Can Conversation-Based Intervention Using Speech-Generating Devices Improve Language in Children With Partially Intelligible Speech?

Abstract

Children with partially intelligible speech and language difficulties are often provided with speech-generating devices (SGDs) to support their spoken communication. However, little is known about how SGDs can be used best to facilitate language development. This study evaluated the effects of a conversation-based intervention using SGDs, on the rate of expressive clause production in children with speech that is partially intelligible (SPI). A multiple probe across participants design was used to evaluate intervention with four children aged 8 to 10 years. Following baseline monitoring, the children received language-focused intervention involving adult recasting, SGD modeling, and prompting to repair. Participants showed increases in their rate, complexity and length of clause use. Gains were sustained and generalized. Intervention using conversation, recasting, and SGD modeling can increase clause production and grammaticality of children with SPI.

Keywords

speech-generating device (SGD)speech that is partially intelligible (SPI)language delay/disorder conversation-based intervention recasting SGD modeling

Background

Augmentative and alternative communication (AAC) is a term used to describe any mode of communication used to support and/or replace natural speech, such as the use of signing, communication books, and electronic speech-generating devices (SGDs) (Griffiths, Price, & Bloch Clarke, 2018). The last decade has seen significant growth in the availability of specialist speech-generating software made available on mainstream mobile technologies such as tablet computers. Falling costs and growing recognition of SGD potential to support language development and communication has led to such SGDs being provided to children with an increasingly diverse range of neurodevelopmental disabilities (e.g., Ganz, 2015).

For children with minimal or no intelligible speech, and for whom use of gesture and signing is severely restricted by significant physical limitations, high dependence on the use of aided AAC systems, particularly SGDs, provides clear theoretical and clinical rationale for intervention. For these children, SGD use supports linguistic development, and provides a bridge between language understanding and their capacity for expression (Griffiths et al., 2018). SGDs are also provided to children with motor speech disorders who have speech that is partially intelligible (SPI), in the absence of severe whole-body physical impairment. This heterogeneous group includes children with a range of conditions such as dysarthria, for example, associated with Worster-Drought syndrome (WDS; Clark, Harris, Jolleff, Price, & Neville, 2010), childhood apraxia of speech (CAS) (Murray, McCabe, & Ballard, 2014), and developmental speech and language disorders related to genetic conditions (Newbury & Monaco, 2010). Such children commonly receive intervention focusing on speech intelligibility as a primary mode of communication in addition to the use of manual signs and gestures. For these children, SGDs are typically used to provide an “intelligible” alternative when speech is not understood, and where signs are unknown to the listener and gestures lack required linguistic sophistication (Binger, 2007; Murray et al., 2014). However, the ways in which SGD provision can support language development for these children is less clearly defined. We propose that although the focus of SGD intervention support and use is on in-the-moment repair of unintelligible speech, child self-repair can also serve as an important language learning mechanism (Clarke, Soto, & Nelson, 2017), and there is strong potential for interventions to exploit the repair function of SGD use for enhancing language.

A key element of adult scaffolding of child language that offers naturally occurring opportunities for child repair of a prior utterance, and which is understood to be particularly effective in facilitating language acquisition, is the recasting of child utterances. In essence, recasts are adult turns that occur straight after the child’s utterance and include features of the child’s turn, including its basic meaning, while also correcting and/or expanding the turn (e.g., Baker & Nelson, 1984). Following an adult recast, the child may repair their prior utterance incorporating part or all of the adult recast. Such interventions reflect social interactive theories of language development which propose that children learn language through conversational interaction with adults who, consciously or not, scaffold child language learning by aligning to, and responding “in sync” with, the child’s focus of language use and interaction (Clark, 2014).

A large body of intervention research has demonstrated the effectiveness of recasts in facilitating language acquisition in children developing typically and those with difficulties learning language (e.g., Camarata, Nelson, & Camarata, 1994). The use of recasts along with prompting for SGD use in child-directed conversation have also proved positive in supporting enhanced language use by children and adolescents with little or no intelligible speech using SGDs (Soto & Clarke, 2017, 2018); For example, in a conversation-based intervention conducted by Soto and Clarke (2017), the participants chose from photographs of themselves in activities (e.g., trips or holidays) as the basis for a conversation with the interventionist. During these conversations, the children were exposed to adult recasts and explicit prompts to repair their prior turns, using their SGD. The children showed gains in grammaticalization and use of clauses that were broadly maintained outside the intervention sessions. Soto and Clarke (2017) propose that the children using SGDs benefited from recasts delivered during motivating conversations, and from promoting of SGD use primarily because it assisted children to learn and retain the critical operational requirements (see also Clarke et al., 2017).

Adult modeling of AAC use has also formed a critical element of interventions supporting language learning by children with complex needs (O’Neill, Light & Pope, 2018; Sennott, Light, & McNaughton, 2016). For example, the impact of modeling as a primary element of intervention for children was examined by Sennott et al. (2016) in a systematic review of research between 1989 and 2013. Sennott et al. argue that during early childhood, typically developing children are involved in rich language interactions with numerous speech models. In contrast, children who are learning to use AAC rarely see adults modeling the use of their AAC system. Sennott and colleagues identified that the modeling interventions led to consistently positive and meaningful improvements in language in areas of morphology, syntax, semantics, and pragmatics. They concluded that using AAC models alongside techniques such as recasting, in naturalistic interactions, enabled children using AAC to improve receptive and expressive language.

