Abstract
Intervention studies targeting developmental language disorder (DLD) have been conducted mainly with monolingual English-speaking children and bilingual children learning English. There is a need to explore the application of current methods to more languages. However, adapting an existing intervention to other languages can be a daunting task as it is not necessarily clear which aspects need modification. This article presents pilot studies that were conducted to explore the applicability of dialogic book reading (DBR) to three languages and countries, and to clarify methodological aspects of the intervention in preparation for the development of a protocol for a larger international efficacy study to be conducted across several languages and countries. The article describes an iterative process conducted by an international team, which included group work on the creation and adaptation of intervention materials, as well as three feasibility clinical trials conducted at different stages of the iterative process in Bulgaria, Austria and Lebanon. Each trial addressed methodological aspects of DBR implementation, including the number and types of books to use in total and per session, the length of sessions, the number and type of target words in total and per session and the types of probes used to keep track of intervention gains. In addition to shedding light on methodological aspects of the implementation of DBR, the study describes the iterative process undertaken by the research team, which could serve as a model or inspiration to other international teams.
Keywords
I. Introduction
The vast majority of studies on developmental language disorder (DLD, previously specific language impairment [SLI]) have focused on English-speaking children. Although the number of cross-linguistic studies on DLD has increased, their primary focus has been on the manifestation of DLD and its identification in languages with different typologies, often testing theoretical accounts originally based on English data by examining error patterns in oral (e.g., Hansson and Nettelbladt, 1995 for Swedish; Scheffer et al., 2022 for Dutch; Taha et al., 2021 for Arabic) and written language (e.g., Kuvač Kraljević et al., 2022 for Croatian, Mackie and Dockrell, 2004 for English). The dominance of English is even greater in the area of intervention. Even intervention studies on bilingual children overwhelmingly include children whose L2 is English (Harvey et al., 2018; Pillinger and Vardy, 2022; Rinaldi et al., 2021). One reason for the scarcity of studies on non-English languages may be that some studies are published only in non-English languages (e.g., Eydal et al., 2019; Michallet and Boudreault, 2014) which limits their impact on the internationally recognised state of art of the field. However, a more important reason may be related to the marked differences in professional traditions across countries. While studies on the underlying nature of DLD are often conducted by linguists or psycholinguists, intervention efficacy studies are more likely to be conducted by researchers with clinical training in speech language therapy (SLT) or who collaborate with SLTs. Doctoral training in SLT has been offered in many English-speaking countries such as the UK, the USA, Canada and Australia for decades, however not in most European countries (National Vignette chapters in Law et al., 2019a). Indeed, the professional field of Speech Language Therapy (also called Speech Language Pathology and Logopedy) has only recently been brought to the university level across European countries by the Bologna Agreement (NetQues Project Report, 2013). The present study thus responds to a need for developing methods of checking the applicability of existing interventions to non-English languages with the necessary adaptations as well as for stimulating clinical SLT research across European countries in which such research does not have a strong tradition. Greater international participation in intervention studies not only provides evidence of whether existing methods are efficacious in these countries, but also, with time, can lead to the creation of methods that include more varied perspectives. The main aim of this study was to conduct pilot studies to explore cross-linguistic adaptations of a dialogic book reading (DBR) method and to clarify methodological aspects with a view to developing a common protocol to be run in several countries in a later study.
In selecting a method to adapt across countries, it is relevant to consider which types of methods are currently used and familiar to clinicians in those countries. Clinical intervention for DLD has used various methods that vary in scope. They may stimulate language broadly or focus on a specific area such as vocabulary or morphology. Target selection has been based on theoretical accounts that vary, for example, in whether domains of language are considered to be modular (e.g., Chomsky, 1980; Fodor, 1983) or non-modular (e.g., Marchman and Bates, 1994). In addition, clinical interventions differ in the extent to which therapy sessions are directed by the clinician versus follow the child's attention and pace (Fey, 1986). A large survey of European practitioners conducted by COST Action IS1406 (see Law et al., 2019a; Law et al., 2019b) found significant differences across countries in the extent to which practitioners have a clear theoretical focus in their work and also in their preferred methods (Saldaña and Murphy, 2019). Vignettes from each of the 39 participating countries revealed, for example, that Vygotskian methods, involving naturalistic interactions, first published on in Russian, were used earlier in Eastern than Western European countries. Overall, Vygotskian and Piagetian methods had the greatest number of supporters across Europe, followed by the socio-pragmatic approach (Law et al., 2019a).
Beyond their theoretical orientation, intervention methods are also coloured by prevailing patterns of interaction in the culture in which they are used. The way in which adults speak to children and the way in which children are expected to respond varies from culture to culture (e.g., van Kleeck, 1994; Verdon et al., 2015). The need to consider cultural aspects when designing and conducting an SLT intervention is widely recognised (see e.g., Kayser, 1996; Kohnert, 2010; Paul et al., 2017). Therefore, it cannot be assumed without verification that an intervention that leads to appropriate clinical gains in one culture will lead to similar gains in another culture. Indeed, even interventions grounded in the same theoretical underpinnings may need to be implemented differently in different cultural contexts to achieve similar language gains in children with DLD. Nevertheless, it is also likely that many interventions that have not yet been tried out or had their efficacy examined outside of an English-speaking context may, in fact, work well in many other contexts.
