For some researchers (e.g., Gierut, 1998 in an important ‘state-of-the-art’ article), the large ‘functional’ component includes children with CAS, children with articulation disorders and children with phonological disorders, all included in the same category. Other scholars (e.g., Ruscello, 2008), who also make the known-versus-unknown aetiological distinction, place children with phonetic (articulatory) and phonemic (phonological) difficulties in the populous ‘unknown origin’ group, followed by small subgroups of children who have at least one observable explanation for their speech difficulties. From Ruscello’s perspective, children in these subgroups may exhibit a range of organic issues: craniofacial anomalies, such as cleft lip and palate or dental malocclusion; sensory impairments, such as hearing loss; and motor speech disorders, such as a dysarthria (Hodge, 2010), apraxia due to a known cause, or idiopathic CAS, affecting speech–motor planning and/or speech–motor execution, in varying combinations. Ruscello uses the overarching term ‘sound system disorder(s)’ (SSD) to embrace all possibilities as opposed to my preferred term, ‘speech sound disorder(s)’ (also SSD). A noteworthy aspect of the ‘functional’ speech disorders is that there are now suggested subgroups that are linked to causes (Flipsen, Jr., 2002).

Drawing on data from several hundred case studies, Shriberg (2006) summarised seven putative subtypes of SSD (listed below) in a Speech Disorders Classification System (SDCS) based on genetic (inherited) and environmental risk factors. Furthermore, he suggested clinical prevalence percentages for the first four of these seven, coding them with ‘working terms’ and abbreviations as subtypes of speech delay. Prevalence was not estimated for a fifth speech delay subtype, Speech Delay-Dysarthria (SD-DYS), but potentially it would amount to less than 2% of the SSD speech delay population. In a separate SSD category of speech errors, Shriberg included Speech Errors-Sibilants (SE-/s/) and Speech Errors-Rhotics (SE-/r/). He added two additional (unnumbered) categories of SSD, also based on genetic and environmental risk factors whose working terms and abbreviations were Undifferentiated Speech Delay (USD) and Undifferentiated Speech Sound Disorder (USSD). An additional category, Motor Speech Disorder Not Otherwise Specified (MSD-NOS) was added in 2009 (Shriberg et al., 2010).

1. Speech Delay-Genetic (SD-GEN), 56% of referrals
   
2. Speech Delay-Otitis media with Effusion (SD-OME), 30% of referrals
   
3. Speech Delay-Developmental Psychosocial Involvement (SD-DPI), 12% of referrals
   
4. Speech Delay-Apraxia of Speech (SD-AOS), <1% of referrals
   
5. Speech Delay-Dysarthria (SD-DYS)
   
6. Motor Speech Disorder Not Otherwise Specified (MSD-NOS)
   
7. Speech Errors-Sibilants (SE-/s/)
   
8. Speech Errors-Rhotics (SE-/r/)

Undifferentiated Speech Delay (USD)

Undifferentiated Speech Sound Disorder (USSD)

Shriberg classified the primary origin or probable aetiology of the putative subtypes as

1. SD-GEN: Polygenic/Environmental
   
2. SD-OME: Polygenic/Environmental
   
3. SD-DPI: Polygenic/Environmental
   
4. SD-AOS: Monogenic? Oligogenic?
   
5. SD-DYS: Monogenic? Oligogenic?
   
6. SE-/s/: Environmental
   
7. SE-/r/: Environmental
   
   
   
   • USD: Any of 1–5
     
   • USSD: Any of 1–7

With regard to the processes affected, Shriberg’s breakdown was

1. SD-GEN: Cognitive–Linguistic
   
2. SD-OME: Auditory–Perceptual
   
3. SD-DPI: Affective–Temperamental
   
4. SD-AOS: Speech–Motor Control
   
5. SD-DYS: Speech–Motor Control
   
6. SE-/s/: Phonological Attunement
   
7. SE-/r/: Phonological Attunement
   
   
   
