Enhancing Form–Meaning Connections in the Language Teaching of Children with Developmental Language Disorder: Evidence from Two Teaching Interventions

Abstract

Focus on form (FonF) teaching interventions have been widely employed to help second language learners notice a target grammatical form while their attention is on meaning, i.e., establishing new form–meaning connections. These interventions can be input-based, focusing on the processing of input (i.e., Processing Instruction), or output-based, focusing on production within communicative activities (i.e., Dictogloss). The current pilot study explored whether such teaching interventions would be beneficial for children with DLD. The study employed Processing Instruction and Dictogloss for the teaching of passives in two groups of Greek school-aged children with DLD. The study applied pre-tests and (delayed) post-tests to explore (a) the potential (long-term) effectiveness of the interventions, and (b) potential differences in their effectiveness within this population. The findings indicate that both Processing Instruction and Dictogloss can be promising interventions for Greek children with DLD since they both led to learning gains and retention two weeks after the interventions across all tasks (comprehension, production, and sentence repetition). However, Dictogloss was more effective in production, while Processing Instruction in sentence repetition (when accuracy scores are measured).

1. Introduction

Current research on language teaching techniques in children with DLD has mostly focused on therapeutic contexts. These techniques often require the explicit teaching and learning of a new symbolic system of rules (i.e., colors/shapes representing different grammatical items and syntactic relations within the sentence), extensive training, equipment, and several weeks of individual treatment. However, less is known about the potential applicability and effectiveness of certain language intervention techniques that could be relatively easily implemented not only in therapeutic but also in classroom settings. Drawing from the field of Second Language Acquisition (SLA) and classroom-based research, the present study employs two Focus on Form (FonF) language teaching intervention techniques in children with DLD: Processing Instruction (PI) and Dictogloss. These FonF techniques aim to enhance the way learners establish new form–meaning connections by “forcing” them to notice the target grammatical form while their attention is focused on meaning. Such techniques have only recently been explored in atypical acquisition (Barcroft et al., 2025; Santamaria et al., 2013 with promising results. This is the first study that provides preliminary but also crucial findings for such intervention techniques in DLD. The present grammatical interventions target passives, a notoriously vulnerable grammatical phenomenon in atypical language acquisition. The two teaching intervention techniques are compared to explore their potential effectiveness for the target population.

To date, several grammatical interventions have been successfully employed to teach passives in children with DLD (see below). However, most of these methods have been mainly employed in English. Currently, there are no studies on grammatical interventions in Greek children with DLD. This pilot study addresses this gap by creating enhanced opportunities for children to notice grammatical forms in language input and/or output and to discover their associated meaning.

Given their pedagogical nature, these interventions are relevant to various stakeholders in education such as (special education and/or language) teachers, language consultants, and specialists as well as clinicians and speech and language therapists (SLTs).

The next section reviews the existing intervention studies for passives in children with DLD. Section 1.2 presents the interventions of PI and Dictogloss and their potential added value in the field of DLD followed by the present study and its results (Section 2 and Section 3, respectively) as well as the discussion of the present findings (Section 4) and conclusion (Section 5).

1.1. Passives in Typical and Atypical Language Acquisition

Passives are a challenging grammatical structure during acquisition both in typically developing children as well as in children with Developmental Language Disorder (DLD) across various languages. Typically developing monolingual children seem to acquire passives relatively late (around age 5; 0) (see Armon-Lotem et al., 2015 and references therein), while certain minor difficulties may persist even at the age of 6; 0 (Marinis & Saddy, 2013 and references therein). In typical language development, these difficulties are expected to mostly recede with literacy development in the first two years of primary school (Fotiadou & Tsimpli, 2010; Paspali et al., 2024a). However, children with Developmental Language Disorder (DLD), who are known to exhibit difficulties in the domain of morphosyntax and its interaction with semantics, pragmatics, and phonology (e.g., Adani et al., 2016; Andreou et al., 2022; Dosi & Gavriilidou, 2020; Leonard, 2014; Poll et al., 2010; Papadopoulou et al., 2009; Rice & Blossom, 2013; Tsimpli et al., 2016), exhibit persistent acquisition challenges with passives (e.g., Du et al., 2024; Leonard et al., 2006; Marinis & Saddy, 2013; Marshall et al., 2007; van der Lely, 1996) compared to their typically developing peers (TDs) or compared to chronologically younger children matched in language abilities. These studies have shown that children with DLD exhibit difficulties in comprehension, production, and processing of reversible passives (e.g., the giraffe is pushed by the rabbit). In comprehension, they tend to misinterpret the subject of the sentence (giraffe) as being the agent of the action while it is the patient, assigning the reverse interpretation (i.e., interpreting the sentence as being active). In production, they tend to produce reversal errors (i.e., the agent in the subject position instead of the patient) or produce active instead of passive verb forms. In terms of real-time comprehension, children with DLD seem to show severe delays in terms of the online processing of passives (Marinis & Saddy, 2013).

The underlying reason for these vulnerabilities in children with DLD has been mostly argued to be their processing limitations, including phonological memory and working memory deficits (e.g., Leonard, 2014; Marinis & Saddy, 2013; Montgomery, 2002) that affect children’s verbal and non-verbal abilities (but see also van der Lely, 1996 for the Representational Deficit Hypothesis). Crucially, limited processing capacity, slower speed of processing, and working memory deficits have been found for several complex syntactic phenomena and embedded structures (e.g., Gillam & Hoffman, 2003; Montgomery & Evans, 2009). These findings are predicted by the Procedural Deficit Hypothesis which posits that impairments in procedural memory are linked to syntactic deficits (Lum et al., 2012; Ullman & Pierpont, 2005), given that the procedural learning system is responsible for detecting and extracting abstract syntactic rules and (ir)regularities from inputs. Thus, grammar is impaired while vocabulary, which is supported by the declarative learning system, remains relatively intact. As a result, declarative memory, which involves some level of explicit learning, may compensate for grammatical deficits (Lum et al., 2012). This suggests that explicit grammar instruction can be beneficial (Balthazar et al., 2020; Ebbels et al., 2024), aligning with the Procedural Deficit Hypothesis (PDH).

