Teaching Latino Parents to Support the Multi-Symbol Message Productions of their Children who Require AAC

Abstract

Many children who require augmentative and alternative communication (AAC) have difficulties shifting from the single- to multi-symbol stage of language development. Ensuring that children who use AAC receive evidence-based interventions to address this problem is critical. Furthermore, there has been little AAC treatment efficacy research to date that addresses the needs of multicultural populations, particularly those of Latino children who use AAC and their families. To address these issues, the current investigation used a mixed methodology, which included a focus group to determine if any changes needed to be made to an existing instructional program; and a single subject, multiple probe, across participants design to evaluate the effects of a caregiver instructional program on the multi-symbol utterance productions of Latino children who used AAC. Specifically, a cognitive strategy instructional approach that has been used to teach turn-taking skills to both Caucasian and African American children who use AAC was modified and used to teach caregivers how to support the production of their children's multi-symbol messages. In order to foster emergent literacy skills, caregivers implemented the strategy during book reading activities. Focus group results revealed that changes to the instructional program may be required for some Latino families. With regard to the instructional program, all caregivers successfully learned to use the instructional strategy, and all children increased their use of multi-symbol messages. Research and clinical implications of the findings are discussed.

Keywords:

• Aided AAC Modeling
• Augmentative and Alternative Communication AAC
• Cognitive Strategy
• Intervention
• Latino
• Literacy
• Multicultural
• Storybooks
• Symbol Combinations
• Two-word Utterances

Introduction

Children who require augmentative and alternative communication (AAC) (e.g., communication boards, speech-generating devices) are at high risk for expressive language difficulties (e.g., Lund & Light, 2007). One particularly challenging transition is the shift from the use of single- to multi-symbol utterances; children using AAC tend to rely on single symbol messages to communicate (see Binger & Light, 2008, for a review). As the transition from single- to multi-symbol utterances marks the emergence of syntax, this transition is critical.

Very little research has been conducted to date to address this stage of language development with children who use AAC. Some initial data have been reported in which productions of multi-symbol messages have increased following intervention (e.g., Romski & Sevcik, 1996). However, only two experimentally controlled studies that involved children who used AAC were located in the literature. Binger and Light (2007) used a single subject, multiple probe design across participants to evaluate the impact of using aided AAC modeling (i.e., pointing to and labeling graphic symbols on a child's aided AAC device) to increase the productions of multi-symbol utterances of children who used AAC. Four of the five children in the study achieved this goal. In a second investigation, Nigam, Schlosser, and Lloyd (2006) used a single subject, multiple probe design across targets to investigate the impact of using a matrix strategy and milieu intervention techniques on the action-object combinations of three children who used AAC. Two of the three children successfully learned to use action-object combinations. Although the results of these two investigations were largely positive, one drawback is that in both studies, intervention was administered by a clinician or researcher; none of the caregivers or other typical communication partners received instruction during the investigations.

Caregivers spend far more time with children who use AAC than the speech-language pathologist, and unfortunately, the transactional effects of disability may result in caregiver behaviors that, under typical conditions, are not facilitative of language development. These behaviors include taking the majority of conversational turns, asking predominantly yes/no questions, and failing to recognize the turns of children using AAC (e.g., Light, Binger, & Kelford Smith, 1994). When attempting to provide children who use AAC with opportunities to develop their language skills, it is important to include instructional components for caregivers and other communication partners in order to ensure that children receive the benefit of facilitative communication strategies from those with whom they most frequently interact.

With these issues in mind, a number of researchers have designed instructional programs working with communication partners, including caregivers and educators, to enhance communication for children who use AAC (e.g., Basil, 1992; Basil & Soro-Camats, 1996; Iacono, Chan, & Waring, 1998; Kent-Walsh, 2003; Kent-Walsh, Hasham, & Stewart, 2004). Typically, communication partners have been instructed to use a combination of techniques designed to improve communication skills, such as the use of expectant delays, wh- questions, and aided AAC models, in addition to increasing responsivity to the child's communicative attempts. On the whole, use of these techniques has resulted in positive communication outcomes for the children using AAC. For example, Kent-Walsh and colleagues (Kent-Walsh, 2003; Kent-Walsh et al., 2004) used experimentally controlled single subject research designs to evaluate the impact of teaching caregivers and educational assistants evidence-based techniques designed to increase the turn-taking skills of children who used AAC. The studies conducted by Kent-Walsh and colleagues differed from most other AAC communication partner programs in two main respects. First, Kent-Walsh and colleagues used story reading contexts to teach communication partners to use an interaction strategy, which included a least-to-most cueing hierarchy designed to elicit turns from the children who used AAC. Steps of the hierarchy included the following: (a) read text + provide aided AAC model; (b) ask wh- question + provide aided AAC model; and (c) provide sample answer to wh- question. The mnemonic RAA RAA RAA! (R = Read, A = Ask, A = Answer) was used to facilitate instruction. An expectant delay was provided between each step. In addition, the adults were instructed to respond contingently to each turn the child took. The communication partners provided systematic instruction to the children each time they read a story through the consistent use of the cueing hierarchy.

Second, Kent-Walsh and colleagues used an eight-step cognitive strategy approach for instructing the partners (Kent-Walsh & McNaughton, 2005). Cognitive strategy instruction is an errorless learning approach that has been used frequently by educational researchers; for example, by Jiménez (1997) in literacy instruction with Latino children. In the studies conducted by Kent-Walsh and colleagues (Kent-Walsh, 2003; Kent-Walsh et al., 2004), the researchers followed the eight steps outlined in Kent-Walsh and McNaughton to teach adults to facilitate the turn-taking skills of children who used AAC. Instruction took place over the course of multiple instructional sessions, with the researcher gradually fading prompts and feedback as sessions progressed. The caregivers in Kent-Walsh et al. (2004) learned to use the instructional strategy consistently after only 2.5 h to 3.5 h of instruction. Once the caregivers began using the interaction strategy techniques, the children's rates of communicative turn-taking and range of semantic concepts expressed increased dramatically. To date, however, no one has investigated the impact of caregiver instructional programs on more advanced linguistic skills such as the expression of multi-symbol utterances.

