Specific Language Disorder

Abstract

Specific language disorder is a language development disorder in which spoken language falls markedly below age expectations despite typical hearing, nonverbal intelligence, and no autism or overt neurological cause, affecting roughly seven percent of children entering school. Grammatical morphology is disproportionately impaired: tense and agreement marking is a sensitive clinical marker, and repeating nonwords or sentences separates affected from unaffected children. Competing accounts locate the deficit in a domain-specific grammar module or in a broader procedural-memory system that is not language-specific. A multinational consensus reframed the condition, dropping the requirement that nonverbal ability exceed language ability and adopting the term developmental language disorder. Three interactive demonstrations trace how the diagnostic criteria, the tense-marking signature, and the arithmetic of screening each shape who is identified.

Keywords: specific language disorder, developmental language disorder, clinical marker

Specific language disorder (SLD), historically termed specific language impairment (SLI) and now more often developmental language disorder (DLD), names a persistent difficulty acquiring spoken language that cannot be attributed to hearing loss, low nonverbal intelligence, autism, or a known neurological lesion (Leonard, 2014). MeSH classifies it as a disorder of language development, and both ICD-11 (as developmental language disorder) and DSM-5 (as language disorder, within communication disorders) list it as a diagnosable condition. The word specific in the older name asserted that language was selectively impaired while cognition was spared — a claim later research has substantially qualified.

What Specific Language Disorder Is

The defining feature is a gap between language and everything else. A child with SLD hears normally, scores in the typical range on nonverbal reasoning, shows no autistic social-communication profile, and has no identified brain injury or syndrome — yet produces and understands spoken language well below age peers (Leonard, 2014). The impairment can be predominantly expressive (producing language), predominantly receptive (understanding it), or mixed; receptive involvement generally predicts a poorer outcome. Although the proximate cause is unknown, the disorder is strongly heritable: twin studies find far higher concordance in identical than in fraternal pairs, placing it among the more genetically influenced developmental conditions (Bishop, North, & Donlan, 1995). Comorbidity is common rather than exceptional: the disorder frequently co-occurs with dyslexia, attention difficulties, and developmental coordination problems, and the CATALISE consensus was careful to specify which of these co-occurring conditions do and do not exclude the diagnosis.

Historically the diagnosis rested on a discrepancy criterion: language scores had to fall a set distance below nonverbal IQ, on the theory that a child whose language merely matched a low IQ had a general delay rather than a specific one. Epidemiological work undermined this. In the SCALES population study, children who met language criteria but had somewhat lower nonverbal ability differed from the narrowly defined group in severity, not in kind, and responded to the same interventions (Norbury et al., 2016). The nonverbal cutoff excluded children with identical needs for no principled reason.

Prevalence is substantial. A stratified kindergarten sample put specific language impairment at 7.4% of five-year-olds, higher in boys than girls but common in both (Tomblin et al., 1997); the SCALES study, using the broader modern criteria, found 7.58% (Norbury et al., 2016). Despite this, SLD attracts a fraction of the research funding and public awareness of rarer conditions, and most affected children are never formally identified (McGregor, 2020).

Clinical Markers

Because SLD has no biological test, diagnosis rests on behavioural markers — language behaviours that reliably separate affected from unaffected children. The most studied is finiteness marking: the grammatical morphemes that signal tense and agreement, such as past-tense -ed, third-person singular -s, and the auxiliaries is, was, and do. English-speaking children with SLD omit these markers long past the age at which peers use them reliably, producing forms such as yesterday he walk or she go home (Rice & Wexler, 1996).

Rice and Wexler framed this as the extended optional infinitive stage. Typically developing toddlers pass through a phase in which they treat tense marking as optional, sometimes producing a bare (infinitive-like) verb where an adult would inflect it. All children leave this stage; children with SLD remain in it far longer, so a composite of finite-verb-morphology accuracy discriminates the groups sharply between ages three and eight (Rice & Wexler, 1996).

Two processing tasks work as efficient screens. Nonword repetition asks the child to repeat pronounceable nonwords of increasing length; accuracy falls off steeply with syllable count in SLD, implicating phonological working memory. Sentence repetition asks the child to reproduce sentences verbatim and taxes the whole language system at once. Both show high sensitivity and specificity and are among the strongest single markers of the disorder (Conti-Ramsden et al., 2001).

Theoretical Accounts

Two families of theory compete over what is broken. Domain-specific accounts hold that language, or some component of grammar, is selectively impaired. Van der Lely's Grammatical-SLI hypothesis argues that a subgroup of children has a specific deficit in computational grammatical complexity — the hierarchical, structure-dependent operations underlying syntax, inflection, and the binding of pronouns — while non-linguistic cognition is spared (van der Lely, 2005). On this view the disorder reveals the modular architecture of the language faculty: damage one component and the rest stands.

