we do not anticipate that hypothesized relations between EEs and EFs are causal in nature, we here test whether EEs serve as an easy-to-use proxy for the emotional aspects of parent–child relationship quality and household climates. If these hypotheses are supported, it is likely due another construct’s shared variance with EEs rather than as a result of EEs directly.

Methods

Participants and Procedure

A sample of 131 middle- to high-income couples in heterosexual partnerships expecting their first child was recruited from obstetric, gynecological, and prenatal hospital visits and parenting classes in New York City as a part of a larger prospective longitudinal study of parental wellbeing, parent-child interactions, and children’s self-regulation across the transition to parenthood. To be eligible, participants had to: (1) be first-time parents, (2) expecting delivery of a healthy single baby, (3) planning to speak English as a primary language with their child, and (4) have no history of severe mental illness (e.g., psychosis) or substance misuse. When children were 14 months of age, 114 families were eligible to participate: Five families withdrew, and 12 families were ineligible to participate due to moving out of state. An additional 10 families did not respond to attempts to contact, yielding a total analytic sample of 104 families. Mothers included in the study identified primarily as white (71%); the remainder were Asian (16%), Black (2%), or identified as a different race (5%). Fathers were also primarily white (73%); the remainder were Asian (11%), Black (3%), or identified as a different race (4%). The remaining parents chose not to disclose a racial identity. Children (53 Female) were nearly 15 months old at time of data collection (M=14.71; SD=0.46; Range=13.83–16.03). Families received $15 for completing a screening questionnaire upon enrollment during pregnancy and $50 for each home visit. A total of four home visits were completed: One when the mother was in her third trimester of pregnancy, and one each when the child was 4-, 14-, and 24-months of age. For the purposes of the present investigation, only data from the 14-month data collection visit are used.

Data collection took place in families’ homes to encourage naturalistic interaction between parents and children so as to better enable both parents’ participation in the study. Both parents were present for the entire study visit (which typically lasted approximately 90 minutes) and each parent completed each of the tasks described below. Tasks were administered in a fixed order as a part of a larger battery of tasks.

Measures

Expressed Emotions

Mothers and fathers were each asked to speak for five minutes about their thoughts and feelings about their child, as well as their relationship with their child. Parents were read the following script in a quiet room without the other parent or child present, after which they were given the sheet of paper with the instructions printed:

I would like to hear your thoughts and feelings about [child’s name], in your own words, and without my interrupting with any questions or comments. When I ask you to begin, I would like you to speak for 5 minutes, telling me what kind of person [child’s name] is and how the two of you get along together. After you begin to speak, I prefer not to answer any questions until after the 5 minutes are over.

After instructions were read, parents were encouraged to ask questions, after which they were told to begin. Responses were collected and timed using a digital voice recorder.

Speech samples were transcribed and coded for EEs using a coding scheme developed by Gottschalk and Gleser (1969) and Magaña and colleagues (1986). EEs include three subscales: Critical comments (e.g., “She is very obnoxious”), EOI (e.g., “I was so worried, I had to skip work”), and quality of relationship (e.g., “Our relationship is great”). Each quality-of-relationship statement was coded as: Strong positive = +2, Weak positive = +1, Neutral = 0, Weak Negative = −1, Strong Negative = −2, where positive statements indicated a positive relationship (e.g., My child and I get along together very well), neutral statements indicated a neutral relationship (e.g., My child and I get along together okay), and negative scores indicated a negative relationship (e.g., My child and I do not get along together and she is always unhappy when I take care of her). Statements were considered “strong” versus “weak” based on the parents’ own qualification of the description. Statements were considered “strong” (positive or negative) if they included an adjective such as “very”, “always”, “really”, “extremely”, or “immensely” (e.g., she always loses her temper); statements were considered “weak” if they included no strengthening or qualifying adjective such as or if they included a qualifying adjective or adverb that signified a small amount such as “a bit”, “kind of”, or “a little” (e.g., she is a bit cheeky). Each speech sample received a score that reflected the total number of critical statements, total number of statements indicating EOI, and a total score based on the number and quality of relationship statements.

A total of 190 speech samples were available. Of the 104 participants who participated in data collection at this time point (described in Participants above), speech samples were available from 100 mothers (3.8% missing) and 90 fathers (13.5% missing). Seventy speech samples (36.8% of 190 available samples) were double-coded to ensure reliability. Inter-rater reliability was computed for number of critical comments, statements indicating EOI, and statements pertaining to the quality of relationship on a 0 to 5 scale; was very high between the two 𝜅 coders, 𝜅 = 0.92, p < .001.

