A practical look at new clinical tools for fairer, more equitable speech-language assessments.
A bilingual child can be labeled as disordered for speaking a different dialect, not for having a disorder. In a 2011 ASHA review, the English version of a bilingual screener misidentified 65% of Spanish- and English-dominant children as impaired. That kind of error is exactly what linguistic justice targets: assessment that treats language variety as a deficit rather than a difference.
The University of Tulsa is developing a linguistic-justice assessment to move SLP evaluation beyond monolingual benchmarks. This is a systems-level shift, not another cultural-competence add-on, and it changes which assessment tools clinicians can defend in real practice.
What does it mean, in practical terms, for a speech-language evaluation to be linguistically just? At its core, linguistic justice in speech-language pathology is the right to be assessed, diagnosed, and treated without penalty for one's language variety, dialect, or multilingual status. It is not the same as cultural competence, which centers on awareness and respect. Linguistic justice goes further by targeting structural bias in the tools clinicians use: standardized language assessments, eligibility formulas, and therapy goal selection.
A guiding principle is that language difference is not a disorder. A child who speaks African American English, Appalachian English, or Spanish-influenced English may produce patterns that a norm-referenced test scores as errors, even though those patterns are rule-governed and typical for their community. For example, a child who says "He run" or "She be at the store" is using grammatical structures common to some English varieties. When a clinician scores those as errors, the child may be mislabeled as having a language impairment. Deficit-based assessment treats the child's linguistic system as broken. Linguistic justice reframes the evaluation around communicative competence, not distance from a single standardized norm. For clinicians, this shift means asking whose norms define "typical" and whether a low score signals a difference, a disorder, or both. It also means refusing to write speech therapy goals that erase a child's dialect or accent in the name of "fixing" speech.
Traditional speech and language assessments can fail diverse clients through at least three forms of bias. Construct bias occurs when a test defines language or communication too narrowly, often treating standard English forms as the only correct target. Content bias enters when items draw on mainstream cultural knowledge or vocabulary that children from other backgrounds may not share. Norming-sample bias is common when the standardization group underrepresents bilingual children, dialect speakers, and children from minoritized language communities.5 If a child is not represented in the norming sample, the resulting standard score cannot be interpreted as a valid comparison.
ASHA guidance is clear that a dialect, accent, or use of a heritage language is not a disorder.1 Multilingualism itself is not a disorder either.2 A true language disorder is expected to show persistent, unexplained difficulties across all of the child's languages.45 Features that match the child's primary speech community or dialect are not errors.4 When a standardized test ignores this, a child can receive a low score on a bilingual speech assessment based on dialect or bilingual development rather than impairment. That score may then trigger special education placement even though the child has a language difference, not a disability, and would not be eligible for services on that basis.3
Because standardized scores alone can misclassify language differences as disorders, ASHA recommends dynamic assessment, ethnographic interviewing, and non-normed criterion measures.3 These tools examine the full communication profile across languages instead of relying on one culturally loaded score.2 This shift protects children from being labeled disordered simply because their language background does not match the test's assumptions.3
Published bias concerns are concentrated in a few widely used instruments. The table below summarizes four assessments with documented norming, dialect, or item-bias issues and the alternatives clinicians commonly recommend for a more linguistically just evaluation.
| Assessment | Age Range | Primary Bias Concern | Linguistically Just Alternative |
|---|---|---|---|
| Clinical Evaluation of Language Fundamentals, Fifth Edition (CELF-5) | 5;0-21;11 | Administered in General American English. Speakers of African American English, Spanish-influenced English, Appalachian English, and Chinese-influenced English may be penalized. Modified scoring can underidentify true language disability in AAE speakers, while unmodified scoring can overidentify typically developing AAE speakers. | Dynamic assessment, test-teach-retest, response analysis, and culturally and linguistically informed scoring. |
| Preschool Language Scales, Fifth Edition (PLS-5) | 0-7;11 | Norming sample was 78.9% Mainstream American English, and English-dominant bilingual children were underrepresented. Reviewers reported culturally and linguistically biased items that can disadvantage non-mainstream cultural and linguistic backgrounds. | Parent-child comparative analysis and other alternative assessment approaches for culturally and linguistically diverse children. |
| Peabody Picture Vocabulary Test, Third Edition (PPVT-III) | N/A | African American children and children whose mothers had low education levels tended to score lower, with a greater tendency than the EVT to unfairly identify African American children and children of low maternal education as at risk for language disorder. | Expressive Vocabulary Test (EVT), described as a better indicator of vocabulary skill in the cited study. |
| Test of Language Development-Primary (TOLD-P) | N/A | ASHA reports an item-bias analysis for the TOLD-P, indicating documented concern about item bias in this assessment. | Parent-child comparative analysis and dynamic assessment for culturally and linguistically diverse children. |
The field's center of gravity is moving away from monolingual benchmarks, but assessment tools have been slow to follow.