Domain-general accounts reject the specificity premise. Ullman and Pierpont's procedural deficit hypothesis locates the problem in the procedural-memory system — the frontostriatal circuitry that also supports motor sequencing and rule-governed skill — and argues it is not specific to language at all (Ullman & Pierpont, 2005). Grammar suffers because rule-based combination depends on procedural memory, while the vocabulary that depends on declarative memory is relatively preserved and can even be recruited to compensate. The hypothesis predicts the subtle motor and sequencing difficulties frequently observed alongside the language deficit, which a purely linguistic account does not.

The evidence has moved against strict specificity. Children with SLD show measurable, if milder, difficulties in nonverbal domains, and the discrepancy-criterion research that removed the IQ cutoff was itself an admission that language impairment does not respect a clean cognitive boundary (Bishop, 2006). Most contemporary researchers treat the condition as a disorder in which language is the most affected system rather than the only affected one.

Worked Example

Consider a nonword-repetition screen applied to a whole kindergarten cohort. Suppose it has a sensitivity of 0.85 (it flags 85% of children who truly have SLD) and a specificity of 0.90 (it correctly clears 90% of those who do not), and take the base rate as 7%.

In a cohort of 1,000 children, 70 have SLD and 930 do not. The screen flags 70 × 0.85 = 59.5 true positives and misclassifies 930 × (1 − 0.90) = 93 unaffected children as false positives. Of every child the screen flags, the proportion who truly have SLD — the positive predictive value — is 59.5 ÷ (59.5 + 93) = 59.5 ÷ 152.5 ≈ 0.39.

So even a marker with excellent sensitivity and specificity identifies a flagged group that is only about 39% true cases at this base rate. The arithmetic is not a flaw in the marker; it is Bayes' rule acting on a 7% prevalence. It is why nonword and sentence repetition are used to screen — to decide who receives a full diagnostic language assessment — rather than to diagnose on their own. The interactive screening demonstration lets the reader vary sensitivity, specificity, and prevalence and watch the predictive value move.

Key Researchers

Dorothy V. M. Bishop (b. 1952). Emeritus Professor of Developmental Neuropsychology at Oxford; led the CATALISE consensus that reframed SLI as developmental language disorder. ORCID

Laurence B. Leonard (Purdue University). Author of the field's canonical monograph, Children with Specific Language Impairment, and a leading voice on cross-linguistic profiles of the disorder. No public ORCID record was located. Google Scholar

Karla K. McGregor (Boys Town National Research Hospital). Contemporary researcher on word learning and on the systematic under-identification of children with the disorder. ORCID

Courtenay Frazier Norbury (University College London). Led the SCALES population study that measured prevalence and questioned the nonverbal-IQ criterion. ORCID

Mabel L. Rice (University of Kansas). With Kenneth Wexler, established tense marking as a clinical marker through the extended-optional-infinitive account. ORCID

Michael T. Ullman (b. 1962). Georgetown neuroscientist; originator of the procedural deficit hypothesis linking the disorder to procedural memory rather than a language-specific module. No public ORCID record was located. Google Scholar

Discussion

SLD sits at the intersection of developmental psychology, linguistics, and clinical practice, and its history is a case study in how a diagnostic label shapes the science done under it. The word specific encoded a theoretical bet — that language could be impaired in isolation — and organised decades of research around proving or puzzling over that isolation. When population studies showed the isolation was never clean, the field faced a choice between defending a criterion that excluded needy children and revising the construct. It revised.

For cognitive psychology the disorder is a natural experiment on the architecture of language. If grammar can be selectively spared or impaired, that is evidence for modularity; if the deficit reliably drags in procedural memory and motor sequencing, that is evidence against it. The unresolved tension between the Grammatical-SLI and procedural-deficit accounts is precisely a disagreement about how much of language is its own system. The practical stakes are equally real: a common, treatable disorder that most affected children never have named is a public-health gap, and the diagnostic arithmetic above explains why better screening alone will not close it.

Current Directions

The most consequential recent development is terminological and conceptual rather than experimental. The two-phase CATALISE Delphi study convened dozens of specialists across several countries to reach consensus on how childhood language problems should be identified and named (Bishop et al., 2016). Phase 2 recommended abandoning the term specific language impairment in favour of developmental language disorder, dropping the nonverbal-IQ discrepancy requirement and clarifying which co-occurring conditions do and do not exclude the diagnosis (Bishop et al., 2017). The change has been widely adopted in research and increasingly in services.