Focusing on Greek, passives are an exceptionally challenging structure in typical and atypical acquisition. They are acquired late by Greek monolingual and heritage bilingual children (Paspali et al., 2024a and references therein). Furthermore, passives cause difficulties during processing in typically developing children and children with language and/or reading difficulties (Nerantzini et al., 2022; Paspali et al., 2024c). Talli and Stavrakaki (2020) also found numerical differences in the acquisition of passives in children with DLD (lower accuracy by 13%) compared to their typically developing peers. These difficulties are also related to the structure itself in languages such as Greek—Greek passives are encoded inflectionally on the verb. Greek has a mediopassive voice system, indicating that some verbs which are morphologically marked as “passive” may have another interpretation such as reflexive or reciprocal and not only a passive one (Alexiadou et al., 2015 and references therein). The present study targets only the passive interpretation, either by including prototypically passive verbs (e.g., push or chase), that can only have a passive interpretation, or by including full reversible passive structures (i.e., with the presence of the by-agent phrase which unambiguously denotes a passive interpretation of the sentence (The grandma is being washed by the girl./The boy is being shampooed by the grandpa)).

1.2. Teaching the Passive in Children with DLD

Previous studies targeting certain grammatical phenomena, including passives, in children with DLD have employed both implicit and explicit teaching intervention approaches in speech and language therapy contexts (Baron & Arbel, 2022). Implicit language intervention approaches are used to enrich the child-directed input so that children with DLD discover the target structures more easily via recasts and expansions (such as recasting and modeling). Explicit language intervention approaches are rule-based including the explanation of grammatical rules, feedback, multimodal cues in the instruction materials, and drill-like formats (Montgomery & Gillam, 2024) such as in Shape Coding, MetaTaal, and complex syntactic interventions. Finally, a combination of both explicit and implicit language intervention approaches has been employed (Finestack, 2018; Neumann et al., 2024). It has been argued that implicit approaches are more appropriate for young children with DLD (preschool-aged), as implicit learning is more natural and cognitively less demanding, especially given their limited metalinguistic abilities (Leonard, 2014; Cleave et al., 2015; Owen Van Horne et al., 2018; Wada et al., 2020). On the other hand, explicit interventions have also been recommended and applied—particularly with school-aged children—since explicit language intervention approaches tend to rely on the relatively stronger learning system in DLD, namely declarative memory (e.g., Balthazar & Scott, 2018; Finestack & Fey, 2009; Zwitserlood et al., 2015). More recently, a combination of both explicit and implicit approaches has been recommended (e.g., Finestack, 2018; Balthazar et al., 2020, p. 227; Neumann et al., 2024) for school-aged children, as unconscious learning and the generalization of grammatical rules to new items, contexts, and modalities require not only explicit instruction but also implicit and inductive learning and practice.

Focusing on passives, Camarata and Nelson (1992) tested preschool-aged children with DLD by means of implicit approaches comparing recasting with imitation and found that children acquired passives faster with recasting. Bishop et al. (2006) conducted an implicit intervention with 24 English-speaking school-aged children with specific and non-specific language impairments targeting the comprehension of reversible passives by means of computerized training (rote training). They found no learning gains for the children with DLD compared to the control group, even though responses speeded up over the course of training. In terms of the explicit methods in the treatment of passives, Ebbels and van der Lely (2001) and Delage et al. (2025) employed the Shape Coding intervention technique (i.e., a technique which uses colors, shapes, and arrows to indicate syntactic and morphological structures) with 4 English-speaking and 18 French-speaking school-aged children with DLD, respectively. Ebbels and van der Lely (2001) found that three out of four children exhibit learning gains after the intervention in both the comprehension and production of passives with retention after 30 weeks for two out of three children. Delage et al. (2025) also found learning gains after the intervention and the generalization of the structure in new sentences (not included in the training). Overall, it seems that both implicit and explicit approaches benefit children with DLD in the acquisition of passives.

1.3. FonF Grammar Teaching Techniques: The Case of Processing Instruction and Dictogloss

Over the last three decades, a grammar teaching approach which has been embraced within SLA research is Focus on Form (FonF) (Long, 1991). This approach aims to draw learners’ attention to the target grammatical form within a meaningful context via certain techniques. In this way, it helps learners create form–meaning connections. Such techniques are Processing Instruction (e.g., VanPatten, 1996), corrective feedback (e.g., Ellis et al., 2006), and textual enhancement (e.g., Simard, 2009) as well as reconstruction and group dictation activities such as running dictation and Dictogloss (e.g., Olioumtsevits et al., 2023; Thornbury, 1997; Willis & Willis, 2007). In the present study, we focus on the teaching of passives by means of PI and Dictogloss, which have been found to be efficient in L2 learners for the teaching of passives (e.g., Qin, 2008). Both techniques are multimodal; they can be both explicit and implicit in terms of grammar instruction, i.e., an explicit grammar teaching component can be added; and they can be inductive followed by feedback on learners’ errors/self-correction (see below).

