RESEARCH DOSSIER
Developmental Language Disorder and Language-Learning Differences
Abstract. Developmental language disorder (DLD) describes persistent developmental difficulties understanding or producing language that restrict everyday participation and are not explained by an identified differentiating biomedical condition. This fundamental critical narrative synthesis examines the history and operational definition of DLD; prevalence and longitudinal change; phonological, lexical, grammatical, discourse, and pragmatic profiles; competing linguistic, cognitive, auditory, and neurodevelopmental explanations; multilingual assessment and differential diagnosis; associations with literacy and psychosocial development; and experimentally studied interventions. Primary experiments, population studies, international consensus reports, clinical research, meta-analyses, and a recent randomized caregiver-mediated intervention are examined according to design and generalizability. The synthesis distinguishes risk factors, co-occurring conditions, explanatory hypotheses, and diagnostic evidence, and it identifies priorities for equitable, functional, longitudinal research. No original experiment, registered systematic review, meta-analysis, medical review, or new validated mechanism is claimed.
Keywords: developmental language disorder; DLD; specific language impairment; language acquisition; language-learning differences; child language; morphosyntax; vocabulary; narrative; language assessment; bilingualism; cross-linguistic assessment; nonword repetition; sentence repetition; language intervention; literacy; dyslexia; CATALISE; longitudinal language development; language and cognition
The Scientific Object: Developmental Language Disorder and Language-Learning Differences
Developmental language disorder (DLD) concerns the acquisition, organization, and everyday use of language when persistent difficulty in understanding or producing language interferes with social, educational, or practical functioning. Language here includes sound patterns as linguistic representations, word meanings, grammar, discourse, and the use of linguistic information for communication. Developmental variation encompasses many ordinary rates and pathways of learning; a clinically significant disorder requires a judgment about persistence, impact, and contextual adequacy. The CATALISE consensus emphasizes prognosis and functional consequences and replaces a narrow discrepancy model with a developmentally grounded assessment of language difficulty Bishop et al. (2016) Bishop et al. (2017) Leonard (2014).
The research object is intrinsically multidimensional. One child may understand conversational instructions but struggle to mark tense and agreement; another may produce fluent sentences while misunderstanding complex relational language; still another may have difficulty learning unfamiliar words, organizing narratives, or following extended classroom explanation. Differences across domains, languages, situations, and developmental stages are data requiring explanation rather than departures from a single idealized presentation. Their scientific examination combines linguistic descriptions, psychometric comparisons, naturalistic samples, longitudinal measurement, and learning experiments. Observed patterns must be related to age, opportunities for exposure, educational expectations, and individual strengths Rice et al. (1995) Conti-Ramsden & Botting (1999) Kidd et al. (2018).
This monograph integrates historical research under the term specific language impairment (SLI) with the contemporary DLD framework. It traces epidemiology and prognosis, evaluates mechanistic hypotheses, examines conditions that complicate identification, and assesses evidence for intervention. A critical narrative synthesis is appropriate because the relevant evidence spans experiments, observational cohorts, clinical guidelines, and systematic reviews with different target constructs. The work identifies each design's inferential reach rather than pooling incompatible estimates or claiming to execute a registered systematic review. The central question is how diverse developmental profiles become recognizable, measurable, and appropriately supported without confusing linguistic difference with clinical impairment Bishop (2006) Bishop et al. (2017) Ebbels et al. (2019).
Historical Genealogy: From Developmental Aphasia to Specific Language Impairment
Twentieth-century studies used several labels for children who showed unexpectedly persistent spoken-language difficulties, including developmental aphasia, developmental dysphasia, and specific language impairment. The vocabulary reflected changing theories of language localization, developmental maturation, and the status of nonverbal intelligence in clinical classification. Tallal and Piercy examined rapid auditory discrimination under the designation developmental aphasia, whereas later research investigated how linguistic forms and processing profiles differed from those of age peers. These historical descriptions can be examined as distinct proposals about causal mechanisms, not treated as evidence that one original syndrome explains all later DLD profiles Tallal & Piercy (1973) Leonard (2014) Bishop (2006).
The SLI research program produced important standardized cohort studies, grammatical experiments, and work on phonological memory. Rice and colleagues' optional-infinitive account investigated atypically prolonged omission of tense marking in English-speaking children, a finding that supplied a precise target for grammatical explanation. Population studies, meanwhile, demonstrated that clinically important language difficulty is often less visible than obvious speech sound errors and may be missed by routine observation. Each approach sharpened measurement but also exposed classification problems: results based on English verb morphology cannot simply be universalized to languages with different grammatical architectures Rice et al. (1995) Tomblin et al. (1997) Leonard (2014).
The move toward DLD arose partly because the designation specific seemed to demand an isolated language deficit and a sizeable gap between language and nonverbal ability. Those exclusions were difficult to reconcile with observations of associated learning, attentional, motor, and social difficulties. CATALISE used an international multidisciplinary Delphi process to establish terminology more closely linked to functional impact, prognosis, and the distinction between differentiating biomedical conditions and co-occurring developmental conditions. The historical transition changes inclusion boundaries for research samples as well as clinical practice, so older SLI evidence requires careful population mapping when used in contemporary DLD synthesis Bishop et al. (2016) Bishop et al. (2017) Norbury et al. (2016).
CATALISE Terminology and the Boundaries of Diagnosis
The CATALISE Phase 2 statements reserve the term language disorder for problems likely to persist and to cause significant functional impairment, and DLD for language disorder without an identified differentiating biomedical etiology. A child can have genetic or environmental risk factors and can also present with attention-deficit/hyperactivity disorder or other developmental difficulties while meeting DLD criteria. Performance on nonverbal reasoning does not have to exceed language performance by a fixed numerical difference. This terminology addresses a recurrent scientific problem: definitions that depend heavily on exclusion may select unusually narrow groups that poorly represent children needing support Bishop et al. (2017) Norbury et al. (2016).
Diagnostic language must also distinguish conditions in which language disorder is associated with an identified biomedical syndrome or condition. In such cases, a clinical formulation describes the language impairment in relation to the differentiating condition, while the child's communication needs remain central to assessment. The conceptual separation is between etiological classification and treatment need; it never implies that those children should be excluded from language support. Co-occurrence, in contrast, refers to another developmental difficulty that can be present alongside DLD without explaining the linguistic phenotype. Contextual factors and unequal educational opportunities require assessment rather than reflexive use as exclusion criteria Bishop et al. (2016) Bishop et al. (2017) Ebbels et al. (2019).
The consensus is a professional agreement informed by evidence, not a randomized experiment that demonstrates one biological disease entity. Its value is to improve the consistency of research recruitment, clinical communication, and access to services while allowing hypotheses about mechanisms to remain open. Investigators should report the operational language cutoff, functional consequences, exposure history, nonverbal measures, and co-occurring conditions used to form each sample. Disagreements among national diagnostic systems or service policies should be made explicit. No single score, test item, or consensus term replaces developmental history, contextualized language observation, and appropriately selected clinical assessment Bishop et al. (2017) Ward et al. (2024) Restrepo (1998).
Population Prevalence and the Problem of Case Ascertainment
The early epidemiological work of Tomblin and colleagues screened a large kindergarten population and provided a frequently cited estimate near seven percent for specific language impairment under the study's operational definition. Norbury and colleagues later examined language disorder in a population design and estimated DLD of unknown cause at approximately 7.6 percent, with uncertainty around that estimate. These are influential estimates for sampled age groups and ascertainment procedures, not universal constants across nations, age ranges, languages, and changing diagnostic categories. Thresholds, exclusion rules, assessment language, and school participation can materially alter the population classified as affected Tomblin et al. (1997) Norbury et al. (2016).
Population prevalence also differs from the proportion of children who have received a formal diagnosis, are referred to a specialist, or appear in an intervention study. Educational services depend on awareness, access, economic resources, and differences in professional referral practices. Children whose difficulties primarily affect comprehension may attract less attention than children whose errors are audible. In multilingual settings, unfamiliarity with a child's home-language development can yield both underidentification and overidentification. Any prevalence claim therefore requires denominator, age, language background, assessment procedure, geographical context, and a precise distinction between screened impairment and clinically confirmed need Norbury et al. (2016) Restrepo (1998) Bishop et al. (2016).
Theoretical inference from epidemiology requires particular restraint. A high population frequency establishes the public importance of DLD, but does not establish a single cause, a unitary cognitive deficit, or a fixed natural boundary between typical and atypical language trajectories. Screening distributions may reflect continuous variation interacting with clinical cutoffs; persistence and functional limitation remain substantive dimensions. Comparative studies should examine how estimates respond to alternative thresholds and to inclusion of co-occurring diagnoses. School-entry estimates are especially useful for planning but cannot be directly transported to adolescent outcomes or other languages without longitudinal and cross-cultural validation Kidd et al. (2018) Bishop et al. (2017) Conti-Ramsden et al. (2018).
Developmental Prognosis and the Meaning of Persistence
Prediction of persistence is a defining clinical challenge because early language development displays substantial individual variability. A toddler with limited expressive vocabulary may rapidly acquire words during the next year, whereas a child with broad receptive and expressive difficulty can face enduring challenges. Bishop and Adams followed children with early language and phonological difficulties and linked later reading outcomes to the developmental course of language impairment. Such prospective designs are more informative than retrospective narratives about a single early milestone because they reveal temporal ordering and the changing predictive value of different language measures Bishop & Adams (1990) Bishop et al. (2016) Kidd et al. (2018).
