Language ability, executive functioning and behaviour in school-age children.

BACKGROUND: Many children with language impairment present with deficits in other areas, including executive functioning (EF), attention and behaviour. Similarly, many children receiving services for attention or behaviour problems have deficits in language ability. AIMS: To evaluate the relations among EF, language ability and behaviour problems in a sample of school-age children with a wide range of language and behaviour profiles. The following research questions were addressed: Does performance on EF tasks predict language ability? Do language ability and EF predict problems with attention, internalizing and/or externalizing? METHODS & PROCEDURES: EF was defined as referring to the separable, yet related, processes of shifting, updating working memory and inhibition as specified in the latent variable model of EF. Children aged 8-11 years recruited from an urban school district completed standardized language and cognitive assessments and a computerized task assessing EF. Their parents completed standardized questionnaires assessing the children's EF and problem behaviours. Regression analyses were conducted. OUTCOMES &
RESULTS: Regression analyses revealed that EF did not contribute to language ability beyond the variance accounted for by nonverbal reasoning. Language ability contributed to attention problems when entered as a single predictor, but was no longer significant when the EF measures were added to the model. Language ability did not significantly contribute to internalizing or externalizing behaviour problems. Parent-reported inhibition was a robust predictor of attention, internalizing and externalizing behaviour problems. CONCLUSIONS & IMPLICATIONS: In this sample of school-age children, language ability was related to attention problems, but not to internalizing or externalizing. Children with behaviour problems may have particular difficulty with inhibition.

What this paper adds?

What is already known on the subject?

Some children with language impairment have deficits in EF. Many children with language impairment present with attention, internalizing or externalizing behavior problems, and many children receiving services for behavior problems present with significant deficits in language ability.

What this paper adds?

In a sample of school-age children with a wide range of language and behavior profiles, EF did not predict unique variance in language ability beyond that accounted for by nonverbal cognition. Inhibition contributed to attention, internalizing and externalizing behavior problems.

Introduction

Children with language impairment (LI) often present with co-morbid deficits, including deficits in executive functioning (EF) (Dibbets et al. 2006, Henry et al. 2012, Im-Bolter et al. 2006). Even after LI seemingly has resolved, subtle deficits in language, cognition and/or behaviour/conduct may persist (Brownlie et al. 2004, Rescorla 2009). Miyake et al.'s (2000) influential individual differences study of EF in young adults revealed that shifting attention, updating working memory and inhibition of prepotent responses are separate functions sharing an underlying commonality.

Updating working memory is often termed working memory in the literature, and will be called working memory here. Investigations into the development of EF in children have found evidence of these same latent variables (Lehto et al. 2003).

However, van der Sluis et al. (2007) revealed only two latent variables: shifting and working memory; conversely, St. Clair-Thompson and Gathercole (2006) identified only working memory and inhibition.

Brownlie et al. (2004) and Villamarette-Pittman et al. (2002) suggested that executive dysfunction could be the underlying connection between LI and delinquency, although Ford et al. (2007) found that receptive language ability was a stronger predictor of behaviour problems than was EF. Children with LI and children with ADHD perform similarly on some measures of language, including mean performance on sentence recall tasks, although they differ in their use of tense marking in sentence recall (Redmond 2005), and lexical diversity and mean length of utterance in conversation (Redmond 2004). Children and adolescents with LI have been shown to perform less accurately than typically developing peers on EF tasks requiring working memory (Henry et al. 2012, Im-Bolter et al. 2006) or response inhibition (Henry et al. 2012). Im-Bolter et al. (2006) found that inhibition contributed to language ability via its relation with attention rather than via a direct pathway. Some behavioural data have indicated no significant differences between the children with LI and their typically developing peers on tasks measuring shifting (Dibbets et al. 2006, Henry et al. 2012, Im-Bolter et al. 2006); however, fMRI data suggest shifting may be more demanding for children with LI than for typically developing children (Dibbets et al. 2006).

