When Will He Talk? An Evidence-Based Tutorial for Measuring Progress Toward Use of Spoken Words in Preverbal Children With Autism Spectrum Disorder

Abstract

Purpose

Professionals face substantial challenges determining whether and when children with autism spectrum disorder (ASD) who are not yet using spoken words will use spoken language as their primary means of communication. This tutorial provides speech-language pathologists with practical guidance on how to measure expressive language predictors for progress monitoring and making intervention decisions for children with ASD who are preverbal.

Method

This tutorial is a repackaging effort that seeks to make the research accessible to clinicians wishing to implement evidence-based practice.

Results

We describe intentional communication, consonant inventory in communication acts, and responding to joint attention as particularly valuable prelinguistic skills to measure. We explain how and when to efficiently assess progress using published assessments periodically and using brief (5-min) communication samples for more frequent progress monitoring.

Conclusions

Communication samples can be used to show how a child performs within a therapeutic setting during teaching (treatment data) and outside of the therapeutic setting (generalization probe data). Both types of data are critical for determining whether the child is exhibiting progress and which aspects of intervention are facilitating progress toward use of spoken words. These recommendations also balance the evidence for best practices for progress monitoring and the demands on clinicians' time and effort. To encourage the measurement of prelinguistic skills of children with ASD who are preverbal in clinical practice, we include (a) example data collection documents, (b) examples with hypothetical data and interpretation, and (c) guidance on communication sampling procedures.

Supplemental Material

https://doi.org/10.23641/asha.13557836

For typical language learners, the milestone of first words is expected around the first birthday (Oller et al., 1999). Typically, by 2 years of age, toddlers have a vocabulary of several hundred words and, between 2 and 3 years of age, children's communication with multiword utterances blossoms. What is perhaps most remarkable is how seamless and effortless learning to talk is for most children. The same is not true for children with developmental disabilities, including autism spectrum disorder (ASD). For children with ASD, first words often do not appear at the anticipated time nor with great ease (Tager-Flusberg et al., 2005). In fact, parents commonly report lack of spoken words as an early concern prior to their child's diagnosis of ASD (Hess & Landa, 2012; Ozonoff et al., 2009).

During the early course of development, children with ASD vary widely in their spoken language skills (Ellis Weismer et al., 2010; Luyster et al., 2008; Pickett et al., 2009). Some, albeit a small percentage, score within age expectations on early language measures (Ellis Weismer et al., 2010). However, many preschool children with ASD are not yet producing any words, and approximately 50%–70% are not using spoken phrases (Ellis Weismer & Kover, 2015; Thurm et al., 2015). Many of those children will learn to use spoken words and phrases. However, approximately 25%–30% of children with ASD are expected to remain minimally verbal and not use spoken language as their primary means of communication (Anderson et al., 2007; National Research Council, 2001; Tager-Flusberg & Kasari, 2013; Tager-Flusberg et al., 2005; Wodka et al., 2013). Age alone is an insufficient predictor for whether a child will talk. Pickett et al. (2009) concluded that, for children with ASD who are not talking at age 5 years, the emergence of spoken words is more likely before age 7 years than after. Yet, some children with ASD reportedly started talking as late as 13 years of age (Pickett et al., 2009; Williams, 1990).

A more robust ability to predict which children with ASD will use spoken words would improve clinical decisions about prioritizing intervention goals, particularly goals related to communication modes. Some children with ASD who are preverbal may benefit most from interventions with a strong emphasis on spoken words. Others may benefit from a strong emphasis on augmentative and alternative communication (AAC). Complicating clinical decisions further, a child's relative benefit from different communication modes may change as his or her skills change. Knowing the degree to which a child is exhibiting the necessary foundational skills for spoken words provides critical information as to whether or not intervention strategies should be adjusted. Examination of prelinguistic skills may signal that a child is progressing toward spoken words and provide encouragement for continuing current intervention efforts. In contrast, if a child is not progressing toward spoken words as expected, the current intervention plan should be reevaluated and adjusted as appropriate (e.g., change intervention strategies or intensity). These questions should be considered on an ongoing and frequent basis. Thus, frequent data collection is needed for making clinical decisions (e.g., whether the strategies, goals, approach, and/or intensity of intervention needs to be altered).

This tutorial applies empirical knowledge about predictors of expressive language growth in children with ASD who are preverbal to evaluating change, including incremental change, in clinical and educational settings. We use the term preverbal to convey that the children are using no or few words at this time and whose treatment goals include acquisition of spoken language. A portion of these children (approximately 25%–30%) are expected to remain minimally verbal based on current evidence. Determining that a child is likely to remain minimally verbal requires examining their prelinguistic skills and progress over time. Thus, the speech-language pathologist (SLP) does not need to know whether a child will use spoken words and phrases to implement the empirically supported assessment approach described in this tutorial. Some children will show progress toward spoken words, and intervention will be tailored to maximize spoken language gains. For other children, careful analysis of their progress toward spoken words, even with changes to intervention strategies, may lead to a reduced emphasis on spoken words and increased emphasis on AAC. In both scenarios, the analysis of key prelinguistic skills when the child is not using spoken words facilitates clinical decisions to improve outcomes.

In this tutorial, we describe the value of measuring prelinguistic skills and then describe three particularly valuable prelinguistic skills for predicting expressive language skills in children with ASD. We then describe how to assess prelinguistic skills periodically and through frequent progress monitoring to capture incremental change. We conclude by explaining how to modify intervention based on assessment data and by describing future research needs.