Longitudinal evidence also complicates static language subtypes. Conti-Ramsden and Botting reported changes in classifications of children with SLI over time; this demonstrates that a phenotype defined at one age need not map cleanly onto a stable category. Development can involve growth with continued relative disadvantage, domain-specific improvement, or changing manifestations as educational demands increase. Investigators must distinguish a stable liability from stable measured performance. Repeated assessments using developmentally comparable tasks, informant reports, and contextual participation measures offer stronger inference than one categorical label assigned from a single testing session Conti-Ramsden & Botting (1999) Leonard (2014) Bishop et al. (2017).
Persistence is not equivalent to absence of learning. A child may acquire sophisticated vocabulary and compensatory strategies while continuing to experience difficulty with inferential language, complex syntax, or rapid classroom instruction. The interpretation of a longitudinal change should separate absolute developmental gain from change in percentile position relative to peers. Later outcomes depend on schooling, literacy opportunities, accommodations, co-occurring difficulties, and support. The work of Conti-Ramsden and colleagues on young-adult education and employment illustrates heterogeneity, making it unsuitable to convert an elevated group risk into an individual forecast Conti-Ramsden et al. (2018) Ebbels et al. (2019) Bishop & Snowling (2004).
Research Designs and Levels of Explanation
Scientific investigation of DLD uses complementary designs addressing different questions. Population sampling estimates frequency and associations; longitudinal cohorts examine trajectories and prospective predictors; cross-sectional experiments compare performance under controlled tasks; training studies test modifiable functions; and randomized interventions estimate treatment effects under specified conditions. The unit of analysis may be an individual response, a linguistic construction, a child, a classroom, or a family interaction. These levels cannot be assumed interchangeable. A study demonstrating weaker sentence repetition in a selected clinical group does not, by itself, establish why the difficulty developed Norbury et al. (2016) Ward et al. (2024) Ebbels et al. (2019).
Measurement is particularly consequential because language itself is used to administer many cognitive tasks. An apparent limitation in verbal working memory may partly reflect the linguistic complexity of instructions or the child's familiarity with the test language. Conversely, restricted language knowledge can depress performance on ostensibly nonverbal activities requiring conceptual explanation or verbal mediation. High-quality experiments manipulate memory load, input rate, lexical familiarity, sentence complexity, and modality independently when possible. Analyses should report age matching, selection rules, language history, reliability, the size and uncertainty of group differences, and whether results generalize across tasks Archibald & Gathercole (2006) Ullman & Pierpont (2005) Kidd et al. (2018).
The evidence assembled here is organized by inferential strength rather than by a single hierarchy applied without context. Random assignment supports causal claims about particular treatments, while population designs support representativeness and estimates of risk; carefully controlled linguistic experiments characterize the computational locus of a deficit. Systematic reviews aggregate designs but remain constrained by the quality and heterogeneity of included studies. The monograph therefore evaluates the fit between each scientific claim and the method that produced it. Assertions about an underlying neurodevelopmental mechanism require convergent observations across measurement modalities, whereas diagnostic recommendations demand prospective accuracy and consequences for children Law et al. (2004) Ward et al. (2024) Bishop et al. (2017).
Phonological Representations, Speech, and Language
Speech sound production and language learning interact, but the constructs refer to distinguishable aspects of communication. Speech sound disorder primarily concerns producing or organizing the sounds of speech, whereas DLD involves the representation, comprehension, and use of linguistic structure across domains. Phonological representation contributes to learning new words, segmenting utterances, and recognizing grammatical contrasts, so co-occurrence is expected in some children. Researchers should distinguish articulation error, phonological processing difficulty, and limited linguistic knowledge through appropriate tasks. The older prospective literature shows why early sound difficulties should be studied alongside broader language measures when investigating later literacy Bishop & Adams (1990) Leonard (2014) Bishop & Snowling (2004).
A child's pronunciation can obscure grammatical competence or make accurate vocabulary production difficult to judge. Conversely, apparently clear speech can coexist with significant receptive, syntactic, or narrative limitations. Transcription of connected speech permits analysis of consonant accuracy, inflectional morphology, lexical diversity, and discourse organization, provided that examiners keep these dimensions separate. A nonword repetition task additionally requires hearing, temporary representation, sequencing, and motor production; poor performance may arise through multiple pathways. Its diagnostic promise comes from measurable group differentiation, while explanatory interpretations require independent process measures rather than treating the task as a pure assay of phonological storage Archibald & Gathercole (2006) Schwob et al. (2021) Ortiz (2021).
Cross-language considerations are central because phoneme inventories, syllable complexity, orthographic depth, and grammatical marking differ. A task using consonant clusters common in one language may reflect unfamiliar phonotactics in another. Speech production standards established in one dialect may misclassify ordinary variation, and test adaptation must evaluate equivalence rather than directly translate prompts. Developmental language research gains explanatory precision when linguistic phenomena are specified within the grammar being studied and when general claims are tested against typologically varied populations. Appropriate assessment considers all of a child's languages and relevant community norms Restrepo (1998) Schwob & Skoruppa (2022) Bishop et al. (2017).
Lexical Learning, Semantic Knowledge, and Word Retrieval
Vocabulary growth depends on extracting forms from speech, linking them to meanings, retaining associations across encounters, and adjusting knowledge as linguistic contexts accumulate. Children with DLD can show difficulty learning unfamiliar labels, retrieving familiar words, interpreting multiple meanings, and expanding semantically organized knowledge. These abilities are distinguishable: recognition of a familiar object word can be preserved when rapid retrieval is difficult, and explicit teaching may improve a particular word without eliminating a broader learning vulnerability. An account of lexical development must specify whether the difficulty concerns encoding, retention, retrieval, semantic integration, or the opportunities available for acquisition Leonard (2014) Bishop (2006) Kidd et al. (2018).
Experimental word-learning tasks vary in the number of exposures, competing referents, contextual richness, required response, and delay to test. Performance differences are only interpretable when researchers identify those parameters and compare recognition with production and short-term success with long-term retention. A lower score on a single fast-mapping task should not be taken as proof of a domain-general memory failure, especially when initial vocabulary knowledge or attention differs between groups. Training designs can separate immediate acquisition from generalization to new exemplars and spontaneous use. Language samples supplement standardized vocabulary tests by capturing word deployment in consequential communication Archibald & Gathercole (2006) Kidd et al. (2018) Ebbels et al. (2019).
Semantic difficulties become especially consequential when education demands abstract, technical, and relational vocabulary. Understanding cause, exception, comparison, and probability may require integrating word meaning with sentence syntax and background knowledge. A child who can name everyday objects may still face extensive difficulty following unfamiliar instructional vocabulary. Family opportunities, exposure to academic language, schooling conditions, and cultural practices influence lexical experiences for all children. The DLD construct therefore should incorporate functional patterns across learning contexts without attributing differences in language input automatically to the presence or absence of a developmental disorder Bishop et al. (2016) Bishop et al. (2017) Conti-Ramsden et al. (2018).
Morphosyntax and the Architecture of Grammatical Difficulty
Morphosyntax includes the inflections and sentence structures that convey tense, agreement, argument relations, negation, and complex propositions. Rice, Wexler, and Cleave demonstrated that the optional use of finite tense morphology could extend developmentally in children selected for SLI. Their findings supplied a specific grammatical hypothesis: particular children may treat forms that are obligatory in adult English as optional for longer than age peers. The explanatory target is the structure and developmental timing of a linguistic contrast, not a generalized claim that every grammatical feature is equally impaired in every language or every child with DLD Rice et al. (1995) Leonard (2014) Bishop et al. (2017).
Grammatical vulnerability may appear through omission of tense markers, difficulty interpreting embedded clauses, incorrect assignment of sentence roles, or limited control of long-distance dependencies. Different tests probe distinct processes: elicited production asks for forms under a designed context, grammatical judgment depends on metalinguistic understanding, and comprehension tasks may introduce additional demands on memory and inference. In natural discourse, short utterances can conceal grammatical limitation because speakers choose familiar constructions and avoid demanding structures. A scientifically informative profile triangulates controlled tasks with spontaneous samples, considering both accuracy and the communicative contexts in which forms appear Rice et al. (1995) Ward et al. (2024) Archibald & Gathercole (2006).
The cross-linguistic distribution of tense, case, agreement, and word order makes a language-specific prediction essential. An English third-person singular marker cannot be treated as a universal diagnostic indicator in a language without a comparable morpheme. A theory claiming a general grammatical learning limitation must explain why the surface phenotype differs across linguistic systems. Comparative work should report the target construction's frequency, prosodic prominence, regularity, morphological complexity, and developmental timing in the community. This allows researchers to test explanations about computational constraints while acknowledging distinct expression in each language Leonard (2014) Kidd et al. (2018) Bishop et al. (2017).
Sentence Comprehension, Integration, and Linguistic Load
Comprehending a sentence requires more than recognizing its component words. Listeners identify grammatical relations, integrate meanings, maintain relevant dependencies, and use discourse context to select plausible interpretations. Children with DLD may be especially challenged by long or structurally complex utterances, instructions containing multiple relations, and sentences in which word order or morphology must override everyday expectations. Difficulty can be obscured by contextually plausible guesses: a child may act correctly when a familiar routine supplies the intended meaning but struggle when syntax is the decisive cue. Controlled comprehension experiments therefore include contrasts that separate linguistic interpretation from general world knowledge Leonard (2014) Archibald & Gathercole (2006) Bishop (2006).