Furthermore, virtually no AAC intervention research has been conducted to investigate the impact of using such programs with caregivers from diverse ethnic or racial backgrounds, with the exception of Kent-Walsh, et al. (2004). However, developing such interventions has been identified as a critical need in communication disorders (American Speech-Language-Hearing Association, 2000). Families from diverse ethnic or cultural backgrounds may relate to mainstream culture in a variety of ways, and it is important for clinicians to ensure that they are not marginalizing beliefs and traditions that are different from mainstream culture (Parette, Huer, & Scherer, 2004). Instead, clinicians must develop an understanding of values from other cultures, recognize how these values may differ from mainstream culture, acknowledge any cultural biases of the intervention, and adapt the intervention to meet the needs of the family (Parette & Brotherson, 2004). There is a particular need for research involving the U.S. Latino population, the fastest growing ethnic group in the US (Battle, 2002).

Thus, the purpose of the study was twofold: (a) to evaluate whether an existing instructional program for caregivers would be valid for Latino parents, and (b) to evaluate the effectiveness of the instructional program on the multi-symbol utterances of three children who required AAC. Specifically, the following research questions were examined: (a) what changes, if any, were required for the proposed instructional program to ensure its appropriateness for Latino populations?, (b) what was the effect of the instructional program on the Latino caregivers' implementation and generalization of the targeted interaction strategy during book reading activities?, and (c) what was the effect of the instructional program on the multi-symbol utterance productions of children who use AAC during book reading activities?

This investigation implemented a mixed methods research design, which included both qualitative and quantitative components (Brinton & Fujiki, 2003). A focus group (Morgan, 1997) was conducted to ensure that the proposed instructional program was appropriate for Latino caregivers. Once the focus group was completed and changes were made to the intervention program, the first author taught Latino caregivers how to use the instructional strategy (i.e., the cueing hierarchy described above). This strategy was designed to facilitate expressive symbol combinations during story reading activities by their children who required AAC.

Focus Group

Method

Research Design

The first component of the investigation was a focus group designed to validate the intervention program that would be evaluated in the quantitative component of the investigation. This step was required because the planned intervention program was developed based on efficacy data with Caucasian and African American families, not Latino families. Focus groups have the advantage of providing direct evidence of the agreements and disagreements of participants' opinions, instead of relying on the post hoc analyses that are required with individual interviews (Morgan, 1997).

Participants

A convenience sample (Patton, 2002) of four experts on Latino culture in Albuquerque, NM was used. Informed consent was obtained for all participants. Two of the participants were professors of education or speech-language pathology, were Latino parents of Latino children, and conducted research focused on Latino populations. The third participant was an AAC expert who had a Ph.D. in speech-language pathology and had extensive experience in working clinically with Latino families in the school context. The fourth participant was a Latino father of a child who used AAC; both father and child participated in the single subject portion of the current investigation.

Procedures

One semi-structured focus group session was conducted by the first author with the four participants at a central location. An interview guide (Patton, 2002), consisting of the themes to be covered and issues to be probed throughout focus group discussion, was developed. This guide focused on the following three topics: (a) the instructional context (i.e., interactive storybook reading), (b) the interaction strategy to be targeted throughout the instructional program (including the component skills of modeling AAC use, using expectant delays, asking wh- questions, modeling appropriate responses, and increasing responsivity), and (c) the instructional components (i.e., video review, role play, live practice, provision of instructor feedback). Although the interview guide helped to frame the discussion during the focus group session, the first author informed the participants that she did not have a strict series of questions to ask and that there was flexibility with the topics to be covered during the conversation. Participants were encouraged to raise any related issues as they arose throughout the course of conversation.

Data Analysis

The video-taped focus group session was transcribed verbatim into written text. A modified version of Boyatzis' (1998) multi-step procedure for inductively developing codes was used to analyze the data. An outline of paraphrased items was prepared, which served as a synopsis of the raw data. This synopsis was then analyzed for occurrence of themes in order to provide some initial organization to the data. Following the identification of the themes, corresponding operational definitions were developed. The final coding themes were used to analyze the entire data sample. Sub-themes were then identified within each coding theme to provide a more detailed description of the issues discussed by the participants.

To address reliability and validity issues, coding reliability and member checks were completed as follows: A graduate student was trained to code the data by applying the operational definitions and then coded 10% of the data. Cohen's kappa (Shrout, Spitzer, & Fleiss, 1987) was used to calculate inter-rater reliability and yielded an excellent reliability value (Kappa = 0.95). To examine the validity of the results, a summary of the focus group findings, including themes and sub-themes, was compiled and presented verbally to the focus group participants (Creswell, 2005; Kvale, 1996), and participants were asked to provide input on the accuracy of the summary in depicting their contributions during the focus group session. All participants indicated that the summary accurately represented their discussions and they each provided encouraging remarks relating to the potential impact of the overall investigation.

Results

Themes and Sub-Themes

The final coding themes and sub-themes were as follows: (a) Research Study (past research, participants, purpose, & procedures), (b) Instructional Program (terminology & instructional activities), (c) Instructional Context (storybooks & home setting), (d) Instructional Strategy (expectant delay, wh- question asking, & modeling AAC use while reading), (e) Cultural Issues (AAC use & culture, diversity across Latinos, & Latino literacy and educational experiences and practices), and (f) Unrelated or Uncodable statements. What follows are brief summaries of the issues arising within sub-themes that had an impact on the instructional program examined in the quantitative component of the investigation. It should be noted that the other themes and sub-themes contained valuable information to validate the instructional program, but they are not described in detail since they did not necessitate instructional changes.

