Evidence-based intervention for preschool children with primary speech and language impairments: Child Talk - an exploratory mixed-methods study

Objectives

• To determine current evidence, practice and user perspectives regarding SLT-led interventions for preschool children with PSLI.

• To identify how best to engage preschool children in the process of developing appropriate interventions.

• To develop a model(s) of intervention that can integrate current evidence, professional expertise and family perspectives in ways that allow the intervention to be individualised to children’s and families’ communicative, physical, social and cultural contexts.

Objectives

• To identify tools that can be developed to ensure the appropriate stratification of interventions and the measurement of outcome.

• To identify the measures required to develop formal economic assessments of SLT-led interventions and care pathways within speech and language therapy services.

• To work with the RCSLT to facilitate the national take-up and ownership of the framework.

Speech and language therapists

At each of the six case study sites, invitations to participate in focus groups were e-mailed to SLTs through the speech and language therapy service leads (Figure 3). Those SLTs who were interested in participating in the study were asked to contact the research team at North Bristol NHS Trust by e-mail. On receipt of expressions of interest, a member of the research team contacted each individual to gain information about his or her case load and experience to determine eligibility: the inclusion criteria were currently practising NHS SLTs with at least 2 years’ experience of working with preschool children with PSLI.

FIGURE 3.

Flow diagram of the recruitment process for SLT focus groups.

Those who met the inclusion criteria were e-mailed a participant information sheet (which were held in the participants’ local area) and were given the opportunity to ask questions by telephone or e-mail before attending. Nine focus groups were held across the six sites with a total of 40 SLTs. The SLTs who attended had been qualified as an SLT for an average of 14 years (range 2–43 years). They also emanated from a range of different university training courses (Table 4).

The nine focus groups, conducted locally on non-NHS sites, lasted between 1 and 1.5 hours. At the start of the focus groups participants were reminded that their contributions were voluntary and that they had the right to withdraw at any time. Consent forms were signed at this point.

The focus group discussions were semistructured, with a topic guide that the moderator followed. An example topic guide is provided in Appendix 8. At the beginning of the focus groups, ground rules for discussion were covered, which included covering the confidential nature of the groups. SLTs were then asked questions about:

• the interventions that they use with preschool children with PSLI and their components

• the ways in which they modify their interventions in relation to child, context and family factors

• the rationales for and purposes of the interventions, including descriptions of how the interventions were thought to cause change.

During the focus groups participants were encouraged to be explicit about their interventions and give detail rather than merely the names of programmes or listing resources. When participants used these, they were encouraged to expand and provide more detail.

At the first focus group fictional vignettes were used; the types of cases that were discussed included the following groups:

• a child aged between 3 and 4 years with speech impairment/disorder

• a child aged between 2 and 3 years with receptive and expressive language delay

• a child aged between 4 and 5 years with social communication problems.

Subsequent to each focus group, audio recordings were transcribed verbatim by a member of the research team. The data were anonymised and participants were given pseudonyms when requested. Thematic and content analysis was conducted using NVivo 9 software (QSR International, Warrington, UK). Issues raised in the earlier groups informed subsequent groups in terms of the questions asked.

Parents

Clinical leads at each case study site were asked to identify children within their service who had received input from the speech and language therapy service, who met the criteria for PSLI and who were aged between 2 years and 5 years 11 months. Invitation letters were sent to the parents of these children by the SLT service on behalf of the research team accompanied by a participant information sheet and an expression of interest slip to be returned in a prepaid addressed envelope to the research team. The invitation letter and participant information sheet were designed in collaboration with the Child Talk parent panel.

Forty-seven parents expressed an interest in taking part (Figure 4). They were contacted by telephone by a member of the research team and asked a number of screening questions to establish eligibility according to PSLI and age. Parents whose level of spoken English would significantly limit participation in an interview or focus group were excluded. After this process, 39 parents were invited to take part, of whom 10 attended a focus group or an individual face-to-face interview (when only one participant was available to attend). Focus groups did not occur in all case study sites because of the poor take-up of invitations. To boost recruitment rates, possible alternative methods of data collection were discussed with the Child Talk parent panel. Following these discussions, an amendment was submitted to the ethics committee to seek approval to contact the parents who expressed an interest and had been unable to attend the focus groups to ask if they would take part in an audio-recorded telephone interview. As a result, an additional 16 parents were interviewed by telephone by a member of the research team (a qualified SLT).

FIGURE 4.

Flow diagram of the recruitment process for parent focus groups.

The Child Talk parent panel identified lack of childcare as a barrier to participation and suggested that parents might be more likely to attend if offered something additional. In response to this, free play-based arts therapy sessions were set up by the research team and advertised locally for children to attend while parents attended a focus group. In total, 1500 flyers were distributed to advertise these sessions across the locality, and parents who had expressed an interest in taking part in a local focus group but who had been unable to attend were also made aware of the events. Two parents attended these sessions.

Following the various methodologies described above, a total of 28 parents took part in this study across England (10 focus groups, 16 telephone interviews and two play-based sessions). This number was lower than we would have anticipated but it has given an insight into the difficulties of engaging with this population and the lessons that should be learned for future studies (see Chapter 4, Discussion). A summary of the numbers of parents participating in focus groups and interviews is provided in Table 5.

Informed consent was taken by a member of the research team at the start of the group or interview. The wording of the consent form and participant information sheet was reviewed by the parent panel. Focus groups lasted on average 1 hour 40 minutes whereas individual interviews were typically shorter, lasting between 25 and 50 minutes. Telephone interviews were recorded using an audio-recorder telephone device.

The discussions were semistructured, with a topic guide that the moderator followed (see Appendix 9). At the beginning of the focus groups ground rules for discussion were covered, which included covering the confidential nature of everything that was raised in the groups. Participants in the face-to-face interviews also had the confidentiality process explained to them. Parents were asked questions about:

• what happened during speech and language therapy sessions, including:

  • activities

  • materials used

  • length of sessions

  • how regularly they were seen

  • advice given

  • targets set

• their understanding of the aim of the activities

• whether or not any change was seen in their child following SLT intervention and their view on how much of this change was attributable to speech and language therapy

• whether or not they modified their behaviour/interactions as a result of going to see the SLT

• if anything was particularly helpful or unhelpful

• if there was anything about what happened that they would change.

An iterative approach was taken to the focus groups whereby issues raised were followed up in subsequent groups and interviews. Telephone interviews used a semistructured format and followed the topic guide used for the focus groups.

Audio recordings were transcribed verbatim by a member of the research team. All data were anonymised and participants were given pseudonyms when requested.

Early years practitioners

Service managers at each of the six case study sites were asked to identify EYPs who met the inclusion criterion of working directly with preschool children with PSLI. Participants were approached and invited to participate through managers of nurseries and children’s centres in the local area, who distributed the invitation to relevant staff members. In addition, local nurseries and children’s centres were identified through the use of a search engine and government websites.

Early years practitioners who expressed an interest in participating were contacted by a member of the research team. There was no specific inclusion criteria other than that EYPs worked directly with children in an early years venue. Participants were e-mailed an information sheet before attending a focus group and were given the opportunity to ask questions by telephone or e-mail before attending. Twenty-three EYPs expressed an interest in participating in the focus groups (Figure 5), of whom 18 attended from five of the six case study sites (there were difficulties with recruiting within case study site 4). The number of EYPs who attended the focus groups varied between groups (mean 4, range 2–5).

FIGURE 5.

Flow diagram of the recruitment process for EYP focus groups.

Because of the low recruitment levels, ethical approval was sought to invite EYPs who had expressed an interest but who were unable to attend a focus group to take part in an audio-recorded telephone interview. Six EYPs were interviewed by telephone (Table 6).

Focus groups took place at non-NHS sites in each of the selected research sites and typically lasted between 1 and 1.5 hours. At the start of the focus groups, written informed consent was taken by a member of the research team. Telephone interviews were conducted at a time convenient to the participants and typically lasted around 25–35 minutes. Consent was taken verbally by reading each statement and the participant answering ‘yes’ or ‘no’. All participants were offered a copy of their consent form, either by e-mail or by post.

The focus group discussions were semistructured, with a topic guide that the moderator followed (see Appendix 10). At the beginning of the focus groups, ground rules for discussion were covered, which included covering the confidential nature of everything that was raised in the groups. Telephone interviews followed the topic guide used for the focus groups. EYPs were asked questions about:

• the interventions that they use with preschool children with PSLI and their components

• the ways in which they modify their interventions in relation to child, context and family factors

• the rationales for and purposes of the interventions, including descriptions of how the interventions were thought to cause change.

Subsequent to each focus group and telephone interview, audio recordings were transcribed verbatim by a member of the research team. All data were anonymised and participants were given pseudonyms when requested.

Underserved groups

To identify groups perceived to be underserved, SLT service managers from the six case study sites were asked to identify groups within their geographical areas that they considered fit into one or more of the following categories:

• groups with poorer attendance rates than other groups in the catchment area

• groups who were, for other reasons, under-represented on the speech and language therapy services’ PSLI caseloads compared with their representation in the general local population

• parents/children to whom managers considered that they do not provide appropriate services.

A number of groups were identified by speech and language therapy service managers:

1. carers of looked-after children (children in public care, who are placed with foster carers, in residential homes or with parents or other relatives)

2. people from minority ethnic groups [including specifically the Somali community and refugees and asylum seekers (RAS)]

3. families from areas of low socioeconomic status (SES)

4. travellers or gypsies, defined as those with a cultural tradition of nomadism or living in a caravan, and all other people with a nomadic habit of life, whatever their race or origin. 91

The first three of these groups were selected for inclusion into this study for reasons related to the feasibility of completing data collection within the time frame of this programme. As these were groups perceived to be less likely to be receiving services, recruitment directly through speech and language therapy services or of those currently receiving intervention was unlikely to be successful. It was therefore decided to identify members of the groups selected who were already attending other groups in their local area as a positive strategy for recruitment. Recruitment was therefore targeted to specific groups, as discussed in the following sections.

Specific Interest Groups

Speech and language therapist SIGs and services focusing on preschool and early years populations were identified through a number of sources:

• local SLTs and the research team’s knowledge of relevant SIGs

• scanning for SIG advertisements placed in the RCSLT Bulletin magazine over the previous 12 months

• scanning for advertisements placed by other relevant organisations or groups, for example early years forums, in the RCSLT Bulletin magazine.

Special Interest Group chairs (n = 7), service leads (n = 2) and leaders of other relevant groups and organisations (n = 1) were contacted by e-mail (taken from their adverts) with information about the project. They were invited to express an interest in hosting research events for Child Talk.

Four groups expressed an interest in hosting a research event (Table 8). The organisers of these groups used their normal advertising routes to alert their members to the events. A week before each event the potential participants were sent copies of the information sheet and consent form for information. The potential participants were invited to contact the research office by e-mail or telephone before attending the event if they had any questions regarding the research. Informed consent was taken at the start of the research event by a member of the research team. A total of 66 SLTs took part across the four groups (see Table 9). Of the activities that were undertaken at the four groups, 64 SLTs participated, with two therapists leaving one of the events before the voting activity (see Activity 1: validity of the themes from the speech and language therapists’ perspective).

Since these research events took place, the RCSLT have renamed SIGs as Clinical Excellence Networks. We have retained the nomenclature of SIG as the groups were called this at the time of recruitment and data collection.

The data collection activities undertaken at these events are described in the following sections (see Activity 1 to Activity 4); as the activities undertaken varied across the events, the numbers of participants taking part in each activity vary.

National consensus events

After the regional SIG events, two national events were held by the research team, one in Leeds and one in London, to assess consensus on the emerging typology and toolkit (assessments and outcomes). These events were open to a range of participants – EYPs, parents and SLTs – including participants who had previously taken part in Child Talk, who were invited to attend by e-mail. In addition, widespread advertising was undertaken in magazines and appropriate publications, on websites and on social media. The parent panel gave suggestions for advertising the event to parents. Anyone interested in taking part in the event was asked to contact the research team to confirm eligibility and register for an event. The inclusion criterion for practitioners (SLTs and EYPs) was experience of working with preschool children with PSLI and that for parents was having a child aged < 7 years and being, or previously being, worried about their child’s talking. Registered participants were sent a participant information sheet and consent form by e-mail at least a week before the consensus event. Informed consent was taken at the start of each consensus event by a member of the research team. Each event lasted a full day and involved a series of exercises suited to the participants’ area of expertise (SLTs, EYPs, and parents). In total, 47 participants attended the event in Leeds (42 SLTs, five EYPs) and 46 participants attended the event in London (44 SLTs, two EYPs). Although seven parents registered to attend, none of them attended on the day. The activities developed for the national events were designed to build on the findings from the SIG event activities (see Specific Interest Groups).