In summary, research indicates that using recasts, prompting for child repair, and modeling within motivating conversations can prove successful in facilitating language development in children with severe speech and physical disabilities. Children with SPI are often supported to use SGDs to repair their unintelligible speech during conversation to enhance communicative intelligibility. Child self-repair can also be a powerful language learning mechanism. Therefore, the purpose of this study was to examine the effects of a conversation-based intervention involving SGDs, and using recasting, repair, and modeling, on the expressive language of children with SPI.

Method

Participants

Child participants

Four children (three boys and one girl) participated in the study. Table 1 gives information for language skills, speech intelligibility and characteristics, and communication modes for each participant. The children were assessed for receptive vocabulary and speech intelligibility by the first author, prior to commencement of the study, and evidence of receptive and expressive language abilities were taken from the children’s Speech and Language Therapy records. All of the children attended the same school, where admission criteria included a standard score of 7 or above on nonverbal subtests of standardized cognitive assessment batteries. The children in this study met the following criteria: (a) aged between 8 years 8 months and 10 years 1 month; (b) presented with a severe speech delay/disorder; (c) presented with a severe receptive and expressive language delay/disorder; (d) had been provided with an SGD which had software allowing for marking of grammar; (e) accessed their SGD directly, using touch-screen and/or keyboard; (f) showed functional communicative competence at Levels II to III on the Augmentative and Alternative Communication Profile (Kovach, 2009); (g) had English as their dominant language; and (h) had hearing and vision within normal functional limits.

Table 1.

Participant Characteristics.

Child	Age	Diagnosis	Receptive vocabulary (age equivalent)^a	Receptive grammar (age equivalent)^b	Expressive grammar (age equivalent)^c	Speech intelligibility^d	Communication modes
Aaron	9;02	CAS	5;05	5;01	<3;06	10%	Speech, manual signs, gestures, Clicker™^e
Ben	8;08	Genetic condition	5;06	<5;00	<3;06	50%	Speech, manual signs, gestures, Grid 2™ on GridPad^e
Claire	10;01	Genetic condition	6;00	5;08	<3;06	3%	Verbalizations, manual signs, gestures, Grid 2™ on GridPad^e
Dilshad	9;05	Worster-Drought syndrome	5;04	5;01	<3;06	35%	Speech, manual signs, gestures, Clicker™ App on iPad^e

Note. Ages and age equivalents are in years;months. CAS = childhood apraxia of speech.

The British Picture Vocabulary Scales–Third edition (Dunn & Dunn, 2009). ^bConcepts & Following Directions Subtest of Clinical Evaluation of Language Fundamentals–Fourth U.K. Edition (Semel, Wiig, & Secord, 2006).^cThe Renfrew Action Picture Test–Revised Edition (Renfrew, 2010). ^dThe Beginner’s Intelligibility Test (Osberger, Robbins, Todd, & Riley, 1994). ^eClicker Communicator App™ for iPad and The Grid 2™ are two symbol and word vocabulary set systems for SGDs which use core vocabulary words, but also allow vocabulary personalization, and creation of spontaneous and novel utterances with grammaticalization.

All of the children used manual signs and gestures in addition to their speech and SGD. Three children used speech as their primary means of communication, and one child used manual signs as the primary mode of expression, as reported by each child’s class SLT. All children were ambulant and physically active. Pseudonyms have been used for all of the children.

Aaron

Aaron was diagnosed with CAS and severe dysarthria of unknown etiology. He also had a profile of poor attention with extreme distractibility. Aaron’s speech was characterized by reduplication and repetitive syllables. His expressive language consisted of single words, with some two-word phrases.

Ben

Ben was diagnosed with the genetic condition 1p microduplication and a severe speech and language delay. He had a severe speech sound disorder, and his expressive language predominantly consisted of nouns, with some verbs and some adjectives. He was able to produce some simple clauses, but this was very inconsistent, and he more typically used two-word phrases.

Claire

Claire was diagnosed with an unspecified genetic condition which presented similarly to WDS, and a severe expressive and receptive language disorder. She had a limited range of speech sounds which she supplemented with finger spelling, and Cued Articulation (Passy, 2010) for speech sounds. She expressed herself mainly through signing of single words and two-word phrases.

Dilshad

Dilshad had a clinical description of WDS and severe speech and language delay. Speech characteristics included devoicing errors, omissions, and inconsistencies. Prior to the intervention, he was reported to have demonstrated an ability to produce clauses containing subject–verb–object and auxiliary verbs in specific therapy settings. He had severe word finding difficulties.

Adult participants

Baseline and intervention sessions were conducted by each child’s class speech and language therapist (SLT), who were all experienced in working with complex pediatric speech, language, and communication needs. Two children were in the same class and received baseline and intervention from the same SLT. The SLTs were blind to the intervention strategy while conducting the baseline sessions. Generalization sessions were conducted with class learning support assistants (LSAs) and a class support volunteer, each familiar to the child with whom they worked. The generalization partners were blind to the intervention.