AAC use and culture

Participants discussed the fact that some Latino families may believe that their children communicate functionally at home and may resist having their children use AAC devices in the home, a finding that is supported by Huer and colleagues (Huer, Parette, & Saenz, 2001). To address this issue, one participant suggested that AAC devices should be introduced as something that “will help your child with school”; all participants agreed that this was a good approach for introducing AAC. When discussing the potential cultural reasons for not using AAC, a participant pointed out, “It's not like you're going to say, ‘Well, I can't do this because it's not within their culture.’[AAC]'s not really within anyone's culture, probably”.

Storybooks

Several considerations for selecting and introducing storybooks were discussed. First, participants discussed the fact that many Latinos have more of an “oral tradition” that may not include storybook reading in the home. However, all participants agreed that storybook reading is more common among New Mexico Latinos today and indicated that they read storybooks to their own children. Several participants discussed the importance of selecting books that contained everyday themes, such as going shopping or taking a bath. As one participant said, “I think the type of book would make a big difference …this Little Critter book has a lot of everyday themes in it, as opposed to … a fairy tale that these kids may not be familiar with.” Another suggestion was to use books in which there are repeating characters with whom children can grow to identify. One participant mentioned that the characters in book series become “familial.”

Terminology

The participants cautioned against using the term training when referring to adult instruction, as “it can be a very sensitive word for people … because it can almost be like, ‘I'm gonna give you this training,’ as if you, the parent, need to be trained”. Another participant stated, “… parents don't want to feel like a failure … [some parents might say,]‘Don't tell me my child needs help. I'm doing the best I can’.” Instead, participants suggested using terms such as learning and instruction.

Expectant delays

When using expectant delays, participants recommended focusing on the pause component and not requiring eye contact. The participants agreed that for some Latinos, pauses that are combined with eye contact and an expectant facial expression may be perceived by the child as a form of discipline.

Wh- questions

The participants noted that some Latinos, particularly those who are less acculturated into mainstream values, may find asking wh- questions for which there is a known answer to be “unnatural” and “unnecessary” (e.g., asking “Where is Little Critter?” when he appears on that page of the book). However, the participants carried this discussion further and again agreed amongst themselves that this approach should still be included, given its relevance to AAC and given that it could be addressed within the instructional program by:

Saying [to the parents] … ‘What you do at home on a regular basis is important and valuable, and you should maintain that. But this is what it's like at school, and this is how you can prepare your child for school.’ …Keep them kind of separate, so you're not replacing … what's going on at home.

Another suggestion aimed at assisting parents who are not used to asking wh- questions was to provide them with videos of others asking wh- questions and allowing them time to practice using such questions.

Diversity across Latinos

The participants also stressed the need to remember that the Latino population in central New Mexico is highly diverse. Many families have resided in the area for several hundred years, and others are recent immigrants. Many families are highly acculturated into middle class U.S. culture and values, while other families maintain a more traditional Spanish or Mexican way of life (Rodriguez & Olswang, 2003). All participants agreed that, although the potential instructional program modifications they suggested may be important for some Latino families, they may not be as critical for families that adhere more to values that are typically associated with the middle class, mainstream population.

Changes to the Program

Based on the focus group results and the fact the investigators anticipated diversity in the participant group for the quantitative component of the investigation, the following changes were made to the instructional program: (a) various components of the program were presented in relation to Latino culture and as being beneficial for preparing children for school: using AAC, reading to their children, and asking wh- questions; (b) only books that contained everyday themes were used in the investigation; (c) the term instructional program was used instead of training program in conversation with the participants; (d) parents were not required to maintain eye contact with an expectant facial expression when using pauses as a prompt. Several other suggestions, such as the recommendation that parents view videos of others asking wh- questions and be provided with practice using this prompt, were already built into the instructional model.

Interaction Strategy and Acquisition of Multi-Symbol Messages

Method

Research Design

The second component, which was the major focus of the investigation, was a single subject, multiple probe design across three participants to evaluate the effectiveness of the instructional program. Single subject designs are particularly well suited for AAC interventions, as they allow for the establishment of experimental control with participants who are from heterogeneous populations (McReynolds & Kearns, 1983).

Participants

Three Latino caregivers and three children from a metropolitan area of central New Mexico participated in the study. The caregivers provided informed consent both for themselves and for their children before the investigation began. Selection criteria for caregivers included the following: (a) parents or guardians of children who required AAC; (b) Latino (i.e., self-identify as Latina/o, Hispanic, Mexican, Mexican-American, or Spanish; Rodriguez & Olswang, 2003); (c) no known speech, language or hearing impairments; (d) at least a high school diploma or equivalent; and (e) implementation of the targeted interaction strategy in fewer than 25% of opportunities during book reading interactions with their children at the onset of the investigation. One father and two mothers participated in the study. All parents were born in the United States and spoke English as a first language; none spoke Spanish fluently. All of the parents stayed at home during the day to care for their children. All three were married, owned their own homes, and had spouses who had full-time employment. All three parents reported reading to their children on a daily basis and that they had over 50 child and adult books in their homes.

The children in the investigation met the following criteria, which were based on Bedrosian's (1999) selection criteria for AAC interactive storybook reading research and Binger and Light's (2007) selection criteria for teaching multi-symbol utterances: (a) were between 2;6 and 7;11 years of age, (b) had hearing and vision within or corrected to be within functional limits, (c) had an expressive vocabulary of at least 25 words/symbols, as indicated on the MacArthur Communicative Development Inventories (Fenson et al., 1993), (d) presented with severe, congenital motor speech impairments (i.e., less than 50% comprehensible speech on Dowden's 1997 Index of Augmented Speech Comprehensibility in Children), (e) comprehended early two-word relations with at least 80% accuracy, based on measures from Miller and Paul (1995), (f) were able to listen to stories and answer simple wh- questions based on stories (e.g., Who? What?), and (g) communicated using telegraphic messages (i.e., no more than 10% of communicative turns consisting of two or more aided AAC symbols during a 10 min storybook reading activity). Hearing and vision status were obtained via parent report. The remaining criteria were evaluated by the first author.

