Psycholinguistic profiling reveals underlying impairments for Greek children with speech disorders

Abstract

This study investigated the possibility that similar speech error patterns in children may arise from different patterns of underlying speech processing difficulties. Four Greek children (aged 4;7–5;6 years) with similar speech output difficulties were assessed with a range of experimental tasks with phonologically matched items in order to profile a selection of the consonants and clusters, with which they experienced production difficulties. To check that the skills and words tested were typically acquired by this age, the tasks were also conducted on a control group of five typically developing children (aged 4;4–5;11 years). Tasks included nonword auditory discrimination, mispronunciation detection, naming, real word repetition and nonword repetition. For the experimental group, comparison of performance across tasks for each consonant or cluster identified different loci of difficulty across children. The findings suggest that this kind of profiling could reveal underlying speech processing difficulties, leading to implications for intervention.

Keywords

assessment children Greek psycholinguistic speech disorder

I Introduction

A screening study concerning the prevalence of communication impairment in children in the urban area of Patras, Greece revealed that, according to teacher perceptions, 11% of kindergarten children had some speech difficulties, including 4.4% who were totally unintelligible to adult listeners (Okalidou and Kampanaros, 2001). The incidence of articulation difficulties is higher than that reported in demographic studies in the USA (Shriberg et al., 1999) and Australia (McKinnon et al., 2007). Information on typical Greek speech acquisition is available; see, for example, Mennen and Okalidou (2007), but this is the only published paper on speech difficulties in Greek-speaking children.

Traditionally children’s speech difficulties have been assessed by naming tasks or word repetition tasks that enable phonological analysis of productions and comparison to adult realization of these words (Grunwell, 1989). Following these principles, the Evaluation Battery of Phonetic and Phonological Development (EBPPD) was developed by the Panhellenic Association of Logopedics (Panhellenic Association of Logopedics, 1995). The test aims to assess the development of consonants and consonant clusters, as well as to describe the phonological processes. Spontaneous picture naming is used to collect a sample of 70 single word utterances. The test provides developmental norms, based on a cross sectional study in a sample of 300 Greek children aged 2;6 to 6;0 years in the Attica region (central Greece). To consider that a consonant or cluster has been mastered in one age group, the researchers set the criterion of 75% of children in this group to have integrated the particular consonant or cluster in their system. This assessment is widely used in everyday clinical practice in Greece in order to assess the phonetic inventory and the phonological processes that characterize a child’s speech output.

There is a growing body of evidence suggesting that speech output difficulties can arise from a variety of underlying sources, and so a comprehensive assessment of a child’s speech should include a detailed assessment of input and representations, as well as output skills (Baker et al., 2001). According to Baker et al. (2001), this psycholinguistic approach helps to clarify the way in which children process speech and language at a cognitive or psychological level and devises hypotheses about the psychological processes or components that may be impaired. It also aids the setting of appropriate therapy goals. Baker et al. (2001) suggest that the model proposed by Stackhouse and Wells (1997) ‘can provide more comprehensive data: data that allow hypothesis testing about the possible problems underlying individual clients’ speech and literacy difficulties’ (p. 692).

1 The Stackhouse and Wells speech processing model

The Stackhouse and Wells speech processing model (Stackhouse and Wells, 1997) has been applied to the assessment of various groups of children with speech difficulties, including children with word-finding difficulties (Constable, 2001), deaf children (Pascoe et al., 2013; Rees, 2009) and children with persistent speech difficulties (Pascoe et al., 2005). The model is depicted in Figure 1.

Figure 1.

The Stackhouse and Wells speech processing model.

Three emboldened boxes represent the child’s stored knowledge about a word’s form (phonological representation), its meaning (semantic representation), and the specific articulatory gestures required for its pronunciation (motor program). Input processes include peripheral auditory processing, speech versus nonspeech discrimination and phonological recognition, where speech is recognized as the input language (e.g. English or Greek) and then decoded and compared with stored phonological representations. When motor programs are accessed, motor planning is involved in assembling them into an utterance. The motor plan is executed and gives rise to speech output at the level of motor execution. The narrow arrows between these boxes indicate the on-line processing of familiar information. The model includes off-line processing units, shown as shaded boxes. Off-line processes can be called upon when circumstances demand the processing of novel or unfamiliar material. For example, in input phonetic information can be recognized and learnt by an off-line level of processing called phonetic discrimination. In output motor programming allows the creation of new motor programs.

Different processing tasks involve different levels and interactions throughout the model. Therefore, by comparing the response to different tasks, the model can be used to identify the levels in the speech processing system that are giving rise to a child’s speech difficulties (Stackhouse and Wells, 1997).

2 Identifying input difficulties

Nonverbal responses to auditory discrimination tasks do not require output skills and so can reveal underlying speech input processing difficulties (Stackhouse and Wells, 1997). It is useful to compare children’s discrimination of real words with their discrimination of nonwords, where they are less likely to be using a lexical route and accessing stored representations. A number of studies have revealed auditory discrimination difficulties in children with speech output problems (Bridgeman and Snowling, 1988; Edwards, 2002; Rvachew and Grawburg, 2006). Some of these have compared real and nonword discrimination. For example, Bridgeman and Snowling (1988) compared a group of children with Childhood Apraxia of Speech (CAS) (aged 7;2–11;0 years) with typically developing reading age-matched controls in nonword and real word auditory discrimination tasks. Results showed that, for these older children, controls and children with CAS could discriminate between real words equally well. For nonwords, the children with CAS performed significantly worse than controls on discrimination tasks when there was a difference in phoneme sequence (e.g. [vɒst]–[vɒts]) but not in tasks with pairs that differed in one phoneme (e.g. [vɒs]–[vɒt]) (Bridgeman and Snowling, 1988). This indicated that the children with CAS had more difficulty with discriminating novel speech stimuli, when lexical representations are not available.