   • USD Processes Affected: Any of 1–5
     
   • USSD Processes Affected: Any of 1–7

Waring & Knight (2013, p. 12) comment, ‘The SDCS is still primarily a research tool driven by a search for genetic factors associated with speech disorders. The value of the theoretical underpinnings of the SDCS may not be fully apparent until the repercussions of genetic research impact upon speech pathology, sometime in the future. If researchers are able to identify specific markers that can be readily employed by clinicians to classify children with SSD of unknown origin into the eight putative SDCS subgroups, the classification system would become clinically useful. Further validation is required before the SDCS is used as a clinical tool’.

When classification systems based around aetiological distinctions with causal subgroups are applied, clinicians attempting differential diagnosis discover that the speech of some children is impossible to pigeonhole because it seems to ‘belong’ in more than one category. Broomfield and Dodd (2004a) discuss the unsurprising elusiveness of neat clinical categorisation, pointing out that it has caused several authors (Fox, Dodd & Howard, 2002; Stackhouse & Wells, 1997) to question the clinical utility of the aetiological approach. Commenting on Shriberg’s (1997) SDCS aetiological system, Broomfield and Dodd note three difficulties: (1) children do not fall easily into one subgroup or another; (2) it bears doubtful universality since it has not been trialled with non-English-speaking children; and (3) it provides no mechanism to account for developmental change. They go on to suggest that one alternative to the medical model approach is the theoretically strong psycholinguistic profiling approach proposed by Stackhouse and Wells (1997). But again, Broomfield and Dodd (2004a) are unconvinced, querying its utility on three points: one, its having little regard to surface phonology; two the lengthy diagnostic process involved in applying the framework; and three, the uncertainty of its universal applicability.

By contrast, Dodd (1995, 2005) proposed a differential diagnosis model with psycholinguistic foundations that is based primarily on linguistic profiling and speech subtypes. In it, specific speech subtypes are matched to discrete areas of psycholinguistic difficulty or breakdown that are ‘testable’ or ‘differentially diagnosable’. Dodd’s model enjoys some support for its universal applicability (Goldstein, 1996; So & Dodd, 1994). It embraces four subtypes that can occur at any age or stage of speech development, plus a fifth category for CAS. They are

• Phonological delay: in which children have a phonemic difficulty and all phonological rules or processes evident in a child’s speech output are attested in typical development, but are characteristic of children chronologically younger than the child in question (this large group accounts for 57.5% of referrals).
  
• Consistent deviant phonological disorder/ also known as consistent atypical phonological disorder: in which children have a phonemic difficulty with co-occurring non-developmental or unusual errors and developmental rules or processes, with the presence of unusual processes signalling that the child has impaired understanding of the target phonological system (this group comprises 20.6% of referrals).
  
• Inconsistent deviant phonological disorder/ also known as inconsistent phonological disorder: in which children have a phonemic difficulty and exhibit delayed and non-developmental error types and variability of production of single-word tokens equal to or greater than 40% (this group covers 9.4% of referrals).
  
• Articulation disorder: in which children had a difficulty at the phonetic level and are unable to produce particular perceptually acceptable phones. They exhibit substitutions or distortions of the same sound in isolation, words and sentences, during imitation, elicitation and spontaneous speech tasks (this group amounts to 12.5% of referrals).
  
• Childhood apraxia of speech: in which children have difficulty at the motor planning, programming and execution levels, exhibiting multiple deficits involving phonological planning, phonetic programming and motor programming implementation (less than 1% of children with SSD have CAS).

Faced with a range of approaches to classification, clinicians and researchers tend to use cover terms differently from each other, and there is variation from clinician to clinician and from researcher to researcher, sometimes relative to ‘what they grew up with’.

