Abstract
Childhood Apraxia of Speech (CAS) is a disorder which affects not only speech-motor planning but also language ability more broadly. The present study attempts to explore the morphosyntactic, socio-cognitive, and prosody comprehension skills of children with CAS. Twenty preschool children with CAS and 20 typically developing children listened to a story, while viewing a wordless PowerPoint, picture presentation on a computer screen. Then, they were instructed to retell the story. Each participant listened to two stories, one with “lively” and one with “flat” prosody. The results revealed poor socio-cognitive abilities which were also evident in the use of verb-complement clauses, and poor prosody comprehension skills in the group with CAS. These findings affirm that morphosyntactic deficits are not the core characteristic of CAS, and suggest a connection between the production and comprehension of prosody, and between prosodic and (socio)-cognitive skills.
Keywords
Introduction
Childhood Apraxia of Speech (CAS) is broadly characterized as a speech disorder affecting speech motor planning and/or programing. It can occur both as an idiopathic disorder, namely limited to a core motor speech deficit in the absence of problems in hearing, intelligence, and language comprehension, or as one of the results of a neurobehavioral or genetic disorder (American Psychiatric Association [APA], 2013; American SpeechLanguage and Hearing Association [ASHA], 2007; see also Wilson et al., 2019). The main symptoms of CAS include; (1) the making of inconsistent errors in phonemic and word production, (2) the phenomena of lengthened and disrupted coarticulatory transitions between sounds and syllables, and the production of unusual prosody, especially in the realization of lexical or phrasal stress (ASHA, 2007, see also Terband et al., 2019). Growing evidence, however, indicates that CAS is not just a motor planning/programing deficit, but one that affects language ability in a broader sense (e.g., Marion et al., 1993; McNeill et al., 2009; Nijland et al., 2003), while research has shown that displaying speech deficits can also affect individuals’ socio-cognitive skills (e.g., Aro et al., 2012; Durkin & Conti-Ramsden, 2007; Tarshis et al., 2020).
Morphosyntax in CAS
In particular, morphosyntax seems to cause problems for children with CAS (Ekelman & Aram, 1983; Lewis et al., 2004; McNeill & Gillon, 2013; Murray et al., 2019). More specifically, Ekelman and Aram (1983) found that although spontaneous speech of children with CAS had an advanced mean length of utterance (MLU), it contained pronoun selection errors and inflectional morpheme omissions in verbs. Similarly, Lewis et al. (2004) reported low performance of children with CAS in sentence comprehension, grammatical completion, and sentence imitation. McNeill and Gillon (2013), on three occasions over a period of 1 year, analyzed the language production skills of three children with CAS. They employed a story retelling task (after listening to a model story), and they examined MLU, based on morpheme count, as well as word and morpheme errors and omissions (e.g., function words, inflectional morphemes such as third-person singular “s,” past tense “-ed,” irregular verbs). Results were not consistent for all three of the children and thus researchers concluded that “morpho-syntactic errors in CAS are less likely to be related to the core deficit in the disorder but rather to concomitant language difficulties that many children with the disorder experience” (McNeill & Gillon, 2013, p. 16). Lastly, Murray et al. (2019) assessed the morphological skills of 26 preschool children with CAS using sentence-recalling, word structure, and expressive vocabulary tests. The results demonstrated that children with CAS had poorer expressive language skills compared to their receptive skills, while half of them demonstrated poor accuracy on a range of morphemes. According to the authors, such a performance cannot be attributed to motor speech deficits of children with CAS and, thus, morphological difficulties can be co-morbid to CAS. Therefore, it appears that, until now, there has been only a very limited number of studies dealing with morphosyntax of a small number of children with CAS and that there are morphosyntactic phenomena that have not been explored yet or they have not been explored sufficiently. Thus, additional exploration of morphosyntax in children with CAS is required in order for safe conclusions to be extracted regarding their morphosyntactic abilities.
Socio-Cognitive Abilities in CAS
Research, however, has demonstrated that speech deficits can affect, apart from language abilities, individuals’ socio-cognitive skills (e.g., Aro et al., 2012; Durkin & Conti-Ramsden, 2007). In particular, preschool children with speech impairments, such as expressive delay, and specific language impairment, have been found to initiate conversation less often, to be less successful in keeping conversational interactions going, to be more likely ignored by peers, and in general to refrain from interacting with their peers more often when compared to typically developing children (e.g., Aro et al., 2012; Durkin & Conti-Ramsden, 2007). Social interactions, however, are reported to help children develop their ability to understand other people’s emotions, mental state, and desires in order to predict others’ behavior (Tarshis et al., 2020), as well as enhance individuals’ ability to take into account the perspective of the listeners, namely, to being sensitive to the prior knowledge and understanding of others and adjusting the way information is conveyed to meet their cognitive and social needs (Kecskes, 2010; Tarshis et al., 2020). These abilities are components of socio-cognitive abilities (Kecskes, 2010; Tarshis et al., 2020; for a review regarding the connection between young children’s social interactions and their psychological understanding of persons see de Rosnay & Hughes, 2006). Although children with CAS are expected to face social communication difficulties and poor psychological understanding of persons, to the best of our knowledge, there are only two experimental studies addressing this issue (among others). In particular, Teverovsky et al. (2009) reported that children with CAS have socio-cognitive deficits in that they face difficulties in understanding non-verbal and spoken messages, in producing non-verbal messages, while they also have a poor ability to avoid dangerous situations because of not recognizing other people’s intentions. Similarly, Rusiewicz et al. (2018) reported that children with CAS feel embarrassed of their speech production, and thus they refrain from speaking in certain situations, have difficulty making friends, and avoid participating in daily activities, events, and social situations (for children with both CAS and developmental dyspraxia see also Sylvestre et al., 2013). These results, however, were based on parents’/caregivers’ answers to relevant questions and they were not extracted by assessing children’s own performance on relevant tasks. Thus, more research is needed when exploring the socio-cognitive skills of children with CAS.
