A survey of Swedish speech-language pathologists’ practices regarding assessment of speech sound disorders

Abstract

Purpose

To explore Swedish clinical practice regarding assessment of suspected Speech Sound Disorders (SSD) in children.

Methods

A web-based questionnaire, regarding assessment of SSD in children 4;6–6;11 (years; months), was distributed to Swedish speech-language pathologists (SLPs) through social media and online forums. The questions concerned the frequency and manner of assessment for seven assessment components, chosen based on a review of international recommendations for SSD assessment.

Results

A total of 131 SLPs responded to the questionnaire. The results show that Swedish SSD assessment practices vary with regards to the frequency and manner of assessment for many components. Speech output is frequently assessed while speech perception, phonological awareness and oral-motor function are assessed less frequently. A variety of manners of assessment, for example, standardised tests, non-standardised material, and informal assessment procedures, such as observation, are utilized by respondents.

Conclusions

Swedish SSD assessment practices are variable. The present paper reveals areas for development within SLP practice and education programmes, and provides a new perspective on present praxis with regards to the assessment of suspected SSD in Sweden.

Keywords:

• Speech sound disorders
• assessment
• speech-language pathology
• evidence-based practice

Introduction

Speech sound disorders (SSD) are developmental impairments that manifest overtly as developmentally inadequate speech output, which often negatively impacts speech intelligibility. In addition to affecting speech output, SSD may also involve difficulties with speech perception and/or imprecise mental representations of speech sounds. Children with SSD are a heterogeneous group with respect to the aetiology and severity of their disorder, their speech characteristics, involvement of other linguistic domains as well as response to intervention [1]. Although classification of SSD subtypes has been subject of much debate, there is presently agreement concerning broad subgroups, such as articulation-based SSD, phonologically based SSD and SSD due to motor planning/programming deficits, for example Childhood Apraxia of Speech (CAS) [2]. However, most children with SSD of unknown origin fall within the phonologically based SSD group, which is very diverse [2]. In this respect, Sweden lags behind; Swedish terminology to describe children with SSD diverges from international classification systems. An SSD of unknown origin, without motor characteristics, is referred to as an articulation disorder, or as “fonologisk språkstörning,” which translates to “phonological language disorder” [3]. Although this term literally indicates a condition that is a specific type of language disorder, a speech-language profile of this type roughly corresponds to a “phonological disorder” (or, a “phonologically based SSD”) and would, in itself, not warrant a diagnosis of developmental language disorder (DLD [4]). While university hospitals and universities are beginning to use the terms “talljudsstörning/SSD” or “talstörning/SSD,” we expect that “phonological language disorder” is still widely used among clinicians.

Regardless of terminological differences, children with SSD comprise a significant proportion of the caseload of speech-language pathologists (SLPs) [5,6] and a comprehensive assessment, including a description of the child’s speech profile, is essential for a correct diagnosis as well as for treatment planning and monitoring of progress.

Recommendations for assessment of SSD

There are a number of recommendations and guidelines available concerning the assessment of suspected SSD in children. Unfortunately, there are, as of yet, no such guidelines concerning the clinical care of children with SSD in Sweden.

As children with SSD, despite comprising a heterogeneous group, all produce speech that does not meet expected targets, assessment of speech output is essential. A phonetic transcription of a representative speech sample, assessment of phonemic inventory, sound stimulability and prosody as well as appraisal of consistency, type and distribution of speech error patterns, are recommended [7–9]. Consistency of speech errors is recognised as particularly important in identifying Childhood Apraxia of Speech (CAS) [10,11] and inconsistent phonological disorder [1].

In addition to speech output, appraisal of oral structure and oral-motor function is significant in SSD assessment in order to rule out any structural or functional deficits, such as a sub-mucous cleft palate or limited muscle strength or mobility [7–9]. As impaired oral-motor function is often found in children with persistent SSD [12], it may also be valuable for prognostic purposes.

Furthermore, many children with SSD have impaired speech perception; according to a recent review, 60 of 73 included studies on speech perception skills in children with SSD reported difficulties with speech perception in some or all participating children [13]. Whilst the nature of the relationship between speech perception and speech production is not yet fully understood [13], speech perception is recommended to be included in SSD assessment [8,13,14] and many researchers emphasise the importance of assessing a child’s perception of sounds that are subject to speech errors in his/her speech [13,15].

Moreover, many children with SSD – particularly those with more severe or persistent disorders – have co-occurring difficulties with phonological awareness [16,17], that is, the awareness of and ability to manipulate, the sounds and sound system of one’s language. As phonological awareness (specifically phonemic awareness) is strongly associated with literacy acquisition [18], the assessment of phonological awareness skills in children with suspected SSD is recommended [8,14,19].