Experimental Design

A multiple probe across participants design was used, which allows progress made by a number of individuals to be attributed to the strategies used in the intervention (Gast, Lloyd, & Ledford, 2014). The threshold to establish a stable baseline was a rate of variation of no more than three productions of clauses, defined as Subject + Verb + Object/Complement/Adverbial, between all baseline sessions. After the intervention was introduced to the first child, the criterion for beginning intervention with the second, and subsequent children, was production, by the participant currently receiving the intervention, of four clauses more than his or her own highest baseline level, for three consecutive sessions. The order of intervention for the children was randomly assigned.

Procedures

Materials

Baseline, intervention, and generalization probes each centered on conversation between the child participant and an adult partner. For each conversation probe, the children’s parents supplied two pictures of their child taking part in an event such as an outing, club, or home activity, along with specific vocabulary relevant to the event (e.g., names of people at a family party). The events were ones in which the children’s conversation partners had not been involved. To ensure that relevant vocabulary was available to the children on their SGDs, the first author created new vocabulary pages before the commencement of the study. These new pages used the same organization across participants, while the number of cells varied slightly depending on the participant’s language level. The new pages were accessible via the SGD home page and were designed to reflect the child’s current language organization strategy (e.g., graphic symbol type, use of color coding). Access to, and use of, these pages was modeled during intervention. An example of the SGD page layout is shown in Figure 1.

Figure 1.

Example of conversation vocabulary page layout on speech-generating device^a (SGD). Reproduced with permission from Crick Software https://www.cricksoft.com/uk

Baseline

Baseline sessions consisted of a 20 to 30 minute conversation. The child was asked to choose a picture that they had not talked about previously and then asked, “Tell me what happened.” Their SGD was available to them throughout the conversation. The SLTs could ask further questions to promote conversation and use expectant pauses and repeat back to the child what they had understood him or her to say, but they were not permitted to use any other therapeutic techniques to support the child’s communication, and were blind to the intervention techniques.

Training of SLTs

Individual SLTs received information about the intervention procedures once they had completed baseline sessions with all children with whom they were working, so that no SLT was conducting baseline and intervention simultaneously with different children. As SLTs conducting the intervention were experienced in complex pediatrics, a 30-minute session was sufficient to complete initial training. Training consisted of verbal instruction from the first author, some role-play of the child–therapist interaction, and provision of a checklist covering the intervention steps which clinicians reviewed at the beginning of each session.

Intervention

Intervention sessions were conducted at each child’s regular allocated individual speech and language therapy session in school. Each child received 12 sessions of intervention, designed to represent an individual speech and language therapy input package for a school half-term period. The experimental design criteria for staggering the commencement of intervention for each child, intervening school holidays, and participant illness meant that for all children the provision of the total 12 sessions varied between once or twice a week and the intervention period for the children ranged from 6 to 12 weeks.

During intervention, each child continued to attend their usual two to three weekly group speech and language therapy sessions within their class, focusing on different objectives to those of the study, such as topic word learning and reading comprehension skills.

Intervention sessions typically lasted 25 to 30 minutes and consisted of a conversation between the child and SLT. At the start of the session, the child was asked to choose a picture from a choice of two not talked about previously. The intervention strategy involved a six-step procedure, here referred to as ARAMPR:

Ask: ask the child about the picture, for instance, “can you tell me what happened.” This may lead to an exchange whereby the SLT and child engage in establishing the intended meaning of the child’s contribution;

Recast: recast the child’s contribution to correct and expand their prior utterance into a target clause incorporating subject + verb + object/complement/adverbial, plus additional elements such as pronouns, determiners, tense markers, prepositions and conjunctions depending on the expressive language structures already used by the child;

Ask: ask if that is what the child meant and amend if necessary;

Model: model the target clause on the SGD;

Prompt: prompt the child to repair their prior turn including using the SGD, for example, saying “You have a go” with further explicit instructions as required;

Respond: make a positive conversational response to the child’s repaired utterance.

For example:

Ask	SLT	“what else do you want to tell me about this picture”
	Child	“(unintelligible speech) . . .show. . . (unintelligible speech)”
	SLT	“show?”
	Child	“no (unintelligible speech), no, friends”
	SLT	“did you show a friend last night, or are you talking about today?”
	Child	“no, class, everybody”
Recast + Ask	SLT	“you showed everybody in class,” is that right?
	Child	“ye”
Model	SLT	models sentence on the SGD which speaks: ‘I showed everybody in class’
Prompt	SLT	“your turn,” and prompts child to find and combine symbols on the SGD
	Child	uses SGD to speak: ‘I showed everybody in class’
Respond	SLT	“well done”

Note. SLT = speech and language therapist; SGD = speech-generating devices.

During each intervention session, the SLTs delivered a minimum of four “ARAMPR” intervention episodes. This minimum number of episodes was frequently exceeded later in the intervention, as the children’s skills developed. The respective levels of recasting, modeling, and prompting for each intervention episode also developed during the intervention in line with the child’s developing skills, and following discussion between the investigator and individual SLTs for each child.

Generalization

Four generalization sessions of between 20 to 30 minutes each were conducted: one during baseline, one during intervention, one at 2 weeks postintervention, and a final session at 4 weeks postintervention. All the pictures used in the study sessions preceding that generalization session were available to the child, and new pictures of recent events were added with corresponding vocabulary on the SGD for the postintervention sessions. Children were also free to talk on a topic of their choice if they did not want to use the pictures.

