Inner Speech and Executive Function in Children With Developmental Language Disorder: Implications for Assessment and Intervention

Abstract

Purpose:

Many children with developmental language disorder (DLD) also have difficulty with executive function. The presence of co-occurring deficits in language and executive function can obscure assessment results and lead to the implementation of ineffective interventions. It is also the case that inner speech, or the use of self-directed language to guide thought and action, often mediates performance on executive function tasks. The aims of this tutorial are to (a) summarize what is known about how inner speech affects executive function performance in typical populations and children with DLD and (b) highlight potential implications for clinical practice and directions for future research. We provide a brief background on inner speech, including theoretical frameworks, typical development, and measurement approaches. We then summarize research on inner speech and executive function involving typical adults and children, followed by a description of the few studies involving children with DLD.

Conclusions:

Work with typical adults and children has concluded that inner speech operates as a self-cueing device to support understanding of task rules, sequencing of task order, and maintenance of task goals. Work involving children with DLD suggests that their inner speech is less mature, less relevant, and less effective overall when completing executive function tasks. However, very few studies have examined the relations between inner speech and executive function in children with DLD. It is important for speech-language pathologists to understand the potential role of inner speech during executive function tasks, given how often these skills are utilized during everyday activities. Although more research is needed, speech-language pathologists are in a unique position to support both language and executive function goals for children with DLD.

Developmental language disorder (DLD) is a significant impairment of language expression and comprehension without obvious causal factors such as intellectual disability or neurological impairment (Bishop et al., 2017; Leonard, 2014). DLD is highly prevalent, affecting roughly 8% of school-age children (Law et al., 2000; Norbury et al., 2016; Tomblin et al., 1997). Among children with DLD, difficulties with executive function are also quite common (Kapa & Plante, 2015). Executive function is a set of cognitive processes that guide goal-directed behavior through the coordination of working memory (i.e., updating), cognitive flexibility (i.e., shifting), and inhibition (Diamond, 2013; Miyake, Friedman, et al., 2000). Together, these core executive function skills support more complex executive functions such as planning, problem solving, and goal setting (S. M. Jones et al., 2016; Marcovitch & Zelazo, 2009). In the classroom, children with executive function challenges often have difficulty navigating the environment and completing complex assignments. Executive function is a growing topic of interest to speech-language pathologists (SLPs) supporting children with DLD because it influences behavior and learning. However, language may also influence executive function—both on formal assessments and in the natural environment. The language demands of some executive function tasks are more obvious than others. For example, many tasks have complex verbal instructions, the stimuli themselves may be spoken or written words, and language may be required to make a response. However, language is also typically involved in tasks we consider to be “nonverbal” or those without spoken or written instructions and stimuli. Inner speech, or the use of self-directed language to guide thought and action, is often overlooked during assessment and intervention, but it has the potential to greatly influence performance. The first section of this tutorial will provide a brief overview of inner speech including theoretical frameworks, typical development, and measurement approaches. For an in-depth review of these topics, see the work by Alderson-Day and Fernyhough (2015) or Vissers et al. (2020). The second and third sections of this tutorial will summarize what is known about inner speech during executive function tasks in typical populations and in children with DLD, respectively. The tutorial concludes with a section on clinical implications and directions for future research.

Inner Speech

Inner speech—also known as self-talk, internal monologue/dialogue, covert speech, or verbal thinking—can be defined as “the subjective experience of language in the absence of overt and audible articulation” (Alderson-Day & Fernyhough, 2015, p. 931). Inner speech has intrigued psychologists for decades and more recently caught the attention of SLPs (e.g., Fahy, 2014). An important assumption to consider is whether language ability and inner speech are synonymous terms or equivalent abilities within an individual. It has been suggested that the neural substrates of inner speech are the arcuate fasciculus and superior longitudinal fasciculus—structures comprising the dorsal language stream and commonly associated with more general language processing (Geva & Fernyhough, 2019). Because inner speech develops in parallel with the maturation of the dorsal language stream, it is plausible that these structures are also responsible for the development and effective use of inner speech. However, most of the direct evidence in support of this theory comes from studies of adults (see Geva, 2018, for a review). For the purposes of this tutorial, we will interpret studies according to the working hypothesis that language and inner speech are closely related. Future work—particularly that involving children with DLD—should further evaluate this assumption.

Integrating Inner Speech With Theories of Executive Function

No formal models of inner speech and executive function have been empirically tested. However, there are two leading theoretical perspectives guiding investigations of inner speech during executive function tasks. The first perspective views language as a method of rehearsal to support a specific executive function: working memory, which is the ability to actively retain and process information for a limited time. Working memory is often demonstrated in digit span or sentence repetition tasks where a participant hears and repeats verbal information, typically with some kind of manipulation such as reversing the order of the digits. There are several models of working memory, but the one most often cited in works on inner speech is the multicomponent model from Baddeley and Hitch (1974). According to this model, working memory is the product of three components: a central executive system for guiding attention, a phonological storage system or “phonological loop,” and a visuospatial storage system or “scratchpad.” The phonological loop, sometimes called the articulatory loop, has been highlighted as a mechanism of verbal rehearsal—or inner speech—during working memory tasks. When this mechanism is disrupted (e.g., through articulatory suppression, which will be discussed in the section on measurement), working memory capacity will be reduced.

The second theoretical perspective considers language as an essential tool for mediating a variety of cognitive processes (Vygotsky, 1987). Indeed, the concept of inner speech was originally put forward by the notable Russian psychologist Lev Vygotsky. His sociocultural theory of development pinpointed language as one of the most valuable tools for mediating intellectual functioning and solving problems. More recently, the hierarchical competing systems model (HCSM; Marcovitch & Zelazo, 2009) has been used to provide a conceptual explanation for the relation between inner speech and executive function (e.g., Larson et al., 2019). The HCSM was developed to explain the emergence and early development of executive function in the context of a simple hidden object or visual search task that draws on multiple executive functions. According to the HCSM, goal-directed behavior is influenced by two hierarchically arranged systems. The rudimentary habit system depends on previous experience to generate automatic responses. The higher order representational system utilizes conscious reflection, typically in the form of inner speech, to modify automatic responses. The process of using inner speech to reflect on and change behavior is also known as verbal mediation. During executive function tasks, verbal mediation influences the habit system to produce the correct response.