1.3.1. PI

PI draws from Input Processing (VanPatten, 1996, 2002), a psycholinguistic model which describes the default strategies that learners use when receiving language input (i.e., the Primacy of meaning principle or the First Noun principle). PI aims to alter these strategies by providing learners with more enriched input to notice the target grammatical form while processing sentences for meaning and to make the target form–meaning connections. Through this process, the input is transformed into the intake (i.e., grammatical features that students have understood and will retain in working memory during comprehension in real time). For this reason, one major component of PI is structured input activities which are constructed in such a way that they enforce the corresponding form–meaning association (i.e., structured input). These activities are divided into referential and affective. In referential activities, learners listen to or read sentences and are forced to focus on the target grammatical form to assign the correct meaning. Referential activities include correct/incorrect answers for each sentence and feedback is given since learners are provided with the correct answers and are asked to correct their errors themselves. Affective activities require the learner to express a belief, opinion, view, or other affective reaction towards the content of a sentence retrieving information in real time, including the target grammatical form. In addition, PI also includes two explicit components before the structured input activities, which include (a) the explicit teaching of the target form and its mapping with a certain meaning, and (b) a component which explicitly draws the learners’ attention to the incorrect processing strategies they employ while processing the target structure.

PI has been shown to be effective with various grammatical structures and across multiple languages and groups of learners, including (older) adults, adolescents, and children, with long-term effects observed after the intervention (see Wong, 2024 for a recent literature review on PI). Furthermore, it has been found to be effective not only in comprehension but also in production (Qin, 2008) and online processing (Lee & Doherty, 2018; Lee et al., 2022). Most importantly, PI teaching protocols have also been implemented in atypical populations. Santamaria et al. (2013) conducted a preliminary study with aphasic adult patients by means of a PI protocol to enhance the connection between a certain morphosyntactic cue and thematic role assignment in Spanish. One out of the two patients with aphasia exhibited improved performance on the target phenomenon as tested with tests prior and after the PI intervention. Barcroft et al. (2025) implemented a PI protocol using auditory–visual modalities (i.e., auditory training which is meaning-oriented) to 78 school-age children who are deaf and hard of hearing in several morphosyntactic forms in English (third person –s, possessive –s, past tense –ed, and plural –s/es). They found that children exhibited improved performance in comprehension (two-choice picture discrimination task) after the intervention as measured via pre-tests and post-tests. Thus, these studies indicate that PI may be useful for increasing syntactic comprehension as well as the perception and comprehension of bound morphemes in atypical child and adult populations.

1.3.2. Dictogloss

Dictogloss (Swain, 1985) is an intervention which focuses on production. This intervention draws from the Output hypothesis (Swain, 2000), which argues that grammar is learned via comprehensible output and, as such, it targets the production of the grammatical form. According to this hypothesis, the systematic effort of students to produce comprehensible output promotes acquisition and the automation of the grammatical forms/structures. In Dictogloss, the teacher introduces the topic of the text for this activity. Then, they read the texts twice. During the second time, students are allowed to take notes if they want to. Then, the students work in groups of two or more to reconstruct the text in either oral or written form by using their notes and pictures (given by the teacher), which represent the actions/events of the text. They are asked to either reconstruct the whole text or to fill out certain parts of an incomplete version of the text that are missing, i.e., the parts including the target grammatical form. In the end, they compare their text with the original/complete version of the text and correct their errors. Dictogloss is a collaborative and interactive task, which subsequently provokes language production (see Sousa et al., 2019). Furthermore, it is an activity designed to practice all language skills while teaching grammar (Qin, 2008). Specifically, the reconstruction stage of a Dictogloss encourages negotiation and thus languaging upon the target grammatical form (Olioumtsevits et al., 2023). Furthermore, learners also have the opportunity to correct themselves and to further notice the target grammatical form of the output. Dictogloss has been found to be efficient in both comprehension and production, with the learning gains remaining long after instruction (Qin, 2008; Gorman & Ellis, 2019).

1.3.3. (FonF) Teaching Interventions in Greek

Focusing on Greek, certain FonF teaching techniques have been found to be effective in second language learners, school-age children with refugee backgrounds, and in Greek-heritage children. Papadopoulou et al. (2010) employed and compared PI with textual enhancement in adult second language learners of Greek for the teaching of adjective-noun agreement. Both techniques were found to be effective in comprehension as well as in production. Agathopoulou et al. (2013, as cited in Agathopoulou & Papadopoulou, 2014) compared PI with Dictogloss in the teaching of verbal aspect in adult second language learners of Greek. They found that both teaching techniques were effective in comprehension and production, with the learning effects remaining three weeks after the teaching interventions. However, learning effects were also observed for the control group in two out of three evaluation tasks, although they did not receive instruction, which could be attributed to training effects during the testing stage. Olioumtsevits et al. (2023) employed Running Dictation to teach tense and agreement in school-aged children with refugee backgrounds. Although no significant learning gains were found from pre- to post-tests, the children changed the error patterns they made, i.e., they exhibited more systematic and predictable error patterns after the interventions, reflecting qualitative changes in their language development. Ouli and Konta (2025) compared PI and a form of Dictogloss (i.e., students had to find the missing words from the text and also correct ungrammatical sentences in groups) in the teaching of past tense in child heritage learners of Greek living in Hungary. Both interventions were found to be effective, while both led to lower learning gains in the more cognitively demanding task (i.e., sentence repetition).