The adults conducting the generalization sessions were blind to the procedures of the intervention. Adult conversation scaffolding techniques mirrored those of baseline sessions, that is, the adults were instructed that they could ask open questions, and use expectant pauses and follow-up questions, and repetition of child contributions to check understanding. They were not to use any other therapeutic techniques to support the child.

Data Analysis

Fidelity

SLTs and LSAs read a procedural checklist before each session to ensure compliance with the procedures. The first author reviewed video-taped recordings of baseline, intervention, and generalization sessions every week to assess for correct implementation. No deviations from the protocol were observed, and SLTs followed procedures for baseline and intervention without difficulty. Intervention adaptation was discussed with SLTs concerning the respective levels of recasting, modeling, and prompting used in response to children’s developing skills. LSAs complied correctly with generalization session procedures throughout.

Transcription

All baseline, intervention, and generalization sessions were video-recorded and were transcribed by the lead author. All child output was orthographically transcribed and annotated for mode of communication. Adult output was annotated for type of utterance.

Dependent measures

The transcriptions of the children’s output, made using the SGD, speech, signing, or a mixture of these modes, were analyzed by the lead author for the following measures of grammaticality:

Rate of self-initiated clause use, by any communication mode. Clauses were defined as basic sentences containing at least three elements: subject, verb, and either object, complement, or adverbial, but which did not have to be in correct grammatical order. Self-initiated clauses were those that were produced spontaneously or were self-repaired by the child. Spontaneous clauses were defined as either initiated by the child or following an adult comment, question, or verbatim repetition for clarification of the child’s utterance. Self-repaired clauses were those in which the child reformulated at least 50% of their original utterance. A measure of rate use was used because session length varied. Rate was calculated for each dependant variable, by multiplying the number of occurrences by 30 and dividing that total by the length of the session in minutes.

Rate of clause use following prompt to repair, produced by any communication mode. These were defined as utterances containing at least three elements: subject, verb, and either object, complement, or adverbial, not necessarily in grammatical order, that followed an adult recast, model or prompt for their production.

Rate of fully grammatical clause use (by any mode), that is, any clauses, either self-initiated or repaired following a prompt, that were completely grammatically correct.

Length of longest utterance (using any mode), that is, the total number of words in all the clauses in the longest conversation turn/utterance (Scott and Stokes (1995) found that phrase structure expands and clauses contain an increasing number of adverbials as syntax in school-aged children develops).

Rate of use of different modes of communication. The mode(s) of communication used by the participants to produce each clause were recorded, that is, SGD, speech, sign, or mixed modes (a combination of two or all of the modes). Rate of use was then calculated for each mode.

Reliability

To determine reliability of the transcriptions, the second transcriber, who was a qualified SLT preparing to return to the profession, independently viewed and transcribed 10% of the videos. This second transcriber reviewed videos from all stages and children. She was masked to the stages of the study and did not know the children or adults involved. Transcriptions of all words and word attempts by the child (by whatever communication mode), plus the coding of every adult utterance preceding a child utterance, were compared. Agreement between the transcribers for each transcript was calculated by dividing the number of agreements by the total number of agreements, omissions, and differences. The mean reliability score for the child words and word attempts was 86% (range of 77–93%), and the mean reliability score for coding of the adult utterances was 86% (range of 77–92%). Transcript differences and omissions which affected utterances meeting the criteria for analysis were reexamined and discussed by the two transcribers until consensus was reached. Each transcript was then altered in line with these decisions.

To determine reliability of the analyses, a second analyst, who was a university research assistant working in the field of neurodisability, independently examined 25% of the transcripts across all stages and participants, in randomized order and masked to the stages of the study. Inter-analyst agreement was calculated by dividing the number of agreements for each dependant variable across the transcript sample by the total number of agreements, disagreements, and omissions for that feature. Mean reliability across the sample was 85% for identification of qualifying clauses, 97% for coding of clauses as self-initiated versus prompted to repair, 86% for the number of words in the clause, and 94% for identification of fully grammatical clauses. The overall mean reliability score across dependent variables was 92% (range of 78–100%).

Visual and statistical analysis

Analysis of the data for the baseline, intervention, and generalization probes was carried out visually and statistically. Single case design (SCD) standards produced for the U.S. Department of Education’s What Works Clearinghouse detail six features to employ in visual analysis: trend, variability, immediacy of effect, overlap, consistency of data patterns, and level of performance (Kratochwill et al., 2010). The data were examined for these features.

Statistical estimates of the effectiveness of the intervention were calculated for all dependent measures for each of the participants, using nonoverlap procedures. Tau-U and Improvement Rate Difference were used in combination and were calculated using an online calculator (www.singlecaseresearch.org). Tau-U is an index designed for single-case research, which uses a combination of trend measures for the intervention, and nonoverlap measures between study phases. IRD is the improvement rate in baseline subtracted from the improvement rate in intervention (Parker, Vannest, & Davis, 2011).

Social validation

The social validity of the intervention and outcomes was explored by seeking the perceptions of the clinicians taking part in the study (Schlosser, 1999). Semi-structured interviews were conducted by the investigator with the SLTs, within 6 weeks of each SLT completing the intervention. The SLTs’ views about any aspect of the intervention were sought including whether, and how, the intervention and components of the intervention affected the child. The interviews were audio-recorded, reviewed, and a written record made. These data were then analyzed qualitatively following the principles of thematic analysis (Braun & Clarke, 2006) by coding features across the data, collating the coded data into themes, and refining and naming these themes.