A second thread is the problem of recognition. Contemporary work documents how a disorder affecting one child in fourteen remains largely invisible to the public, to educators, and to funders, and argues that the field's own past insistence on narrow criteria contributed to that invisibility (McGregor, 2020). Alongside this, population cohorts continue to refine prevalence estimates and to test whether the boundary with typical variation and with related conditions is best drawn categorically or dimensionally (Norbury et al., 2016).

Glossary

CATALISE consensus.

A multinational, multidisciplinary Delphi study that reached agreement on identifying and naming children's language disorders, leading to the term developmental language disorder.

Clinical marker.

A behaviour that reliably distinguishes children with a disorder from those without it, used to support diagnosis in the absence of a biological test.

Comorbidity.

The co-occurrence of two or more conditions in the same individual, such as language disorder alongside dyslexia or attention difficulties.

Developmental language disorder.

The current preferred term for a persistent language deficit of unknown cause, adopted by the CATALISE consensus in place of specific language impairment.

Discrepancy criterion.

The now-abandoned requirement that language scores fall a fixed distance below nonverbal IQ before a specific language impairment could be diagnosed.

Expressive language.

The production of spoken language, including vocabulary, grammar, and connected discourse, as distinct from its comprehension.

Extended optional infinitive.

A prolonged stage in which a child treats tense and agreement marking as optional, producing bare verb forms long past the typical age.

Grammatical-SLI.

A proposed subgroup whose impairment centres on computational grammatical complexity — hierarchical syntax, inflection, and binding — with non-linguistic cognition spared.

Nonverbal IQ.

A measure of reasoning that does not depend on language, historically used to establish that a language deficit was specific rather than part of a general delay.

Nonword repetition.

A task requiring a child to repeat pronounceable nonwords of increasing length, taxing phonological working memory and serving as a sensitive screen for the disorder.

Positive predictive value.

The proportion of individuals flagged by a screen who truly have the condition, which depends jointly on the screen's accuracy and the base rate.

Procedural deficit hypothesis.

The account that the disorder stems from impaired procedural memory in frontostriatal circuits, sparing declarative memory, and is not specific to language.

Receptive language.

The comprehension of spoken language; receptive involvement in the disorder generally predicts a poorer developmental outcome than expressive difficulty alone.

Sentence repetition.

A task requiring verbatim reproduction of spoken sentences, which loads the whole language system and is among the strongest single markers of the disorder.

Specific language impairment.

The older name for the disorder, asserting a language deficit selective to language and sparing general cognition — a claim later research substantially qualified.

Frequently Asked Questions

How is specific language disorder different from a speech disorder?
A speech disorder affects the articulation or fluency of sounds, whereas specific language disorder affects the underlying system of words, grammar, and meaning; a child may speak clearly yet still struggle to build or understand sentences (Leonard, 2014).

Does the child simply have a low IQ?
No. By definition nonverbal intelligence is in the typical range, and population research showed that even children whose nonverbal ability was somewhat lower differed only in severity, which is why the IQ-discrepancy criterion was dropped (Norbury et al., 2016).

How common is it?
It affects roughly 7% of children entering school, making it one of the most common developmental disorders, though it is far less recognised than rarer conditions (Tomblin et al., 1997).

Why was the name changed to developmental language disorder?
An international consensus study concluded that the word specific implied a cognitive isolation that the evidence did not support and that a clearer, less exclusionary label would improve identification and services (Bishop et al., 2017).

What are the best signs to look for?
Persistent omission of tense and agreement markers, difficulty repeating nonwords or sentences, and language that lags well behind peers despite normal hearing are among the most reliable indicators (Rice & Wexler, 1996).

Is the disorder specific to language, as the old name implied?
Probably not entirely; affected children often show mild difficulties in procedural memory and motor sequencing, which the procedural deficit hypothesis takes as evidence that the impairment extends beyond language (Ullman & Pierpont, 2005).

Can a single test diagnose it?
No. At a 7% base rate even an accurate screen flags a group that is mostly false positives, so screening identifies who needs a full diagnostic assessment rather than delivering the diagnosis itself (Conti-Ramsden et al., 2001).

Do children grow out of it?
Many improve with intervention, but for a substantial proportion the difficulties persist into adolescence and adulthood, and receptive involvement in particular predicts more lasting impact (McGregor, 2020).

Support Organizations

Organizations that provide information, assessment guidance, and advocacy for developmental language disorder and related communication conditions.

References