Within Greek DLD research, there is a lack of studies focusing on teaching interventions. Most studies on Greek children with DLD focus on language acquisition, with limited research exploring the feasibility and effectiveness of teaching interventions in this population. Bachourou et al. (2024) examined the effectiveness of cognitive and linguistic training in promoting gains in verbal short-term and working memory among school-aged children with DLD. Cognitive training included digit span tasks and word/non-word/sentence repetition tasks, while linguistic training included the comprehension and production of complex syntactic structures (e.g., relative clauses) by means of visual representations (shapes/colors), guided sentence building, recasting, and structured elicitation. They found both cognitive and linguistic training were effective. However, Bachourou et al. focused on improving children’s memory. The present study is the first to apply two teaching interventions to support children with DLD in acquiring a linguistic structure.

1.4. Goals and Research Questions of the Current Study

The current pilot study aims to explore whether the enhancement of form–meaning connections by means of FonF language teaching techniques can help children with DLD with their grammatical difficulties. Specifically, the study explores (a) the extent to which PI and Dictogloss are effective grammatical interventions in children with DLD, (b) whether any potential learning gains are maintained after the interventions (i.e., long-term effects of the interventions), and (c) whether there is a difference between the two in terms of their effectiveness. Thus, the current study developed two pedagogical protocols (for PI and Dictogloss, respectively) focusing on the teaching of passives in children with DLD. The novelty of these protocols lies in the fact that they exploit FonF activities for the first time in children with DLD to create opportunities for noticing the target grammatical form, while they are engaged with meaning and content.

This is a pilot and exploratory study and is not meant to produce final conclusions. Instead, the study explores the feasibility of two FonF teaching interventions employed in DLD by means of measuring their effectiveness on children’s language learning gains in the acquisition of passives. Specifically, a pre-test–post-test design was conducted. Grammar evaluation tests (pre- and post-tests) targeted passives before and after the implementation of the teaching protocols. The research questions (RQs) were:

RQ1: (a) To what extent are PI and Dictogloss feasible and effective grammar teaching interventions in children with DLD? And (b), to what extent are any potential learning gains maintained after the interventions (i.e., long-term)?

If PI and Dictogloss are feasible and effective grammar teaching interventions in children with DLD, higher accuracy in the post-test compared to the pre-test is expected. If learning gains are found which can be maintained long-term, accuracy scores in the post-test will be maintained in the delayed post-test.

RQ2: Is there a difference between the two teaching interventions in terms of their effectiveness?

If there is a difference between the two teaching interventions, a significant interaction is expected between the intervention type and the post-test accuracy scores. Given that the study is exploratory, and no previous research has been conducted with FonF interventions in this population, no specific predictions are made regarding the direction of the interaction or the tasks in which any differences may emerge.

2. Materials and Methods

2.1. Participants

The three criteria for participation were:

(a)

Children should have been officially diagnosed with DLD by developmental-behavioural pediatricians, and by qualified speech and language therapists in the absence of autism, intellectual disability, deafness, etc. (Bishop et al., 2017);

(b)

They should have been monolingual native speakers of Greek;

(c)

They should have been between the third and fifth grade of primary school (8–11 years old), given that Greek passives are acquired towards the end of the second grade of primary school in typically developing children (Paspali et al., 2024a).

A total of 23 school-aged children with DLD (third and fourth grade of primary school) took part in the study (mean age in months = 105, SD = 6.3, range = 95–115, 8 girls). The control group consisted of 3 children who only conducted the evaluation tests and did not take part in the teaching interventions (mean age in months = 104, SD = 10.1, range = 95–115, 1 girl), the Dictogloss group consisted of 10 participants (mean age in months = 105.7, SD = 5.9, range = 98–115, 4 girls), the Processing Instruction group consisted of 10 participants (mean age in months = 105.1, SD = 6.3, range= 95–114, 3 girls). They all had Greek as their first and only language and they had all been diagnosed with developmental language disorder by experienced clinicians (i.e., developmental-behavioural pediatricians as well as qualified speech and language therapists) in the absence of a differentiating condition (such as deafness, intellectual disability, and ASD)1.

The children were all receiving speech and language therapy at the time of testing. Their regular interventions at the time of testing focused on reading comprehension skills, as reported by the clinicians (e.g., main idea, inferencing, and decoding). Crucially, they have never received intervention on the target structure. The children’s parents gave their written informed consent, and children gave their oral informed consent according to the Declaration of Helsinki. This study was approved by the Research Ethics Committee of the [Aristotle University of Thessaloniki] (36707/2023).

2.2. Materials and Procedure

Two teaching protocols were constructed to target the instruction of passives—a PI teaching protocol and a Dictogloss teaching protocol. The protocols were implemented in a speech and language therapy center in Northern Greece by one SLT and two teachers, the latter of whom helped children in the organization and preparation of various school subjects. They implemented the teaching protocols voluntarily after giving written informed consent and they received a 4 h training session by the author on the methodology and implementation plan before implementing them. They also received the lesson plans in written format together with detailed step-by-step instructions. Both teaching protocols were implemented individually for each child as in the two previous studies implementing FonF teaching protocols in clinical populations (Barcroft et al., 2025; Santamaria et al., 2013). All sessions were voice-recorded. Each intervention (see

Table 1. Timeline of the teaching interventions and evaluation across groups.