EFs Assessments

Three EFs assessments were administered in the fixed order presented below with both parents and research assistants present. A detailed description and reliability, and validity of these tasks are reported elsewhere (Devine et al., 2019).

Inhibition.

Inhibition was measured using the Prohibition Task (Friedman et al., 2011). In this task, the examiner showed the child a glittery “wand” for up to 15 seconds. The wand was placed within arm’s reach of the child, and the experimenter told the child not to touch it. The trial ended either when the experimenter had waited 30 seconds, or the child touched the wand. Time was measured from when the experimenter turned away from the child until the child touched the toy or when the task was over. A longer wait before touching the wand implied greater inhibition. Scoring took place offline, and double-coding of 20% of videos revealed excellent interrater reliability, ICC=.99, p<.001.

Working Memory.

Working memory was assessed using the Three Boxes Task (Miller & Marcovitch, 2015). In this task, children saw three colored garages with correspondingly colored cars. The experimenter showed the child each car, then hid the colored cars in the correspondingly colored garages (i.e., the blue car with the blue garage). The experimenter then hid the garages and counted aloud for five seconds. The experimenter then asked the child to find a car and encouraged the child to point to a garage. In the first trial, the infant was always correct. The experimenter praised the child and removed the retrieved car and placed it out of sight. The same protocol was used on all subsequent trials. If the child searched an empty box and did not find a new car, the experimenter said: “It’s not there. Let’s have another go.” and repeated the procedure. The task was discontinued when the child found all three cars or failed to find a car for three consecutive trials. Fewer trials implied better working memory. Scoring took place offline, and double-coding of 20% of videos revealed perfect interrater reliability, =1.00. 𝜅

Cognitive Flexibility.

Cognitive flexibility was measured using the Ball Run task in which children saw a structure with three differently colored holes on the top (red, yellow, and green) and a ball which corresponded in color to one hole (red or green). The toy was placed out of reach from the child, and both the yellow and either the green or red hole were locked with the adapted sliding mechanism. The experimenter placed the red/green ball in the corresponding hole (i.e., the red/green hole). The ball rolled down and played a musical reward. The ball was then handed to the child, and the child was instructed to try it themself. If the child placed the ball correctly, they received verbal affirmation from the experimenter as well as the musical reward. If the child attempted to place the ball into the wrong hole, it would fail to drop through and was counted as a trial. Six trials were completed. If the child correctly completed four or more trials, the experimenter proceeded on to the shifting stage; If not, testing was discontinued. In the shifting stage, the experimenter used the differently colored ball (red/green) and correspondingly colored hole. One demonstration trial was given, and the same procedure as with the previous ball was used. Scoring took place offline, and double-coding of 20% of videos revealed perfect interrater reliability, 𝜅=1.00.

EFs Factor Scoring.

Details of the process of factor score estimation for these tasks are available elsewhere (Devine et al., 2019). Factor scores were simultaneously estimated for working memory and cognitive flexibility in which performance on each trial of the respective task (0 = incorrect; 1 = correct) loaded onto the latent factor. The two latent factors and a manifest variable representing latency to touch were allowed to correlate in the estimated model. As there was only a single trial of the inhibition task, we were unable to generate a latent variable and thus only a single manifest variable was included. The model fit the data well, 𝜒² (102) =155.70, p=.001, RMSEA=0.07 90% CI [0.05, 0.09], CFI=0.94. The three outcome variables were uncorrelated with one another.

Covariates

Several covariates were added to analyses. We controlled for household income as well as child sex. Additionally, to ensure EEs are not simply a proxy for parental sensitivity and that the association between EEs and EFs is not due to other characteristics of parent–child relationships, we controlled for parental sensitivity and non-intrusiveness as coded from video recordings of five-minute free play sessions with both mother and father (Ainsworth et al. 1974); intercoder reliabilities (interclass correlation, single rater, absolute agreement) were >.70 for all dyads on both sensitivity and non-intrusiveness scales. Scores representing anxiety (State-Trait Anxiety Inventory; Spielberger et al., 1970), depression (Center for Epidemiological Studies-Depression Inventory; Radloff, 1977), and general wellbeing (General Health Questionnaire; McDowell, 2006) were also included. Internal consistency was adequate for all survey measures (⍺ > .80). Finally, we controlled for parents’ EFs as measured by a delta score (reaction time for accurate responses in a mixed-side block minus reaction time for accurate responses in the same-side “Hearts” block) from a Hearts and Flowers task (Davidson et al., 2006) to ensure associations were not due to intergenerational transmission of skills (Deater-Deckard, 2014).