Assessment tool survey

A short online survey was developed using Snap Survey software. This survey was designed to gather information on:

• which three assessments therapists use most frequently with preschool children with PSLI

• the reasons why these are selected

• what information the assessments provide to support clinical decision-making.

Appendix 14 provides a screenshot of the online survey. The survey was limited to three assessments to keep the time required to complete it to a minimum and to encourage therapists’ participation. A covering explanatory e-mail and link to the survey was circulated to the 40 SLTs who had participated in the focus groups (see Speech and language therapists) and who had agreed to further involvement in the research programme. Consent for undertaking the survey was implied through completion and submission of survey responses. Of the 40 SLTs invited, 25 took part in the survey.

The first survey question asked participants to identify the three assessments that they use most frequently when working with children with PSLI. The following questions were then asked about the assessments that the SLTs had identified (as free-text boxes):

• What is your primary reason for selecting this assessment?

• What information does this assessment provide to support your clinical decision-making?

Summative content analysis was used to analyse the survey data. 93,94

Speech and language therapist intervention surveys

Two surveys were developed, with the first contributing to the development of the second as well as producing data in its own right. The longer second survey was designed to gather more detailed data on the components of interventions and the child, family and contextual factors that might lead to their modification. The focus of the research was on the activities and strategies used rather than on which programmes or resources are used, although, within survey 2, activities and strategies were not explicitly separated.

Parent survey

The aim of this parent survey was to obtain the views of parents on the commonly used interventions identified by SLTs (see Chapter 2, Study 2.2: identifying the interventions used by speech and language therapists). Following the challenges with recruitment of parents to the focus groups (see Chapter 4, The perspectives of parents), and in discussion with the Child Talk parent panel, the inclusion criteria for this study were broadened to include any parents who had been worried about their child’s speech and language development. This reflects that fact that parents do not need to have experienced speech and language therapy to be able to feed back on how achievable and/or easy to understand the strategies and activities that SLTs recommend are. It was felt that the inclusion criteria did need to specify that parents needed to have been worried at some time about their child’s speech and/or language development so that they were likely to have experienced similar emotions and therefore be able to relate to parents who had received some degree of speech and language therapy and who had been given these strategies to perform in their home environment. It was also decided that, rather than undertake focus groups, this study would utilise an online survey to elicit parents’ views.

Recruitment took place through the internet, flyers, parent-focused publications and social media. In addition, a recruitment video was developed, scripted and voiced by the parent panel (whiteboard animation), which introduced the Child Talk project and described what parents were being asked to do in a lay and engaging manner. This video was placed on the research unit website [http://speech-therapy.org.uk (accessed 12 January 2015)] and YouTube (YouTube, LLC, San Bruno, CA, USA) [http://youtu.be/fn3ebCd1vc0 (accessed 15 December 2014)] and provided a link to direct parents to the parent survey.

The survey started with a set of nine videos, made by the parent panel and research team, which demonstrated the strategies typically recommended to parents by SLTs and the activities used to model these. The strategies and activities were selected from data provided by SLTs on their current practice from the BCRP’s national survey89 and through content analysis of the Child Talk SLT focus groups (see Chapter 2, Study 2.2: identifying the interventions used by speech and language therapists). The videos included a brief description of the strategies and the rationale behind them followed by a demonstration of the activities and toys that the SLT survey had shown were typically used by SLTs (see Chapter 2, Study 2.2: identifying the interventions used by speech and language therapists). Both the explanation of the strategies and the demonstrations were provided by an experienced SLT [see www.youtube.com/watch?v = 3jM9QZ9nq74 for an example (accessed 23 December 2014)]. Following the video, parents were asked to express their views using a user-friendly rating system (Snap Survey software). The questions were asked to:

• gain an indication of if parents feel that they understand the rationales behind the strategies

• investigate if parents would consider it ‘realistic’ or ‘achievable’ to carry out these strategies and, if not, whether this was because of time pressures, not feeling confident to carry them out or other barriers

• understand how parents might feel about being asked to use these strategies

• gauge how parents feel about the resources/toys used by SLTs.

Appendix 17 provides a screenshot of the questions and responses for one of the videos. The videos and survey were hosted on the research unit’s website. In addition to parents completing the online survey at home, the research team sought to tap into existing parent groups and organised coffee morning events to allow opportunities for parents to view the videos and rate them in a community setting. This also allowed opportunities for parents who did not have access to the survey to take part. Four groups watched the complete set of videos and completed the survey, including a pre-existing support group for parents of children with SLCN, parents at an event organised by a school and parents at two coffee mornings arranged directly by the research team.

Parents accessing the videos online were not required to respond to all of the videos to submit data and therefore the order of the videos was monitored and adjusted to provide an even response rate to all videos. A total of 56 parents (online and groups) responded to some part of the survey, with at least 28 responding to all of the videos.

Data were entered into SPSS software and descriptive statistics were calculated.

Outcome survey (Delphi round 3)

Delphi round 3 was an online questionnaire, using SurveyMonkey, developed for ranking the outcome indicators within each outcome domain for each typology theme.

The survey was circulated to 109 SLTs who had indicated at the SIG events, at the national consensus events or by e-mail contact with the research team that they would like to be involved further in the research. A covering e-mail and explanation was circulated with the survey. Consent for taking part in the Delphi questionnaire was implied through completion and submission. In total, 79 responses were received.

The themes were split into two groups and participants were randomly assigned to the groups. Within each group the questions were asked in two different orders to account for any participants not completing all questions. This split meant that group 1 respondents were asked to complete 25 ranking questions (totalling 93 individual items to be ranked) and group 2 were asked to complete 32 ranking questions (totalling 101 individual items to be ranked).

The survey asked SLT participants to rank the outcome indicators for each outcome domain. Consensus was reached when > 65% of SLTs ranked a particular outcome indictor in the top three. For outcome domains that had less than three outcome indicators, therapists were asked, ‘Is “X” an acceptable indicator for the outcome “Z”? If “No”, can you suggest any other indicators that could be used to evidence change and/or improvement for the outcome “Z”?’, with responses provided in a free-text box.

Search strategy

The Booth and Fry-Smith98 PICO model (population, intervention, comparison, outcome) guided the development of the search strategy. The ‘population’ of interest was defined as preschool children between the ages of 2 years and 5 years 11 months with PSLI. The papers had to include an empirical intervention, although we did not specify the nature of the intervention. A comparison group was not a requirement in the included papers but there had to be at least one outcome measurement of speech or language.

The research team developed the search strategy using previous publications in similar areas,29,35 clinical expertise, information specialist advice and identification of terms used in key papers. The same resources were also used to determine a priori inclusion and exclusion criteria, specify search terms (see Appendix 18) and decide which databases to search. 29,35 The databases were selected to ensure maximum inclusion of published data, unpublished data and conference proceedings. Table 9 provides a summary of the databases searched and the number of studies identified from each database.

To be included, studies had to meet the following requirements:

• be an empirical evaluation of an intervention, including randomised controlled trials (RCTs), experimental and quasi-experimental studies and case studies, that included multiple baseline or other systematic manipulation of the intervention

• at least 80% of the sample was required to be within the age range 2 years to 5 years and 11 months at the start of the intervention or at recruitment

• the intervention had to be led by a SLT (or equivalent profession)

• language and communication outcomes: at least one of the primary outcome measures of included studies had to address the child’s language (semantic, morphological and/or syntactic aspects), communication (social use of language) or interaction (verbal or non-verbal).

Studies were excluded if they related to:

• children whose language appeared to be developing typically with no evidence to suggest that their language was ‘at risk’

• children whose language delays were associated with other developmental or pervasive conditions such as learning difficulties, autism, cleft palate and cerebral palsy

• social or behavioural outcomes that were not language or communication related.

Search procedure

Subject terms were adapted with similar terms to account for specific variations between database(s) thesauruses. Examples of subject heading terms that were used in different databases are provided in Appendix 19.

No date limit or language restrictions were placed on the reference search. A total of 55,271 references were found across the 18 databases. All search results were downloaded into Reference Manager software (version 11; Thomson ResearchSoft, San Francisco, CA, USA), in which duplicates were removed, leaving 33,325 references.

To assess reliability, two members of the research team independently reviewed the titles of 325 of the papers to screen for relevance, removing animal studies and any others that did not fit the exclusion and inclusion criteria. These authors then reviewed this process and there was 100% consensus. The remaining 33,000 references were shared between these two authors and references were assessed for inclusion at the title level. This process led to the retention of 4574 references, which were divided up for review at the abstract level. This abstract review was undertaken by four members of the research team, including two SLTs, to ensure that at least two people reviewed each manuscript and that one was a SLT. When disagreements occurred the rationale was discussed within the team and consensus was reached. At this point 473 papers were retained and the full papers were reviewed in light of the inclusion and exclusion criteria. Once again, three of the reviewers met and consensus was reached, leading to the final retention of 147 studies. These studies were then quality appraised using one of two tools deemed suitable for each study’s research design/methodology.

Quality appraisal

The quality appraisal tools to be used in the review had to meet certain criteria: they had to be of relevance to the research designs used in the included papers, use a scoring system with a cut-off point indicating an acceptable quality and, preferably, have been used previously in similar content areas. The cut-off point is an indicator of acceptable methodological quality. It is not unusual to undertake a quality appraisal of papers as part of a systematic review process. We aimed for transparency in this process by using quality appraisal tools for which training was available, working to consensus on ratings and operating the same standards for inclusions.

Initial scrutiny of the included papers showed that a range of designs had been used, including RCTs, quasi-experimental group designs and single case experimental designs (SCEDs). A number of quality appraisal tools were considered. Tools from the Critical Appraisal Skills Programme (CASP)99 were considered but could not be used because they either did not have a checklist for single case designs or did not have readily quantifiable criteria. The checklist of Downs and Black100 was also considered and rejected because it is aimed at RCTs and quasi-experimental group designs but is not appropriate for SCEDs. It is considered more valid to use a consistent approach across the study designs.

Finally, two quality appraisal procedures, the Physiotherapy Evidence Database (PEDro)-P scale101 and the SCED scale,102,103 were selected for use in this research. These two tools were adopted by speechBITE, the Australian-based service that evaluates research papers in speech and language therapy across all three methodological types. The PEDro-P scale was designed to appraise the methodological quality of RCTs and non-RCTs. The SCED scale is a parallel tool designed to appraise studies using SCEDs. Scores for the PEDro-P scale range from 0 to 9 whereas scores for the SCED scale range from 0 to 10. The higher the score, the greater the quality of the methodology applied and reported within the study. speechBITE provides online training for the use of these tools, which was undertaken by all reviewing members of the research team.

Of the 100 papers assessed using the PEDro-P scale, scores ranged from 0 to 9, with a mean of 4.2, and of the 47 papers assessed using the SCED scale, scores ranged from 2 to 10, with a mean of 6.8. Previous reviews have used a score of ≥ 6 on these two scales to indicate a high-quality study. 104,105 A score of ≥ 6 was therefore used to determine the studies of acceptably high quality to be retained for inclusion in this work (n = 58).

Data extraction

Data extraction was undertaken on the 58 retained studies (see Chapter 3 for details) and a summary of each was produced detailing the following characteristics:

• country of origin of the study

• number of child participants

• age and gender split of the child participants

• number of other participants (excluding the research team)

• demographics of the other participants

• context/location (e.g. home, clinic, preschool)

• cultural context

• study design

• study focus

• aim/hypothesis of the intervention

• description of the intervention

• whether the intervention was focused on PSLI or whether PSLI was measured as an outcome of a differently focused study

• intervention type (e.g. parent–child interaction, language enrichment)

• assessment measures

• outcome measures

• who delivered the intervention

• if the child participants had been diagnosed as having PSLI or were ‘at risk’ of developing PSLI

• findings/conclusions of the study.

Mapping retained studies onto the Child Talk typology categories

On completion of the data extraction process the studies were reviewed in light of the typology themes identified in Chapter 2 (see Study 2.1: identifying the themes of speech and language therapy practice). Using the definitions of the categories, four members of the research team mapped the studies. This entailed assessing the content of each of the studies, specifically the intervention and outcome measures, and deciding which (if any) of the typology categories were represented by the study. Each study could be represented in more than one theme. Two members of the research team then reviewed the categorisation and reached 100% consensus across all studies.

Synthesis

The synthesis process consisted of two stages. A wide range of characteristics of the studies was extracted. The first were broadly contextual and related to the country, culture and language(s) of the researchers and participants. From this we were able to determine if a diversity of countries was represented in the final data set and if particular interventions or designs were associated with practice in particular countries. The second stage addressed the characteristics of the participants, as age (within the inclusion criterion), gender and diagnosis compared with ‘at risk’ might be associated with particular types of intervention or categories of practice within the typology. The following list gives the factors extracted from the retained papers:

• evidence of positive change, within each study’s own parameters

• country of origin/culture

• size of the sample

• child participant demographics: age, gender

• study design

• what the authors stated the intervention was about

• who delivered the intervention

• where the intervention was delivered

• assessment measures used

• outcome measures used

• number of treatment sessions

• interventions (whole and individual activities and strategies) used in the study mapped against the activities and strategies provided by SLT participants in the Child Talk focus groups and online surveys.