Although it would have been preferable for all of the children to have had significant prior exposure to AAC, such children are difficult to locate. A significant number of speech-language pathologists still receive little to no instruction in AAC in their graduate programs (Ratcliff, Koul, & Lloyd, 2008) and may not be providing AAC services for young children who could benefit from AAC. Furthermore, a recent survey revealed that even for those preschoolers who do receive AAC, a very small percentage (i.e., 15% in one state in the U.S.) are provided with access to voice output devices (Binger & Light, 2006). It is therefore not surprising that the participants in the current investigation had minimal prior exposure to AAC. All participants, however, were exposed to the graphic symbols prior to the onset of the baseline phase, and all children identified the symbols included on the communication boards/pages with at least 90% accuracy before the baseline phase began. When a child did not comprehend a symbol, a paired instructional paradigm was used to teach the meaning of that symbol (Schlosser & Lloyd, 1997). All participants had access to their communication devices/boards during the baseline phase. The children's names used in this report are pseudonyms.

Antonio

At the onset of the investigation, Antonio was age 4;1 and had a profound phonological process disorder with no other known cognitive, language, or motor impairments (see Table 1). Antonio's speech sound repertoire consisted of several consonants and vowels. His speech productions were consistent, and he had normal intonation patterns. Antonio attended a preschool classroom for children with disabilities and received individual and group speech services. He lived at home with his mother, father, and older brother. Antonio's older brother had a history of a severe speech and language impairment and was diagnosed with a Pervasive Developmental Disorder (PDD) during the time of the investigation. Antonio did not demonstrate any PDD-like behaviors. Antonio had received an AAC evaluation at school just before the start of the investigation, and the school provided him with a MiniMerc™ toward the end of this investigation. Thus, he had minimal exposure to AAC prior to the study. Antonio used a Mercury™ throughout the investigation. Antonio's father, who participated in the study, was 34 years old, had a high school degree, and had completed approximately nine college credits.

TABLE 1 Participant characteristics including chronological age, sex, primary disability, speech comprehensibility and context on the I-ASCC, TACL-3 Scores, and communication modes expressed.

Participant			I-ASCC		TACL-3 scores
Name	Age, sex	Disability	No context	Semantic context	Stand. score	%ile	Communication modes
Antonio	4;1 M	Profound phonological process disorder	3%	10%	106	65th	Natural speech, signs, gestures, Mercury™ with Functionally Speaking
Angela	3;4 F	Suspected VCFS, profound VPI, suspected CAS	0%	0%	109	73rd	Natural speech, idiosyncratic signs, gestures, low tech communication boards
Julia	2:11 F	Subpalatal cleft, profound VPI	0%	0%	106	65th	Natural speech, idiosyncratic signs, gestures, MightyMo™

Notes. Pseudonyms have been used. CAS = Childhood apraxia of speech. I-ASCC = Index of Augmented Speech Comprehensibility in Children (Dowden, 1997). TACL-3 = Test of Auditory Comprehension of Language, 3rd ed. (Carrow-Woolfolk, 1999). VCFS = Velocardiofacial Syndrome (also known as DiGeorge Syndrome). VPI = velopharygeal insufficiency.

Angela

Angela was age 3;4 and was suspected of having Velocardiofacial Syndrome (VCFS, also known as DiGeorge Syndrome) and childhood apraxia of speech (CAS). She had profound velopharyngeal insufficiency (VPI), characterized by a flaccid palate with limited anterior/posterior mobility, incomplete contact with the pharyngeal wall during speech production, and a pulsating pharyngeal wall. Her parents chose not to have her tested for VCFS and wished to delay pursuing surgical options. Angela's speech was extremely hypernasal and consisted of nasal consonants, /h/, and several nasal vowels. During the investigation, the Verbal Motor Production Assessment for Children (VM-PAC) was administered (Hayden & Square, 1999). Angela's scores for four of the five subtests (Global Motor Control, Focal Oromotor Control, Sequencing, and Speech Characteristics) fell at or below the 5th percentile, indicating significant difficulties with a wide range of oral-motor skills. Her intonation patterns were relatively intact. Angela had no other known disabilities. She did not attend preschool and spent her days at home with her mother. She lived with her mother, father, and two older sisters. Angela did not have any exposure to AAC prior to the onset of the investigation. Her parents did not wish to pursue her using a voice output communication device but were willing for her to try using communication boards; thus, Angela used low-tech communication boards for the study. Angela's mother was 44 years old, had a bachelor's degree, and had completed approximately 12 credits of Master's level coursework.

Julia

Julia was aged 2;11 when the investigation began. She was diagnosed with a subpalatal cleft. She had no other known disabilities. She had profound VPI and her speech was extremely hypernasal; her speech sound repertoire consisted of several nasal consonants and vowels. Her intonation patterns appeared to be normal. Julia's parents were gathering information about surgical options to repair Julia's cleft during the investigation but were not prepared to schedule a date for surgery. At the time of the investigation, Julia spent her days at home with her mother. She lived with her parents and older brother. Julia did not have any exposure to AAC prior to the study. Julia used a MightyMo™ to communicate during the investigation. Julia's mother was 26 years old, had a high school degree, and was in the process of completing her first few college level courses.