3 Assessment of phonological representations

For some input tasks, accessing the lexicon is a requirement, rather than an option. The definitive nature of such tasks can mean stronger hypotheses can be generated about the development and integrity of phonological representations. For example, in a mispronunciation detection task, where children are asked to decide if the target stimulus they hear matches the picture stimulus presented, the child has to access their phonological representation to make the comparison (Stackhouse et al., 2007).

Some children with speech impairments have been found to have difficulty with tasks tapping phonological representations (Rvachew and Grawburg, 2006; Sutherland, 2005). Rvachew and Grawburg (2006) studied a group of 95 four- and five-year-old children with speech disorders who were receiving speech and language therapy. Half of them had significant difficulty with a mispronunciation detection task. Sutherland and Gillon (2007) conducted a longitudinal study on the performance of nine children with moderate or severe speech impairment on tasks designed to tap underlying phonological representations, speech production and phonological awareness. Children with speech impairments scored significantly below typically developing controls on speech production tasks and had greater difficulty on word accuracy decision tasks. These findings suggest that some children with speech impairments have difficulty in storing and/or accessing accurate phonological representations.

4 Assessment of speech output

Picture naming has been widely used to assess speech output (Anthony et al., 1971; Grunwell, 1989; Panhellenic Association of Logopedics, 1995). Mispronunciations when naming indicate inaccurately stored motor programs. If the child produced the same inaccuracies when repeating the same words or matched nonwords, this would indicate additional difficulties with motor execution. Conversely, correct repetition would indicate relatively intact motor skills. Therefore investigations have compared naming with repetition tasks using real words and nonwords in order to distinguish between word storage difficulties and motor difficulties (Nathan, 2004; Vance et al., 2005). The EBPPD (Panhellenic Association of Logopedics, 1995) only assesses spontaneous naming and so does not distinguish between word storage difficulties alone and the possibility of additional difficulties with motor programming or motor execution.

5 Psycholinguistic profiling

Psycholinguistic profiling can be used to compare the child’s skills in a range of areas, such as phonological awareness, speech discrimination or motor execution (Pascoe et al., 2005). For this kind of profiling it is particularly important to use standardized or norm-referenced tests.

Psycholinguistic profiling can also be used for an in-depth investigation of a speech difficulty experienced by an individual child, where standardized or norm-referenced tests are less likely to be used. This difficulty is one not expected in a child of the same age and may be very specific, such as difficulty with a particular set of consonants and clusters that the child has difficulty producing. Ebbels (2000) and Rees (2009) demonstrated how profiling contrasts marked less typically by deaf children could reveal different loci of difficulty within the same child and across children. When profiling single consonants or clusters it is important to match test items across tasks as closely as possible. Then differences in performance are less likely to be due to the nature of the test items and more likely to be due to difficulty with the accuracy of representations or a skill (e.g. motor execution).

6 Greek phonology

Presently there are no studies using psycholinguistic profiling with Greek-speaking children with speech difficulties. There are some important similarities and differences between the English and Greek phonological system that need to be considered in such an investigation. The syllable structure of Greek can be described in the formula C(0–3)VC(0–1) (Mennen and Okalidou, 2007). For example: /ˈɐloɣo/ (‘horse’) V-CV-CV, /ˈmilo/ (‘apple’) CV-CV, /ˈspiti/ (‘house’) CCV-CV, /stɾɐˈtçotis/ (‘soldier’) CCCV-CCV-CVC, /ˈpoɾtɐ/ (‘door’) CVC-CV. Consonant clusters are not allowed in final position except in borrowings from Katharevaousa (Holton et al., 1998) and loan words (Setatos, 1974). Greek has 31 consonants (including the affricates /ts/ and /dz/) (Mennen and Okalidou, 2007). Any consonant can be used in word-initial and syllable-initial within-word positions. Setatos (1974) lists as many as 65 clusters (including /ks/) in syllable-initial and word-initial position. Research data suggest that the first syllabic structures to appear in a child’s speech are open type syllables (Kappa, 2002) that consist of plosives, nasals and laterals plus a vowel (Magoula, 2000). Then the palatal fricatives (and their allophones) appear and, subsequently, dental fricatives and trills (Magoula, 2000). Papathanasiou et al. (2012) report that 2-syllable; 3-syllable and multisyllabic words that consist of open syllables are accurately produced by four-year-old children, whereas close syllables are only accurate in word-final and not in within-word position. Their data indicated that no phonological processes are active in children aged 4;0–6;0 years and that the phonetic inventory is fully developed by the age of 5;6 to 6;0.