The intra- and interprofessional miscommunications that arise from having several approaches to the classification and description of SSD, and the tendency for professionals to mix-n-match them with abandon, spill over into intervention-related terminology and nomenclature. Gierut (1998), for example, talks about four ‘phonological treatments’: traditional sensory-motor articulation therapy, cycles therapy, the Metaphon approach and conventional minimal pair therapy. Summarising the traditional sensory-motor articulation approach, Gierut cites Van Riper and Emerick (1984) and Winitz (1969, 1975), whose work in speech intervention well and truly predated any explicitly motivated phonological intervention, and likely predated the term ‘phonological therapy’. She then provides an account of cycles (Hodson & Paden, 1991), with its traditional and metaphonological flavour (Kamhi, 2006, p. 275), about which Fey (1992) famously remarked, ‘there is nothing inherently phonological about the use of cycles’ (p. 279). At the same time, he added that because the approach aims to encourage gradual system-wide change, it is highly consistent with the goal common to all phonological approaches of facilitating reorganisation of the child’s system. The third account Gierut provides is of conventional minimal pair therapy (Weiner, 1981), the only one on her list of four to meet all of Fey’s (1985, reprinted 1992) exacting criteria for a ‘phonological’ therapy following the three ‘phonological principles’ outlined below. In Weiner’s approach, the therapist works at real-word (meaning) level, confronting children with their own homonymy, providing a semantic motivation to change production, thereby facilitating phonological reorganisation. The fourth ‘phonological treatment’ Gierut outlines is Metaphon (Dean, Howell, Waters & Reid, 1995), in which the SLP/SLT aims to eliminate persisting phonological processes via metalinguistic awareness tasks (Howell & Dean, 1994) involving imagery, minimal contrast activities, feigned listener confusion and guided discussion to promote self-monitoring of output.

Like Grunwell (1975) and Ingram (1976), Fey (1992) observed three basic principles underlying what he called phonology-based approaches to treatment such as conventional minimal pairs therapy (Weiner, 1981).

1. Modifying groups of sounds, or syllable structures
   The first principle concerned the modification of groups of sounds attacked according to an organising feature or systematic rule (of the child’s), rather than focusing on the ‘correction’ of individual phonemes. Targeting groups of sounds indicates acknowledgement of the systematic nature of phonology and the prospect of promoting generalisation of new learning across the child’s speech sound system. The speech production errors that the systemic rules represented fell into two main categories: substitution errors (also called systemic errors) where one sound or sound class is substituted for another (as in stopping, fronting, gliding, backing and assimilation); and structural errors, where the structure of the syllable or word changes (as in cluster reduction, diminutisation, schwa insertion or epenthesis, final consonant deletion, initial consonant deletion and weak syllable deletion). For example, if a child who was ‘stopping’ treated all fricatives as voiced stops, producing, systematically, fun as /bʌn/, sum as /dʌm/ and shoe as /du/, and so on, the therapist would target fricatives as a sound class, or frication as a feature, rather than treating, say, /f/ then /s/, then /ʃ/, and so on, position by position in word-initial, within-word and word-final contexts. Similarly, if a child’s speech were characterised by prevalent final consonant deletion, with many open syllables and productions like /kɒ/ for cough, /bʌ/ for bus, /pϵ/ for pet and /ϵ/ for edge, the therapist would encourage final consonant inclusion across his or her system, in preference to treating word-final /t/, /f/, /s/ and /dʒ/ individually and serially as word-final singleton omissions.
   