Sentence repetition is often used as a compact measure because exact reproduction typically requires lexical knowledge, structural representations, perception, and temporary retention. Ward and colleagues' systematic review and meta-analysis assessed its accuracy in discriminating children with DLD from comparison groups, finding that the usefulness of tasks depends on construction and scoring choices. Such a task can support assessment but should not be presented as a direct or exclusive measurement of grammatical competence. The examiner must consider dialect, hearing, attention, memory load, age, and the linguistic familiarity of the stimuli before interpreting errors as evidence of developmental language impairment Ward et al. (2024) Bishop et al. (2017) Archibald & Gathercole (2006).
Listening demands vary systematically across ordinary situations. A classroom lecture offers fewer opportunities to request clarification than a one-to-one conversation, and rapid transitions between instructions may tax working memory or suppress opportunities to integrate complex sentences. Functional difficulty depends partly on these task demands. Researchers can measure the contribution of linguistic complexity by holding information content approximately constant while manipulating clause embedding, sentence length, or prosodic boundaries. Practitioners can document how comprehension changes with visual context, repetition, and processing time, while distinguishing temporary accommodation benefits from durable improvement in language learning Ebbels et al. (2019) Bishop et al. (2016) Kidd et al. (2018).
Discourse, Narrative, and Pragmatic Communication
Narrative language requires selecting relevant events, ordering them in a comprehensible sequence, maintaining reference, expressing causal relationships, and adjusting information for an audience. Children with DLD may show difficulty with global story organization, local grammatical cohesion, or the vocabulary needed to convey internal states and temporal transitions. Winters and colleagues synthesized narrative performance studies and documented a substantial average group disadvantage, alongside meaningful methodological variation. A narrative deficit can derive from multiple linguistic domains, so measures of story grammar, utterance complexity, and audience comprehension should be analyzed distinctly instead of combined into an unexplained global score Winters et al. (2022) Leonard (2014) Bishop et al. (2017).
Pragmatic functioning concerns the use of language in interaction: turn-taking, repair, inferential interpretation, reference to shared knowledge, and management of social goals. Its relationship with DLD is complex because children may struggle conversationally when underlying vocabulary or syntax makes rapid interpretation difficult. Social interaction is also shaped by peer responses and communication environments. Identical observable breakdowns can have different explanations, including insufficient linguistic comprehension, unfamiliarity with conversational norms, limited opportunities to participate, or other developmental conditions. A careful assessment integrates structured interaction, contextual observation, informant reports, and direct examination of language form and meaning Bishop et al. (2016) Bishop et al. (2017) Conti-Ramsden et al. (2018).
Narrative sampling provides a bridge between experimental validity and everyday communication, but it requires methodological discipline. Researchers must specify whether a child retells a familiar story, generates a novel narrative, answers prompts, or tells a story from pictures. Performance may change with topic familiarity, visual support, culture-specific narrative conventions, and examiner scaffolding. Scoring systems should have documented interrater reliability and appropriate age and language norms. Comparing narrative performance before and after intervention must additionally distinguish learning the practiced story from generalized improvement in discourse organization, an essential test of educational relevance Winters et al. (2022) Ebbels et al. (2019) Restrepo (1998).
Adolescent and Adult Developmental Outcomes
Developmental language disorder remains relevant after early childhood because increasingly complex academic, occupational, and interpersonal settings introduce new linguistic demands. At secondary school, students encounter disciplinary vocabulary, implicit classroom inference, written argument, and extended directions; in early adulthood, vocational preparation and employment require communication in less familiar social environments. Conti-Ramsden and colleagues documented heterogeneous education and employment outcomes in young adults with childhood DLD histories. This evidence supports longitudinal attention to functioning without presenting an individual's future opportunities as determined by a childhood diagnosis Conti-Ramsden et al. (2018) Bishop & Adams (1990) Bishop et al. (2017).
The manifestations of difficulty can change with development and compensation. A person may learn common grammatical constructions while continuing to expend substantial effort when interpreting unfamiliar or time-pressured language. Literacy, social confidence, organizational strategies, and institutional supports may shape whether a linguistic vulnerability becomes an everyday barrier. Consequently, longitudinal researchers need measures that remain meaningful across age ranges, including educational attainment, self-reported participation, communicative autonomy, and workplace requirements. Absolute improvement should be documented alongside relative comparisons, and studies should be alert to attrition that disproportionately excludes participants with the most complex difficulties Conti-Ramsden et al. (2018) Conti-Ramsden & Botting (1999) Ebbels et al. (2019).
Research on adult outcomes also demands respect for the distinction between statistical risk and personal capacity. An association between childhood language impairment and lower mean attainment may reflect indirect pathways involving schooling access, reading comprehension, peer experiences, and socioeconomic circumstances. Observational cohorts rarely identify a uniquely causal developmental mechanism without strong designs for confounding and mediation. Qualitative accounts and participatory research can clarify goals that laboratory measures miss, such as negotiating assistance and understanding institutional communication. The objective of life-course study is a more accurate account of functional pathways and support requirements across contexts Conti-Ramsden et al. (2018) Bishop & Snowling (2004) Bishop et al. (2016).
Competing Accounts of Developmental Language Difficulty
Explanatory theories of DLD differ in what they identify as the primary limitation. Some focus on the representation or learning of grammatical structure; others emphasize temporary memory, auditory timing, procedural sequence learning, statistical learning, or interactions among multiple vulnerabilities. A mechanism becomes scientifically useful when it predicts a pattern of performance across tasks and distinguishes that pattern from effects of age, prior language knowledge, and general task difficulty. An account of limited processing resources, for example, must specify which computations become less efficient and why some constructions are more vulnerable than others. Research has not established one necessary and sufficient mechanism shared by every child classified as having DLD Bishop (2006) Ullman & Pierpont (2005) Leonard (2014).
Models differ in explanatory level. A linguistic theory may describe an error as delayed acquisition of tense features; a cognitive theory may attribute the same error to limited processing capacity or weaker representations resulting from learning history. These proposals are compatible in some respects and rival hypotheses in others. Evidence for one level cannot automatically be taken as evidence for a proposed neural realization. Rigorous model comparison tests distinct predictions using independent tasks and specified developmental trajectories. Language background, stimulus frequency, item familiarity, and co-occurring attention or reading difficulties must be accounted for before a group contrast is interpreted as confirmation of a domain-general or language-specific theory Rice et al. (1995) Archibald & Gathercole (2006) Kidd et al. (2018).
The strongest synthesis is an explicit multilevel one. It asks which observable linguistic structures are difficult, which processing operations the relevant tasks require, how individual children learn from exposure, and whether biological or environmental factors explain meaningful variance. The answers may vary by developmental period and linguistic domain. Treating every association as causal would conceal the distinction between a risk correlate and an intervention target. Conversely, acknowledging causal heterogeneity does not eliminate coherent research questions; it motivates measurement that separates components and systematic testing of predicted interactions. Scientific explanation advances through discriminating evidence rather than the rhetorical search for a universal deficit Bishop (2006) Bishop et al. (2017) Norbury et al. (2016).
Auditory Temporal Processing and Speech Perception
Early auditory-processing accounts proposed that difficulty distinguishing brief or rapidly sequenced acoustic events may limit acquisition of speech contrasts and later language structure. Tallal and Piercy's experimental work on nonverbal auditory perception in children labeled as having developmental aphasia was influential because it proposed a measurable lower-level process that might affect many linguistic outcomes. The theory offers testable predictions about stimulus duration, temporal order, discrimination accuracy, and the relation of these performances to language learning. It remains necessary to distinguish the historical demonstration of task differences from proof that such differences cause a heterogeneous developmental language disorder Tallal & Piercy (1973) Bishop (2006).
A critical methodological question is whether auditory tasks measure perception itself, task comprehension, attention, auditory working memory, or response selection. If children must remember the order of several tones or learn a novel response mapping, difficulty can emerge even when the relevant acoustic contrasts are perceptually available. Auditory theories therefore require carefully matched control conditions and prospective evidence that early auditory sensitivity predicts later linguistic outcomes independently of shared confounders. The acoustic structure of languages also matters: changes in vowel duration, prosody, and rapidly changing consonant cues place different demands on learners. A generalized temporal explanation should survive tests across these dimensions Tallal & Piercy (1973) Archibald & Gathercole (2006) Kidd et al. (2018).
Auditory training is an especially demanding test of causal theory. Improvement on a practiced sound-sequencing task establishes learnability of that task, whereas clinically meaningful evidence requires transfer to untrained words, grammar, comprehension, and participation. A mechanism can be implicated without being the sole intervention target, and observed treatment gains may arise through alternative processes such as attention and motivation. Claims that correcting a low-level auditory weakness resolves DLD require stronger controlled evidence than correlations between test scores. The continuing value of the account lies in specifying sensory contributions and in challenging experiments to distinguish access to speech cues from broader language learning Bishop (2006) Law et al. (2004) Ebbels et al. (2019).
Phonological Working Memory and Temporary Linguistic Representations
Temporary representation of phonological information supports repetition of unfamiliar words, retention of instructions, and the integration of words across an utterance. Archibald and Gathercole studied short-term and working-memory performance in children with SLI, contributing to evidence of differences that extend beyond conventional vocabulary tests. Yet verbal memory tasks are partly linguistic: prior familiarity with syllable patterns and lexical structure can influence the very representations held in memory. A correlation between nonword repetition and language proficiency does not specify whether memory limitation precedes language difficulty or whether weaker linguistic knowledge makes nominally equivalent memory tasks harder Archibald & Gathercole (2006) Leonard (2014) Schwob et al. (2021).