The present research was conducted by an international research group that was part of a European Union-funded COST Action that had as its central goal to study DLD across countries of Europe and to enhance DLD research in countries where it had previously been relatively scarce. COST Actions function by bringing together experienced and emerging researchers as well as professionals from all European countries and neighbouring countries to stimulate knowledge transfer and collaborations. Action members interested in intervention study design and evidence-based intervention attended a training school within the COST Action taught by the last author of this article. There they developed the idea of jointly conducting an international intervention efficacy study, similar to other cross-country studies conducted as part of the Action, such as practitioner and public surveys (Law et al., 2019b; Elin Thordardottir et al., 2021). After a week of sharing many views from the different countries represented, the idea was advanced to attempt to conduct an international efficacy study. Recognising the importance of cultural similarities and differences, the group reached a consensus that the DBR procedure was a promising approach to implement across the participating countries, given the widespread use of book-reading in these countries and the method’s balance between structure and naturalistic interaction. The hypothesis was that the structured aspect would facilitate comparable implementation across countries, while the naturalistic interaction would provide some flexibility for culturally relevant interactions. The end goal of the undertaking described in this article was both to test the feasibility of developing a research intervention protocol that could be carried out across many countries, and to gauge the applicability of the intervention approach across the countries and languages by assessing how it was received by clinicians and families. As the group embarked on this process, however, a number of questions arose regarding methods and the process of adapting the method to other languages and ensuring its consistent implementation by clinicians who were not familiar with it. This article describes an iterative process that developed gradually through these exchanges and that culminated in the development of important aspects of the international protocol.
1. Dialogic book reading
Shared reading interventions include a large set of types of activities that are centred around book reading. They vary in their aims, being used to stimulate various aspects of oral language (e.g., Restrepo et al., 2013) or to promote emergent literacy skills (e.g., Chrifi and Maillart, 2018) and can use various methods such as reading out loud, reading with errors, direct teaching of target words, the use of signs, images using low and high tech methods, and of object manipulation depending on the needs of the children being worked with (Godin et al., 2015; Grevesse et al., 2020). In this study, the focus was on oral vocabulary development, using DBR(Whitehurst et al., 1988), a specific reading activity to boost oral language, based on the assumption that children need to use language and receive feedback on this use through adult–child interaction to promote their language development with the adult providing scaffolding to support the child's learning using the natural context of storytelling and the child's active participation in it. The basic idea, based on a Vygotskian interactionist framework, is that support from an adult leads to language development that goes beyond spontaneous development (Whitehurst and Lonigan, 1998), and that the child's active participation in reading leads to greater gains than passively listening to the text being read (Mol et al., 2008). DBR can be used to target various oral language goals. As such, in this method, the book serves as a vehicle or catalyst for a conversation during which a clinician or a parent, while reading a book, uses naturalistic prompts and response techniques to support and expand the child´s language abilities. By participating in the activity, the child is exposed to and brought by the adult conducting the intervention to use words and linguistic structures in a meaningful and interesting activity rather than in a contrived teaching activity (Zevenbergen and Whitehurst, 2003). The responses and prompts provided by the clinicians are also consistent with a naturalistic approach (Fey, 1986).
Books can be chosen based on their language content, for example in terms of the specific vocabulary, specific morphology or sentence structures. In English, lists of books are available that guide clinicians to childreńs books that are particularly suitable for working on specific clinical targets (Paul et al., 2017). One of the questions facing our working group was to find books that are applicable across cultures and languages to be included in the study.
Numerous studies have been conducted describing the DBR procedure or investigating its efficacy in different English-speaking populations of children with language difficulty, including Down syndrome (e.g., Jordan et al., 2011; Quinn et al., 2020), autism spectrum disorder (ASD) mostly using the modified DBR intervention because of these children's verbal and social behaviours, modifications in the context of delivery and modifications of prompts etc. (e.g., Fleury and Schwartz, 2017; Grygas Coogle et al., 2018). DBR has also been used with children with limited vocabulary (e.g., Hargrave and Sénéchal, 2000), and DLD (Lavelli et al., 2019). Moreover, studies on DBR have been conducted in various settings, for example, a group setting (Jordan et al., 2011), a parent–child interaction (Lavelli et al., 2019) and have targeted various populations, age ranges and language skills. DBR has also been used in studies involving bilingual children who speak English and Spanish (Restrepo et al., 2013; Tsybina and Eriks-Brophy, 2010). A few studies have also been published on the use of shared book reading interventions with monolingual children who speak non-English languages. At the time the study was conducted, only a few such studies were found, some of which did not include children with DLD (Desmarais et al., 2013; Elmonayer, 2013; Opel et al., 2009). A systematic review published after the present study was conducted included 46 studies conducted in the United States, Hong Kong, Canada, Australia, Turkey, the UK, Egypt, Bangladesh, Brazil, Mexico and South Africa on the effect of DBR on the language abilities of children aged 2–9 years. Of the 46 studies, 35 were conducted in English and only 11 in a non-English language (Arabic, Bangla, Chinese and Spanish). The studies included a wide range of study design and a variety of language structures that were targeted. They also included a large range of populations, among them typically developing children, dual language learners, deaf and hard-of hearing children, children at risk for language delay and children with confirmed developmental disorders. Results showed that DBR is effective in improving a wide range of language skills and reading outcomes, both in English and in the other languages that were part of the studies. Also, it was shown that DBR can improve the motivation of children under 5 years for joint activities with the parents, the positive experience of reading books, and parental confidence (Pillinger and Vardy, 2022). Other recently published studies in different languages have shown a positive effect of DBR on preschool children's inferential skills (a study in German, with some children speaking other languages, such as Turkish, English, French, Polish, Russian and Spanish; Grolig et al., 2020), receptive vocabulary, listening comprehension, reading interest (a study in Chinese; Dong et al., 2024) and expressive language (a study conducted in linguistically and culturally diverse country, Indonesia; Asrifan et al., 2022).