Results

Descriptive Statistics

Descriptive statistics and bivariate correlations are shown in Table 1. We first tested within-couple patterns of EEs. Paired-sample t-tests revealed that mothers tended to say more critical statements (M = 3.48; range = 0–14 statements) than did fathers (M = 2.04, range=0–7 statements), t(89) = 3.79, p < .001. Mothers were also deemed to express a higher relationship quality with their child than fathers, t(89) = 2.34, p=.022. No differences were found between mothers’ and fathers’ EOI scores. No effect was found of child’s gender on any of the parent’s EEs scores. Furthermore, we found a within-couple correlation for EOI, r(90) = .41, p < .001, such that if mothers said more EOI statements, so did fathers within the same couple. Descriptive statistics mentioned here and presented in greater detail in Table 1 were similar to the findings of studies of parents of older children in clinical populations (e.g., Malla et al., 1991; McCarty & Weisz, 2002; Rienecke et al. 2016; van Furth et al., 1993).

Table 1.

Bivariate correlations and descriptive statistics for all variables

		1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22
1	M—Criticisms	--
		.0
2	F—Criticism	6	--
		.0	.0
3	M—EOI	7	5	--
		.0	.1	.41
4	F—EOI	1	7	***	--
			-
		.0	.1	-	-
5	M-Relationship	4	3	.16	.04	--
		-
		.2	.0			.0
6	F—Relationship	3*	6	.13	.10	4	--
		-
		.0	.0	-	-	.0	-
7	C—Female	7	5	.05	.02	6	.12	--
			-					-
		.0	.2			.1	-	.0
8	C—Inhibition	0	2	.03	.01	4	.08	2	--
		-	-
	C—Working	.0	.1	-	-	.1		.0	-
9	Memory	4	0	.04	.06	9	.09	1	.18	--
						-		-
1	C—Cognitive	.1	.1			.0		.0	-	.0
0	Flexibility	0	0	.05	.08	4	.01	1	.08	4	--
										-
1		.0	.0			.0	-	.0		.0
1	M—Sensitivity	4	1	.13	.14	5	.07	7	.21	9	.07	--
		-								-
1	M—Non-	.0	.1			.0	-	.1		.0		.85
2	Intrusiveness	6	4	.13	.04	0	.07	0	.11	7	.06	***	--
		-						.3
1		.0	.0			.1	-	2 *	-	.1	-	.31	.2
3	F—Sensitivity	7	4	.01	.14	1	.13	*	.01	4	.12	**	5*	--
1	F—Non-	.0	.0			.0	-	.2	-	.1	-	.26	.2	.85
4	Intrusiveness	6	2	.09	.09	7	.21	6*	.04	1	.05	*	2*	***	--
		-	-			-		-		-			-		-
1	M—Executive	.2	.1	-	-	.0		.2	-	.0	-	-	.0	-	.2
5	Function	1*	0	.04	.07	9	.04	3*	.12	1	.06	.12	6	.13	1*	--
													-		-
1	D—Executive	.0	.0			.1	-	.0		.1	-		.0		.0
6	Function	2	0	.01	.05	3	.02	1	.01	5	.17	.02	7	.06	2	.05	--
						-				-					-
1		.1	.0		.30	.0	-	.1	-	.0		-	.0		.0
7	M—GHQ	6	7	.04	**	6	.14	4	.17	8	.11	.02	2	.11	3	-.07	.06	--
		-				-				-					-
1		.0	.1	.24		.1	-	.0	-	.0		-	.0	-	.1			.48**
8	M—STAI	2	3	*	.19	5	.11	4	.18	4	.07	.06	1	.14	4	-.09	.11	*	--
						-				-					-
1		.0	.0	-		.0	-	.0	-	.0			.1		.0			.74**	.57
9	M—CESD	4	6	.01	.03	5	.12	2	.14	1	.06	.02	0	.07	2	.03	.16	*	***	--
		-	-			-
2		.0	.1	-	-	.0	-	.1		.0			.2		.0					.31
0	F—GHQ	3	0	.03	.18	2	.11	6	.01	7	.06	.12	2	.03	5	.01	.08	.28*	.22	**	--
2		.1	.1	-	.25	.0	-	.2	-	.0		-	.0		.0				.23	.39	.56
1	F—STAI	1	0	.18	*	3	.10	7*	.04	2	.13	.04	4	.09	1	-.12	.04	.35*	*	**	***	--
2		.0	.0	-	-	.0		.1		.0		-	.0		.0				.33	.32	.72	.73
2	F—CESD	6	9	.09	.17	2	.05	8	.06	4	.00	.01	3	.02	4	-.10	.06	.27*	**	**	***	***	--
		10		10		10		11		10	10	10	10
	N	0	90	0	90	0	90	4	77	3	3	4	4	92	92	102	94	91	90	90	80	80	80
		3.	2.	2.0	18.	2.	2.0	1.	14.	.0		5.8	5.	5.8	5.	247.	227.		12.	31.	2.1	12.
	Mean	33	04	1	56	84	3	53	45	1	.09	0	48	9	70	05	39	2.35	48	26	4	65	31.78
		2.	1.	9.0	1.0	2.	2.4	.5	11.	.7		1.6	1.	1.6	1.	58.1	63.5		3.3	7.6	3.0	3.6
	SD	71	98	5	3	46	5	0	69	2	.58	8	86	7	75	6	7	2.76	9	7	0	9	8.45
										-	-						-
										.7	.79					119.	1.08
										9-	-					82-	-
		0-	0-	4-	2-	0-	-1-	1-	.45	1.	1.4		1-		2-	418.	377.		6-	20-	0-	6-	20-
	Range	14	7	46	48	12	12	2	-30	72	2	2–9	9	2–9	9	31	12	0–10	24	58	12	20	53