The Value of Measuring Prelinguistic Skills

Prelinguistic skills emerge prior to the onset of spoken words and include the use of nonverbal means of communication. Prelinguistic behaviors include gestures (e.g., waving, reaching, and pointing), vocalizations (e.g., cooing and canonical babbling), and eye gaze (e.g., looking back and forth between an object and another person and looking at something another person is pointing toward; Watt et al., 2006). Unlike spoken words, prelinguistic forms of communication are not symbolic. Nonetheless, prelinguistic communication is purposeful and conveys meaning. For example, when a child reaches toward a container of bubbles on a high shelf while looking back and forth between the bubbles and an adult, she is likely requesting that the adult retrieve the bubbles for her. As another example, a child lifts his arms while looking at his caregiver in an apparent request to be picked up. The child may add a vocalization, if the initial gesture is ineffective. A child also may extend her arm toward an adult to show a ball or other favorite toy in order for the adult to enjoy that item as she does. Thus, even without spoken words, children direct requests and comments about ongoing activities toward other people.

Importantly, prelinguistic skills predict expressive spoken language skills in children with and without ASD. For example, Watt et al. (2006) reported that joint attention and consonant inventory in the second year of life predicted age 3 years expressive language outcomes for children with typical development. Morales et al. (2000) also found that responding to joint attention accounted for unique variance in later vocabulary development in children with typical development. For children with ASD, a recent meta-analysis revealed a strong association between vocalizations and later expressive language (mean r = .50, 95% CI [0.23, 0.76]; McDaniel et al., 2018). Other prelinguistic skills, such as responding to joint attention (Paul et al., 2008; Thurm et al., 2007; Yoder et al., 2015), intentional communication (Charman et al. 2005; Plumb & Wetherby, 2013; Yoder, 2006; Yoder et al., 2015), and gestures (Luyster et al., 2008), also have predicted later expressive language outcomes in children with ASD. Describing a child's prelinguistic skills provides a rich picture of the child's progress toward spoken words.

Particularly Valuable Prelinguistic Skills for Predicting Expressive Language Skills in Children With ASD

Selecting prelinguistic skill predictors that yield the most useful information increases progress monitoring efficiency and provides data for evidence-based, data-driven intervention planning. Replicated findings support the value of three prelinguistic skills for predicting expressive language growth in preverbal children with ASD: intentional communication, consonant inventory in communication acts (also known as diversity of key consonants used in communication acts), and responding to joint attention (Charman et al., 2005; Paul et al., 2008; Plumb & Wetherby, 2013; Thurm et al., 2007; Wetherby et al., 2007; Woynaroski et al., 2016; Yoder et al., 2015). Measuring these skills improves the prediction of expressive language outcomes above and beyond other tested variables. Thus, investing time in evaluating these three predictors is expected to yield more “bang for your buck” than other predictors.

Yoder et al. (2015) identified these three predictors (as well as parent linguistic responses) as improving the prediction of expressive language skills even when accounting for seven other empirically supported predictors (i.e., receptive language, attention during child-directed speech, motor imitation, nonimitative oral motor functioning, object play, cognitive impairment level, and ASD severity). They evaluated the predictors across 16 months (beginning at a mean age of 34.7 months). Across a longer period of time, Charman et al. (2005) found that the rate of nonverbal intentional communication acts at age 2 years predicted language skills at age 7 years. Similar to Yoder et al. (2015), Wetherby et al. (2007) identified consonant inventory in communication acts as a particularly valuable predictor of expressive language skills in children with ASD at age 3 years. Siller and Sigman (2008) identified responding to joint attention as the strongest predictor of language development in children with ASD. Thus, replicated findings support using intentional communication, consonant inventory in communication acts, and responding to joint attention for monitoring spoken language development in children with ASD who are not yet talking or just beginning to talk.

In the following sections, we first define intentional communication, consonant inventory in communication acts, and responding to joint attention and describe how each of these predictors connects with expressive language. We then explain the development of these skills in children with typical development and children with ASD in Table 1.

Table 1.

Comparison of particularly valuable prelinguistic skills in children with typical development and children with autism spectrum disorder.