In order to identify potential risk factors, but also to encompass the functional consequences of the disorder, a detailed case history, as well as assessment of intelligibility and functional communication are recommended for inclusion in SSD assessment [7–9]. “Intelligibility” refers to how much of the child’s speech listeners understand [20], and is central in identifying the disorder, in determining the need for intervention, as well as for evaluating therapeutic outcomes [20,21]. “Functional communication” designates the child’s communication in a broader sense, and provides information regarding the child’s ability to communicate in daily life, thus, involving consideration of the participation component of the International Classification of Functioning, Disability and Health (ICF) framework [22].

Hearing is also important to take into account in assessment of suspected SSD [8,9] as it may affect perception of speech, convey aetiological information, and have implications for intervention. Moreover, screening for language difficulties is central as many children with SSD also have comorbid DLD [23]. Children whose speech problems persist after 5 years of age are at increased risk of more pervasive language difficulties [24], which should be taken into consideration in assessment.

International surveys of clinical practice regarding children with SSD

Surveys of SLP practice can provide insights into clinical reasoning and the extent to which recommendations from published research are implemented in the clinic. McLeod and Baker [6] surveyed 231 Australian clinicians regarding SSD assessment and found that the most frequently included (51–90%) assessment components were parental interview containing the child’s case history, a single word test, assessment of intelligibility, stimulability and the child’s phonological processes and phonetic inventory. However, phonemic awareness, oral-motor skills and speech perception were less frequently (16–26%) assessed [6]. This pattern is mirrored in other surveys of SLP practice; Skahan et al. [25] found that speech output is frequently assessed by clinicians in the United States, while speech perception, phonological awareness and oral-motor function are less often included in SSD assessment. Unfortunately, not all surveys report [5,26] on practice concerning speech processing abilities.

In previous surveys, assessment procedures are often measured by inviting clinicians to describe which assessments they use or by choosing alternatives from a list of standardised tests [5,6,26], sometimes including follow-up questions concerning assessment of speech output [6,25]. The manner in which other components, such as intelligibility or speech perception, are assessed, has not been reported. As there is evidence suggesting that speech perception skills, phonological awareness and oral-motor function are important to consider in SSD assessment, we wanted to explore the extent to which they are included, and how they are assessed, in Swedish SLP practice.

A note concerning standardised tests

Assessment of SSD is often conducted by means of standardised tests. Standardised tests are effective, administered in a consistent manner and designed to enable clinicians to compare a child’s speech to that of his/her peers (norm-referenced) or criteria from published research (criterion-referenced). However, as argued by Fabiano-Smith [27], most (norm-referenced) standardised tests provide neither the sensitivity nor specificity needed to function as the sole foundation of SSD assessment and should, therefore, be supplemented with other manners of assessment.

Previous surveys of clinical practice show that SLPs use standardised tests in a non-standardised way, for example by omitting sections or items that are inappropriate for their clients [26,28]. An unstandardised use entails that one cannot use the tests scoring system or norms [27] which may impede assessment and follow-up of progress. On the other hand, flexible use of standardised test materials may enable clinicians to adapt their assessment to each individual child and his/her circumstances.

Given that observation and informal procedures and materials are also used in SSD assessments [6,25,26,28], relying on standardised tests in surveys of clinical assessment practices may result in an incomplete understanding of the variability of practice. Surveys of SSD assessment that offer clinicians the opportunity to describe their practices are, therefore, needed in order to understand how a child’s (cap)abilities are assessed.

Swedish speech-language pathology practice

Little is known about the assessment practices in Sweden regarding children with suspected SSD and there are currently no Swedish national guidelines or recommendations concerning SSD assessment or intervention. In a survey of 178 Swedish clinicians concerning knowledge, estimated occurrence and typical speech characteristics of CAS, Malmenholt et al. [29] found that a majority of respondents felt unsure about diagnosing CAS and a total of 89% stated that they lacked competence about the disorder. Many respondents expressed a need for increased knowledge regarding current research, intervention and differential diagnosis. These findings suggest that differential diagnosis of SSD subtypes is a challenge for Swedish clinicians. However, the assessment practices for SSD, in general, remain elusive.

Aim

Based on assessment components recommended for inclusion in SSD assessment by research literature and international clinical guidelines, this study aimed to explore the frequency and manner in which the components are assessed in Swedish clinical practice.