Prior research has found relations between behaviour problems and EF. Deficits in inhibition have been linked to ADHD and externalizing behaviour problems in young children and adolescents (Ford et al. 2007, Friedman et al. 2007, Riccio et al. 2011, Willcutt et al. 2005). Investigations are needed to assess the relations between shifting and internalizing and externalizing behaviour problems. Toplak et al. (2009) found that adolescents without ADHD performed significantly better on a trail-making task, assessing set-shifting, than adolescents with ADHD. Willcutt et al.'s (2005) meta-analysis indicated small effects and inconsistent results when examining the relations between shifting and ADHD. Friedman et al. (2007) revealed that lower levels of attention problems significantly predicted better shifting abilities at some ages (7, 9, 10 and 11), but not at others (8, 12, 13 or 14). The literature on the relations between working memory and internalizing and externalizing behaviour problems is limited. Working memory may be indirectly related to risk-taking behaviour via its relation with impulsivity in preadolescents (Romer et al. 2009). A relation between working memory deficits and ADHD has been found (Friedman et al. 2007, Toplak et al. 2009, Willcutt et al. 2005); however, Friedman et al. (2007) revealed a stronger link between ADHD and inhibition than between ADHD and working memory.

Purpose of the current investigation

The current investigation explores EF in children with a range of linguistic and behaviour profiles relative to the latent variable model of EF (Miyake et al. 2000). Research on EF has highlighted the importance of incorporating both laboratory measures and measures that assess performance in real-world settings (Riccio et al. 2011). Parent-report measures of EF provide information about an individual's ability to achieve goals, whereas laboratory measures assess cognitive efficiency under optimal conditions (Toplak et al. 2013).

Investigations into the relations among language and related constructs traditionally compare individuals with LI to individuals without LI; however, the present study examined these relations using language as a continuous variable, based on the perspective that individuals with LI represent the lower tail of an ability distribution, rather than a distinct category (e.g., Rescorla 2009). The following research questions were addressed: Does performance on EF tasks predict language ability? Do language ability and EF predict attention, internalizing and/or externalizing behaviour problems?

Methods

Participants

The 42 students (22 male, 20 female), aged 8:1–11;7 years;months (mean = 114.78 months, SD = 12.43 months) enrolled in this preliminary investigation were recruited from the after-school programmes of a large, urban school district in the Midwestern United States. The mother's years of education was used as a measure of socioeconomic status (range = 12–20 years, mean = 15 years, SD = 2.56). The racial/ethnic breakdown of the sample was 52% African American, 33% Caucasian and 15% more than one race/ethnicity. All participants were monolingual speakers of English, with normal or corrected to normal vision, normal hearing and no known neurological deficits. Per parent report, three participants had been diagnosed with LI, three received therapy for deficits in articulation, three received services for deficits in reading, two received tutoring for general academic needs and one had been diagnosed with behaviour problems.

The current investigation included children with a broad range of cognitive abilities, although no children fell into the range of intellectual disabilities. Recent research has found similarities between children and adolescents with specific language impairment (SLI), typically defined as LI with nonverbal cognitive scores within the average range, and those with nonspecific language impairment (NLI), who score below the average range on tests of both language and nonverbal cognition (e.g., Miller et al. 2006).

Tasks

The core language composite standard score of the Clinical Evaluation of Language Fundamentals—Fourth Edition (CELF-4; Semel et al. 2003) was used as an omnibus measure of language ability. The fluid intelligence composite score of the Leiter International Performance Scale—Revised (Leiter; Roid and Miller 1997) evaluated nonverbal cognition. One student was not administered the Leiter due to early pick-up.

Parents completed the Conners-3P (Conners 2008), a parent-report ADHD index, to assess the students’ attention; the Child Behaviour Checklist (CBCL; Achenbach and Rescorla 2001) to evaluate the students’ internalizing and externalizing behaviour problems; and the Behaviour Rating Inventory of Executive Function (BRIEF; Gioia et al. 2000) to assess the students’ EF. The t-scores for ADHD index of the Conners-3P, the internalizing and externalizing indices of the CBCL, and the inhibit, shift and working memory scales of the BRIEF were analysed.