Although the HCSM has not been vetted across a wide variety of executive function tasks, its focus on goal-directed behavior rather than a specific executive function is aligned with the established unity and diversity model of executive function, developed by Miyake and Friedman (Friedman et al., 2008; Miyake & Friedman, 2012; Miyake, Friedman, et al., 2000). In this model, a “common executive function” underlies the three core executive functions of updating, shifting, and inhibition. Updating is the constant monitoring and revising of information in working memory, shifting is the flexible switching between tasks or mental sets, and inhibition is the intentional with-holding of a dominant response. The common executive function is the “ability to maintain task goals and goal-related information and use this information to effectively bias lower-level processing” (Miyake & Friedman, 2012, p. 11). Figure 1 connects verbal mediation from the HCSM (Marcovitch & Zelazo, 2009) to the unity and diversity model of executive function (Friedman et al., 2008; Miyake & Friedman, 2012; Miyake, Friedman, et al., 2000) to provide a conceptual framework for how inner speech might influence executive function task performance. According to this figure, executive function task performance is based on one or more specific executive function skills and the common executive function ability, which is mediated by conscious reflection in the form of inner speech. Conceptually, strong verbal mediation skills would support the common executive function to achieve the desired performance. Weaknesses in verbal mediation ability and/or the common executive function could lead to impaired task performance.

Figure 1.

The shared theoretical foundation of inner speech and executive function. Friedman et al. (2011) found that inhibition does not account for unique variance in executive function (EF) task performance after considering shifting-specific skills, updating-specific skills, and common EF.

Typical Development of Inner Speech

Vygotsky (1987) suggested that language directs thought, and over the course of development, it gradually progresses from external to internal forms. Recent reviews have refined and classified Vygotsky’s theory about the internalization of language into four concrete stages of inner speech development (Alderson-Day & Fernyhough, 2015; Vissers et al., 2020). These stages are summarized in Table 1. Stage 1 aligns with early language acquisition, and it involves the overt and social use of spoken language to communicate with others. Stage 2 begins around 3–4 years of age and involves a pivot from dialogue with others to overt conversation with the self (i.e., private speech). Often, children will imitate the language used by their caregivers during independent play to regulate the actions of toys and other manipulatives. Vygotsky classically referred to this behavior as egocentric speech and described the importance of learning to put psychological space between the self and the immediate situation (Vygotsky & Luria, 1994). Stage 3 begins around 6–7 years of age and involves a shift from overt to covert self-talk (i.e., from private to inner speech) in which children begin to regulate their own thoughts and actions using an internalized and self-directed form of language. Some studies describe an even finer progression from private to inner speech as it moves from intelligible to unintelligible, from voiced to unvoiced (Winsler & Naglieri, 2003). This progression may feel counterintuitive to SLPs who are focused on the accuracy of language forms; however, in the context of verbal mediation, private speech that is unintelligible or unvoiced signals a more internalized and thus more mature ability to regulate thought and action. The fully silent form of inner speech used to be considered developmental mastery; however, Stage 4 is a more recent advancement of inner speech theory. It involves a transition from the intentional use of concrete grammatical phrases or sentences to an unconscious use of condensed or abstract concepts to guide behavior.

Table 1.

Prospective model of typical development of inner speech.

Stage	Form	Description	Typical age behavior starts	Example
1	External dialogue	Overt, social use of language to communicate with others.	1–2 years	“Mommy, play blocks.”
2	Private speech	Overt use of language to communicate with the self. Language is becoming more internalized: unintelligible, unvoiced. Lips might move during focused tasks.	3–4 years	“I can build it…. Put the red block here.”
3	Inner speech	Covert use of language to regulate own thoughts and actions. Language is often a grammatical phrase or sentence.	6–7 years	No audible speech: First put the red block and then add the blue one.
4	Condensed inner speech	Fully internalized use of language that is reduced to a condensed or abstract form.	8+ years	No audible speech: First red, then blue.

Note. This developmental structure was adapted from existing reviews of inner speech (Alderson-Day & Fernyhough, 2015; Vissers et al., 2020). There is consensus about the sequence of stages, but there are limited empirical data to support the specific age at which each stage begins.

In summary, the development of inner speech progresses from social to self-directed, external to internal, and concrete to abstract. A similar trajectory can be observed in the development of executive function (Barkley, 2012). For example, infants rely on their caregivers to provide self-regulatory soothing, and toddlers depend on external commands to inhibit unsafe behaviors. Young children imitate the language modeled by their caregivers to direct the actions of their toys and peers before they produce covert, self-directed speech to guide their own actions. Eventually, children learn to visualize their future actions and related outcomes, and they internalize the language used to guide their actions.

Measurement of Inner Speech

Due to the internal nature of the behavior, inner speech is rather difficult to measure directly. Several indirect methods have been reported in the literature, each with pros and cons depending on the population and purpose of measurement. Here, we describe the measures that are most relevant for use with children with DLD and in the context of completing executive function tasks. These include self-report measures such as questionnaires or experience sampling, private speech observation, and articulatory suppression. A discussion of assessment for clinical purposes is provided later in the implications section of this tutorial.

Self-report.