Method

A web-based questionnaire was developed and distributed to clinically active SLPs who provide services to children with SSD. The questionnaire focused on practice concerning seven assessment components that are considered important [7–16] for SSD assessment; speech output,¹ consistency of speech errors (henceforth referred to as consistency), oral-motor function, speech perception, phonological awareness, speech intelligibility and functional communication. While we acknowledge the importance of hearing and non-verbal IQ in SSD assessment, the components included in the questionnaire were selected based on their relevance for Swedish SLP practice (hence excluding questions concerning assessment of hearing and non-verbal IQ). Moreover, screening for language difficulties is important, but the focus of the current survey was assessment of abilities related to speech processing and production.

The questionnaire

A pilot version of the questionnaire was sent to three SLPs who were asked to comment on readability, structure and content. The SLPs were asked to review whether all concepts were easy to understand, whether the layout and text were clear and whether the length of the questionnaire was reasonable. The SLPs did not suggest any changes, and consequently, no alterations were made.

The questionnaire consisted of 16 main items, of which 11 included multiple choice and/or free-text follow up questions. No questions were compulsory.

In the introductory information, the study’s aims were described and contact details to the first author of this paper were provided. It was explained that participation was voluntary and that no personal identifiable information would be collected. The introductory text also contained a clarification concerning the term “SSD” and its chosen Swedish translation “talstörning”: “Speech Sound Disorders encompass difficulties that manifest in speech, regardless of cause (i.e. phonological, articulatory, structural and/or motor difficulties)”.

The following section of the questionnaire included four demographic questions regarding work setting, clinical experience, geographic region and frequency of clinical contact with children with SSD (“How often do you see children with SSD?”).

The next section contained seven sets of multiple-choice questions concerning the frequency and manner of assessment for each of the above-mentioned assessment components. SLPs were instructed to answer the questions on the basis of the assessment they perform in order to make a diagnosis and plan intervention, that the child is between 4;6 and 6;11 (years; months), and that they suspect a SSD based on the referral, or through observation.

Respondents were asked to indicate how often they assess each component (“Do you assess component X in the children you see?”) by choosing one pre-determined answer (always, often, sometimes, seldom, never). For each component, this question was followed by a query concerning manner of assessment, in which participants could choose several of the fixed answer options (standardised tests, home-made materials, observation of the child, anamnesis, other). Respondents were subsequently asked to describe their manner/s of assessment in writing (free-text response).

Consistency of speech was provided with a definition; “Do you assess whether a child’s speech errors are consistent or inconsistent (i.e. whether the target sound always changes in the same way or not)?,” but no other components were defined in the survey.

The participants were also invited to describe their transcription practices, through choosing one answer concerning frequency of transcription of speech (always, often, sometimes, seldom, never) and one for type of transcription (orthographic transcription, broad/phonematic transcription, fine/phonetic transcription, a mix of the above).

Finally, respondents were asked to describe any frequently occurring aspect of their SSD assessment that the questionnaire had overlooked in writing.

The questionnaire also included three questions concerning diagnose-code use, the components’ importance and the frequency of speech error patterns, that will not be presented here.

The questionnaire was administered through Google Forms and was distributed to SLPs through social media, an online SLP forum and to clinical supervisors via e-mail. The questionnaire accepted responses during the summer of 2019. Information concerning response rate is not available, as there was no direct contact with any of the respondents, nor is there information concerning how many clinicians interacted with the social media posts.

Ethical considerations

Participants were informed that participation was voluntary and consented to participation through submitting a response to the questionnaire. As the study collected no personal or identifiable information, an ethics review was not necessary.

Analyses

Descriptive statistics (response counts and percentages) were compiled for all multiple-choice and Likert-scale questions, using R [30]. Responses to open-ended questions were labelled and categorised with respect to their contents, that is, a surface level analysis (i.e. coding and counting of explicit meanings of written responses, without attempting to interpret implicit ideas). First, all free-text answers were manually reviewed by each author independently, to identify meaningful descriptive labels. The individual inspection was followed by a joint discussion during which the authors agreed upon a set of appropriate descriptive labels for each question. Labels were determined on the basis of being mutually exclusive, relevant to the question and of not being repetitive of any of the fixed response options (e.g. “observation” for manner of assessment). For example, the label “qualitative approach” was used for the following statements describing manner of assessment of functional communication; “In play and through observation,” “This is assessed qualitatively all the time, during every session. However, I have never performed a quantitative assessment of this.,” “In interaction with the child on his/her level.” After the discussion, the authors individually counted and classified all responses with regards to the determined labels. The answers were then tallied, and token-to-token interrater agreement was calculated. All three authors were in complete agreement concerning the number of instances for 72 of 110 labels, and were near agreement (1–3 tokens difference) for 32 additional labels. In cases of discord, answers were discussed by all authors, until consensus was reached.