7 Study overview

This study aimed to investigate the value of profiling the psycholinguistic processing system of a group of Greek-speaking children with speech difficulties. It predicted that the use of this kind of profiling may reveal different underlying speech processing patterns across children and across the consonants and clusters tested. The specific hypothesis tested was:

For a group of Greek-speaking children with difficulties in producing particular consonants and clusters, the combination of loci of difficulties identified by profiling will vary across children and, for each child, across consonants and clusters.

The possible loci of difficulties investigated were:

auditory discrimination of the target consonants/clusters without reference to lexical representations;

phonological representations of words beginning with the target consonants/clusters;

motor programs of words beginning with the target consonants/clusters;

motor execution of the target consonant/cluster.

II Method

1 Design

This study presents four single case studies of Greek children with speech difficulties. Applying the Stackhouse and Wells (1997) model of speech processing, a range of tasks with matched stimuli were designed to explore the possible loci of difficulties for a selection of each child’s speech patterns.

The tasks were checked with a small group of typically developing children in the same age group to ensure tasks were age appropriate. A qualitative comparison was made between the profiles of each child and, for each child, between the profiles of each of the consonant and cluster examined. Those examined were those with which at least two of the children had production difficulties.

2 Participants

All the participants were monolingual speakers of Greek and they were tested in Greece by testers whose first language was Greek.

a Experimental group

Participants were selected from a larger group of children with identified speech difficulties. The following four were chosen as they showed similar errors in their speech production and were in an age range that differed by twelve months or fewer:

BT (male, 4;7 years)

AM (male, 4;9 years):

CK: (female, 4;11 years)

CL (male, 5;6 years)

All, except AM, had already been receiving speech and language therapy for 3 to 6 months. AM’s parents had decided not to opt for speech therapy sessions, hoping his difficulties would resolve in time. An oral examination of structure and function of each child revealed no major structural or functional abnormalities of the speech organs. It was not possible to compare the motor execution and motor programming skills with typically developing children as norms for these skills are not available for Greek-speaking children. All four children were attending local kindergarten schools. They were reported by their teachers to have age-appropriate vocabulary, understanding of grammar and nonverbal skills. AM had a history of otitis media with effusion, but the most recent hearing tests indicated that none of the children had hearing difficulties.

b Control group

Tasks developed for the experimental group were piloted with a small group of five typically developing children (4;4–5;11 years) attending the same kindergarten as one of the participants with speech difficulties, in order to check that the tasks were tapping skills typically acquired by this age group.

3 Materials

a Published assessments

The experimental group were assessed with the Evaluation Battery of Phonetic and Phonological Development (EBPPD) (Panhellenic Association of Logopedics, 1995), to date the only test available for speech assessment of Greek-speaking children. This single word picture naming test consists of 70 black and white pictures for target words that contain all the phonemes and phonotactic structures of the Greek language. The test identifies which consonants and clusters are missing from the child’s phonetic inventory and which systemic and structural phonological processes characterize their speech. It also provides norm references for the age of acquisition of phonemes. For each of the main four participants, the consonants and clusters realized incorrectly were consistently substituted with the same inaccurate realizations. All the participants realized /s/, /ts/ and /ks/ incorrectly in all permissible word positions in spontaneous picture naming and, for some, there were further difficulties with the production of /r/, /θ/ and /ð/. In Greek /ts/ is an affricate consonant and /ks/ is a consonant cluster (and so only appears in syllable-initial position). According to EBPPD norms these consonants and cluster are expected to be acquired by children at that age. Structural phonological processes such as cluster reduction and final consonant deletion in syllables were also present in the speech of three participants. As open syllables are much more common in Greek than closed syllables (Holton et al., 1998), the EBPPD refers to the phonological process of ‘final consonant deletion in syllables’, which includes deletion of consonants at the end of any syllable, whether they occur at the end of a word (for example /ˈðɾɐkos/ (‘dragon’) (CCV-CVC) produced as [ˈðɾɐko], or within the word (for example /ˈpoɾtɐ/ (‘door’) (CVC-CV) produced as [ˈpotɐ]). A summary of the results of the EBPPD for the main participants is shown in Table 1.

Table 1.

Results of EBPPD for main participants.

Child	Consonants and clusters absent from phonetic inventory	Systemic phonological processes	Structural phonological processes
BT	/r/, /ks/, /ts/	• /r/ realized as [l]	• Cluster reduction
		• /ks/ realized as [s]	• Final consonant deletion in syllables
		• /ts/ realized as [s]	• Final consonant deletion in syllables
AM	/ks/, /ps/, /ts/, /dz/, /s/, /z/, /θ/, /ð/	• Lateralization: /ks/, /ps/, /ts/, /dz/, /s/, /z/ realized as [kɬ], [pɬ], [tɬ], [dɮ], [ɬ], [ɮ]	• Cluster reduction
		• Fronting: /θ/→[f], /ð/→[v]	• Final consonant deletion in syllables
CK	/r/, /ks/, /ps/, /θ/	• /r/ realized as [l]	• Cluster reduction
		• /ks/ realized as [ts]
		• /ps/ realized as [ts]
		• /θ/ SIWW realized as [s]	• Final consonant deletion in syllables
CL	/ks/, /ps/, /ts/, /dz/, /s/, /z/	Fronting: /s/→[θ], [z]→[ð]	Structural processes were not present in the speech of CL (although clusters were often realized inaccurately)

Note. SIWW = syllable-initial within-word.