	Week 1	Week 2	Week 3	Week 5
Group
Control group	pre-test	-	post-test	-
PI	pre-test	PI teaching protocol 2 sessions	PI teaching protocol 2 sessions + post-test	delayed post-test
Dictogloss	pre-test	D teaching protocol 2 sessions	D teaching protocol 2 sessions + post-test	delayed post-test

) was conducted in 4 sessions for each child in two consecutive weeks (Week 2 and Week 3) and each session lasted 45–60 min. In the first session, both the PI and Dictogloss groups received explicit grammar instruction in passives (Supplementary Materials). In this session, the teacher presented an active sentence followed by a passive version of the same sentence. The passive sentence had an inanimate subject and no by-phrase so that the child focused on the different verbal morphology between active and passive and the corresponding meaning depicted in the picture below the sentence. Then, another active–passive pair of sentences with the corresponding picture was presented to the child including two animate nouns in the active sentence (an agent in the subject position and a patient in the object position) and one animate noun in the passive sentence (a patient in subject position). Finally, another active–passive pair of sentences with the corresponding picture was presented to the child. Both included two animate noun phrases (i.e., an agent in the subject position and a patient in the object position in active sentences, and a patient in the subject position and an agent within the by-phrase position in passive sentences (see Supplementary Materials)). The same structure was repeated two more times with new pairs of sentences. In the second, third, and fourth sessions, the children were presented with a set of materials depending on the group they belonged to (see below). Three similar sets of materials were constructed for each teaching intervention and each set was implemented in one of the three sessions. The three sets of each intervention were similar to each other. The occurrence of passive sentences was similar between the two teaching interventions. Specifically, 16 passive sentences and 9 active sentences were encountered in each session (i.e., 48 passive and 27 active sentences in total for each teaching intervention). Below, the PI (Section 2.2.1) and Dictogloss (Section 2.2.2) teaching protocols are described followed by the evaluation materials (Section 2.2.3) and data analyses (Section 2.2.4).

The learning objectives of the two teaching protocols for the children were: (a) to develop connections between active and passive voice, as expressed morphologically on the verb, and their difference in meaning, (b) to observe the syntactic position of the nouns in the sentence and the inverted thematic roles in passive sentences in contrast to active sentences, and (c) to become familiar with the presence of the prepositional by-phrase, which is an additional cue for the passive interpretation, but also with its absence in a passive sentence.

2.2.1. Processing Instruction Protocol

Each set consisted of three activities (see Supplementary Materials), two referential activities, and one affective activity. All of them were presented both auditorily by the teacher but also in written format in the booklet they had in front of them. The first activity aims to draw children’s attention to the target verbal form and the associated meaning via alternation of passive and active sentences (6 passive and 3 active sentences) and to the fact that the subject of the sentence is the agent in active sentences but the patient in passive sentences to help them alter their processing strategy (i.e., they tend to interpret the subject of the sentence as being the agent). Children were presented with a sentence, active or passive, and were asked to select the correct picture each time (the two pictures depicted the same event with reversed thematic roles). The second activity aims to further increase their awareness of the processing strategy employed and to make them realize the presence of the by-phrase as a helpful cue towards the correct interpretation in passives and the possibility of omission of this by-phrase. The teacher read aloud each sentence, and the child had to select one of the two pictures with the teacher providing feedback afterwards. There were 6 passive and 3 active sentences. The third activity aimed to make children express their view on whether they have witnessed certain incidents while they were exposed to the target grammatical phenomenon. Specifically, the teacher read a sentence and the child had to indicate whether they have witnessed such an incident and, if so, where In this activity there were no correct/incorrect answers. Visual cues of the depicted incidents were also provided to further promote comprehension of the target structure. There were 4 passive and 3 active sentences in this activity. Note that, since this is a PI teaching protocol, participants were not required to produce the target form.

2.2.2. Dictogloss Protocol

Each set consisted of two texts (see Supplementary Materials), which were different episodes about the adventures of an elf in the forest. These texts were accompanied by pictures depicting each event of the episode. During the introduction stage, children were asked whether they like fairy tales and to name their favorite one. Then, they listened to the episode. The first time they just listened to the pre-recorded audio file at normal speed. The second and third times of listening to the same episode, they listened to the pre-recorded audio file at a lower speed, and they were also presented with the relevant pictures depicting the actions of the episode in serial sequence. Then, the children were presented with a booklet containing different pages for every two episodes which included the relevant sentences and the corresponding pictures below the text. The text included gaps for the verbs (passive or active verbs), for the by-phrases, or for the subject of the passive sentences. The teacher started reading the story aloud and the children were asked to fill out the gaps in the text with the aid of the pictures. The child or the teacher had to write each answer in its missing gap. At the feedback stage, the teacher and the child started reading the text aloud together and the teacher provided the child with the correct answers orally. Thus, oral feedback was always provided to the child. When errors occurred, the child had to correct their initial answers by writing down the correct answer they had heard using a pen with a different color. Orthographical errors were corrected at this stage (intervention) but did not count as inaccurate responses in the evaluation stage. In each session, there were two episodes consisting of 16 passive sentences and 9 active sentences overall. The step of making notes was skipped due to the age of the children (note that even typically developing children at this age face difficulties with taking notes while listening to a story). The pictures were carefully and meticulously designed to help children find the missing word depicted (i.e., the verb of the depicted action or the NP of the agent/theme), hence reducing the memory demands. Thus, children were able to focus on the production of the correct grammatical form. During our piloting, typically developing children of the same age were able to identify the missing words easily after listening to the text just twice.