Results

Participants

A total of 131 SLPs from a variety of work settings (most commonly from hospitals, n = 46/131) and geographical regions (19/21 regions in Sweden) responded to the questionnaire. Many clinicians had 5 years of experience (n = 48), although there was sizeable variation concerning work experience among the respondents. Most clinicians reported seeing children with SSD every day (66/131) or every week (53/131). Demographic information for the participating SLPs can be seen in Table 1.

Table 1. Description of the participants’ work setting, clinical experience and frequency of clinical contact with children with SSD.

Work setting
Hospital	46
University hospital	18
Outpatient clinic	21
Private practice	13
School/pre-school	21
Habilitation clinic	10
Other^a	2
Total	131
Years active
<1 year	12
1–5 years	48
6–10 years	31
11–15 years	17
16–20 years	10
>20 years	13
Total	131
Clinical contact
Every day	66
Every week	53
Every month	12
More seldom	0
Total	131

^aChild and adolescent mental health services (n = 1), school health services (n = 1).

Frequency of assessment

The responses show that the most frequently assessed components were speech output (66% always, 27% often) followed by consistency (60% always, 33% often), functional communication (44% always, 31% often) and intelligibility (35% always, 34% often). Less frequently assessed abilities were phonological awareness (6% always, 32% often), oral-motor function (11% always, 21% often) and speech perception (11% always, 38% often). See Figure 1 for a summary of responses.

Figure 1. Reported frequency of assessment for each component. The components are presented in decending order, from most to least frequently always assessed.

Frequency of assessment across health care regions and work settings

Respondents were asked to specify within which county they worked. Nineteen of twenty-one Swedish counties were represented. Three respondents did not provide an answer regarding county. To facilitate an evaluation of whether there were differences in frequency of assessment due to geographic location, the counties were categorised into Sweden’s six health care regions (Swe: “sjukvårdsregioner”); North, Mid Sweden, Stockholm-Gotland, South-East, West and South health care regions (see Figure 2).²

Figure 2. Reported frequency of assessment for each component, across Sweden's six health-care regions.

Visual inspection of responses reveal that speech perception is the most commonly component to “never” be assessed in the Mid Sweden, Stockholm-Gotland and South-East health care regions. Oral-motor function was reportedly “never” assessed to the highest degree in the South health care region and was rarely “always” or “often” included in assessment by clinicians in the Mid-Sweden, South and South-East health care regions. Frequency of assessment concerning speech output and consistency were similar across the six regions. See Figure 2 for a summary of responses.

Respondents also specified the work setting in which they were active. Frequency of assessment for each component across all work settings is illustrated in Figure 3. Visual inspection of the responses in Figure 3 reveals that oral-motor function is seldom or never assessed by many clinicians in preschool or school settings and seldom routinely included (always or often) by SLPs in hospitals. Clinicians in habilitation clinics assessed functional communication to a high degree while some SLPs in hospitals, outpatient clinics and school settings responded that they seldom or never assess functional communication. Frequency of assessment of consistency and speech output were reportedly comparable across work settings.

Figure 3. Reported frequency of assessment for each component, across work settings.

Manner of assessment

For all seven assessment components, respondents indicated the manner in which they assessed each component, with fixed response options. Respondents were not limited in their number of responses. The answers reveal that standardised tests are often used in assessment of speech output, consistency, phonological awareness and oral-motor function (n = 71–121), but less often for assessment of functional communication and intelligibility (n = 15–19). Observation was the most frequently cited manner of assessment for oral-motor function (n = 83), intelligibility (n = 96) and functional communication (n = 103), often in combination with anamnesis or home-made materials. Speech perception was reportedly most commonly assessed through standardised tests (n = 47) and home-made materials (n = 37).

A majority of respondents reported using several manners of assessment for numerous assessment components, with an overall average of around 2 manners per component. However, there were differences across the seven components; clinicians cited a larger repertoire of assessment manners for the more frequently assessed components. See Table 2 for an overview of responses.

Table 2. Reported manner of assessment for the seven assessment components.

	Standardised tests	Home-made materials	Observation	Anamnesis	Other	Responses	Respondents
Speech output	121	30	91	66	14	322	131
Consistency	110	25	82	24	14	255	130
Oral-motor function	71	60	83	51	13	278	124
Speech perception	43	37	18	19	8	125	105
Phonological awareness	72	49	15	12	16	164	117
Intelligibility	19	11	96	61	9	196	118
Functional communication	15	10	103	82	3	213	121

The most frequently cited manner of assessment is indicated in bold. The components are listed in the order that they were presented in the questionnaire. Response options were fixed.

In the subsequent questions, the respondents were asked to describe their manner of assessment for each component in free text. Answers varied in character, depending on responses to the previous question. Responses that occurred less than four times, or that were repetitions of previous responses (e.g. “observation”) were not included. See Table 3 for an overview of tests cited in free text-responses.

