Growing interest is mounting on the cognitive characteristics of children with developmental language disorders (DLDs). This diagnostic label has been recently introduced to replace previous ones such as specific or primary language impairments. It refers to children who have language disorders not associated with biomedical conditions such as brain injury, neurodegenerative diseases, cerebral palsy, or other conditions linked to genetic or neurological causes [1
]. Crucially, it is now well-accepted that such linguistic difficulties may co-occur with a variety of other cognitive impairments affecting, for example, procedural memory [2
], motor control [3
], and executive functions (EFs; [4
]). Because of their relevant role in cognitive and linguistic development, the study of EFs in DLDs may be particularly important for an accurate interpretation of the communicative and language difficulties observed in these children.
According to Miyake and colleagues [7
], EFs are set of cognitive skills (moderately correlated components in a confirmatory factor analysis) necessary to (1) update the ongoing information in working memory while monitoring the execution of a task (updating), (2) inhibit unnecessary or non-pertinent pieces of information (inhibition), and (3) efficiently shift between mental sets (shifting). More recently, Miyake and Friedman [8
] proposed a unity/diversity framework for the organization of EFs. According to their model, these three components would share common underlying processes (unity, i.e., what these components have in common) but might depend on component-specific processes (diversity). For example, the ability to maintain the focus on the aim and execution of a task might significantly rely on the joint contributions of the inhibitory (e.g., allowing a speaker to inhibit the production of irrelevant or tangential information while producing a discourse), shifting (needed to flexibly shift between topics and/or perspectives), and updating (implicated in the ability to link different pieces of information exchanged during a conversation) components of EFs. Garon and colleagues [9
] suggest that such components develop hierarchically in children. Namely, the development of EFs may take place only after the development of sustained attention, i.e., the ability to focus on a specific target for prolonged time periods. As a second step, children would begin to develop the ability to keep the information active in their working memory (i.e., updating). Afterwards, inhibition and eventually shifting would begin to develop. Crucially, later abilities appear to rely on previously acquired skills and may significantly affect complex behaviors such as those required in efficient, communicative exchanges. From this point of view, a task requiring children to shift between mental sets should require the ability to inhibit irrelevant stimuli, keep the relevant ones in working memory, and focus on the task as long as it is needed. This also means that difficulties in the development of earlier skills (e.g., attention skills) might negatively impact on the development of later abilities, such as working memory, inhibition, and/or shifting.
Growing evidence suggests that children with DLDs may have difficulties in maintaining sustained attention on verbal [11
] and non-verbal tasks [14
] in phonological working memory, which is crucial for the updating component of EFs [16
], and on tasks assessing inhibitory control and/or cognitive flexibility [4
]. This apparently also applies to preschoolers with DLDs (e.g., [18
]) and supports the possibility of a domain-general executive function deficit in these children. An interesting issue regards the possible effects of difficulties in EFs on the linguistic and communicative abilities of children with DLD. Message production and comprehension are likely the result of complex multistage processes (e.g., [21
]) where the three executive components might play relevant roles. Indeed, according to Miyake and colleagues [7
] and Mozeiko et al. [23
may be involved in the generation of complete episodes within a narrative discourse, in the selection of informative words, and in the ability to monitor the communicative flow; updating
may be required to generate and understand sentences as well as recall former episodes or episodic contents for an accurate organization of a story; inhibition
may be critical for monitoring the production of extraneous comments and derailments while telling a story and for the ability to inhibit the semantic competitors while producing or understanding words. The current investigation aims at exploring the potential relation between difficulties in two of the executive components described in Miyake and colleagues [7
] (i.e., updating and inhibition) and the linguistic and narrative skills of a cohort of preschoolers with a diagnosis of DLD. Namely, a group of preschoolers with DLDs and one of children with typical development were administered tasks assessing their abilities to (1) keep verbal information in their phonological short-term and working memory while completing a task (updating), and (2) inhibit the production of prepotent responses under specific conditions. Furthermore, they also received tasks assessing their linguistic and narrative abilities. We hypothesize that (1) the participants with DLD would perform worse than controls on tasks assessing inhibition and updating, (2) they would have significant difficulties on tasks assessing production and comprehension, and (3) their performance on the tasks assessing inhibition and updating would be related to their linguistic and narrative difficulties.