Prelinguistic skill	Children with typical development (TD)	Children with autism spectrum disorder (ASD)
Intentional communication	• Emerges around 8–9 months (R. L. Carpenter et al., 1983; M. Carpenter et al., 1998) • Examples include an infant raising her arms to request being picked up or handing a caregiver an object to request help • Use of conventional communication (e.g., pointing and nodding) and symbolic communication (e.g., spoken words and manual signs) emerges around 12 months (Bates et al., 1975; Wetherby & Prizant, 1989) • Combine multiple forms of communication to increase effectiveness (e.g., add vocalizations to pointing when mother is not attending to child's object of interest; Wu & Gros-Louis, 2014)	• May use forms of communication not observed in children with TD (e.g., another person's hand as a tool; Lord, 1995; Stone et al., 1997) • Lower rates of intentional communication (Wetherby et al., 2007) and smaller inventory of conventional gestures than children with TD and children with other developmental disabilities (Charman et al., 2003; Watson et al., 2013) • Rate of nonverbal communication acts for 2-year-old children with ASD predicted outcomes in multiple domains (e.g., social and communication skills) 5 years later (Charman et al., 2005)
Consonant inventory in communication acts	• Initially produce reflective, primitive sounds based on general state (e.g., comfortable, hungry, or tired)and not intentional (Oller, 2000; Oller et al., 1999) • Vocalizations become intentional and more complex in first year of life (Oller, 2000; Oller et al., 1999) • Produce precanonical babbling (e.g., marginal babbling, squeals, and full vowels) around 3–8 months (Nathani et al., 2006; Oller, 2000; Oller et al., 1999) • Produce canonical syllables (i.e., consonant–vowelssyllables with adultlike transitions between the consonant and vowel, such as “buh” and “dah”) around 5–10 months • Use canonical syllables within intentional communication, which emerges around 8–9 months (e.g., “buh-buh” while reaching for a toy and looking toward an adult or using “dadada” to gain an adult's attention)	• Smaller consonant inventories than children without ASD (Landa et al., 2013; Wetherby et al., 2007) • Produce more nonspeech and atypical vocalizations (e.g., high-pitched squeals and low-pitched growls or glottal sounds), which tend to include few or no consonants, than children without ASD (Landa et al., 2013; Plumb & Wetherby, 2013; Sheinkopf et al., 2000; Wetherby et al., 2007)
Responding to joint attention	• Follow an adult's point under certain conditions around 9 months (M. Carpenter et al., 1998; Lempers, 1979) • Reliably respond to bids for joint attention around 12–15 months (Bakeman & Adamson, 1984; M. Carpenter et al., 1998) • Ability to follow another individual's point usually precedes the ability to follow a person's gaze to an object (M. Carpenter et al., 1998; Lempers, 1979)	• Decreased responsivity to bids for joint attention compared with children with TD (M. Carpenter et al., 1998, 2002; Chawarska et al., 2012; Wetherby, 2006) • Difficulty following adults' points, other gestures, and eye gaze (Baron-Cohen, 1989; Baron-Cohen et al., 1997; M. Carpenter et al., 2002) • As a group, children with ASD followed the same pattern of joint attention as children with TD (e.g., sharing attention, then following attention, and then directing others' attention), but 25% of them followed a different pattern (e.g., directing others' attention before following bids for joint attention; M. Carpenter et al., 2002) • Some children with ASD appear to follow gaze later than expected relative to other joint attention skills (Paparella et al., 2011)

Intentional Communication

Intentional communication can be defined as a child's use of gestures, vocalizations, gaze, and/or spoken words to convey meaning to another person (Wetherby et al., 1988). Gestures used by young children who are not yet talking or just beginning to talk can be classified as contact gestures (e.g., reaching, showing, and giving objects), distal gestures (e.g., raising arms to be picked up and pointing at something at a distance to request or direct another person's attention), and representational gestures (e.g., waving, nodding and shaking head, and blowing a kiss). When a child displays intentional communication, which includes attention to another individual and an object or event, these behaviors or groups of behaviors are referred to as communication acts. Communication acts may include multiple forms of communication, such as pointing to an object while looking back and forth between an object and a person or vocalizing while giving an object in apparent request for help.

Intentional communication is important to capture because children with ASD who frequently engage in intentional communication acts to express their wants, needs, and thoughts and to direct others' attention show an underlying desire to communicate (Bruner, 1974). Caregivers may respond to their children's prelinguistic intentional communication acts with linguistic input that supports the child's understanding of spoken words, a necessary skill for expressive language use (McDaniel et al., 2017; Yoder, 2006). One example of such linguistic input is linguistic mapping, wherein the adult describes the child's action and/or underlying message. For example, a child points to a ball and says, “Uh,” and then the caregiver says, “That's a ball,” as a linguistic map (Yoder & Warren, 2002). Children who frequently exhibit intentional communication acts may elicit more frequent and/or more complex linguistic input from caregivers that supports language development (Gros-Louis et al., 2006). They also may engage in more interactions that facilitate object knowledge. Such object knowledge is one cognitive basis for learning early words (McDaniel et al., 2017).

Consonant Inventory in Communication Acts

Consonant inventory in communication acts (the number of different consonants used communicatively) is a measure of phonological complexity. Use of more consonants within communication acts indicates more developed phonological skills than a relatively restricted consonant inventory. Consonant inventory in communication acts also includes a communicative component because only consonants within communication acts are counted. Thus, consonants produced in vocalizations that do not appear to serve a communicative function (e.g., repetitive or self-stimulating behaviors) are not counted. Children who use a wide variety of consonants during communication acts may be attempting to say words they understand prior to those words being understood by listeners (Woynaroski et al., 2016). Those children who use a wide variety of consonants during communication acts also may be exhibiting oral motor skills shared with spoken language (McDaniel et al., 2017).

Responding to Joint Attention

Responding to joint attention requires a child to shift his or her gaze from a current focus to the referent of the adult's bid for attention (Yoder et al., 2015). Adults' bids can include pointing, spoken words, and/or eye gaze (Bottema-Beutel, 2016; Yoder et al., 2015). For example, a child may be looking down at a book when his father points at a bird outside the window while saying, “Look!” The child then lifts his head and shifts his gaze to his father and follows his father's point to look at the bird. In contrast to initiating joint attention, responding to joint attention can only happen if and when a communication partner provides a bid to the child. Thus, this skill necessitates that the child is monitoring another person's communication and that monitoring leads to a response. When children consistently respond to joint attention, they may attend to and benefit more from language input from adults than when they do not respond to joint attention or respond inconsistently. For example, children with ASD who tend to respond consistently to bids for joint attention may benefit more from linguistic mapping, follow-in comments, and expansions compared with children who less consistently respond to bids for joint attention (McDaniel et al., 2017; Siller & Sigman, 2008). As a result, children with ASD with greater joint attention skills may also develop greater expressive language skills compared with children with ASD with lower joint attention skills.