Prosody Comprehension in CAS
Impaired production of prosody is considered a primary diagnostic criterion for CAS (ASHA, 2007; see also Haley & Jacks, 2019; Terband et al., 2019), while difficulties in prosody comprehension have also been reported. In particular, Nijland et al. (2003) and Marion et al. (1993) reported that children with CAS display difficulties in the comprehension of prosodic cues, such as the duration of words and phonemes (e.g., “ice cream” vs. “I scream”) (Nijland et al., 2003) or the comprehension (and production) of rhyme (Marion et al., 1993).
According to the Perceptual Salience Approach (Echols & Newport, 1992), prosody aids the development of phonological processing since more phonological detail is understood for perceptually salient syllables than is for less salient ones (those with a higher pitch and longer duration, e.g.,). This claim agrees with earlier research, which showed that speech delivered using good prosody improved typically developed individuals’ verbal sentence comprehension and, consequently, phonological memory (e.g., Martzoukou et al., 2020; Shintel et al., 2014), while deficits in prosody comprehension have been reported to extend to memory processes which store and retrieve lexical representations (Shriberg et al., 1997, 2012). Therefore, impaired prosody can affect both comprehension and memory. Despite the key role of prosody, very few studies have explored the prosody perception abilities of children with CAS (Marion et al., 1993; Nijland et al., 2003). Thus, more research is clearly needed.
To summarize, the language, socio-cognitive, and prosody comprehension abilities of children with CAS remain poorly understood. In particular, there is a substantial area of language development in children with CAS that has not been rigorously investigated, and what is more, socio-cognitive abilities have not been properly addressed in previous studies, given that findings were based on parental/caregiver reports, rather than on objective assessments of children’s performance. Similarly, given the limited number of studies on prosody perception in children with CAS, it is clear there is a need to further explore this aspect. The present study aims at vanquishing critical paucity in our knowledge about CAS by systematically examining how all these variables operate within the same group of children with CAS. To this end, a narrative retelling task was adopted.
Narratives as a Means of Language, Socio-Cognitive and Prosodic Skills Evaluation
An ecologically valid way for clinicians and researchers to investigate not only children’s morphosyntactic abilities but also their socio-cognitive skills is by making use of narrative telling/retelling tasks (Gagarina et al., 2019; Lorusso et al., 2007; McCabe & Rollins, 1994; Tsimpli et al., 2016). Narratives can be assessed at the two levels of microstructure and macrostructure. Narrative microstructure refers to children’s morphosyntactic abilities by examining several aspects such as the MLU in words or morphemes, the number of main and dependent clauses, the lexico-grammatical features (e.g., connectives, prepositional phrases), and lexical diversity (e.g., Tsimpli et al., 2016). On the other hand, macrostructure focuses on the narrator’s ability to provide the listener with all the necessary information regarding the characters, the context, and the episodes of the story, as well as of the characters’ actions, reactions, and intentions with the use of the corresponding internal state terms (e.g., Gagarina et al., 2019; Lorusso et al., 2007; McCabe & Rollins, 1994; Tsimpli et al., 2016). The ability to take into consideration what the listener needs to know in order to understand the narration, combined with the ability to understand and present the characters’ internal states, which set the goals, trigger their attempts, and express their reactions, are indicative of the success of someone’s socio-cognitive skills. Therefore, it can be said that the analysis of children’s narrative tellings/retellings is likely to provide us with valuable information about their language and socio-cognitive skills.
To date, there is no published study exploring both the microstructure and the macrostructure level of the narratives produced by children with CAS. The present study aims to fill this gap by using a narrative retelling task. Examining the narrative abilities of children with CAS at both levels could give us a more holistic view of the disorder. Moreover, since the prosodic characteristics of an utterance have been found to affect individuals’ comprehension and phonological memory abilities (Martzoukou et al., 2020; Shintel et al., 2014), the narrations of the present study were pre-recorded by an actress, so that all participants would hear the stories produced in exactly the same way, while manipulations of the prosodic cues of intonation and timbre would enable us to address whether intensive prosodic cues can improve the comprehension and phonological memory abilities of children with CAS.