The Dimensional Change Card Sort (DCCS; Zelazo 2006) computerized version, developed for children aged 8–11 years, was used as an experimental measure of EF. Children were shown a centred target picture with two side-by-side pictures beneath it on a 15-inch laptop computer using E-Prime® 2.0 software. A touch-screen add-on was connected via the USB port. In the first set of five trials, the children selected the picture that matched the target according to one dimension (e.g., colour). For the next five trials they selected the picture that matched according to the other (e.g., shape). In the third set of trials, the mixed condition, the dimension by which they were to select the picture switched, requiring matching by shape on some trials and by colour on others. The word shape or colour was displayed on the screen and presented auditorally to cue the correct dimension. The children matched by the dominant dimension on 40 trials, and by the non-dominant dimension on 10. Age-residualized scores for mixed condition reaction time were analysed. Due to a technical error, one child did not complete the task. The DCCS has been shown to involve multiple cognitive processes; thus, it is not a ‘pure’ measure of a single EF component (Waxer and Morton 2011).

Procedure

Prior to recruitment, approval was obtained from the Institutional Review Boards of the University of Wisconsin-Madison and of the participants’ school district. The same speech–language pathologist tested each child individually in an empty, quiet room in his or her school. The order of task administration was varied across participants to avoid order effects. Parents completed consent and background forms, the CBCL, the BRIEF and the Conners-3P.

Results

Descriptive data

Table1 summarizes the scores on the measures. Of the 40 children in the study, 13 (33%) fell within the clinical classification of LI, scoring 1.25 SDs or more below the mean on two or more subtests or composites on CELF-4. On the Conners-3P, three (8%) students earned borderline scores, two (5%) earned high scores and three (8%) earned very high scores; thus, 21% of the students in the study could be considered at risk for attention problems. On the CBCL, none of the students’ scores fell into the borderline range for internalizing behaviour problems, and five (13%) fell within the clinical range. On the externalizing index, four (10%) scored within the borderline range and six (15%) scored within the clinical range, with a total of 10 (25%) of the students at risk for externalizing behaviour problems.

Table 1

Descriptives scale

	Range	Mean	Standard deviation (SD)	N
Conners-3P ADHD Indexa	42–90	52.14	13.97	42
CBCL Externalizinga	33–77	47.04	13.63	42
CBCL Internalizinga	33–72	46.52	12.09	42
CELF-4 Core Languageb	54–123	90.33	17.76	42
Leiter Fluid Reasoningb	73–125	94.76	13.86	41
BRIEF Inhibita	37–87	46.81	10.42	42
BRIEF Shifta	36–83	46.47	11.67	42
BRIEF Working Memorya	36–83	46.38	10.14	42
DCCS	–1.87 to 3.66	0.72	0.99	41

Notes

T-scores: mean = 50, SD = 10.

Standard scores: mean = 100, SD = 15.

Correlations

Table2 displays the correlations among the measures. The demographic variables of age and SES were significantly correlated only with language scores, and nonverbal cognition was correlated only with language and DCCS. All the parent-report measures (BRIEF inhibit, working memory and shift; CBCL internalizing and externalizing; and Conners-3P ADHD) were significantly correlated with one another. The experimental measure of EF was significantly correlated with language and nonverbal cognition, but not with the parent-report measures of EF or behaviour problems.

Table 2

Correlations

Age	SES	FR	CELF	INH	SHF	WM	DCCS	ADHD	INT	EXT
EXT	0.06	−0.11	−0.10	−0.24	0.76***	0.58***	0.60***	0.20	0.63***	0.72***
INT	0.10	0.14	0.08	−0.03	0.76***	0.63***	0.65***	−0.30	0.59***
ADHD	0.14	−0.15	−0.24	−0.43**	0.76***	0.71***	0.72***	0.13
DCCS	0	0.23	−0.35*	−0.35*	−0.21	−0.02	0.02
WM	0.24	−0.02	−0.21	−0.31*	0.77***	0.82***
SHF	0.24	0.13	−0.15	−0.19	0.74***
INH	0.04	−0.03	−0.11	−0.23
CELF	−0.29*	0.33*	0.60***
FR	−0.09	0.29
SES	0.11
AGE

Notes: EXT = CBCL Externalizing t-score, INT = CBCL Internalizing t-score, ADHD = Conners-3P ADHD Index t-score, DCCS = Dimensional Change Card Sort, WM = BRIEF Working Memory t-score, SHF = BRIEF Shift t-score, INH = BRIEF Inhibit t-score, CELF = CELF Core Language Standard score, FR = Leiter Fluid Reasoning Index standard score, SES = socioeconomic status.

p ≤ 0.05,

p ≤ 0.01,

p ≤ 0.001.