b Psycholinguistic assessments

A selection of those consonants and clusters incorrectly realized by the participants in the EBPPD were targeted for an investigation into the children’s processing. Consideration was given to those that were incorrectly realized by more than one child. All three selected (/s/, /ts/ and /ks/) were consistently realized in the same (incorrect) way by each child. A maximum of three sounds were selected for each child and only tested in syllable-initial position, to keep the testing time to a minimum, and they were elicited in three output tasks (naming, word repetition and nonword repetition). In the output tasks they were tested in word-initial and within-word positions and in the input tasks only in word-initial position. They were paired with their realizations in the child’s speech to form contrasts used for two input tasks (mispronunciation detection and nonword auditory discrimination). As the tasks were designed to form hypotheses about loci of difficulty, items were carefully matched across the tasks. Some output stimuli were used with several children (e.g. the same output tasks to elicit /ts/ were used with three children), but all the input tasks had to be tailor-made for each child, as the children all realized the targets tested in different ways. For a description of the targets chosen and tested for each child, see Table 2.

Table 2.

Consonants, clusters and contrasts selected for follow-up psycholinguistic tasks.

Participant	Targets selected	Realizations in EBPPD	Targets tested in output tasks	Contrasts tested in input tasks
BT	/ts/	[s]	/ts/	/ts/–[s]
	/ks/	[s]	/ks/	/ks/–[s]
AM	/s/	[ɬ]	/s/	/s/–[ɬ]
	/ts/	[kɬ]	/ts/	/ts/–[kɬ]
	/ks/	[kɬ]	/ks/	/ks/–[kɬ]
CK	/ks/	[ts]	/ks/	/ks/–[ts]
CL	/s/	[θ]	/s/	/s/–[θ]
	/ts/	[tθ]	/ts/	/ts/–[tθ]
	/ks/	[kθ]	/ks/	/ks/–[kθ]

Output tasks involved eliciting each target sound three times (15 times in total) to test consistency of production. All the children’s responses were transcribed phonetically by a qualified speech and language therapist. Input tasks involved 12 trials for the each contrast to ensure that performance was not due to chance. The stimuli used for profiling /s/–[θ] for CL are given as an example in Appendix 1.

The five experimental tasks were as follows:

Picture Naming: This task was designed to tap the child’s ability to access accurate stored representations for real words. Three words with the targets in syllable-initial position were each elicited five times by asking the child to name coloured pictures. The 15 items were elicited in a random order. For all the word and nonword repetition and for input tasks the tester sat to the side of the child when producing the spoken stimuli so that the children were not able to use visual cues.

Word Repetition: This task was designed to tap the child’s ability to produce a word without necessarily accessing stored representations. Children were asked to repeat the same list of stimuli words used in the naming task.

Nonword Repetition: This task was designed to evaluate motor programming and motor execution without needing to rely on lexical representations. Nonwords matching the real words used in the previous output tasks were formed by changing one consonant while keeping the target consonant or cluster, vowels and stress the same. For example, the nonword /siˈcea/ was formed to match the word /siˈmea/ (‘flag’). Porpodas (2005) suggests changing a consonant of a real word in order to form a nonword, while Paraskevaidis (2010) suggests that either a consonant or vowel can be changed. As there are only five vowels in Greek it was not possible to change a vowel in some of the target words without forming another real word. For consistency purposes it was decided to change a consonant in all words to form nonwords. Each of the nonwords was said by the tester in random order five times and the child was asked to repeat them. When nonwords were presented, the tester firstly placed a parrot puppet at the side of her face, telling the child that the puppet only produced silly words and not real words.

Mispronunciation Detection: This input task was designed to test the accuracy of stored phonological representations of words containing the problematic sounds. For each contrast tested, one of the words from the naming task was used. The child was shown the corresponding picture and the tester said the name six times correctly and six times erroneously by substituting the problematic phoneme or cluster for the child’s own individual realization. For example, if the child realized /s/ as [θ], the target word /siˈmea/ (‘flag’) would be produced six times correctly and six times as [θiˈmea]. The correct and incorrect productions of each word were presented in random order and nontested variables such as volume and intonation were kept consistent. Before the spoken stimuli were presented, the children were asked: ‘Are these the right words for the pictures?’ They were encouraged to respond with ‘Yes’ or ‘No’ and these responses were recorded. Judgements were scored as correct or incorrect. In order to ensure that the child understood the task, it began with a trial using a practice-item with initial consonant/s that were realized correctly as shown by the EBPPD.

Nonword Auditory Discrimination: This task was designed to test the child’s ability to discriminate the contrast tested without reference to lexical representations. For each contrast tested one nonword pair was formed. One of the nonwords in the pair was derived from the word used in the mispronunciation detection task. For example, the word /siˈmea/ (‘flag’) was changed to the nonword [siˈcea] by changing one of the consonants and keeping the targeted consonant/s, vowels and stress constant. The other nonword was formed by substituting the first phoneme in the first nonword for the child’s realization (e.g. [θiˈcea]). From each nonword pair (e.g. [siˈcea]–[θiˈcea]) a set of 12 task items was derived. Six were ‘same-pair’ items (3 × [siˈcea]–[siˈcea] and 3 × [θiˈcea]–[θiˈcea]) and 6 were ‘different-pair’ items (3 × [siˈcea]–[θiˈcea] and 3 × [θiˈcea]–[siˈcea]). Items in each set were presented in a random order, and nontested variables such as volume and intonation were kept consistent. The child was asked to judge whether the pair presented was the same or different. Judgements were scored as correct or incorrect. In order to make sure that the child understood the task, it began with a trial using a practice pair, where the initial consonants were both used correctly in naming (EBPPD task). When nonwords were presented, two pairs of animal puppets, a pair of identical ones, and a pair of different ones were placed in front of the child to demonstrate the concept of same/different. The child was told that the puppets only produced silly words and not real words, and that the identical puppets would sound the same whereas the different puppets would sound different. When a pair of auditory stimuli was presented the child could either respond verbally, by saying ‘same’ or ‘different’, or by pointing to the relevant pair of toys.