Results

Figures 2 to 4 present the results for the use of self-initiated clauses, clauses following a prompt to repair, fully grammatical clauses, and the length of the longest utterance. These provide visual representation of trend, variability, immediacy of effect, and consistency. Visual inspection of Figures 2 to 4 suggests that experimental control was maintained for all dependent variables across all participants. For all participants, the use of self-initiated clauses, clause use following prompt to repair, and the use of fully grammatical clauses increased during intervention and were generalized above preintervention levels once the intervention ended, with the exception of generalization of fully grammatical clauses for two of the children. The length of the longest utterance increased during intervention, and was generalized and maintained above preintervention levels for three of the children.

Figure 2.

Rate of clause use.

Figure 3.

Rate of fully grammatical clause use.

Figure 4.

Length of longest utterance in each session.

The mean rate of use of self-initiated clauses versus clauses following prompt to repair in intervention is shown as part of Table S1 (Supplementary Materials). All children increased their mean rate of use of self-initiated clauses in both intervention and generalization sessions when compared with baseline levels.

The communication modes used to produce clauses in each phase are presented in Figure 5. All participants increased the mean rate of clauses produced using their SGD during intervention to above baseline levels. All of the children also increased the mean rate of clauses they produced by speech or mixed modes of communication in these sessions.

Figure 5.

Mean rate of clause production by each mode of communication in each phase.

Statistical Nonoverlap Measures of Effect

The Tau-U and IRD scores (see Table S2 in Supplementary Materials) show highly comparable results. Intervention effect for Tau-U and IRD scores can be interpreted in relation to benchmarks as follows, Tau-U: questionable <0.65, effective 0.66 to 0.92, very effective >0.92; IRD: >0.7 very effective, 0.5 to 0.7 effective, <0.5 questionable. IRD and Tau-U indicate that the intervention was effective or very effective for a majority of dependent measures. A summary of the results for each child is presented below.

Participants

Aaron

Aaron did not produce any self-initiated clauses in baseline. In intervention, his level of use of the target measures was 2 (range of 0–5) for self-initiated clause use and 5 (range of 3–7) for prompted clause use. His level for use of fully grammatical clauses was 1 (range of 0–5), and his level for the number of words in his longest utterance was 8 (range of 5–12). Aaron generalized use of self-initiated clauses, producing a level of 2 (range of 1–3) in generalization probes compared with 0 at baseline. The length of his longest utterance in generalization sessions was also above baseline at a level of 5 (range of 3–6) compared with 0 at baseline. Aaron did not produce fully grammatical clauses during generalization sessions.

Aaron used a variety of means to produce clauses during intervention and generalization, in comparison with no use of clauses at baseline (see Figure 5). His rate of clause production using his SGD was 0 at baseline, 5 in the intervention sessions, and 1 in the generalization sessions. He used mixed modes to produce 0 clauses at baseline but a level of 1 clause during both intervention and generalization sessions. He used speech to produce 0 clauses at baseline and generalization but a level of 1 clause during the intervention sessions.

Ben

Ben’s use of all the target measures increased during intervention as indicated by the change in levels between baseline and intervention sessions. Level of self-initiated clause use was 5 (range of 4–7) at baseline and 9 (range of 1–13) in intervention. Level of clause production following prompt for repair in intervention was 5 (range of 1–8). Levels for rate of fully grammatical clause use were 0 at baseline and 2 during intervention (range of 0–6). Levels for the number of words in his longest utterance were 8 at baseline (range of 5–11) and 10 in intervention (range of 7–20). Ben generalized use of self-initiated clauses, with a level of 14 (range of 10–19) compared with 5 (range of 4–7) at baseline. His level of use of fully grammatical clauses was 1 (range of 0–3) in generalization probes compared with 0 at baseline. Level for number of words in Ben’s longest utterance, however, did not increase in the generalization sessions, with a level of 7 (range of 6–9) compared with the 8 (range of 5–11) at baseline.

Ben increased his use of clauses using speech and mixed modes in intervention and generalization sessions when compared with baseline sessions as follows: level of clause use by speech was 4 at baseline, 7 during intervention, and 9 in generalization; levels of use of mixed modes was 1 at baseline, 2 during intervention, and 5 during generalization. Ben increased his rate of production of clauses using the SGD between baseline and intervention, where levels increased from 1 to 5. He did not use his SGD in generalization sessions.

Claire

Claire’s use of grammatical measures also showed increases. Her level of self-initiated clause use was 2 (range of 2–3) at baseline and 3 (range of 0–6) in intervention, with a level of 7 (range of 3–11) for prompted clause use. Claire’s level for rate of fully grammatical clause use was 0 at baseline and 1 (range of 0–5) during intervention. The level for the number of words in her longest utterance was 3 (range of 3–4) at baseline and 8 (range of 6–11) during intervention. She generalized the use of self-initiated clauses, producing a level of 4 (range of 4–5) in generalization probes, compared with 2 (range of 2–3) at baseline. The length of longest utterance showed some improvement from a level of 3 (range of 3–4) at baseline to 4 (range of 0–6) in generalization. Claire did not use fully grammatical clauses in baseline or generalization.