How to Assess Prelinguistic Skills

Monitoring progress is a critical component of providing evidence-based intervention. SLPs must determine the degree to which a child is responding to an intervention to determine whether and how services need to be adjusted. Data collection for clinical decisions needs to be proactive and frequent. Assessment does not just occur during the child's first appointment or during his or her eligibility evaluation. Furthermore, the type of assessment data collected during therapy sessions for progress monitoring differs from that collected during an initial or annual evaluation to determine eligibility and general areas of strength and need. The selected published assessments for evaluating prelinguistic skills described below are more appropriate for initial or annual assessments than progress monitoring due to the time required to administer and score. For progress monitoring, we recommend using brief, frequent communication samples during therapy sessions to provide data needed for clinical decisions. Such progress monitoring helps SLPs determine the degree to which a child is responding to an intervention by assessing performance within and outside of a therapeutic context (Olswang & Bain, 1994).

Selected Published Assessments for Evaluating Prelinguistic Skills Periodically

Several published assessments use specific tasks or activities to assess intentional communication, consonant inventory in communication acts, and/or responding to joint attention. See Table 2. These assessments look very different than traditional speech-language assessments that use books of picture stimuli and ask children to respond to questions and directions. Published assessments that focus on early communication skills provide a framework for observing skills that may be particularly helpful to professionals with less experience assessing early communication skills. The framework helps SLPs identify salient features of early communication that may be underappreciated or hard to notice without training.

Table 2.

Selected published communication assessments for children with ASD who are preverbal.

Assessment	Description	Developmental level	Skill area(s) assessed
Communication and Symbolic Behavior Scales Developmental Profile Behavior Sample (Wetherby & Prizant, 2002)	• Norm-referenced assessment of communicative competence • The 30-min behavior sample includes communication temptations, book sharing, symbolic and constructive play probes, and a language comprehension probe	6–24 months	• Twenty types of behaviors are scored including rates of intentional communication, consonants used in communication acts, and responding to joint attention • Social, speech, and symbolic composite scores are provided
Communication Complexity Scale (Brady et al., 2012)	• A scripted assessment with a series of 12 play-based activities • 12-point scale that describes the early expressive communication skills of individuals who primarily communicate with presymbolic or early symbolic means • Approximately 20–30 min to administer	Birth to word combinations (words, signs, and/or augmentative and alternative communication [AAC] symbols)	• Scale encompasses preintentional (e.g., environmental awareness and attending to object or person), intentional nonsymbolic (e.g., gestures and vocalizations), and beginning symbolic communication (e.g., words, signs, or AAC symbols)
Communication Matrix (Rowland, 2004)	• An expressive communication measure designed for individuals with complex communication needs (e.g., difficulty producing natural speech) • Based on a series of questions and observations, a graphical display of the individual's communication skills is generated • Approximately 10–60 min to complete	Birth to 24 months	• Inventories how an individual communicates for four functions (i.e., refuse, obtain, social interaction, and information exchange) and at seven stages of communication development, including preintentional, intentional, presymbolic, and symbolic communication
Early Social Communication Scales (Mundy et al., 2003)	• A 15- to 25-min structured observation designed to measure nonverbal communication skills • The examiner presents nine tasks, such as object spectacle, social interaction, gaze following, and turn-taking tasks	8–30 months	• Focuses on the child's ability to initiate and respond to three types of behavior: joint attention, behavioral requests, and social interaction

Several limitations of these published assessments should be acknowledged. Most notably, they require substantial time and expertise to score and interpret. The administration times are comparable to or shorter than common language assessments, but scoring video-recorded sessions for specific behaviors can be quite time-consuming. Therefore, even though some of the published assessments can capture subtle changes in prelinguistic skills, they were not designed to be administered frequently (e.g., weekly). Because some of these published assessments are not readily available in a user-friendly form (e.g., Early Social Communication Scales; Mundy et al., 2003), SLPs may have to devote substantial time to understanding the administration procedures and gathering all of the supplies. Despite these limitations, the published assessments can yield useful information when used with children displaying early communication skills. Additionally, they can provide a framework for communication samples. The manuals often include helpful descriptions of key behaviors, such as definitions and examples of joint attention and intentional communication.

Frequent Progress Monitoring of Prelinguistic Skills to Capture Incremental Change

Progress monitoring measures must be relatively brief in time to administer and score so that they can be administered repeatedly across short time intervals. In addition, progress monitoring measures must capture change at relatively small increments (not just large changes) to document change over these short time intervals. It is essential to capture not only the achievement of a goal (i.e., large change) but also any incremental changes so as to inform intervention planning on a week-to-week basis. For example, let us assume a long-term goal of spoken words for a child with ASD who is 3 years old and not yet talking. Intervention focused on spoken language may yield many changes that are evidence of moving toward the goal long before the goal of spoken words, or even the subgoal of first words, is actually achieved. A progress monitoring measure that captures change only at the level of number of spoken words will not be adequate to individualize intervention.

Selecting and administering appropriate progress monitoring assessments can be difficult for an SLP working with young children with ASD. Children with ASD may have difficulty understanding assessment tasks, and social impairments may interfere with test performance (Barokova & Tager-Flusberg, 2018; Tager-Flusberg, 2000). Furthermore, methods for progress monitoring prelinguistic skills are fewer than for progress monitoring skills beyond the preverbal stage (Tager-Flusberg et al., 2009).