This review of research shows the flexibility of DBR in targeting various language domains across various populations and age groups. At the same time, although the studies are consistent in that they use books to target language structures using naturalistic prompts to scaffold the child's learning, they vary greatly in many methodological aspects. Thus, the literature does not provide clear guidelines as to how to ideally implement the DBR approach for particular age groups or populations or for specific linguistic targets. Notably, studies differed in terms of the number and types of books used, the criteria for their selection as well as the implementation of the protocol in terms of length and number of sessions, the number or density of prompts in relation to reading text. For example, Hargrave and Sénéchal (2000) using DBR for emergent literacy training, selected 10 books based on the number of illustrations, text length, age and cultural appropriateness, presence of rhyme, and prior exposure. The intervention lasted four weeks (20 sessions). The order of presentation of the books was not specified. Lavelli et al. (2019), examining the impact of parent-based shared book reading on the conversational engagement and linguistic production of Italian-speaking children with DLD, used eight books in an 8-week intervention. The only reported criterion for book selection was that they contain lots of pictures and very little text. Parents read the same book four times a week. Inconsistencies were also found in the overall number of books and whether the same book was used across multiple sessions. These differences are likely to be due to differences in the populations served with varying needs stemming, for example, from age level and the severity of the language difficulty or disorder. With an eye to developing a common DBR protocol to eventually be administered across several countries and languages to older preschoolers with DLD, the international group of researchers went through a series of steps which are described below and include literature reviews, group discussions and group work, and clinical feasibility trials.
II. Methods – A series of trial stages
A team of researchers including a larger team of 10 researchers (6 of which also have clinical SLT training) from seven countries (referred to in Figure 1 as the larger working group) embarked on a process of exploration to decide on aspects of the methodology of their planned international study, such as the age range to be targeted and inclusionary and exclusionary criteria. This iterative process included several stages which are depicted in Figure 1, including group work, educated guesses and trial and error, starting with the emergence of the idea for an international intervention study in a training school on EBP organised by COST Action IS1406 held in March 2017 to the development of a final protocol for such a study. Following the tradition of the Action, the idea was subsequently pitched to all interested members of the Action at the next Action meeting, held in Limerick in May 2017 with the understanding that representatives from all countries were invited to participate. Subsequently, interested members communicated via e-mail to organise the next steps. A special DBR group meeting was held in Zagreb in the fall of 2017, attended by around 10 COST Action members who were committed to participating in conducting the study (larger group). The pilot studies reported on here were undertaken by a smaller group (referred to as pilot group) made up of researchers most of whom were also members of the larger team.
Illustration of the decision-making process for the DBR study.
1. Zagreb work meeting, September 2017
At this meeting, the larger group discussed aspects of the eventual international study planned as the end goal of this project, including the target population, age range, and methods of identifying DLD across languages. Tentative decisions were also made regarding the length of the study in terms of number of weeks, based on previous intervention literature and feasibility based on clinical traditions and structure of service in each of the countries. Each participating country was to be required to include at least one child, for a single-subject-with-replication study design. The child was to be of older preschool age (4;00–5;6 years of age; not yet in school) and to have been identified as having DLD. Acknowledging that the participating countries vary in the availability of diagnostic tests, it was agreed that each country would at the least ensure a good description of the language abilities of the participants using the test traditions of their country. Another criterion was that the child should not have received SLT intervention within the three months immediately preceding the study. Vocabulary progress was to be tracked through comprehension and/or production probes administered in each session. Vocabulary was selected as the target as it is easier to track across languages than are grammatical structures that vary across languages and whose developmental trajectory also varies across languages leading to different manifestations or error patterns across languages leading to differences in appropriate target selection (Stavrakaki et al., 2011; Elin Thordardottir, 2015). In addition to being able to target specific structures, DBR also involves conversations and story-telling. It was envisioned that possible broader gains in language skills could be measured with standardised tests and narrative language samples pre- and post-treatment. The team tentatively selected lists of age-appropriate target vocabulary items based on word lists such as the MacArthur-Bates Communicative Development Inventories which are available across multiple languages (Fenson et al., 2006), clinical experience and previous studies conducting vocabulary interventions with children with DLD in a similar age range (e.g., Restrepo et al., 2013;Elin Thordardottir et al., 2015).
The specific DBR prompting procedures to be used in the planned study were those of Zevenbergen and Whitehurst (2003). For younger children (2–3 years), the adult asks what questions, follows answers with questions, repeats, helps when needed, praises, encourages, pursues interests and promotes fun. For older children (4–5 years), two acronyms summarise the prompts: PEER (Prompt, Evaluate, Expand, Repeat) and CROWD (Completion, Recall, Open-ended, Wh-, and Distancing prompts). The specific prompts need to be adapted to the specific language skills targeted by the intervention. In this study, we used the PEER and CROWD prompts for older preschoolers.
Beyond these decisions of the larger group, a series of questions emerged needing further discussion and pilot testing before the final protocol could be formulated:
Books. What is the appropriate content level for the target age group? How many books should be used in total and within each session? How much time should be spent on the DBR activity within each session? Should the same book be used in multiple sessions? The group also discussed whether it would be better to use paper books or digital books but settled on paper books. Vocabulary. Which and how many words should be targeted? Here, a balance was needed between a sufficient number of words while avoiding overloading the child. Articulatory complexity of each word in the different languages was also discussed as was the fact that the words needed to have translation equivalents in each of the languages being targeted and to be of an appropriate difficulty level for older preschoolers with DLD. Probes. Is it feasible to use a probe to track progress? When should the probe be used, after each session or after several sessions? Should it target both production and comprehension? Procedure. Once the procedure has been established, how can clinicians be trained to ensure a consistent delivery method? Dosage and duration. How many intervention sessions and of what duration? How many vocabulary items should be targeted per session and how many times should each be presented during each session and overall? For example, Restrepo et al. (2013) targeted 5 words per book. In their first 3 weeks, a new book was used over multiple sessions in the same week, with a new book introduced in the following week. Participants. Should the international study target monolingual or bilingual children or both?