Table 3. Tests cited for assessment for the seven assessment components.

	Test	Responses (n)	Total responses (n)
Speech output	Hellqvist fonemtest [31]	67	83
	LINUS [32]	29
	DYMTA [33]	16
	SVANTE [34]	10
	Nya Lundamaterialet [35]	4
Consistency	DYMTA	23	59
	Hellqvist fonemtest	22
	LINUS	10
Oral-motor function	NOT-S [36]	26	66
	STORM [37]	22
	ORIS [38]	9
Speech perception	Nelli	12	47
	B.A.S	4
Phonological awareness	B.A.S [39]	15	55
	Bedömning av språklig medvetenhet [40]	9
	Nelli [41]	6
	UMESOL [42]	4
	MiniDUVAN [43]	4
Intelligibility	ICS [44]	8	28
Functional communication	Pragmatic profile in CELF [45]	4	27
	FOCUS [46]	4

Tests, or parts of tests, mentioned at least four times in free text answers are included in the table. The components are listed in the order that they were presented in the questionnaire.

The following section details results from free-text answers. Concerning speech output, most respondents reported using picture-naming tests (see Table 3) and some (n = 37) described using spontaneous or connected speech in assessment. Few (n = 6) mention conducting audio or video recordings of speech. A number of clinicians described using the same procedure for assessment of consistency as for speech output (n = 18). Oral-motor assessment was conducted by means of tests (see Table 3), sections of tests (n = 8), pictures (n = 14) and unspecified exercises (n = 8). For assessment of phonological awareness, rhymes (n = 11), minimal word pairs (n = 10), non-word repetition (n = 5), auditory discrimination (n = 7) and a number of different phonological awareness exercises, such as phoneme segmentation or identification, (n = 9) were cited, in addition to the tests presented in Table 3 (some citing only certain sections/items of these tests, n = 8). A number of respondents (n = 11) answered that assessment of phonological awareness depended on the child’s age (often not assessed prior to 5 or 6 years of age). Speech perception was often assessed through minimal pairs (n = 25), auditory discrimination (n = 16) and parts of standardised tests (n = 10). A few clinicians cited phonological awareness exercises (n = 4) and some reported assessing speech perception during treatment (n = 6). A rather large number of respondents (n = 12 of 47 responses) expressed uncertainty regarding speech perception; some answered that they did not understand the term, while others described using receptive language tests for assessment. For intelligibility and functional communication, many respondents describe a “qualitative” approach to assessment (n = 22, 16 respectively), for example through observation and interaction with the child during the assessment, and/or questions posed to caregivers and teachers.

When queried concerning their transcription practices, the vast majority responded that they frequently transcribe children’s speech (n = 78 “always,” n = 42 “often”), and few clinicians transcribe speech sometimes, occasionally or never (n = 8, 2, 1, respectively). Phonetic transcription was employed by 33 clinicians, broad transcription by 34 and orthographic transcription by 11. A large group (n = 53) utilised a combination of the above-mentioned levels of transcription when documenting children’s speech output.

The following question, concerning whether the questionnaire had overlooked any aspect of assessment that is always or often performed, was answered by 15 SLPs. Comments included assessment of language skills (n = 5), speech motor skills (as opposed to non-speech oral movements, n = 2), consideration of the child’s age (n = 2) and responsiveness to intervention (n = 2).

Discussion

The present study aimed to explore assessment practices of Swedish SLPs regarding children with suspected SSD. Based on assessment components that are cited in international research and guidelines, the study explored frequency and manner of assessment for each component. A web-based questionnaire was distributed through social media and online SLP forums, and was answered by 131 clinicians.

Frequency of assessment

The results show that some assessment components, such as speech output and consistency of speech errors, are assessed more frequently than others, for example speech perception, oral-motor function and phonological awareness. This pattern is similar to findings from previous international surveys of clinical practice regarding that speech processing abilities are assessed less often than speech production abilities [6,25]. However, the results suggest that Swedish SLPs evaluate intelligibility to a lesser degree (35% always) than clinicians in the United States (75% always) [25] and Australia (55% always) [6].

It is arguably reasonable that all components are not always included in SSD assessment, as the assessment is contingent on a child’s communicative profile and his/her difficulties. A detailed evaluation of speech output is, naturally, always necessary for a SSD diagnosis. However, it is conceivable that factors, such as time constraints in the clinic and the fact that some children may have difficulties participating in long assessment sessions, may lead to the occasional exclusion of relevant components. Nonetheless, it is disconcerting that some clinicians respond that they never assess certain components. Speech perception, phonological awareness and oral-motor function, which were “never” assessed by a number of SLPs, are highly relevant, and research shows that some children with SSD have difficulties in these areas [e.g. 12–17,19]. It has been argued that a screening of oral structure and oral-motor function should be obligatory for all children suspected of presenting with an SSD [9]. In light of these recommendations, our results indicate that there is room for improvement in Swedish SSD practice concerning the frequency of assessment of oral-motor function. An increase in the frequency of assessment of speech perception and phonological awareness also appears warranted based on responses in this survey.