One progress monitoring tool that is designed for the prelinguistic stage is the Early Communication Indicator (ECI; Greenwood et al., 2019). It is one of the Individual Growth and Development Indicators developed to assess progress in children birth to 3 years of age (Greenwood et al., 2019, 2011, 2006; Luze et al., 2001; Priest et al., 2001). The ECI assesses total communication, gestures, vocalizations, single words, and multiple words during a 6-min play-based assessment with a standardized set of toys (i.e., house or farm set). The ECI could be used to measure intentional communication. However, it is not designed to be scored in sufficient detail to collect data on the consonant inventory in communication acts and does not include specific opportunities for responding to joint attention. Thus, we describe how to use data from therapeutic activities and communication sampling to collect the necessary data for evidence-based clinical decisions in place of or in addition to published assessments.

We apply the framework described by Olswang and Bain (1994) to discuss the details of data collection for making intervention decisions (i.e., progress monitoring) for children with ASD who are preverbal. It often is very helpful to know how a child performs within a therapeutic setting when teaching is occurring (i.e., treatment data) as well as outside of the therapeutic setting (i.e., generalization probe data; Olswang & Bain, 1994). Seeing progress within a treatment session provides encouraging data; however, ultimately, the child needs to be able to use those skills in everyday environments with a variety of communication partners. We consider (a) what to measure, (b) how to measure, and (c) when to measure to make intervention decisions for children with ASD who are preverbal.

What to Measure

For what to measure, we focus on intentional communication, consonant inventory in communication acts, and responding to joint attention. As we have already discussed, these variables are particularly predictive of expressive language skills for children with ASD who are not yet talking.

Supplemental Material S1 includes blank example data collection documents for monitoring intentional communication, consonant inventory in communication acts, and responding to joint attention in communication samples. These documents can be adapted by SLPs for specific children. They do not need to be used in their current form. Instead, SLPs are encouraged to modify the number of target trials, communication sample context, format, level of prompting provided, and other features to meet the progress monitoring needs of the children they serve. The details of the progress monitoring procedures also should continue to evolve as the evidence base for intervention for children with ASD develops.

Intentional communication. Intentional communication may be quantified in different ways, including the total number of intentional communication acts or the number of communication acts with a certain feature (e.g., total types of functions and forms) within a specified period of time (e.g., 5 min). Examining a child's repertoire of functions and forms may highlight specific areas of need for intervention. Because intentional communication includes prelinguistic as well as linguistic communication acts, measures of intentional communication are appropriate for children at the prelinguistic level and those beginning to use linguistic forms, such as spoken words or signs. SLPs can weight higher level forms of intentional communication with more points to show growth over time. For example, children receive 1 point for nonsymbolic intentional communication acts, 2 points for communication acts with a single spoken (or signed) word, and 3 points for communication acts with spoken (or signed) phrases for the weighted frequency of intentional communication (also known as weighted triadic communication; Greenwood et al., 2006; Yoder et al., 2009). The example data collection form for intentional communication in Supplemental Material S1 divides the observed communication acts into nonsymbolic, single spoken words/signs, and spoken/signed phrases to permit differential weighting, if desired. Alternatively, one could focus on nonsymbolic communication with separate boxes for acts that use vocal communication, gestural communication, or both. Such a system would specify how a child is using nonsymbolic communication and whether over time he or she is more frequently combining modes (i.e., vocalization plus gesture) to communicate more effectively.

For children who are not producing intentional communication acts, consider marking intentional behaviors that they demonstrate, which an adult can interpret. Intentional behaviors are actions that the child does on purpose but that are not yet directed toward another individual. For example, the child may smile and/or vocalize when he or she sees a favorite toy character without looking between the character and a person. Thus, the child produces an intentional behavior(s) that suggest he or she would like the character but does not direct the behaviors to another person. Even though the act is not intentional communication, it includes several skills foundational to intentional communication (e.g., apparent interest in object/activity, vocalizing, and displaying appropriate facial expressions). As another example, the child may push away a food in apparent protest while only looking at the food. The child signals that he or she does not want the food but does not direct that message toward a communication partner. In contrast, when the child hands food back to his communication partner, that action would be classified as intentional communication. Intentional behaviors, such as those described in these examples, may become the child's means of intentional communication especially with targeted intervention. Monitoring the frequency and diversity of intentional behavior can indicate forms and/or functions of intentional communication most likely to develop with continued support.

Consonant inventory in communication acts. The diversity of key consonants used in communication acts is an empirically supported way to measure consonant inventory in communication acts (Wetherby et al., 2007; Woynaroski et al., 2016; Yoder et al., 2015). To calculate the diversity of key consonants used in communication acts, count the number of specific “key” consonants (i.e., /m/, /n/, /b/ or /p/, /d/ or /t/, /g/ or /k/, /j/, /w/, /l/, /s/, and /ʃ/) that the child uses communicatively. These consonants were identified as “key” because most appear relatively early in typical development and listeners can reliably identify them. Because voiced–voiceless cognates (i.e., /b/ vs. /p/, /d/ vs. /t/, and /g/ vs. /k/) are hard to distinguish reliably, children receive only 1 point for each pair regardless of whether they use one or both of the consonants. Thus, the maximum score is 10. The consonants may be produced in nonword vocalizations or spoken words, as long they are in communication acts. Whether the consonant was used in a communication act is the focus of the analysis, not whether it was produced in a word (McDaniel et al., 2019). Consonants produced outside of communication acts are not counted. An SLP can use a form similar to the blank example data collection form for consonant inventory in communication acts in Supplemental Material S1 to circle the consonants observed within a communication sample.