Research Questions and Hypotheses
In particular, the present study will attempt to provide answers to the following questions: (1) In comparison to typically developing (TD) children, do children with CAS have similar morphosyntactic skills? (2) Do children with CAS have similar socio-cognitive abilities to those of TD children as evaluated by their ability to (a) present the sub-components of the story (story structure), and (b) the main character’s internal states in order to effectively convey narratives? (3) Would the performance of the two groups (CAS vs. TD) on comprehension questions indicate a similar understanding of the narratives? (4) Would the use of intensive prosody have an impact on children’s phonological memory (as assessed through the structure of their narratives) and comprehension of the narratives (as evaluated through comprehension questions)?
Based on previous studies: (1) children with CAS are expected to perform worse in morphosyntax (microstructure) compared to how do the controls (Ekelman & Aram, 1983; Lewis et al., 2004; McNeill & Gillon, 2013; Murray et al., 2019)
(2) As far as socio-cognitive abilities (macrostructure) is concerned, the group with CAS is expected (2a) to produce narratives with poorer story structure compared to those of the typically developing, age-matched peers, due to their limited social interactions which might lead to a restricted perspective-taking ability. In other words, children with CAS might consider several pieces of information as already being known and so, unnecessary to be spoken (e.g., Kecskes, 2010; Tarshis et al., 2020). On the other hand, their speech difficulties might lead them to make shorter productions due to their feeling of embarrassment (Rusiewicz et al., 2018). Lastly, the difficulties they face with phonological working memory (e.g., Shriberg et al., 2012) might make them forget and not present all parts of the narratives. As for the use of internal state terms, based on the reported connection between speech impairments, limited social interactions and the understanding of others’ feelings, desires and intentions (Tarshis et al., 2020; Teverovsky et al., 2009; see also de Rosnay & Hughes, 2006), (2b) the group with CAS is expected to use fewer internal state terms than do the control group, as they would not be able to recognize the main character’s internal states.
Regarding the comprehension questions, (3) children with CAS are expected to perform similarly to those in the control group since their disorder does not seem to affect their language comprehension skills (APA, 2013; ASHA, 2007), and, thus, they are reported to have better receptive than expressive language skills (see also Murray et al., 2019)
Lastly, as for the prosodic manipulations, (4) intense intonation and timbre are expected to improve TD children’s phonological memory (as assessed through the story structure of their narratives) and comprehension skills (as evaluated through comprehension questions) (Shintel et al., 2014). In contrast, children with CAS are not expected to show such an improvement given that data from previous studies have reported difficulties in the comprehension of prosodic cues (Marion et al., 1993; Nijland et al., 2003).
Method
Participants
The authors contacted two private speech-language therapy centers that specialize in dealing with children with speech disorders, and informed clinicians about the purpose of the study and the characteristics of the children needed. Over a period of 2 year, 20 preschool children with CAS that did not result from a known neurological impairment, aged from around 4 to 6 year old, were identified as suitable for the present study. To ensure that those with sensory, hearing, cognitive, and neurological abnormalities were excluded, developmental-behavioral pediatricians, neurologists, and otolaryngologists had previously evaluated each child. Children were then referred to national Medical and Pedagogical Centers, certified by the Ministry of National Education, where they were diagnosed as having CAS, according to DSM-5 (APA, 2013), “Verbal Dyspraxia profile” (Jelm, 2001), “Screening test for the Developmental Apraxia of Speech” (Blakery, 1980) and the Greek version (Georgas et al., 1997) of WISC-III (Wechsler, 1992). From the total of 23 children addressed to us, two were excluded for the reason that they had already started elementary school by the time their assessment was completed and one had failed to complete all the experimental tasks. Thus, 20 children with CAS took part in the present study.
Apart from the children with CAS, 20 typically developing (TD) children matched with the clinical group on a one-to-one basis regarding age and gender were recruited for the present study through personal contacts with parents and parental groups (see Table 1). These TD children were reported as not having any sensory, hearing, or cognitive impairment or any neurobehavioral or genetic disorder.
Each Group’s Demographical Information and Performance on Speech Assessments Tests.
Note. CAS = children with apraxia of speech; TD = typically developing children; F = females; M = males, SD = standard deviation.
All children were monolingual, native speakers of standard Greek, and came from homogeneous, somewhat middle-class socioeconomic backgrounds. Following the principles of the Declaration of Helsinki, all of the children’s parents/caregivers were informed of the study’s objective and process, both orally and in writing, and they were ensured that their child’s personal data would be kept confidential. They were, then, asked to sign a consent form before their child could participate in the study. Participants were free to leave at any time without giving a reason. Stickers and drawing materials were given to all children as a reward for their participation.