2. Establishing feature contrasts
   The second principle was around establishing feature contrasts as opposed to perfecting articulatory execution sound by sound and word position by word position. Phonetic placement techniques, such as hunting (Van Riper & Irwin, 1958), the ‘trail blazing’ (sic) progressive approximations method (Van Riper, 1963), the successive approximation procedure (Kaufman 2005; McCurry & Irwin, 1953), and shaping (Bernthal & Bankson, 2004; Shriberg, 1975) are all goal attack strategies used as intermediary steps towards adult-like phonetic execution of a therapy target. In the process of fostering such articulatory precision, clinicians may verbally encourage children to produce ‘a good crisp /s/’, ‘a clear /tʃ/’, ‘a perfect /k/’, ‘a sharp /t/’, ‘a beautiful /ŋ/’, or ‘a lovely /l/’. But in phonological therapy, the child is rewarded for creating contrast by using a sound in the target sound class, or a reasonable approximation of the target. For instance, with a child working at the systemic (substitution process) level to eliminate stopping of fricatives, a production like /ʃʌn/ for sun rather than /dʌn/ for sun would be rewarded, because /ʃ/ and /s/ are in the same (fricative) sound class, and both are voiceless. With a child working at the structural level, trying to learn final consonant inclusion, a production such as /biːm/ for bean rather than /biː/ for bean would be rewarded, because a final consonant was included, and furthermore the /m/ is a nasal consonant (like /n/). The clinician’s acceptance of phonemic contrast and the lack of emphasis on fine tuning of phonetic form, particularly in the early stages of therapy, can be difficult to explain or ‘sell’ to parents and caregivers (and even some SLPs/SLTs), especially if they are anxious and eager to see progress. Their expectation of a clinician’s role may be that we are ‘supposed to be’ encouraging perfection, and they can find it hard to understand that, if a goal in therapy is to eliminate stopping of fricatives, then /fɪp/ for ship is ‘more correct’ than /dɪp/ for ship and /sɪp/ for ship is even better!
   
3. Working at word level and making meaning
   The third principle had to do with the goal of making meaning, with the implication that the therapy itself must perforce be constructed around listening to, discriminating between, decoding and saying ‘real’ words. Indeed, it is a truism that phonological therapy must be at word level or above (i.e., word, phrase, sentence or conversational level) in order to signal to the child that the purpose of having a system of sound contrasts, or a phonological system, is to communicate (or to make meaning). The child discovers that homophony must be avoided and appropriate contrast established. If come, crumb, drum, gum, plum, some and thumb are all collapsed and realised homonymously as [dʌm], they come to appreciate that something (phonological) has to change.

Stoel-Gammon and Dunn (1985) reviewed the small (at the time) literature on the relationship between normal and disordered child phonology, finding a general view that, while there were many similarities between normal and disordered speech sound systems, there were also substantial differences between them. Their useful list of the most frequently described characteristics of developmental phonological disorders (as distinct from ‘phonetic’ or ‘articulation’ disorders) included the presence of

1. Static speech sound systems that had plateaued at an early level of development.
   
2. Extreme variability in production, without gradual improvement.
   
3. Persistence of phonological processes beyond typically expected ages.
   
4. ‘Chronological Mismatch’ (Grunwell, 1981) with mastery of ‘later’, ‘difficult’ sounds and structures and errors with sounds usually acquired early in development.
   
5. Idiosyncratic rules or processes that rarely occur in normal phonology.
   
6. Restricted use of contrast.

There is a regrettable tendency among SLPs/SLTs to ‘improve on’ the term phonological disorder, replacing it with: ‘phonological processing disorder’, ‘phonological process disorder’, or ‘phonological processes disorder’. Although they have crept into the vernacular, achieving prominence in the workplace, on the Web, in e-mail discussion, and on professional association websites, none of these three inappropriate terms is an acceptable synonym for either ‘phonological disorder’ or ‘developmental phonological disorder’.

Phonological disorder in the area of speech (Grunwell, 1987; Ingram, 1989a), phonological processes in the area of speech (Stampe, 1969), phonological processes in the area of literacy (Scarborough & Brady, 2002) and phonological processing in the area of literacy (Snowling, Bishop & Stothard, 2000) are four different things.

It its first 15 years, my website (Bowen, 1998b) attracted approximately 14 000 emails containing questions and/or requests for advice from among over 1 000 000 unique site visitors from around the world at the rate of between 10 and 20 emails a week. About half of the email came from SLP/SLT professionals and students and the other half from parents, adults with communication disorders and interested members of the general public.

