ABA Research

LSHSS

Research Article

Explicit Grammar Intervention in Young School-Aged Children With Developmental

Language Disorder: An Efficacy Study Using Single-Case Experimental Design

Samuel D. Calder,a Mary Claessen,a Susan Ebbels,b,c and Suze Leitãoa

Purpose: This study evaluated the efficacy of an explicit, combined metalinguistic training and grammar facilitation intervention aimed at improving regular past tense marking for nine children aged 5;10–6;8 (years;months) with developmental language disorder. Method: This study used an ABA across-participant multiple-baseline single-case experimental design. Participants were seen one-on-one twice a week for 20- to 30-min sessions for 10 weeks and received explicit grammar intervention combining metalinguistic training using the SHAPE CODING system with grammar facilitation techniques (a systematic cueing hierarchy). In each session, 50 trials to produce the target form were completed, resulting in a total of 1,000 trials over 20 individual therapy sessions. Repeated measures of morphosyntax were collected using probes, including trained past tense verbs, untrained past tense verbs, third-person singular verbs as an extension probe, and possessive ’s as a control probe. Probing contexts included expressive morphosyntax and grammaticality judgment. Outcome measures also

included pre–poststandard measures of expressive and receptive grammar. Results: Analyses of repeated measures demonstrated significant improvement in past tense production on trained verbs (eight of nine children) and untrained verbs (seven of nine children), indicating efficacy of the treatment. These gains were maintained for 5 weeks. The majority of children made significant improvement on standardized measures of expressive grammar (eight of nine children). Only five of nine children improved on grammaticality judgment or receptive measures. Conclusion: Results continue to support the efficacy of explicit grammar interventions to improve past tense marking in early school-aged children. Future research should aim to evaluate the efficacy of similar interventions with group comparison studies and determine whether explicit grammar interventions can improve other aspects of grammatical difficulty for early school-aged children with developmental language disorder. Supplemental Material: https://doi.org/10.23641/asha. 11958771

D evelopmental language disorder (DLD) refers to a condition in which children experience lan- guage difficulties in the absence of known bio-

medical conditions or acquired brain injury (Bishop,

Snowling, Thompson, Greenhalgh, & CATALISE-2 Con- sortium, 2017). Compared to typically developing peers, children with DLD present with particular difficulties in morphosyntactic skills, such as the use (Rice, Wexler, & Hershberger, 1998) and judgment (Rice, Wexler, & Redmond, 1999) of grammatical morphemes associated with tense.

Finiteness marking is challenging for children with DLD (see Leonard, 2014, for a review). Finiteness refers to the obligatory marking of verbs indicating subject–verb agreement and tense, including affixation of morphemes –ed (e.g., the girl walked) and –s (e.g., the girl walks) to verbs for past and present tenses, respectively. Within English and cross-linguistically, finiteness is a quality of well-constructed clauses (Dale, Rice, Rimfeld, & Hayiou-Thomas, 2018). There is evidence supporting disordered finiteness as a distinct etiological construct and predictive marker of lan- guage growth for DLD (Bishop, Adams, & Norbury, 2006).

aSchool of Occupational Therapy, Social Work and Speech Pathology, Curtin University, Perth, Western Australia bMoor House Research and Training Institute, Moor House School & College, Oxted, United Kingdom cLanguage and Cognition, University College London, United Kingdom

Correspondence to Samuel D. Calder: [email protected]

Editor-in-Chief: Holly L. Storkel Editor: Amanda J. Owen Van Horne

Received March 19, 2019 Revision received July 1, 2019 Accepted October 5, 2019 https://doi.org/10.1044/2019_LSHSS-19-00060 Publisher Note: This article is part of the Forum: Morphosyntax Assessment and Intervention for Children.

Disclosure: The authors have declared that no competing interests existed at the time of publication.

Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020 • Copyright © 2020 American Speech-Language-Hearing Association298

Children’s grammar difficulties are a primary source of pa- rental concern when considering referral for clinical services (Bishop & Hayiou-Thomas, 2008).

Grammar Interventions Treatment for DLD aims to accelerate language growth

and remove barriers to functional communication by har- nessing strengths (Justice, Logan, Jiang, & Schmitt, 2017). Ebbels’s (2014) review indicates an emerging evidence base for the effectiveness of grammar intervention for school- aged children with DLD. Current evidence is parsed into implicit and explicit approaches to intervention. According to Ebbels’s framework, implicit interventions target production and understanding of grammar using grammar facilitation techniques implicitly by responding to children’s errors in a naturalistic way (Fey, Long, & Finestack, 2003). Children’s learning and the knowledge acquired are not necessarily associated with awareness. Explicit interventions target in- creased awareness of the goals of intervention with a pre- established concept of the criteria for success: Learning is conscious and deliberate, and information can be recalled on demand (Shanks, 2005). Within each approach to inter- vention, specific techniques are used to improve acquisition of grammar.

Implicit Interventions Using Grammar Facilitation Intervention and scaffolding techniques used in im-

plicit approaches are described as grammar facilitation (e.g., Fey et al., 2003), which aims to facilitate the acquisition of grammar by increasing the frequency and quality of target forms in input and output. Greater exposure to and oppor- tunities to learn and use language theoretically accelerates the likelihood of language growth (Leonard, 2014). Studies have empirically tested grammar facilitation techniques sup- porting their use with expressive morphosyntax targets, including imitation (Nelson, Camarata, Welsh, Butkovsky, & Camarata, 1996), modeling (Weismer & Murray-Branch, 1989), focused stimulation (Leonard, Camarata, Brown, & Camarata, 2004), and conversational recasting (see Cleave, Becker, Curran, Van Horne, & Fey, 2015, for a review). Recently, Van Horne, Fey, and Curran (2017) reported on a primarily implicit intervention, in which procedures in- cluded a combination of sentence imitation, observational modelling, storytelling and focused stimulation, recasting, and cueing for incorrect responses. All eighteen 4- to 10-year- old children with DLD enrolled in the study improved their use of regular past tense. Notably, as participants were dismissed from the study following 36 sessions, many still did not achieve mastery of the intervention target. In general, outcomes following implicit intervention are favor- able for morphosyntax in preschool-aged children (Leonard, 2014); however, mastery of intervention targets is rarely reported.

Explicit Intervention Using Metalinguistic Training Difficulties with morphosyntax often persist into

school age for children with DLD (Bishop, Bright, James,

Bishop, & Van der Lely, 2000). An alternative approach may be required because children with DLD may have difficulty learning grammar through implicit grammar fa- cilitation. Metalinguistic training aims to improve children’s learning of the rules of grammar by creating conscious awareness of grammar through explicit metacognitive teach- ing (Ebbels, 2014), allowing children to actively reflect on language targets. Meta-awareness is enhanced, so rules of grammar are learned explicitly in a compensatory way.

Metalinguistic techniques can be used explicitly to teach grammar through metacognitive strategies using visual supports and graphic organizers (Ebbels, 2014). The SHAPE CODING system is designed to explicitly teach oral and written syntax to children with language disorder (Ebbels, 2007). Ebbels, van der Lely, and Dockrell (2007) compared use of the SHAPE CODING system with semantic therapy and a no-treatment control group with 27 children with DLD aged between 10 years and 16;1 (years;months). The authors concluded that the SHAPE CODING system is a viable and efficacious treatment approach to improve verb argument structure in older school-aged children. Although evidence for improvement in grammar comprehension is mixed (e.g., Zwitserlood, Wijnen, van Weerdenburg, & Verhoeven, 2015), children may be able to consciously re- flect upon the rules of grammar through explicit interven- tions in the presence of receptive language difficulties to improve understanding, especially older children (Ebbels, Marić, Murphy, & Turner, 2014).

Grammar intervention approaches effective for chil- dren above 8 years old should be tested with younger chil- dren to address the concerning gap in evidence for this age group (Ebbels, 2014). Furthermore, Ebbels (2014) suggested there may be benefit to integrating therapy techniques to include grammar facilitation and metalinguistic training in a range of activities (e.g., Fey et al., 2003). Combined ap- proaches are yet to be explored extensively.

Combined Intervention Approaches In an early-stage efficacy study, Finestack (2018)

used a combined implicit/explicit metalinguistic approach compared to an implicit approach to teach novel morphemes to 6- to 8-year-old children with DLD. The combined ap- proach was more efficacious than the implicit approach, with gains being maintained and generalized. In a random- ized controlled trial of 31 preschool-aged children, Smith- Lock, Leitão, Prior, and Nickels (2015) used explicit teaching principles combined with a systematic cueing hierarchy, which was effective in improving use of expressive morpho- syntax when compared to conversational recasting alone. Importantly, the study included a metalinguistic component where children in the explicit group were aware of the thera- peutic goal (Smith-Lock et al., 2015). Kulkarni, Pring, and Ebbels (2013) conducted a clinical evaluation of the SHAPE CODING system combined with elicited production and recasting to improve the use of past tense for two children with DLD aged 8;11 and 9;4. Both made significant gains in their use of the target structure.