Appendix 20 provides summaries for each paper with regard to the following characteristics: (1) the number of participants, (2) child participant demographics, (3) the context/location of the intervention, (4) the study design, (5) who the intervention was delivered by and (6) the quality rating. Appendix 21 provides summaries for each paper with regard to the following characteristics: (1) assessments used and (2) outcome measures used.

The studies were categorised by typology theme. This consensus was achieved by at least two, but more usually three, members of the research team working independently to map each study against the nine themes. Any disagreements were resolved by discussion. It is worth noting that mapping is dependent on the study authors explicitly stating that they are doing ‘X’ or ‘Y’. This does not mean that the authors of the studies did not undertake activities; some activities might be a part of the assumed procedures of an intervention. However, to ensure rigorous interpretation of the evidence, only those studies that explicitly reported an activity or strategy of relevance to a typology theme were mapped against that theme.

Once the data were synthesised the level of evidence available for each theme was graded A, B or C using the grading scheme formulated by the Clinical Outcomes Group of the NHS Executive. 106 This is outlined in Table 10.

An additional step in the analysis of the studies (in relation to the typology themes) was undertaken by mapping the study intervention activities and strategies to those identified by SLTs during focus groups, the online survey and SIGs. This was carried out by SLTs. This allowed identification of aspects of the interventions:

1. undertaken by practitioners and not represented in the literature

2. undertaken by practitioners and represented in the literature

3. represented in the literature but not reported in the other areas of this research programme.

Recording of the children’s groups

A number of methods were used to record data. The drama and movement therapist (DMT) wrote comprehensive field notes and reflections both during and immediately following her observations of the speech and language activities. When she took a more active role in the sessions the notes were made from both her recollections and those of other team members.

All sessions were video recorded with two ‘environmental video cameras’ (Figure 6a). There were also two head-mounted Kiddicams. One was a wide-angle mini waterproof Camcorder 720p [see www.dogcamsport.co.uk/dogcam-bullet-hd-camera-WIDE.html (accessed 15 December 2014)]. This video recorder did not have a monitor; placement was carried out by the research team, who attempted to ensure that it was positioned on a headband (Figure 6b) and captured what the child was looking at.

FIGURE 6.

(a) Environmental video cameras mounted on tripods (‘X’); and (b) wide-angle head-mounted Kiddicam.

The other Kiddicam was a matchstick camera with a portable digital video recorder [see www.dogcamsport.co.uk/lawmate-ss42-matchstick-camera.html (accessed 15 December 2014)]. The monitor provided with this camera allowed the researchers to adjust the angle of the camera on the headband to ensure that it was pointing in a direction that matched the child’s line of sight. Our intent was to generate video images that capture a child’s-eye view, to gain some insight into the children’s experiences, from their perspective.

Although the children were too young to be able to give informed consent, their assent was considered throughout. The individual activities were explained to the children in suitable language by experienced members of the research team. If they chose to remove themselves from an activity by moving out of the group then the team ensured that they were safe and if appropriate encouraged them to re-engage using similar methods to those that they might use within a nursery setting. If a child continued to display a desire to stop it was taken as a removal of assent and the child was taken back to his or her nursery base. The Kiddicam was introduced to the children by the DMT and the children were asked which of them wanted to wear one of them. When a child volunteered to wear a camera they were assisted by a team member to put it on. If no child volunteered then a member of the research team wore one and opportunities were given between activities for a child to wear it if he or she showed an interest. If a child had been wearing a Kiddicam but wanted to take it off then they were helped with this and other children were offered the opportunity to take over.

Analysis

The principles of the framework analysis approach were used to examine and explore the data. The framework analysis approach was developed by Ritchie and Spencer;109 it is frequently used for large qualitative data sets and allows the analysis to be conducted transparently in a way that can be shared by a number of analysts. This approach was felt to be suitable for this study (see Box 5).

In excess of 72 hours of video data were collected across the groups. One of the researchers attended the majority of the groups (16/24) and reviewed and edited all of the video files. One of the team had extensive experience in framework analysis and intentionally selected 10 sessions as follows: two sessions were selected in which the activities were child led; in two sessions the children were used to a very structured educational setting; two sessions included the youngest children within them; two groups were selected in which the DMT activities took place before the SLT activities; and two groups were selected to allow analysis of, and comment on, changes over the course of the four site visits. Thus, the analysis sampled sessions that varied according to contextual variables that were likely to impact on the children’s experiences of therapy.

1. Each group session was broken down into separate SLT activities and the aims and other variables, such as the materials used for each activity, were recorded in a Microsoft Excel spreadsheet.

2. Two members of the team watched video footage from video cameras on tripods and the Kiddicam cameras simultaneously from each of the 10 group sessions (Figure 7). Each session was an hour in length. Descriptions of what the children were doing and saying in response to the activities, the SLTs and the facilitators were made by the researchers and entered into the Microsoft Excel spreadsheet. The researchers also integrated any field notes from team members (such as the DMT) into the descriptive analysis and this included rich interpretations of children’s perspectives. After describing two groups, the researchers discussed their descriptions to check for consistency and ensure that there was continued consistency in subsequent analyses. A final verifier examined the data set to ensure consistency across researchers.

3. Once all data sets had been coded by the two researchers, the categories were refined and themes generated. Any differences were agreed.

4. Key quotes were extracted from the descriptions of sessions to enable the overall story from the framework set to be told.

5. Biases were reduced by ongoing liaison and meetings between the team members and verification strategies were set in place to ensure trustworthiness. These processes enabled critical review of the analysis.

FIGURE 7.

Multiview Digital Replay System software (University of Nottingham, Nottingham, UK; http://thedrs.sourceforge.net) used to watch and analyse groups (DReSS node, National Centre for e-Social Science, University of Nottingham).

Building consensus on the typology themes

We wanted to explore with SLTs at the SIG and national events whether or not the draft typology was comprehensive and relevant to the work that they do. We also aimed to build consensus over adaptations to the draft typology that would lead to the final typology. The following sections present data that aimed to explore the validity of the draft typology.

Speech and language therapists at SIG events were asked whether they considered the 10 themes in the typology to be essential, desirable or not used in their work with preschool children with PSLI. Table 14 presents the findings of this exercise. Of the 64 therapists who completed this activity, 80% (n = 51) indicated that they thought that the themes covered all aspects of their work with preschool children with PSLI.

The pattern of response was similar across the four events. For seven of the themes, all participants agreed that they would be either essential or desirable to their work. Of the 10 themes, ‘self-monitoring’ and ‘speech’ achieved the lowest level of agreement that these were essential to the work. Some participants provided explanations for not working on certain areas, for example ‘speech’ work was seen as being relevant only with older preschool children and not all SLTs worked with these older children. ‘Self-monitoring’ work was frequently associated in the discussions with ‘speech’ work and therefore similar explanations for its lack of use were given. Despite this, there was general agreement that these two themes were legitimate parts of the SLTs’ practice with preschool children.

Explaining the themes

We next wanted to explore whether or not the initial descriptions of the typology themes were robust, inclusive and able to be accurately interpreted by SLTs. We also required the descriptions of the themes to be understandable by lay audiences. To gauge these factors, participants at the SIGs were asked to describe how they would explain each theme to a parent. To establish SLTs’ ability to accurately interpret the typology themes, descriptions were reviewed for their consistency, that is, the extent to which SLT descriptions were similar to, expanded or were different from the descriptions that we provided.

Thirty-seven participants at SIG events completed explanations of the themes for parents. Of these, 26 provided a description for all 10 themes. For all themes, the majority of participants provided explanations that were consistent (i.e. either the same or expanded) with the given definition. However, some participants wrote explanations that were different in some way for all of the themes except comprehension. These discrepancies consisted of:

• confusion with/or inclusion of another theme

• emphasis on a concept not previously included in that theme

• poor definitions.

Table 15 indicates the numbers of participants who provided written explanations that were very close in content to the original definitions that the team provided (Child Talk definitions), the numbers who expanded on the definitions and the numbers who provided different definitions. Decisions about whether the definitions were the same, expanded or different were made by two members of the research team, who were also qualified SLTs.

Vignette exercise (final validation)

As a next step we wanted to establish whether SLTs were able to describe their interventions in relation to the typology themes. Participants attending the national events were asked to indicate the interventions used for each theme of the typology for a child from their caseload. Responses were received from 62 participants. The responses of participants were analysed by SLTs from the research team. Responses were explored for consistency of the interventions with the theme descriptions, that is, the extent to which interventions were deemed relevant to that theme. This was used as a form of validation of the themes and their descriptions. This exercise also intended to establish if the typology themes were comprehensive, that is, if SLTs were able to describe all of their work in relation to the themes.

Table 16 reports the number of therapists who provided intervention information in relation to each theme. It also provides information on the number of therapists who explicitly stated that a theme was not applicable to the work that they did with that child. Missing data, that is, when the total number of responses for any theme is < 62, reflect the fact that some SLTs did not provide intervention information or report that the theme was not applicable to their work.

The largest number of responses from SLTs was under the theme of ‘adult understanding and empowerment’ (n = 60). The theme with the least number of responses was ‘self-monitoring’ (n = 31), closely followed by ‘comprehension’ (n = 35). Reasons reported for not working on these areas included that these areas were still being monitored or that they had not been worked on to date with that particular child.

All themes had relevant or consistent interventions reported for them. Most SLTs provided examples of interventions that matched the existing descriptions of the themes. However, the theme ‘participation’ included a relatively large number of intervention activities that were not consistent with the theme (n = 10). These focused on the child’s immediate participation in the session or attendance at clinic, with tasks being based around motivating or engaging the child in the session, for example ‘child chose favourite games or activities at beginning of session’ (B26).

There was a large amount of variance in the level at which therapists worked on ‘adult–child interaction’, although all of the intervention descriptions were consistent with the original theme descriptions. Although some therapists described targeted/detailed work, including programmes such as the Hanen programme [see www.hanen.org (accessed 19 March 2015)], others referred to adult–child interaction as something that they only touched on, for example ‘general advice re parent–child interaction given at initial appointment’ (B36).

Many of the responses for ‘speech’ included interventions that could be categorised under ‘sound awareness’, including phonological awareness and discrimination activities, indicating overlap of these categories. These sound awareness activities were often described alongside speech production activities. Several participants (B13, B29 and A52) also reported the use of minimal pairs, which incorporate aspects of both sound awareness and speech production.

The findings from the vignette activity confirmed that SLTs are able to describe their interventions in relation to the typology themes. Although not all themes were relevant for all of the vignettes used by SLTs, there were no interventions that SLTs reported that they could not describe in relation to at least one of the themes. There were a number of interventions, however, that SLTs reported in relation to several of the themes. Although this might indicate overlap of the themes, it appeared that these interventions were broad (e.g. parent–child interaction, language groups) and were intended to target multiple aspects of the typology at a time.

Refinement of and adaptations to the typology themes

The data gathered from the activities described were reviewed iteratively as data emerged, in terms of emerging challenges, issues and gaps in the typology. At the national events, participants were asked to directly vote on and discuss many of the issues that emerged. Three of the theme changes involved a change of name, with ‘child participation’ becoming ‘functional communication’, ‘structure and content’ becoming ‘expressive language’ and ‘parent/adult–child interaction’ becoming ‘adult–child interaction’. The other change was broader and involved merging the two themes ‘speech’ and ‘sound awareness’. The rationale behind these changes is discussed below.

Data from the SIG events indicated confusion over the theme ‘child participation’, with 42% of participants providing a description that was classified as ‘different’ for this theme. In particular, many participants referred to participation in the session or in therapy rather than to the broader aspects of participating in life events, which was the emphasis in the original theme description. Similarly, at the national events, there were mixed responses: although some therapists felt that ‘participation’ was a label that captured the theme adequately, others indicated a lack of certainty about what it encompassed. Participants used a variety of words and phrases as potential alternatives or to signal their interpretation of the concept of ‘participation’, including ‘functional communication’, ‘engagement’, ‘motivation’ and ‘social use of language’. The predominant term that was used was ‘functional communication’ and thus this term was used in the final typology.

‘Structure and content’ was noted by SLTs to be an unfamiliar phrase and 36% of participants at the national events provided a description that was different from the description that we provided. Adjustments to the term ‘structure and content’ were not discussed at the national events. However, the name of this theme was changed to ‘expressive language’ to be consistent with the phrasing of the theme ‘comprehension’ and to incorporate the use of widely accepted SLT terminology.