The current investigation explores executive, linguistic and narrative skills in children with DLDs and TD and the potential relation between measures of updating and inhibition and the abilities to produce and discriminate phonemes, words, and sentences, as well as the ability to organize a narrative discourse at the macrolinguistic level (i.e., global coherence) and utter words that are appropriate from both linguistic and communicative points of view (i.e., lexical informativeness). Overall, the cohort of children with DLDs scored lower than controls on tasks assessing updating (i.e., phonological short-term and working memory) and inhibitory skills (confirmation of Hypothesis 1). As for language, they showed a complex pattern of results with some impaired abilities (e.g., articulation, phonological discrimination, grammatical comprehension and production, and the ability to produce appropriate words on the narrative production task) and other ones where they did not score differently than controls (e.g., lexical production and comprehension skills, and the production of errors of global comprehension while telling a story) (partial confirmation of Hypothesis 2). Finally, updating and inhibitory skills were found related to the difficulties observed in the participants with DLDs (confirmation of Hypothesis 3). These results will be discussed in light of current models of message processing.
From a cognitive point of view, this study confirms previous investigations highlighting in children with DLD potential difficulties in inhibition [18
] and phonological working memory, a cognitive system that plays a crucial role in the updating process (e.g., [17
]). According to Baddeley [38
], working memory allows individuals to momentarily store and process visual and/or verbal information through different components. Namely, verbal information is assumed to be processed in a phonological working memory system formed by a phonological short-term memory system, where incoming verbal information is momentarily stored, and a subvocal rehearsal process that keeps this information active until needed. Participants with DLDs performed worse than controls on both forward and backward digit recall tasks, suggesting a difficulty not only in the passive storage of verbal information in working memory (as investigated via the forward digit recall task) but also in the active manipulation of such information (as investigated via the backward digit recall task). This supports the possibility that reduced phonological working memory skills are a characteristic feature of DLDs. Indeed, a difficulty in keeping track of linguistic information in short-term memory and eventually process it in working memory might result in slowed vocabulary acquisition (e.g., [39
]) and affect the linguistic development of children with DLDs [35
]. In line with this hypothesis, phonological short-term and working memory limitations correlated with a range of linguistic measures in the current study. Similarly, the participants with DLDs produced significantly more errors than controls on the two subtests of the inhibition task of the NEPSY-II. It is noteworthy, however, that their performance on such tests was within normal range, as shown by their mean scalar scores (see Table 3
). On the naming subtest, 69% of them scored above the 50th centile while none of them scored below the 25th centile; on the inhibition subtest, their performance was more variegated with 44% of them scoring at or above the 50th centile, and 43% of them at or below the 25th centile (8% among children with TD). Altogether, these data suggest a relative weakness in inhibitory control in the participants with DLDs, which is not necessarily pathological. Such variability may be among the causes of the heterogeneous findings reported in the literature on inhibition in children with DLDs, who may present with difficulties in other components of EF (e.g., in [20
] where a cohort of twenty-two preschoolers with DLD showed difficulties on both verbal and non-verbal tasks assessing updating and shifting but not on tasks assessing inhibition).
From a linguistic point of view, as a group, the preschoolers with DLDs showed a range of linguistic difficulties affecting both production and comprehension (see also [1
]). In line with several previous reports, such difficulties involved mostly phonological and syntactic skills [42
]. A careful evaluation of our results suggests that the children with DLDs did not have articulatory difficulties. Indeed, in the expressive tasks, they did not show articulatory impairments while producing words. On the contrary, their errors were mostly characterized by phonological substitutions (i.e., phonological paraphasias). Together with their significant difficulties in the phonological discrimination task, this rules out the possibility that their difficulties in dealing with the Articulation task were phonetic. Rather they were likely due to a difficulty in phonological processing, i.e., in the ability to categorize phones in abstract phonemic categories. Phonological problems were accompanied by difficulties in both expressive and receptive syntax [45
] but not by difficulties in naming or lexical comprehension. Indeed, morphology and syntax are usually more affected than vocabulary in these children [47
]. Their utterances were characterized by reduced grammatical accuracy and omissions of both content and function words that made their sentences incomplete [47
]. Interestingly, even in absence of naming difficulties and more general impairments in the ability to generate a mental model of the story to describe (as shown by the lack of group-related differences in the production of errors of global coherence), the participants with DLDs produced significantly fewer words that were scored as informative during the narrative production task (see also [17
]). This is a particularly interesting result as it suggests that, even if they were able to correctly select the target words in a decontextualized setting (as it is the case for naming tasks), they could not efficiently select words within a given narrative context. In other terms, they had difficulties in using contextual elements in the process of lexical selection (see below).
We would also like to stress the heterogeneity of the linguistic performance of the children with DLDs. Indeed, as can be seen in Table 4
, Table 5
, Table 6
and Table 7
, the performance of these participants is quite variable. Even if, as a group, they scored lower than controls on measures assessing phonology (articulation and phonological discrimination), syntax (% complete sentences and grammatical comprehension), and the ability to produce contextually adequate words (% lexical informativeness), in all of these cases, there were individuals scoring up to 2 SDs below the expected mean, as well as children performing 1 or even 2 SDs above the mean. At the same time, significant within-group variations were also found on measures tapping lexical skills (naming and lexical comprehension) with children scoring 2 SDs below the mean. This further confirms the need for a comprehensive assessment of language skills in these children in order to derive a clear picture of their individual relative strengths and weaknesses. These will eventually be necessary for an adequate description of their linguistic profile and for the generation of effective rehabilitation programs.