Calder et al.: Grammar Intervention in Young Children With DLD 299

Although grammar facilitation is generally considered implicit (Ebbels, 2014; Fey et al., 2003), there is evidence that the techniques can be used explicitly. In a pilot efficacy study, Calder, Claessen, and Leitão (2018) combined the SHAPE CODING system with the systematic cueing hier- archy detailed in Smith-Lock et al. (2015) to improve gram- mar in three children with DLD aged 7 years. Importantly, systematic cueing as a grammar facilitation technique in this study was explicit. Cues ranged from least to most sup- port, and there was a focus on teaching correct production of grammar through errors to avoid the child perceiving the error to be semantic in nature, as may be the case when using conversational recasting without stating the goal of intervention first. The findings provided early evidence supporting the use of combined intervention approaches to improve receptive and expressive grammar, particularly production of regular past tense following 5 weeks of inter- vention. Notably, participants made gains in expressive grammar following only 10 intervention sessions across 5 weeks, which is markedly shorter duration than reported in many intervention studies. However, the authors ac- knowledge that including measures of teaching, maintenance, and generalization (e.g., Finestack, 2018) would have broad- ened understanding of treatment effects and that a longer period of intervention might be necessary.

Grammar Interventions in Clinical Practice Recently, Finestack and Satterlund (2018) reported

on a national survey of speech-language pathology practice in the United States. Past tense verb production was a common intervention goal for practitioners in both early (40%) and elementary education settings (60%). Interest- ingly, overall between 60% and 70% used explicit presenta- tions as an intervention procedure, despite relatively little investigation in this area until recently. Therefore, it appears explicit instruction to improve past tense may not only be supported by an emerging evidence base but is also fre- quently used in clinical practice.

The Current Study For early school-aged children, preliminary data sug-

gest that explicit, combined metalinguistic and grammar facilitation approaches are efficacious in treating the use of tense marking and for improving receptive grammar more generally (Calder et al., 2018). Building on early-stage studies of treatment efficacy is required to determine if treat- ment procedures are considered evidence based. Fey and Finestack (2008) outline the need for a programmatic ap- proach to pursuing intervention research, specifically noting the value of small-scale studies aimed at exploring and identifying specific components of intervention approaches and their effects on specific populations. This study forms a part of a program of research to design, develop, and evaluate the efficacy of an explicit, combined grammar in- tervention in line with Robey’s phases of clinical research (Robey, 2004). We report on a range of measures to evaluate the efficacy of explicit intervention to improve grammar.

Single-case experimental design (SCED) methodology was used to test the following confirmatory hypotheses and is reported as per the Single-Case Reporting guideline In BEhavioural interventions (Tate et al., 2016):

1. For young school-aged children with DLD (specifically aged 5;10–6;8), there will be a significant treatment effect on trained past tense verbs and a generalized effect to untrained verbs across 20 sessions of explicit intervention combining metalinguistic and grammar facilitation techniques.

2. These children will improve significantly on pre– poststandardized measures of expressive and receptive grammar.

Method Research Design Design

The current study was an ABA across-participant multiple-baseline SCED, including a minimum of five data points (i.e., sessions) for each phase (Kratochwill et al., 2012). Multiple baselines were conducted for varied dura- tions across participants, and introduction of treatment to participants was staggered. Repeated measures were col- lected throughout the intervention phase and posttreatment maintenance phase (Dallery & Raiff, 2014), including the target behavior (past tense verbs), an extension of the tar- geted behavior (third-person singular [3S] verbs), and a control behavior (possessive ’s). This design is noted for robustness regarding strengths of internal validity and exter- nal validity when compared to other SCEDs (Tate et al., 2016). As a Phase I–II study, we replicated and built on findings from Calder et al. (2018) by refining intervention protocols, determining optimal dosage, and evaluating du- ration of therapeutic effect (Robey, 2004).

Randomization To improve internal validity further, participants

were randomly assigned to one of three predetermined staggered onset to intervention conditions. To ensure con- cealed allocation, participants were assigned a code that was entered into a random list generator by a blinded re- searcher. Participants received five (P1, P3, P8), seven (P5, P7, P9), or nine (P2, P4, P9) pre-intervention baseline sessions over as many weeks; 20 intervention sessions over 10 weeks; and five postintervention sessions to evaluate maintenance. Participants were also randomized to gram- maticality conditions described below.

Blinding Participant caregivers and teachers were aware chil-

dren were receiving grammar intervention but were blinded to the intervention target. Postintervention measures were collected via blinded assessment using trained student speech- language pathologists (SLPs).

300 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

Participants Selection Criteria

Participants included nine early school-aged children diagnosed with DLD. The inclusion criteria were aged between 5;6 and 7;6, English as a primary language, and grammar difficulties associated with DLD. Exclusionary criteria included a neurological diagnosis, a cognitive im- pairment, and hearing outside normal limits. Participants were recruited from a specialized educational program for students diagnosed with DLD. Ethical approval for the study was obtained from the Curtin University Human Research Ethics Committee (approval number: HRE2017- 0835) and the Western Australian Department of Educa- tion. The principal consented school participation and then provided information letters and consent forms to the parents/carers of potential participants identified by SLPs and teachers employed at the educational program. Parents returned the completed consent forms if they wished their child to participate. The study reached capacity at nine participants so we could achieve three replications over three baseline conditions as per reporting standards (Kratochwill et al., 2012).

Participant Characteristics The participants’ school enrolment package was

accessed, including the assessment protocol and the most recent standardized assessment scores available. Data in- cluded Clinical Evaluation of Language Fundamentals Preschool–Second Edition (Wiig, Secord, & Semel, 2004), a test of nonverbal IQ, and a comprehensive exploration of previous medical history to identify contributing factors to language difficulties, such as acquired neurological dam- age, or hearing loss. These factors combined are consid- ered evidence of a diagnosis for DLD (Bishop, Snowling, Thompson, Greenhalgh, & CATALISE Consortium, 2016). Participants then passed a hearing acuity test. All partici- pants passed the Phonological Probe from the Test of Early Grammatical Impairment (Rice & Wexler, 2001) for artic- ulation of phonemes necessary for morphosyntactic pro- duction targets.

All demographic information is presented in Table 1. Participants included eight boys and one girl aged between 5;10 and 6;8 at initial assessment. Ages at enrolment to

the specialist school varied from 3;8 to 5;11. P1, P2, P4, and P8 were in their third year of placement at the school; P3, P5, P7, and P9 were in their second; and P6 was in her first.

Measures Repeated Measures

Repeated measures of morphosyntax were collected at every data point using various probes, including trained probes, untrained probes, an extension probe, and a control probe (elaborated in the following sections). Probing con- texts included both expressive morphosyntax and grammati- cality judgment. Grammaticality judgment was selected as a method of measuring grammatical progress, as there is evidence performance on such tasks mirrors production tasks (Rice et al., 1998, 1999). As grammaticality judgment is a clinical marker of DLD (Dale et al., 2018; Rice et al., 1999), identification of grammatically correct sentences in the studied participants was expected to be below chance levels of accuracy prior to intervention.

Trained probes. Regular past tense (–ed) repeated measures of trained verbs were probed in two conditions: 12 –ed verbs trained within sessions were measured, and 12 –ed verbs from the previous session were measured. All –ed verbs were predetermined at the outset of intervention and selected based on their suitability to intervention activi- ties. We also chose verbs that were not in the Grammar Elicitation Test (GET; described below; Smith-Lock, Leitão, Lambert, & Nickels, 2013) to allow comparison between trained and untrained verbs. These probes were adminis- tered during the intervention phase at the end of Session 2 (i.e., data point B1 the first week of intervention) and every even session thereafter.

Untrained probes. Repeated measures of untrained expressive morphosyntax probes were selected from an adapted version of the GET. This experimental test was designed to elicit multiple instances of specific expressive morphosyntax targets, including 30 items probing the treated grammatical structure (–ed). Repeated measures were also developed for a grammaticality judgment task including 30 –ed probes. Videos of actions depicting the declarative clauses containing –ed were created as stimuli for untrained probes. Accompanying audio for each task item, both

Table 1. Demographic information.

Participant ID Sex Age at enrolment to school

(years;months) Current year at specialized

educational program Age at initial assessment for study (years;months)

P1 Male 4;0 3rd 6;3 P2 Male 3;11 3rd 6;2 P3 Male 4;7 2nd 5;10 P4 Male 5;4 3rd 6;8 P5 Male 5;2 2nd 6;6 P6 Female 5;11 1st 6;2 P7 Male 5;3 2nd 6;7 P8 Male 3;8 3rd 6;0 P9 Male 4;9 2nd 6;1

Calder et al.: Grammar Intervention in Young Children With DLD 301

grammatical and ungrammatical (e.g., The girl painted a picture. vs. The girl paint* a picture.), was recorded by a woman with an Australian accent, blinded to the purpose of the research. Each video with corresponding audio was embedded into a Microsoft PowerPoint presentation. Participants wore Sony noise-cancelling headphones during administration and were required to decide if the sentence “sounded right” by pressing “yes” or “no” on a tablet app. Items were counterbalanced for grammaticality, so partici- pants did not receive the same combination of grammatical/ ungrammatical items, and there was no pattern in the pre- sentation of grammatical/ungrammatical items to counter- act a priming effect.