The name ‘parent/adult–child interaction’ was changed after a lack of consensus emerged from participants regarding how best to configure this theme and its constituent concepts. A decision was made at the level of the research team to adjust the name of the theme minimally to ‘adult–child interaction’. This distinguishes it from the specific intervention called ‘parent–child interaction therapy’ and includes the broader reference to all adults who interact with the child, that is, in addition to parents.

Two of the original themes ‘sound awareness’ and ‘speech’ were merged into a broader version of the theme ‘speech’, whose new definition incorporates both speech and sound awareness. Responses from SLTs at the SIG events with regard to explaining the themes to parents had indicated that SLTs view these two themes as overlapping. Table 17 shows all items that were taken to vote at the national event and the results. As can be seen, the majority of SLTs felt that the two categories ‘sound awareness’ and ‘speech’ should be merged.

The description of the characteristics of the theme ‘speech’ with respect to sound awareness was also adjusted to refer exclusively to phonological awareness rather than basic or environmental sound awareness. Although the original description of the sound awareness theme included both speech sound awareness and environmental sound awareness, in the descriptions that SLTs provided for parents at SIG events the majority of participants referred only to speech/sound awareness. Only five of the 29 participants who completed this exercise referred to the more basic skills of identifying and discriminating between environmental sounds. Similarly, in the vignettes activity at the national events only 3 of 62 participants referred to more generic listening activities in the sound awareness theme. Voting at the national events produced > 90% agreement across both sites that more basic aspects of sound awareness were more appropriate to the theme ‘foundation skills’ (see Table 17).

Other challenges to the typology

Three other key challenges to the typology themes emerged from the SIG events:

1. References to children’s emotional well-being or confidence and where this fits in the typology.

2. Many SLTs reported that the training of other adults formed a major part of their interventions (parents and professionals). It was not clear whether this aspect of therapy was a characteristic of ‘parent/adult–child interaction’ or warranted a new theme ‘communication environment’.

3. There was feedback that the themes ‘self-monitoring’ and ‘generalisation’ seemed to overlap.

Discussion of, and voting on, these issues at the national events indicated that no changes to the themes should be made in relation to these three areas. Box 6 provides a summary of the findings from discussions that took place regarding points 1–3 that did not lead to changes in the descriptions of the themes.

Modelling how the original 10 themes relate to each other

Participants (SLTs, n = 64) at the SIGs were asked to model how the 10 themes from the draft typology related to each other. This task was intended to explore how SLTs perceived the relationship, interaction or overlap between the typology themes. The models produced by participants at the SIG events can be divided into two groups, hierarchical and modular, with some models combining elements of both. In the following sections the two types of models are described separately and examples of relevant models are provided for each.

Hierarchical models

Hierarchical models were presented as lists, pyramids, trees and flow charts and suggested an order to the 10 themes. These orders seemed broadly developmental or progressive, suggesting that some themes were considered more basic or fundamental than others. Examples of these types of models are given in Figures 9 and 10.

FIGURE 9.

Hierarchical tree model representing how the themes might relate to each other.

FIGURE 10.

Pyramid model representing how the themes might relate to each other.

Figure 9 shows a drawing of a tree, which was used to represent a hierarchical structure: at the roots are ‘foundation skills’, ‘parent/adult understanding’, ‘parent–child interaction’ and ‘child participation’, suggesting that these were considered to be underpinning skills; the trunk represents ‘comprehension’ and, further up, ‘structure and content’ of language; these in turn lead to the branches, which represent ‘sound awareness’, and finally the leaves, which represent ‘speech’. A cloud hovering over the tree represents ‘self-monitoring’ and ‘generalisation’, suggesting processes that facilitate growth in the tree.

Figure 10 shows a pyramid, a model used by a number of participants. In this version, the base of the pyramid represents ‘parent understanding’, ‘child participation’ and ‘parent–child interaction’; the next layer represents ‘generalisation’ and ‘foundation skills’, which is followed by ‘comprehension’ and then ‘content’. The layer above this represents ‘sound awareness’ followed by ‘speech’ and finally ‘self-monitoring’. The exact configuration of each pyramid model varied slightly from each other and from the previous tree model but there are clear similarities in both the content and ordered structure of the models (see Figures 9 and 10).

Modular models

Modular models were presented visually as boxes, balloons and Venn diagrams. The themes were then grouped within these, sometimes suggesting orders within a module. Examples are shown in Figures 11 and 12.

FIGURE 11.

Balloon and box modular model representing how the themes might relate to each other.

FIGURE 12.

Venn diagram modular model representing how the themes might relate to each other.

Figure 11 shows boxes containing themes – ‘speech’, ‘comprehension’, ‘structure and content’, ‘self-monitoring’ and ‘generalisation’ – with dotted lines between them. Above the box are three balloons, with each balloon representing a different theme: ‘foundation skills’, ‘parent empowerment’ and ‘parent/adult interaction’. Bidirectional arrows connect the balloons and the boxes. The model indicates that assessment drives the selection of relevant themes and that factors such as time and availability, background and age modify these selections.

Figure 12 shows a Venn diagram with overlapping circles representing a speech module and a language module. Within the language module are the themes ‘foundation skills’, ‘comprehension’ and ‘structure and content’; within the speech module are the themes ‘sound awareness’, ‘speech’, ‘self-monitoring’ and ‘generalisation’. There is an implied order within the modules; an arrow from ‘language’ to ‘speech’ further implies an order in which language is tackled before speech. In the overlap of the Venn diagram are the themes ‘parent empowerment’, ‘parent–child interaction’ and ‘child participation’, suggesting that these elements are included in both language and speech modules.

Final typology of speech and language therapists’ practice with preschool children with primary speech and language impairment

Figure 13 summarises our current understanding of how therapists might combine interventions from the themes to construct an intervention. The figure uses notation from systemic grammar networks. The notation illustrates the choices available and how choices are combined. 114 The following explanation therefore provides a working hypothesis, based on the studies contained in this chapter, of how the themes might be considered in the planning of an intervention.

FIGURE 13.

Working hypothesis of the typology of speech and language therapy for preschool children with PSLI.

Following assessment, therapists would focus interventions on a particular stage and aspect of a child’s speech and language development (‘speech’, ‘expressive language’ and ‘comprehension’). Depending on the child’s profile, they might choose one or more aspects to focus on. They would, simultaneously, be considering how to support a child’s increasing meta-linguistic or self-monitoring skills and the process of generalisation. The particular focus would depend on the stage of therapy. As well as these ‘content’ areas, therapists would also consider how well the child’s ‘foundation skills’ support their ongoing learning, for example whether or not the child’s attention is sufficient to support learning in the particular content area. They would also consider a child’s ‘functional communication skills’ to ensure that any targets were functionally useful to the child and that he or she had the social communication skills to support functional communication. Finally, therapists would consider what support needs to be in place for adults who are in regular contact with the child, ensuring that they understand the nature of the child’s language learning impairments and are able to interact with the child in a way that supports his or her speech and language development.

Content analysis of focus groups

A coding framework was developed by a member of the research team in NVivo 9 software. The framework sought to classify interventions into different categories. The categories were designed to incorporate the different types of disorder within PSLI (e.g. expressive or receptive language, speech, social communication) as well as the different types of intervention (i.e. activities, strategies, programmes or resources). Categories in the framework were not mutually exclusive, so, for example, an intervention might be coded as both an expressive and a receptive language activity. The results of the content analysis are shown in Table 19.

As shown in Table 19, a large range of interventions was coded. It can be seen that, although participants were encouraged to give detail rather than list programmes or resources, a large number of programmes and materials were coded. Another noteworthy finding is that a large number of strategies for language were coded (n = 150) relative to receptive (n = 43) and expressive (n = 33) language activities. By contrast, speech activities (n = 90) were more commonly coded than speech strategies (n = 51). There were also a large number of strategies (n = 123) that were classified as general speech, language and communication strategies.

Therapy rationale

One of the aims of the focus groups was to explore SLTs’ rationales for therapy. There were 178 references to rationales coded in NVivo 9. These are not going to be reported in full here; however, some noteworthy points are considered. SLTs reported a wide range of rationales for their intervention use, depending on the intervention being delivered. For example, a rationale provided for ‘adult–child interaction’ was ‘to adapt their communication style to then give the child a chance to learn words more effectively’ (SLT_107). In relation to information-carrying word (ICW) activities, by contrast, an example of a rationale provided was ‘you can do verbs and you can do vocabulary development at the same time, as you’ve got a structure you can embed other things in to’ (SLT_025). Thus, rationales showed great variation depending on which particular activity or strategy was being referred to.

Rationales were also provided at a broader level, for example relating to later outcomes [‘importance of language as the foundation of learning and literacy’ (SLT_107)] and evidence of effectiveness [‘I think there’s been some recent evidence that actually it’s pretty strong and robust in terms of being effective intervention’ (SLT_108)].

It was noted that SLTs appeared to find it difficult to provide explicit rationales for their interventions, with many rationales provided in terms of the intervention activity aiming to improve the area that it was targeting, without a clear description of the mechanism of change. Making rationales explicit often involved detailed probing from the facilitator.

Therapy modifications

The focus groups also took a preliminary look at factors that might lead to the individualisation or modification of therapy. Many SLTs talked about the importance of adapting therapy and appeared to find it difficult to describe work in relation to a generic child. Therapists often prefaced their descriptions about the use of an intervention with ‘well it depends’ or ‘depending on . . .’, with ‘depends/depending’ being said 62 times across the focus groups. The following quote from a SLT who attended the focus groups indicates how important the role of adaptations is in planning therapy:

. . . but it’s not one size fits all and you can’t be, that’s not what we‘re about as therapists, now I’m aware that as we’re talking, we’re talking so much about adapting and changing and in professional judgement and those split second judgements that you make and then adapt your own interaction or your own information.

SLT_095

Modifying factors were coded in NVivo 9. Table 20 shows the numbers of factors coded that relate to the child, the family and the context. Examples of commonly raised factors that lead to modifications in relation to these areas are described in more detail in the following three sections.

The frequency of use of interventions

Interventions and modifying factors identified from the focus group content analysis were used to populate two surveys. Alongside interventions identified from other sources, such as the BCRP,37,89 the two surveys aimed to explore which intervention components SLTs use with preschool children with PSLI, as well as to examine what factors lead SLTs to adapt or modify their use of intervention components. Both surveys utilised predetermined response options. The focus was on components of interventions, that is, activities and strategies, rather than on specific programmes or resources, which are more generalised and may include a range of components within them.

In the first survey SLTs were asked to report the frequency of use of intervention activities and select whether they use them with all preschool children (aged between 2 years and 5 years 11 months) with PSLI, those with primarily language difficulties or those with primarily speech difficulties. They were also asked to select factors that led them to adapt their interventions. In the second survey respondents were asked to base their responses on a specific preschool child with PSLI. In the second survey SLTs were not asked directly about factors that were reported to lead to adaptations; instead, the survey was designed to gather more detailed child, family and contextual information, the purpose being to explore whether or not particular profiles of child/family/context could be linked to particular patterns or individualisation of interventions.

The lists of interventions utilised in survey 1 was refined for survey 2 following the results of that survey and more detailed analysis of the SLT focus groups. Further, to reduce the length of the lists of interventions that SLTs were presented with in survey 2, relevant lists of intervention activities or strategies were presented in relation to the case that they had selected (i.e. speech, language or social communication). These differences mean that, although the surveys are reported together here, there are not matching intervention data for all of the presented interventions. It is also noteworthy that a higher proportion of SLTs would be expected to select interventions in survey 1 than in survey 2, as survey 1 refers to general intervention use whereas in survey 2 responses are based on a specific child.

Differences between the use of strategies and the use of activities

Table 23 displays intervention activities and Table 24 displays intervention strategies that SLTs reported in survey 1 that they used always/sometimes with ‘all children’. It can be seen that a much higher proportion of SLTs reported using strategies than activities.

Table 25 displays activities and Table 26 displays strategies that were reported to be used for children with primarily language difficulties. Comparison between the two indicates that higher percentages of SLTs reported using strategies than activities for children with language impairments.

Table 27 displays activities and Table 28 displays strategies that were reported to be used for children with primarily speech difficulties. Comparison between the two indicates the opposite pattern, with slightly higher percentages of SLTs reporting using activities than strategies for children with speech impairments.

These findings from both surveys indicate that there are different patterns of intervention use in relation to strategies and activities. The high percentage of SLTs who report that strategies are relevant for ‘all children’ indicates that they are more generalised approaches that are adopted for children with PSLI. Activities, by contrast, are more targeted and specific and thus are less universally applicable to all children. The higher proportion of SLTs reporting using activities than strategies for children with speech delay also indicates that work for speech delay, in particular, is more targeted. Although strategies appear to be important for ‘all children’ they appear to take a more predominant role in therapy for children with language delay.