Claire increased her rate of production of clauses using her SGD and mixed modes in intervention and generalization sessions when compared with baseline sessions (see Figure 5). Levels for rate of clause use produced via her SGD were 0 at baseline, 8 during intervention, and 2 in generalization, and for mixed modes were 0 at baseline, 1 during intervention, and 2 during generalization. Claire’s rate of production of clauses using signing decreased, with levels of 2 at baseline, 1 during intervention sessions, and 0 in generalization sessions.

Dilshad

Dilshad was able to produce more self-initiated clauses than the other participants at the start of the study (in baseline), though none of his clauses were fully grammatical. He made large increases in his rate of production of the grammatical measures. His levels for rate of self-initiated clause use were 18 (range of 16–19) at baseline and 19 (range of 7–30) during intervention. His level of clause use following prompt to repair was 8 (range of 4–14). Dilshad’s level of fully grammatical clauses increased from 0 at baseline to 12 (range of 1–17) during intervention. The level for number of words in his longest utterance was 11 (range of 10–14) at baseline and 19 (range of 13–30) during intervention. Dilshad was also able to generalize these increases, producing a level of 26 (range of 18–38) self-initiated clauses in generalization probes compared with 18 (range of 16–19) at baseline. Fully grammatical clauses also increased from a level of 0 at baseline to 8 (range of 1–15) in generalization. A particularly large increase was observed in the length of his longest utterance from a level of 11 (range of 10–14) at baseline to 26 (range of 12–48) in generalization. Dilshad’s gains continued to grow strongly for all targets after the intervention ended (see Figures 2 –4).

Dilshad increased his rate of production of clauses using the SGD and speech in intervention and generalization sessions when compared with baseline sessions (see Figure 5). Levels for rate of clauses produced using the SGD were 2 at baseline, 12 during intervention, and 10 in generalization, and levels for clauses produced using speech were 9 at baseline, 12 during intervention and 12 during generalization. Using mixed modes, Dilshad produced levels of 7 clauses at baseline, 3 during intervention, and 4 during generalization.

Social Validation

One central theme concerned observed improvements in the children’s use of grammar, with the SLTs commenting positively on improvements during intervention and after completion of the study measures. For instance, Aaron had a “much better understanding of word-order” and was “now beginning his sentences with a person.” A second theme concerned features of the intervention itself. For example, one SLT reported that “recasting enabled the child to see how to put the words into a grammatically correct sentence.” However, difficulties arose for children when the SGD did not contain vocabulary items which the child wanted to use. SLTs commented also that the intervention had highlighted new areas of strength and hitherto unrecognized areas of difficulty, for example, for Ben, this related to difficulties with topic maintenance. A third highlighted theme was that the children had begun to demonstrate increased “ownership” of their SGDs and confidence in communication as a consequence of experiencing the benefits of its use. For instance, Ben was reported to use his SGD “frequently” and “proudly,” despite sometimes struggling with its functionality, and Claire was reported to have started using her SGD spontaneously to communicate, rather than in response to prompting. Dilshad’s SLT reported that since participating in the intervention, he “took a breath” to think and formulate his language before he spoke, and that he appeared “empowered” by his increased intelligibility through using his SGD. SLTs also commented that feedback received from parents was very positive. For instance, Dilshad’s parents said he could now “add in the details of his life” and he was “so much happier and more confident.”

Discussion

The purpose of this study was to examine the effects of a conversation-based intervention using child language scaffolding including SGDs on the expressive language of children with SPI. The children had receptive and expressive language delay/disorder and communicated using combinations of communication modes, including speech, signing, and SGDs.

The intervention was effective in increasing the rate of production of self-initiated clauses by the children, and three of the four children also showed increases in the number of words used within the clauses they produced, although, arguably, fewer gains were recorded in the use of fully grammatical clauses. When compared with baseline, increases shown during intervention were generalized to conversations with familiar adults who were blind to the intervention strategy. Many of the generalized gains were greatest at 2 and 4 weeks postintervention (see Figures 1 –3).

The findings from this study contribute to a growing body of research examining the effects of conversation-based interventions using adult scaffolding techniques on expressive language in children provided with SGDs (Soto & Clarke, 2017, 2018). In this study involving children with partially intelligible speech, SGDs represent one modality of expression that may or may not be used interchangeably with children’s speech. As previously reported in the AAC literature (e.g., Soto & Clarke, 2017), adult recasts provide children with opportunities to directly compare their utterance with one that is grammatically improved. In this study, as only corrective recasts were used (i.e., when the child produced an incomplete or erroneous utterance), the capacity for recasts to stimulate child “noticing” and contrasting their utterance with the adult reformulation was enhanced (Clarke et al., 2017). Adult modeling of the target utterance on the SGDs also provided the children with a template to imitate when producing their own utterances (Sennott et al., 2016). Explicit prompting for the child to repair their prior utterance using the SGD offered focused experiential learning opportunities for language construction and familiarization with the language infrastructure of the SGD, for example, using the grammatical color coding system. These episodes of clause building invariably slowed the pace of the conversation. This, together with the children’s active language work, may have further enhanced opportunities for embedding and retaining new language structures. Conducting the intervention through conversations about the children’s own experiences, as chosen by the children, further stimulated the children’s commitment to the activity, reflecting research evidence for the benefits of this approach in language intervention (Khan, Nelson, & Whyte, 2014), and the views expressed by the SLTs who carried out the intervention, who felt strongly that conversation on topics important to the children and previously unknown to the SLTs was extremely motivating and engaging for the children.