NOTE:

^***p < .001,

^**p < .01,

^*p < .05. M—Mother; F—Father; C—Child; GHQ—General Health Questionnaire; STAI—State-Trait Anxiety Inventory; CESD—Left for Epidemiological Studies–Depression Index

Path Analyses

Two separate path analyses (one for mothers and the other for fathers) were conducted to test study hypotheses while controlling for correlations among predictors to allow for correlations among the three variables representing EFs. Analyses were limited to children for whom at least one parent completed a speech sample; for participants who had parent speech data, missing data was accounted for using Full Information Maximum Likelihood estimation. Full Information Maximum Likelihood estimation takes into account the covariance matrix for all available data on the independent variables to estimate parameters and standard errors. This approach provides more accurate estimates of regression coefficients than do list wise deletion or mean replacement (Enders & Bandalos, 2001). Thirteen children were excluded from analyses for having no available direct assessment data. Results of path analyses are shown in Table 2.

Table 2.

Linear regressions predicting child executive function scores

		Inhibition				Working Memory				Cognitive Flexibility
	Mother	β	SE	t	p	β	SE	t	p	β	SE	t	p
Expresse d Emotion	Critical Comments	- 0.0 7	0.1 2	- 0.5 9	.55 9	0.0 0	0.1 0	- 0.0 1	.98 8	0.0 7	0.1 1	0.6 6	.51 0
s	EOI Score	0.0 8	0.1 3	0.5 8	.56 4	0.0 2	0.1 1	0.1 6	.87 3	0.0 3	0.1 2	0.2 9	.77 3
	Relationship Quality	0.0 5	0.1 2	0.4 6	.64 4	0.2 0	0.0 9	2.1 0	.03 6	- 0.0 4	0.1 0	- 0.4 0	.69 0
Child & Househo ld	Child Female	- 0.0 5	0.1 2	- 0.3 7	.71 5	0.0 1	0.1 0	0.0 9	.93 2	- 0.0 3	0.1 0	- 0.3 2	.75 3
Covariat es	Household Income	0.0 3	0.1 2	0.2 5	.80 0	0.3 2	0.1 7	1.8 8	.06 0	0.1 3	0.1 9	0.6 9	.48 8
Parent Behavior s	Sensitivity	0.3 0	0.2 1	1.4 4	.15 0	- 0.1 7	0.1 8	- 0.9 3	.35 3	0.0 2	0.1 9	0.0 8	.93 9
	Non-Intrusiveness	- 0.1 8	0.2 2	- 0.8 1	.41 8	0.0 6	0.1 8	0.3 2	.75 1	0.0 4	0.1 9	0.2 0	.83 9
Parent Covariat es	General Wellbeing Problems	- 0.0 8	0.2 0	- 0.4 0	.68 7	- 0.2 1	0.1 6	- 1.3 6	.17 4	0.1 2	0.1 7	0.6 8	.49 4
	Mother Anxiety	- 0.1 8	0.1 6	- 1.1 3	.26 1	- 0.0 2	0.1 4	- 0.1 2	.90 4	- 0.0 4	0.1 5	- 0.2 6	.79 7
	Mother Depression	0.0 6	0.2 1	0.2 9	.77 1	0.1 7	0.1 7	1.0 1	.31 3	- 0.0 1	0.1 8	- 0.0 8	.93 7
	Mother Executive Function	- 0.1 3	0.1 1	- 1.1 8	.23 8	- 0.0 4	0.1 0	- 0.3 7	.70 9	- 0.0 6	0.1 1	- 0.5 5	.58 3
	Father	β	SE	t	p	β	SE	t	p	β	SE	t	p
Expresse d Emotion	Critical Comments	- 0.2 8	0.1 2	- 2.3 1	.02 1	- 0.1 2	0.1 0	- 1.1 9	.23 3	0.1 2	0.1 0	1.1 7	.24 2
s	EOI Score	0.0 6	0.1 5	0.4 0	.69 0	- 0.2 1	0.1 2	- 1.7 0	.08 9	0.1 8	0.1 3	1.4 2	.15 4
	Relationship Quality	- 0.1 5	0.1 2	- 1.2 9	.19 7	0.1 2	0.1 1	1.1 6	.24 5	0.0 3	0.1 1	0.2 6	.79 6
Child & Househo Id	Child Female	- 0.0 4	0.1 2	- 0.3 1	.75 3	- 0.0 3	0.1 0	- 0.3 3	.74 5	0.0 0	0.1 0	- 0.0 3	.98 0
Covariat es	Household Income	0.1 1	0.1 6	0.7 0	.48 1	0.4 3	0.1 8	2.4 4	.01 5	0.1 2	0.2 1	0.5 7	.56 6