Clinical Speech-Assessments
Both groups of children were also evaluated by the speech-language therapist in the author group of the present study, based on children’s speech and language medical history, orofacial and myofunctional evaluation, and ASHA (2007) criteria. Moreover, the speech abilities of all the children were evaluated by use of the following tests (Table 1):
(a) The “AnOmilo Test” (Panhellenic Association of Speech-Therapists, 2005; adaptation from Maeder & Roy, 2000) which is a screening tool for the detection of speech and language disorders in children aged from 3.9 to 4.6years old. The AnOmilo Test is divided into five domains: articulation of consonant clusters, with 16 as the maximum score and a score of ≥10 as a normal performance cut-off; picture description task, with a maximum score of 9 and a normal performance cut-off score of ≥7; repetition of three sentences with progressive difficulty and a score of 19 as the maximum score and ≥16 as the cut-off score; descriptive evaluation of speech flow (typical vs. not typical); and descriptive evaluation of voice quality (hoarse, whispering or typical vocalization).
(b) The “Apraxia characteristics checklist” (Crary, 1984.), which consists of 17 items within which the first three can be considered as describing the main criteria for the diagnosis of CAS and thus scored with 2 points each, whereas each of the rest of the 14 items is scored with 1 point. A total score above 5 is considered indicative of CAS.
(c) The “Diadochokinetic rate assessment” measures how quickly children can accurately repeat individual syllables (i.e., pa/ta/ka) or the combination of them (i.e., pataka) with the use of different parts of their mouth, namely front of the mouth (lips) for the pa, middle of the mouth (tip of the tongue) for the ta and back of the mouth (soft palate) for the ka. Children were asked to repeat each of the syllables or the combination of them for 1 minute. Their performance was measured in terms of iterations per second (see also Fiori et al., 2016).
(d) Lastly, the “Children’s Phonetic and Phonological Development Assessment” (Panhellenic Association of Speech-Therapists, 1995; adaptation from Grunwell, 1988), which descriptively evaluates the phonological system of children aged from 2.6 to 6 year, was administered to all participants. This test explores several phonological errors, such as consonant sequencing problems, metathesis, word inconsistency, difficulty with multisyllabic, and low intelligibility after repeat. The target words elicit all vowels in syllable initial and final positions, all consonants, and all diphthongs of the Modern Greek language.
Battery of Pre-tests
Before the main task of narrative assessment was administered, all children took a battery of tasks in order for their working memory and morpho-syntactic abilities, auditory short-term memory, vocabulary knowledge, and non-verbal cognitive abilities to be evaluated as to whether they were possible predictors of their narrative performance.
In particular, a Sentence Repetition test for preschool children (Stavrakaki & Tsimpli, 2000), was administered to all participants. According to Tsimpli et al. (2016), sentence repetition examines not only participants’ working memory but also their morpho-syntactic abilities, since repeating syntactic forms also requires syntactic knowledge to divide the sentence into smaller morphosyntactic constituents and to reconstruct them in a grammatical way. In order to make sure that all participants would listen to sentences produced in exactly the same way, all 16 sentences had been pre-recorded and produced by a trained female phonetician. During the administration of the test, each sentence was presented separately, and each child was instructed to verbally repeat it after its presentation. Three points were awarded for each sentence correctly repeated by the child, two points in cases where the child made one error, one point for two errors, and zero points in cases where the child made more than two errors. Phonology was not taken into account in the manner we coded children’s production. The maximal score was 48 points.
Auditory short-term memory was explored with the use of the Forward Digit Recall sub-test from the Athena Test (Paraskevopoulos et al., 1996). In this sub-test, every child was presented with a row of digits, starting with two and increasing to seven digits in the last block. The criterion for moving on to the next block was a correct recall of the sequence of digits in either the first or the second attempt. The procedure stopped if the child failed to correctly recall a digit sequence twice.
An Expressive Vocabulary Test (Vogindroukas et al., 2009; adaptation from Renfrew, 1997), which assesses children’s ability to name 50 items displayed one by one on a black and white picture, was administered to all participants. Words produced were considered correct if they could be recognized by the examiners as attempts of the children to produce the target name of the depicted object, even if the pronunciation contained serious phonological errors. Each correct word was given one point.
Lastly, the Raven’s Colored Progressive Matrices test (RCPM; Raven et al., 1990) was used to assess the children’s non-verbal cognitive abilities. RCPM is devised for children from 4 to 11 year old. It consists of 36 colored graphics which are incomplete in the bottom right end corner, and participants have to select the one that best completes the pattern from six alternatives.
Materials
The LITMUS-MAIN tool (Peristeri et al. in Gagarina et al., 2019) was used to assess story retelling. In particular, two of the four stories of the tool were selected, the Baby Birds story, and the Dog story. Each story has four main characters and consists of three episodes with each of them having (i) a goal statement for the protagonist, (ii) an attempt by the protagonist to reach the goal, (iii) an outcome of the attempt in terms of the goal, and (iv) internal states which initiate the events and express reactions. Each story is depicted through six colored pictures and, for the needs of the present study, was recorded as produced by an actress. In order for prosody comprehension to be explored, each story was recorded twice, once with “lively” and once with relatively “flat” prosody (see also Martzoukou et al., 2020).