For children who are not producing consonants, recording the types of sounds the child is producing (in addition to continuing to listen for consonants) and the degree to which he or she is using sounds communicatively may be informative. Progress may be seen first in a reduction of atypical vocalizations (e.g., high-pitch squeals and low-pitch growls) or an increase in using vowels within communication acts. This approach to documenting progress focuses on what the child is doing, rather than what he or she is not doing. Observing what the child is doing enables the SLP to make week-to-week decisions about whether and how the focus and/or approach of intervention needs to change in light of the incremental progress and long-term goals. For example, a child may exhibit increasing use of vowels, but not consonants, within communication acts. A logical focus for intervention would be building from that communicative intent and providing many opportunities for the child to use consonants within meaningful contexts to elicit communicative use of consonants. Possible strategies include recasts, follow-in comments, and linguistic mapping from the adult to provide rich linguistic input.

Responding to joint attention. Children must be provided with opportunities to demonstrate their responses to bids for joint attention because responding to joint attention inherently follows another individual's action. Sufficient opportunities are required for stable estimates. The examiner also must attend to the type(s) of cues he or she provided (e.g., pointing toward versus looking toward the object of interest). The example data collection form for responding to joint attention in Supplemental Material S1 provides spaces to note the type of joint attention bids provided, accuracy for each trial, and overall accuracy for the session.

For children who currently do not respond to bids for joint attention independently, documenting the type and amount of support necessary to elicit a response can capture progress even though the child is not demonstrating the skill independently. SLPs can adjust the level of support to identify the child's current skill level and then increase difficulty over time. For example, saying “Look!” while looking toward and pointing to a picture on the wall provides more support than saying, “Wow!” when looking at an exciting toy because the first includes a pointing gesture.

How to Measure

Communication samples address multiple weaknesses of published assessments (e.g., long administration and/or scoring time and limited flexibility). They provide an excellent opportunity to collect treatment data and generalization probe data regarding prelinguistic skills of interest in clinical and educational settings (Barokova & Tager-Flusberg, 2018; Kasari et al., 2013; Tager-Flusberg et al., 2009). Intentional communication, consonant inventory in communication acts, and responding to joint attention can be (but do not have to be) measured within a single communication sample. We use the term communication sampling rather than language sampling to emphasize the assessment of prelinguistic skills rather than spoken words. Nonetheless, most language sampling principles and recommendations apply. Communication sampling typically involves video-recording a child in a context that offers opportunities to communicate and identifying and analyzing specific behaviors by reviewing the video recording. If only a limited number of behaviors are of interest, live communication sampling may be possible. However, recording samples is preferred to permit multiple viewings of subtle cues or details, which may not have been of interest initially and to share with other team members. In this section, we describe how communication samples can be used to collect generalization probe data as well as treatment data.

For generalization probe data, we are interested in how the child performs outside of a therapeutic context. When using communication samples for generalization probe data, it is advantageous to keep many of the features of the sample constant across repeated measures. For example, for each sample, use the same communication partner, same room, same length (e.g., 5 min), and the same set of toys for the communication samples, ones that are not used in therapy. This consistency not only reduces the number of factors that may influence the child's performance day to day but also makes the sampling procedure more routine and less cognitively demanding for the SLP. The generalization probe data can be gathered from the treating SLP interacting with the child without using all of the supports that are typical in treatment. The SLP sets up the environment to support engagement and communication. Engagement is supported by following the child's lead and joining in play at the child's demonstrated level of play. Communication is supported by talking about topics related to the child's focus of attention, monitoring utterance length and complexity (while maintaining grammatical utterances), and avoiding directives (except when needed for safety). This interaction style minimizes modeling and prompting of communication and language, which differentiates it from interaction styles often used in therapeutic activities. Consequently, this communication sample is likely to resemble those used during evaluation sessions more than a typical therapy session.

The sample data collection pages (see Supplemental Material S1) include spaces to record the context of the sample (e.g., activity, communication partner, and setting). If an SLP uses the same interaction style for all generalization probe communication samples, a shorthand phrase, such as “communication sampling procedure,” can be used to describe the support. See the intentional communication page in the Appendix for an example. Generalization data also can be collected during communication samples with a peer or teacher who is unlikely to use therapeutic strategies during the sample. This communication sample could take place in the therapy room, classroom, cafeteria, playground, or other common location. The data collection documents in Supplemental Material S1 can still be used. They include spaces to write the context of the sample.

For treatment data, we are interested in how the child performs with therapeutic support. Thus, the communication sampling context is the therapeutic context rather than using the interaction style described for collecting generalization probe data. For example, an SLP may provide opportunities for the child to request and comment while building Mr. Potato Head to facilitate intentional communication acts. The SLP can record the number of intentional communication acts observed during the sample, which includes support from the therapeutic strategies (e.g., withholding objects, offering choices, follow-in comments, and linguistic mapping). See the Appendix for an example. The example document divides the intentional communication acts into nonsymbolic acts, single words, and phrases to capture key changes in the onset of spoken words. The use of nonsymbolic communication acts is expected to precede spoken words and to increase across the prelinguistic period. We also record the activity and supports on the document to provide context for interpreting the results.

Similarly, SLPs can use the example documents for consonant inventory in communication acts and responding to joint attention. They can record the data live during a therapeutic activity (which is the communication sample) or watch the brief video recording after the session. If a recording is used, it would be possible to document multiple prelinguistic skills (i.e., intentional communication, consonant inventory in communication acts, and responding to joint attention) from a single sample. SLPs can weigh these options based on the degree to which it is possible to record data live, the logistics of recording sessions, and the frequency with which data need to be collected, which we consider in the next section.