Materials and Instrumentation

A modified version of the guidelines developed by Kent-Walsh and colleagues (Kent-Walsh, 2003) was used for selecting books for the investigation. The books selected (a) had illustrations; (b) incorporated text and storylines that were appropriate to each child's receptive language level, cultural background, and interests; and (c) included at least six double-page spreads (i.e., 12 pages). Two book series were selected: Mercer Mayer's Little Critter books and Dora the Explorer books (various authors). Culturally and developmentally appropriate symbols were added to each child's aided AAC system. Each symbol represented a single concept (e.g., “Little Critter,” “throw,” “ball”). To ensure access to appropriate vocabulary, separate “pages” were created for each book and were organized using Fitzgerald keys (McDonald & Schultz, 1973); that is, symbols were organized from left to right, following typical word order patterns (agents, actions, descriptors, objects, etc.). Vocabulary items corresponded with the major agents, actions, descriptors, and objects of the narrative and illustrations for each book, in addition to symbols representing the concepts “who,” “what,” and “where.” Each page contained 30–35 symbols (all are available at www.cathybinger.com). The same pages were used for all three children, although the devices were different.

For each child, each set of books was randomly designated as Set A books, to be used for the intervention phase, or as Set B books, to be used for the generalization phase described below.

Dependent Measures

The dependent measures were: (a) accuracy of caregivers' implementation of the targeted strategy in obligatory contexts (i.e., at least once on each double-page spread in each book), and (b) frequency of children's multi-symbol utterance productions (including both spontaneous and imitated messages) within a 10 min book reading activity. In addition, for the child data, the number of different multi-symbol messages (e.g., DORA EAT versus DORA PLAY) and the number of spontaneous messages within each session were recorded. Spontaneous messages were defined as messages for which the parent did not provide an aided AAC model during any prior turn within the cueing hierarchy that the parent had been implementing (i.e., within the Read, Ask, or Answer step; see below for details) or during the parent's prior turn (if the parent was not using the cueing hierarchy at the time).

Procedures

Baseline phase

Prior to intervention, baseline measures were taken to establish current levels of performance on the dependent measures. Separate baseline measures were taken for the instructional context (i.e., Set A books) and generalization context (i.e., Set B books). The appropriate communication displays (i.e., the “pages” designed to accompany each book) were given to the child, but the caregiver-child dyads were not provided with feedback or instruction; caregivers were instructed to read to their children as they typically did. Baseline measures were taken during at least three sessions for each dyad. This continued until there was little variation across measurements, with no evident trend of increasing performance (McReynolds & Kearns, 1983).

Intervention phase

After stable baselines were established, instruction was provided for each caregiver. Strategy instruction was completed with the first caregiver, who then provided intervention for his or her child; once a treatment effect was demonstrated for the first dyad, the second caregiver received instruction, and so forth (McReynolds & Kearns, 1983). Set A books (i.e., either Little Critter or Dora the Explorer) were used for caregiver instruction and for all intervention phase sessions with the children; the other set (Set B books) was used for generalization.

To facilitate multi-symbol utterance productions, the caregivers were taught to use a modified version of the interaction strategy developed by Kent-Walsh and colleagues (Kent-Walsh, 2003; Kent-Walsh, Hasham, & Stewart, 2004). In the original strategy, the number of aided AAC symbols that were modeled in each step of the hierarchy was not specified. In the current investigation, parents were taught to provide models that contained two symbols during the Read, Ask, and Answer (or RAA RAA RAA!) steps of the hierarchy. Thus, parents implemented as many of the following steps as necessary to elicit a turn from each child on each page of the book: (a) read text + provide two-symbol aided AAC model, (b) ask wh- question + provide two-symbol aided AAC model, and (c) answer wh- question + provide two-symbol aided AAC model. In addition, caregivers provided pause time between each of the above steps (at least 5 s) and responded contingently to each of their children's turns. This responsivity component included providing a two-symbol aided AAC model following each turn that the child took. For example, if the child selected the aided symbol for LITTLE CRITTER, the caregiver could respond contingently by saying Yes, Little Critter hit the ball, while selecting the aided symbols for LITTLE CRITTER and HIT. These parental responses took a variety of forms, including imitations, expansions, and recasts. Figure 1 depicts an illustration of the interaction strategy; each parent was provided with a copy of the strategy.

Figure 1. Interaction strategy used by Latino parents to teach their children to use multi-symbol messages.

To teach the parents how to use this strategy, the first author used a cognitive strategy approach to instruct the parents. The instructional protocol included the following eight steps: (a) pretest and solicit the parent's commitment to learning the targeted strategy, (b) describe the strategy, (c) demonstrate use of the strategy, (d) provide verbal practice of the strategy steps (using the RAA RAA RAA! mnemonic), (e) practice implementing the strategy in controlled contexts (i.e., in role plays), (f) practice implementing the strategy in natural contexts (i.e., book reading with the children), (g) complete post-test and solicit the parent's commitment to long-term implementation of the strategy, and (h) demonstrate generalized use of the strategy. Instructional sessions continued until caregivers used the targeted strategy with 90% accuracy in obligatory contexts (i.e., at least once on each double-page spread). As soon as a caregiver mastered the strategy, data collection on the dependent variable began for the intervention phase for that dyad, that is, the child's use of multi-symbol utterances. The caregiver used the strategy during book reading sessions with his or her child and continued until the child produced a minimum of 10 symbol combinations within a 10 min book reading session for three consecutive sessions.

Generalization phase

Following completion of the intervention phase, generalization probes were taken using Set B books (the set not previously used in the intervention phase). Measures were taken for both dependent variables (parent use of the strategy and child use of multi-symbol utterances).

Maintenance phase

Maintenance probes were conducted 2, 4, and 8 weeks following completion of the generalization phase to determine whether (a) caregivers continued to use the strategy, and (b) children continued to produce symbol combinations. The caregivers did not receive feedback during this phase. If the caregivers' performance had dropped below 80%, additional “booster” sessions would have been implemented. However, all caregivers continued to use the strategy consistently, so no additional instruction was necessary.