In the “lively prosody” version, the recording had wide variations of intonation (baby birds’ story pitch range: 412 Hz, dog’s story pitch range: 427 Hz) and changes of the timbre, according to which character was talking each time (e.g., the cat looking at the birds, or the dog looking at the delicious sausages). In the “flat prosody” version, stories were produced with relatively small variations in the intonation pattern (baby birds’ story pitch range: 213 Hz, dog’s story pitch range: 207 Hz) and no changes of the timbre. The duration of the stories was almost the same for each of them in both conditions so as to allow children to have the time to process the auditory information.
Both stories were presented to each child, but no child listened to the same story twice and no child listened to the two stories in the same prosody condition. In particular, for every five children in each group, the following order/prosody condition combinations were created:
(a) Little birds-story with lively prosody—Dog-story with flat prosody
(b) Little birds-story with flat prosody—Dog-story with lively prosody
(c) Dog-story with flat prosody—Little birds-story with lively prosody
(d) Dog-story with lively prosody—Little birds-story with flat prosody
Lastly, children’s comprehension skills were also assessed through the use of the ten comprehension questions per story which required an open-ended response (LITMUS-MAIN tool; Gagarina et al., 2019).
Procedure
Each child was examined individually in a quiet room at the speech therapy center or at their home. The child was shown three colored envelopes on the computer screen and was informed that each of them contained a different story. The child was then asked to choose one, creating the (false) impression that they made the choice, when in fact the order and the content of the story were pre-arranged by the researchers. Then the child listened to the story through headphones, while viewing two pictures per slide on the computer screen. After listening to the entire story, the child was shown all six pictures on a single slide on the computer screen and was asked to re-narrate the story to the examiner who was not present in the testing room. The examiner reminded the child that she (the examiner) did not know the story, ensuring that the child did not assume any prior knowledge on the examiner’s part. After the child’s retelling of the story, she asked each child a set of comprehension questions by pointing each time to the relevant picture or pictures (for details of the procedure see also Martzoukou et al., 2020). The experiment was divided into two sessions on different days; one for each story. An OLYMPUS digital voice recorder (VN-8500pc) was used to record all experimental sessions.
Data Analysis
For the microstructure analysis, the following measures were used: (1) diversity of content words, namely the ratio of different word stems to the total number of content-word tokens, (2) diversity of function words, namely the ratio of different function words divided by the total number of function words, (3) MLU in words (and not in morphemes, since Greek is a highly inflected language); utterance segmentation was based on pauses between phrases, (4) subordination index that is, number of subordinate sentences divided by the total number of sentences and (5) number of adverbial, (6) relative and (7) verb-complement clauses in each child’s narratives (Tsimpli et al., 2016).
Regarding the macrostructure analysis, MAIN (Gagarina et al., 2019) conventions were used. In particular, each narrative is assessed based on the following six sub-components: (1) setting component (reference of time and place), (2) internal state terms as initiating event, (3) goal, (4) attempt, (5) outcome, and (6) internal state terms as a reaction. The last five components occur three times per story. The first component is awarded 2 points (1 for time and 1 for place) and 1 point is awarded for each of the remaining components, leading to a maximum score of 17 points. Internal state terms, functioning as initiating events or as reactions to events or actions were calculated based on the, in total, six relevant components (two components × three times per story) of the MAIN conventions.
Two of the authors transcribed and analyzed the story retellings of the children. The percentage of agreement between the two coders was about 95%, whereas in the remaining cases, disagreements were viewed, and modifications were made as needed. As for the comprehension questions, one point was awarded for every correct answer to each story’s three questions about goal statements, five questions about the internal state of the characters, and two questions aiming at eliciting children’s predictions about the internal state of one of the main characters of the story. Thus, the maximum accuracy score was 10.
All statistical calculations were performed using JASP, an open-source, statistical software (University of Amsterdam). Normality was assessed using the Shapiro–Wilk test. Differences between the TG and the CAS were estimated via Independent T-test in the case of nominal distribution, and with the Mann-Whitney U test in the case of non-normal variables. In the case of the speech assessments, though, multiple tests showed a statistically significant difference between the two groups and, thus, MANOVA was used in order for the effect of the independent variable (Group) to be explored.
To evaluate the effect of prosody on phonological memory and comprehension, as explored via Story structure and Comprehension questions respectively, comparisons of the paired mean difference of the two prosody conditions (flat vs. lively) in each group were performed with the use of Paired Sample T-tests, in case of normal distribution, or Wilcoxon signed rank-test, in case of non-normal distribution. For the effect sizes, rank biserial correlation (r) was used for non-parametric tests and the Cohen’s d (|d|) for parametric tests. The level of significance was set at .05 for all the analyses.
Results
Demographical and Clinical Characteristics
MANOVA analyses revealed significant differences between the two groups in all tests and subtests, and a significant effect for the independent variable Groups [Wilks’ Lambda < 0.001, F = 4201.167, p < .001] (see Table 1).
Battery of Pre-tests
Results revealed that the two groups performed statistically differently only on the Sentence Repetition Task (U = 116.500, p = .024, r = −.417) (see Table 2).