The following sections outline the gradual decision-making process of the international team and describes three pre-experimental case trials carried out in Bulgaria, Austria and Lebanon, deliberately varying in scope and structure as they were conducted at different stages of the process.
a. Development of DBR prompts
In clinical interventions using DBR, clinicians may choose books that address the specific treatment goals of an individual child. However, DBR is also frequently used in groups and therefore, must suit the needs of a group of children. In our case, a set of books was to be used in a common protocol to be delivered in clinics across several countries. Many books are translated into many languages, and many are about animal characters or things that appear familiar to children in many cultures. At the Zagreb meeting, members of the larger group were tasked with bringing books that they thought would be suitable for the intervention based on previous group discussions in meetings and through on-line correspondence. The group examined the books together and started to work in smaller teams on selecting potential vocabulary items that could be targeted using particular books. The team discussed the appropriate complexity level, the types of vocabulary items and the number of words to target. First, based on the items on vocabulary tests and clinical expertise regarding words that are functional for young children and are frequently targeted in intervention with preschoolers with DLD, the team tentatively agreed on four vocabulary categories with the idea that each of these could be targeted in a series of books: emotions, household-related vocabulary, vocabulary related to problem solving and school readiness. Given that verbs have been found to be more difficult to learn than nouns by young children with DLD (Svaldi et al., 2024), a tentative decision was made to include nouns, verbs and adjectives in order to include a range of difficulty. The selection of words was also discussed in terms of the articulatory complexity of each word in the different languages and whether they had translation equivalents in the different languages.
Starting with an initial set of books brought in by different team members and subsequently agreed on by the whole team as culturally appropriate for all countries involved, the larger team was divided into groups, each working on developing specific target words and prompts (Whitehurst et al., 1988; Zevenbergen and Whitehurst, 2003) with each team focusing on separate word categories. Several issues were raised in the group discussions, including the age appropriateness of certain words and the appropriateness of targeting the basic meaning or figurative meaning of words (e.g., ‘a grey morning’). It also became clear in this work session that the team members’ familiarity with the procedure varied greatly. The reasons for this are manifold. One of them is that representatives of some of the countries were not themselves SLTs but researchers who ultimately worked with clinicians in their country to carry out the study. Another reason is that some of the clinicians on the team were more familiar with more clinician-directed methods and while intrigued by a more naturalistic approach were not experienced in using it. Subsequently, time was spent familiarising the team members with the intervention and practicing implementing the DBR reading and prompting procedure (Zevenbergen and Whitehurst, 2003) focusing on formulating prompts, finding opportunities for prompting while reading a book, planning such opportunities ahead of time, delivering the prompts in a lively and engaging manner and reacting to child utterances elicited by the prompting with additional naturalistic responses. The last author, a researcher and certified SLT, conducted demonstration sessions to the group and group members were directed to view demonstration videos of the approach available on the internet. Given this familiarisation, part of the reason for the clinical case trials was to test the feasibility of conducting this type of intervention in the different countries. During this group work and familiarisation sessions, the idea of developing scripted prompts for each book to be used was born. Scripts would ensure uniform administration and would also serve as a teaching tool demonstrating the procedure to those less familiar with it.
b. Development of test probes
Various target outcome measures for the eventual intervention study were discussed. The outcome measure needed to provide a coherent and consistent way of measuring progress across sessions. A probe was developed by the pilot team representing each target word by a picture differing from those used in the books. The probes were to be administered in a receptive and productive form at the beginning and end of each therapy session, thus assessing progress within each session and retention between sessions. The production probe was to be administered before the receptive probe to avoid modelling the words before the production probe. A page with all target words was used for the receptive assessment with the child being asked to point to the word spoken by the examiner. For the productive part, the picture of the target word was, for some items, elicited by a short sentence/phrase, for example “Birds can …” (target word: fly).
2. Period in between the Zagreb and Vienna meetings
During the period between the Zagreb and Vienna work meetings (see Figure 1), aspects of the DBR method to be used were tested in clinical case trials conducted by members of the pilot team in three countries, Bulgaria, Lebanon and Austria as representatives from these countries either conducted the trial themselves or had connections with clinics in which the intervention could be tried out. The three trials are different in structure reflecting that they were carried out at different points in time and that at the time they were conducted, the work on selecting books and associated target words was at different stages.
The Bulgarian trial used a book that was subsequently not included because the trial indicated that it was not suitable for the age range targeted. The Lebanese and Austrian trials included books that were subsequently retained. Differences between the trials also reflect different clinical traditions across countries regarding diagnostic procedures, as had been expected.
Ethics approval for the study was obtained from the Ethics Committee of the Center for Interdisciplinary Research in Rehabilitation of the Greater Montreal (CRIR) which comprised analysis of the international data, which other countries used to obtain necessary ethics approval in their country. Each team was responsible for administrating and obtaining the consent forms from the parents of participating children.
III. Trial 1: Bulgaria
This first trial involved only one book. It took place in a university clinic in Sofia and was conducted in Bulgarian, with a single child as participant. Beyond the basic goal of trying out the procedure with a Bulgarian child, the aim of trial 1 was to check the complexity of the selected words that refer to household items and how the child with DLD engaged in the DBR intervention. The intervention was carried out by an SLT in an individual work room within one session of 20 min. The SLT was a member of the pilot team described above and trained during team meetings. The trial also tested the proposed procedure for documenting the child's DLD diagnosis and profile given the tests used in Bulgarian clinics.
The DBR procedure was implemented within one session, with the main goal of assessing the appropriateness of the book (content, child's level of interest) and the prompting procedure. Specific prompts followed the scripts developed ahead by the members of the Bulgarian research team. The presentation was done in two phases: (1) The SLT read the book to the child without emphasising the target words and without prompts. (2) The book was read a second time using the script with the prompts. The specific prompts and the child's responses of the child are listed in Table 1. They included fill in the blank prompts, wh-prompts, recall prompts and a distancing prompt. In each presentation of the book, the target words were presented once or twice (towel: 2, exercise 2, hoop 1, help 2, yard 2, burn 1). Due to time constraints, it was not possible to conduct baseline testing before the intervention.