Procedural Reliability

A procedural standard, developed by Kent-Walsh (2003), was revised for the current investigation and followed by the first author for all parent instructional sessions. These sessions were videotaped and later viewed and evaluated by trained research assistants. The coders used a procedural checklist, based on the procedural standard, to determine the reliability of the parent instructional sessions. The coder watched videotaped recordings of the instructional sessions while filling out the checklist. Procedural reliability was evaluated for 100% of the parent instructional sessions and was calculated as follows: the number of instructional steps correctly implemented divided by the sum of the number of steps correctly implemented, incorrectly implemented, and omitted. Procedural reliability across participants averaged 97% (range = 93–100%), indicating that the instructional protocol was followed during parent instructional sessions.

Coding

Each session was videotaped and transcribed for data collection purposes. To create the transcripts, trained coders (undergraduate and graduate students in communication disorders) independently generated transcripts while repeatedly reviewing digital videos. Both the parent and child communications were transcribed. For parents, all spoken words (both reading the text and other speech) were recorded as well as actions that helped to describe events (e.g., turning a page, pointing to a picture in the story). For children, all communication modes were transcribed (i.e., aided AAC, unaided AAC, speech/vocalizations, gestures, head nods/shakes) as well as relevant actions. Once the transcripts were complete, the dependent measures were coded by reviewing the videotapes and transcripts. For the caregiver measures, a modified version of Kent-Walsh's (2003) definitions was used. The coder indicated whether or not the caregiver completed each step of the interaction strategy (e.g., read + model two symbols; pause; ask wh- question + model two symbols; etc.). The percentage of steps the caregivers implemented correctly was divided by the total number of steps (i.e., implemented correctly, implemented incorrectly, or missing) to calculate the percent correct for each session. To calculate child measures, the coders indicated each instance of a multi-symbol aided AAC message during each story reading session. Multi-symbol messages were defined as the selection of two or more aided AAC symbols with no more than a one s pause between each symbol (Binger & Light, 2007).

Reliability of Transcription and Data Collection

Reliability was calculated separately for the transcripts and for recording of dependent measures. To calculate transcript reliability, a second trained research assistant re-analyzed randomly selected sections (26%) of transcripts for each book reading session for each phase of the investigation (i.e., at least 20% of the baseline transcripts for the first dyad, at least 20% of the intervention transcripts for the first dyad, etc.). Inter-rater agreement was calculated for behaviors that directly impacted the dependent variables (i.e., all aided AAC messages and spoken words for the caregivers and all aided AAC messages for the children). Percent agreement was calculated by dividing the number of agreements by the sum of the agreements, disagreements, and omissions. The overall mean transcript reliability was 96% for caregiver behaviors (range = 91–99% per session) and 92% for child behaviors (range = 83–100% per session), indicating the transcripts were reliable. When disagreements between transcribers arose, the transcribers re-examined the videos and transcripts together until they mutually agreed upon changes to the transcripts.

For reliability of data collection, a second trained coder independently reviewed 30% of videotapes and transcripts of the primary transcriber and recorded the dependent measures (i.e., in the same manner in which the data were collected by the primary transcriber). Cohen's kappa (Shrout, et al., 1987) was used to calculate inter-rater reliability for the parent and child dependent measures. Kappa for the parent measures averaged 0.85 (range = 0.84–0.86 across parents) and 0.97 for the child data (range = 0.95–1.0 across children), indicating that caregivers' implementation of the instructional steps and children's multi-symbol messages were reliably recorded by the coders.

Data Analysis

Data from the baseline, intervention, generalization, and maintenance phases were graphed and visually inspected for changes in trend, slope, and level (McReynolds & Kearns, 1983). Furthermore, the percentage of non-overlapping data (PND) points (the percentage of points during the intervention phase that exceeded the highest point during baseline) was calculated to measure the effectiveness of the intervention (Scruggs & Mastropieri, 1998). PND scores were rated as follows, according to Scruggs and Mastropieri: greater than 90 indicates that the treatment is very effective, 70 to 90 indicates that the treatment is effective; 50 to 70 indicates that it is questionable; and below 50 indicates that it is ineffective.

Social Validation

Focus group results were used at the onset of the investigation to ensure that the planned program was socially valid. In addition, two measures were taken at the end of the investigation. First, caregivers who participated in the study completed a questionnaire about the perceived benefits for themselves and for their children. Second, the spouses of each adult participant watched two randomly selected, randomly ordered, five-minute pre- and post-intervention video clips of their spouse and child reading storybooks. The spouses then filled out a form containing three forced-choice questions: “In which tape do you think your spouse better supported your child's language skills?”, “In which tape do you think your child used better language skills?,” and “In which tape do you think your child communicated more effectively?.”.

Results

Caregivers

None of the three parents demonstrated use of the instructional strategy prior to instruction (i.e., zero % during each baseline session; see Figure 2). Following the strategy instruction sessions, (which took place in between the baseline and intervention sessions), all three parents consistently used the interaction strategy for the remainder of the investigation (i.e., 80% or higher). The percentage of non-overlapping data between baseline and the remaining phases was 100% for all three parents, indicating that the intervention was highly effective. There was little variation in the instructional time required for each parent (i.e., 2.7 h for Antonio's father and 2.4 h for Angela's mother and Julia's mother).

Figure 2. Percentage of interaction strategy steps correctly implemented by parents within 10 min storybook sessions.