Each Group’s Mean Scores (SDs) and Score Range on Pre-tests.
Note. CAS = children with apraxia of speech; TD = typically developing children; SRT = sentence repetition test; FDR = forward digit recall; EVT = expressive vocabulary test; RCPM = Raven’s colored progressive matrices; max. = maximum; SD = standard deviation.
Mann-Whitney U test. bIndependent T-test.
Morphosyntax (Microstructure)
Comparisons between the two groups’ performance on morphosyntax revealed significant differences only on the number of Complement clauses used by each group [U = 599.000, p = .44, r = −.251] (see also Table 3).
Each Group’s Mean Scores (SDs) and Score Range on Microstructure Measures.
Note. CAS = children with apraxia of speech; TD = typically developing children; SD = standard deviation.
Mann-Whitney U test.
Independent T-test.
Socio-Cognitive Abilities (Macrostructure)
Between group comparisons demonstrated significant differences in the use of internal state terms in the story structure [U = 693.500, p = .300, r = −.133], but not in the story structure of children’s narrative retellings (see also Table 4).
Each Group’s Mean Scores (SDs) and Score Range on Macrostructure Measures and Comprehension Questions (Mann-Whitney U test).
Note. CAS = children with apraxia of speech; TD = typically developing children; IST = internal state terms; max. = maximum; SD = standard deviation.
Comprehension Questions
Results revealed significant differences between the two groups in the accuracy in Comprehension questions [U = 591.500, p = .042, r = −.261] (see also Table 4).
Since children with CAS used significantly fewer internal state terms in story structure, and since seven out of 10 Comprehension questions aim at eliciting internal states of story characters, further analysis was conducted to explore whether the poor performance in Comprehension questions of children with CAS was due to their poor comprehension of characters’ internal states. In particular, children’s performance in the seven Comprehension questions eliciting internal states was measured and compared separately from the rest of the questions. Results revealed that the difference between the two groups’ performance was statistically significant [U = 567.500, p = .022, r = −.291.
The Role of Prosody
Since prosody has been found to affect both comprehension and phonological memory (e.g., Martzoukou et al., 2020; Shintel et al., 2014), we wanted to explore whether prosody affected participants’ story structure skills and accuracy in comprehension questions.
Between group comparisons, using Mann-Whitney Tests, showed that TD children’s performance was significantly better than that of children with CAS when prosody was lively in both Story structure [U = 111.000, p = .015, r = −.445], and Comprehension questions [U = 118.000, p = .024, r = −.410], but when prosody was flat, no differences between the two groups were found [Story structure: p = .300; Comprehension questions: p = .576] (see also Table 5).
Each Group’s Mean Scores (SDs) and Range of Accuracy with Lively and Flat Prosody.
Moreover, within group comparisons revealed that TD children’s performance on Story Structure was significantly better when the prosody was lively compared to their performance when the prosody was flat [t = 3.715, p = .001, /d/ = 0.831]. No difference was found, though, as far as their performance on Comprehension questions is concerned (p = .353). As for children with CAS, their performance did not differ between the two conditions of prosody either on Story Structure (p = .358) or Comprehension questions (p = .617) (see also Table 5).
Discussion
The aim of the present study was to add to the, to date, scarce research findings regarding children with CAS and to offer a clearer and more complete view of their language and socio-cognitive abilities. To this end, a narrative retelling task accompanied by comprehension questions was presented to 20 children with CAS who were from around 4 to 6 year old, and 20 TD children matched for gender and age. An additional prosodic manipulation enabled us to address whether or not lively prosody can enhance participants’ phonological memory as assessed through the structure of their narratives) and comprehension of the narratives (as evaluated through comprehension questions). Therefore, participants’ microstructure, macrostructure, and prosody comprehension abilities were evaluated.
Battery of Pre-tests
Before the main experiment, a battery of assessments consisting of a Sentence Repetition test (Stavrakaki & &Tsimpli, 2000), a Forward Digit Recall sub-test (Paraskevopoulos et al., 1996), an Expressive vocabulary test (Vogindroukas et al., 2009), and the Raven’s Colored Progressive Matrices (Raven et al., 1990), a test exploring children’s non-verbal cognitive abilities, was administered to all participants. Results revealed similar performance between the two groups in the Raven’s Colored Progressive Matrices, but lower scores for children with CAS in the rest of the assessments compared to the controls, although the difference between the two groups reached significance only in the case of the Sentence Repetition test. As Tsimpli et al. (2016) suggest, Sentence Repetition tests mainly examine participants’ morpho-syntactic abilities since they have to divide and reconstruct in their memory well-formed, syntactic chunks. Poor performance of children with CAS in some aspects of morphosyntax has been previously reported (Ekelman & Aram, 1983; Lewis et al., 2004; McNeill & Gillon, 2013; Murray et al., 2019). We refrain however from attributing our results to a pure morpho-syntactic deficit since the Sentence Repetition test also demands phonological and memory processing, which have been also reported to be poor in children with CAS (e.g., Shriberg et al., 2012). Further support for this caution comes from the results of ANCOVAs which revealed that Sentence Repetition scores did not affect performance on any microstructural variable.