Measures of self-report, such as questionnaires and experience sampling, can provide an indirect assessment of inner speech that occurs during a specific activity. These measures can provide qualitative and quantitative information about the type and frequency of inner speech from the perspective of the “speaker.” Although they are most often used with adults (Alderson-Day et al., 2018; Brinthaupt et al., 2009; McCarthy-Jones & Fernyhough, 2011; Morin et al., 2011), there is a precedent for their use with children as well. For example, Winsler and Naglieri (2003) administered a trail-making task from the Cognitive Assessment System (Naglieri & Das, 1997) to a large, cross-sectional sample of children and adolescents aged 5–17 years. Immediately after completing the task, participants were asked, “Tell me how you did these,” or “How did you complete these pages?” The examiner recorded whether participants indicated use of private or inner speech as a strategy. Results suggested that about 60% of participants, across all age groups, reported using some kind of inner speech strategy. There was also a linear increase in participants’ self-report of inner speech strategy use from 5 (8%) to 17 (37%) years of age. The authors concluded that, overall, awareness of verbal mediation strategy was low, but it increased with age (Winsler & Naglieri, 2003). The need for metacognitive awareness about one’s use of inner speech is a potential limitation of self-report methods with younger children. Furthermore, the need to use receptive and/or expressive language to understand and respond to questionnaires puts children with DLD at a disadvantage in reporting potential strategy use. Nonetheless, self-report measures are a highly feasible source of information that can be gathered in addition to the primary outcome of task performance.

Private speech observation.

Another indirect measure of inner speech is to observe verbal output and behavior during task completion. Several studies have utilized a coding scheme (adapted from Berk, 1986) that captures the use of self-directed language along a continuum of internalization (Al-Namlah et al., 2006; Fernyhough & Fradley, 2005; Lidstone et al., 2011). It begins with social speech (i.e., spoken language addressed to others), then private speech (i.e., spoken language addressed to self), and finally observable movements suggestive of inner speech such as unintelligible whispering or silent lip and tongue movements. Although simple observations may or may not capture the use of private speech, some studies manipulate tasks to promote its use (Fernyhough & Fradley, 2005; Lidstone et al., 2011, 2012; Sturn & Johnston, 1999). For example, the Tower of London task requires moving balls or disks, one at a time, along three vertical pegs from a starting position to match a target position. Fernyhough and Fradley (2005) modified the task instructions to include, “Sometimes children like to talk aloud when they play this game. You can do that if you like. I bet in class you have to be quiet! While playing this game you can talk and say whatever you want to” (p. 109). It should be noted that as a child’s awareness of the observation increases, the authenticity of their private speech behavior may decrease. If children with DLD do not use private speech as often or in the same ways as their peers with typical development, this method of observation may be limited. Indeed, “failure to observe the child using covert verbal strategies does not necessarily indicate the absence of their use” (Winsler & Naglieri, 2003, p. 669).

A similar line of work by Kuvalja et al. (2014) went beyond qualitative descriptions and frequency analyses to more objectively examine the co-occurrence of observed private speech and self-regulatory behavior. They audio- and video-recorded 6-year-old children with and without DLD while they completed a structured planning activity using Lego bricks. Basic frequency analysis found no significant differences between groups in the total amount of self-directed speech or self-regulatory behavior. The authors then used lag sequential analysis (LSA) to identify pairs of events (i.e., verbal and nonverbal behaviors) within a specified time window. Although LSA revealed pairs of behaviors that co-occurred throughout the sample, it did not account for temporal order (i.e., which behavior occurred first), and it could not detect patterns of more than two behaviors. Therefore, t-pattern analysis was used to extract recurring patterns from the continuous stream of behavior. The t-pattern analysis revealed that children with DLD produced longer patterns of behavior with more complex temporal relationships than their peers with typical development. Kuvalja et al. (2014) suggest that t-pattern analysis allows for the unbiased identification of complex patterns that would otherwise go unnoticed with LSA or frequency analysis. It is a robust method with the potential to elucidate the role of inner speech during real-world tasks involving executive function.

Articulatory suppression.

Another approach to studying inner speech is to prevent it from occurring. By comparing task performance with and without the use of inner speech, researchers can make inferences about the impact of inner speech on a specific task. To prevent the use of inner speech, researchers often use a “dual-task paradigm” in which participants are asked to simultaneously complete the primary task and a secondary language-based task such as repeating a word or phrase at 1-s intervals. Even very simple articulatory suppression tasks such as repeating a single word (e.g., “the, the, the”) can disrupt the use of inner speech and impair executive function task performance (Baddeley et al., 2001). The change in performance across conditions (with/without articulatory suppression) tells us how helpful inner speech was for task completion. If there is a decrease in task performance with articulatory suppression compared to without, it suggests the participant was effectively using inner speech to complete the task. Combining simple articulatory suppression with an executive function task has become a common method for empirical investigations of inner speech in both adults and children (Emerson & Miyake, 2003; Fatzer & Roebers, 2012; Karbach & Kray, 2007; Kirkham et al., 2003; Kray et al., 2004, 2008; Miyake et al., 2004; Saeki & Saito, 2004). It has even been used in the few studies of children with DLD (Larson et al., 2019; Lidstone et al., 2012). Details and findings from these studies will be reported in the following sections.

Inner Speech and Executive Function in Typical Populations

In this section, we will describe what is known about how inner speech influences executive function in populations with typical development. Most studies use a combination of the measurements described above and focus on a single executive function skill (e.g., updating, shifting, or inhibition). However, it is important to remember that the influence of inner speech is likely directed toward the common executive function underlying all goal-directed tasks (Miyake & Friedman, 2012). Converging evidence for the effects of inner speech across multiple executive function tasks (Fatzer & Roebers, 2012) further supports the involvement of a common executive function ability.