Complete sets of 30 untrained –ed verbs were probed pre- and postintervention. Sets were randomized for ad- ministration at the initial assessment (Timepoint 1), 1 week prior to intervention commencing (Timepoint 2), 1 week following intervention (Timepoint 3), and 5 weeks following cessation of intervention (Timepoint 4). Both expression and grammaticality judgment were assessed.

Reduced randomized sets were generated for each other data point using nine expressive probes and 12 gram- maticality judgment probes. All possible allomorphs were included (i.e., [d], [t], and [əd]) and equally distributed. Probes were administered via laptop during the pre-intervention baseline phase, at the beginning of Session 3 (i.e., data point B2 in the second week of intervention) and every odd ses- sion thereafter during the intervention phase, and in the postintervention maintenance phase.

Extension probes. Expressive repeated measures of 3S served as an extension of the treated structure. Items in- cluded 30 probes and were taken from the GET. A gram- maticality judgment task was also developed as per the untrained –ed probes (e.g., The man sneezes. vs. The man sneeze*.). 3S was considered an extension measure due to the structure’s relative complexity compared to –ed, since bare stem forms are grammatical when used with first- person subject pronouns or plural subject nouns (e.g., I like ice cream vs. The boys like ice cream vs. The boy likes ice cream). We also expected there might be improvement in 3S due to the frequent instances of input during therapy (see Intervention section) and increased awareness of the need for tense marking.

Control probes. Similarly, expressive repeated measures of possessive ’s served as a control probe. Items included 30 probes and were taken from the GET. As above, a grammaticality judgment task was developed (e.g., The spider is living on a leaf. This is the spider’s leaf. vs. The spi- der is living on a leaf. This is the spider* leaf.). For ’s, still images of nouns depicting ownership were retrieved from copyright-free image sources. ’s was considered a control as this noun possession was not taught as part of therapy and therefore should remain stable throughout the inter- vention period.

For extension and control probes, all possible allo- morphs were included (i.e., [s], [z], [əz]) and equally distributed. Randomized sets of nine expressive and 12 grammaticality judgment items were generated and administered as per the

untrained –ed probes during pre-intervention, intervention, and postintervention phases.

Pre–Post The Structured Photographic Expressive Language

Test–Third Edition (SPELT-3; Dawson, Stout, & Eyer, 2003) and the Test of Reception of Grammar Version 2 (TROG-2; Bishop, 2003) were administered both pre- and postintervention as expressive and receptive standardized grammar measures, respectively. The SPELT-3 measures expressive morphosyntax using 54 items across a range of structures and was normed on children aged 4–9 years. To address discriminant accuracy of the test, Perona, Plante, and Vance (2005) determined 90% sensitivity and 100% sen- sitivity at 95 cutoff (−0.33 SD). This cutoff score was used for the current study based on the recommendation, al- though it is noted that while other studies applied this cut- off with older children (e.g., Van Horne et al., 2017), Perona et al. (2005) sampled children aged 4–5 years. The TROG-2 measures a total of 20 different grammatical structure con- trasts and was normed on children aged 4–16 years. Dis- criminant accuracy was evaluated on a sample of 30 children aged 6;2–10;11, which confirmed the test is sensitive to identifying communication difficulties in children (Bishop, 2003). Both tests have strong reliability and appropriate validity.

Reliability A blinded researcher scored 20% of all measures au-

dio- and video-recorded throughout the study. Interrater reliability for experimental measures was calculated using intraclass correlation coefficients (ICCs) using absolute agreement and single measures in a two-way mixed-effects model. Interpretation of ICC values is as follows: < .40 = poor, .40–.59 = fair, .60–.74 = good, and .75–1.00 = excel- lent (Cicchetti, 1994). For trained –ed probes, the ICC for expressive measures was .879, and the ICC for grammati- cality judgment was .977. The ICC for expressive un- trained –ed, 3S, and ’s probes was .937, and the ICC for the grammaticality judgment of untrained –ed, 3S, and ’s probes was .985. Therefore, excellent agreement was observed across all experimental measures.

Intervention All intervention sessions were videotaped and carried

out in a quiet space at the site of the educational program. Procedures were similar to those reported by Calder et al. (2018) and are explained within the model suggested by Warren, Fey, and Yoder (2007) for describing treatment intensity. The dose was 50 trials within 20- to 30-min ses- sions; dose form was explicit intervention combining meta- linguistic training using the SHAPE CODING system (Ebbels, 2007) with a systematic cueing hierarchy (Smith- Lock et al., 2015); dose frequency was twice a week, total intervention duration was 10 weeks, and cumulative interven- tion intensity was 50 trials × 2 times per week × 10 weeks, resulting in a total of 1,000 trials over 20 individual

302 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

therapy sessions through roughly 7–10 hr of therapy. This is double the intervention duration in the pilot study (Cal- der et al., 2018), where authors suggested that participants may demonstrate larger treatment effects following a lon- ger duration. Training of morphosyntax was embedded within engaging and naturalistic activities suited to early school-aged children, including playdough, board games, and playing with puppets as well as farm and sea creature manipulatives. Target morphemes were presented in syntactic structures as they occurred felicitously within these activities. The first author (S. D. C.), a trained SLP, delivered all interventions.

Each session began with a short recap of the aims: to say WHAT DOING words (verbs) that have already happened and to add the sounds ([d], [t], [ǝd]) onto the end of those words. Next, the SLP would direct the child’s atten- tion to the laminated shapes and arrows used as a visual organizer throughout session activities. See Figure 1 for es- sential shapes, including the oval (subject noun phrase WHO/ WHAT?), the hexagon (verb phrase WHAT DOING?), and the rectangle (object noun phrase WHO/WHAT?). Additional visual cues included three separate laminated cards that depicted a “left down arrow” to depict –ed and an orthographic representation of the allomorphs (i.e., “d” for [d], “t” for [t], and “ed” for [ǝd]). The SLP said, “Last time, we used our shapes and arrows to help us. Like this: ‘We move our shapes and arrows. What did we do? We moved [bring ‘d’ arrow into the WHAT DOING? hexagon] our shapes and arrows. The [d] at the end of moved lets us know it’s already happened.” The participant was reminded, “I (the SLP) will say what we do in the session (i.e., present tense) and you will say what we did (i.e., past tense).” This was followed by two activities that were designed to give the participants ample opportunities to produce –ed verbs in response to an interrogative (e.g., What did you do? Did you just VERB? Tell me….).

Each activity began with explicit instruction of how to apply –ed inflection, using one exemplar from each of the allomorphic categories. Within each activity, there were approximately 25 opportunities for the child to respond to an interrogative (e.g., You roll the playdough! What did you do?) using –ed verbs while the SLP gestured to the shapes and arrows (see Figure 1). The child was therefore encour- aged to respond using a subject–verb–object syntactic frame, consistently. If the child responded with an unmarked verb (i.e., bare stem) or overgeneralized form (e.g., playded ), he or she was supported with a systematic cueing hierarchy moving from least to most support outlined in Figure 2.

As much as possible, verbs were blocked according to allo- morphs and presented from least to most difficult (i.e., [d]➔[t]➔[ǝd]) in accordance with Leonard (2014) and Marshall and van der Lely (2006). At the end of every ac- tivity, the SLP recapped what the participant had learned using the shapes and arrows and comprehension questions. For example, if the target sentence had been “I rolled play- dough,” the SLP would gesture to the WHO?/WHAT? oval and ask, “WHO rolled the playdough?” Then, gesture to the WHAT DOING? hexagon while bringing down the “d” left down arrow and ask, “What DID you DO?” and finally gesture to the WHO?/WHAT? rectangle and ask, “WHAT did you roll?” A plausible response to all of these questions is “I rolled the playdough,” giving further oppor- tunity to reinforce production using a consistent syntactic frame. If an error occurred, the same systematic cueing hierarchy described above was employed. The shapes and arrows were then removed, and the interrogative (What DID you DO?) was repeated without visual support for an exem- plar from all three allomorphic categories, reinforcing in- ternalization of the grammatical rule. If a child had achieved 80% success over three sessions on any measure, “silly sentences” were introduced, a metalinguistic subactivity whereby three sentences were said, either grammatically or ungrammatically (i.e., –ed morphemes were either included or omitted), and the child would decide if the sentence “sounded right.”

These procedures were repeated for a second activ- ity, giving 50 opportunities to use –ed inflection during the activity that was bookended with explicit teaching and comprehension questions using three exemplars from each allomorphic category. At the end of each session, the child was reminded of the goal of the session and why it is im- portant to say the sounds at the end of “WHAT DOING?” words that have already happened and to listen out for those sounds.