A large number of the questions received relate to classification and terminology. For example, the parent of an unintelligible 4 year old asked, ‘Can you please tell me the difference between an articulation disorder and a phonological disorder; how can you tell them apart; and are they treated differently?’ Another wrote, ‘My five year old was diagnosed by one therapist who said he had a phonological disorder, but the therapist who is actually treating him says he has “a phonological processing disorder” and that we need to work on “his artic”. I am so, so confused by this: help!’ And this came from a colleague: ‘Although I have been an ASHA-certified school-based SLP for over 20 years, and most of my caseload is “artic”, I have to say I am confused about the difference between phonetic speech sound disorders, and phonemic speech sound disorders. In simple terms, what exactly is the difference, and can they exist concurrently in the same child?’

Such questions prompted the development of a plain English response on the website, a variation of which is shown in Box 2.1. Predictably, in light of the prevalence of SSD, it is one of the most-retrieved pages there and much used by students, clinical educators, and practitioners, including those preparing talks for consumer groups.

Within the overall SSD client population, there are two major sub-groups of children: (1) poorly unintelligible preschoolers with low percentages of consonants correct (PCCs) and multiple errors; and (2) acceptably intelligible school-aged students with high PCCs and ‘residual errors’. In practice, clinicians often reserve the terms ‘articulation disorder’ and ‘functional articulation disorder’ for the reasonably intelligible children of all ages, with one or just a few speech production difficulties, characteristically involving /s/, /z/, /l/ and /.ɹ/ and, in some settings, /θ/ and /ð/ also.

A high proportion, but we do not know precisely how high, of the unintelligible pre-schoolers have a phonological impairment entailing linguistic difficulties with organising speech sounds into patterns of sound contrasts. A small proportion (<1% of the paediatric SLP/SLT SSD caseload) has CAS, thought to be due to a deficit in speech motor control (ASHA, 2007; Shriberg, 2004, 2006; Shriberg, Campbell, Karlsson, McSweeney & Nadler, 2003). Each and every SSD can co-occur (e.g., phonetic and phonemic issues in the same child), and each can occur with other communication disorders (e.g., CAS and stuttering in the same child) or with other conditions (e.g., phonological disorder and ADHD in the same child).

Speech assessment involves careful, informed observations and hypothesis testing. The process typically begins with the referral followed by a preliminary, informal screening procedure in which the SLP/SLT listens to and watches the child speak. Usually without discussion or collaboration, the SLP/SLT develops, independently, a tentative explanation, or hypothesis, about the nature of any apparent difficulties with a view to conducting further investigations if needed. This initial screening may be as simple as making observations of the child in conversation, either with the therapist or with a parent, sibling or peer, or it may involve a short screening test. Appropriate screening reflects sensitivity to cultural and linguistic diversity (Goldstein, A19). In family-centred practice (Watts Pappas, A30), which is by no means universal, there is collaboration around the nature and conduct of further assessment. The SLP/SLT provides pertinent information, but it is a family’s decision whether to proceed, who should be present in assessment sessions, and so on.

A person who is not an SLP/SLT may be called upon to conduct initial screening and it may involve the use of computer software. For example, the 66-picture computerised Phoneme Factory Phonology Screener developed in the United Kingdom is designed for teachers to administer to children whom they suspect have speech sound difficulties, before referring them to SLP/SLT services for assessment. The teacher listens as the student names the pictures, writing down alphabetically or phonetically the child’s production of one particular sound per word.

The software has the capability of generating a report, based on what the teacher records, specifying the errors and patterns revealed, with normative comparisons and an indication of whether the child’s speech difficulties are ‘developmental’ or ‘disordered’. Any recommendation to refer to speech therapy is based on this report. The report also guides the teacher to appropriate activities to use in an associated software title in the series, the Phoneme Factory Sound Sorter program.