2.2.3. Evaluation

The pre-test was administered one week before the first session of the teaching intervention and the post-test was administered immediately after the end of the last session of each intervention. To test for retention, a delayed post-test was also conducted two weeks after the last intervention session. The evaluation (Table 1) was conducted by the author. The researcher did not know at the time of the evaluation which intervention each child had received. All tests were voice-recorded. Each test included a comprehension task, a production task, and a sentence repetition (SR) task. The comprehension task included 8 passive sentences accompanied by the corresponding picture illustrating the meaning of the sentence. In each sentence, one of these three parts were missing: the verb, the subject, or the by-phrase. The children were asked to listen to each sentence and look at the corresponding picture. They were then asked to select the correct answer out of three options—when the verb was missing, the options were the verb in active form, passive form, or in another inflectional form; when the subject was missing, the options were a noun in nominative, in accusative, or in genitive/prepositional phrase; when the by-phrase was missing, the options were the noun within a by-phrase, the noun in nominative, and the noun in accusative. The production task included 8 passive sentences accompanied by the corresponding picture, illustrating the meaning of the sentence. In each sentence, there was a gap and the missing phrase was given in the initial form in brackets (verb, subject, by-phrase). The children had to orally produce the correct form of the missing phrase. In the SR task, there were 8 passive sentences. The children listened to a sentence and had to repeat it. Two scoring methods were calculated: (a) accuracy scores measured whether the child repeated the sentence verbatim (0–1 points), and (b) structural scores measured whether the child repeated the target structure correctly (0–1 points), regardless of other errors or omissions. The first scoring method has been argued to measure overall language processing abilities, while the second scoring method measures language knowledge of the target form in production (Marinis & Armon-Lotem, 2015). For pre- and (delayed) post-tests, 72 sentences were constructed (24 by test, 8 by task) and they were equally assigned to the three test versions2. Half of the sentences employed verbs that children had listened to during the teaching interventions but in new sentences (i.e., the lexical items for the subject and the by-phrase differed). The rest of the sentences included verbs that children had not listened to during the intervention3.

2.2.4. Data Analyses

The data analysis was conducted in R (Version 4.4.1; R Core Team, 2023). Mixed-effects binary logistic regression models were used using the glmer function; these models can account for random variability (i.e., in participants and items) and are suitable for repeated measures in which observations are not independent (Baayen et al., 2008). Given that the control group (i.e., the group who did not receive any teaching intervention at all) comprised only three children, it was not included in the statistical analysis, and it was reported before the results of the PI and Dictogloss group. Thus, the fixed effects of the models were Group (PI, Dictogloss), Test (pre-test, post-test, delayed post-test), and their interaction. Sum contrasts (assimilating ANOVA contrasts) were applied to Group and reverse Helmert contrasts were applied to Test. Thus, the first regression coefficient of Test (i.e., Test 1) across all Tables in Section 3 reports the learning gains of the two interventions together, i.e., contrasting post-test to pre-test. The second regression coefficient of Test (i.e., Test 2) reports retention, i.e., contrasting level 3 (delayed post-test) and levels 1 (pre-test) and 2 (post-test) combined. The interactions indicate whether there is a significant difference between the two teaching interventions either in terms of learning gains (first interaction coefficient) or in terms of retention (second interaction coefficient). Random effects were included for participants and items. A random slope for Test was included within the Participant intercept to capture individual differences in learning trajectories (e.g., some participants improve more than others). A random slope for Group was not included, as this is a between-subject factor, i.e., participants cannot show variation within Group since each receives only one of the two interventions. Random slope simplification was applied when the model did not converge or due to 1/−1 correlations in the random effects. Significant interactions were further explored with pairwise comparisons (adjusted for multiple comparisons with Bonferroni correction) using the emmeans package (Lenth, 2025). Throughout the current study, b, SE, z, and p-values are reported.

3. Results

3.1. Control Group

The children with DLD in the control group did not exhibit learning gains in any of the tasks as their mean scores indicate (

Table 2. Mean accuracy scores by task and teaching group in pre-test, post-test, and delayed post-test with standard deviation (SD) in parentheses.

		Pre-Test	Post-Test	Delayed Post-Test
Task	Group
Comprehension task	Control group	0.46 (0.51)	0.42 (0.50)	0.29 (0.46)
	Dictogloss	0.61 (0.49)	0.88 (0.33)	0.76 (0.43)
	PI	0.54 (0.50)	0.76 (0.43)	0.65 (0.48)
Production task	Control group	0.38 (0.50)	0.25 (0.44)	0.33 (0.48)
	Dictogloss	0.48 (0.50)	0.94 (0.24)	0.86 (0.35)
	PI	0.50 (0.50)	0.81 (0.40)	0.78 (0.42)
SR—accuracy scores	Control group	0.25 (0.44)	0.12 (0.34)	0.21 (0.42)
	Dictogloss	0.24 (0.43)	0.29 (0.46)	0.19 (0.39)
	PI	0.21 (0.41)	0.46 (0.50)	0.46 (0.50)
SR—structural scores	Control group	0.38 (0.50)	0.38 (0.50)	0.33 (0.48)
	Dictogloss	0.59 (0.50)	0.79 (0.41)	0.74 (0.44)
	PI	0.61 (0.49)	0.69 (0.47)	0.69 (0.48)

). This is also confirmed for each child within this group when looking at the individual means. Thus, the present findings show that any improvement in passives after the teaching intervention is less likely to be attributed to training effects from the tests themselves or to other reasons. However, given the difficulties in recruiting this population, there were only three children with DLD in this group and this finding should be interpreted with caution.

3.2. PI and Dictogloss Groups

3.2.1. Comprehension Task

The mean accuracy in the comprehension task is reported in Table 2 and Figure 1.

Table 3. Mixed-effects model results in the comprehension task.

	Estimate	Std. Error	z	p
(Intercept)	1.212	0.314	3.86	<0.001
Test 1	0.831	0.218	3.81	<0.001
Test 2	−0.008	0.121	−0.07	0.946
Group	0.862	0.572	1.51	0.132
Test 1: Group	0.410	0.308	1.33	0.183
Test 2: Group	0.011	0.167	0.07	0.948

Note: p-values in bold indicate statistical significance.

shows the model results. In the comprehension task, both intervention groups exhibited increased accuracy in the post-test compared to the pre-test indicated by the significant effect of Test 1. The delayed post-test did not exhibit significantly lower or higher performance compared to the post-test, reflecting retention. The two intervention groups did not differ from each other since no significant interactions were found between the two, reflecting similar learning gains and retention.