The experimental participants completed these tasks for the sounds and contrasts that were selected for them as individuals (see Table 2). The control group completed all the output tasks that were used with the participants (i.e. those for /s/, /ts/ and /ks/). As each of the participants realized their problematic sounds differently, experimental input tasks were designed for nine different contrasts. Therefore, in order to restrict the time spent testing, the control group completed input tasks for just three of these contrasts (/s/–[θ], /ts/–[s] and /ks/–[s]).

All the participating children were assessed by a qualified speech and language therapist in a small quiet room in their kindergarten school, with tasks presented in the following order:

Picture Naming

Nonword Auditory Discrimination

Mispronunciation Detection

Nonword Repetition

Word Repetition

The responses in the output tasks were audio-recorded as well as transcribed at the time of recording. The audio recordings were transcribed again by the tester and by another speech and language therapist. When all of the sets of transcriptions were compared there was no disagreement about whether the target phoneme or cluster had been realized correctly or incorrectly.

c Scoring for psycholinguistic tasks

For each input task, correct judgements for each stimulus provided a raw score. Raw scores of 11 or 12 (out of 12) were taken as an indication of a good performance that was not likely to be due to chance as the probability of 11/12 occurring by chance is 0.006 (Siegel and Castellan, 1988). A raw score of less than 11 was taken as an indication of difficulty with discriminating the contrast tested as the p values of these occurring by chance are greater than 0.01 (Siegel and Castellan, 1988).

For each output task, a response was marked as correct if the problematic phoneme or cluster elicited in the word or nonword was realized correctly. Minor phonetic variations were accepted if the realizations of target phonemes could not be perceived as other phonemes. A score of 15 in the naming tasks would indicate that the target phonemes and cluster were well specified in the motor programs. A score of less than 15 could indicate how consistently the target was realized correctly. Any correct responses in the repetition tasks would indicate a child’s potential to execute a phoneme or cluster, and the exact score would indicate the degree of stimulability of the sounds.

III Results

1 Control group

The group of five typically developing children in a similar age range performed at or near ceiling on the experimental tasks.

Input tasks: For the nonword discrimination, three children scored 11/12 for one task and 12/12 for the remaining two tasks, and the other two children scored 12/12 for all three tasks. For the mispronunciation detection, one child had two scores of 11/12 and two of 12/12, three children had one score of 11/12 and three scores of 12/12. The remaining child had full scores for all four tasks. The probability of the score of 11/12 occurring by chance is 0.006 (Siegel and Castellan, 1988).

Output Tasks: Apart from one score of 14/15 in one nonword repetition for one child, all children scored 15/15 for all the output tasks. The mean scores for each participant in the control group are presented in Table 3.

Table 3.

Control participants’ mean raw scores on experimental tasks.

Participant	Age	Input tasks		Output tasks
Participant	Age	Nonword discrimination (max=12)	Mispronunciation detection (max=12)	Naming (max=15)	Word repetition (max=15)	Nonword repetition (max=15)
1	4;4	11.33	11.50	15.00	15.00	15.00
2	4;7	11.33	11.75	14.75	15.00	15.00
3	4;8	12.00	11.75	15.00	15.00	15.00
4	4;11	12.00	12.00	15.00	15.00	15.00
5	5;4	11.33	11.75	15.00	15.00	15.00

2 Experimental group (with speech difficulties)

All four participants had difficulties with the experimental tasks.

a Input tasks

For the nonword discrimination, CL’s scores were 9/12, 11/12 and 12/12. The three remaining participants scored 12/12 for all tasks. For the mispronunciation detection, raw scores ranged from 6/12 to 12/12, with AM and CK scoring at chance level for all the tasks.

b Output tasks

For the naming task there were no accurate productions of /ts/, /ks/ and /s/ either in SIWI (syllable-initial word-initial) or SIWW (syllable-initial within-word) position for any of the participants. BT was able to repeat words and nonwords containing /ts/ with 60% and 80% accuracy respectively. The remaining three children scored 0 for all the output tasks. The results of the main participants are presented in Table 4.

Table 4.

Main participants’ raw scores on experimental tasks for each consonant/cluster.

Participant	Age	Problematic consonant/cluster	Input tasks		Output tasks
Participant	Age	Problematic consonant/cluster	Nonword discrimination (maximum = 12)	Mispronunciation detection (maximum = 12)	Naming (maximum = 15)	Word repetition (maximum= 15)	Nonword repetition (maximum = 15)
BT	4;7	/ts/	12	12	0	9	12
		/ks/	12	12	0	1	0
AM	4;9	/s/	12	6*	0	0	0
		/ts/	12	9*	0	0	0
		/ks/	12	9*	0	0	0
CK	4;11	/ks/	12	7*	0	0	0
CL	5;6	/s/	9*	12	0	0	0
		/ts/	12	11	0	0	0
		/ks/	11	12	0	0	0

Note. * These scores are at chance (p < 0.01).