Linear regression modelling

Linear regression modelling was employed using block entry to evaluate predictors of language (table3), attention problems (table4), internalizing (table5) and externalizing (table6). BRIEF inhibit, working memory and shift t-scores and residualized age scores for DCCS reaction time were entered as predictors; CELF-4 was also entered as a predictor for attention problems, internalizing and externalizing. The demographic variables did not improve the model fit when entered as predictors of attention problems, F(2,38) = 1.36, R = 0.26, R2 = 0.07, p = 0.27; internalizing, F(2,37) = 0.34, R = 0.13, R2 = 0.02, p = 0.71; and externalizing, F(2,37) = 0.35, R = 0.14, R2 = 0.02, p = 0.71; thus, the final models for attention problems, internalizing and externalizing did not include the demographic variables. CELF-4 did not significantly contribute to internalizing, F(1, 38) = 0.03, R = 0.03, R2 = 0.001, p = 0.87, or externalizing, F(1, 40) = 2.40, R = 0.24, R2 = 0.06, p = 0.13; thus, the final models for internalizing and externalizing did not include CELF-4. Nonverbal cognition significantly predicted the CELF-4 score. When entered without nonverbal cognition, DCCS and BRIEF Working Memory together were significant predictors of CELF-4, F(2,37) = 5.07, R = 0.48, R2 = 0.19, p = 0.008, but were no longer significant with nonverbal cognition in the model. SES and age were not significant predictors of CELF-4, and were not included in the final model.

Table 3

Linear regression models predicting the Core Language Composite Standard score on the Clinical Evaluation of Language Fundamentals—Fourth Edition

Block	Predictor	B	SE B	β
1	Leiter	.77	.17	.59***
	F(1,38)=20.48, R=.59, R2=.35, p<.001
2	Leiter	.61	.18	.47**
	BRIEF Inhibit	−.35	.39	−.20
	DCCS	−4.02	2.67	−.22
	BRIEF Working	.42	.47	−.23
	Memory
	BRIEF Shift	.32	.36	.21
	F(5,34)=5.48, R=.67, R2=.45, R2 Δ=.07, p=.001

Notes: Leiter = Composite Standard Score on the Fluid Reasoning Index. SES = mother's years of education; DCCS = age-residualized score on the Dimensional Change Card Sort; BRIEF = t-score on the Behavior Rating Inventory of Executive Functions.

p ≤ 0.001,

p < 0.01.

Table 4

Linear regression models predicting Conners-3P ADHD Index t-score

Block	Predictor	B	SE B	β
1	CELF	−0.34	0.12	−0.43**
F(1,39) = 8.46, R = 0.43, R2 = 0.18, p < 0.01
2	CELF	−0.14	0.08	−0.17
	BRIEF Inhibit	0.75	0.22	0.56*
	BRIEF Shift	0.31	0.21	0.26
	BRIEF Working Memory	0.02	0.27	0.01
	DCCS	2.76	1.53	0.19
F(5,34) = 16.77, R = 0.84, R2 = 0.71, R2Δ = 0.53, p < 0.001

Notes: CELF = Clinical Evaluation of Language Fundamentals—Fourth Edition Core Language Composite Score; BRIEF = Behavior Rating Inventory of Executive Functions t-score.

p < 0.05,

p < 0.01.

Table 5

Linear regression models predicting Child Behavior Checklist internalizing t-score

Block	Predictor	B	SE B	β
1	BRIEF Inhibit	0.61	0.21	0.53*
	BRIEF Shift	0.13	0.19	0.13
	BRIEF Working Memory	0.16	0.25	0.13
	DCCS	0.13	0.19	0.13
F(4,35) = 14.58, R = 0.79, R2 = 0.63, p < 0.001

Notes: BRIEF = Behavior Rating Inventory of Executive Functions t-score; DCCS = age-residualized reaction time on the Dimensional Change Card Sort.

p ≤ 0.01.