3.2.2. Production Task

In the production task, the mean accuracy is represented in Table 1 and Figure 2.

Table 4. Mixed-effects model results in the production task.

	Estimate	Std. Error	z	p
(Intercept)	1.319	0.217	6.07	<0.001
Test 1	1.186	0.212	5.59	<0.001
Test 2	0.193	0.117	1.66	0.097
Group	0.627	0.369	1.70	0.089
Test 1: Group	0.724	0.325	2.23	0.026
Test 2: Group	0.016	0.179	0.09	0.928

Note: p-values in bold indicate statistical significance.

shows the model results. There was a significant effect of Test 1, reflecting an overall increased accuracy in the post-test compared to the pre-test. The delayed post-test did not exhibit significantly lower or higher performance compared to the post-test, reflecting retention. Crucially, there was a significant interaction between the intervention group and Test1, reflecting that the two intervention groups exhibit differences from each other in their post-test accuracy (compared to the pre-test, i.e., differences in the learning gains they exhibit). Pairwise comparisons revealed that accuracy significantly increased in both intervention groups in the post-test compared to the pre-test (PI: b = 1.65, se = 0.464, z = 3.55, p = 0.001; Dictogloss: b = 3.097, se = 0.596, z = −5.196, p < 0.001). However, these learning gains were higher for the Dictogloss group compared to the PI group, as indicated by the between-group comparison in the post-test (b = 1.334, se = 0.612, z = 2.18, p = 0.029). No significant difference was observed between the two interventions in the pre-test (b = −0.113, se = 0.425, z = −0.266, p = 0.791). Thus, both intervention groups exhibited learning gains in the post-test. However, the Dictogloss group exhibited significantly higher learning gains in the post-test compared to the PI group.

3.2.3. SR Task—Accuracy Scores

In the SR task, the mean accuracy for the accuracy scores is represented in Table 1 and Figure 3.

Table 5. Mixed-effects model results in the SR task—accuracy scores.

	Estimate	Std. Error	z	p
(Intercept)	−1.041	0.270	−3.86	<0.001
Test 1	0.446	0.140	3.18	<0.001
Test 2	0.041	0.081	0.51	0.613
Group	−0.676	0.536	−1.26	0.207
Test 1: Group	−0.590	0.279	−2.12	0.035
Test 2: Group	−0.412	0.162	−2.55	0.011

Note: p-values in bold indicate statistical significance.

shows the model results. The results show that there was a significant learning gain with significantly increased accuracy in the post-test. However, this effect was mediated by interactions between Group and Test, reflecting differences between the two groups. Pairwise comparisons revealed that learning gains were obtained only after PI since that accuracy was higher in the post-test compared to the pre-test for PI (b = 1.48, se = 0.406, z = 3.65, p < 0.001) but not after Dictogloss (b = 0.30, se = 0.385, z = 0.78, p > 0.05). Furthermore, retention of those learning gains was also found for PI since the difference between post-test and delayed post-test is not significant (b < 0.001, se = 0.363, z = 0.00, p > 0.05). In the delayed post-test, accuracy was higher in PI compared to Dictogloss (b = 1.50, se = 0.63, z = 2.39, p = 0.017), given that only PI revealed learning gains. No differences in terms of accuracy were found between the two intervention groups in the pre-test (p > 0.05).

3.2.4. SR Task—Structural Scores

The mean accuracy for the structural scores is represented in Table 1 and Figure 4.

Table 6. Mixed-effects model results in the SR task—structural scores.

	Estimate	Std. Error	z	p
(Intercept)	0.918	0.236	3.90	<0.001
Test 1	0.373	0.149	2.51	0.012
Test 2	0.051	0.086	0.59	0.556
Group	0.335	0.452	0.74	0.460
Test 1: Group	0.363	0.261	1.39	0.165
Test 2: Group	0.061	0.151	0.40	0.688

Note: p-values in bold indicate statistical significance.

shows the model results. The results show that both intervention groups exhibited increased accuracy in the post-test compared to the pre-test indicated by the significant effect of Test 1. The delayed post-test did not exhibit significantly lower or higher performance compared to the post-test, reflecting retention. The two intervention groups did not differ from each other since no significant interactions were found between the two, reflecting similar learning gains and retention.

4. Discussion

The current pilot study employed two FonF teaching intervention techniques in Greek school-aged children with DLD. The study provides preliminary evidence on the feasibility and effectiveness of the respective interventions in supporting children with DLD in the acquisition of passives. FonF intervention techniques force learners to notice the target grammatical form while their attention is on meaning, i.e., establishing new form-meaning connections. These techniques can be input-based, focusing on the processing of input (i.e., Processing Instruction), or output-based focusing on producing the target grammatical form within communicative activities (i.e., Dictogloss). The current study explored whether such teaching interventions would be beneficial for children with DLD, building on two previous studies (Barcroft et al., 2025; Santamaria et al., 2013) that implemented similar interventions with atypical populations and reported promising results. The present study targeted the teaching of passives which is a quite vulnerable grammatical phenomenon in children with DLD by means of the Processing Instruction and Dictogloss teaching interventions. The study employed pre-tests, post-tests, and delayed post-tests to explore (a) the potential (long-term) effectiveness of the teaching interventions (RQ1–2), and (b) potential differences in the effectiveness of the two teaching interventions (RQ2). The present findings indicate that both Processing Instruction and Dictogloss could be promising for children with DLD since both led to learning gains in passives across all tasks: comprehension task, production task, and sentence repetition task. Specifically, in the comprehension task, both were equally effective, leading to learning gains and retention. In the production task, Dictogloss was significantly more effective than PI, even though both interventions led to learning gains and retention. In the SR-task, learning gains and retention were observed only for PI when the correct repetition of the whole sentence was measured. When correct repetition of the target structure was measured, both Dictogloss and PI were equally effective leading to learning gains and retention.