Interventions with variable use

Interventions could be organised into three groups: (1) those used by a high proportion of therapists; (2) those used by very few therapists; and (3) those used by about half of the sample (40–60%). This last group is interesting as it represents those interventions with the highest variability in use, indicating a lack of consensus regarding their value.

In survey 1, this middle range is particularly interesting to explore as therapists reported on work that they did ‘always or sometimes’ for all children with PSLI. This means that responses should reflect the participating SLTs’ usual practice and thus any differences indicate variability. In survey 2, by contrast, the SLTs were reporting work that they did with a specific child. In this instance it can be expected that many interventions will be used by about half of the therapists, as they may or may not be relevant for that particular child.

Table 29 presents those interventions that were used always/sometimes or occasionally/never by 40–60% of participants according to survey 1. There are four interventions that show this variability in use; all four of these interventions are activities rather than strategies.

Additional responses

In both surveys therapists were able to give additional responses to either provide more detail of their interventions or raise interventions that may not have been covered in the surveys. In survey 1, 122 respondents provided further information about the interventions that they use for preschool children with PSLI. In survey 2, 158 respondents felt that the listed activities and strategies covered all of the interventions that they had used with the specified child. The remaining 17% (n = 32) did not feel that this survey covered all of the interventions that they had used and they therefore listed additional activities and strategies. For details of these responses from both surveys see Appendix 28.

Factors that lead to adaptations of interventions

Both surveys explored factors that are important in the individualisation of therapy. Tables 30–32 indicate the responses obtained from survey 1 in relation to a range of factors that may or may not be influential in modifying the practice of SLTs. Table 30 provides those factors for which there appears to be high consensus that they are influential in modifying practice (> 60% reported that ‘yes’ or ‘usually’ they would adapt their interventions in relation to these factors); Table 31 provides those factors for which there appears to be a lack of consensus that they are influential in modifying practice (40–60% reported that ‘yes’ or ‘usually’ they would adapt their interventions in relation to these factors); and Table 32 provides those factors for which there appears to be high consensus that they are not influential in modifying practice (< 40% reported that ‘yes’ or ‘usually’ they would adapt their interventions in relation to these factors). It can be seen that there are a large number of factors (16/28) for which there is a high level of consensus that they are influential in modifying practice. This indicates that SLTs adapt their interventions in relation to a wide range of factors.

As with the patterns of variable intervention use, the middle range or 40–60% is interesting as it represents those factors with a lack of consensus in terms of whether or not they are influential in modifying practice, in this case maternal depression, culture, medical history, views of other professionals around the child and parental preference.

Cluster analysis was conducted for survey 2 to explore patterns of individualisation of interventions. Cluster analysis classifies a set of observations into two or more mutually exclusive groups. The purpose is to discover a system of organising groups, in which groups share properties in common. The specific aims of cluster analysis in the present research were to establish whether or not any patterns of intervention emerged. Cluster analysis for both speech and language cases generated two clusters. These clusters appeared to indicate differences in the use of strategies; however, the quality of these clusters was rated as poor and fair. Statistical analysis was then conducted to see if these clusters were related to any child, family or contextual variables (i.e. to see if these factors influenced certain patterns of intervention). The intention was to explore whether or not patterns of intervention were associated with specific variables, which might provide information on the way that SLTs individualise their therapy. Although there were a number of statistically significant findings, the clinical significance of these is questionable, as the quality of the clusters was rated as only fair or poor. Furthermore, a large number of (chi-squared) tests of difference were conducted, increasing the likelihood of false positives. However, the full analysis and significant chi-squared test findings, in relation to child, family and contextual independent variables, are reported in Appendix 29.

Relating intervention components to the developing typology

Having established a wide range of interventions that SLTs report using with preschool children with PSLI, the next step was to explore how they relate to the therapy aims or the typology themes that have emerged. Two key methods for relating intervention components to the typology themes were adopted:

• focus group cross-tabulations (see Chapter 1, Focus groups)

• SIG activities: essential/advisory/not used activity and sorting exercise activity (see Chapter 1, Consensus events).

Final collation and analysis

As part of the process of evaluation, it was necessary to collate the various interventions that had emerged from the programme and to examine the validity of their inclusion and association with the typology themes. Using a data-driven process, interventions associated with each theme of the typology were collated. Interventions were added to a typology theme only when there was evidence from one of the data sources that it was associated with that theme. The following data sources were used (in the following order): focus group cross-tabulation of themes and content analysis, card-sorting exercises, survey 1, survey 2 and the national validation check.

In a final checking exercise, any interventions that could be considered to be ‘programmes’ or ‘resources’ were removed. Intervention activities and strategies that were similar were also collapsed. It was not possible to aggregate data across the activities, as the questions all differed slightly in terms of how participants rated or organised their responses and therefore in terms of how the responses can be interpreted. However, by collating the data from the various sources, it was possible to build a picture of the relative popularity of interventions in relation to the typology themes.

The collating exercise produced a list of interventions for each theme of the typology, with the resulting lists varying in length. Table 33 provides a summary of the number of relevant intervention activities and strategies that emerged for each theme. A total of 144 different intervention activities and strategies emerged after removal of duplicates and programmes or resources. The theme ‘speech’ had the largest number of interventions assigned to it (n = 45); however, there were relatively similar numbers across the themes, with seven of the nine themes having between 35 and 45 relevant interventions assigned to them. ‘Generalisation’ (n = 28) and ‘self-monitoring’ (n = 16), in particular, had lower numbers of relevant interventions assigned to them. It should be noted that the interventions were not mutually exclusive, that is, the same intervention could be linked to multiple themes, although it is likely that implementation of an intervention will vary according to the context of the theme.

Table 34 indicates the number of interventions associated with multiple themes. It can be seen that, although most were linked to only one theme (n = 81), many could be linked to multiple themes.

Strength of associations of interventions with typology theme

Rather than presenting the large amount of data on intervention components, it is more interesting to review the collated data to understand the associations between interventions and typology themes. To evaluate the overall strength of the data supporting the association of an intervention with a theme, the intervention components were rated in relation to whether they appeared to have ‘strong’ or ‘medium’ support from the SIG and national consensus day activities. The strongest indication of a link between an intervention and a theme comes when there is a high level of agreement between participants that an intervention is relevant to a particular theme. For this reason, the following definitions were used to classify a strong or medium association in relation to the SIG data:

• strong: the intervention was regarded as ‘essential’ to the theme by > 60% of participants or > 60% of participants allocated the intervention to that theme in the sorting exercise

• medium: the intervention was regarded as essential by 40–60% of participants or was regarded as desirable by > 40% of participants.

The survey data were classified as ‘strong’ when > 60% of SLTs in either survey reported using an intervention and as ‘medium’ when 40–60% in either survey reported using an intervention. As different percentages for use of the interventions were obtained from the two surveys, when differences in classification of the strength of the data occurred between the two surveys, the highest value was taken. This decision was taken as the higher values obtained in survey 1 reflect general use rather than use with a specific child.

In Tables 35–43, the interventions that have been identified for each theme of the typology are presented. The tables present in columns those interventions that have strong and medium evidence from the sorting activities at the SIGs. The rows indicate corresponding survey evidence when applicable. If either there were no survey data available or there was only weak support from survey data the term ‘less support from survey data’ is used. The concepts of strong and medium support described earlier are used throughout.

Interventions that had only weak support from the SIG data to link them to a theme are not shown. It is acknowledged, however, that this does not rule them out as being relevant to a theme; it indicates only that they are less likely to be a priority for that theme. For example, for those interventions that spanned multiple themes, SLTs might have instigated a ‘split vote’. Therefore, the activities and strategies that are linked to each theme from the typology are not necessarily a complete list of all of the activities and strategies that could be used for each theme. In addition, one would not expect all activities and strategies to be appropriate for every preschool child with PSLI. Assessment of each child is of course required to identify the most appropriate targets and, hence, the related interventions. Nonetheless, Tables 35–43 provide an overview of the commonly used activities and strategies for each of the themes.

What was the focus of the studies?

Of the total of 33 studies included in this review, 26 have change in a child’s speech sound system as either their primary or their secondary focus. 140,144–159,161–167,169 For 25 studies, change in the sound system was the primary focus; for the remaining study,139 change in the sound system was an additional focus. Specifically, the primary focus was change in word-finding skills and measurement of change in the speech sound system was carried out as a by-product of the intervention received. 139

Out of the remaining seven studies, one focused on change in prosody of speech rather than the speech sound system, specifically change in the stress pattern of multisyllabic words and sentences. 140 This study observed change in the spoken productions of stress patterns in multisyllabic words and phrases in children with speech and language impairment. Although this is a study of speech production, the focus is on speech prosody rather than the speech sound system and, as a consequence, this study is considered separately to the other 26 studies focusing on children’s speech sound systems.

The focus of the other six studies141–143,160,168,171 was change in a child’s phonological awareness skills following specific intervention targeting phonological awareness. In each study the participants had speech impairment, language impairment or both. In many cases, phonological awareness tasks are used in speech and language therapy intervention with children of preschool age to facilitate improvements in speech or phonological development in particular. These studies, therefore, were focusing on the changes observed in children’s phonological awareness when speech and language skills themselves are impaired. Phonological awareness tasks include activities such as rhyme identification and production, phoneme addition and deletion, phoneme segmentation and phoneme blending. In some cases they also include alphabetical tasks such as letter naming and nonsense word naming.

Country of origin of the papers

Of the 32 papers included, 22 were from the USA,139–159,170 five were from Canada,160–164 three were from Australia165–167 and two were from the UK. 168,169

Design

The most common research design was that of a case series [18 of the studies (in 17 papers139–145,148–150,152–154,158,165–167) employed a case series design]. These case series studies used a combination of multiple baselines, staggered starts and a crossover design. Four studies used a RCT design,157,162,163,168 five used a between-groups method,156,160,161,164,169 one used a single subject multiple baseline design,159 one used an alternating treatment design,147 three were case studies146,151,155 and one used a crossover design with multiple baselines. 171

Size of the sample

By nature of the study designs the sample sizes were typically small. Sample sizes ranged from 1146,151,155 to 159. 169 Ten of the studies had > 10 participants143,157,160–164,168,169,171 but the most common number of participants was two. The total number of children represented in this theme was 542, with a mean of 16.4 (median 4.0).

Child participant demographics

In this category all 33 studies recruited child participants with a language impairment (i.e. none was classed as ‘at risk’).

Not all of the studies reported the gender split of participants. Among the available data there were 170 boys and 62 girls across the studies in this theme. The average age of the participants across all 33 studies was 51.6 months, ranging from 32.0 months to 66.0 months. 141,169

Assessment measures

Delivery of the interventions

For studies in the area of speech production or articulation there was relatively little variation in who delivered the interventions and where they took place. In all but 5 of the 33 studies, the intervention was delivered by a SLT. In those five studies the intervention was carried out by a combination of SLT and parent,143 a research assistant,163 educators and paraprofessionals160,171 and a graduate student SLT. 144 Twenty-three studies were undertaken in a clinic setting. 146,148,151–159,161–164,168,171–175 One was carried out in the home context,143 one was split between the clinic and preschool169 and another four were carried out purely in a preschool context. 141,154,160,171 The remaining four studies did not state where the intervention was undertaken. 139,159,161,164

Number of treatment sessions

It is difficult to make comparisons across all 33 studies of the number of treatment sessions. The information detailed across the studies is not uniform but overall the interventions were most likely to be delivered either twice a week (seven studies145,146,154–156,165,167) or three times a week (nine studies141,142,147,148,151,152,157,159,171). The duration of an intervention varied from 30 to 60 minutes. The lowest number of sessions delivered was just six,161 with most studies reporting 10–120 sessions. The longest intervention lasted for 9.5 months; in this study a child was seen 67 times. 158 In most of these studies the number of treatment sessions also varied from one child to the next as the intervention was stopped once a child reached 50–90% generalisation on the probes.

Comparison with speech and language therapists’ practice

Table 48 shows the activities and strategies identified by SLTs who participated in the Child Talk focus groups as being ways of working or elements of practice that they use with children with PSLI. It also shows how essential to working on improving speech these interventions are, as rated by SLTs participating in consensus group events (SIG) (see Chapter 1, Methodology overview). Finally, Table 48 shows how many of the studies included in the review used each of these interventions.

None of the SLTs specifically mentioned use of picture cards or computer-based programmes but these could also have been classed as fun activities and some of the studies that explained the computer-based programmes detailed built-in feedback within the computer games.