Parent Behavior s	Sensitivity	0.3 1	0.2 5	1.2 3	.21 8	0.2 5	0.2 0	1.2 2	.22 2	- 0.3 9	0.2 0	- 1.9 5	.05 2
	Non-intrusiveness	- 0.2 8	0.2 4	- 1.1 8	.24 0	- 0.0 8	0.2 1	- 0.3 6	.71 6	0.2 4	0.2 0	1.1 9	.23 5
Parent Covariat es	General Wellbeing Problems	- 0.1 1	0.1 9	- 0.5 8	.56 0	0.0 3	0.1 6	0.2 0	.83 9	0.1 1	0.1 6	0.6 7	.50 6
	Father Anxiety	- 0.1 9	0.1 9	- 1.0 0	.31 6	- 0.1 3	0.1 7	- 0.7 8	.43 5	0.3 5	0.1 6	2.1 9	.02 8
	Father Depression	0.2 4	0.2 3	1.0 7	.28 4	0.1 3	0.2 0	0.6 5	.51 8	- 0.2 7	0.2 0	- 1.3 4	.18 2
	Father Executive Function	- 0.0 1	0.1 4	- 0.0 4	.96 5	0.0 7	0.1 1	0.6 7	.50 6	- 0.1 8	0.1 1	- 1.6 3	.10 4

We first tested whether the number of critical comments uttered in a five-minute speech sample was related to children’s EFs. Total number of critical statements uttered by fathers was negatively related to child’s score on the Prohibition Task (β = −0.28, p = .027), but no other EFs task. For mothers, there was no significant relation between critical comments and any outcome measure. We next sought to test whether the number of utterances indicating EOI was related to children’s EFs. We again found a relation between fathers’ utterances and children’s performance on the working memory task in the expected direction (β = −0.21, p = .088), but not with other EFs tasks. Again, we did not find a relation between mothers’ utterances and children’s EFs.

Finally, we tested whether overall parent-child relationship quality as indicated by EEs was associated with children’s EFs. The quality of the relationship between mother and child was positively related to child’s score on the Three Boxes Task (β = .20, p = .038), but no other EFs task. Fathers-child relationship quality was unassociated with child EFs.

Discussion

This is the first study to our knowledge to examine parents’ EEs when children are toddlers and to suggest EEs may relate to important cognitive outcomes of children. We found substantial variability in variables extracted from coding of EEs, suggesting this may be a useful measure that requires little time or cost investment from future researchers. In this study, we hypothesized negative relations between parents’ negatively valanced EEs (statements that convey criticism and EOI) and children’s EFs, as well as positive relations parents’ positively valanced EEs (those that convey a positive relationship quality) and children’s EF. While this is ultimately a small study and we choose not to draw firm conclusions about practical or statistical significance as to any causal relations between EEs and EFs, results of this exploratory study lent partial support for our hypotheses: Fathers’ criticism scores were negatively related to their child’s inhibition skills, whereas mothers’ relationship quality scores were positively related to their child’s working memory. Additionally, some evidence for a negative relation between fathers’ EOI and their child’s working memory was found.