At the University of Tulsa, Clinical Assistant Professor Christy Hedges is developing a linguistic-justice assessment in the Department of Communication Sciences and Disorders. The project, announced in a University of Tulsa news release, is explicitly framed around the idea that all languages and language varieties are equal, so speakers of nonstandard dialects or multilingual backgrounds are not treated as deficient. For SLPs trained on norm-referenced tests built around monolingual English speakers, the tool signals a substantial rethink of which behaviors count as errors.
The tool's stated purpose is to reduce the higher rates of misdiagnosis that multilingual speakers face under standard evaluation tools. Conventional tests often flag typical multilingual development or rule-governed varieties like African American English as errors. The linguistic-justice assessment aims to distinguish a genuine communication disorder from a language difference by moving away from monolingual benchmarks and avoiding penalties for nonstandard language varieties. Early design direction includes integrating translanguaging and code-meshing, though the university has not yet detailed the instrument's structure, languages or dialects covered, or exact scoring approach.
The target population is multilingual speakers, but no specific age bands are listed in the announcement. As of 2026, the assessment is under development, not completed or validated. The news release does not describe psychometric validation, normative data, or publication plans, so clinicians should treat it as a promising direction rather than a ready-to-use tool. What remains to be studied includes which language pairs and dialects will be represented, how dynamic features are administered, and how the tool performs across ages and settings. Until those validation steps are published, the practical implication is clear: use the framework to question monolingual assumptions and consult bilingual SLP resources, but do not substitute it for existing, psychometrically supported measures when eligibility decisions are on the line.
Linguistically just assessment shifts from a one-time snapshot to a teaching-informed observation of how a student learns. The table below contrasts the core features of conventional tools with linguistically just alternatives.
| Traditional Standardized Assessment | Linguistically Just Assessment |
|---|---|
| Static, norm-referenced, single point-in-time score with no teaching phase between administrations. | Dynamic test-teach-retest method that includes an explicit teaching phase and retest to measure modifiability. |
| Fixed items and standard administration; examiner feedback and scaffolding are restricted to preserve standardization. | Mediated learning experience where the clinician intentionally teaches, prompts, and provides feedback, then observes responsiveness. |
| Compared to normative samples that may not match the student's cultural or linguistic background, leading to potential bias and misidentification. | Interpreted with attention to cultural and linguistic background, using community reference databases and dialect-sensitive analysis. |
| Product-oriented outcomes (right/wrong scores) that do not measure how the child learns during assessment. | Process-oriented evaluation of learning processes and modifiability, focusing on qualitative change across teaching. |
| Administration rules typically restrict rewording directions, modifying items, or adding practice items. | Allows reviewing items for cultural relevance, repeating or rewording directions, modifying or adding practice items, and documenting adaptations. |
| Limited ability to distinguish speech-language differences due to second language learning or dialect from true impairments. | Test-teach-retest helps distinguish language difference from disorder, especially for culturally and linguistically diverse learners. |
In a 2011 ASHA review, the English version of a bilingual screener incorrectly identified 65% of Spanish- and English-dominant children as impaired. By fifth grade, a 2026 study found emergent bilinguals had 2.31 times the odds of receiving speech or language impairment services compared with monolinguals.
Clinicians can reduce bias now, even before new tools reach the field. Below are practical, defensible steps for assessment and planning.
When a district or insurance reviewer expects a standard score, do not discard it silently. Instead, document that the score was interpreted with caution because the test was not normed on the child's language background. Then pair it with a detailed dynamic assessment note. Describe what you taught, the level of support needed, and how the child responded across mediated learning trials. For example: "After modeling and two practice items, the child independently produced the target structure in three of four opportunities." That evidence carries more diagnostic weight than a single static score.
Make language sampling and narrative retell the center of your assessment, not an add-on. Collect a conversational sample in the child's home language and English when possible. Use dialect-sensitive scoring that credits grammatical features of the child's variety rather than marking them as errors. Compare communicative effectiveness, not deviation from mainstream forms.
Before finalizing eligibility, ask yourself: Would this child's pattern be typical for a same-age peer from the same language community? Is there evidence of breakdown across settings, or only in one context? Did the child show modifiability during dynamic assessment? If the answer is no, reconsider whether a speech-language disorders label is appropriate.
Therapy goals should target functional communication: narrative clarity, turn-taking, listening comprehension, or phonological awareness. Never write goals that seek to eliminate an accent or dialect. Those goals reinforce deficit views and lack support from evidence-based practice in speech-language pathology.