Memory assessment gains interpretive value by manipulating phonological similarity, sequence length, wordlikeness, semantic support, and delay. A child may show good performance on highly familiar syllables and poorer performance on novel forms, suggesting that prior learning shapes apparent capacity. Comparing verbal with visual or spatial tasks can reveal whether a weakness is more specialized, but such comparisons must consider their different reliability and attentional requirements. Models should make predictions about the relationship between trial-level errors and the learning of new forms rather than treating a low span score as a complete explanatory result. Longitudinal measurement can explore temporal ordering without claiming automatic causal identification Archibald & Gathercole (2006) Kidd et al. (2018) Bishop (2006).
An educational consequence follows from the task analysis: instructions containing several unfamiliar terms may be difficult because vocabulary, syntax, and temporary memory demands accumulate. Visual reminders, shorter chunks, and opportunities to request repetition can improve immediate access even when underlying language learning proceeds slowly. Such accommodations should be evaluated through actual comprehension and participation, not inferred to cure memory mechanisms. The effectiveness of any training targeting temporary retention should be assessed against active controls and generalization measures. Functional accessibility, direct language teaching, and the mechanism of a child's errors are complementary but empirically distinct questions Ebbels et al. (2019) Bishop et al. (2016) Law et al. (2004).
Procedural Learning, Statistical Regularity, and Sequencing
The procedural deficit hypothesis proposed by Ullman and Pierpont relates language difficulty to dysfunction in systems supporting the learning and execution of rule-governed sequences and skills. The account connects aspects of grammar with broader procedural learning rather than restricting explanation to abstract linguistic representation. It yields testable predictions about sequence-learning performance, relationships with morphosyntactic difficulty, and dissociations between more explicit lexical knowledge and the acquisition of regular structures. The model's interest lies in its ability to generate cross-domain predictions; establishing a correlation between grammatical performance and another sequential task remains a preliminary step rather than evidence of one shared neural cause Ullman & Pierpont (2005) Leonard (2014) Bishop (2006).
Statistical learning accounts emphasize extraction of regularities from input, including the frequency of word sequences, transitional probabilities, distributional patterns, and correlations between linguistic forms and meanings. Such learning is necessary for typical language development, but a poor experimental statistical-learning score may reflect attention, fatigue, strategy, or the complexity of stimuli. The crucial question is whether tasks using artificial sequences predict real-language learning over time and whether differences remain when relevant confounds are controlled. Children acquire language through rich interactions rather than laboratory streams alone, so ecological generalization must be tested rather than assumed Kidd et al. (2018) Bishop (2006) Ullman & Pierpont (2005).
Mechanistic diversity is a central unresolved issue. Some children may experience difficulty when representations are weak or input is sparse, others when new patterns must be consolidated, and others when structures need rapid online retrieval. A broad learning-disorder hypothesis should accommodate the different domains and severity profiles observed in DLD without explaining every outcome retrospectively. Computational models can improve precision by estimating learning rates and transfer across training conditions, but model parameters should be identifiable and stable across tasks. Longitudinal tests of learning mechanisms, genetic moderation, and developmental change are needed to move beyond cross-sectional evidence Kidd et al. (2018) Conti-Ramsden & Botting (1999) Bishop et al. (2017).
Genetic, Neurodevelopmental, and Environmental Contributions
Developmental language abilities arise through interactions among inherited variation, neurological development, experiences, and educational environments. Genetic influences on language-related traits do not imply a single inherited DLD gene or a uniform mechanism for every diagnosed child. Bishop's analysis of possible causes emphasizes that developmental language difficulties can reflect multiple risk processes whose contributions depend on phenotype and developmental context. A useful causal research program distinguishes genetic association, molecular function, neural consequences, and behavioral expression. Each transition requires evidence that is stronger than a cross-sectional association between a trait and a biological marker Bishop (2006) Leonard (2014) Bishop et al. (2017).
Neural studies face the challenge of identifying signals specific to language learning rather than common to attention, movement, or general maturation. Differences in group-average anatomy or activation do not automatically define a diagnostic biomarker, especially when clinical samples contain variable language profiles and co-occurring conditions. Any proposed indicator needs replication, out-of-sample discrimination, and evidence of added value beyond well-characterized behavioral assessment. Developmental changes and compensatory systems make single-time-point brain measurements difficult to interpret etiologically. Biological plausibility should be integrated with behavioral tests and longitudinal data that specify whether a mechanism precedes, accompanies, or follows linguistic difficulty Ullman & Pierpont (2005) Bishop (2006) Norbury et al. (2016).
Environmental conditions influence the resources and opportunities available for language growth, including interaction, schooling, and access to assessment. Such conditions should be studied through measurable pathways and structural context, rather than used to attribute DLD to families or to confuse bilingual exposure with pathology. A child can have a developmental vulnerability and also experience differences in instructional language or unequal access to support. Statistical adjustment for socioeconomic and educational variables may reduce confounding, but it also risks concealing genuine social mechanisms if used indiscriminately. Clear causal diagrams and triangulation across designs improve interpretation of these interconnected pathways Bishop et al. (2016) Restrepo (1998) Norbury et al. (2016).
A Developmental Systems Synthesis of Language Learning
The linguistic and cognitive research programs can be brought together by distinguishing representation, processing, learning, and participation. Representation concerns the structures and meanings that a learner has acquired; processing describes the operations required to recognize and use them under temporal constraints; learning concerns how these capacities change with experience; participation concerns the opportunities and demands through which language matters in daily life. An observed error may emerge from one or several of these levels. This analytical organization provides a framework for evaluating evidence without claiming discovery of a unique biological mechanism or replacing established models with an untested universal theory Kidd et al. (2018) Bishop (2006) Bishop et al. (2017).
A developmental systems approach invites more discriminating experiments. A child with weak verb inflection might be tested with familiar and unfamiliar verbs, varying auditory presentation and memory load, while tracking acquisition across repeated sessions. Improvement limited to trained words differs from generalization to a new structural rule. Repeating such experiments at different ages could establish whether difficulties change as representations become more robust. Cross-linguistic extension tests the same computational hypothesis with language-appropriate structures. The resulting evidence would constrain linguistic and domain-general explanations rather than allowing either to absorb every possible outcome after observation Rice et al. (1995) Archibald & Gathercole (2006) Schwob & Skoruppa (2022).
The synthesis also recognizes that clinically meaningful progress need not correspond to normalization of all laboratory measures. Better access to classroom discussion, ability to express preferences, or improved understanding of complex instructions may be substantial functional outcomes despite continued differences on standardized language tests. Conversely, a laboratory gain that fails to transfer has limited practical significance. Evaluation should therefore combine primary linguistic change, generalization, participation, and meaningful longitudinal consequences while documenting intervention intensity and environmental conditions. This is an agenda for research integration grounded in existing findings and methodological distinctions Ebbels et al. (2019) Roberts & Kaiser (2011) Conti-Ramsden et al. (2018).
Clinical Assessment as Multidomain, Contextualized Measurement
Assessment of DLD asks whether linguistic difficulties persist, affect everyday functioning, and warrant specialized support. A robust evaluation combines developmental history, hearing and communication context, standardized assessment where appropriate, direct observation, language sampling, and perspectives of caregivers, teachers, and the child. Separate measures are needed for comprehension and expression because apparently competent conversational production can obscure difficulty understanding complex language. Test interpretation must include the quality of normative evidence, sensitivity to age, dialect, multilingual exposure, and the practical consequences of a false positive or false negative conclusion. No single low score fully captures the scientific or clinical construct Bishop et al. (2016) Bishop et al. (2017) Ebbels et al. (2019).
Assessment batteries typically sample vocabulary, sentence structure, discourse, and learning, but the meaning of their composite scores depends on the included tasks. Examiners should record what was measured, whether different subscales share stimulus or response demands, and how reliably they distinguish children with functional impairment. Language samples add ecological information but are sensitive to elicitation context, transcriber decisions, and the amount of speech collected. Repeated observation can separate stable difficulty from temporary inattention or unfamiliarity with a task. A diagnostic conclusion is an integration across evidence sources, rather than an attempt to make heterogeneous measurements appear perfectly interchangeable Ward et al. (2024) Winters et al. (2022) Restrepo (1998).
Validity includes the consequences of measurement. A cutoff with attractive group separation may nevertheless yield many false positives when used in a low-prevalence screening population, and a test developed in a highly selected clinic may perform differently in a school. Sensitivity, specificity, and predictive value should be reported with the population in which they were estimated, including confidence intervals and reference-standard limitations. Functional impact should be documented separately from test performance because some children develop compensatory strategies that mask difficulty under brief standardized conditions. An ethically and scientifically adequate assessment asks what findings imply for educational access and support Norbury et al. (2016) Bishop et al. (2017) Ortiz (2021).
Sentence Repetition, Nonword Repetition, and Diagnostic Accuracy
Sentence repetition and nonword repetition are frequently examined because they sample demanding forms of linguistic processing in relatively brief procedures. Sentence repetition involves perception, lexical and grammatical representations, temporary retention, and speech production; nonword repetition adds sensitivity to sound-sequence familiarity, phonotactics, and articulation. These dependencies explain why both can distinguish groups in some studies while remaining imperfect proxies for an underlying mechanism. Ward and colleagues' 2024 systematic review and meta-analysis addressed sentence-repetition diagnostic accuracy and task variation, reinforcing the need to describe stimulus construction and scoring criteria rather than treating every implementation as the same test Ward et al. (2024) Archibald & Gathercole (2006) Leonard (2014).