Target words, specific prompts used in the presentation of the Bear book and child's responses.
Target words for each book in the Austrian study.
1. Continued group work developing scripts and assessment probes
Over the next few months, work on selecting target vocabulary and writing scripts continued by members of the larger group of researchers working in smaller groups on selecting target vocabulary and embedding it in books with various prompts, with a continued focus on the four vocabulary categories. In terms of the size of the target word pool, precise guidelines were hard to find in the literature for DBR specifically. It was observed, however, that in vocabulary intervention studies with preschool children using DBR and focused stimulation methods (Restrepo et al., 2013; Elin Thordardottir et al., 2015), the intervention groups increased their productive vocabulary by an average of 15–20 words over 12–14 sessions, with individual children's gains ranging up to 30 words. Based on these considerations, the group tentatively agreed on a target word pool of 18 words and focused on a set of three books to use in further trials. In this version of the protocol, a total of 18 target words were to be targeted in the three books, with six words being targeted in each book. The books targeted three of the original four vocabulary categories: emotions, problem solving, and household items. It was further decided that each target word would be modelled 10 times per session/book and that each word would be targeted using three different types of prompts. In order to ensure a reliable procedure involving modelling presentation of the target words with consistent use of DBR prompts in all languages, specific examples of modelling and prompts were developed by the whole group in English, which were then translated and adapted into other languages.
The smaller working groups of the larger team of researchers and particularly those of the pilot team continued to develop scripts for prompts to use in each of the three books being developed. Some challenges were encountered in fitting the same vocabulary across all three books and in adapting specific books to each country for example because words did not always have direct translation equivalents across the languages. The work of the smaller working groups was shared and discussed by the larger group. This work was coordinated by the Canadian team, and due to the time-consuming nature of this work, parts of the scripts were further developed by a research assistant on that team, who also proposed additional words that could fit the pictures and story of each book, as well as short additions to the texts of the books in order to fit the desired vocabulary. The research assistant also developed an assessment probe for the target words with a picture depicting each word. Additional discussions took place on the adaptability of words across languages. The scripts and assessment probes for the three books were finalised in January 2018. These results of the larger team were then used by the smaller pilot team in their clinical trials.
IV. Trial 2: Austria
This trial used the three books for which scripts were finalised in early 2018 (see Tables 1 and 2). It was conducted in German in a private SLT practice in Vienna, Austria. Parents of previous clients received information on the study and a questionnaire to check whether their child met the study inclusionary criteria.
These German standardised assessments were administered: (1) ‘Psycholinguistische Analyse kindlicher Aussprachestörungen-II’ (PLAKSS-II; Fox-Boyer, 2014), which indicated no Speech Sound Disorder (SSD); (2) A vocabulary test for children aged three to five (‘Aktiver Wortschatztest für 3- bis 5-jährige Kinder – Revision (AWST-R)’; Kiese-Himmel, 2005), which placed the semantic and lexical performance clearly below average (T-Value = 32, percentile rank = 3); (3) A test for grammar comprehension (‘Test zur Überprüfung des Grammatikverständnisses’ (TROG-D); Fox-Boyer, 2016), showed a T-value of 40 and a percentile rank of 15 – results just within the average range.
Although the scripts and assessment probes had been developed and the number of target words had been tentatively decided, the appropriate timeline of the approach was still open to question. The Austrian trial tested a particular intervention schedule that was found to work within its service delivery tradition and the constraints of the clinic and participating family. Two SLTs conducted the sessions, usually twice a week for 8 weeks and altogether 13 sessions (assessment, probes and intervention sessions). Table 3 shows the timetable of the intervention procedure including the administration of the standardised assessment test sessions, the probes and the intervention sessions. Baseline assessment used receptive and productive probes which had been developed by the international team to later compare the child's performance from before and after the intervention. Once the intervention started, only production probes were administered.
Timetable of the Austrian pilot study procedure.
Only one session took place in week 3 due to illness of the SLT; therefore, two books were covered in the next session to achieve the desired exposure to each book (hence the prolonged duration of the session).
This trial also addressed another aspect needing clarification: the number of times to use each book. Each book was presented three times during the intervention study. With the exception of intervention session 4 in week 2, a single book was used in each intervention session. The only exception was that the SLT was ill and therefore absent from the second session in week 3 (usually two sessions are held per week), so two books were covered in the following session to reach the total number of books used (each book being used three times in total). The three books (see Table 2) were used alternately, as indicated in Table 3.
The intervention sessions began in week 3 for a total of 7 weeks. Within each, first, the productive probe was carried out. Second, the content of the book was read aloud without explicit modelling. Third, the entire book was read aloud again with the scripted DBR prompts. Fourth, the productive probe was carried out again. Each target word was modelled a total of 13 times in each session as described above. In week 8, the receptive and productive probes were carried out again.
Receptive and productive probes were conducted at the beginning and end of each session. This was done to examine learning within sessions as well as maintenance across sessions. In the productive probe, the child was asked what he/she saw in the picture (of the target word). If there was no response or an incorrect response was provided by the child, the SLT noted the child's initial response on the record sheet and gave the child a prompt in the form of semantic and/or phonemic cue. For example, “if you have a brother or a sister this is your ____.”). These prompts and the results in terms of the child's utterances were also noted on the record sheet. The child's utterance following the prompt by the SLT was the final response that was used to analyse the probe data. In the receptive probe, the SLT produced a word, and the child was asked to point to a picture. Figure 2 shows the receptive probe results in the baseline sessions before the intervention and in the assessment sessions after the intervention ended. In each of the sessions, the probe was conducted twice.
Receptive probes from before (baseline sessions) and after the DBR intervention phase (probe session), in each case conducted at the beginning and the end of the session.