Morphosyntax (Microsturucture)
Regarding the findings of the main experiment, a significant difference between children with CAS and TD children was found only in the use of Complement clauses. Therefore, our first hypothesis [hypothesis (1)], according to which children with CAS were expected to perform worse in morphosyntax, was only partially confirmed. By examining each variable separately, we can see that no statistically significant differences were found between the two groups, as far as which the use of function words is concerned. Results from previous studies are inconsistent since Ekelman and Aram (1983) and Murray et al. (2019) reported mainly pronoun selection errors in children with CAS, whereas variability among children has been also reported (McNeill & Gillon, 2013; Murray et al., 2019). If we assume that the difficulties of children with CAS in using function words are more evident in the use of pronouns, then a reason this difficulty was not obvious in the present study might be because Greek is a highly inflectional language with morphological marking for grammatical person on the verb which makes the use of subject pronouns optional (pro-drop language).
Moreover, the study of language development has shown that children start producing function words and subordinate clauses at around the age of 3, while development is not yet complete before the age of 6 (e.g., Hoff, 2013). The use of function words and complex sentences (subordinate structures) leads to higher MLU scores. In the present study, the children were younger than 6 year old at the time of testing and their performance on microstructure variables was low for both groups (CAS &. TD). Therefore, it is likely that their age is too early an age for great morpho-syntactic differences between the two groups to be found in the variables examined in the present study, and that if children were to be tested later in childhood these differences could be more pronounced. This might then be the reason why MLU scores as well as the ratio of subordinate structures and the use of adverbial, and relative clauses were similar between the two groups.
As for verb-complement clauses, they are selected as complements of a verb in the higher clause. Mental verbs (such as think, believe, wonder, realize, and remember) which together with emotional terms are regarded as internal state terms, need a complement clause. This might suggest that children with CAS lack the ability to articulate other people’s feelings, beliefs, and thoughts, as they have not mastered yet the use of complement structures which permit the representation of embedded sentences about internal states. Several studies have reported a connection between the use of complement clauses and the use of internal state terms (e.g., Martzoukou et al., 2020; Schick et al., 2007; Villiers et al., 2009). Thus, examining the performance of children with CAS on the use of internal state terms can only be of great interest.
Socio-Cognitive Abilities (Macrostructure)
Regarding the children’s performance in macrostructure variables and especially those in story structure, our hypothesis [hypothesis (2a)] was that children with CAS narrative retellings would be poorer, compared to those produced by the TD group. This hypothesis was not confirmed as no statistical differences between the two groups were found. In particular, children with CAS were expected to produce less informative retellings due to their limited perspective-taking ability (Kecskes, 2010; Tarshis et al., 2020). The development of perspective-taking ability in children typically starts to emerge in the preschool years and continues to develop throughout childhood and adolescence (Tarshis et al., 2020). The results revealed poor performance for both groups, thus, again their ages might be too young for perspective-taking differences between the two groups to be found. Moreover, the feeling of embarrassment created in children with CAS due to their speech impairments (Rusiewicz et al., 2018) was suggested as being another reason for their expected less informative productions when compared to the ones produced by the TD children. We should keep in mind, though, that the connection between speech impairment and embarrassment was extracted by parents’/caregivers’ reports and not by children’s own reports. Thus, the “lack of experience hypothesis” suggested by Yovetich et al. (2000), according to which young children do not have enough exposure to verbal experiences for their self-esteem to be affected by their speech impairments could also account for the results of the present study. Lastly, the difficulties children with CAS have been reported to have with phonological memory (e.g., Shriberg et al., 2012) in combination with the poor performance demonstrated by children with CAS of the present study on the Sentence Repetition test (phonological working memory) could have accounted for the producing of narratives with poor story structure. If we take a closer look at the procedure of the present study, however, we note that children did not have to rely only on phonological (working) memory, but they could support their narrative retellings through the use of the pictures since, during their story-retellings, they were presented with the six pictures of the story placed in the correct order. Indeed, during the experimental procedure, it became clear to us that children with CAS, in particular, found this feature really useful and supportive. This could be the reason why the performance on the Sentence Repetition test did not affect participants’ performance in story structure.
One of the most noteworthy findings of our study is that children with CAS performed poorly in the use of internal state terms in the story structure, and, thus, the second part of hypothesis (2) was confirmed [hypothesis (2b)]. To our knowledge, this variable has not been explored before in children with CAS, although research has shown social communication difficulties in such children (Teverovsky et al., 2009; see also Tarshis et al., 2020) and an association between social interactions and socio-cognitive abilities (de Rosnay & Hughes, 2006). Thus, it seems that children with CAS failed to recognize the character’s internal state. On the other hand, though, it could be suggested, that, as we saw before, children with CAS have not mastered yet the use of complement structures that permit the representation of embedded sentences about internal states, and, thus, they lack the ability to present other people’s feelings, beliefs, and thoughts. Similarly, one could say that motor planning difficulties could account for the fact that children with CAS restricted their narratives to the presentation of the core information of the stories (e.g., reference of time and place, goals, attempts, and outcomes) and excluded less important information about the evolution of the story, such as the feelings of the characters before and after the completion of the episodes. These proposals, though, cannot explain their performance in the next variable tested, namely that of Comprehension questions.