Articulatory Suppression Uniquely Impairs Executive Function Performance

In the dual-task paradigm, one could reasonably conclude that completing two tasks simultaneously would impair performance. However, the effects of articulatory suppression appear to have an impact on executive function task performance that is distinct from other types of dual-task conditions. Most studies that use articulatory suppression compare it to a secondary task that is nonlinguistic, and this is nearly always a motor task (Baddeley et al., 2001; Emerson & Miyake, 2003; Fatzer & Roebers, 2012; Kray et al., 2004; Miyake et al., 2004; Saeki & Saito, 2004; Wallace et al., 2017). A seminal example comes from Emerson and Miyake (2003), who examined inner speech during shifting using an arithmetic switching task. Typical adults completed the task in three conditions in random order: (a) with articulatory suppression, (b) with foot tapping, and (c) without any secondary task or the control condition. The articulatory suppression condition resulted in significantly poorer performance than the foot tapping condition (Emerson & Miyake, 2003). This finding leads to two possible conclusions: (a) preventing inner speech made the task more difficult or (b) the dual task of articulatory suppression simply has greater cognitive demands than foot tapping. We can determine which conclusion is more accurate by examining switch cost or the increase in reaction time for trials that require a shift in response strategy (i.e., rule or mental set) versus non-switch trials where the previous response strategy is repeated. In Emerson and Miyake’s study, switch cost increased by 62% during the articulatory suppression condition compared to the foot tapping condition. The significant negative impact of articulatory suppression leads to the conclusion that preventing inner speech is more detrimental to task performance than other types of dual-task demands.

Similar effects of articulatory suppression on executive function task performance have also been found in children with typical development (Fatzer & Roebers, 2012; Kray et al., 2004). Fatzer and Roebers (2012) examined all three core executive functions (updating, shifting, and inhibition) in children aged 6 and 9 years. On the measures of updating and shifting, children were less accurate in the articulatory suppression condition compared to a tapping condition (i.e., secondary motor task) and a control condition (i.e., no secondary task). The difference between the articulatory suppression and tapping conditions only reached significance for the updating task, which, by design, had the highest demand on memory and the need to use inner speech. There was no effect of articulatory suppression on the inhibition task, which had the smallest demands on memory and language.

In summary, when articulatory suppression interferes with executive function task performance, the conclusion is that participants are unable to utilize inner speech to support the common executive function. Without inner speech to maintain the task goals, responses may be slower and/or less accurate. Furthermore, a larger negative effect from articulatory suppression has been associated with more positive self-monitoring scores on the Behavior Rating Inventory of Executive Function for Adults (Wallace et al., 2017). This suggests that articulatory suppression—or more likely the tendency to use inner speech while completing executive function tasks—is related to real-world executive control.

Task Characteristics Influence the Use of Inner Speech

Executive function tasks differ in countless ways, and it is no surprise that some tasks require more inner speech than others. There are some general task features that have been examined with regard to inner speech including task difficulty, task familiarity, and cue types.

Task difficulty.

Depending on how executive function performance is measured, the difficulty of the task may influence how much inner speech is used. As task difficulty increases, so does the general use of inner speech; however, the type of inner speech used is not qualitatively different (Emerson & Miyake, 2003; Fernyhough & Fradley, 2005). Fernyhough and Fradley (2005) examined 5- and 6-year-old children with typical development using a Tower of London task and private speech observation. They varied the difficulty of the task by increasing the number of moves required to match the target (i.e., two-, three-, four-, and five-move trials). Utterances produced while completing the task were classified as either social or private and then coded for frequency, task relevance, and overtness. As expected, it took more time and a greater number of moves to complete the more difficult problems. When comparing task difficulty with the use of private speech, they found that more private speech occurred at the intermediate levels of difficulty than the easiest or hardest levels. There were no differences, however, in the frequency of private speech subtypes (i.e., levels of overtness) at different levels of difficulty. This finding is broadly in line with the educational concept of the “zone of proximal development” (Vygotsky, 1987), with inner speech being unnecessary at easier levels and ineffective at harder levels.

Task familiarity.

Inner speech may not be required for highly familiar tasks (Fernyhough & Fradley, 2005; Kray et al., 2008). In addition to task difficulty, Fernyhough and Fradley (2005) also examined familiarity effects for the Tower of London across two sessions that occurred 1 week apart. Each session contained just four trials of the Tower of London, but there was a significant interaction between performance during Sessions 1 and 2. The easiest problems on the Tower of London (i.e., problems requiring two to three moves) were solved faster in Session 2 than in Session 1. Notably, they did not find a significant interaction between session and the amount or type of private speech observed. In other words, overall performance on the Tower of London improved with practice or familiarity, but the use of inner speech remained relatively low for easier items across both sessions. Therefore, the effect of task familiarity on inner speech is still unclear for young children with typical development.

Kray et al. (2008) utilized a more comprehensive procedure where both child participants (aged 7–9 and 11–13 years) and adult participants (aged 20–27 and 66–77 years) completed the same switching task and conditions over the course of four or five sessions. The first one or two sessions comprised the introduction phase, and the remaining three sessions comprised the experimental phase. The switching task involved alternating between two rules: decide whether the pictured animal (a) swims or flies or (b) is shown in color or in black and white. The use of inner speech was manipulated across conditions in which participants read the task rule aloud, read an irrelevant word, or read/verbalized nothing. Over the course of the three experimental sessions, Kray et al. found that practice effects significantly reduced processing time, error rates, and the impact of inner speech. More specifically, the costs and benefits of verbalization (i.e., time spent reading irrelevant or relevant words, respectively) were reduced from the first to third sessions of the experimental phase. The authors concluded that the language processes of inner speech are primarily required when a new task set needs to be implemented to strengthen stimulus–response rules and task goals.

Cue types.