Other interventions mentioned in the studies but not by therapists in the research programme were:

• core vocabulary165

• drills146,164,165

• giving an auditory model for the student to imitate144–147,149–153,159,164

• continuous reinforcement144,145,159

• correcting the child’s errors (corrective modelling)144,145,152,157

• spontaneously saying words containing the target sound147,149–153,159,164

• nonsense words147–149,151–153,159

• contrasting supralaryngeal and laryngeal distinctives153

• focused stimulation of key words154

• adult cues, for example descriptions of sounds (long sound, etc.)154

• sound identification training (SAILS program)156,163

• broad target recasts157

• cuing for word retrieval139

• speech perception training161

• teaching prosodic patterns (big and little)140

• auditory bombardment. 140

Evidence of outcomes

Overall, the outcomes were mixed in terms of evidence of positive change. Five studies reported no positive changes/differences between different interventions,143,151,156,158,169 for example it was suggested that minimal pairs treatment was not successful for the children in that study. 158

When positive change was reported, studies found that the following was observed: phonemic expansion,148 greater phonological awareness,142 improved rhyme production,171 increased intelligibility,157,165 improved articulation accuracy,163 greater consistency of word production165 and improved phonemic perception. 163

Success was achieved through a computer-driven speech perception training programme, provided concurrently with sound production training,163 a rhyming module of a phonological training intervention in combination with systematic training in the alphabetic principles,142 a rhyme intervention,171 a maximal opposition approach (two studies),146,147 broad target recast,157 targeting segmental compositions of individual children148 and development of core vocabulary. 165

What was the focus of the studies?

All seven studies assessed children’s comprehension as an outcome. Five of the seven studies targeted children’s comprehension in addition to other aspects of language. 169,172,173,175,176 Only two studies targeted children’s comprehension specifically. 32,174

Country of origin of the papers

Of the seven papers included in this theme, three were undertaken in the USA32,172,173 and four were undertaken in the UK. 169,174–176

Design

Four studies used a RCT methodology,32,169,174,175 two used a case series design with multiple baselines172,173 and one was a cohort study with pre and post measures. 176

Size of the sample

In total, 294 children were represented by the seven studies in this theme. Study sample size ranged from 2 to 159, with a mean of 22.5 and a median of 30.0.

Child participant demographics

One study recruited children who were either language impaired or at risk of language impairment. 32 The other studies in this category all recruited children who were language impaired. In total, 195 boys and 68 girls were specified as participating in the studies, with an additional 36 children not having their gender reported. The mean age of the children was 40 months (range 27.5–50.0 months).

Assessment measures

Sixteen different measures were used across the studies. Eleven of these measures were used only once and many did not assess comprehension. Of the comprehension measures used, only three were used in more than one study: the Peabody Picture Vocabulary Test (PPVT),201 the Preschool Language Scale (PLS)202 and the Reynell Developmental Language Scales (RDLS). 203

Outcome measures

Twenty-one different outcome measures were used across the studies in this theme; however, only one measure of comprehension (RDLS) was used twice.

Delivery of the interventions

Two interventions in this typology was delivered by a SLT,169,172 two were delivered by both a SLT and parents,174,176 one was delivered by a parent alone following parental training by SLTs,175 one was delivered by trained research assistants32 and one is reported to have been delivered by a clinician but it seems likely that this person was a SLT. 173

A range of settings was used across the studies. One study specifically stated that therapy was undertaken in a community clinic169 and another reported that therapy was undertaken in ‘a clinic’;173 the other studies took place in a Head Start centre,32 the home175 and preschool/nursery and the home,174,176 with one study not specifying the location (given the nature of the study it is likely to have taken place in a clinic setting). 173

Number of treatment sessions

The number of treatment sessions ranged from 8 to 78, with these usually being delivered over a 3- to 6-month period. Studies reported that sessions lasted between 15 and 75 minutes. Comparison between studies that focus on comprehension alone and those that include production is difficult. Comprehension-focused studies included from 30 to 40 comprehension sessions whereas those studies that included a focus on production included fewer sessions (16 sessions of 15 minutes over 8 weeks, i.e. twice a week).

What was the intervention?

The types of intervention varied across studies, for example two studies involved picture activities,172,173 one involved a book-sharing activity,32 two involved a range of games and therapy techniques174,175 and one focused on parent–child interaction. 176 The remaining study was a pragmatic RCT and stated that the children in the therapy group ‘received the one-to-one speech and language therapy routinely offered by the therapist . . . and tailored to their individual needs’ (p. 2). 169

Comparison with speech and language therapists’ practice

Table 49 shows the activities and strategies identified by SLTs who participated in the Child Talk focus groups as being ways of working or elements of practice that they use with children with PSLI. The table also shows how essential to working on children’s comprehension these interventions were judged to be by SLTs participating in consensus group events (see Chapter 1, Methodology overview) and how many of the studies included in the review explicitly mentioned using each of these interventions.

Evidence of outcomes

All studies reported improvements in aspects of children’s language. Six of the seven studies saw improvements in comprehension32,169,173–176 and one reported evidence of generalisation from children’s comprehension skills to production skills. 172 Comprehension was seen to improve when the intervention was delivered by parents and also when it was delivered through direct speech and language therapy; between-group differences were found when control groups consisted of ‘no intervention’ but when intervention groups were compared with groups receiving speech and language therapy in a different form then comprehension improvements tended to be comparable in both groups.

What was the focus of the studies?

Although these 27 studies have been grouped together under the theme of ‘expressive language’, the detailed foci of the studies varied. Foci included working on specific aspects of expressive language (e.g. syntax, utterance length, vocabulary) and training parents to support their children’s language development. Of these 27 studies, 25 included interventions to improve expressive language as the primary focus of the study. 70,135,137,139,154,157,169,170,173–187 For two studies,134,168 expressive language was a secondary focus of the study. The following list shows the various expressive language foci:

• expressive and/or receptive language (10 papers representing 11 studies70,134,135,175,176,177–181)

• language delay (three studies182–184)

• parental intervention for language delay (one study174)

• social communication/language (four studies137,138,170,185)

• phonological awareness (one study168)

• language and grammar (two studies173,186)

• teaching syntax (one study154)

• phonology and utterance length (expressive language) (one study157)

• operant conditioning for non-verbal/minimally verbal children (one study187)

• language and phonology (one study169)

• word-finding difficulties (one study139).

Country of origin of the papers

Fifteen papers in this theme were from the USA,134,135,137–139,154,157,170,173,181,183,184,185–187 seven were from the UK,70,168,169,174–176,179 two were from Canada,180,182 one was from Korea178 and one was from Germany. 177

Design

A case series multiple baseline design (n = 5134,139,178,179,183) was the most common study design chosen for the studies that fell into this theme, followed by a RCT design (n = 5157,168,174,175,177). The following study designs were also used: case series (n = 1154), multiple baseline across dyads (n = 4138,170,184,185), single-subject alternating treatment design (n = 1181), pre/post between groups (n = 4135,176,180,182), between groups (n = 370,175,186), case study (n = 1187), between groups, both comparison (n = 2137,169) and between subjects with a crossover treatment design (n = 1173).

Size of the sample

The total number of child participants across all of the studies in this theme was 913. In total, 13 studies had < 10 participants. 134,138,139,154,170,173,178,179,181,183,184,185,187 However, overall, the sample size in this theme was relatively large compared with that in the other categories, with the mean number of children per study being 32.6 (median 15.5 children per study).

Child participant demographics

Of the 27 studies (26 papers), only three studies recruited ‘at-risk’ children,135,137,138 with one study recruiting both children at risk and those with a language impairment. 135 The remaining studies all recruited children with existing language impairment. A number of the studies do not report the gender of the child participants. However, when recorded, there were 257 boys and 154 girls; thus, gender was not provided for 502 participants. The mean age of the child participants was 43.1 months, with a range of 25.0–66.0 months.

Assessment measures

The total number of different assessment measures used across all of the studies in this theme was 71, with 22 of these being identified as appropriate for use as a measure of expressive language.

The three most commonly used assessments that measure expressive language were MLU (n = 10154,157,175,176,178,179,181,182,186), the Preschool Language Scale-3 UK Edition (PLS-3-UK) (n = 8134,138,157,169,170,173,176,185) and the RDLS (n = 470,174,175).

Outcome measures

The total number of different outcome measures used across all of the studies in this theme was 72, with 26 being identified as appropriate for use as a measure of expressive language.

The most commonly used outcome measures to evaluate changes in expressive language were MLU (n = 13138,154,157,170,175,176,178,179,181,182,185,186) and measures of specific expressive language parameters (rate of noun use, number/mean of different words) (n = 8134,135,137,169,173,174,181,182).

Studies that were included in this theme frequently targeted other skills, such as functional communication, speech and generalisation, and frequently measured these other aspects of speech and language too, such as receptive vocabulary using the Peabody Picture Vocabulary Test – Revised (PPVT-R)204 (n = 4134,139,180,183) and the comprehension scale of the RDLS (n = 470,174,175).

Delivery of the interventions

Eleven of the studies stated that a SLT carried out the intervention. 70,139,154,157,168,169,178,180,183,186,187 In four studies it was a combination of a SLT and a parent,174–176,177 in three studies it was a parent alone134,175,182 and in three studies it was a teacher. 170,179,185 Other studies reported that the intervention was carried out by a researcher acting as a teacher,138 EYPs working with parents,137 an assistant SLT184 and graduate student ‘clinicians’181 and that training to teachers was delivered by SLTs,135 with one study stating that the intervention was delivered by a ‘clinician’ but giving no further details. 173 Five of the studies were carried out in a preschool or nursery environment,70,135,178,179,186 five studies were carried out at a child development/Head Start centre,137,138,170,184,185 five were carried out in a speech and language therapy clinic157,168,173,186,187 and a further study was carried out at both a clinic and a preschool. 169 Two studies were carried out within the home environment134,175 and a further three were carried out at both home and at a preschool/nursery. 174–176 Not all of the studies stated where the intervention was carried out.

Number of treatment sessions

There are a large number of studies in this theme and therefore summarising the number of treatment sessions that the children received is not straightforward. In some cases the studies described long-term interventions, for example there was a 3-month programme,181 a 4-month programme182 and one study took place over an entire academic year. 135 At the other extreme a few studies described an intervention taking place over 12 sessions that were delivered twice weekly (e.g. Gallagher and Chiat174). The duration of the sessions varied from 30 to 75 minutes. Finally, some studies described home-based parent training interventions in which the interventions were delivered at least in part by parents (e.g. Gibbard175); these were more individualised in terms of the number and duration of ‘sessions’.

What was the intervention?

This section summarises what took place during the interventions that fall into the ‘expressive language’ theme. Approaches included modelling (n = 9137,169,174,176,177,181,182,184,186), imitation (n = 4173,174,176,178), targeting specific phrases (n = 1138) and sentence recasting (n = 1174).

In terms of materials used during the interventions, the use of picture books or storybooks was commonly mentioned in the studies (n = 6134,135,138,174,177,183), as was a more general use of pictures (often cards) (n = 4154,173,184,187). Toys and play materials were also frequently used (n = 9135,138,169,170,180,182–184,185). These were varied and included dramatic play toys (e.g. kitchen, grocery store and hospital), role-play materials (e.g. clothing, objects) and manipulative toys (e.g. cars, blocks, trains). These items were commonly used for listening tasks and talking tasks and were also used by children to give instructions to other children or to other toys. In some cases the toys were not part of the intervention but were used during the session as a reward.

Parents were also commonly involved in the interventions described (n = 7134,137,174–176,177,182). Examples of how parents were involved included encouraging sharing of picture books,177 introducing parents to language modelling techniques176 and assignment of home practices. 174,183 Feedback was sometimes given to parents through home visits and/or by using videotapes of the child and mother during free play. 176

Reinforcement was frequently integrated into the interventions. Positive reinforcement was given in the form of verbal praise (n = 5138,154,178,186,187), correct responses were reinforced with the stimulus requested (n = 1;187 tokens were given for correct responses with tokens being able to be redeemed for a reward such as ‘playtime’) and, as indicated above, in some studies toys and games were available after the children had performed the therapy task. 173,187 Negative reinforcement was also used (n = 2178,187), for example inappropriate responses were punished continuously by presenting the verbal stimulus ‘no!’ and concurrently withdrawing the stimulus picture for a brief period and avoiding eye contact187 and by partial responses and a request to the child to repeat. 178

Comparison with speech and language therapists’ practice

Table 50 provides details of the strategies and activities mentioned by the SLTs who participated in the Child Talk focus groups in relation to work that aims to improve the children’s expressive language, in quantity, vocabulary or structure. The table also provides the percentage of SLTs ranking each intervention as essential and how many of the studies included in the review explicitly mentioned using each intervention.