Moreover, a number of clinicians seldom or never appraise functional communication and intelligibility, which is troubling, as many argue that the main goal of SLP intervention should be to improve a child’s communication in everyday life [47]. As the functional consequences of an SSD cannot be directly inferred from the severity of the disorder [48], functional communication and intelligibility should habitually be considered in the assessment of SSD.

It should be noted that for some components (most notably speech perception and consistency), confusion concerning terminology was exposed in responses, and the results concerning the frequency of assessment for these components should, therefore, be interpreted with caution. See the following sections for further discussion concerning terminology.

Frequency of assessment across health care regions and work settings

In Sweden, as in other countries, the health care service must provide equal care to all, regardless of, for example, age or geographical whereabouts (The Health and Medical Services act (SFS 2017:30) [49]). The results from this survey indicate that there is some variation concerning how often different components are assessed, across Sweden’s six health care regions and across the work settings represented. Some variation across different work settings can be expected, as case-load characteristics may vary depending on the clinical context. For example, SLPs in habilitation clinics reportedly assessed functional communication to a higher degree than clinicians in other work settings, while a proportionately small number of habilitation SLPs assessed speech perception and phonological awareness. As children who receive habilitative services have comorbid neurological and/or motor difficulties, it is conceivable that clinicians in these settings tend to focus less on assessment and remediation at the body structure/function level of the ICF [22] (e.g. articulatory proficiency), and instead prioritize participation (e.g. functional communication in everyday life).

However, these results should be cautiously interpreted, as there are large differences in the number of respondents from each health care region (n = 7–32) and work setting (n = 10–46). Nevertheless, the findings do suggest a need for cross-regional discussions regarding SSD assessment practices.

Manner of assessment

Respondents reported on manner of assessment through fixed alternatives (e.g. “standardised tests,” see Table 2) and were invited to describe the chosen manner in free-text (e.g. the name of the standardised test, see Table 3). The results reveal variation concerning the manner in which different components are assessed, and many clinicians reported using several manners of assessment for many components. More manners of assessment were cited for frequently assessed components (e.g. speech output) than for less frequently assessed components (e.g. speech perception), indicating a larger catalogue of means of assessment for frequently assessed abilities.

Speech output was often assessed with picture-naming tests, similar to reports from previous surveys of SLP practice in Australia [6], the United Kingdom [5] and the United States [25]. Close to half of the respondents in the present study mention using spontaneous or connected speech in assessment. Comparably, previous surveys show that although many clinicians in the United States [25] and Australia [6] elicit connected speech in SSD assessment, only around one third of SLPs always do so. While single word tests can aid a clinician in determining a child’s phonetic inventory and in appraisal of speech error patterns, they are insufficient for assessment of prosody (e.g. appropriate use of phrasal stress and pauses [7]). Moreover, children’s production of isolated single words does not necessarily reflect their speech proficiency in spontaneous speech – connected speech may bring to light difficulties that are not evident in single-word productions (e.g. between-word simplifications [50]). To ensure ecological validity and representativeness of the speech sample, single word tests should, therefore, be supplemented with a connected speech sample of at least 75 words [51].

The results further show that a majority of the respondents frequently transcribe speech, although different levels of transcription, and combinations thereof, were cited (i.e. orthographic, phonematic and/or phonetic transcription). Depending on the purpose (e.g. for assessing speech production proficiency or intelligibility), a flexible use of transcription levels may well be warranted. However, a phonetic transcription of a representative speech sample is explicitly recognised as an integral part of speech assessment [7,9]. Unfortunately, the present study does not provide insight into when and how the different levels of transcription are used. However, very few respondents describe recording speech for analysis. As audio recording is necessary to reliably transcribe a speech sample, these findings indicate a need for more detailed scrutinization of current transcription practices.

Many of the abilities that were most frequently assessed were also most often assessed using standardised materials (e.g. speech output and consistency). However, intelligibility and functional communication were reportedly often assessed in a “qualitative” manner, primarily through observation. Although this may be sufficient to gain a first impression, as a screening for areas of concern, this finding is rather surprising, considering that standardised assessment materials indeed exist. To gauge the extent to which the child makes themselves understood in daily life, the ICS provides a valid screening [44]. On the other hand, for assessment of intelligibility in a narrower sense, that is, what proportion of the child’s speech is intelligible, the ICS is insufficient [48]. It is possible that many respondents have had this narrower aspect of intelligibility in mind, and that their responses reflect the fact that there is currently no agreed upon method for conducting assessments of intelligibility in children’s speech [48]. As the survey provided no specification of “intelligibility,” further analysis is not possible within the present study.