Productive probes in baseline sessions (before intervention started), in intervention sessions and post intervention. In each session, the probe was conducted at the beginning and end.
Figure 3 shows an increase of 10 words produced correctly between the first and second baseline sessions. In the following intervention sessions, the number of words produced correctly remained similar (e.g., intervention sessions 2 and 3), showing that the child was able to produce 17 out of 18 words of the probes at the end of the study, while only four words were named correctly at the beginning of the study. Furthermore, in the first sessions there is a clear increase between the probes conducted at the beginning and the end of the session. In the later sessions, the difference between the probes at the beginning and end is also observed, but to a lesser extent.
Observations of the child's compliance and attention were made to determine which setting and type of book worked best. In general, the treating SLT reported that the child chose the setting (working on the floor or at the table). Five sessions were held on the floor. The SLT observed greater physical restlessness and less focus and concentration when working on the floor. In general, the participant's compliance was observed to decrease the more often the probes or a particular book was used. The probes seemed to bore the child after the fifth presentation, while presenting the same book twice during a session led to less concentration and focus during the second round. In addition, the sessions in week 3 (Table 3) lasted 45 min, which seemed too long for the participant as he was unable to engage at all during the last 15 min. Therefore, all subsequent sessions were shortened to 35 min, which had a significantly better effect in terms of focus, concentration and compliance.
The results of the Austrian trial indicated that it was possible to engage the child in the procedure by using a set of books where each was repeated only once. The content of the books appeared to be more appealing for the child than the book used in the first pilot trial. Even though an overall pre- to post increase in performance was seen, it is noteworthy that much of the progress made by the child in naming the probe pictures occurred between the two baseline sessions, that is, before the intervention started. The relatively poor performance in the first baseline session indicates that the target words were not overall too easy. However, the learning that is seen in the second baseline session was interpreted as a learning effect partly involving the probes themselves. Indeed, in these sessions, comprehension probes were also administered in which the clinician asked the child to point to a picture corresponding to a word, thereby modelling the target words directly in connection with the probes. It was concluded that the probes needed to be more resistant to learning effects and that the use of comprehension probes introduced a confound making it harder to determine the extent to which the intervention itself contributed to learning.
V. Trial 3: Lebanon
This trial was conducted roughly concurrently with the Austrian trial and used the same books, target words and assessment probes. However, a difference was that this trial targeted a bilingual child. Indeed, one of the team's questions was whether to include bilingual children in the study. Recruitment was done by an SLTs engaged in private practice across diverse urban regions within Lebanon. A comprehensive project explanation was disseminated to potential participants.
The administration of the Parental Bilingualism Questionnaire (PABIQ; Tuller, 2015) shed light on the language use in the child's linguistic environment indicating that the child had early exposure to both French and Lebanese Arabic, with a predominant use of the latter in her immediate surroundings. The Linguistic Richness Index (LRI) was relatively higher in Lebanese Arabic than in French (LRILib = 36; LRIFr = 19) showing Lebanese Arabic to be used in interactions with parents, close family members, family friends and acquaintances (e.g., classmates) and exposure to French being limited to the media (television, YouTube) and school. Despite these observations, the questionnaire results showed the child to prefer expressing herself in French.
Parents reported a delay in the onset of the first words and word combinations, which caused considerable concern even before the age of four and led to early counselling and follow-up in speech language therapy. Analysis of the results of the Evaluation of Oral Language in Lebanese (ELO-L, Zebib et al., 2019) and French (ELO, Khomsi, 2001) standardised for the bilingual Lebanese child population (ISO-USJ, n.d.) showed significant challenges in oral language performance, both for French and Lebanese. The results fall in the low or pathological range.
The books were the same as those used in the Austrian study (see target words in Arabic and French in Table 4).
Target words for each book in the Lebanese study.
The timeline of the intervention was, however, different. The intervention protocol was implemented in eight sessions over a period of 4 weeks administered by two SLTs each treating in a different language, with one session delivered in the second language (French) and the other in the native language (Lebanese-Arabic). Each of these sessions lasted between 30 and 70 min and was recorded for an in-depth analysis of the results. To mitigate possible fatigue and maintain the child's engagement, the protocol was initiated with a single book (The Crocodile book) in the first week. In the second week, a second book (The Cow book) was introduced after the Crocodile book. In the third week, a third book (Ana book) was introduced, read after the Cow book. Finally, in the fourth week, only the Ana book was read to the child. To ensure consistency and to allow each book to be read twice in both languages, a predetermined order was followed when introducing the books. In addition, each intervention session followed these steps: (1) Phase I: Reading according to the original text of the book. (2) Phase II: Reading the original text alongside the prepared interactive script. (3) Phase III: Assessment of the acquisition of target words from the respective books, focusing exclusively on expression to avoid re-modelling during the designation task. To prevent potential bias, the list of target words was not shared with the parents. Playful activities were sometimes included separately from the DBR session, to encourage the child's sustained attention during each session.
Productive vocabulary probe scores across sessions during the implementation of the DBR protocol administered in French (left panel) and in Lebanese Arabic (right panel). Scores are shown as unique to the language of intervention and as conceptual vocabulary computed from responses in French and Lebanese Arabic code-switches, counting translation equivalents only once.
During the baseline sessions before intervention, it was found that the child sometimes produced target words in a language other than the language being assessed. For this reason, it was decided to record two scores at the end of each session; the first being the conceptual score of the target words (correct production in either language counting translation equivalents produced only once) and the second score being that of the target language of that session (unique productions in each language). Figure 4 presents the scores of the productive probes by target language obtained at the time of the different baseline sessions (2 before and 1 after the intervention), as well as in each intervention session.