Schwob and colleagues' systematic review and meta-analysis evaluated nonword repetition for identifying DLD in monolingual and bilingual children, while Ortiz synthesized diagnostic accuracy among bilingual populations. Their contributions indicate the scientific relevance of task properties and language exposure: repeating a sequence that resembles familiar words draws on accumulated knowledge in addition to new phonological encoding. Comparisons between language-specific and crosslinguistic stimuli, including work by Schwob and Skoruppa in French and Portuguese, illustrate why instrument portability requires validation. Task choice should follow a documented question about the learner, linguistic context, and desired diagnostic inference Schwob et al. (2021) Ortiz (2021) Schwob & Skoruppa (2022).
A scientifically defensible account separates accuracy in research case-control samples from performance in unselected classrooms. When cases and controls are selected partly using language measures related to the index test, estimates may exaggerate diagnostic usefulness through circularity or spectrum bias. Cross-validation, blinded scoring, independent reference standards, and assessment of error costs are essential before recommending a cutoff for service decisions. A child who performs adequately on one repetition task may still experience a significant language disorder involving discourse or comprehension. Conversely, limited repetition performance in a recently learned language warrants evaluation of exposure rather than immediate classification Ward et al. (2024) Ortiz (2021) Bishop et al. (2017).
Bilingualism, Multilingual Development, and Diagnostic Equity
Multilingual acquisition introduces additional dimensions of experience into the measurement of language learning. Vocabulary, grammatical proficiency, and processing speed may differ across languages according to age of exposure, proficiency, literacy, conversational use, and the domains in which each language is practiced. Difference between a child's strongest language and the language used in school is not itself evidence of DLD. When a developmental disorder is present, concern may emerge in more than one language, but its visible forms depend on the grammar, exposure conditions, and assessment opportunities of each. CATALISE distinguishes limited knowledge of a language of instruction from persistent developmental language disorder Bishop et al. (2017) Restrepo (1998) Kidd et al. (2018).
Restrepo investigated identification among predominantly Spanish-speaking children, illustrating the importance of language samples and culturally appropriate information when standardized English-only measures are inadequate. More recent meta-analyses of nonword repetition in bilingual populations show the potential and limits of using compact language-learning tasks to reduce dependence on one particular language. Crosslinguistic assessment does not mean assuming that the same cutoff or error has the same meaning in every language. Age-matched monolingual norms may be inappropriate for children whose exposure is distributed differently, and dialect features must be distinguished from clinically relevant errors Restrepo (1998) Ortiz (2021) Schwob et al. (2021).
Equitable assessment requires a detailed account of language history, opportunities to learn, input quality, schooling, family communication preferences, and the specific linguistic structures of each tested language. Interpretation may benefit from trained bilingual clinicians, interpreters, dynamic assessment, convergent reports, and observation of learning over supported trials. The purpose is to distinguish an enduring acquisition difficulty from predictable differences due to linguistic experience while recognizing that both can coexist. Clinical services should document when instruments lack valid norms and avoid converting uncertainty into categorical certainty. Cross-language research that includes communities outside commonly studied English-speaking groups improves both fairness and scientific generalizability Schwob & Skoruppa (2022) Bishop et al. (2016) Restrepo (1998).
Developmental Dyslexia, Literacy, and Spoken-Language Overlap
Developmental dyslexia typically concerns difficulty acquiring accurate or fluent word-level reading and spelling, whereas DLD concerns spoken-language understanding or expression and may extend into written discourse. Bishop and Snowling analyzed the overlap and distinctions between dyslexia and SLI, arguing for multidimensional examination of phonological and broader language skills. Some children exhibit primarily a word-decoding difficulty, others have broad oral-language problems that undermine reading comprehension, and still others have overlapping profiles. The presence of both difficulties is theoretically informative because it prevents a simplistic partition between supposedly separate language and literacy systems Bishop & Snowling (2004) Bishop & Adams (1990) Leonard (2014).
Reading comprehension draws on decoding accuracy, lexical and grammatical understanding, discourse integration, and background knowledge. A child with strong decoding may struggle with text meaning if vocabulary and syntax are limited, while one with good spoken-language comprehension can find decoding inefficient. Longitudinal research is needed to distinguish which difficulties preceded literacy instruction from those amplified by reduced access to written language. Measurement should include word recognition, decoding, reading comprehension, spelling, and oral narrative separately. Educational intervention can then target the actual bottleneck instead of assuming that one standardized reading score defines the developmental mechanism Bishop & Snowling (2004) Bishop & Adams (1990) Bishop et al. (2017).
The practical implications extend beyond early literacy instruction. Adolescents with DLD may face substantial demands in interpreting examination questions, following written procedures, learning disciplinary terminology, and constructing extended written explanations. A profile that looks mild in ordinary conversation can produce serious academic constraints. Literacy services and spoken-language support may therefore need coordinated goals, especially when the child has distinct weaknesses in both domains. Research that evaluates intervention success should include transfer to comprehension of unfamiliar school materials and participation in learning, rather than focusing solely on accuracy with trained items or the surface fluency of oral responses Conti-Ramsden et al. (2018) Ebbels et al. (2019) Bishop & Snowling (2004).
Other Developmental Conditions and Differential Formulation
Language challenges occur in many developmental contexts, including conditions with identifiable biomedical etiologies and co-occurring attentional, learning, or behavioral difficulties. The CATALISE framework separates differentiating conditions from co-occurring developmental disorders and from risk factors. Such separation improves clarity without denying access to support when clinical presentations are complex. The assessment problem is to identify the child's language profile and functional needs while documenting other factors that influence performance. Descriptions should state whether difficulties concern language comprehension, expression, speech production, social communication, hearing, general learning, or several domains together Bishop et al. (2017) Bishop et al. (2016) Norbury et al. (2016).
An assessment of receptive language may require attention to hearing status and to whether spoken directions are supported by gesture or context. A child who has had limited access to a particular language might perform poorly on language-specific vocabulary tasks despite age-appropriate learning in another language. Co-occurring attention difficulty can affect consistency across trials, while general cognitive delays may change the developmental profile without making language intervention irrelevant. Diagnostic reasoning therefore integrates history, direct observation, linguistic tests, relevant medical evaluation, and ecological evidence. The absence of a sharp score boundary calls for transparent reasoning about prognosis and participation Bishop et al. (2016) Restrepo (1998) Ebbels et al. (2019).
The scientific literature also distinguishes impairment classification from causal explanation. A co-occurring reading problem does not establish that reading causes DLD, just as a correlation with working memory does not establish one cognitive etiology. Evidence of shared vulnerabilities needs longitudinal, family, and intervention research with appropriate comparison groups. Exclusionary criteria inherited from older SLI studies can alter the observed strength of associations and should be described when comparing historical results. A comprehensive clinical formulation remains adaptable when new evidence about language, cognition, or biomedical conditions emerges over time Bishop & Snowling (2004) Bishop (2006) Bishop et al. (2017).
Measurement Quality, Replication, and Transportability
Evidence quality in DLD research depends on whether investigators can reproduce procedures and distinguish scientific effects from sample selection. Reports should specify recruitment, age, language exposure, diagnostic criteria, co-occurring conditions, psychometric properties, and the full scoring scheme. Reliability should be assessed for standardized tests and examiner-scored speech or narrative samples. Predefined hypotheses and independent replication help reduce the interpretive flexibility created by many possible outcome measures. A significant group difference offers limited information without a magnitude, precision interval, and account of potential confounders. Clinical relevance requires evidence that a task meaningfully predicts or changes functional outcomes Ward et al. (2024) Winters et al. (2022) Norbury et al. (2016).
Generalization across cultures and languages is particularly challenging when many commonly used measures are English-specific and normed on restricted populations. Research portability should be treated as an empirical question involving adaptation, validation, measurement invariance, and equivalent sampling rather than simple translation. School selection, clinical referral, regional service availability, and parental language preferences may produce systematic differences among groups. Diagnostic models trained on highly selected research cohorts can exhibit reduced accuracy in ordinary educational settings. Independent evaluation across populations and transparent reporting of errors improve the practical usefulness of psycholinguistic measurements Ortiz (2021) Restrepo (1998) Schwob & Skoruppa (2022).
Research synthesis should distinguish the strength of convergence from the number of publications describing similar findings. Multiple papers drawing on one cohort do not constitute independent replication, and repeated use of a single diagnostic instrument can propagate shared measurement bias. Meta-analysis can summarize average performance while still concealing substantial heterogeneity in participants, task designs, and scoring. Narrative synthesis, as undertaken here, can connect these distinctions but cannot replace a prospectively registered systematic search or calculate a new pooled effect. The resulting evidence is best presented with explicit attention to designs, limitations, and the questions each study can answer Ward et al. (2024) Winters et al. (2022) Bishop et al. (2016).
Early Identification, Screening, and Decision Thresholds
Early identification aims to recognize language-development risk while avoiding unwarranted classification of children undergoing ordinary variation. Screening questions and brief tasks can identify candidates for further evaluation, but their clinical significance depends on the age distribution, underlying prevalence, and likelihood that a difficulty will persist. A screening tool optimized for sensitivity may appropriately trigger more assessments while producing many false positives; a highly specific tool may miss children whose comprehension difficulty is less conspicuous. The threshold for referral and the threshold for assigning a developmental diagnosis need different justifications and should not be conflated Tomblin et al. (1997) Bishop et al. (2016) Norbury et al. (2016).