The conversational approach, and the facility for the SLTs to work collaboratively with the first author in reviewing the intervention sessions, gave occasion to amend the respective levels of recasting, modeling, and prompting in the intervention as appropriate to changes in child performance (Light & McNaughton, 2015). For example, complete recasts and models used early in the intervention for Claire were often scaled down by the end of the intervention to a simple prompt “can you say that altogether on your SGD” alongside some support for accessing word categories on her SGD. Similarly, in later sessions, Dilshad often required less adult scaffolding as he began to reformulate his own utterances spontaneously and then produced them on the SGD. In contrast, for Aaron, who had difficulties with attention and distractibility, full recasts, models, and prompts were maintained throughout the intervention, whereas Ben’s difficulties with sequencing meant he continued to need scaffolding support to use the SGD throughout intervention.

The intervention was less effective at stimulating the children’s production of fully grammatical clauses; however, qualitative improvements were noted, not only in using the subject-verb-object/adverbial/complement clause structure but also in children’s use of a wide range of grammatical items including pronouns, tense and plural markers, adverbials, determiners, and prepositions. This is exemplified by clauses such as “Nadia sign her name in books” from Claire and “I wear France a helmet” from Aaron.

It is proposed that conversations that involve child self-repair, in the context of scaffolding techniques including adult recasting, can also facilitate language development in children using SGDs (Clarke et al., 2017). Although the primary aim of the intervention examined in this article was to facilitate improvement in language, increased use of SGDs, either in isolation or combined with speech, gesture, and signing, has clear potential to improve the overall communicative intelligibility of children with SPI. In a review of the effects of the use of low-tech communication aids (e.g., charts and books) and signing in intervention on speech production by children with autism and learning disabilities, Millar, Light, and Schlosser (2006) reported that natural speech and AAC should both increase in frequency if they are used together in the intervention alongside child reinforcement. The impact of conversation-based SGD mediated intervention on the speech production and overall communicative intelligibility of children with SPI is a promising area for further research. Notably also, qualitative improvements in the use of clause formations delivered using speech and/or sign are indicative of improvements in these children’s core language representations, rather than representing improved competence in their use of the SGDs.

Limitations

An important consideration for this study was the concern that extended baseline conditions, when children were not receiving support or reinforcement, can result in “aversive experiences for learners” (Gast et al., 2014, p. 259). Consequently, the baseline period for the study was limited to three sessions. Although this meets established requirements to demonstrate effect (Kratochwill et al., 2010), a greater number of baseline and generalization sessions would have enhanced the standard of evidence. We note also that the semi-structured interviews used for assessing social validity were conducted by the first author, which may have influenced the views provided by SLTs conducting the intervention.

Conclusion

SGDs are increasingly offered to children with SPI who have language difficulties; however, there is a dearth of research evidence about how SGDs may best be used to support these children’s language learning. This study provides positive evidence for the use of conversation-based intervention employing both child-chosen topics of personal events and structured scaffolding of child language including modeling of SGD use and child self-repair, on the language skills of children with partially intelligible speech.

Supplemental Material

Supplemental_Materials_Table_S2_IRD_and_Tau-U_002_plus_MC_2 – Supplemental material for Can Conversation-Based Intervention Using Speech-Generating Devices Improve Language in Children With Partially Intelligible Speech?

Supplemental material, Supplemental_Materials_Table_S2_IRD_and_Tau-U_002_plus_MC_2 for Can Conversation-Based Intervention Using Speech-Generating Devices Improve Language in Children With Partially Intelligible Speech? by Jessie M. Luckins and Michael T. Clarke in Communication Disorders Quarterly

Supplemental Material

Supplemental_Material_S1_Mean_Values_for_Dependent_Measures – Supplemental material for Can Conversation-Based Intervention Using Speech-Generating Devices Improve Language in Children With Partially Intelligible Speech?

Supplemental material, Supplemental_Material_S1_Mean_Values_for_Dependent_Measures for Can Conversation-Based Intervention Using Speech-Generating Devices Improve Language in Children With Partially Intelligible Speech? by Jessie M. Luckins and Michael T. Clarke in Communication Disorders Quarterly

Footnotes

Acknowledgements

The authors thank the children, staff, SLTs, and research assistant who took part in this study and made everything possible, and the parents and school administrators who gave their support. The authors also thank Prof Gloria Soto for comments on earlier drafts of this paper.

Authors’ Note

This research was completed in partial fulfillment of the first author’s MRes study at University College London. Preliminary results were presented at the Communication Matters International AAC Conference in Leeds, United Kingdom, in September 2018.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

Supplementary material for this article is available on the Communication Disorders Quarterly website along with the online version of this article.

References

Baker

N. D.

Nelson

K. E.

(1984). Recasting and related conversational techniques for triggering syntactic advances by young children. First Language, 5, 3–21.

Binger

(2007). Aided AAC intervention for children with suspected childhood apraxia of speech. Perspectives on Augmentative and Alternative Communication, 16(1), 10–12.

Braun

Clarke

(2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3, 77–101.