A final issue regards the need to interpret the linguistic difficulties described so far within a theoretical framework of message production and comprehension. According to some of the most influential models of language processing, the production and comprehension of messages and/or narratives are the result of multistage processes characterized by an interplay between cognitive and linguistic skills (e.g., [22
]). For example, according to the Structure Building Framework (SBF; [22
]), in order to produce a story, a speaker needs to generate a structure or mental depiction of its contents that will serve as a foundation for its development. In this preliminary phase, the ability to focus the attention to the goal at hand and to inhibit potentially distracting actions is crucial. Once the story structure has been generated, the speaker will need to organize it in sequences that must be converted in propositions and eventually verbalized through processes of lexical selection, access, and production (e.g., [21
]). During these phases, the possibility to inhibit the generation of irrelevant (e.g., tangential) propositions and the selection of wrong words, as well as the ability to keep all this information active in working memory, are crucial. As the information flows, the speaker also needs to monitor its consistency with the previously generated structure(s) and, in case of inconsistency, shift and generate a new structure. Additionally, message comprehension is likely the result of a multistage process that entails the interaction between bottom-up and top-down processes (e.g., [52
]). After decoding the phones uttered by the interlocutor, the listener needs to map the perceived frequencies onto phonemic categories, select the target words, and get access to all of their information that will guide the generation of the meaning of the perceived sentence. The organization of such meanings will eventually trigger the generation of structures and mental depictions of the perceived messages. Furthermore, the comprehension of a message is the result of a complex interplay between cognitive (e.g., EF and attention) and linguistic skills. The cohort of children with DLDs had difficulties in the generation of informative words that, as stated earlier in this discussion, were not the consequence of lexical selection/access difficulties. Indeed, their performance on both naming and lexical comprehension tasks (that directly assess lexical selection/access abilities) was similar to that of the participants with typical linguistic development. It is then likely that the reduced levels of lexical informativeness found in the narratives by preschoolers with DLDs are related to other factors.
Coherently with the above-mentioned models of message production, in our study, the % of lexical informativeness showed a moderate positive correlation with the composite score of digit recall and a large negative correlation with the production errors in the inhibition task. Considering that these two measures of EF shared 10% and 30% of variance with that measures of linguistic efficiency, our findings support the possibility that the ability to select contextually appropriate words during the production of a narrative may be related to the abilities to keep the already generated structures active in working memory (updating) while inhibiting irrelevant and/or inadequate pieces of information. Interestingly, growing experimental evidence suggests that the ability to actively suppress irrelevant, conflicting information that otherwise could cause interference emerges functionally intact in early childhood [53
]. Therefore, our data support the possibility of difficulty in developing a function that should be in place in children aged 4 to 5 years.
The participants with DLDs also had difficulties in later stages of message production. Indeed, even in absence of relevant difficulties of lexical selection, they had significant impairments in the production of complete sentences on the narrative production task and in the articulation of the selected words on the articulation and naming tasks. These findings are coherent with previous investigations (e.g., [47
]) and, interestingly, also in these cases, measures of EF showed a significant correlation. Namely, the production of errors on Part B of the inhibition task showed a medium negative correlation with the % of complete sentences. With 22% of shared variance, this correlation suggests that lower inhibitory skills (i.e., the production of more errors on the inhibition task) might significantly affect the process of sentence generation during storytelling. Qualitative inspection of the transcripts confirmed that the majority of the sentences produced by children with DLDs were not complete as they were abruptly interrupted. Usually, these participants searched for the appropriate words that could be inserted in the subsequent utterances. Therefore, reduced efficiency in the ability to suppress irrelevant information might play a role while generating grammatical structures. Finally, a significant medium positive correlation was found between the Composite Digit Span Recall score and the measure of articulation with 17% of shared variance highlighting the possibility of a relation between updating and the ability to correctly articulate words. Nonetheless, significant correlations were also found between measures of updating and inhibition and measures of linguistic comprehension. The ability to identify minimal pairs had a moderate positive correlation with the Composite Digit Span Recall score with 16% of shared variance and a strong negative correlation with the production of errors in Part B of the inhibition task. In this case, the two variables shared 59% of the variance. Therefore, the process of phonological discrimination is also likely affected by EF. In the same vein, grammatical comprehension showed a significant positive correlation with the Composite Digit Span Recall score, with 38% of shared variance and a strong negative correlation with the production of errors on Part B of the inhibition task with 49% of shared variance. Overall, then, these correlations highlight the complexity of linguistic development and processing in children with DLDs.