A useful decision model considers both the probable future course and the opportunity cost of waiting. For a toddler whose expressive vocabulary is delayed but receptive language is strong, repeated monitoring may yield better information than one categorical assessment. Broad difficulties in understanding and producing language can justify earlier comprehensive evaluation, even while causal explanations remain uncertain. The research priority is to identify combinations of prospective predictors that improve discrimination beyond individual milestone counts. Longitudinal cohorts must report loss to follow-up, changes in measurement across age, and how the outcome was defined to avoid exaggerated predictive accuracy Bishop & Adams (1990) Conti-Ramsden & Botting (1999) Bishop et al. (2017).
Decision thresholds also reflect service systems. Where specialist resources are scarce, eligibility criteria may inadvertently become substitutes for scientific definitions and exclude children with genuine functional limitations. Transparent assessment separates descriptive evidence from administrative decisions about service access. Screening studies should document the consequences of incorrect classification, representativeness of language communities, and whether follow-up evaluations were performed independently of the screening result. Families and educators need clear communication about uncertainty, the reasons for monitoring, and the practical supports available before formal identification is established Norbury et al. (2016) Restrepo (1998) Ebbels et al. (2019).
Dynamic Assessment and Responsiveness to Mediated Learning
Dynamic assessment considers how a learner responds when a language task is supported through prompts, modeled examples, explanation, and opportunities to revise an answer. Its conceptual appeal lies in separating unfamiliarity with the examiner's expected language from the capacity to learn under structured assistance. In multilingual contexts this distinction is particularly relevant because static scores often reflect unequal exposure to the particular language of assessment. Yet dynamic procedures vary in instructional content, amount of support, scoring, and reliability, and their diagnostic performance must be established for the actual child population being evaluated Restrepo (1998) Bishop et al. (2016) Schwob & Skoruppa (2022).
A research-quality dynamic assessment specifies the pretest performance, instructional protocol, targeted structure, number of opportunities, and post-teaching measurement. The examiner should avoid simply rewarding compliance with familiar prompts or teaching the precise items that will count as evidence of broad learning. Transfer to novel instances and retention after delay offer stronger indicators than immediate imitation. Although such designs can illuminate modifiability, they require blinded scoring and predictive validation before replacing established elements of clinical assessment. Differences in learning rate may themselves be influenced by task comprehension, attention, and prior knowledge Kidd et al. (2018) Archibald & Gathercole (2006) Ward et al. (2024).
Dynamic measures are best interpreted as one component of a wider developmental formulation. A learner who improves quickly with scaffolding may still require structured opportunities to transfer skills into less predictable classroom interaction; a learner who improves slowly may nevertheless make substantial progress with extended teaching. Examining response patterns over repeated sessions allows research on dosage and developmental moderators, but an individual's few assisted trials cannot identify a permanent neurocognitive trait. The practical value lies in connecting assessment with teachable goals while preserving the distinction between an immediate task response, stable language learning, and everyday participation Ebbels et al. (2019) Hadley et al. (2026) Bishop et al. (2017).
Cross-Linguistic Generalization and Construct Equivalence
Comparing DLD across languages requires identifying which components of a construct can be meaningfully shared and which must be adapted. A measure of grammatical tense marking may be sensitive in one language but irrelevant in another, while complex sentence comprehension can be realized through different morphosyntactic cues. Structural differences influence what errors are possible, how frequently a form appears in input, and when a typically developing child is expected to master it. Generalization should therefore operate at the level of a specified theoretical prediction rather than the identical surface item. Detailed linguistic analyses are central to equitable diagnostic science Leonard (2014) Rice et al. (1995) Schwob & Skoruppa (2022).
Construct equivalence also demands attention to score interpretation. The same raw number of repeated syllables can represent different difficulty if test stimuli differ in familiarity and articulatory complexity, and a narrative rubric can privilege a culture-specific form of storytelling. Within- and between-group comparisons require evidence of reliability, comparable elicitation conditions, and either measurement invariance or an explicit justification for non-equivalent measures. The bilingual nonword-repetition literature provides examples of how accuracy varies with stimulus design; these findings are a reason for further validation rather than a universal cross-language diagnostic cutoff Ortiz (2021) Schwob et al. (2021) Restrepo (1998).
Research programs that include typologically varied languages can discriminate explanatory theories. If a procedural-learning limitation is hypothesized, investigators can ask whether different morphological systems show predictable sensitivity to sequential regularity. If phonological memory is implicated, comparable manipulation of syllable familiarity should be examined within different phonotactic systems. These comparisons require adequate samples, qualified language expertise, and prospective analysis plans. Their results may support shared learning mechanisms while demonstrating language-specific behavioral manifestations. The objective is a generalizable science in which diversity provides experimental leverage rather than an obstacle to be removed Ullman & Pierpont (2005) Archibald & Gathercole (2006) Kidd et al. (2018).
Longitudinal Causal Inference and Developmental Mediation
Developmental trajectories create opportunities to examine causal ordering, but repeated measurement alone does not remove confounding. A child's early grammatical score may predict later reading comprehension because language directly supports text interpretation, because other cognitive abilities influence both, or because educational access shapes both development and assessment. Causal models should identify possible common causes, time-varying influences, and mediation pathways before interpreting a statistical association. Longitudinal studies offer the advantage of temporal evidence and within-person change while remaining sensitive to missing data, test revisions, and the stability of measurement across age Bishop & Adams (1990) Bishop & Snowling (2004) Conti-Ramsden et al. (2018).
Strong designs would follow multiple language domains, literacy, educational experiences, and functional participation at several developmental periods. Predictors should be distinguished from correlates measured after the outcome has begun to emerge. Researchers need to test whether a proposed mechanism adds predictive information beyond prior language performance and relevant contextual variables, and whether predictive relations vary across subgroups. Changes in classification described by Conti-Ramsden and Botting make latent trajectory approaches attractive, although model fit by itself does not establish a natural diagnostic subtype. Results should be interpreted with uncertainty and validated in independent cohorts Conti-Ramsden & Botting (1999) Norbury et al. (2016) Kidd et al. (2018).
Intervention trials can contribute causal evidence when they deliberately change an exposure or teach a specific language target, but they also have boundaries. A treatment that improves grammatical scores establishes an effect of that instructional procedure and comparison condition on measured outcomes; it does not necessarily identify the original developmental cause of DLD. Mechanistic mediation claims require independent measurement and appropriate temporal ordering. The 2026 caregiver-mediated trial illustrates both the promise of randomized designs and the importance of differentiating grammar outcomes, vocabulary outcomes, population representation, and interim follow-up. Combining such trials with developmental cohorts may progressively sharpen explanation Hadley et al. (2026) Roberts & Kaiser (2011) Ebbels et al. (2019).
Artificial Systems, Language Tools, and a Changing Communication Ecology
Digital communication increasingly includes automated captions, speech recognition, translation, reading assistance, and generative language tools. These systems can alter how people encounter information, formulate messages, and receive feedback, including for learners who face persistent language difficulties. Their potential relevance to DLD lies in practical questions about accessibility, instructional design, feedback reliability, and communicative autonomy. The effectiveness and safety of any particular artificial tool require direct evaluation with the intended population. Familiarity with computational language generation should not be treated as evidence that such systems reproduce a child's developmental language mechanisms or provide independently validated clinical interventions Ebbels et al. (2019) Kidd et al. (2018) Bishop et al. (2017).
A scientifically appropriate evaluation should specify the tool's function and intended outcome. Automated simplification may reduce sentence complexity yet remove essential meaning; speech-to-text may facilitate classroom participation while introducing errors that users must identify; generated practice items may increase exposure but vary in accuracy and developmental appropriateness. Studies can compare assisted and unassisted comprehension, attention demands, transfer to unsupported language tasks, user control, and the reliability of responses across linguistic varieties. Accessibility benefits should be studied separately from durable changes in underlying language knowledge. Privacy and accessibility requirements are particularly important when children's language samples become inputs to digital services Bishop et al. (2016) Ward et al. (2024) Ebbels et al. (2019).
The Artificial Era introduces a wider ecology of human–artificial communication in which linguistic support may be personalized and available across settings. The relevant scientific task is to determine which interactions help children and adults participate more effectively, which create dependency or misleading feedback, and how outcomes differ by language background and developmental profile. The category of artificial cognition belongs to a separate interdisciplinary inquiry; conclusions about human DLD should be grounded in human developmental and clinical evidence. Comparative research can be valuable when definitions of learning, context, and assistance are explicit, and when claims about human benefit are tested against concrete outcomes Kidd et al. (2018) Conti-Ramsden et al. (2018) Ebbels et al. (2019).
Research Priorities and a Reproducible Evidence Agenda
A credible research agenda begins with the heterogeneity that existing studies document. Investigators should recruit linguistically diverse cohorts and specify which linguistic domains, functional outcomes, and co-occurring difficulties define each sample. Multisite prospective designs can help evaluate whether explanatory associations reproduce across settings and remain stable through development. Publicly documented instruments, analysis plans, and uncertainty estimates reduce the risk that a preferred cognitive theory is affirmed by post hoc choice among multiple outcomes. Research should also preserve meaningful information about schooling, access, socioeconomic context, and the languages used at home and in educational institutions Norbury et al. (2016) Conti-Ramsden & Botting (1999) Bishop et al. (2017).
Mechanistic research needs interventions and tasks capable of distinguishing competing predictions. Auditory perception, short-term memory, grammatical learning, and sequential learning should be measured with designs that separate language knowledge from shared task demands. Studies of treatment should report intensity, fidelity, the duration of follow-up, and whether change extends from trained examples to novel language and real communication. Outcomes must be chosen for practical relevance as well as statistical sensitivity. Combining standardized language measures, ecological samples, and carefully defined participation endpoints improves the connection between basic science and the daily consequences of DLD Tallal & Piercy (1973) Archibald & Gathercole (2006) Hadley et al. (2026).