Children

All three children demonstrated limited use of multi-symbol messages via aided AAC during the baseline phase (see Figure 3). After their parents learned to implement the interaction strategy, all three children used symbol combinations consistently and with greater frequency, and all achieved criterion (i.e., at least 10 symbol combinations per 10 min session for three consecutive sessions). The percentage of non-overlapping data between baseline and intervention (i.e., for Set A books) was 100% for all three children, indicating that the intervention was highly effective. Antonio immediately began using more symbol combinations once intervention began. Angela's progress was slower, but by the eighth session (i.e., following 1 h and 20 min of intervention), frequency of use for multi-symbol messages increased. Julia produced at least 10 symbol combinations per session by the fourth intervention session.

Figure 3. Number of multi-symbol messages produced by children within 10 min storybook sessions.

The mean number of different multi-symbol messages that the children used during each phase of the investigation are shown in Table 2. The number of different multi-symbol messages increased dramatically following the baseline phase for each child. The percentage of spontaneous messages during the last three phases (intervention, generalization, and maintenance) was calculated for each child. The majority of messages were produced spontaneously (86% for Antonio, 97% for Angela, and 56% for Julia).

TABLE 2 Mean number of different multi-symbol messages per 10-min session for baseline, intervention, generalization, and maintenance phases.

Participant	Phase
	Baseline	Intervention	Generalization	Maintenance
Antonio	2.0	11.8	11.3	13.6
Angela	0.5	6.5	16.5	14.7
Julia	1.3	10.0	8.5	12.0

Generalization

All three parents used the interaction strategy with at least 80% accuracy during the generalization phase, indicating that they generalized use of the strategy when reading novel storybooks. All three children produced at least 10 multi-symbol messages within three generalization sessions, indicating generalized production of symbol combinations for a novel storybook series.

Maintenance

All three parents continued to use the interaction strategy with at least 80% accuracy for 2 months after the generalization phase ceased. For the children, both Antonio and Angela used symbol combinations consistently 2, 4, and 8 weeks after the generalization phase (minimum = 13 combinations for Antonio, 14 for Angela). Julia produced 17 combinations at her 2 week maintenance session. Her family was on a vacation at the four week point, and she produced eight multi-symbol messages at the 8 week session.

Social Validation

On the parent questionnaire, all participating parents reported very high levels of satisfaction with the program. They all indicated that they would participate in a similar program again and would recommend the program to others. Antonio's father and Julia's mother indicated that their children were talking more. Julia's mother also indicated that her daughter was “participating more during book reading time.” Reported strengths of the program included building the children's vocabulary, working to identify the children's needs, and having an instructor who cared about the program, children, and parents. No one reported changes that they would make to the program. The spouses who viewed the pre- and post-instruction videos all indicated that the participating parents supported their child's language skills better in the post-instruction videos. Furthermore, the spouses indicated that their children's language and communication skills were superior in the post-instruction videos.

Discussion

Revising and Teaching the Interaction Strategy

For the current investigation, Kent-Walsh's (2003) interaction strategy was modified slightly before instruction began with the Latino caregivers. The focus group results indicated that only minimal changes needed to be made to the instructional program to ensure its appropriateness for various New Mexico Latino populations, specifically, the method in which the program was presented to caregivers, the types of storybooks that were used, and the way that expectant delays were administered. The resulting program was used to teach Latino parents to support the early expressive symbol combinations of children who used aided AAC. All of the caregivers in the current study readily learned the interaction strategy, and all three children learned to consistently combine symbols to form multi-symbol messages. The changes that were made to the interaction strategy developed by Kent-Walsh, therefore, did not appear to have a negative impact the effectiveness of the strategy. Furthermore, all three participating caregivers reported very high satisfaction levels with the program, and they appreciated the one-on-one time between researcher and parent and between parent and child.

There are a number of reasons that may have contributed to the success of the Latino parents learning and effectively using the interaction strategy. First, there is much evidence that using a cognitive strategy instructional approach can yield positive outcomes for individuals who are learning a wide range of skills (e.g., Ellis, Deshler, Lenz, Schumaker, & Clark, 1991; Jiménez, 1997), including parents who are learning to support the communication skills of their children who use AAC (Kent-Walsh et al., 2004). Another potential contributing factor relates to the specific characteristics of the parents who participated in the current study. Unlike some other Latino populations in the U.S., many New Mexican Latinos have lived within the U.S. for centuries and are highly acculturated to mainstream U.S. society. The majority of New Mexican Latinos who graduate from high school (a requirement for inclusion in the proposed investigation) are fully fluent in English, and in fact, many New Mexican Latinos speak English as a first language and may not speak Spanish (U.S. Census Bureau, 2000). This was the case with all of the parents in the current study. Furthermore, the components of the interaction strategy (i.e., aided AAC models, expectant delays, wh- questions, and contingent responses) have proven to be effective for improving the communication skills of individuals from mainstream American society (e.g., Binger & Light, 2007) and have been taught to adults within this group (see Kent-Walsh, 2003, for a review). Therefore, it may have been relatively easy for the highly acculturated Latinos in this study to learn the steps of the interaction strategy. Along similar lines, all of the Latino parents in the current study reported reading to their children on a regular basis, and they all asked their children wh- questions during the baseline phase of the investigation (albeit without the presence of aided AAC models), which is not the case with some Latinos (as evidenced by the focus group results; also, see van Kleeck, 1994). Thus, the parents were already familiar with certain aspects of the intervention program. Any or all of these factors may have contributed to the success of the intervention for the particular parents who participated. Teaching the strategy to other Latino populations, such as those who are recent immigrants or who have more traditional values, may require more extensive modifications and additional instructional time (e.g., Parette, Huer, & Scherer, 2004).