The data also show that the results vary between French and Lebanese. Despite the fact that the participant had Lebanese as her dominant language, as shown by the PABIQ scores, the school language, French, was used more in both the Lebanese and French sessions, which led us to compute a conceptual score in which productions in either language are counted but in which translation equivalents are counted only once. In general, the results showed an increase in the use of target words across sessions, but they were produced more often in French (11/18) than in Lebanese (1/18 for the probe session 3).
More specifically, analyses show that in Lebanese, the child produced the target word siblings /ixwe/ at the end of the first intervention session. This word was not produced at baseline and was not included in the list of target words for that session. However, by the end of the second session, she produced the target word fire /nār/, which had been modelled several times during the reading of the second book (The Cow book) in intervention session 2, achieving a target language score of 2. Notably, the target words produced in session 2 were different from those in session 1. Moreover, the conceptual scores remained consistent at 6 for both sessions. Specifically, in session 2, the words idea, dragon and diapers from the initial active stock were produced in French, in addition to the target word market. Following the third reading session, the child produced the word climbing in Lebanese /ʕam jʕarbiʃ/ in addition to fire /nār/, previously produced at the end of session 2, thus raising the target word score to 3/18. Additionally, the word market was reused, and the word dive was added to the list of newly produced words. However, after the last intervention session, there was a decrease in the target words score for Lebanese (1/18). Specifically, at the fourth intervention session, the target word fire /nār/ was replaced by its French equivalent feu, and the word climbing /ʕam jʕarbiʃ/ was not produced. Moreover, the child used the word supermarket instead of market. Therefore, the child only produced two target words after three sessions, and this new vocabulary was not acquired in Lebanese.
Conversely, in French, the child was able to produce two target words in the first session. After the second session, the number of produced target words increased by two, including laundry, shelf, dive and problem, in addition to the target words already in the child's active stock (score intervention session 2 Fr = 10/18). However, there was a decrease by two towards the end of the third session (8/18), with only dive and problem being produced. This score increased again to 12 at the end of the intervention (12/18), as the two target words try and solution were added to the list of acquired words, and shelf and market reappeared. The results at probe session 3, two weeks post-intervention, showed maintenance of most of the acquired lexicon (11/18). Finally, it is noteworthy that across interventions in both languages, the child was not able to produce the adjectives difficult and easy by the end of the sessions. However, she acquired a total of five target words in French, including three nouns and two verbs.
In general, the SLTs delivering the intervention reported that the child was often motivated and captivated at the beginning of the sessions, and was specifically fond of books with engaging storylines and animal characters. This was a key factor in maintaining her interest. However, the protocol's method of reading each book twice in each language followed by the probe assessment seemed to be boring for the child. The impact of repetition was highly dependent on the specific book and the child's motivation. Despite her general willingness, the child frequently expressed a desire to discontinue the session, especially when faced with the prospect of rereading a book she had already encountered, particularly during the switch to the second language.
1. Group meetings in Vienna and Setubal in February and May 2018
At the Vienna and Setubal meetings (see Figure 1), the Lebanese and Austrian groups presented their trials to the larger group and the implications of the findings were discussed. These are outlined in the main findings and general findings and discussion section. Overall, the trials offer evidence that DBR can be positively received by SLT clinicians and children and their families across different countries and languages and also shows that this requires training as the procedure is not equally well known across countries. It provides valuable methodological indications regarding implementation with preschool children with DLD, notably on the need for presenting a number of interesting books across sessions, the ideal length of each session, the number of words to target, and on using probes that are not too easy to give rise to significant learning effects.
VI. Main findings and general discussion
The iterative exploratory process including trial studies conducted by the piloting team and methodological decisions and discussions involving larger team of researchers described in this paper was conducted with the aim of evaluating a tentative DBR protocol developed by an international team of researchers to be used in several countries to treat preschoolers with DLD. Beyond contributing to finalising this protocol, this feasibility study provides important novel insights into several aspects of efficient DBR protocols for children with DLD, including the types of books, the overall number of books, the number of books per session, the overall number of target words and the number of target words per session, the adequacy of the probes that had been developed, and other procedural aspects including whether to initially present books without prompting strategies. Prior to these trial studies, no clear answers to these detailed questions were found in the literature. The main findings are summarised below.
1. Types of books, number of books per session
The Bulgarian trial clearly pointed to the need for having more than one book and books of higher cognitive complexity than the one used in that study. The particular book used in the Bulgarian study proved not to be interesting for the Bulgarian participant, which, according to clinical observations, contributed to the child's lack of motivation and engagement. Subsequently, the larger international research team carefully considered potential books not only for their language content but for the intricacy level of the story and the pictures. The book used in the Bulgarian study contains simple sentences, but also relatively simple content, which is probably more suitable for younger children who have less world experience and world knowledge than older preschool children. This observation is consistent with the discrepancy between language abilities and cognitive development of children with DLD, and underscores the need to present materials that match the children's language level without insulting their capacity and need for more engaging stories.
The trials differed in aspects of their procedures that concern how often the same book was used and whether the books were first presented without prompts and then again with prompts. Both the Austrian and Lebanese studies indicate that repetition of the same book is not conducive to maintaining the child's engagement. The Lebanese study highlights the need for a delicate balance between repetition for learning and the need to make the educational experience engaging and varied in order to maintain the child's interest and participation. The differences between the trials in this regard were due to the fact that these aspects had not been specified in the pilot protocols.