Evidence synthesis itself should be reproducible. A future systematic review of this literature should prospectively specify the research question, databases, search dates, eligibility criteria, risk-of-bias procedures, and planned heterogeneity analyses; a meta-analysis additionally needs compatible effect estimates and justified statistical aggregation. The present monograph is a critical narrative integration with traceable sources, designed to state hypotheses, documented findings, and clinical consensus at their appropriate evidentiary levels. Its durable contribution is the organization of distinct scientific questions that can be revisited as longitudinal cohorts, multilingual measures, and intervention trials provide new evidence Law et al. (2004) Ward et al. (2024) Winters et al. (2022).
Principles of Language Intervention and Meaningful Outcomes
Language intervention seeks measurable improvement in communication, learning, and participation through procedures matched to a child's developmental profile and everyday context. Treatment targets may include vocabulary, grammatical constructions, comprehension strategies, narrative organization, or opportunities for successful interaction. The evidence-based pathways described by Ebbels and colleagues emphasize developmental needs and the available research rather than assuming one therapy applies equally to all language profiles. A clear intervention account specifies which linguistic behavior is expected to change, how it is taught, how progress is measured, and whether gains transfer beyond practiced activities. The relevant outcome is both linguistic learning and its practical availability in daily life Ebbels et al. (2019) Law et al. (2004) Bishop et al. (2017).
Intervention evidence must be evaluated by design. Randomized allocation can estimate the effect of a particular instructional package under specified conditions; nonrandomized comparisons and case series can provide useful clinical observations but are more vulnerable to selection and maturation effects. Studies should report frequency, duration, the training of adults delivering support, adherence, and the exact comparison condition. A child improving during an intervention is expected in development, so improvement alone cannot identify a treatment effect. Immediate changes, maintenance after treatment, and generalization to untrained language and settings are distinct outcomes requiring separate measurements Law et al. (2004) Roberts & Kaiser (2011) Hadley et al. (2026).
Intervention is not a singular event separated from ordinary language environments. Children participate in family conversation, classrooms, peer groups, and increasingly digital settings that differ in linguistic demand and opportunities to practice. Responsive interaction, accessible classroom speech, and targeted structured instruction can address different barriers. The precise combination should depend on assessment and evidence, with attention to what the child values and can use. The scientific task is to study dose, format, age, targets, and contextual moderators, avoiding the assumption that a successful outcome in one narrow domain establishes generalized recovery across language functions Ebbels et al. (2019) Conti-Ramsden et al. (2018) Bishop et al. (2016).
Vocabulary, Grammar, and Narrative Treatment Evidence
Targeted intervention studies vary substantially in outcome domain and method. Vocabulary work may provide repeated semantic and phonological encounters with words, contextual examples, and opportunities to retrieve and use them; grammatical interventions may concentrate on controlled exposure, elicitation, recasts, or explicit attention to a construction. The meta-analysis by Law and colleagues synthesized treatment studies in children with developmental speech and language difficulties, illustrating the value of comparing outcomes while acknowledging differences across participants and procedures. Its conclusions must be interpreted in relation to the terminology, interventions, and study quality available at its publication date Law et al. (2004) Ebbels et al. (2019) Leonard (2014).
Language-specific gains should be examined separately from broader standardized scores. Teaching a child several verbs may improve their use in a structured session while leaving spontaneous application and comprehension of new verbs uncertain. Grammatical change is most persuasive when improvement appears with untrained lexical items, in multiple contexts, and after the immediate teaching phase. Narrative work similarly distinguishes the practiced content of stories from general organization and causal explanation. A program can be successful for a narrow target and still require additional support for other domains. Reporting this specificity makes the evidence more useful to children and clinicians Rice et al. (1995) Winters et al. (2022) Ebbels et al. (2019).
Several methodological barriers complicate comparison. Investigators may use different definitions of DLD, recruit children at different ages, or combine receptive and expressive outcomes into one aggregate. Small treatment samples make estimates sensitive to baseline differences, and the absence of blinded assessments can favor expected gains. Reviews must therefore distinguish statistical significance from clinically meaningful improvement and examine the duration of follow-up. Future trials should specify a primary language outcome, reliable functional participation measures, active controls where feasible, and procedures for monitoring maintenance and transfer across settings Law et al. (2004) Roberts & Kaiser (2011) Hadley et al. (2026).
Caregiver-Mediated Interventions and the 2026 Randomized Trial
Caregivers can implement language-learning strategies through everyday interaction when the procedures are well specified, supported, and adapted to family contexts. Roberts and Kaiser reviewed parent-implemented language interventions in a meta-analysis and found evidence relevant to the development of child language outcomes, while highlighting intervention and study differences. Caregiver-mediated practice is a family-participation model, not an explanation that families caused the child's language disorder. Its scientific rationale is to increase useful learning opportunities in recurring, meaningful contexts. Evaluation should distinguish caregiver strategy adoption from changes in children's language knowledge and spontaneous use Roberts & Kaiser (2011) Ebbels et al. (2019) Bishop et al. (2017).
Hadley and colleagues reported a 2026 randomized trial of an enhanced milieu teaching approach with sentence-focused components, enrolling 108 English-home-language children at risk for DLD and their caregivers. Caregivers received telehealth coaching and children were compared with a control group. The reported structural equation model found a significant short-term group effect for a grammar latent variable at 42 months (p = .047), while the corresponding vocabulary latent-variable effect at 36 months was not significant (p = .458). A secondary analysis detected a difference on one receptive vocabulary test, emphasizing the importance of prespecified primary outcomes and transparent interpretation of supplementary findings Hadley et al. (2026).
The trial contributes unusually recent controlled evidence on developmental timing and caregiver-supported grammar learning, but its generalizability requires careful qualification. Participants primarily came from White, non-Hispanic, college-educated families and spoke only English at home; the results should not automatically be generalized to multilingual children or other educational systems. The study assessed interim outcomes during a longer multiphase intervention and did not establish universal long-term remediation. Grammar difficulties remained in many children despite detected treatment effects. The appropriate synthesis is that one defined program improved certain measured outcomes relative to control under studied conditions, with additional follow-up and independent replication needed Hadley et al. (2026) Roberts & Kaiser (2011) Ebbels et al. (2019).
Educational Access, Accommodations, and Coordinated Support
Educational participation depends on the match between students' language profiles and demands embedded in classroom instruction, assessment, and peer interaction. A student with DLD may benefit from explicit explanation of unfamiliar vocabulary, visual organization of multi-step instructions, opportunities for comprehension checks, and additional time to formulate responses. These adjustments should be evaluated in conjunction with direct language teaching rather than used as substitutes for identifying a persistent learning need. The goal is access to curricular meaning, accurate demonstration of knowledge, and communication participation. Documentation should distinguish increased access from changes in the underlying linguistic skills being learned Bishop et al. (2016) Ebbels et al. (2019) Conti-Ramsden et al. (2018).
Coordination between speech-language specialists, educators, families, and the learner is important because the same linguistic problem can appear differently across settings. Classroom teachers may notice errors with subject-specific terminology or written explanation that are absent in a brief clinic task, while clinicians may identify particular sentence structures that account for classroom misunderstanding. Functional goals can be operationalized by observing whether the student follows novel instructions, contributes to discussion, or comprehends unfamiliar texts. Outcomes should be defined collaboratively and evaluated at appropriate intervals. This approach links the research object to observable educational functioning without substituting an impressionistic classroom judgment for a comprehensive assessment Ebbels et al. (2019) Bishop & Snowling (2004) Bishop et al. (2017).
Support requires developmental continuity. Transitions between primary and secondary education, higher education, and employment alter the communication ecology and introduce new demands on interpretation, planning, and self-advocacy. Young-adult outcome research documents variation in education and employment among people with DLD histories and underscores the need to study successful pathways as well as risk. Research should examine whether support produces durable gains in participation and autonomy and how institutional accessibility affects outcomes. A diagnosis can guide planning while the person's own goals and communication strengths remain central to practical decisions Conti-Ramsden et al. (2018) Bishop et al. (2017) Kidd et al. (2018).
Critical Scientific Discussion: Evidence, Heterogeneity, and Open Problems
The central scientific challenge is the coexistence of measurable regularities and substantial individual heterogeneity. Grammatical, lexical, memory, and narrative group differences are reproducible research targets, yet they do not imply that every affected individual shares one profile or causal route. Longitudinal changes in language classifications further challenge fixed subtype models. A successful explanatory program must say which observations are expected under a mechanism, which would count against it, and whether predictions generalize across linguistic systems. The historical distinction between a phonological risk marker and an extended optional-infinitive grammatical profile illustrates how different findings can coexist without collapsing into a single dimension Conti-Ramsden & Botting (1999) Rice et al. (1995) Bishop (2006).
Important evidence gaps remain in causal identification. Behavioral test scores often combine multiple abilities; epidemiological associations can reflect confounding; neurodevelopmental group differences may arise through different routes; and early diagnostic thresholds can be sensitive to a child's linguistic experience. Stronger work requires prospective multi-wave samples, independently measured mechanisms, transparent inclusion rules, language-representative recruitment, and preregistered tests of competing predictions. Intervention experiments offer evidence about modifiability but do not necessarily establish why the difficulty originally arose. Claims about a biological cause therefore should be connected to discriminating longitudinal or intervention-based evidence rather than inference from descriptive classifications Norbury et al. (2016) Ullman & Pierpont (2005) Kidd et al. (2018).