Expressed Emotions in Toddlerhood

Extant research using the expressed emotions framework has been dedicated largely to older populations (e.g., Rienecke et al. 2017); indeed, the youngest populations for whom EEs have been examined included preschool-aged children (e.g., Hastings et al., 2008). Here, we establish the expressed emotions coding paradigm as a promising research tool for early child development. It is easy and quick to administer, and coders can be trained quickly on the coding scheme (i.e., coders in this study were trained within a week). We found substantial variability in parents’ EEs even when children are 14 months of age, as well as interesting within-dyad differences: On average, mothers tended to be more critical than fathers, and some fathers indicated having a weak negative relationship with their child, but they did not differ in their EOI. However, it is important that findings be replicated in the future with a broader and more generalizable sample.

Unlike prior research, which demonstrated relations between negative aspects of parents’ EEs (i.e., criticisms and EOI) and children’s outcomes (e.g., Musser et al., 2016; Rienecke et al., 2017), the current study did not indicate a relation between mothers’ EEs and children’s self-regulatory skills in a typically developing toddler population. However, results indicate a negative relation between fathers’ EEs and child’s score on the inhibition and working memory tasks. Additionally, we found significant relations between a positive measure of mothers’ EEs (relationship quality) and child outcomes, but no parallel for fathers.

We did not find that a single facet of EEs related to all of children’s EFs, which is not unsurprising. Prior studies using this dataset found differential associations between children’s environment and components of EFs (e.g., Devine et al., 2019). This specificity might also be due to several factors. First, mothers may have been more likely to spend more time with the child as tends to be the norm in the US. If this is the case, in fathers’ relatively limited time with their children, negative comments might be more salient. Second, there is reason to believe parents’ EEs might be differentially related to child outcomes at different developmental periods in the child’s life. At younger ages, children spend more time with their parents overall, whereas as children age, there is an increasing number of significant others in their lives, including other children (e.g., peers, teammates) and other adults (e.g., teachers, coaches, parents of friends) (e.g., Berndt, 1979; Parke et al., 2006). Thus, there may be greater diffusion of EEs over the day.

Limitations and Conclusions

Though this study adds to both the field’s repertoire of measurement of very young normally-developing children and understanding of predictors of very early EFs, there are several limitations. Importantly, this study does not permit causal interpretation. Parents may be responsive to their children’s EFs and are more critical or positive in response to children’s ability to regulate their emotions and behaviors. Future research should consider bidirectional and longitudinal associations and stability in EEs and EFs, as well as geographic and cultural differences. Additionally, it is important to note that the researcher’s presence might have affected parents’ speech samples in the current study, as data collectors were in the room as parents were speaking. Parents might not have felt comfortable speaking honestly in the presence of others. Another note-worthy limitation is that parent’s affect or tone is not included in the EE coding due to EE being coded from transcribed speech samples. This potentially limits the validity of the EE measure since the tone and affect of the parent’s statement can change the meaning and intent of the parent’s words. For example, a parent saying “my child is very energetic” can be seen as a positive statement if it was said in an enthusiastic tone or as a negative statement if it was said in an upset or annoyed tone. However, all utterances were coded within the broader context of the overall speech sample, meaning some nuance could be interpreted from context clues of surrounding utterances. As EEs become a more widely used tool, we look forward to investigations that can shed light on the nuances of not just what parents say, but how they say it.

Due to the importance that researchers have found that EFs and EEs play in our lives, finding a relationship between these two phenomena could aid in our understanding of EFs, to parenting and to a child’s future. EEs are an easy tool to deploy for the coding of speech data and further research should consider the interplay between EE and other aspects of parents’ speech, including mind-mindedness. Conducting more research on the impacts of EEs—and/or on the associations between EEs and other aspects of parenting—on a child’s future could have major implications for parenting practices and a child’s life.

Highlights.

• Expressed emotions can be reliably, easily, and validly coded from speech samples

• Parents’ expressed emotions relate to children’s executive functions

• Children of more critical fathers have worse inhibition

• Children of fathers who are emotionally overinvolved have worse working memory

• Children of mothers who express high relationship quality have better working memory