Camarata

S. M.

Nelson

K. E.

Camarata

M. N.

(1994). Comparison of conversational-recasting and imitative procedures for training grammatical structures in children with specific language impairment. Journal of Speech and Hearing Research, 37, 1414–1423.

Clark

E. V.

(2014). Pragmatics in acquisition. Journal of Child Language, 41, 105–116.

Clark

Harris

Jolleff

Price

Neville

B. G.

(2010). Worster-Drought syndrome: Poorly recognized despite severe and persistent difficulties with feeding and speech. Developmental Medicine and Child Neurology, 52, 27–32.

Clarke

M. T.

Soto

Nelson

(2017). Language learning, recasts, and interaction involving AAC: Background and potential for intervention. Augmentative and Alternative Communication, 33, 42–50. doi:10.1080/07434618.2016.1278130

Dunn

L. M.

Dunn

D. M.

(2009). The British Picture Vocabulary Scales (3rd ed.). London, England: GL Assessment.

Ganz

J. B.

(2015). AAC interventions for individuals with autism spectrum disorders: State of the science and future research directions. Augmentative and Alternative Communication, 31, 203–214.

10.

Gast

D. L.

Lloyd

B. P.

Ledford

J. R.

(2014). Multiple baseline and multiple probe designs. In Gast

D. L.

(Ed.), Single case research methodology: Applications in special education and behavioural sciences (pp. 251–296). London, England: Routledge.

11.

Griffiths

Price

Bloch Clarke

M. T.

(2018). Alternative and augmentative communication. In Cowan

Najafi

(Eds.), The handbook of electronic assistive technology (EAT) (pp. 181–214). London, England: Elsevier.

12.

Khan

Nelson

Whyte

(2014). Children choose their own stories: The impact of choice on children’s learning of new narrative skills. Journal of Child Language, 41, 949–962.

13.

Kovach

T. M.

(2009). Augmentative & alternative communication profile—A continuum of learning. East Moline: LinguiSystems.

14.

Kratochwill

T. R.

Hitchcock

Horner

R. H.

Levin

J. R.

Odom

S. L.

Rindskopf

D. M.

Shadish

W. R.

(2010). Single-case designs technical documentation. Retrieved from http://ies.ed.gov/ncee/wwc/pdf/wwc_scd.pdf

15.

Light

McNaughton

(2015). Designing AAC research and intervention to improve outcomes for individuals with complex communication needs. Augmentative and Alternative Communication, 31, 85–96. doi:10.3109/07434618.2015.1036458

16.

Millar

D. C.

Light

J. C.

Schlosser

R. W.

(2006). The impact of augmentative and alternative communication interventions on the speech production of individuals with developmental disabilities: A research review. Journal of Speech, Language, and Hearing Research, 49, 248–264.

17.

Murray

McCabe

Ballard

K. J.

(2014). A systematic review of treatment outcomes for children with childhood apraxia of speech. American Journal of Speech-Language Pathology, 233, 486–504.

18.

Newbury

D. F.

Monaco

A. P.

(2010). Genetic advances in the study of speech and language disorders. Neuron, 68, 309–320.

19.

O’Neill

Light

Pope

(2018). Effects of interventions that include aided augmentative and alternative communication input on the communication of individuals with complex communication needs: A meta-analysis. Journal of Speech, Language, and Hearing Research, 61, 1743–1765.

20.

Osberger

M. J.

Robbins

A. M.

Todd

S. L.

Riley

A. I.

(1994). Speech intelligibility of children with cochlear implants. Volta Review, 96, 169–180.

21.

Parker

R. I.

Vannest

K. J.

Davis

J. L.

(2011). Effect size in single-case research: A review of nine nonoverlap techniques. Behavior Modification, 35, 303–322.

22.

Passy

(2010). Cued articulation: Consonants and vowels (Rev. ed.). Melbourne, Victoria, Australia: ACER Press.

23.

Renfrew

. (2010). The Action Picture Test: Revised Edition (The Renfrew Language Scales). Oxford: Speechmark Publishing Ltd.

24.

Schlosser

R. W.

(1999). Social validation of interventions in augmentative and alternative communication. Augmentative and Alternative Communication, 15, 234–247.

25.

Scott

C. M.

Stokes

S. L.

(1995). Measures of syntax in school-age children and adolescents. Language, Speech and Hearing in Services in Schools, 26, 309–319.

26.

Semel

Wiig

E. H.

Secord

W. A.

(2006). Concepts & Following Directions Subtest of Clinical Evaluation of Language Fundamentals–Fourth U.K. edition. London: Pearson.

27.

Sennott

S. C.

Light

J. C.

McNaughton

(2016). AAC modeling intervention research review. Research and Practice for Persons with Severe Disabilities, 41, 101–115.

28.

Soto

Clarke

M. T.

(2017). Effects of a conversation-based intervention on the linguistic skills of children with motor speech disorders who use augmentative and alternative communication. Journal of Speech Language and Hearing Research, 60, 1980–1998. doi:10.1044/2016_JSLHR-L-15-0246

29.

Soto

Clarke

M. T.

(2018). Conversation-based intervention for adolescents using augmentative and alternative communication. Augmentative and Alternative Communication, 34, 180–193. doi:10.1080/07434618.2018.1490926

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB

0.09 MB