Procedural Fidelity A blinded researcher scored 20% of videotaped ses-

sions on percentage accuracy using a priori established cri- teria for intervention procedures. A total of 19 items were scored for sessions (see Appendix A for a checklist for scoring intervention procedure fidelity). Note that if chil- dren were introduced to “silly sentences,” sessions were scored against an additional two (total 21) items. Intra- observer agreement was calculated using ICC. The aver- age score was 97.1% for percentage accuracy, and the ICC for treatment procedures was .996.

Figure 1. Visual depiction of visual cues used during the intervention phase.

Calder et al.: Grammar Intervention in Young Children With DLD 303

Analysis Single-Subject Analyses

Treatment effects of teaching, generalization, and maintenance through repeated measures of morphosyntax were statistically evaluated using Tau-U by combining nonoverlap and trend of data (Parker, Vannest, Davis, & Sauber, 2011) across all phases and data points. Tau-U uses Kendall’s S to interpret significance testing and outputs p values. Raw scores on probes were converted to percent- age correct. Baselines were contrasted using the Tau-U online calculator (Vannest, Parker, Gonen, & Adiguzel, 2016), and the Tau value was checked for trend of baseline in pre- and postintervention phases. For pre-intervention baseline, Tau values above 0.40 (increasing trend) or below −0.40 (decreasing trend) were deemed unstable and cor- rected, as recommended by Parker et al. (2011). This was repeated for all applicable baseline versus intervention con- trasts. Finally, phase contrasts were aggregated to provide an omnibus effect size for study participants, where, using Cohen’s standard, 0.2 is small, 0.5 is medium, and 0.8 is large.

To evaluate performance on the full sets of untrained –ed verbs, a concurrent within-group approach was used (e.g., Zwitserlood et al., 2015) where Friedman nonpara- metric two-way analysis of variance (ANOVA) tested dif- ferences between Timepoints 1 and 2 pre-intervention and Timepoints 3 and 4 postintervention scores. Participant scores determined a group mean and standard deviation in expressive and grammaticality judgment probes within each timepoint. Post hoc Wilcoxon signed-ranks tests made pairwise comparisons between testing points. These statistics were computed using IBM SPSS Version 25.

Kratochwill et al. (2012) outline standards for analysis of repeated measures via visual inspection to report on a functional relation between dependent and independent var- iables, which includes comments on level, trend, and vari- ability within phases and comments on immediacy, overlap, and consistency between phases. For the current study, within-phase level performance was evaluated with group

statistics. Furthermore, Tau-U handles within-phase level, and trend and variability within and between phases, as well as overlap between phases. Therefore, reporting on visual inspection is limited to the immediacy of the func- tional relation between –ed use and understanding, and the staggered introduction of intervention across participants.

Pre–postanalyses. Pre–postdifferences on standard- ized measures were tested in a case series approach by calculating the Reliable Change Index (RCI; Unicomb, Colyvas, Harrison, & Hewat, 2015). The RCI statistic cal- culates whether an individual’s change in score (i.e., pre– postdifference in standard scores) is statistically significant by using the reliability values of a standardized test. The RCI is calculated using the formula x2 – x1 / Sdiff, where x1 is the participant’s pretest score, x2 is the same participant’s posttest score, and Sdiff is the standard error of difference between the two test scores. An RCI above 1.96 is consid- ered statistically significant at 0.05 significance level.

Results Sequence Completed

All participants completed planned sessions within pre-intervention baseline (A), intervention (B), and post- intervention maintenance (A) phases. There was an average of 50.74 (SD = 1.2, range: 48–56) trials for each participant to produce –ed. Out of the nine participants, six (P1, P2, P3, P4, P5, and P7) demonstrated at or above 80% perfor- mance on at least one measure of –ed marking over three sessions. These participants engaged in the “silly sentences” aspect of intervention procedures as described in the Inter- vention section.

Outcomes and Estimation Single-Subject Treatment Effects (Expressive)

Data not reported in tables are available in the Sup- plemental Materials (see list of Supplemental Materials in Appendix B). Pre-intervention baselines on production

Figure 2. Systematic cueing hierarchy used when the child produced the target verb in error.

304 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

of –ed verbs taken from the GET were stable for 4/9 par- ticipants. P1 (Tau = −0.70), P3 (Tau = −0.70), P4 (Tau = 0.58), P8 (Tau = 0.60), and P9 (Tau = −0.71) had baselines corrected for subsequent analyses. Data from expressive repeated measures are presented in Figures 3–5, and results from Tau-U analyses are reported in Table 2. Of the nine participants, eight (P1–P7 and P9) demonstrated statisti- cally significant trend in production of trained verbs tested within session during the intervention phase (see Figure 3). Phase contrasts were combined and yielded an aggregated effect size of 0.88, which is considered large. For trained verbs tested between sessions (see Figure 4), seven (P1–P5,

P7, and P9) of nine participants demonstrated statistically significant performance during the intervention phase, with a large aggregated effect size of 0.83. Seven (P1–P7) of nine participants demonstrated a statistically significant trend in production of untrained –ed verbs during the intervention phase (see Figure 5), yielding a medium effect size of 0.64.

Analysis of Tau scores revealed a significant negative trend in performance for P1 (Tau = −0.40), P6 (Tau = −0.40), and P7 (Tau = −0.40) across five data points in the post- intervention maintenance phase. Note the Tau values for these three participants are at the minimum level for baseline trend (Tau = ± 0.40) corrections according to Parker et al. (2011).

Figure 3. Percentage correct on expressive trained within-session probe repeated measures for Groups 1–3.

Calder et al.: Grammar Intervention in Young Children With DLD 305

For expressive 3S extension probes, P7 (Tau = 0.62), P8 (Tau = 0.60), and P9 (Tau = 0.57) demonstrated an unstable baseline with a positive trend. During the inter- vention phase, P6 demonstrated significant improvement (p = .03), and P9 demonstrated significant decline (p = .03). Phase contrasts yielded a nonsignificant (p = .65) aggre- gated effect size of −0.05. P1 (Tau = 0.80), P2 (Tau = 0.40), and P4 (Tau = 0.70) demonstrated a positive trend in the postintervention maintenance phase.

For expressive ’s control probes, P2 (Tau = 0.69) and P4 (Tau = 0.61) showed unstable baselines with positive trends, while P9 (Tau = −0.43) showed an unstable base- line with a negative trend. Of the nine participants, both P1 (p = .013) and P3 (p = .004) demonstrated significant improvement during the intervention phase. Phase con- trasts yielded a nonsignificant (p = .33) aggregated effect size of 0.10. P5 (Tau = 0.40) continued to show a positive trend in the postintervention maintenance phase, while P7

Figure 4. Percentage correct on expressive trained between-session probe repeated measures for Groups 1–3.

306 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

(Tau = −0.50), P8 (Tau = −0.40), and P9 (Tau = −0.40) showed a negative trend.

Single-Subject Treatment Effects (Grammaticality Judgment)

Pre-intervention baselines for past tense grammati- cality judgment probes were stable for all participants. Only one participant (P5) improved significantly in correctly judging grammaticality on trained verbs tested within ses- sions (p = .02). P1 (p = .04) and P4 (p = .04) improved sig- nificantly on trained verbs tested between sessions, and a

small (0.26) yet significant (p = .009) effect size across participants was calculated. Only one (P2) participant demonstrated a significant trend in correct grammaticality judgment of untrained –ed verbs during the intervention phase (p = .02).

For grammaticality judgment 3S extension probes, P8 showed an unstable baseline with a negative trend, Tau = −0.40. P4 demonstrated a significant improvement during intervention (p = .02), and P8 demonstrated a significant negative trend (p = .02). P2 (Tau = −0.80). Phase contrasts yielded a small, yet significant (p = .03) aggregated effect

Figure 5. Percentage correct on expressive untrained probe repeated measures for Groups 1–3.

Calder et al.: Grammar Intervention in Young Children With DLD 307

size of 0.22. P8 (Tau = −0.40) demonstrated a negative trend in the maintenance phase, while P3 (Tau = 0.53) demon- strated a positive trend.

For grammaticality judgment ’s control probes, P4 demonstrated a negative trend, while P7 (Tau =0.65) and P8 (Tau =0.90) demonstrated a positive trend during base- line. P2 demonstrated a significant positive trend during intervention (p = .02). Phase contrasts yielded a nonsignifi- cant (p = .76) aggregated effect size of 0.03. P4 demonstrated a negative trend in the maintenance phase, Tau = −0.40.

Within-Group Concurrent Approach Mean scores and standard deviations for –ed pro-

duction and grammaticality judgment at four timepoints are presented in Table 3. A Friedman two-way ANOVA demon- strated that the production of untrained –ed verbs differed significantly between timepoints, χ2(F = 22.47), df =3,

p < .001. Post hoc Wilcoxon signed-ranks tests and a Bonferroni-adjusted α of .0167 (0.05/3 comparisons: Time- point 1 vs. Timepoint 2, Timepoint 2 vs. Timepoint 3, and Timepoint 3 vs. Timepoint 4) showed –ed production was significantly higher at Timepoint 3 (mean rank = 3.78) than at Timepoint 2 (mean rank= 1.56), z = –2.67, N-Ties = 9, p = .008. Differences between other timepoints were non- significant, suggesting a stable pre-intervention baseline, an observable treatment effect between pre- and postinterven- tion testing points, and a maintenance of gains at a group level. Tests for grammaticality judgment were nonsignificant.