Several studies have manipulated cue type to better control how inner speech operates during executive function tasks (Chevalier & Blaye, 2009; Emerson & Miyake, 2003; Kray et al., 2004, 2008; Miyake et al., 2004). In general, cues can range from high to low transparency according to a few key characteristics. High-transparency cues are concrete and informative; they are often complete linguistic forms (e.g., written or spoken words). Low-transparency cues are abstract or arbitrary symbols that provide partial information (e.g., letters or colors). Given a high-transparency cue, inner speech might be used to read or label the response rule. Given a low-transparency cue, inner speech is required to translate the cue into something relevant to the task. During articulatory suppression, when the ability to use inner speech is prevented, it is more difficult to translate cues and mediate task performance. For example, Miyake et al. (2004) examined typical adult performance on a cued task-switching paradigm. Participants had to select either the shape or the color of a geometric figure based on a cue. In the high-transparency word cue condition, participants saw the word “SHAPE” or “COLOR” before and during the stimulus presentation. In the low-transparency letter cue condition, participants saw the letter “S” or “C.” Both cue conditions were presented with and without articulatory suppression. There was an increased switch cost during articulatory suppression in the letter cue condition that was not present in the word cue condition. This can be interpreted to mean that retrieval of the task rule was most difficult for low-transparency cues when inner speech was prevented. The task rule was retrieved more easily when the cue was highly transparent, regardless of whether inner speech was used.

Similar results for cue manipulations have been reported in children (Chevalier & Blaye, 2009; Kray et al., 2004). For example, Chevalier and Blaye (2009) used an advanced Dimensional Change Card Sort (DCCS) task to examine shifting and inner speech in children aged 5–9 years. Their task required children to match a stimulus picture with one of two target pictures by color or shape. There were three cue types: transparent auditory cues (i.e., the spoken words “color” or “shape”), transparent visual cues (i.e., a multicolor string for the color rule or a black square for the shape rule), arbitrary or low-transparency visual cues (i.e., a black square or gray circle, with rule assignment counterbalanced across participants). In general and like adults, children were more accurate and faster to respond when cues were highly transparent.

Finally, there is evidence that more active processing of cues may lead to better performance on executive function tasks (Kirkham et al., 2003). For very young children, aged 3 years, performance on a DCCS task was significantly better in a labeling condition compared to a standard condition. In the labeling condition, participants answered the questions, “What’s this one? Where does it go?” before responding to each trial. In the standard condition, the examiner provided the cue, “Here is a red one. Where does it go?” The authors posit that active labeling may serve to inhibit one mental set, redirect attention to another mental set, and verbally mediate performance. For slightly older children, aged 4 years, there were no significant differences between the standard and labeling conditions. It is possible that better attention and/or better linguistic comprehension accounts for these results.

The Effects of Inner Speech Change Across the Life Span

As some of the prior evidence suggests, inner speech, executive function, and the interaction between them change across the life span. It is well established that executive function ability generally increases during childhood and adolescence, reaches a ceiling during young adulthood, and then slowly declines with old age (Cepeda et al., 2001; Zelazo et al., 2004). A similar developmental trajectory has also been documented in studies examining the effects of inner speech during executive function tasks (Kray et al., 2004, 2008). Specifically, Kray et al. (2004) compared three age groups: children (8–10 years), younger adults (20–25 years), and older adults (61–72 years) on a cued task-switching paradigm. They found that younger adults performed better (i.e., had lower set selection costs) than both children and older adults. During a verbal secondary task condition, they found that reading irrelevant and conflicting verbal cues, similar to an articulatory suppression task, interfered with performance for older adults but not children. Reading relevant verbal cues, or scaffolding inner speech, facilitated task performance for children but not older adults.

In a later study, Kray et al. (2008) used a similar switching task with various cue manipulations to examine four age groups. The overall design was similar to the 2004 study with groups of younger adults (20–27 years) and older adults (66–77 years); however, children were also divided into younger (7–9 years) and older (11–13 years) groups. In addition to replicating the findings from the previous study, they also found that age-related differences in task switching were increased when irrelevant cues were present (i.e., during articulatory suppression). The differences were most extreme for younger children and older adults, which suggests that these populations rely on internal cueing more than older children and younger adults. The authors conclude that the effect of inner speech on executive function (or at least shifting ability) differs across the life span. Specifically, inner speech may serve as a compensatory mechanism in younger children and older adults. Additional studies are needed to confirm a causal relationship between inner speech and executive function over time, but these results are promising. They also support the possibility that differences in language ability—whether due to development or disorder—underlie the connection between inner speech and the common executive function.

Summary of Inner Speech and Executive Function in Typical Populations

In summary, much is known about the development of inner speech and its effects on executive function task performance in typical populations, despite some methodological limitations. The use of articulatory suppression within a dual-task paradigm yields evidence that inner speech has a unique effect on executive function. More specifically, the suppression of inner speech impairs executive function task performance as measured by increased reaction times and switch costs. As tasks increase in difficulty, so does the use of inner speech (as determined by private speech observation or inferred through articulatory suppression). On the other hand, as tasks become more familiar, the need to use inner speech decreases. The presence of task cues seems to facilitate more efficient use of inner speech, especially when cues are highly transparent and require active processing.

Many have concluded that inner speech serves as a self-cueing device that can be used to support understanding of task rules, sequencing of task order, and maintenance of task goals (e.g., Cragg & Nation, 2010; Emerson & Miyake, 2003). The dynamic role of inner speech supports the idea that it acts upon the common executive function (i.e., goal-directed behavior; Miyake & Friedman, 2012), rather than any one executive function skill. Thus, verbally mediated goal-directed behavior interacts with shifting, updating, and inhibition abilities to perform executive function tasks. Given that suppressing inner speech or preventing verbal mediation has a negative effect on executive function task performance, it is important to consider how the inner speech of children with language impairment might differ from their typical peers.