IV Discussion

The scores for the control group children suggest that the tasks designed for the main participants were tapping skills typically acquired by children from 4;4–5;11 years. All the main participants in the experimental group had difficulties with the majority of the tasks conducted but their individual profiles differed.

BT’s input test scores suggest that he was able to discriminate between all the contrasts tested in nonwords, and that the targets /ts/ and /ks/ were well specified in the phonological representations of the words tested. Although BT never realized /ts/ and /ks/ correctly in naming tasks he had some success in repetition tasks. For /ts/, scores for word and nonword repetition respectively were 9/15 and 12/15. This suggests that he had the potential to execute this consonant but was less likely to do so when accessing lexical representations. The only success he had in realizing /ks/ was a score of 1/15 in word repetition. Although this indicates that /ks/ is less stimulable than /ts/ it does suggest that he has the potential to execute the cluster.

AM’s input test scores suggest that he was able to discriminate between all the contrasts tested in nonwords (despite his history of otitis media), but that /s/, /ts/ and /ks/ were not well specified in the phonological representations of the words tested. AM never realized these sounds correctly in any of the output tasks. There was therefore no evidence that he had the motor execution skills to produce these consonants.

For CK, only /ks/ and the contrast /ks/–[ts] was profiled. Her input test scores suggest that she could discriminate the contrast in nonwords, but that /ks/ was not well specified in the phonological representation of the word tested. CK never realized /ks/ correctly in the output tasks. Therefore there was no evidence that she had the motor execution skills to produce this cluster.

CL’s input scores for nonword discrimination suggest that he could discriminate between /ts /–[θ] and /ks/–[kθ] but had difficulty with /s /–[θ]. He was able to discriminate this contrast in the mispronunciation detection task when he was accessing his phonological representation. This could indicate that he needed top-down processing to aid his discrimination of /s/–[θ]. His scores on the mispronunciation tasks suggest /s/, /ts/ and /ks/ were well specified in the phonological representations of the words tested. CL never realized /s/, /ts/ and /ks/ correctly in naming tasks and real and nonword repetition tasks. There was therefore no evidence that he had the motor execution skills to produce these targets.

These profiles reveal that the participants differed in performance across tasks, indicating different loci of difficulties. For the input tasks all the errors involved judging different pairs to be the same or an incorrect version of a word to be correct. Both the same and different pairs and the correct and incorrect stimuli were balanced in number. This suggests that the children understood the nature of the task but sometimes had difficulty with auditory discrimination of the differences tested. Except for CL’s responses to one nonword auditory discrimination task, all the children demonstrated a good ability to discriminate the contrasts tests in nonwords. Bridgeman and Snowling (1998) found that their group of children with CAS (aged 7;2–11 years) did not have difficulty with nonword pairs that differed in one phoneme (e.g. [vɒs]–[vɒt]) (Bridgeman and Snowling, 1988). Similarly, the children in this study had very little difficulty with nonword discrimination that does not require access to lexical representations.

Two of the four children (AM and CK) had difficulties in discriminating the contrasts tested in the mispronunciation detection tasks when they were required to access phonological representations. This indicates that, for AM and CK, the target sounds were not well specified in the phonological representations of the words tested, despite an ability to discriminate the same contrast in nonwords. This phenomenon is described by Bryan and Howard (1992), who proposed that, for some children, phonological representations become frozen in that they are resistant to change, despite an improved ability to discriminate the contrast. The other two children (BT and CL) had no difficulty with the mispronunciation detection tasks, despite their speech output difficulties. This suggests that the target sounds were well specified in phonological representations but not in motor programs. This finding concurs with the findings of Rvachew and Grawburg (2006) and Sutherland and Gillon (2007) that a subgroup of children with speech disorders have problems with underlying phonological representations.

All four children had difficulty with the naming tasks, never realizing the target sounds accurately. This was expected from the previous performance on the EBPPD and confirmed that the children’s speech realizations were consistently inaccurate. This suggests that all children had inaccurate motor programs for the words tested due to the target sounds being poorly specified. Three of the children (AM, CK and CL) were also unable to produce the target sounds in word repetition or nonword repetition. This suggests that they had an additional difficulty with the motor execution of these sounds in potential words. The oral examination of structure and function did not discover any anatomical anomalies or problems with nonspeech oral movements, indicating that these three children had the physical potential to produce the targets in simpler phonetic environments, but were unable to execute them in the nonwords tested. It would have been informative to assess the children’s ability to imitate the targets in other simpler phonetic environments (e.g. in isolation). Nevertheless, the profiling indicated that BT was the only child demonstrating an ability to execute his target sounds in words or nonwords. His high degree of success (80%) in producing /ts/ in nonwords suggests that, when there was less interference from the inaccurately stored motor programs, he could execute the phoneme in segment combinations that were very similar to real words. Interestingly BT (4;07 years) was the youngest child. The findings of Vance et al. (2005) suggested that young children make more use of bottom-up processes for repetition tasks, and so this may explain why the older children were not able to execute the target sounds in repetition tasks. If they were using top-down processing, their inaccurate motor programs would be more likely to interfere with the task.