Research equity is part of scientific validity. Instruments normed on a narrow set of language communities and education systems can reproduce apparent disparities in classification, reducing both accuracy and service access. The evidence base needs validation in less-studied languages, diverse bilingual profiles, different socioeconomic conditions, and adolescents or adults whose needs are less visible. Future syntheses should quantify diagnostic performance in representative settings and compare the consequences of misclassification. Meta-analyses of narrative ability and repetition tasks demonstrate the value of aggregating evidence but also reveal how definitions and procedures shape average results. Their heterogeneity is informative, not merely statistical noise Ortiz (2021) Ward et al. (2024) Winters et al. (2022).
Conclusion
Developmental language disorder is a persistent, functionally important pattern of difficulty acquiring and using language that can affect comprehension, expression, learning, communication, and participation across the life course. Its contemporary definition rests on developmental prognosis and functional impact while recognizing that co-occurring difficulties and diverse language-learning histories shape the observed profile. The field has advanced from narrow historical categories toward a framework that allows population research, psycholinguistic experiments, and practical assessment to address related but distinct questions. Its evidence supports substantial developmental variability and the need for individualized language profiles rather than a universal behavioral signature Bishop et al. (2017) Norbury et al. (2016) Conti-Ramsden et al. (2018).
Scientific explanation requires integrating the structures of language with how they are processed, learned, and deployed in actual communication. Grammatical investigations, memory and auditory-processing experiments, procedural-learning hypotheses, and crosslinguistic assessment each contribute discriminating observations, while none establishes one mechanism sufficient for every individual. Robust inference depends on clear operational definitions, independent comparison of predictions, accurate language-history measurement, longitudinal designs, and validation across communities. In clinical and educational settings, combining multiple evidence sources protects against confusing language difference, a co-occurring condition, and developmental language disorder Rice et al. (1995) Archibald & Gathercole (2006) Schwob et al. (2021).
Treatment research offers grounds for targeted support and for further trials of particular intervention packages, including recent randomized evidence for caregiver-mediated grammar outcomes. The unanswered questions concern generalization, dosage, equitable identification, developmental timing, mechanisms of response, and outcomes that matter to children and adults. Future research should combine precise linguistic measurements with functional participation and life-course evidence, examine diverse languages and settings, and identify which interventions work for which learners under what conditions. This synthesis provides a source-grounded account of the research landscape while preserving the distinction between established results, clinical consensus, competing hypotheses, and questions still open to scientific investigation Hadley et al. (2026) Ebbels et al. (2019) Bishop et al. (2017).
References
- Archibald, L. M. D., & Gathercole, S. E. (2006). Short-term and working memory in specific language impairment. International Journal of Language & Communication Disorders, 41(6), 675–693. https://doi.org/10.1080/13682820500442602
- Bishop, D. V. M., & Adams, C. (1990). A prospective study of the relationship between specific language impairment, phonological disorders and reading retardation. Journal of Child Psychology and Psychiatry, 31(7), 1027–1050. https://doi.org/10.1111/j.1469-7610.1990.tb00844.x
- Bishop, D. V. M., & Snowling, M. J. (2004). Developmental dyslexia and specific language impairment: Same or different? Psychological Bulletin, 130(6), 858–886. https://doi.org/10.1037/0033-2909.130.6.858
- Bishop, D. V. M. (2006). What causes specific language impairment in children? Current Directions in Psychological Science, 15(5), 217–221. https://doi.org/10.1111/j.1467-8721.2006.00439.x
- Bishop, D. V. M., Snowling, M. J., Thompson, P. A., & Greenhalgh, T. (2016). CATALISE: A multinational and multidisciplinary Delphi consensus study. Identifying language impairments in children. PLOS ONE, 11(7), e0158753. https://doi.org/10.1371/journal.pone.0158753
- Bishop, D. V. M., Snowling, M. J., Thompson, P. A., & Greenhalgh, T. (2017). Phase 2 of CATALISE: A multinational and multidisciplinary Delphi consensus study of problems with language development: Terminology. Journal of Child Psychology and Psychiatry, 58(10), 1068–1080. https://doi.org/10.1111/jcpp.12721
- Conti-Ramsden, G., & Botting, N. (1999). Classification of children with specific language impairment: Longitudinal considerations. Journal of Speech, Language, and Hearing Research, 42(5), 1195–1204. https://doi.org/10.1044/jslhr.4205.1195
- Conti-Ramsden, G., Durkin, K., Toseeb, U., Botting, N., & Pickles, A. (2018). Education and employment outcomes of young adults with a history of developmental language disorder. International Journal of Language & Communication Disorders, 53(2), 237–255. https://doi.org/10.1111/1460-6984.12338
- Ebbels, S. H., McCartney, E., Slonims, V., Dockrell, J. E., & Norbury, C. F. (2019). Evidence-based pathways to intervention for children with language disorders. International Journal of Language & Communication Disorders, 54(1), 3–19. https://doi.org/10.1111/1460-6984.12387
- Hadley, P. A., Roberts, M. Y., Harrington, E. K., Jones, M., Preza, T., Zanzinger, K. E., Kaat, A. J., Grauzer, J., & Kaiser, A. (2026). Short-term effects of a caregiver-mediated intervention for preschoolers at risk for developmental language disorder: A randomized controlled trial. Journal of Speech, Language, and Hearing Research, 69(7), 3290–3313. https://doi.org/10.1044/2026_JSLHR-25-00387
- Kidd, E., Donnelly, S., & Christiansen, M. H. (2018). Individual differences in language acquisition and processing. Trends in Cognitive Sciences, 22(2), 154–169. https://doi.org/10.1016/j.tics.2017.11.006
- Law, J., Garrett, Z., & Nye, C. (2004). The efficacy of treatment for children with developmental speech and language delay/disorder: A meta-analysis. Journal of Speech, Language, and Hearing Research, 47(4), 924–943. https://doi.org/10.1044/1092-4388(2004/069)
- Leonard, L. B. (2014). Children with specific language impairment (2nd ed.). MIT Press. https://doi.org/10.7551/mitpress/9152.001.0001
- Norbury, C. F., Gooch, D., Wray, C., Baird, G., Charman, T., Simonoff, E., Vamvakas, G., & Pickles, A. (2016). The impact of nonverbal ability on prevalence and clinical presentation of language disorder: Evidence from a population study. Journal of Child Psychology and Psychiatry, 57(11), 1247–1257. https://doi.org/10.1111/jcpp.12573
- Ortiz, J. A. (2021). Using nonword repetition to identify language impairment in bilingual children: A meta-analysis of diagnostic accuracy. American Journal of Speech-Language Pathology, 30(5), 2275–2295. https://doi.org/10.1044/2021_AJSLP-20-00237
- Restrepo, M. A. (1998). Identifiers of predominantly Spanish-speaking children with language impairment. Journal of Speech, Language, and Hearing Research, 41(6), 1398–1411. https://doi.org/10.1044/jslhr.4106.1398
- Rice, M. L., Wexler, K., & Cleave, P. L. (1995). Specific language impairment as a period of extended optional infinitive. Journal of Speech and Hearing Research, 38(4), 850–863. https://doi.org/10.1044/jshr.3804.850
- Roberts, M. Y., & Kaiser, A. P. (2011). The effectiveness of parent-implemented language interventions: A meta-analysis. American Journal of Speech-Language Pathology, 20(3), 180–199. https://doi.org/10.1044/1058-0360(2011/10-0055)
- Schwob, S., Eddé, L., Jacquin, L., Leboulanger, M., Picard, M., Ramos Oliveira, P., & Skoruppa, K. (2021). Using nonword repetition to identify developmental language disorder in monolingual and bilingual children: A systematic review and meta-analysis. Journal of Speech, Language, and Hearing Research, 64(9), 3578–3593. https://doi.org/10.1044/2021_JSLHR-20-00552
- Schwob, S., & Skoruppa, K. (2022). Detecting developmental language disorder in monolingual and bilingual children: Comparison of language-specific and crosslinguistic nonword repetition tasks in French and Portuguese. Journal of Speech, Language, and Hearing Research, 65(3), 1159–1165. https://doi.org/10.1044/2021_JSLHR-21-00017
- Tallal, P., & Piercy, M. (1973). Defects of non-verbal auditory perception in children with developmental aphasia. Nature, 241, 468–469. https://doi.org/10.1038/241468a0
- Tomblin, J. B., Records, N. L., Buckwalter, P., Zhang, X., Smith, E., & O'Brien, M. (1997). Prevalence of specific language impairment in kindergarten children. Journal of Speech, Language, and Hearing Research, 40(6), 1245–1260. https://doi.org/10.1044/jslhr.4006.1245
- Ullman, M. T., & Pierpont, E. I. (2005). Specific language impairment is not specific to language: The procedural deficit hypothesis. Cortex, 41(3), 399–433. https://doi.org/10.1016/S0010-9452(08)70276-4
- Ward, L., Polišenská, K., & Bannard, C. (2024). Sentence repetition as a diagnostic tool for developmental language disorder: A systematic review and meta-analysis. Journal of Speech, Language, and Hearing Research, 67(7), 2191–2221. https://doi.org/10.1044/2024_JSLHR-23-00490
- Winters, K. L., Jasso, J., Pustejovsky, J. E., & Byrd, C. T. (2022). Investigating narrative performance in children with developmental language disorder: A systematic review and meta-analysis. Journal of Speech, Language, and Hearing Research, 65(10), 3908–3929. https://doi.org/10.1044/2022_JSLHR-22-00017