When to Measure

For when to measure treatment data, we recommend collecting some data during every treatment session. Such data will not necessarily cover every goal in every session. Nonetheless, the data should be related to the child's long-term goal. For children with ASD who are preverbal, assessing intentional communication, consonant inventory in communication acts, or responding to joint attention would all relate to the long-term goal of use of spoken words. For example, if the SLP sees a child 2 times per week, the SLP might collect treatment data for each of two of the three subgoals during one activity per session. That plan would permit treatment data from one or two sessions per week to inform planning for the next week.

For generalization probe data, we recommend starting by collecting weekly 5-min communication samples with children with ASD who are preverbal. This frequency and length balance the available empirical evidence and the current demands on clinicians' time and effort. For many variables, we do not have empirical support for the exact amount of time needed to collect a stable estimate that results in individuals being similarly ranked across testing occasions separated by short periods of time (e.g., 2 or 4 weeks). In contrast, an unstable measure fluctuates between administrations and is influenced by extraneous factors beyond the skill being assessed (e.g., fatigue, attention, and social engagement). Samples increase in stability as they increase in length and as the child's level of skills increases (McDaniel, in press; Sandbank & Yoder, 2014; Yoder et al., 2018). When a child is just beginning to show a new skill, more variability is observed. Relevant to assessing prelinguistic skills, we found that consonant inventory in communication acts was sufficiently stable with only two 6-min ECI (Greenwood et al., 2019, 2011) samples for young children with ASD (13–30 months of age; McDaniel, Yoder, et al., 2019). Twelve months later, only one 6-min ECI was needed for this same sample of children with ASD.

Based on the available empirical evidence and known demands on SLPs' time, we anticipate that this level of treatment data (i.e., each session) and generalization probe data (i.e., weekly) will be sensitive to change but not overly burdensome for collection and interpretation. The data are likely to provide the SLP with sufficient information to readily describe the child's progress and current skills, including the information needed to update annual Individualized Family Service Plans (before age 3;0 [years;months]) or Individualized Education Programs (after age 3;0). For students receiving services from a third-party payer, assessments may be required to demonstrate the need for continued services, in addition to the treatment and generalization probe data. The published assessments described in the prior section and Table 2 may fulfill that need. Nonetheless, some children are going to require more frequent and/or longer samples for adequate stability to show incremental change. In such cases, consider increasing the length of the sample or combining samples from multiple sessions in an effort to reduce variability from one observation to the next (e.g., Wetherby et al., 1988; Yoder et al., 2018). For example, consider collecting a 5-min sample 3 times per week, adding the results together, and examining the results on a week-to-week basis rather than session-by-session basis to make patterns and progress more salient. This viewpoint is analogous to examining the stock market over a period of weeks or months rather than day to day. Also consider confirming that the features of the sample context are as consistent as possible to minimize other potentially influential factors (e.g., communication partner, activity/set of toys, time of day, and level of fatigue). For example, the SLP may discover that a child shows greater success in the morning relative to the afternoon. Collecting the sample at the same time of day would be expected to reduce the observed variability. Another approach to gathering more data would be to ask the child's caregivers or teachers to provide input. The SLP could ask the caregivers or teachers about how the child is communicating with them or using different consonant sounds, for example.

Modifying Intervention Based on Assessment Data

The progress monitoring data enables SLPs to evaluate week-to-week progress to adjust intervention strategies as needed to maximize outcomes. By keeping track of the child's level of success with different levels of support, the SLP can interpret which supports are helpful, which are ineffective, and which can be faded because the child exhibits the skills in at least some contexts without support.

For data interpretation and intervention planning, it is important to distinguish between predictors of expressive language growth and malleable factors in children with ASD who are preverbal. Being a predictor of expressive language does not in and of itself indicate that the predictor should be targeted. For example, chronological age often correlates with expressive language; however, we cannot target the child's chronological age as an intervention target. Nonetheless, if a predictor is a malleable factor with theoretical and empirical support that its relation with expressive language is or could be causal, it can be considered for an intervention target. Recognizing the underlying rationale for why these predictors might affect language growth and the expected patterns of development in children with typical development and children with ASD can guide intervention planning decisions. The three predictors we describe in this article differ in the state of the evidence. Intentional communication presents with more supportive evidence than consonant inventory in communication acts and responding to joint attention. Nonetheless, the latter two still present with supportive theoretical and empirical evidence. When implementing the progress monitoring procedures described above, the SLP will be able to determine the degree to which these variables are changing and make informed clinical decisions regarding changes in intervention approaches and focuses.

Intentional Communication

A number of intervention strategies including, but not limited to, Hanen More Than Words (Carter et al., 2011), Parent-Mediated Communication-Focused Treatment in Children With Autism (Green et al., 2010), Picture Exchange Communication System (Yoder & Stone, 2006), Pivotal Response Treatment (Hardan et al., 2015), and Responsive Education and Prelinguistic Milieu Teaching (Yoder & Stone, 2006) have demonstrated success in improving language skills of children with ASD. Thus, intentional communication can be addressed via multiple avenues. Targeting intentional communication necessarily involves attending to how the child uses a variety of forms of communication. Intentional communication includes nonverbal forms of communication (e.g., gestures, vocalizations, and eye gaze) as well as signs and aided forms of AAC. Children with ASD have demonstrated benefit from AAC for developing intentional communication and other language skills (Ganz et al., 2012; Kasari et al., 2014). An SLP may consider introducing AAC strategies and track the degree to which the child uses those strategies for intentional communication and whether increased use of spoken language is used simultaneously. For example, when introduced to a speech-generating device (SGD), a child may initially show increased requesting with the SGD, followed by increases in vocalizations. This pattern would support continued use of the SGD to augment spoken communication. In contrast, another child who is also introduced to an SGD may initially show increased requesting with the SGD but no increases in vocalizations. This pattern would support increased emphasis on the use of the SGD and potentially other forms of AAC to support intentional communication (Schlosser et al., 2007). In addition to considering the frequency and forms (e.g., use of vocalizations, signs, and AAC) of communication, functions of communication should be considered as well when selecting more specific intervention targets.