For each child, their profile for the target sounds tested was very similar. For example, AM performed at chance in the mispronunciation tasks for all three targets tested and did not produce the targets in any output task, and BT had full scores for the input tests for both targets tested and was able to execute both targets in repetition tasks. This may be further evidence that there are subgroups of children with speech impairments that have a general difficulty with an aspect of speech processing, such as the formation of accurate phonological representations, as shown by Rvachew and Grawburg (2006) and Sutherland and Gillon (2007).

Four children were investigated in this study and, in order to select consonants and clusters that all the children had difficulty producing, only two consonants and one cluster were profiled. In order to minimize testing time for each child these were only tested in word-initial position. Therefore it is not possible to generalize the results to other consonants and clusters and other word positions.

It would be useful to examine the robustness of the tasks used in this study to see if they would yield similar results if used with the same child with speech difficulties over a short period of time without intervention. It would be useful to conduct investigations profiling a wider range of consonants and clusters in different word and syllable positions with a larger number of children with speech delay and disorder.

The data collected did suggest different loci of difficulty for the four children who had similar difficulties with speech output. Previous studies involving profiling consonants and clusters inaccurately produced by deaf children in this way have also revealed different loci of breakdown (Ebbels, 2000; Rees, 2009).

V Implications for intervention

Clinical decisions regarding intervention for children with speech disorders will depend on a combination of factors, including the clinician’s theoretical perspective, the nature of the child’s difficulty, the efficacy of different approaches and the child’s response to trial therapy (Kamhi, 2006). To fully understand the nature of a child’s speech difficulty it is important that the assessment reveals patterns both in the production of spontaneous speech and any underlying impairments. For example, it is important to know if the child has the potential to execute any problematic sounds in repetition tasks, indicating their stimulability. Outcomes for intervention are generally better when stimulable sounds are targeted (Rvachew, 2005). This study suggests that it is also important to know whether sounds are well specified in phonological representations, as this will vary from child to child. Inaccurate phonological representations imply that the child is unable to discriminate the difference between their inaccurate productions of words and the target productions, thus suggesting some auditory training as part of the intervention, where the child is familiarized with the auditory difference in productions. Auditory training is often more effective if it is combined with output work, as intervention should take advantage of the whole connected speech processing system (Rees, 2001). Rvachew (2005) found that outcomes for unstimulable sounds were improved by including phonemic perception training alongside phonetic placement procedures.

The investigations conducted on the four children in this study led to different implications for intervention. For example, intervention with BT and CL could focus on output work for the selected targets, in the knowledge that input skills seem to be relatively intact. As BT can imitate the targets, intervention can focus on updating motor programs including the targets. However, CL would need more help with producing the targets in isolation and simpler phonetic environments. Intervention with AM and CK could integrate auditory training with output work, knowing that the targets are not well specified in phonological representations.

The psycholinguistic tests used in this study did not require expensive resources. They required knowledge of the spoken language being assessed and the understanding of a set of psycholinguistic principles. Therefore this kind of follow-up profiling could potentially be used with children with speech difficulties in a wide range of languages.

Footnotes

Appendix

Appendix 1.

Stimuli in experimental task for participant CL for /s/→/θ/.

Test/s	Stimuli
Naming and word repetition	/siˈmɛɐ/ (‘flag’) × 3
	/sɐliˈɡɐɾi/ (‘snail’) × 3
	/sokoˈlɐtɐ/ (‘chocolate’) × 3
Nonword repetition	[siˈcɛɐ] × 3
	[sɐliˈtɐɾi] × 3
	[sofoˈlɐtɐ] × 3
Mispronunciation detection	For picture of /simɛɐ/ (‘flag’):
	/simɛɐ/ × 6
	[θimɛɐ] × 6
Nonword auditory discrimination	[sicɛɐ]–[sicɛɐ] × 3
	[θicɛɐ]–[θicɛɐ] × 3
	[sicɛɐ]–[θicɛɐ] × 3
	[θicɛɐ]–[sicɛɐ] × 3

Acknowledgements

The authors would like to acknowledge financial support from the Greek Scholarship Foundation (IKY). They would also like to thank Ilias Papathanasiou for referring participants, the children and their parents for taking part, the children’s teachers for their co-operation, and Maggie Vance for her comments on an earlier draft of the article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Anthony

Bogle

Ingram

McIsaac

(1971) Edinburgh articulation test. Edinburgh: Churchill Livingstone.

Baker

Croot

McLeod

Paul

(2001) Psycholinguistic models of speech development and their application to clinical practice. Journal of Speech, Language, and Hearing Research 44: 685–702.

Bridgeman

Snowling

(1988) The perception of phoneme sequence: A comparison of dyspraxic and normal children. British Journal of Disorders of Communication 23: 245–52. doi:10.3109/13682828809011936

Bryan

Howard

(1992) Frozen phonology thawed: the analysis and remediation of a developmental disorder of lexical phonology. European Journal of Disorders of Human Communication 27: 343–65.

Constable

(2001) A psycholinguistic approach to word finding difficulties. In: Stackhouse

Wells

(eds) Children’s speech and literacy difficulties 2: Identification and intervention. London: Whurr, 330–65.