With regards to functional communication, formal materials are available for assessment of Swedish children’s communication in daily life (e.g. FOCUS [46]). However, the clinicians’ responses indicate that these are not routinely included in assessment of suspected SSD. As omitting standardised assessment procedures may restrict evaluation of intervention progress with respect to impact on functional communication and intelligibility, these findings ought to summon discussions concerning current practice.

Similarly, many respondents reported using observation and homemade materials for assessment for oral-motor function, despite the fact that standardised screenings of oral-motor function are readily available [NOT-S; 36]. Exercises and pictures of unspecified type were mentioned in description of manner of assessment of oral-motor function by many clinicians. Unfortunately, these responses do not allow for an analysis of how assessment is conducted or what aspects of oral-motor function are considered.

Regarding consistency of speech errors, there are two operationalizations of the concept; phonemic inconsistency (i.e. inconsistent production of the same target phoneme across different lexical contexts) and token-to-token variability (i.e. inconsistency across multiple productions of the same word or syllable), each type requiring a different approach to assessment. Consistency of speech errors is acknowledged as important for identifying CAS [10,11], inconsistent phonological disorder [1], and for differentiating between progressive and non-progressive variability [7]. Although there is diverging evidence surrounding which type of inconsistency is most valuable in differentiating between CAS and speech delay [11], identifying cases of inconsistent phonological disorder requires assessment of token-to-token variability [1]. The way in which the respondents define consistency was not directly probed in the present study. This, in combination with the fact that the definition provided in the questionnaire allows for both interpretations, could affect the validity of the responses concerning the frequency of assessment for this component. However, many free-text responses reveal that clinicians use single-word tests for estimations of consistency (often citing the same procedure as for assessment of speech output in general), suggesting appraisal of phonemic consistency. While some clinicians report using tests in which token-to-token variability is assessed (i.e. DYMTA [33]), these findings suggest that the assessment procedure for consistency described by the respondents, in general, may not be sufficient to identify children with inconsistent phonological disorder [1].

Free-text responses indicate that assessment of speech perception is often conducted in ways that may be too crude to identify more subtle difficulties, that is, through perceptual judgments of clear prototypical productions of words or sounds. Exercises mentioned include correct/incorrect judgements and minimal word pair identification, based on productions spoken by the SLP. Previous research shows that few children fail on such tasks, although they may still have difficulties with speech perception [15]. Furthermore, responses show that a number of SLPs express confusion regarding speech perception (Swe: “talperception”). Some clinicians explicitly acknowledged uncertainty regarding the concept, while others interpreted it to mean receptive language or phonological awareness. The variable terminology and methodological differences in previous research concerning speech perception [13] may conceivably influence Swedish clinicians’ understanding of the concept. The fact that reported practice does not seem to meet recommendations from published research is understandable, in light of the known lack of reliable and appropriate assessment materials for speech perception [13] and the cited confusion concerning terminology. Nevertheless, these findings reveal a knowledge gap that should be addressed by clinicians and SLP educational programmes alike, to make sure that speech processing abilities are given due notice in SSD assessment.

Assessment of phonological awareness was reportedly conducted in a number of different ways. For some, the reported manners of assessment are in line with recommendations (e.g. through the use of standardised tests including segmentation, blending or elision of phonemes or syllables). However, in many free-text responses, the descriptions of the reported assessment procedures are more obscure (e.g. “minimal pairs” and “auditory discrimination”). These responses do not allow for an analysis of what aspects of phonological awareness are assessed, that is what type of manipulation (e.g. elision or blending) or linguistic level (e.g. phoneme, onset-rhyme or syllable) is used. Thus, a review of phonological awareness assessment remains a task for future investigation.

For some respondents, assessment of phonological awareness was contingent on the child’s age; a number of clinicians did not assess phonological awareness in children prior to their enrolment in compulsory school (5–6 years of age). Indeed, it has been argued that phonological awareness is difficult to reliably assess before the age of four due to task processing demands [52]. In a recent paper, Erskine and colleagues [52] found that younger children’s performance on nonword repetition predicts their later performance on tasks that probe phonological awareness. They, therefore, suggest that nonword repetition may be a suitable task for children who are too young for phonological awareness tasks, such as phoneme blending or elision [52]. As the importance of proper phonological awareness assessment is widely recognised [19], we would welcome a discussion regarding how and at what age to screen children with SSD for difficulties with phonological awareness in Sweden. Such discussions should particularly acknowledge children with atypical speech errors or omissions, for whom the risk of difficulties with phonological awareness is higher [53,54].