Analysis of Pre–Postresults Pre- and postintervention standard scores on the

SPELT-3 and the TROG-2 are reported in Table 4. Ex- ceeding the RCI of 1.96 indicates statistically significant improvement. All but one participant (P6) exceeded the

Table 2. Summary of expressive repeated-measures baseline versus treatment phase contrasts on trained and untrained targets.

Participant ID Kendall’s S z score p value Tau 90% CI

Within session P1a 55 3.37 < .001* 1.1 [0.56, 1] P2 88 3.60 < .001* 0.98 [0.53, 1] P3a 51 3.12 .002* 1.02 [0.48, 1] P4a 69 2.82 .005* 0.77 [0.32, 1] P5 70 3.42 < .001* 1 [0.52, 1] P6 66 2.70 .007* 0.73 [0.29, 1] P7 56 2.73 .006* 0.80 [0.32, 1] P8a 15 0.92 .36 0.30 [0.24, 0.84] P9a 85 4.15 < .001* 1.21 [0.73, 1]

Aggregated ES Group — — < .001* 0.88 —

Between session P1a 57 3.49 < .001* 1.14 [0.60, 1] P2 88 3.59 < .001* 0.98 [0.53, 1] P3a 57 3.49 < .001* 1.14 [0.60, 1] P4a 57 2.33 .02* 0.63 [0.19, 1] P5 70 3.42 < .001* 1.00 [0.52, 1] P6 37 1.51 .13 0.41 [−0.04, 0.86] P7 48 2.34 .02* 0.69 [0.20, 1] P8a 15 0.92 .36 0.30 [−0.24, 0.84] P9a 85 4.13 < .001* 1.21 [0.73, 1]

Aggregated ES Group — — < .001* 0.83 —

Untrained P1a 40 2.67 .007* 0.89 [0.34,1] P2 79 3.49 < .001* 0.98 [0.52, 1] P3a 30 2.00 .05* 0.67 [0.12, 1] P4a 56 2.47 .01* 0.69 [0.23, 1] P5 45 2.38 .02* 0.71 [0.22, 1] P6 73 3.22 .001* 0.90 [0.44,1] P7 44 2.33 .02* 0.70 [0.21, 1] P8a 13 0.87 .39 0.29 [0.26, 0.84] P9a −8 −0.42 .67 −0.13 [0.62, 0.37]

Aggregated ES Group — — < .001 0.64 —

Note. Em dashes indicate data not available. CI = confidence interval; ES = effect size. aUnstable baseline corrected.

*Significant.

308 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

RCI for the SPELT-3. Furthermore, for the majority of participants, postintervention standard scores exceeded the manual-reported confidence intervals (90% and 95%) around their pre-intervention standard scores. Note, however, that even though P1’s RCI was significant, his post-SPELT-3 standard score of 76 does not exceed the 90% and 95% confidence intervals around his pre-SPELT-3 standard score of 69. One participant (P8) exceeded the RCI for the TROG-2 (2.12).

Adverse Events In the case of absence during the intervention phase,

participants (P5, P6, P7, P8, and P9) attended a make-up ses- sion in the final week of intervention in which within-session and between-session teaching probes were collected. Due to issues with attention and engagement, procedural changes occurred for P6, who received 30 trials per session, and the systematic cueing hierarchy was limited to elicited imitation.

Discussion This study evaluated the efficacy of an explicit gram-

mar intervention combining metalinguistic training and grammar facilitation aimed to improve regular past tense (–ed) marking for nine children with DLD aged 5;10–6;8. Intervention taught –ed marking through explicit rule

instruction and visual supports using the SHAPE COD- ING system. A systematic cueing hierarchy (Smith-Lock et al., 2015) was used to support participants. This study con- tributes to the design, development, and evaluation of inter- vention efficacy by moving through levels of evidence and analogous research designs (Robey, 2004).

Treatment Effects Single-Subject Analyses

We hypothesized participants would improve signifi- cantly on –ed verbs trained and probed within sessions and between sessions. Most participants improved on expres- sive repeated measures of trained verbs with large effects, indicating this intervention is efficacious for improving pro- duction of –ed verbs taught in sessions. Furthermore, most participants improved on untrained verbs with medium effects, suggesting generalization. Within-group Friedman nonparametric two-way ANOVA also demonstrated a gen- eralized treatment effect, which was maintained for 5 weeks. For grammaticality judgment, only three participants im- proved on trained verbs, one improved significantly on un- trained verbs, and another continued to improve 5 weeks postintervention. Few gains were observed across partici- pants on an extension measure (3S) and on control mea- sures of ’s both production and grammaticality judgment. Limited progress on control probes strengthens our ability to

Table 3. Mean scores on complete sets of untrained past tense verbs across four timepoints.

Measure

Pre-intervention Postintervention

Timepoint 1 Timepoint 2 Timepoint 3 Timepoint 4

Expressive (/30) 7.44 (SD = 4) 7.44 (SD = 5.47)† 22.89 (SD = 5.97)* 21.89 (SD = 7.23)††

Grammaticality judgment (/30) 15.22 (SD = 1.87) 16.22 (SD = 1.03)† 19.25 (SD = 4.97) 18.78 (SD = 6.25)

*Significant difference between pre- and postintervention timepoints = observed treatment effect. †Nonsignificant difference between pre-intervention baseline timepoints = stable baseline. ††Nonsignificant difference between postintervention timepoints = maintained treatment effect.

Table 4. Pre- and postintervention standard scores.

Participant ID

SPELT-3 TROG-2

Pre-intervention Postintervention Pre-intervention Postintervention

P1 69 76 (2.78)* 74 76 (0.24) P2 90 111 (9.33)* 97 95 (0.24) P3 79 102 (6.83)* 86 93 (0.83) P4 71 105 (13.54)* 81 83 (0.24) P5 57 90 (13.14)* 81 86 (0.35) P6 72 78 (0.64) 65 58 (−0.83) P7 84 100 (6.37)* 62 74 (1.42) P8 69 88 (7.54)* 79 97 (2.12)* P9 57 78 (8.33)* 65 67 (0.24)

Note. Scores are standard scores with a mean of 100 and SDs of 15. SPELT-3 = Structured Photographic Expressive Language Test–Third Edition; TROG-2 = Test of Reception of Grammar Version 2.

*Statistically significant, that is, above 1.96.

Calder et al.: Grammar Intervention in Young Children With DLD 309

attribute improvement on –ed production to intervention. Results support the efficacy of intervention to improve –ed production on trained and untrained verbs; however, we ob- served limited gains on grammaticality judgment measures.

Visual inspection of expressive repeated measures reflects results from statistical analysis regarding the imme- diacy of the functional relation between –ed production and intervention. That is, a positive trend is observable upon the staggered introduction of intervention across participants. Specifically, trained expressive probes appeared to improve more rapidly, as early as Week 1 of intervention, whereas for untrained verbs, gains are observable around the 5-week mark across participants. Finally, visual inspection revealed production of –ed on untrained verbs remained relatively stable for all children during the postintervention phase, supporting findings from a within-group statistical analysis.

Pre–Postcomparisons Pre–postcomparisons of standard measures of ex-

pressive and receptive grammar across participants mirrored single-subject analyses. Of the nine participants, eight exceeded the RCI for expressive grammar, and one child exceeded the RCI for receptive grammar. Overall, pre– postanalyses suggest the intervention had a broad effect on expressive grammar captured through standardized grammar measures. However, effects on measures of grammar com- prehension were modest compared to expressive grammar.

General Discussion Results from the current study support and build upon

findings in the literature. Finestack (2018) demonstrated the efficacy of explicit–implicit instruction using novel morphemes, suggesting that the experimental approach may yield quicker gains, and improvement closer to mastery compared to existing implicit-only intervention procedures. Furthermore, Finestack called for an evaluation of treat- ment effectiveness using true English morphemes across measures of maintenance and generalization to progress the clinical applicability of research findings. Calder et al. (2018) piloted the intervention with a small group of early school- aged children diagnosed with DLD. Findings suggested intervention implemented over 5 weeks, twice per week with- out predefined dosage, improved –ed production of untrained verbs and standard measures of expressive and receptive grammar. The authors concluded maintaining the consis- tent dosage (i.e., 50 trials) and extending the duration (i.e., 10 weeks) may improve production on untrained verbs and discern optimal dose to allow replication for clinical practice.