Evidence of outcomes

Overall, the studies that looked at work that aimed to improve children’s expressive language, in quantity, vocabulary or structure, report fairly positive outcomes. As the studies had varied aims, comparison is difficult (e.g. some of the studies aimed to examine differences between different types of interventions), but overall only three of the studies reported no improvement in language for the child participants. 135,169,183

In summary, despite numerous different approaches to intervention, the interventions when grouped together led to a reduction in word-finding substitution,139 an increase in social communication behaviour,138 an increase in the use of verbal behaviours and target vocabulary words,138 improved expressive language skills,175 an increased amount of child talk during play,185 an increase in the number of requests and descriptive talk and an increase in the complexity and diversity of language. 185

In terms of working with parents, the studies demonstrated mixed outcomes. Successful outcomes were achieved through increasing the number of interactions between a mother and her child, which in turn led to increases in the child’s vocabulary and grammatical complexity. 182

Some studies reported that not all of the participants benefited from the intervention. As the sample sizes were generally small this could mean that, for example, two out of three children showed improvement (e.g. Hart and Gonzalez154).

In some studies a comparison was made between two different approaches to therapy, for example in one study significantly more children improved in a ‘phonological awareness’ group than in a ‘language stimulation’ group. 168

Country of origin of the papers

Both papers representing this aspect of the typology were undertaken in the USA.

Design

One study consisted of three case studies154 and the other used a between-group study design with four dyads. 138

Size of the sample

The total number of child participants across the two studies was only 11 (eight in one study138 and three in the other154).

Child participant demographics

It should be noted that, although the studies in this category all recruited children with a language impairment, one study recruited children who were at risk of impairment. 138 The gender split was not detailed in either of the studies. The average age of the participants across both studies was 51.8 months, ranging from 43 months to 60 months.

Assessment measures

Five different assessment measures were used in this typology, with each one used only once; none of them measured self-monitoring:

• child behaviour checklist (Caregiver Teacher Report Form; CTRF)205

• HAPP191

• PLS-3-UK195

• Preschool Language Scale-4 UK Edition (PLS-4-UK)206

• Social Skills Rating System (SSRS). 207

Outcome measures

Nine different outcome measures were used in this typology, with each one used only once; none of them measured self-monitoring:

• HAPP191

• MLU

• Peer Language and Behaviour Code208

• PLS-4-UK206

• relatedness

• total occurrences of major phonological deviations (TOMPD)

• turn taking

• use of child’s name

• use of target vocabulary.

Delivery of the interventions

One of the studies described interventions undertaken by a researcher acting as the teacher138 and the other study was undertaken by SLTs. 154

Both studies were conducted in preschool settings. One was reported as being a Head Start centre, suggesting that this study was undertaken in an economically deprived area. 138

Number of treatment sessions

In the study by Stanton-Chapman et al. 138 there were 12 individual treatment sessions, delivered twice weekly for 30 minutes. In the study by Hart and Gonzalez, three to four 20-minute sessions were delivered per week. 154 The first dyad had 16 sessions, the second had 14 sessions and the third had 12 sessions. 154 With only two studies represented in this theme of the typology, a comparison of the number of treatment sessions delivered across studies is of limited value.

Comparison with speech and language therapists’ practice

Interventions (whole and individual activities and strategies) used in the studies were mapped against the activities and strategies provided by SLT participants from the Child Talk focus groups and online surveys. Table 51 shows which interventions were ranked as essential within consensus group events (SIGs) and how many of the studies used each of these interventions.

In addition to the interventions in Table 51, the two studies used other activities and strategies:

• shared book reading138,154

• role play138

• reflective statements (redirects – models, direct instructions, indirect instructions)138,154

• modelling138,154

• labelling138,154

• praise. 138

Evidence of outcomes

There is preliminary support for the effectiveness of communication-centred approaches in facilitating change in children’s phonological system. 154 However, neither study provided any theory or suggestion of how self-monitoring helps this to happen. Peer-directed interventions were generally effective in increasing social communication behaviours of children who had fewer social interactions at baseline. 138 Because of the different aims of the studies included in this theme (i.e. phonology, social interaction), it is not possible to draw conclusions as to the effectiveness of self-monitoring.

What was the focus of the studies?

Although each of these studies included ‘generalisation’ in the research design, the overall focus of each of the studies was more varied. The 12 studies with a main focus on generalisation are detailed below:

• Comprehension treatment using cards. 172

• Effects of a multicomponent intervention, implemented in a play context, which aimed to improve children’s social–communicative interactions. 170

• Maximal oppositions. 146

• Acquisition of word-initial three-element clusters. 152

• Intervention to learn the correct use of three pronouns and the verb form ‘are’. 187 The target behaviours were trained individually and positively reinforced when correctly used, or modelled, with children punished for inappropriate responses.

• An intervention with two parts: (1) a script-based intervention in which individualised targeted semantic constructions were trained and (2) generalisation of semantic changes. 178

• Investigation of the effects of a multicomponent intervention strategy to increase peer-directed social communication. 138

• Treatment designed to facilitate both sentence length and speech intelligibility. 157

• Investigation of how phonological information contributes to the effectiveness of treatment. 139

• Minimal pairs treatment. 158

• Articulation – effectiveness of treatment with repeated probes. 159

• The effect of a prosodic intervention on children’s metrical patterns. 140

Country of origin of the papers

The dominance of research based in the USA in relation to encouraging generalisation is striking. Of the 24 studies included in this typology, 20 were from the USA,138–140,144–147,149–152,154,157–159,170,172,173,179,187 one was from Australia,168 one was from South Korea,178 one was from the UK174 and one was from Canada. 164

Design

Fourteen studies used a case series design (including multiple baseline, staggered start, multiple baseline; single subject design with staggered baseline; multiple baseline alternating crossover design). 138–140,144,145,150,152,154,158,167,170,172,178,179 Case studies were used three times (using an AB design or multiple baselines). 146,151,187 Other designs included RCTs,157,174 alternating treatments,147 an across-subject staggered multiple baseline design,149 a between-group design,164 a single subject multiple baseline design159 and between subjects with a crossover treatment design (including multiple baseline measures). 173

Size of the sample

In total, 204 children were represented across the 24 studies. Only three studies had > 10 participants, with 24 participants,174 48 participants164 and 52 participants. 157 The mean number of children per study was 8.5, with a median of 3.5 per study.

Child participant demographics

It should be noted that, although the studies in this category all recruited children with a language impairment, one study also recruited children who were at risk of impairment. 138 Four of the studies did not provide information that could be used to identify the gender of the participants. Of the remaining studies, there were 73 male and 23 female participants. The age range of children across the studies was 35.0–66.0 months, with an average age of 50.8 months.

Assessment measures

Across the 24 studies, 53 different assessment measures were used. The most frequently used assessment measures relevant to generalisation were MLU, used seven times,140,147,157,167,172,178,179 and percentage consonants correct, used four times. 140,144,157,167

Outcome measures

Seventy-seven different outcome measures were used across the 24 studies, with 71 being used in only one study. The majority of the outcomes reported across the studies were not relevant to generalisation. The only two outcome measures used four or more times were MLU157,170,172,178,179 and percentage consonants correct,144,157,164,167 and when used pre and post intervention these are appropriate measures to evaluate ‘generalisation’.

Delivery of the interventions

In 18 of the 24 studies the intervention was delivered by SLTs;139,140,145–147,149–152,154,157–159,164,167,172,178,187 of the other studies two were delivered by teachers,170,179 one by a clinician,192 one by a graduate student SLT,144 one by a combination of SLT and parent174 and one by a researcher acting as a teacher. 138 The studies were undertaken in a range of settings: a clinic (n = 14140,144–147,149–152,157,158,167,173,187), home and preschool,174 a special language preschool,179 a preschool154,178 and a Head Start centre. 138,170 In four studies it was not stated where the intervention took place. 139,159,164,172

Number of treatment sessions

In this typology it was most common for an intervention to be delivered three times a week (n = 7147,151,152,157,159,173,178), followed by twice weekly (n = 4144,146,154,167). Interventions lasted between 25 and 60 minutes. There were generally between 12 and 46 sessions, which took place over 6–24 weeks (although in one study the intervention took place over 9.5 months for one child158).

Comparison with speech and language therapists’ practice

Table 52 shows the activities and strategies that were identified by SLTs who participated in the Child Talk focus groups in connection with the theme of ‘generalisation’. The table also provides the percentage of SLTs ranking each intervention as essential and how many of the studies included in the review explicitly mentioned using each intervention.

Evidence of outcomes

In relation to EBP, generalisation is most frequently a secondary focus within the published studies. The 12 studies in this theme whose primary focus was generalisation showed mixed results:

• The results of one study indicated generalisation of preposition–object phrases not only to untreated two-word combinations but also to comprehension of four-word phrases. In addition, there was also some generalisation of production of two-and four-word structures. 172

• In one study, children learned the three-element clusters but were unable to generalise them. 152

• Generalisation of trained responses to untrained stimuli was obtained for each of four target ‘linguistic behaviours’ (she, him, he, are). 187

• Script-based training was effective at increasing all participants’ use of semantic relation skills and generalising acquired semantic relation skills across non-trained scripts and pictures. All participants maintained increased semantic relation performance across the 3-week maintenance assessment period. 178

• The results of one study suggested that broad target recasts facilitate speech intelligibility of only the most developmentally immature speech producers. However, there was no support for this being transferred into speech outside the clinic or home setting. 157

• Phonologically based treatment resulted in a reduction of both phonological word-finding substitutions and semantic word-finding substitutions. Generalisation effects were not as powerful as the authors had hoped; however, no phonological errors occurred during post-treatment testing. 139

• Two subjects met the various performance criteria at each step of the treatment intervention but they did not generalise modified speech sounds production of treated phonemes to untreated words or to untreated phonemes. 158

• In one study an intervention was generally effective in increasing the social communication behaviours of children who had fewer social interactions during baseline conditions. Specifically, the intervention increased the use of verbal behaviours, social communication skills and target vocabulary words in unstructured play with peer partners. 138

• How often a list of targeted treatment words (probe lists) was administered did not have an effect on the extent of generalisation or the occurrence of a practice effect. 159

• Generalisation probes were structured to evaluate the use of specific grammatical forms and syllabic shapes over the course of up to 30 sessions. The results indicate that the intervention procedures facilitated change in children’s productions of w-S stress patterns in untrained multisyllabic words, although less so in untrained phrases. 140

As indicated above, the 24 studies included in this theme represented a wide range of primary foci of intervention. Accordingly, a wide range of interventions and assessment measures was used. These, and the outcome measures, relate to the focus of the intervention rather than to generalisation.

What was the focus of the studies?

Of the three studies focused on ‘foundations skills’, two focused on an adult supporting the children to organise play (including assigning roles, identifying and naming toys to be used), play in dyads and self-evaluation with the support of an adult. 170,185 Impact on the children’s social communication was measured by the amount and complexity of language used and the amount of time spent on each type of play;170,185 one study aimed to increase peer-directed social communication. 170 This was particularly related to initiations, responding appropriately to other’s initiations/questions or actions, appropriately obtaining a peer’s attention and maintaining a social interaction (turn taking). 138

Country of origin of the papers

Of the four papers included in this typology theme, three were from the USA138,170,185 and one was from the UK. 70

Design

One study used a between-group study design. 70 The most common research design was that of a multiple baseline case series using dyads. 138,170,185

Size of the sample

Fifty-nine children were represented by the four studies, with a minimum of 6 children and a maximum of 39 children across the studies (mean 14.8, median 7.0).

Child participant demographics

It should be noted that, although the studies in this category all recruited children with a language impairment, one study also recruited children who were at risk of impairment. 138 Of the 59 children, 37 were male. The ages of the children ranged between 37 and 60 months, with an average of 44.6 months.

Assessment measures

Eight different measures were used across the studies in this theme; however, only two were directly relevant to the theme:

• peer play code170,185

• peer language behaviour code138,170,185 – subdivided into:

  • measuring communication behaviours = descriptive (peer-directed comments, play organiser statements and acknowledgement responses) and request utterances (information requests, yes/no questions, action and stop action requests and clarification requests)185

  • interventionist behaviour measures. 170,185

Outcome measures

Twelve different outcome measures were used across the four studies in this theme. Eight measures were used only once. Two were used more than once and were directly relevant to the theme of ‘foundation skills’ (child communication measures, peer play code). A social validation scale (which was a teacher rating scale for impressions of verbal engagement during play) was used in one study. 170

Delivery of the interventions

Three of the studies described interventions undertaken by teachers (with research experience). 138,170,185 One study was undertaken by SLTs. 70 In three of the four studies the interventions were carried out in Head Start centres138,170,185 and in the final study the intervention was carried out in a preschool nursery. 70

Number of treatment sessions

As the number of studies in this theme was small, it is possible to detail the number of treatment sessions for each study. In the study by Barratt et al. ,70 children were randomised to a weekly or an intensive intervention. The weekly intervention was delivered once a week for 40 minutes for up to 24 sessions, whereas the intensive intervention was delivered on 4 days a week for 40 minutes for up to 24 sessions. In the study by Stanton-Chapman et al. ,138 intervention sessions were conducted on 5 days per week and lasted for 25 minutes (with a total of 20 sessions). The studies by Craig-Unkefer and Kaiser170,185 included 20-minute interventions three to four times per week, with the first dyad having 16 interventions, the second having 14 and the third having 12.