Comprehension Questions
Children with CAS were expected to have similar accuracy scores to the controls in the Comprehension questions [hypothesis (3)] since the disorder seems to affect their production and not their comprehension skills (APA, 2013; ASHA, 2007; Murray et al., 2019). This hypothesis, however, was not confirmed by the findings. Children with CAS performed worse than did the control group. A closer look, though, revealed that seven out of 10 comprehension questions were about internal states. Thus, if we take into consideration, that children with CAS used fewer internal state terms in their narrative retellings and that most comprehension questions were about internal states, an interesting pattern is revealed. More specifically, it appears that children with CAS face problems in recognizing the emotions of the characters in the story and this is the probable reason they neither use internal state terms as much as do TD children; nor are able to reconstruct them in response to Comprehension questions. Consequently, it seems that narrative retellings of children with CAS did not contain enough complement clauses, not because of their poor morphosyntactic abilities that did not allow them to find the proper means to express the main character’s internal states, but on account of their not having an ability to recognize them.
The Role of Prosody
Regarding our final hypothesis, according to which children with CAS were not expected to benefit from the employment of intense intonation and timbre [hypothesis (4)], the findings confirm the hypothesis. In contrast to TD children, children with CAS performed worse in Story structure and Comprehension questions when prosody was lively, and in fact, their performance was slightly worse in this condition compared to when prosody was flat. On the contrary, TD children’s performance was better when prosody was lively than when it was flat, and this difference gained significance in the Story Structure measures. Therefore, it seems that the difficulties that children with CAS face in comprehending prosodic cues, such as duration and rhythm (Marion et al., 1993; Nijland et al., 2003), also extend to the recognition of intonation and timbre. Consequently, the introduction of prosodic manipulations cannot be of use when the improving of poor comprehension and phonological memory is an aim (Marquardt et al., 2002, 2004; McNeill et al., 2009; Shriberg et al., 2012) of children with CAS.
Conclusion
The aim of the present study was to investigate the microstructure (morphosyntax), macrostructure, and prosody comprehension skills of children with CAS in order to tease out the effects of the deficit on different aspects of language and socio-cognitive ability. Results revealed problems regarding the expression and comprehension of characters’ internal states, which led such children to a restricted use of verb-complement clauses, as well as difficulties with the comprehension of prosody. These results conform with McNeill and Gillon’s (2013) suggestion that morphosyntactic difficulties in these children are not the core characteristic of the disorder, but the result of other difficulties. Nevertheless, there is a need for more research on morphosyntactic abilities with older children with CAS as it could be the case that the lack of differences between the two groups in the rest of the morphosyntactic variables explored in our study was due to their similarly low performance which in turn may be associated with all the participants’ young age.
As for socio-cognitive abilities, there appears to be a common cause or a link between the poor performance of such children in the use of internal state terms and their difficulties in Comprehension questions, which were located in their type of comprehending of the main characters’ internal states. This assumption though is in need of further exploration.
Regarding prosody comprehension, the results demonstrated that prosodic manipulations cannot improve the cognitive abilities of children with CAS (phonological memory and passage comprehension). Moreover, if we consider recognition of prosody as well as part of the socio-cognitive ability, in that prosody helps individuals understand from voice other people’s emotional states, then it could be claimed that children with CAS generally face a more difficulty with the recognition of the socio-cognitive related aspects of language use. In order to reach sound conclusions regarding these children’s socio-cognitive ability, though, further research must be conducted. Lastly, our findings regarding prosody might suggest a connection between prosody production, according to the criteria for CAS, and prosody comprehension in children with CAS.
Therefore, while the present study does add to the existing literature, the research in it raises new issues and questions regarding the characteristics of CAS and the features that need to be considered in developing when designing interventions that would help children with this specific impairment.
Footnotes
Acknowledgements
The authors thank the children and their parents for participating in this study
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.
Ethics Statement
When the study was conducted the Department of Speech and Language therapy was part of the Technical Educational Institute of Epirus. The Educational Institute of Epirus did not have an ethics committee to evaluate and approve research proposals.
We can affirm, however, that we followed the ethical guidelines set forth both by the Educational Institute and by the Declaration of Helsinki regarding the ethical treatment of human participants in research. Thus, all children’s parents/caregivers were informed about the purpose and the procedure of the study, both in oral and written form and they were ensured that their child’s personal data would be kept confidential. They were, then, asked to complete written consent forms before their child’s participation in the study. Participants had the right to withdraw at any point without having to provide a reason. Stickers and drawing materials were given to all children as a reward for their participation
Data Availability Statement
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.