The current study applied recommended changes to intervention dose and intensity, and predictions were sup- ported. Furthermore, using measures of verbs trained in session and those from previous sessions allowed the analysis of within- and between-session gains (e.g., Finestack, 2018). We saw that children showed greater and more rapid im- provement on trained verbs probed within and between sessions compared to untrained verbs. However, gains in

standard measures of receptive grammar were not observed to the extent reported in Calder et al. (2018). It is likely that reduced improvement on the measure is attributable to the baseline performance of the participants from the current study. That is, the baseline scores of the current group of participants were higher than those reported in Calder et al., which may suggest fewer gains were to be made on such a measure. This finding is consistent with literature suggesting that receptive grammar is less amenable to improvement when compared to expressive grammar (Ebbels, 2014).

From a theoretical perspective, limited improvement on receptive measures may be due to the status of inter- nal representations of language remaining relatively fixed. However, increased production practice may establish new representations, such as those practiced within sessions, which are generalizable to similar targets, such as other verbs marked for –ed or 3S. This pattern was observed with two participants (P2 and P4, respectively), so future research is needed to explore this claim further. Alternatively, the current standard measures of receptive grammar may fall short of their aim. Recently, Frizelle, Thompson, Duta, & Bishop (2019) found multiple-choice grammar tasks may underesti- mate children’s abilities compared to truth-value tasks. In the current study, probing grammaticality judgment of trained and untrained verbs allowed investigation of im- provement of obligatory tense marking as a specific behav- ior, although improvement was limited across participants. This may provide evidence of the persistent nature of lan- guage disorder (e.g., Dale et al., 2018). Alternatively, the task may be implicated by other cognitive factors, such as phonological short-term memory. Regardless, further re- search is needed to unpack effective methods to treat recep- tive language difficulties.

Current findings are comparable to recent studies targeting –ed marking in children with DLD. For example, in a study using similar procedures to the current study, Smith-Lock et al. (2015) demonstrated explicit rule instruc- tion coupled with a systematic cueing hierarchy was more effective in improving morphosyntax in preschool children with DLD when compared to recasting alone. A key differ- ence to intervention procedures reported in this study is the inclusion of visual metalinguistic training and the explicit use of the cueing hierarchy. That is, cues in this study were presented to highlight that the targeted behavior was not observed, and so the children were encouraged to reflect on the rule they had just been taught with the support of visuals and to self-correct. Furthermore, the current study implemented over double the cumulative intensity than Smith- Lock et al. (2015), although trials were not specified in that study, so it is challenging to make direct comparisons. Finally, Van Horne et al. (2017) reported positive treatment outcomes following intervention targeting –ed production. Importantly, the primarily implicit intervention procedures outlined in Van Horne et al. were effective in improving –ed for both studied groups following 36 sessions, which is markedly longer than dose duration reported here and by Smith-Lock et al. (2015), suggesting that explicit inter- ventions may be more time efficient in improving expressive

310 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

grammar outcomes. Future research is needed to compare the superiority of the two approaches to intervention.

This study further extends on a body of research eval- uating the efficacy and effectiveness of explicit interventions using visual support strategies to improve grammatical knowledge for children with language difficulties, specifi- cally the SHAPE CODING system (Ebbels, 2007). Positive results of use of the system have been reported with older children with DLD (Ebbels et al., 2014, 2007; Kulkarni et al., 2013), younger children with DLD (Calder et al., 2018), and children with complex learning needs (Tobin & Ebbels, 2019). It should be noted that positive results were reported by Finestack (2018) where metalinguistic training without visual support was efficacious in improving grammar in young children with DLD. Continued research in this area will discern the extent to which the visual aspect of the SHAPE CODING system is responsible for positive treat- ment effects.

We saw that children showed greater and more rapid improvement on verbs trained in session when compared to untrained verbs, suggesting children with DLD may have difficulty generalizing grammar skills, particularly those relying upon sequence learning, such as finiteness marking. Therefore, we are more likely to see immediate improvement in verbs trained via intervention compared to untrained verbs. We also expected there might have been improvement on verbs marked for 3S; however, this was not widespread across participants, with P6 improving during intervention and three (P1, P2, and P4) improving postintervention. This finding suggests that, generally, grammar targets should be taught directly, even if they are linguistically related to existing intervention targets for children with DLD. Further- more, production practice did not seem to affect grammati- cality judgment; however, metalinguistic training may have. That is, regardless of practice trials being held consistent, children for whom “silly sentences” were introduced (P1, P2, P3, P4, P5, P7) appeared to perform better on repeated measures of grammaticality judgment (see Supplemental Materials S10, S11, and S12). Therefore, introducing the subactivity at the onset of treatment, rather than awaiting the 80% accuracy criterion, may result in improvement of grammaticality judgment.

Other factors to consider when evaluating treatment effectiveness are environmental. For example, the partici- pant with the lowest performance in general (P6) had attended the specialist school for the least amount of time, compared to P2 and P4, the strongest performers who were in their third year at the specialist school. It could be that these children were primed to learn during language-based tasks more so than P6. However, P6 also had the lowest pre-intervention language scores and received fewer trials throughout the intervention phase. Nonetheless, P6 still im- proved significantly despite these potential barriers. Through SCEDs, evaluating individual treatment responses allows researchers and clinicians to extricate factors related to re- sponsiveness to intervention that may otherwise be lost in group treatment studies (Plante, Tucci, Nicholas, Arizmendi, & Vance, 2018).

Limitations There are limitations to this study. Firstly, generaliz-

ability of results using SCED must be applied with caution. Although the methodology allows for analysis of treatment effects for individuals, the lack of a control group and rela- tively small sample size inhibit the ability to make causal inferences regarding treatment effectiveness in relation to the general population. Furthermore, within-participant analysis does not control for the influence of external factors, such as classroom instruction, when compared to robust randomized group comparison studies. Nonetheless, SCEDs provide a useful methodology for establishing an early evidence base for newly developed interventions (Fey & Finestack, 2008). In fact, Horner et al. (2005) suggest re- sults from a minimum of five studies totaling at least 20 par- ticipants across three different research teams are necessary to determine intervention efficacy using high-quality SCEDs prior to effectiveness being tested using clinical trials. The current study was designed using guidelines developed by Kratochwill et al. (2012) and Tate et al. (2016) to meet minimum standards for SCED to interpret treatment efficacy. Note that an independent rater did not collect repeated measures within the baseline and intervention phases as per Kratochwill et al.’s recommendation. However, strong interrater reliability values addressed potential observer bias. Secondly, the current study used convenience sampling to recruit participants from a specialized school designed to provide intensive language and literacy support to young children with DLD. While nonverbal IQ was not directly measured as part of this study, all participants were enrolled into an educational program for children with DLD in the presence of average nonverbal IQ. Furthermore, the socio- economic status of participants was unknown, and the majority (8/9) of participants were male. Therefore, the current sample may not be representative of the population of children with DLD at large. Lastly, the current efficacy study was limited to the analysis of –ed production and grammaticality judgment and standard expressive and re- ceptive grammar scores. More naturalistic measures, such as narrative or conversation sampling, may better serve as true measures of generalization in future studies.

Clinical Implications A recent survey of U.S. speech pathologists investi-

gating current clinical practices for grammar intervention found that, although a regular component of practice, spe- cific aspects of grammatical interventions are not well un- derstood (Finestack & Satterlund, 2018). Furthermore, –ed marking is often targeted as a treatment goal, and explicit presentation is often used in intervention procedures. How- ever, relatively little research has been reported using ex- plicit intervention for teaching –ed to early school-aged children. Fey and Finestack (2008) proposed a framework for conceptualizing intervention components. The current intervention is summarized in Table 5. This framework may serve as a point of reference for clinicians planning to

Calder et al.: Grammar Intervention in Young Children With DLD 311

implement intervention to improve production of –ed for early school-aged children with DLD. Clear intervention procedures and maintaining consistent dose throughout the intervention phase also allow clinicians to replicate findings. It appears generally that this intervention is less efficacious for improving grammaticality judgment of –ed, with only a small intervention effect (0.26) observed. However, a sim- ilar effect (0.22) was observed for grammaticality judgment of 3S but not for the production or grammaticality judg- ment of ’s. Since 3S was not targeted directly but is linguis- tically related, perhaps improvement for some children was due to the phonological saliency of /z, s/ compared to –ed /d, t/, providing a learning advantage to the morpheme when combined with metalinguistic training.

Conclusions Results continue to support the efficacy of explicit

grammar interventions to improve –ed marking in early school-aged children. Future research should continue to evaluate the efficacy of similar interventions, for example, using more clinically relevant dosage (e.g., one session per week). It is also important to determine whether explicit grammar interventions can improve other aspects of gram- matical difficulty for younger children with DLD, such as copula/auxiliary use or wh-questions. Overall, findings contribute to the understanding of efficacious intervention procedures for early school-aged children with DLD, sug- gesting children are able to apply knowledge acquired through explicit instruction.

Acknowledgments The authors acknowledge the contribution of an Australian

Government Research Training Program Scholarship in supporting this research.

The authors thank the specialized school, participants, and student speech-language pathologists Taylor Gartner, Kelly Mackean, Jenni Brennan, Becky Balchin, and Siobhan Kavanagh involved in this study. They also thank Lizz Hill as the blinded assessor and Karen Smith-Lock for supporting the use of the Grammar Elicitation

Test and sharing intervention sessions to adapt for the purpose of this study.