A note concerning standardised assessment materials

The results show that standardised tests are frequently used for assessment of a number of components, although most frequently speech output. Standardised tests should not be used as the sole foundation for assessment and diagnosis of a SSD [27] and, in this respect, the findings from the current study are reassuring, as they hint that clinicians do not rely exclusively on tests, but rather utilise several manners of assessment. Additionally, Swedish clinicians reported a flexible use of standardised tests – using certain sections or items of tests in assessment, analogous to practices described in previous surveys [26,28]. Nonetheless, discussions are merited concerning why standardised tests for intelligibility and functional communication are not widely used in current clinical practice, as they could facilitate evaluation of the effects of intervention on children’s communication in everyday life.

Limitations and future research

There are a number of limitations with the current work. First, the data upon which the article is based were collected during the summer of 2019, making it somewhat dated. However, to the best of our knowledge, no national guidelines or recommendations concerning the assessment or intervention of SSD in children have been published since the data was collected. We, therefore, believe that the data presented here are still relevant.

With regards to the representativeness of the sample of clinicians in the present study, 2359 SLPs were actively employed, in 2018 in Sweden. Of these, 1530 worked in healthcare services (Swe: “Hälso- och sjukvård”) and 262 in compulsory school [55]. Unfortunately, there are no statistics available concerning how many of these clinicians see children with SSD, as healthcare services and school settings encompass adult and adolescent patients as well as children with other communicative difficulties, in addition to children with SSD. With this in mind, we believe that the results can be expected to represent a meaningful portion of Swedish SLPs who see children with SSD. However, the variability in work settings and work experience may have affected the results. Many clinicians worked in hospitals and a large number reported having 5 years, or less, of work experience. Future studies of SLP practice would benefit from active recruitment to ensure a more representative balance across work settings and work experience.

In order to keep the questionnaire a reasonable length, we chose to focus on assessment components related to the processing and production of speech. This led to the exclusion of certain assessment components that are relevant for SSD assessment, such as a screening for language difficulties. However, few clinicians provided an answer when queried if the questionnaire had missed any components that were routinely included in their assessment of SSD.

Although the current survey provided clinicians with the opportunity to describe their practices, the level of detail in responses was not always sufficient to understand the manner in which assessment was conducted. For example, in responses concerning manner of assessment of oral-motor function, speech perception and phonological awareness, some clinicians simply wrote “pictures” or “exercises,” but did not specify how they were used. Future studies could further our understanding of clinical reasoning in SSD assessment by exploring unstandardised manners of assessment, for instance through interviews with clinicians. Additional research is also needed to investigate Swedish clinical practice concerning target selection and intervention for children with SSD, regarding which little is known. Finally, as other health-care professions can provide crucial information for SSD assessment, it is pertinent to explore the extent to which Swedish children with suspected SSD receive formal evaluation of hearing and non-verbal IQ, by audiologists and psychologists, respectively.

Clinical implications

We hope that the issues revealed in this survey (variability regarding frequency of assessment, confusion concerning terminology as well as issues with the manner of assessment for a number of components) will launch national discussions and encourage the creation of clinical guidelines for SSD assessment. The present study includes a number of suggestions concerning improvement of praxis; however, we believe that discussions concerning why current practice does not meet recommendations, with regards to SSD assessment, are necessary for successful integration of research findings in clinical practice. Swedish terminology to describe children with SSD also needs to be updated, so that Swedish clinicians are able to translate and apply international research findings to their own practice. We hope that national recommendations will lead to increased structure in SSD assessment across Sweden and thus contribute to equal care of high quality for all children with suspected SSD.

Conclusions

The results from this survey show that Swedish practice concerning assessment of SSD in children is variable with regards to how often, and in what manner, components, cited as important in international research and assessment guidelines, are assessed. This variability reveals areas for development for SLP education programmes and in-service training, and should motivate the development of Swedish clinical guidelines for SSD assessment. The current study is useful for Swedish SLPs who wish to scrutinise their own assessment practices as well as for international comparisons of SSD assessment.

Acknowledgements

The authors would like to thank all the clinicians who contributed with their time and expertise through responding to the questionnaire, as well as the SLPs who provided feedback on the pilot version of the questionnaire.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data available on request from the authors.

Notes

1 Note that the literal translation of the term used in the questionnaire would be “phonological output” (Swe: “fonologisk output”).

2 Note that Halland county’s municipalities are divided between Sweden’s West and South health care regions – in these analyses, the whole county (four respondents) is included in the West health care region.