Child AAC Outcomes

The interaction strategy yielded positive outcomes for all three child participants. All of the children occasionally used multi-symbol messages during the baseline phase, indicating that for them, this was an emerging skill. This may help explain the rapid acceleration in Antonio's and Julia's use of symbol combinations (i.e., 11 and 10 multi-symbol messages, respectively, during the first intervention session). Angela took somewhat longer to produce such combinations consistently but still produced at least 10 symbol combinations per session after only 1 h and 20 min of instruction (i.e., her eighth intervention session). It is interesting to note that of the three children, Angela was the only one who did not use a voice output device, which may have had an impact on her rate of learning. Similar results were noted in Binger and Light (2007); that is, of the four children who successfully completed the intervention program, the two children who used voice output communication devices progressed more quickly than the two who used communication boards. However, no conclusions on this point can be drawn from either study, as a number of other factors may have had an impact on the outcomes (e.g., motivation, cognitive level, etc.).

A number of factors may have contributed to the children's success in learning to produce symbol combinations. First, it should be noted that the cognitive strategy approach used with the parents required that the children participate in the last step of their parent's instructional program; thus, the children benefited from their parents using the interaction strategy prior to the beginning of the intervention phase. Second, all of the components of the strategy have been used to support the communication skills of children who use AAC (e.g., Kent-Walsh, 2003); and aided AAC modeling, in particular, has proven successful for teaching children who use AAC to combine symbols (Binger & Light, 2007). Third, modifications were made to the strategy to specifically encourage productions of two-symbol messages. For example, parents were taught to model two-symbol messages consistently, which was not a component of the original strategy (Kent-Walsh). Finally, all of the children had extensive prior experience sharing books with their parents. As this was not a new experience for them, the children may have had ample cognitive resources available to attend to the novel components of story reading – specifically, with their parents' use of the RAA RAA RAA! strategy. This prior experience may have contributed to the positive results of the generalization phase as well; that is, the children readily used symbol combinations with a book series that was not used during the intervention phase.

Limitations

Several factors limit the external validity of the findings. First, only one focus group with only four participants was conducted. Furthermore, convenience sampling was used to identify the focus group participants, which is the least desirable sampling technique (Patton, 2002). With regard to the single subject design, only three dyads participated in the investigation. In addition, both the parents and children represented fairly specific sub-groups. All of the parents were Latinos who spoke English as a first language, were highly acculturated into middle class U.S. society, had taken or were taking college courses, and provided their children with rich literacy experiences in their homes. The children in the study had no motor impairments (except for their speech), and receptive language skills were within normal limits. It is possible, then, that the results from the investigation may not generalize to other Latino populations (Huer et al., 2001) or to children who use AAC who have different profiles (e.g., cognitive delays, receptive language impairments, severe motor impairments). However, it is very encouraging to note that a very similar version of the instructional program has been shown to increase the turn-taking skills of children with different profiles, including children with severe speech and physical impairments and children with cognitive delays (Kent-Walsh, 2003; Kent-Walsh et al., 2004). With respect to the contexts for intervention, it is not known if the results from the storybook reading context can be generalized to other settings, such as activities of daily living (e.g., getting dressed, taking a bath). In addition, the long-term maintenance of the use of multi-symbol messages is unknown, as maintenance data were only collected for 8 weeks.

Directions for Future Research

Future research should include other Latino populations, such as Latinos who are bilingual or primarily Spanish-speaking, are recent immigrants, have a different socioeconomic status, and are less acculturated to mainstream US society. For example, Rodriguez and Olswang (2003) found that many Mexican-American mothers (all of whom were either born in Mexico or were first generation U.S. citizens) tended to hold more traditional, authoritarian child-rearing beliefs and values than Anglo-American mothers. Replicating the current investigation with less acculturated Latinos is needed to determine whether further changes to the program are warranted when working with such populations (Parette et al., 2004). More extensive focus groups or other types of qualitative research with various Latino populations may be appropriate in future studies, following the guidelines suggested by Huer and Saenz (2003). External validity would be strengthened by replication with children who use AAC who have different profiles from those in the current investigation, such as children with cognitive impairments or severe physical impairments.

Another major area for future research is to extend this type of intervention to more advanced grammatical skills.. Much of the AAC intervention literature has focused on pragmatic skills (e.g., Kent-Walsh, 2003; Light & Binger, 1998), and recent investigations (in addition to the current study) have begun to look at two-symbol combinations (Binger & Light, 2007; Nigam et al., 2006). To ensure that individuals who use AAC achieve their full linguistic potential, interventions also must address higher level grammar skills, such as the use of grammatical morphemes.

Further research is also required to investigate the impact of communication partners withdrawing the use of prompts on children's multi-symbol utterance productions. The parents continued to use the interaction strategy during the maintenance phase, and it is not known what the children would have done in the absence of prompting. In another intervention study, several children completed a phase in which aided AAC models were withdrawn, and all children who completed this phase continued to produce symbol combinations above baseline levels, although one was below criterion (Binger & Light, 2007).

Finally, two parents in the current investigation commented that they noticed changes in not only their children's use of AAC, but also their speech, even though the program used in the current investigation did not specifically target improvements in speech (no post-instruction speech measures were taken). Further research that involves children with profiles similar to those in the current investigation (i.e., with severe motor-speech disorders) is required to determine the impact of aided AAC on speech (Millar, Light, & Schlosser, 2006).

Author Note

Cathy Binger, Department of Speech and Hearing Sciences, University of New Mexico; Jennifer Kent-Walsh, Department of Communication Sciences and Disorders, University of Central Florida; Jacqueline Berens, Stephanie Del Campo, and Donna Rivera, Department of Speech and Hearing Sciences, University of New Mexico.

This research was supported in part by a New Investigator Research Grant from the American Speech-Language-Hearing Foundation. Preliminary results were presented at the Annual Conference of the American Speech-Language-Hearing Association in Miami, FL, November, 2006. The authors offer their thanks to: Carey Marquez, Michael Martinez, Sandra Nettleton, Annette O'Connor, Barbara Rodriquez, Rebecca Sanchez, Stacy Taylor, and Lauryn Waits. The authors also thank the parents and children who made this study possible.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.