Inner Speech and Executive Function in DLD

Much less is known about the use of inner speech during executive function tasks in children with DLD. Limited or impaired inner speech has been reported in other populations with atypical language abilities, including adults with aphasia poststroke (Geva et al., 2011; Langland-Hassan et al., 2015), children with hearing loss (Hall et al., 2017; Jamieson, 1995), and children with autism spectrum disorders (Holland & Low, 2010; Larson et al., 2020; Lidstone et al., 2009; Russell-Smith et al., 2014; Wallace et al., 2009; Williams & Jarrold, 2010). Most studies that examine inner speech in children with DLD have used observational and/or dual-task articulatory suppression in the context of problem-solving tasks such as the Tower of London (Kuvalja et al., 2014; Larson et al., 2019; Lidstone et al., 2012; Sturn & Johnston, 1999). More recent work has examined children with DLD in the context of shifting tasks (Kapa, 2021). What follows is a summary of the current evidence for how inner speech influences executive function performance in children with DLD.¹

Sturn and Johnston (1999) examined the inner speech of preschool children with DLD (aged 4 years) in comparison to an age-matched control group and a younger language-matched control group (aged 3 years). Children completed a play-based problem-solving activity in familiar peer dyads, with at least one member of the dyad demonstrating typical language development. Each session was audio-/video-recorded, and each child’s utterances were coded for relevance (i.e., narrow relevance, broad relevance, or irrelevant), function (i.e., regulating, affective, or word play), and addressee (i.e., private or social). Regulatory utterances that had narrow relevance to the task were classified as “problem-solving language.” These utterances reflect verbal mediation in the form of private speech. Older children with typical development produced the most problem-solving language, followed by children with DLD, and then younger children with typical development. Sturn and Johnston suggested that children who speak less overall will produce less problem-solving language and, therefore, will experience fewer cognitive benefits from language. The explanation for reduced problem-solving language could be that (a) children with DLD have experienced the costs associated with incorrect utterances (e.g., increased problem-solving time or decreased problem-solving accuracy), and so they avoid them, and/or (b) children with DLD are using various forms of inner speech that are actually impeding their cognitive efficiency. The authors cautioned that their small sample size (n = 6 children per group) and considerable variance within groups make their findings less reliable. There is also the possibility that groups differed in internalized forms of inner speech not observable to the researchers.

Using the dual-task paradigm with articulatory suppression, studies have produced evidence that school-age children with DLD demonstrate limited verbal mediation abilities that negatively influence executive function task performance (Larson et al., 2019; Lidstone et al., 2012). For example, Lidstone et al. (2012) investigated children (aged 7–11 years) with DLD and typical development while completing a problem-solving task with and without articulatory suppression. The task was a Tower of London with modifications to encourage thoughtful planning before moving the pieces. At baseline (without articulatory suppression), children with DLD were significantly less accurate at solving problems than children with typical development (accuracy: DLD = 42%, typical development = 57%). During articulatory suppression, children with typical development demonstrated reduced problem-solving abilities, whereas children with DLD did not experience a negative effect (accuracy: DLD = 43%, typical development = 50%). Children with typical development appear to use their intact language skills to facilitate executive function performance, which is negatively impacted when inner speech is suppressed. Children with DLD are less efficient in using language to facilitate executive function, so performance is relatively unchanged when inner speech is suppressed. This study also reported that during planning, the inner speech of children with DLD was less internalized and, thus, less mature than would be expected for their age.

Larson et al. (2019) employed a very similar methodology and investigated children (aged 8–12 years) with DLD and typical development. They found that children with DLD achieved a similar level of accuracy as their peers with typical development at baseline and during articulatory suppression. However, they observed that, on average, children with DLD spent significantly less time planning and, thus, less time using inner speech. Interestingly, children with DLD who had relatively better language ability spent less time planning than children with DLD who had poorer language ability. This pattern was reversed for children with typical development. Taken together, these results support a relationship between language and problem solving, but more work is needed to understand how this differs among children with DLD and typical development.

Only recently has the inner speech of children with DLD been examined outside problem-solving tasks. Kapa (2021) examined shifting in children (aged 4–5 years) with DLD using a standard version of the DCCS task. This task required matching a stimulus picture with one of two target pictures by color or shape. In the “target removed” condition, the target pictures were temporarily removed from view when it was time to switch rules or mental sets. This condition was designed to support attention and provide a visual cue to the switch. In the “labeling” condition, the child had to give a verbal response to the prompt, “What is this one?” before sorting each card. This condition was designed to facilitate the use of inner speech and verbally mediate the response. In a sample of preschool children, performance was better under the target removal condition than the labeling condition. Some children were able to accurately label the cards but still sorted them incorrectly. In other words, effectively scaffolded inner speech did not have a positive impact on performance. Kapa concludes that, for children with DLD, supporting visual attention by removing the target may have a greater impact on executive function task performance than verbal mediation. This contrasts with evidence from typical populations that suggests children as young as 3 years old will perform better with verbal mediation compared to without (Kirkham et al., 2003).

In summary, the inner speech used by children with DLD has been described as less mature, less relevant to the task, and less effective overall when compared to children with typical development. However, given the limited evidence among a wide range of ages and tasks, there are exceptions to this statement, and understanding the causal role of inner speech during executive function tasks requires further investigation.

Clinical Implications for Children With DLD

The extant literature provides a solid foundation for strategically considering the effects of inner speech on executive function. Although relatively less is known about these effects in children with DLD, there are some important clinical implications for assessment and intervention.

Assessment

We know that inner speech plays an influential role in mediating performance on a wide variety of tasks. We can safely assume that many assessments, even those measuring “nonverbal” skills, utilize inner speech to mediate goal-directed behavior and improve performance. We can also predict that children with DLD are using inner speech less often and less efficiently than their peers with typical development. Therefore, when a child with DLD performs poorly on an executive function assessment, how do we know whether it is due to poor executive function, weak inner speech, or both? Given what we currently know about inner speech and verbal mediation, we need to reconsider how executive functions are assessed. Only by controlling for verbal mediation in some way can we obtain an accurate picture of the executive functions targeted by an assessment.