What was the intervention

Interventions (whole and individual activities and strategies) used in the study were mapped against the activities and strategies reported by SLT participants from the Child Talk focus groups and online surveys.

Comparison with speech and language therapists’ practice

Table 53 shows the activities and strategies identified by SLTs who participated in the Child Talk focus groups that targeted children’s foundation skills. The table also provides the percentage of SLTs ranking each intervention as essential and how many of the studies included in the review explicitly mentioned using each intervention.

Intervention activities or strategies mentioned within studies in the review but not by therapists in focus groups include:

• book reading138,170

• role play170,185

• reflective statements (redirects – models, direct instructions, indirect instructions)70,170,185

• modelling138,185

• labelling138,170,185

• talking to their friends. 138

Evidence of outcomes

Most of the studies appear to indicate that the treatments were successful but, again, overall, the studies report fairly low levels of specific intervention success. Peer-directed interventions were generally effective at increasing social communication behaviours of children who had fewer social interactions at baseline. Modelling of interaction to a mother increased verbal interaction with her child and suggested increased utterances in response from her child. Play–plan–report therapy increased social communication behaviours. The four studies represented in this theme are also included in other themes. As such, they are very diverse, with few common elements for any conclusions to be drawn. Two aspects of SLT-recommended interventions – maintaining attention and playing language games – were well represented within the four studies and showed positive change following the interventions. 138,170,185 Again, book reading is included in the research studies but was not mentioned by therapists in focus groups.

Country of origin of the papers

All five of the studies representing this area of the typology were undertaken in the USA.

Design

Four of the studies used a case series design, often with multiple baselines. 138,170,184,185 The other study used a between-group design. 137

Size of the sample

The five studies representing this area of the typology included a total of 82 children. The number of children included ranged from 5 to 57, with a mean of 16.4 and a median of 6.0.

Child participant demographics

Two studies recruited children who were ‘at risk’137,138 whereas the other studies all recruited children with language delay. One study did not specify the gender split of the children involved. 137 The other studies included a total of 11 boys and 14 girls. One study simply stated that the children were 5 years of age and gave no other details. 137 The other studies included children aged between 42.0 and 54.0 months (average 48.1 months).

Assessment measures

Seven different assessments were used across the five studies. However, only two – the Child Behavior Checklist/2–3 (CBCL/2–3)209 and the SSRS207 – could be identified to be relevant to functional communication.

Outcome measures

Across the five studies 19 outcome measures were identified. Fifteen of these were used once and the majority were not relevant to the theme of functional communication. However, child communication measures and the peer play code were used in two studies170,185 and can be argued to measure functional elements of communication.

Delivery of the intervention

There was no consistency across the studies in relation to who delivered the intervention. In one study the intervention was delivered by a combination of early years professionals and parents,137 whereas in another an assistant SLT delivered the intervention. 184 One study used a researcher to act as a teacher138 and the final two studies used the children’s preschool teachers. 170,185 Four of the interventions were undertaken in Head Start centres138,170,184,185 and one was undertaken in a child development centre. 137

Number of treatment sessions

In the longest study in this typology, children received the intervention for 5 days a week for 50 weeks a year from 6 months to kindergarten (although age was not specified), with the parents of these children receiving home visits every 1.5 weeks as well as monthly group meetings. 137 The remaining four studies carried out interventions lasting between 7 and 25 minutes, delivered three to five times per week, for a total of 12–23 sessions. 138,170,184,185

What was the intervention?

In addition to the interventions mapped in Table 54 to the typology theme, the included studies used other activities and strategies:

• labelling toys138

• direct and indirect instructions170,185

• role play170,185

• reading138,184

• classroom discussions138,184

• praise. 138

Comparison with speech and language therapists’ practice

Table 54 shows interventions that were identified as being used to have a positive impact on children’s functional communication skills. The table also provides the percentage of SLTs ranking each intervention as essential and how many of the studies included in the review explicitly mentioned using each intervention.

Evidence of outcomes

In the studies in this theme there seemed to be only limited concordance between the approaches used in the studies and those recommended by SLTs.

The findings suggest that the interventions were generally effective in increasing children’s social communication. Specifically, the interventions increased the use of verbal behaviours, social communication skills and target vocabulary words138 and results in improvements in preschoolers’ generation of personal narratives. 184 The findings of Craig-Unkefer et al. 170,185 suggest that the interventions increased the amount of child talk during play and the number of requests and descriptive talk. However, although the studies report successful outcomes, their methodologies are too diverse for a grading of the evidence.

What was the focus of the studies?

The sole study is a PhD thesis and investigates whether low educational-level migrant mothers can be effectively trained to implement dialogic reading techniques, and what effects the mothers’ implementation of dialogic reading techniques have on the oral language production of migrant preschool children. One element of this thesis looks at the level of understanding, in relation to dialogic reading, of the parents before and after undertaking training.

Country of origin of the paper

The study was based in the USA.

Design

The study used a case series with staggered starts and multiple baselines.

Size of the sample and child participant demographics

Only four children (two boys and two girls) are represented in this typology theme. They were recruited from a mixture of at-risk children and those with a diagnosed language delay. The children were aged between 4.0 and 5.7 years.

Assessment measures

None of the assessment measures used in the study were relevant to measuring the knowledge and understanding of adults/parents regarding preschool language delay.

Outcome measures

In relation to the theme of ‘adult understanding’, parents underwent training in dialogic reading and had to score 100% on an assessment tool designed specifically for the project. Another measure used in this study was social validity questionnaires. These targeted the importance, effectiveness and practicality of the intervention and thus examined parents’ perspectives of the intervention rather than specifically their understanding of the intervention; they are thus only of partial relevance to the typology theme.

Delivery of intervention

The study was undertaken in the homes of patients.

Number of treatment sessions

The number of sessions administered ranged from 5 to 11. The lengths of the sessions were comparatively long, with 3-hour and 2-hour sessions in some cases. These sessions were delivered over periods of between 3 and 6 months. The scale of the interventions varied from two instruction sessions followed by a mother being observed on six occasions, up to one home visit made every 1.5 weeks from age 3 months to elementary school (approximately 200 sessions over 5 years).

What was the intervention?

The intervention in this theme focused on increasing parent knowledge about preschool language delay. This was carried out through an educational programme with parents and book reading. 134

Comparison with speech and language therapists’ practice

Table 55 shows interventions that SLTs identified as being used to enhance adult understanding. The table also provides the percentage of SLTs ranking each intervention as essential and how many of the studies included in the review explicitly mentioned using each intervention.

Evidence of outcomes

The study undertook parent training sessions, with two assessments of parents’ understanding of dialogic reading. Parents who did not obtain a score of 88% at the end of the course were given an additional coaching session and subsequently gained 100% when reassessed. Although apparently successful at changing parents’ interactions and parenting styles, the small number of parents included (n = 4) and the lack of reporting of data showing the levels of pre-training understanding mean that there is insufficient information relating to increases in adult knowledge and understanding to make any judgements about the effectiveness of the intervention in improving child language development.

What was the focus of the studies?

The majority of the studies included in this theme had a primary focus on the treatment of expressive language delay, with the others looking at expressive and receptive language delay. None of the included studies specifically focused on the success of adult–child interactions.

Country of origin of the papers

One study was undertaken in Germany,177 four were undertaken in the USA,134,136,137,143 one with Spanish-speaking families,134 one was carried out in Canada182 and three were carried out in the UK. 175,176

Design

Three of the studies within this theme were RCTs. 175,177 Two studies used a pre-test/post-test design (allocation to groups: random, location),176,182 two were case series with multiple baselines134,143 and two used a between-group design. 136,137

Size of the sample

The nine studies in this theme included 1011 children, with a mean number of children per study of 112.3 (median 36.0, range 4–731).

Child participant demographics

Two studies recruited children who were at risk. 136,137 One study recruited both children at risk and those who had been diagnosed with a language delay. 134 The other studies recruited children with a diagnosed delay/impairment. Several studies did not report on the gender split of the participants but among those that did 450 boys and 383 girls were investigated. 134,137,176,177 The ages of the children ranged from 24 to 58 months.

Assessment measures

Thirty-one measures were used across the nine studies. These assessed various levels of the children’s abilities. Lunkenheimer et al. 136 undertook four observational measures of parenting in the home. They focused on parent involvement using the Home Observation for Measurement of the Environment (HOME) Inventory,210 the two subscales of the Relationship Process Code (RPC) (positive reinforcement and engagement interaction)211 and proactive parenting using the Coder Impressions Inventory. 212

Outcome measures

These studies used multiple outcome measures; however, very few measures were related to adult–child interaction. Six studies used language samples of children and mothers,134,136,143,175,176,182 often using Systematic Analysis of Language Transcript (SALT) to support coding,143,182 and thus looked only at the child’s output and not at the adult–child interaction. Another measure asked mothers to complete a satisfaction questionnaire that assessed whether they/their child enjoyed participating in the intervention and whether they thought that it had helped their child. 143

Delivery of the interventions

Parents were instructed to enhance their child’s speech and language learning in a number of ways. Four studies used the Hanen programme or a modified version of the Hanen programme. 175,176,182 Buschmann et al. 177 taught parents how to promote interaction and undertake language modelling using the Heidelberg Parent-based Language Intervention. 213 Roberts et al. 137 undertook parent interaction interventions from when the child was 3 months old until the child started elementary school. These interactions ‘educated’ the family in learning games, child management and parent problem-solving. The other three studies undertook short-term assessments and education, with one study providing feedback on parent–child interactions from the pre-intervention assessment. 136 The other two studies taught strategies such as modelling, pausing and praise. 134,143 Only one study assessed the understanding of the techniques taught to the parents. 134

The interventions for the children were delivered in a number of ways. In four of the studies the intervention was delivered by SLTs working with parents. 175,176,177 In a further three it was reported that parents delivered the intervention134,175,182 and in a further one it was reported that the intervention was facilitated by parents and EYPs. 137 In the final study it was reported that the intervention was delivered by a ‘family/consultant’ dyad, but this is not clear. 136 The interventions were delivered in children’s homes and nursery settings.

Number of treatment sessions

The interventions varied in length from approximately 2–3 months for parent training programmes to a whole academic year for preschool programmes. Sessions lasted from 10–15 minutes for a home reading programme up to 2.25 hours for group sessions.

Comparison with speech and language therapists’ practice

Table 56 shows interventions that SLTs identified as being used to have a positive impact on adult–child interactions. The table also provides the percentage of SLTs ranking each intervention as essential and how many of the studies included in the review explicitly mentioned using each intervention.

In addition to the interventions mapped in Table 56 to the typology theme, the included studies used other activities and strategies:

• demonstrate/inform about techniques175

• teaching strategies175

• group discussion137

• books134,137,143,177

• motivate child to interact182

• motivational interviewing. 136

Evidence of outcomes

The review found some evidence that interventions can be effective in their intended goals of modifying the behaviour of parents in their child’s environment,134,136,176,182 although one study found that parental language support actually declined over the course of the intervention. 143 The case series, which examined book reading, focused on the process of change throughout the interventions. 134,143 The evidence for the effectiveness of these interventions was varied and came from studies with small sample sizes.

The heterogeneity of the included studies, particularly in terms of intervention type and study design, prevents clear conclusions from being drawn about the effectiveness of adult-based interventions. However, the review findings suggest that, overall, involvement in these early preschool interventions can lead to language gains for children with, or at risk of, PSLI.

There was some evidence that environmental approaches to speech and language therapy can have comparable results to those of individually delivered clinic sessions, but that particular approaches may be best suited to children according to their profiles of language delay. 175,176 Around 75% of children who had taken part in interventions were shown to make significant continued language improvements in two studies that carried out 12-month follow-up assessments;176,177 in one study these children were shown to ‘catch up’ to age-appropriate levels of language. 177 However, it is important to draw attention to the fact that a quarter of these children therefore had persisting difficulties with language despite intervention. Furthermore, not all children were found to make equivalent gains as a result of intervention. 175

Body language

Body language included any position or movement of the body that communicated something about the child’s participation, including the position and orientation of their body, movements associated with the expression of emotions, general body activity and movement and also fidgeting. Different aspects of the child’s body language gave insight into the child’s perspective.

Table 58 displays the different aspects of body language that were explored.