Ebbels

(2000) Psycholinguistic profiling of a hearing-impaired child. Child Language Teaching and Therapy 16: 3–22.

Edwards

(2002) Final consonant discrimination in children: Effects of phonological disorder, vocabulary size, and articulatory accuracy. Journal of Speech, Language, and Hearing Research 45: 231–242.

Grunwell

(1989) Assessment of phonology. In: Grundy

(ed.) Linguistics in clinical practice. London: Whurr, 108–36.

Holton

Makridge

Philippaki-Warburton

(1998) Grammar of Greek language (in Greek). Athens: Pataki.

10.

Kamhi

(2006) Treatment decisions for children with speech–sound disorders. Language, Speech, and Hearing Services in Schools 37: 271–79.

11.

Kappa

(2002) On the acquisition of syllabic structure in Greek. Journal of Greek Linguistics 3: 1–52.

12.

Magoula

(2000) The acquisition of phonological system in Modern Greek: The use of consonants (in Greek). Unpublished PhD thesis, University of Athens, Athens, Greece.

13.

McKinnon

McLeod

Reilly

(2007) The prevalence of stuttering, voice, and speech–sound disorders in primary school students in Australia. Language Speech and Hearing Services in Schools 38: 5–15.

14.

Mennen

Okalidou

(2007) Greek speech acquisition. In: McLeod

(ed.) The international guide to speech acquisition. Clifton Park, NY: Thompson Delmar Learning (USA), 398–411.

15.

Nathan

(2004) The development of early literacy skills among children with speech difficulties: A test of the ‘critical age hypothesis’. Journal of Speech, Language, and Hearing Research 47: 377–91.

16.

Okalidou

Kampanaros

(2001) Teacher perceptions of communication impairment at screening stage in preschool children living in Patras, Greece. International Journal of Language and Communication Disorders 36: 489–502.

17.

Panhellenic Association of Logopedics (1995) Evaluation Battery of Phonetic and Phonological Development. Athens: PAL.

18.

Papathanasiou

Dimitrakopoulou

Ntaountaki

Vasilou

(2012) Phonetic and phonological development in Greek children aged 4:0–6;0. Unpublished paper presented at the ASHA Convention 15–17 November, Atlanta, GA, USA. Retrieved from: http://www.asha.org/events/convention/handouts/2012/5037-phonetic-and-phonological-development-in-greek-children-age-4-6 (April 2015).

19.

Paraskevaidis

(2010) Norms on tests assessing verbal working memory using tasks of reproduction of lists of words and pseudowords (original in Greek). Unpublished MSc thesis, Aristotle University of Thessaloniki, Thessaloniki, Greece.

20.

Pascoe

Randall-Pieterse

Geiger

(2013) Speech and literacy development in a child with a cochlear implant: Application of a psycholinguistic framework. Child Language Teaching and Therapy 29: 185–200.

21.

Pascoe

Stackhouse

Wells

(2005) Phonological therapy within a psycholinguistic framework: Promoting change in a child with persisting speech difficulties. International Journal of Language and Communication Disorders 40: 189–220.

22.

Porpodas

(2005) Educational approaches and materials for the evaluation and management of learning difficulties in primary school EPEAEK 2000–2006 (original in Greek). Unpublished research project, Greek Ministry of Education, Athens, Greece.

23.

Rees

(2001) Principles of psycholinguistic intervention. In: Stackhouse

Wells

(eds) Children’s speech and literacy difficulties 2: Identification and intervention. London: Whurr, 41–65.

24.

Rees

(2009) Deaf children’s acquisition of speech skills: A psycholinguistic perspective through intervention. Unpublished doctoral thesis, University College London, London, UK. Retrieved from: http://discovery.ucl.ac.uk/1348395 (April 2015).

25.

Rvachew

(2005) Stimulability and treatment success. Topics in Language Disorders Clinical Perspectives on Speech Sound Disorders 25: 207–19.

26.

Rvachew

Grawburg

(2006) Correlates of phonological awareness in preschoolers with speech sound disorders. Journal of Speech, Language, and Hearing Research 49: 74–87.

27.

Setatos

(1974) Phonology of Modern Greek (in Greek). Athens: Editions Papazisis.

28.

Shriberg

Tomblin

McSweeny

(1999) Prevalence of speech delay in 6-year-old children and comorbidity with language impairment. Journal of Speech Language and Hearing Research 42: 1461–81.

29.

Siegel

Castellan

(1988) Nonparametric statistics for the social sciences. New York: McGraw-Hill.

30.

Stackhouse

Wells

(1997) Children’s speech and literacy difficulties: A psycholinguistic framework. London: Whurr.

31.

Stackhouse

Vance

Pascoe

Wells

(2007) Compendium of auditory and speech tasks: Children’s speech and literacy difficulties 4. Chichester: John Wiley.

32.

Sutherland

(2005) Assessment of phonological representations in children with speech impairment. Language, Speech, and Hearing Services in Schools 36: 294–307.

33.

Sutherland

Gillon

(2007) Development of phonological representations and phonological awareness in children with speech impairment. International Journal of Language and Communication Disorders 42: 229–50.

34.

Vance

Stackhouse

Wells

(2005) Speech production skills in children aged 3–7 years. International Journal of Language and Communication Disorders 40: 29–48. doi:10.1080/13682820410001716172