If the goal of assessment is to determine executive function ability, then we need to create conditions that either bypass or facilitate the use of verbal mediation. Many of the studies presented in this tutorial achieved this goal by using articulatory suppression to prevent inner speech or by providing written cues to prompt and guide the use of inner speech. If the goal of assessment is to compare executive function between children with and without DLD, articulatory suppression is a particularly helpful approach. As seen through numerous examples in this tutorial, preventing the use of inner speech creates a common baseline condition in which inner speech cannot be used to mediate performance. The resulting performance should reflect a relatively pure executive function ability. Unfortunately, there are currently no standardized assessments that utilize the articulatory suppression approach. Although research studies can counterbalance the order of tasks administered with and without articulatory suppression across many participants, clinical assessment of a single participant must consider test–retest reliability—the second administration of a given task will be influenced by the first. Creating valid and reliable assessments of executive function that account for inner speech is an important direction for future research.

Alternatively, if the goal of assessment is to measure inner speech itself, SLPs can adapt their current methods with relative ease. During observations of language and social communication, SLPs can informally record the presence of private or inner speech behaviors. Similarly, problem-solving activities such as bridge building or map making could be structured to facilitate and observe the use of verbal mediation. Although the presence or absence of these behaviors is not a definitive measure of verbal mediation, these observations could provide valuable information about a child’s existing strategy use and be helpful for planning interventions. Observational assessment methods have the potential to shape how we understand and support executive function challenges in school-age children; however, like any informal assessment, their results must be interpreted with caution.

We also want to acknowledge that SLPs may or may not be involved in the direct assessment of executive function in their work setting. Therefore, our primary recommendation is that clinicians focus on building awareness of and advocating for the important role of language in assessment and everyday activities, especially those that require the common executive function or goal-directed behavior. In school settings, executive function is often assessed by a neuropsychologist who shares standardized scores with the educational team. In instances where a child achieves low scores across multiple academic, cognitive, and language assessments, the SLP might want to initiate a discussion among team members about how language abilities, in the form of inner speech, could have affected performance in all areas. This discussion might contribute to more effective planning of long-term goals and allow the SLP to advocate for an appropriate service delivery model—one that includes ample time and resources for SLP services. The evidence presented in this tutorial can support the argument that language impairments are likely to influence all tasks and activities that require goal-directed behavior.

Intervention

Given that verbal mediation is an inherently language-based strategy, SLPs are well positioned to implement inner speech interventions. There are a few features of inner speech that make it a flexible and feasible target for intervention. First, it can be used in any environment (i.e., classroom, home, and playground) and is not necessarily dependent on physical materials. Second, inner speech can be applied to virtually any task that requires executive function or goal-directed action such as focusing/shifting attention or problem solving. Third, inner speech routines can provide a context for generalizing the use of language forms initially targeted with other intervention goals. In its most basic form, inner speech is a tool for self-cueing. As seen throughout the examples in this tutorial, children benefit from external cues while learning to use inner speech effectively. Concrete and physical cues (e.g., written words or scripts) may be necessary in the early stages of learning to use inner speech as a reminder to inhibit actions, verbalize a plan, or shift to a new task. After gaining proficiency, a transition to more transparent cues (e.g., small symbols or spoken reminders) may also be effective. In general, more research is needed to determine the effectiveness of inner speech interventions; however, some efforts have been made to measure the potential efficacy of experimental inner speech training programs (Deák et al., 2004; Doebel & Zelazo, 2016; van de Sande et al., 2016), although only one has examined children with DLD (Abdul Aziz et al., 2016).

Abdul Aziz et al. (2016) investigated whether children with DLD (aged 4–7 years) could improve their use of inner speech over the course of a 5-week training program in comparison to wait list and typically developing control groups. The training program involved two 30-min sessions per week for a total of 10 sessions during which an experimenter encouraged and scaffolded overt verbal planning during collaborative Lego-building activities. There was a general structure to each session, but no specific scripts were used. When the child successfully used “self-regulatory speech,” they were rewarded with stickers. The goals for each session were as follows: (1 and 2) encourage the use of verbal mediation for focusing and paying attention, (3 and 4) sustain attention and refocus attention when distracted, (5 and 6) develop a sequential plan for completing the task, (7–9) use verbal mediation when feeling challenged on more difficult tasks, and (10) summarize all the skills learned and discuss how to use them at home and in school. Pre- and post-intervention, children completed a Tower of London task with a score of 0–3 given for each item based on accuracy and number of moves or attempts. Post-intervention, both children with typical development and DLD performed better on the problem-solving task compared to pretest, but children with DLD made greater gains than their typical peers. This study also reported increased frequency and internalization of verbal mediation behavior and decreased social language use in children with DLD following training, which is in line with typical developmental patterns of inner speech (Abdul Aziz et al., 2016).

The protocol implemented by Abdul Aziz et al. (2016) could be adapted for use in multiple contexts and to meet a variety of goals. Although Lego building is one kind of problem-solving activity, many tasks can be addressed within a problem-solving framework. For example, craft projects, worksheets, or even written assignments can serve as the “problem” to be solved in a classroom or home environment. In individual or small group sessions, SLPs can model and facilitate the types of inner speech used to complete that type of problem. Although the study by Abdul Aziz et al. demonstrated positive outcomes for children with DLD within a feasible treatment model (i.e., two 30-min sessions per week), it is unclear what the “active ingredient” was in changing problem-solving performance. Are some phrases or types of inner speech more effective than others? How much clinician modeling is required for a child to learn the strategy? Which of the session goals described above (e.g., use inner speech to sustain attention, use inner speech to develop a sequential plan, and implement inner speech during increasingly challenging problems) contributed the most to improved problem solving? These questions and numerous others could be asked of any multicomponent intervention program and are excellent topics for future research.