References Bishop, D. (2003). Test for Reception of Grammar Version 2

(TROG-2). The Psychological Corporation. Bishop, D. V. M., Adams, C. V., & Norbury, C. F. (2006). Distinct

genetic influences on grammar and phonological short-term memory deficits: Evidence from 6-year-old twins. Genes, Brain and Behavior, 5(2), 158–169. https://doi.org/10.1111/j.1601-183X. 2005.00148.x

Bishop, D. V., Bright, P., James, C., Bishop, S. J., & Van der Lely, H. K. (2000). Grammatical SLI: A distinct subtype of developmental language impairment? Applied Psycholinguistics, 21(2), 159–181. https://doi.org/10.1017/S0142716400002010

Bishop, D. V. M., & Hayiou-Thomas, M. E. (2008). Heritability of specific language impairment depends on diagnostic criteria. Genes, Brain and Behavior, 7(3), 365–372. https://doi.org/10.1111/ j.1601-183X.2007.00360.x

Bishop, D. V. M., Snowling, M. J., Thompson, P. A., Greenhalgh, T., & CATALISE-2 Consortium. (2017). Phase 2 of CATALISE: A multinational and multidisciplinary Delphi consensus study of problems with language development: Terminology. The Journal of Child Psychology and Psychiatry, 58(10), 1068–1080. https://doi.org/10.1111/jcpp.12721

Bishop, D. V. M., Snowling, M. J., Thompson, P. A., Greenhalgh, T., & CATALISE Consortium. (2016). CATALISE: A multi- national and multidisciplinary Delphi consensus study. Identi- fying language impairments in children. PLOS ONE, 11(7), e0158753. https://doi.org/10.1371/journal.pone.0158753

Calder, S. D., Claessen, M., & Leitão, S. (2018). Combining im- plicit and explicit intervention approaches to target grammar in young children with developmental language disorder. Child Language Teaching and Therapy, 34(2), 171–189. https://doi.org/ 10.1177/0265659017735392

Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6(4), 284–290. https:// doi.org/10.1037/1040-3590.6.4.284

Cleave, P. L., Becker, S. D., Curran, M. K., Van Horne, A. J., & Fey, M. E. (2015). The efficacy of recasts in language interven- tion: A systematic review and meta-analysis. American Journal of Speech-Language Pathology, 24(2), 237–255. https://doi.org/ 10.1044/2015_AJSLP-14-0105

Table 5. Framework for conceptualizing intervention components proposed by Fey and Finestack (2008).

Intervention component Experimental intervention

Children Children with DLD aged 5;10–6;8 Goals Regular past tense (–ed) production and grammaticality judgment Service delivery 1:1 with a speech-language pathologist in clinical contexts (within

a specialized school) Dosage 50 trials, two× sessions per week for 10 weeks: 1,000 trials over

20 sessions and approximately 7–10 hr of intervention Procedures Explicit intervention using metalinguistic training with visual support

combined with a systematic cueing hierarchy Activities Naturalistic games with opportunities to produce –ed verbs (e.g.,

playdough, puppets, board games) Measurement of outcomes Standard grammar measures and criterion-referenced measures

of –ed production and grammaticality judgment

Note. DLD = developmental language disorder.

312 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

Dale, P. S., Rice, M. L., Rimfeld, K., & Hayiou-Thomas, M. E. (2018). Grammar clinical marker yields substantial heritability for language impairments in 16-year-old twins. Journal of Speech, Language, and Hearing Research, 61(1), 66–78. https:// doi.org/10.1044/2017_JSLHR-L-16-0364

Dallery, J., & Raiff, B. R. (2014). Optimizing behavioral health interventions with single-case designs: From development to dissemination. Translational Behavioral Medicine, 4(3), 290–303. https://doi.org/10.1007/s13142-014-0258-z

Dawson, J. I., Stout, C. E., & Eyer, J. A. (2003). Structured Photo- graphic Expressive Language Test–Third Edition (SPELT-3). Janelle Publications.

Ebbels, S. (2007). Teaching grammar to school-aged children with specific language impairment using shape coding. Child Language Teaching and Therapy, 23(1), 67–93. https://doi.org/10.1191/ 0265659007072143

Ebbels, S. (2014). Effectiveness of intervention for grammar in school-aged children with primary language impairments: A review of the evidence. Child Language Teaching and Therapy, 30(1), 7–40. http://doi.org/10.1177/0265659013512321

Ebbels, S. H., Marić, N., Murphy, A., & Turner, G. (2014). Im- proving comprehension in adolescents with severe receptive language impairments: A randomized control trial of interven- tion for coordinating conjunctions. International Journal of Language & Communication Disorders, 49(1), 30–48. https:// doi.org/10.1111/1460-6984.12047

Ebbels, S. H., van der Lely, H. K., & Dockrell, J. E. (2007). Inter- vention for verb argument structure in children with persistent SLI: A randomized control trial. Journal of Speech, Language, and Hearing Research, 50(5), 1330–1349. https://doi.org/10.1044/ 1092-4388(2007/093)

Fey, M. E., & Finestack, L. H. (2008). Research and develop- ment in child language intervention: A five-phase model. In R. G. Schwartz (Ed.), Handbook of child language disorders (pp. 513–531). Psychology Press.

Fey, M. E., Long, S. H., & Finestack, L. H. (2003). Ten principles of grammar facilitation for children with specific language impairments. American Journal of Speech-Language Pa- thology, 12(1), 3–15. https://doi.org/10.1044/1058-0360(2003/ 048)

Finestack, L. H. (2018). Evaluation of an explicit intervention to teach novel grammatical forms to children with developmental language disorder. Journal of Speech, Language, and Hearing Research, 61(8), 2062–2075. https://doi.org/10.1044/2018_ JSLHR-L-17-0339

Finestack, L. H., & Satterlund, K. E. (2018). Current practice of child grammar intervention: A survey of speech-language pathol- ogists. American Journal of Speech-Language Pathology, 27(4), 1329–1351. https://doi.org/10.1044/2018_AJSLP-17-0168

Frizelle, P., Thompson, P., Duta, M., & Bishop, D. V. (2019). Assessing children’s understanding of complex syntax: A com- parison of two methods. Language Learning, 69(2), 255–291. https://doi.org/10.1111/lang.12332

Horner, R. H., Carr, E. G., Halle, J., McGee, G., Odom, S., & Wolery, M. (2005). The use of single-subject research to identify evidence-based practice in special education. Exceptional Chil- dren, 71(2), 165–179. https://doi.org/10.1177/001440290507100203

Justice, L. M., Logan, J., Jiang, H., & Schmitt, M. B. (2017). Algorithm-Driven Dosage Decisions (AD3): Optimizing treat- ment for children with language impairment. American Journal of Speech-Language Pathology, 26(1), 57–68. https://doi.org/ 10.1044/2016_AJSLP-15-0058

Kratochwill, T. R., Hitchcock, J. H., Horner, R. H., Levin, J. R., Odom, S. L., Rindskopf, D. M., & Shadish, W. R. (2012).

Single-case intervention research design standards. Remedial and Special Education, 34(1), 26–38. https://doi.org/10.1177/ 0741932512452794

Kulkarni, A., Pring, T., & Ebbels, S. (2013). Evaluating the effective- ness of therapy based around shape coding to develop the use of regular past tense morphemes in two children with language impairments. Child Language Teaching and Therapy, 30(3), 245–254. https://doi.org/10.1177/0265659013514982

Leonard, L. B. (2014). Children with specific language impairment. MIT Press.

Leonard, L. B., Camarata, S. M., Brown, B., & Camarata, M. N. (2004). Tense and agreement in the speech of children with specific language impairment: Patterns of generalization through intervention. Journal of Speech, Language, and Hearing Research, 47(6), 1363–1379. https://doi.org/10.1044/1092-4388(2004/102)

Marshall, C. R., & van der Lely, H. K. J. (2006). A challenge to current models of past tense inflection: The impact of phono- tactics. Cognition, 100(2), 302–320. https://doi.org/10.1016/j. cognition.2005.06.001

Nelson, K. E., Camarata, S. M., Welsh, J., Butkovsky, L., & Camarata, M. (1996). Effects of imitative and conversational recasting treatment on the acquisition of grammar in children with specific language impairment and younger language- normal children. Journal of Speech and Hearing Research, 39(4), 850–859. https://doi.org/10.1044/jshr.3904.850

Parker, R. I., Vannest, K. J., Davis, J. L., & Sauber, S. B. (2011). Combining nonoverlap and trend for single-case research: Tau-U. Behavior Therapy, 42(2), 284–299. https://doi.org/10.1016/ j.beth.2010.08.006

Perona, K., Plante, E., & Vance, R. (2005). Diagnostic accuracy of the Structured Photographic Expressive Language Test–Third Edition (SPELT-3). Language, Speech, and Hearing Services in Schools, 36(2), 103–115. https://doi.org/10.1044/0161-1461 (2005/010)

Plante, E., Tucci, A., Nicholas, K., Arizmendi, G. D., & Vance, R. (2018). Effective use of auditory bombardment as a therapy ad- junct for children with developmental language disorders. Lan- guage, Speech, and Hearing Services in Schools, 49(2), 320–333. https://doi.org/10.1044/2017_LSHSS-17-0077

Rice, M., & Wexler, K. (2001). Test of Early Grammatical Impair- ment. The Psychological Corporation.