Table 6

Linear regression models predicting Child Behavior Checklist externalizing t-score

Block	Predictor	B	SE B	β
1	BRIEF Inhibit	0.91	0.25	0.70*
	BRIEF Shift	0.07	0.23	0.06
	BRIEF Working Memory	0.02	0.30	0.01
	DCCS	−0.71	1.63	−0.05
F(4,35) = 12.72, R = 0.77, R2 = 0.59, p < 0.001

Notes:

p ≤ 0.001.

BRIEF = Behavior Rating Inventory of Executive Functions t-score; DCCS = age-residualized reaction time on the Dimensional Change Card Sort.

CELF-4 was a significant predictor of the Conners-3P ADHD index score when entered as a single predictor, but was no longer significant when the EF measures were added to the model.

The BRIEF Inhibit score significantly predicted the Conners-3P score, and was the sole significant predictor of both internalizing and externalizing behaviour problems.

Conclusions

The first research question examined the relation between language ability and EF. Scores on CELF were correlated with the experimental measure of EF and with the parent-reported measure of working memory. However, when nonverbal cognition was included in the model, neither was a significant predictor. Language ability was neither correlated with nor predicted by parent-reported shifting or the DCCS, which also required the use of shifting to shift attention to the current card-sorting dimension when it changed. The lack of a significant relation between language ability and shifting is consistent with prior behavioural investigations of LI and EF (e.g., Henry et al. 2012, Im-Bolter et al. 2006). The lack of a significant relation between language ability and inhibition was surprising, given that inhibition has been found to predict scores on tests of reading, spelling, and writing (St. Clair-Thompson and Gathercole 2006), and differences in mean levels of performance in children with LI as compared with children with typical language ability have been found (Henry et al. 2012).

Results of the current investigation provide support for the notion that inhibition and language ability may not have a direct relation in school-age children, as suggested by Im-Bolter et al. (2006).

The second research question addressed predictors of behaviour problems. The results of the current investigation provide further evidence of a relation between inhibition and behaviour problems. Parent-reported inhibition significantly predicted attention problems and internalizing and externalizing behaviour problems. Language ability had a significant relation with attention problems, but this relation was no longer significant when the EF measures were added to the model. It was contrary to expectations that language was not a significant predictor of internalizing or externalizing behaviour problems, and was not significantly correlated with either. It may be the case that a significant relation between language and behaviour problems was not found due to the small number of children in the study falling into the borderline or clinical ranges for behaviour problems. Other investigations into the relation between language ability and behaviour problems have used clinically referred populations. The nature of the relation between language and behaviour problems may differ in individuals with severe behaviour problems.

EF has been proposed as a potential link between decreased language ability and behaviour problems (Brownlie et al. 2004, Villamarette-Pittman et al. 2002). The current findings suggest that inhibition and language ability both contribute to attention problems, with EF (inhibition) being the stronger predictor. However, language ability was neither predicted by nor correlated with inhibition; thus, it seems unlikely that EF provides a link between language ability and attention problems. Language ability did not have significant relations with internalizing and externalizing behaviour problems, whereas inhibition contributed to both.

These findings have clinical relevance for children with attention and/or behaviour problems. Children with attention problems, internalizing behaviour problems or externalizing behaviour problems may benefit from interventions that include focusing on inhibitory processes. Additionally, children with attention problems may benefit from intervention targeting language ability.

It should be noted that the limitations of the current investigation prevent the drawing of strong conclusions, given the relatively small sample size and the use of an EF task that taps multiple cognitive processes. Future studies should assess the relations among language, EF, and behaviour problems in a larger sample using multiple tasks, including tasks considered ‘purer’ indices of the components of EF. Additionally, measures of higher-level language processes, such as the comprehension and production of spoken and written discourse, might be assessed relative to EF and behaviour problems.