In testing the software, 408 children were assessed on the screener by a teacher–researcher and by an SLT using the phonology subtest of the DEAP (Dodd, Zhu, Crosbie, Holm & Ozanne, 2002). These two measures of the children’s speech were used to determine the screener’s sensitivity (71%), specificity (99%) and positive predictive values (81%). The order of testing was randomised (i.e., sometimes the children were assessed first by the teacher and sometimes first by the SLT) so as to control for order effects.

The first author of the Phoneme Factory software packages, Dr. Sue Roulstone, is Emeritus Professor of Speech and Language Therapy, University of the West of England, Bristol and a co-director of the Bristol Speech & Language Therapy Research Unit. She is a Fellow of the Royal College of Speech & Language Therapists and was Chair from 2004 to 2006. She received an Honorary Doctor of Health degree from Manchester Metropolitan University in 2013. Her research interests include evaluation of SLT service delivery, child and family perspectives on speech and language impairment, epidemiology of speech and language impairment and professional judgement and decision-making, and her response to Q8 reflects these interrelated topics.

Child speech assessments for the purpose of diagnosis are prompted by: (1) referral, including referral by a child’s family; (2) a child’s medical, sensory or developmental status, for example, the speech of children with cleft palate is routinely assessed in most of the industrialised world (Golding-Kushner, A17); or (3) failing a speech–language screening (see ASHA, 2004c, for further information). They are conducted by appropriately credentialed SLPs/SLTs, working individually or as members of collaborative teams that may include the child, family and others (Watts Pappas, A30).

Speech assessments are administered to children as needed, requested or mandated or where there are indications that individuals have articulation and/or phonology impairments associated with their body structure/function and/or communication activities/participation (McLeod, A1).

Depending on the presenting picture, the SLP/SLT typically examines the phonetic, phonological, perceptual, phonotactic, prosodic, speech motor and intelligibility aspects of the child’s speech. In evidence-based practice (E³BP), the particular tests chosen depend on the child’s presentation, the educated preferences and theoretical orientation of the clinician, and client/patient values and wishes.

The case history interview and/or a history questionnaire provide helpful information about the child and the family that may assist the therapist to manage the assessment and intervention process sensitively and appropriately. Ideally, information gathering is conducted with an eye to the potential ‘red flags’ for speech impairment (summarised in Box 2.2), including family history (Stein et al., 2011) or the risk and protective factors (Harrison & McLeod, 2010) that alert the clinician to a range of important risk factors and ‘leads’ to pursue.

Overby and Caspari (2013) note that SLPs/SLTs are hesitant to diagnose CAS in children younger than 2 years of age because of the paucity of information about early (birth – toddler) characteristics of the disorder, low volubility (little talking (DeThorne, Deater-Deckard, Mahurin-Smith, Coletto & Petrill, 2011)) in children with suspected CAS, and phonological overlap between CAS and other SSDs. Using home videos provided by children’s parents, Overby and Caspari (2013) compared the early sounds made by four children diagnosed with CAS between the ages of 3 and 5 years old to those of two typically developing children. Sounds examined in this study were organized into the following groups: vegetative sounds (burps, hiccoughs, sneezes, gulps, reflexive grunting, etc.); fixed signals (recurring motor sequences such as laughs, cries and moans that do not vary across individuals); protophones (quasivowels that are speech-like but not transcribable); and fully resonant productions that were transcribable (marginal babbling and canonical babbling for variable purposes). Because they were interested in tracking children’s development of perceptually accurate motor productions (an assessment consideration in the identification of CAS), they included fully resonant vowels and speech (meaningful speech or made up words) as fully resonant productions.

Emphasising the preliminary nature of this work, the individual variation among children with CAS, the current small sample size and the need to expand their comparison group to infants and toddlers with non-CAS severe SSDs, they found that children later diagnosed with CAS demonstrated, as infants and toddlers, a lack of diversity in place, manner and voicing compared to typically developing children. Children with CAS had a ‘place’ preference for bilabials and alveolars; a ‘manner’ preference for stops and nasals; limited babbling defined by reduced syllable shapes (dominated by vowels and CVs); and consonants that disappeared.