The present findings confirm that Processing Instruction can be effective not only in comprehension but also in production (VanPatten & Cadierno, 1993; Shintani, 2015). However, Dictogloss seems to be more effective in production since it is an output-based activity. The advantage of Dictogloss over Processing Instruction in a production task has also been found in a previous study on the teaching of passives in second language learners of English (Qin, 2008). On the other hand, the present study did not find an advantage for Processing Instruction over Dictogloss in the comprehension task, which is in line with previous studies (Agathopoulou & Papadopoulou, 2014; Ouli & Konta, 2025). This could probably relate to the fact that during the reconstruction stage of a Dictogloss task, learners need to think and negotiate the target grammatical form, which may lead them to observe potential errors they make and pay closer attention to the input they receive (see Olioumtsevits et al., 2023 and references therein). Furthermore, in the present study, children were also shown pictures depicting the corresponding events while listening to the text for the second and third time. Thus, visual cues during the listening stage of the Dictogloss task may have further enhanced their interpretation of the passive form.

Furthermore, when accuracy scores were measured in the SR task, only Processing Instruction was effective (i.e., repetition of the sentence verbatim). This is an important finding given that the accuracy scores of SR are considered to tap more into language processing skills and not only into language knowledge of the target structure (Marinis & Armon-Lotem, 2015). Crucially, Processing Instruction is a teaching intervention which aims to alter the incorrect/default processing strategies of the learners and force them to process new form-meaning connections in real time. The present findings reflect the advantage of Processing Instruction over Dictogloss when processing skills come into play for Greek children with DLD. Similar findings on the effectiveness of Processing Instruction in the real-time processing of passives have been previously found in the literature (Lee & Doherty, 2018; Lee et al., 2022). However, when Structural scores in the SR task were measured in children with DLD (correct/incorrect repetition of the passive structure regardless of other errors/omissions), the two teaching interventions did not significantly differ from each other (note though, that the learning gains for the Dictogloss group are numerically higher). This aligns with the view that the structural scores of SR are more related to language knowledge of the target structure and less with language processing (Marinis & Armon-Lotem, 2015).

The present findings also show that children exhibit better performance in the production and comprehension tasks, while in SR they exhibit poor performance in the pre-test which remains below/at chance levels in the post-test despite the learning gains. This is crucial because it shows that children with DLD exhibit severe difficulties in highly demanding tasks which require more processing resources. Notice also that the SR task was the only task of the evaluation in which the sentences were presented only orally and no visual cues (i.e., pictures) were given to the child as opposed to the other two tasks. Sentence repetition tasks have been argued to measure the perception of a grammatical form and not its comprehension, the latter of which is associated with meaning (Barcroft et al., 2025). Thus, the present findings reflect that children with DLD improved comprehension (and production) of the target grammatical form when this form is overtly associated with its target meaning (e.g., by means of visual cues), while they experience great challenges in perception-based tasks in which the grammatical form is present without support from the meaning (i.e., visual cues).

The present study also had certain limitations. The small sample size within each group is one of them. In the control group, there were only three participants which were not included in the statistical analysis and the two teaching intervention groups had 10 participants each. Furthermore, the items in the evaluation were relatively limited (eight items by task). Additionally, the delayed post-test was conducted two weeks after the intervention had ended, which is usually the minimum interval to test for retention effects of a FonF teaching intervention. Thus, future research should address the gaps in this pilot study.

Pedagogical Implications and Future Directions

In the present study, the teaching interventions were employed with each child separately following most previous studies on language intervention in children with DLD (e.g., Delage et al., 2025). Since (long-term) learning gains were observed with these two FonF activities, these preliminary results appear to be promising for supporting grammar development in children with DLD. Teachers may benefit from integrating both approaches to support different aspects of teaching the passive voice—using Dictogloss primarily to enhance production, and PI to strengthen comprehension and recall of passive sentences. The techniques and materials should be tailored to the needs of children with DLD, and multiple sessions are recommended to support long-term learning gains.

Furthermore, FonF activities and teaching interventions are primarily classroom-based. A previous study shows, though, that group work is not always efficient for children with DLD (Plante et al., 2019). Future research could extend such studies within classroom settings (i.e., students working in small groups). This is important when considering the sense of inclusivity along with (linguistic) support of this population within mainstream classrooms. Furthermore, these activities could also benefit typically developing children; all children will have the chance to develop their metalinguistic abilities, inductive learning, and teamwork skills (i.e., active listening, conflict resolution, respect and empathy, accountability, flexibility and adaptability, and collaborative problem-solving). Consequently, the sense of inclusivity (Chahboun et al., 2024) will be enhanced in the classroom. Furthermore, the teaching protocols implemented in the current study were adapted to match the needs of the specific population (they were conducted orally, lower pace of speech in the narrations of Dictogloss, ample visual cues, etc.). These are all important administrative and instructional details that teachers need to take into consideration when designing language activities for children with DLD.

5. Conclusions

This is the first study which developed and tested novel teaching tools to support language teaching in Greek children with DLD by means of FonF activities (both input- and output-based). The results of the present study appear to be promising and valuable for educators, as the teaching interventions of PI and Dictogloss supported children with DLD and led to (long-term) learning gains. Further research is needed in this direction to explore similar methods for supporting the language development of children with DLD.