2. Vocabulary: Selection of target words
Yet another issue that these trials raise concerns about is the selection of target words. In clinical intervention and in many clinical intervention studies, therapy targets are customised for individual children based on probe results and functional needs specific to the child. Such customisation is not as feasible in interventions that administer the same intervention to groups of children using a predetermined set of materials such as is the case for the planned international study, as it was decided relatively early in the development of the protocol that the books would need to be scripted to ensure a uniform administration procedure across languages. This requires a common set of target words for participants from different countries. Even though administering the same set to all participants of an intervention, individually or in groups, diminishes customisation, it increases efficacy in some other ways and can allow more intervention sessions to be presented to each child. The same set of words was used in the second and third trials. The Lebanese child made considerable progress over the course of the intervention, while the Austrian child was observed to perform close to the ceiling quite early on in the intervention. In clinical intervention or in a study using customised target selection, it would have been possible to introduce new target words towards the middle of the intervention and thus achieve even greater progress. Such introduction of new words as previously presented words are mastered have been used in intervention studies targeting vocabulary (e.g., Elin Thordardottir et al., 2015). As this is not possible when a scripted approach is used, the larger international team responded to this finding by revising the target word selection in an attempt to select words that are not too easy, yet are functional words for children in the intended age range and that, across languages, include words that are of approximately similar phonological complexity. Further, the total number of words used in the final protocol was increased from 18 to 36.
3. Vocabulary probes
The vocabulary probes are another aspect that was modified based on the trials. The Austrian study in particular indicated that the probes needed to be revised to avoid learning effects. The Austrian study included three books, each prompting six words belonging to different categories (emotions, problem solving and household items). This study indicated great progress over the course of the study. However, most of the progress was made in the first half of the intervention and a big increase in performance was seen in the baseline period, before the intervention had started. The baseline session did involve a receptive probe in which the target words were spoken by the examiner in direct relation to the probe administration. The Austrian results suggest that the child may have learned the probe words in part from the receptive probe itself. However, the observed progress may not implicate real learning beyond association between the probe picture and the word. More real learning is more likely to be achieved in the naturalistic context of the book reading. The clinicians in the Austrian study reported the impression that the probe was too easy for the child. Further, conducting the probe twice per session, at the beginning and end, proved to be too time-consuming and non-engaging for the child. In response to these observations about the probes, the probe was subsequently revised for the final international protocol, to include only a production probe conducted only at the end of each session. Further, to minimise learning effects, three different versions of the probes were developed that used different picture materials none of which were the same as the pictures appearing in the books and presented the target words in a different order.
Another issue concerning the probes was seen in the Lebanese study which delivered intervention in two languages given that the participating child was bilingual. In this study, the child was found to code switch in the probe context, giving responses in either language regardless of the target language of the probe. The complexity of the bilingual administration of the DBR intervention and the probes contributed to the decision to focus the final protocol on monolingual children in the subsequent study. This decision was made to avoid methodological complexity in a study that was already complicated by an international and multilingual scope, not because the DBR method did not appear to be appropriate for bilingual children. In many respects, the Lebanese pilot study demonstrated a viable means of providing bilingual intervention. It also showed that the scoring has to be adapted for the bilingual children. The procedure used, including keeping track of correct productions in each language as well as analysing the probe productions as conceptual vocabulary (correct productions regardless of the language in which they are produced) provides a useful way of tracking a bilingual child´s progress over time. Children who are raised in a bilingual context can show varying degrees of improvement, depending on language dominance, language use and other language and culture-related factors.
4. Cross-country differences in assessment measures
The three trials also confirmed what the team foresaw already at the Zagreb working meeting, that different countries would differ in the assessment measures available to document children's language levels and confirm their diagnostic status, consistent with survey results on diagnostic methods in paediatric SLT across countries of Europe (Elin Thordardottir, 2015). As outlined in the tentative protocol elaborated in early steps by the larger team, each trial reports detailed test results, however, the types of measures reported vary considerably across the three trials reflecting differences in available tests and clinical traditions. While these differences complicate a direct comparison of children from different countries in terms of the nature and severity level of their language disorder, it remains important to present a documentation of the children's language abilities and difficulties.
5. Final protocol
The lessons learned from the group work, including the work and discussion of the larger group and the clinical material development and clinical trials of the pilot group, served to develop the final protocol for an international study which was subsequently conducted in six countries with preliminary results presented at the final meeting of the Cost Action in Sofia, Bulgaria, in April 2019 and final results in Authors (under review). Based on the findings of the case trials, the final protocol uses books with an interesting storyline and attractive pictures, uses one book per session and minimises repetition of books across sessions. Further, based on these results, as well as other intervention studies (Elin Thordardottir et al., 2015; Restrepo et al., 2013) and the study on dosage by Storkel et al. (2017), the total number of target words was increased from 18 to 36 to avoid ceiling effects and the probes were reconfigured to minimise learning effects. To achieve this with one book presented per session, the number of books in the final protocol was increased to nine books in three sets, with each set targeting 12 words. The final protocol included 12 sessions, with nine presenting a new book and with three books repeated once. A script was prepared for each of the books allowing not only consistent implementation across countries but also control of the dosage, or the number of times each target word was modelled. Further, new assessment probes were developed that comprise three sets of pictures, none of which are identical to pictures used in the books and within each of the three probes, the target words are presented in a different order.
To conclude, these pilot studies show that DBR can be positively received across the different countries and languages participating in the trials and also shows that this requires training as the procedure is not equally well known across countries. It provides valuable methodological indications, notably on the need for presenting a number of interesting books across sessions, the ideal length of each session, the number of words to target, and on using probes that are not too easy to give rise to significant learning effects. These lessons were applied to the final protocol used an international DBR efficacy study subsequently conducted. Determining whether existing intervention methods can be applied in new contexts and languages requires trying them out in these new contexts using research designs that have been developed for this purpose. However, prior to this step, one must consider whether some initial adaptations are necessary to ensure that the materials and method are culturally appropriate, and that clinicians in the new contexts are receptive to the method. The methods for this first step are less well documented in the literature. This study illustrates the iterative decision process undertaken by a small group of researchers working within a larger international team. The steps of this process and the lessons learned prove valuable to other researchers interested in conducting studies involving multiple languages and cultural settings.
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article. This work was supported by the COST | European Cooperation in Science and Technology (grant number COST Action IS 1406 (partial funding)).