Consonant Inventory in Communication Acts

Consonant inventory in communication acts helps maintain the focus of intervention on the communicative use of consonants and the complexity of vocal communication rather than on nonspeech oral motor skills. Nonspeech oral motor exercises should be avoided due to a lack of supportive evidence, even if used in an attempt to expand a child's consonant inventory (McCauley et al., 2009). Although limited information is known about the specific means through which children with ASD expand their consonant inventory in communication acts, intentional communication and parent linguistic responses may facilitate its growth through receptive vocabulary (Woynaroski et al., 2016). Increases in intentional communication and parent linguistic responses may first result in receptive vocabulary growth and then expansion of the child's consonant inventory in communication acts. Theoretically, once children understand words, they begin to try to say those words in communication acts, even if they are not identifiable words to the listener (Woynaroski et al., 2016). Additional evidence shows that expanded consonant use is at least partially responsible for the increase in intentional communication leading to increased expressive language skills (McDaniel et al., 2017). Thus, seeing change in consonant use in communication acts is encouraging for progress toward use of spoken words. However, we do not yet know the degree to which directly targeting consonant use improves expressive language skills in children with ASD who are preverbal. If progress is not observed in the child's use of consonants in communication acts despite data-driven adjustments to the intervention approach, the SLP may consider shifting toward initiating or emphasizing AAC strategies (e.g., SGD, picture symbols, or signs) to support expressive language skills. The child's consonant inventory in communication acts should still be monitored across that time to know whether a shift toward spoken language is warranted as the child develops further.

Responding to Joint Attention

To document incremental progress in responding to joint attention, careful observation of the cues provided (e.g., proximal point, distal point, or eye gaze with or without a verbalization such as “Look!”) and the child's response (e.g., looking to speaker's hand or looking to the object of interest) is needed. Because the cues from the communicative partner can dramatically change the communicative difficulty level, the specific cues provided should be noted during intervention. Improvements in responding to joint attention following developmental and behavioral interventions have been reported (Kasari et al., 2006, 2010; White et al., 2011). For example, Kasari et al. (2006) reported that preschool children with ASD exhibited greater improvement in responding to joint attention after a joint attention intervention that combined developmental and applied behavioral analysis strategies than did children randomly assigned to the symbolic play group or the control group.

Future Research Needs

Continued research is needed to predict more accurately the communication and language outcomes of children with ASD who are preverbal, to develop and refine valid and reliable measures of prelinguistic and early language skills in children with ASD for use in clinical practice, and to develop interventions that maximize spoken language outcomes for children with ASD. Measures for clinical practice must consider feasibility including the amount of training, administration time, and scoring time required. These measures may use a combination of direct observation, caregiver and/or teacher report, and automated (computer) analyses for vocal analysis for example (LENA Research Foundation, 2015). Related to intervention development, more evidence about the rate and type of progress expected during specific interventions will improve SLPs' and related professionals' abilities to determine whether an individual child is making expected progress or whether his or her progress is slower than expected and intervention changes need to be considered. As mentioned above, we also need additional evidence of the degree to which interventions directly targeting consonant use improve expressive language skills in children with ASD who are preverbal. Achieving these goals could improve dramatically the long-term outcomes of children with ASD who are preverbal and support them and their families across the life span.

Conclusions

In this tutorial, we have discussed the identification, measurement, and application of three particularly strong predictors of expressive language growth in preverbal children with ASD (i.e., intentional communication, consonant inventory in communicative acts, and responding to joint attention). Changing the questions posed for progress monitoring for children with ASD who are preverbal could influence intervention practices positively. By changing the question from “How many words does he use?” to questions about predictors of expressive language growth, such as “Which sounds is he using communicatively?” and “How is she expressing her wants and needs?” the response changes from “Zero,” to descriptions of what the child with ASD can do. Focusing on skills rather than deficits is not only more encouraging but also more informative for intervention. Evaluating predictors of expressive language growth could elucidate small but meaningful changes in communication skills and, when needed, provide the impetus for changing the course of intervention. Observing these meaningful changes or the lack of these meaningful changes is particularly significant for SLPs serving children with ASD who are preverbal and their families. Currently, there is limited guidance for decisions regarding how to measure expressive language predictors for progress monitoring and making intervention decisions for children with ASD who are preverbal. By using the clinical procedures described in this tutorial, SLPs can be more confident in their decisions to continue with a current intervention approach or to change course. These decisions enable children with ASD who are preverbal to receive high-quality, effective intervention services that are tailored to their specific needs and growth over time.

Appendix

Data Collection Documents With Hypothetical Data

Funding Statement

All authors read and approved the final manuscript. This research was supported by a U.S. Department of Education Preparation of Leadership Personnel grant (H325D140087; PI: Schuele) and the National Institute of Child Health and Human Development (U54HD090216; PI: Colombo Kansas Intellectual and Developmental Disabilities Research Center).