One way to investigate the effectiveness of complex executive function programs is to examine components individually. For example, “stop-and-think” strategies can be implemented in isolation and, as such, are relatively well studied. Typical children (aged 3–5 years) can improve performance on a switching task when given metacognitive reminders to stop and think about the current rule before each trial (Deák et al., 2004). A similar approach involving prompts to stop and verbally state a plan before submitting responses had positive effects on early literacy performance for typical kindergarten children aged 6 years (van de Sande et al., 2016). What is unique about the work by van de Sande et al. (2016) is that the stop-and-think routine was implemented in the context of completing a computer-administered preliteracy program. Groups of three to six children were stationed at individual computers and given a stuffed animal. A single examiner instructed the group of children to formulate their thoughts out loud to their animal before reaching for the computer mouse to select their answers. After this initial instruction, children worked independently, and the examiner intervened only when absolutely necessary. Children who completed the preliteracy program with the stop-and-think routine made greater gains in preliteracy skills than a control group who completed the preliteracy program only. Furthermore, children with poor executive function skills at pretest benefited from this combined intervention more than children with strong executive function. Investigations of the stop-and-think strategy in older school-age children and children with DLD are warranted.

In contrast to direct instruction of verbal mediation strategies, it is possible that indirect exposure to relevant language forms influences the form of inner speech, which then affects executive function task performance. Doebel and Zelazo (2016) found that exposure to contrastive language (e.g., “This one is not round. It’s different.”) improved typical preschooler’s performance on executive function tasks requiring shifting and resolution of conflicting rules. Specifically, participants completed two sessions containing five tasks designed to mimic natural language exposure. Language was used to describe picture cards and (a) emphasize contrast between pairs of objects, attributes, and actions; (b) emphasize contrast between categories; (c) highlight contrast between associated versus non-associated items; (d) highlight contrast between dimensional values such as shape or color; and (e) highlight contrast between actions. Importantly, the exposure procedures did not include any direct instruction about how to use the language forms during the executive function tasks. Children with DLD may or may not be able to make this generalization given possible difficulty with implicit learning (see Lammertink et al., 2017, for a meta-analysis of statistical learning abilities in children with language impairment). Regardless, this type of intervention is relevant for SLPs whose primary goal is to improve language form with the secondary benefit of improved language facilitating performance on other tasks.

Limitations and Future Directions

In general, more research is needed to fully understand the connections between language and executive function in children with DLD (see Kapa & Plante, 2015). Within this broad aim, there are many directions for investigating the specific effects of inner speech on executive function performance. Initially, the field must try to understand whether the effects of inner speech are the same across a wider variety of executive function tasks—this includes both formal assessments and informal daily activities. A worthwhile line of research would be to establish a representative sample of inner speech or verbal mediation behavior in the context of commonly used executive function tasks. These tasks might utilize the dual-task paradigm with articulatory suppression as discussed throughout this tutorial. The results of such standardized, norm-referenced measures would improve our ability to understand how populations with and without language impairments use inner speech on executive function tasks.

Most of what is known about inner speech in children with DLD comes from problem-solving tasks. It is worth noting that problem-solving tasks are notoriously “impure,” meaning that they engage in multiple executive functions to varying degrees (Miyake, Emerson, & Friedman, 2000). This introduces variability across different problem-solving tasks and across participants with different executive function skill profiles, making it difficult to identify the specific function of inner speech during the task. Less is known about how children with DLD use inner speech on tasks of updating, shifting, and inhibition, but it is certainly worth exploring. There is a sufficient depth of theory (e.g., Miyake & Friedman, 2012; Marcovitch & Zelazo, 2009) and evidence from samples of typical adults and children to create informed hypotheses. For example, based on work by Miyake, Kray, and others, one could predict that children with DLD will experience a smaller effect of articulatory suppression during shifting tasks than compared to children with typical development, but this assumption has yet to be tested. We cannot assume that the common executive function is activated in the same way across all tasks.

It is also important to test the theoretical perspectives currently guiding work on inner speech and executive function. Future research should consider multiple theories and structural models of executive function to guide study design and interpretation. The unitary model by Brydges et al. (2012) is particularly applicable to work involving children, given evidence that the structure of executive function changes over time (Huizinga et al., 2006). Until this work is done, we cannot assume that the role of inner speech during executive function tasks is the same in children and adults. We also cannot assume that the structure of executive function is the same in children with and without developmental impairments such as DLD.

If we tentatively accept the conclusion that children with DLD are likely to produce ineffective inner speech across a variety of tasks and contexts, it is worth exploring how to improve their use of verbal mediation strategy. Interventions reported in the literature have been designed to promote the use of inner speech and/or change the content of inner speech. For children with DLD, both approaches will likely be necessary for more effective verbal mediation of executive function tasks. Expanding on the current intervention literature, work is needed involving larger samples of children with and without DLD from both preschool and school-age populations. Effective intervention protocols—the presence of various cue types, prompts to label stimuli or planned actions before responding, or programmatic exposure to self-directed speech—should be replicated across a variety of tasks and task manipulations and in different educational settings.

A final topic that warrants further investigation is the use of measurements that allow us to examine inner speech more directly—measurements that are not dependent on language output or experimental inference. These currently include functional magnetic resonance imaging and positron emission tomography to examine the neural substrates of inner speech (S. R. Jones, 2009; Shuster & Lemieux, 2005) and electromyography to measure neuromuscular activation of the face and vocal tract during inner speech (Lœvenbruck et al., 2018). These methods are highly objective, but they come with a high cost and limited clinical relevance. Although accurate measurement will continue to be a challenge and limitation of work on inner speech and executive function, there are many opportunities for improving our understanding of inner speech assessments and interventions.

Conclusions

In conclusion, verbal mediation of executive function is a dynamic process of using private or inner speech to guide thought and action to accomplish goal-directed behavior. Children with DLD may have difficulty using inner speech and/or impaired executive function skills. These two abilities have yet to be fully dissociated and understood in the research. However, there is promising evidence that supporting inner speech can have a positive impact on executive function performance. Clinically, SLPs are well positioned to both facilitate the use of inner speech and support executive function in children with DLD.