Rice, M. L., Wexler, K., & Hershberger, S. (1998). Tense over time: The longitudinal course of tense acquisition in children with specific language impairment. Journal of Speech, Language, and Hearing Research, 41(6), 1412–1431. https://doi.org/10. 1044/jslhr.4106.1412

Rice, M. L., Wexler, K., & Redmond, S. M. (1999). Grammati- cality judgements of an extended optional infinitive grammar: Evidence from English-speaking children with specific lan- guage impairment. Journal of Speech, Language, and Hearing Research, 42(4), 943–961. https://doi.org/10.1044/jslhr.4204.943

Robey, R. R. (2004). A five-phase model for clinical-outcome re- search. Journal of Communication Disorders, 37(5), 401–411. https://doi.org/10.1016/j.jcomdis.2004.04.003

Shanks, D. R. (2005). Implicit learning. In K. Lamberts & R. Goldstone (Eds.), Handbook of cognition (pp. 202–220). Sage.

Smith-Lock, K. M., Leitão, S., Lambert, L., & Nickels, L. (2013). Effective intervention for expressive grammar in children with specific language impairment. International Journal of Language & Communication Disorders, 48(3), 265–282. https:// doi.org/10.1111/1460-6984.12003

Smith-Lock, K. M., Leitão, S., Prior, P., & Nickels, L. (2015). The effectiveness of two grammar treatment procedures for

Calder et al.: Grammar Intervention in Young Children With DLD 313

children with SLI: A randomized clinical trial. Language, Speech, and Hearing Services in Schools, 46(4), 312–324. https://doi.org/ 10.1044/2015_LSHSS-14-0041

Tate, R. L., Perdices, M., Rosenkoetter, U., Shadish, W., Vohra, S., Barlow, D. H., . . . Willson, B. (2016). The Single-Case Reporting guideline In BEhavioural interventions (SCRIBE) 2016 statement. Physical Therapy, 96(7), e1–e10. https://doi.org/ 10.2522/ptj.2016.96.7.e1

Tobin, L. M., & Ebbels, S. H. (2019). Effectiveness of interven- tion with visual templates targeting tense and plural agree- ment in copula and auxiliary structures in school-aged children with complex needs: A pilot study. Clinical Linguistics & Phonetics, 33(1–2), 175–190. https://doi.org/10.1080/02699206. 2018.1501608

Unicomb, R., Colyvas, K., Harrison, E., & Hewat, S. (2015). Assessment of reliable change using 95% credible intervals for the differences in proportions: A statistical analysis for case- study methodology. Journal of Speech, Language, and Hearing Research, 58(3), 728–739. https://doi.org/10.1044/2015_JSLHR-L- 14-0158

Van Horne, A. J. O., Fey, M., & Curran, M. (2017). Do the hard things first: A randomized controlled trial testing the effects of exemplar selection on generalization following therapy for grammatical morphology. Journal of Speech, Language, and

Hearing Research, 60(9), 2569–2588. https://doi.org/10.1044/ 2017_JSLHR-L-17-0001

Vannest, K. J., Parker, R. I., Gonen, O., & Adiguzel, T. (2016). Single case research: Web based calculators for SCR analysis (Version 2.0) [Web-based application]. Texas A&M Univer- sity. Retrieved from http://singlecaseresearch.org

Warren, S. F., Fey, M. E., & Yoder, P. J. (2007). Differential treat- ment intensity research: A missing link to creating optimally effective communication interventions. Mental Retardation and Developmental Disabilities Research Reviews, 13(1), 70–77. https://doi.org//10.1002/mrdd.20139

Weismer, S. E., & Murray-Branch, J. (1989). Modeling versus modeling plus evoked production training: A comparison of two language intervention methods. Journal of Speech and Hearing Disorders, 54(2), 269–281. https://doi.org/10.1044/jshd.5402.269

Wiig, E. H., Secord, W., & Semel, E. M. (2004). Clinical Evalua- tion of Language Fundamentals Preschool–Second Edition. Pearson/The Psychological Corporation.

Zwitserlood, R., Wijnen, F., van Weerdenburg, M., & Verhoeven, L. (2015). ‘MetaTaal’: Enhancing complex syntax in children with specific language impairment—A metalinguistic and multi- modal approach. International Journal of Language & Com- munication Disorders, 50(3), 273–297. https://doi.org/10.1111/ 1460-6984.12131

314 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

Appendix A

Checklist for Scoring Intervention Procedure Fidelity

Step Explanation 1/0

1. Explicit teaching Remind the child of the goal of the session. 1a. Activate prior knowledge ❒ 1b. Explain goals ❒

ACTIVITY 1 2. Check vocabulary Child asked to label materials from session linked to subject/object nouns ❒ 3. Goal Demonstrate 3× SV/O sentences using one exemplar from each of the allomorphic categories.

Introduce “left down arrow cues” each alongside its corresponding shape. ❒

4. Practice 25 trials to produce past tense –ed with systematic cueing 4a. Coding Lay large shapes on the floor and student to use as cues to produce SV/O sentences. ❒ 4b. Trials 22–28 trials achieved ❒ 4c. Cueing Are errors cued appropriately? ❒

5. Consolidation At the end of the session, review the 3× SV/O sentences using one exemplar from each of the allomorphic category.

5a. Comprehension task Student to produce subjects, verbs, and objects following comprehension questions ❒ 5b. Production Student says phrase ❒ 5c. Repeat without shapes Student says phrase (cue as necessary) ❒

ACTIVITY 2 6. Check vocabulary Child asked to label materials from session linked to subject/object nouns ❒ 7. Goal Demonstrate 3× SV/O sentences using one exemplar from each of the allomorphic categories.

Introduce “left down arrow cues” each alongside its corresponding shape. ❒

8. Practice 25 trials to produce past tense –ed with systematic cueing 8a. Coding Lay large shapes on the floor and student to use as cues to produce SV/O sentences. ❒ 8b. Trials 22–28 trials achieved ❒ 8c. Cueing Are errors cued appropriately? ❒

9. Consolidation At the end of the session, review the 3× SV/O sentences using one exemplar from each of the allomorphic category.

9a. Comprehension task Student to produce subjects, verbs, and objects following comprehension questions ❒ 9b. Production Student says phrase ❒ 9c. Repeat without shapes Student says phrase (cue as necessary) ❒

10. Summarize Remind the child of the goal of the session. ❒

Total: /19 Percentage accuracy: %

Note. SV/O = subject–verb–object.

Calder et al.: Grammar Intervention in Young Children With DLD 315

Appendix B

List of Supplemental Materials

S1: Expressive raw scores of participants on trained past tense verbs within-session. S2: Expressive raw scores of participants on trained past tense verbs between-session. S3: Expressive raw scores of participants on untrained past tense verbs. S4: Expressive scores of participants on third-person singular (extension). S5: Summary of Tau-U analyses for expressive repeated measures baseline versus treatment phase contrasts on untrained third-person singular targets (extension). S6: Graph of % correct on expressive third-person singular repeated measures (extension). S7: Expressive raw scores of participants on possessive ’s (control). S8: Summary of expressive repeated measures baseline versus treatment phase contrasts on untrained possessive ’s targets (control). S9: Graph of % correct on expressive possessive ’s repeated measures (control). S10: Grammaticality judgment raw scores of participants on trained past tense verbs within session. S11: Grammaticality judgment raw scores of participants on trained past tense verbs between-session. S12: Grammaticality judgment raw scores of participants on untrained past tense verbs. S13: Summary of grammaticality judgment repeated measures baseline versus treatment phase contrasts on trained and untrained targets. S14: Graph of % correct on grammaticality judgment within-session repeated measures. S15: Graph of % correct on grammaticality judgment between-session repeated measures. S16: Graph of % correct on expressive untrained repeated measures. S17: Grammaticality judgment raw scores of participants on third-person singular (extension). S18: Summary grammaticality judgment repeated measures baseline versus treatment phase contrasts on untrained third-person singular targets (extension). S19: Graph of % correct on grammaticality judgment third-person singular repeated measures (extension). S20: Grammaticality judgment raw scores of participants on possessive ’s (control). S21: Summary of grammaticality judgment repeated measures baseline versus treatment phase contrasts on untrained possessive ’s targets (control). S22: Graph of % correct on grammaticality judgment possessive ’s repeated measures (control).

316 Language, Speech, and Hearing Services in Schools • Vol. 51 • 298–316 • April 2020

Copyright of Language, Speech & Hearing Services in Schools is the property of American Speech-Language-Hearing Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.