What does a fair SLP assessment look like when a bilingual speech pathologist evaluates an adult with aphasia, and how does it differ from evaluating a child whose home language differs from school English? The answer shifts the clinical focus from comparing a child to monolingual norms toward measuring change from an adult's own known baseline.
For children, the core question is often whether a language pattern reflects a difference or a disorder.1 For an adult with an acquired disorder, the clinician instead asks what has changed relative to how this person communicated before the injury or illness. That requires documenting premorbid literacy, education, and multilingual status before interpreting any score.2
A bilingual adult with aphasia or cognitive-communication changes needs a detailed language history and premorbid proficiency in each language first, as recommended in multilingual assessment practices. Test performance must be interpreted against that baseline, not against monolingual norms. Stimuli should be selected using culturally relevant assessment and kept comparable across languages, because naming items can behave differently when word frequency varies by language. Include functional communication, caregiver input, and discourse samples as multiple evidence sources.2
For speakers of African American English, Spanish-influenced English, and other language varieties, document which languages were used, who administered the assessment, whether interpreters were involved, and any adaptations made. Record the person's preferred language for participation and reporting. This paper trail reduces misdiagnosis risk and keeps the evaluation grounded in the person's full language repertoire.3 Fair evaluation reflects the person's lived linguistic reality, not simply a translated test.
A request to "reduce" a child's accent or dialect is usually a request to treat a communication difference as if it were a disorder. Accent and dialect are natural, rule-governed language variations, not speech errors. In a school setting, writing "accent reduction" into an IEP is both clinically unsupported and legally risky. Before any evaluation, clarify whether the concern is intelligibility, social participation, or simply a listener's expectation.
Larry P. v. Riles (upheld in 1984 and expanded in 1986) bars California schools from using tests that produce racially disparate impact unless they are validated for the specific population. That principle extends beyond IQ tests to language and articulation measures that score dialect features as errors. The California Department of Education requires schools to treat home dialects, including African American English, as valid linguistic systems.1 ASHA's position statement on accents and dialects is equally clear: accents and dialects are communication differences, and discrimination on that basis is prohibited.2 Accent modification, when offered, must be voluntary and non-pathologizing.2
If a teacher or parent asks for accent or dialect reduction, try this respectful reframe:
Treating dialect features as errors can lead to over-identification, placing children in stigmatizing categories, or under-identification, denying needed services.3 Both can trigger OCR complaints, due process hearings, and professional ethics complaints under Title VI, Section 504, and IDEA.3 Document the difference-versus-disorder analysis, including language background and samples, and seek local legal guidance when a standardized score seems to penalize dialect. When in doubt, consult the district's legal counsel and the ASHA Code of Ethics.
Fair assessment should give every child access to services, not require them to give up their language. Linguistic justice measures skills, not distance from a single standard.
Every evaluation team can adopt a five-part assessment battery audit, regardless of SLP career setting, to catch patterns that routinely misclassify students. The audit should document home language history, compare norming sample demographics to the client, check whether scoring accounts for dialect variation, include dynamic assessment and language sampling, and record family input before final eligibility decisions. A missing home language history or a monolingual norming sample is not a neutral gap; it is a bias signal.
Teams should monitor referrals, evaluations, eligibility, placement, and dismissal by language status, race, and ethnicity. Disproportionality is not a single number; it is a ratio at each stage. Without those ratios, a team can celebrate progress on one measure while quietly overtesting bilingual students at another.
Eligibility meetings default to clinician vocabulary and clinician time. Shared decision-making means documenting whether families had an interpreter, whether their stated priorities appear in the report, and whether they can describe why a child qualifies or does not. If family members cannot explain the decision in their own words, the meeting did not meet a linguistic justice standard.
Name the suspected difference-versus-disorder question before reviewing scores. Review test evidence with an established critical appraisal checklist1, and assign one person to play the role of dialect advocate. Those small norms make bias visible before it becomes a label.
The move toward linguistic justice is finally pushing assessment beyond a single standardized score, but new technology raises the stakes for fairness.
AI-based screening and scoring tools are entering SLP practice faster than validation studies can keep up. To reduce harm rather than automate it, developers must train models on dialect-diverse speech samples, including African American English, Spanish-influenced English, Appalachian English, and other varieties. Validation should test accuracy across dialects, ages, and multilingual speakers, not just majority monolingual groups.
Clinicians should treat AI output as one piece of evidence, never as the sole basis for an eligibility decision. A low score on an automated language sample cannot determine whether a child has a disorder or a difference. Graduate programs should embed linguistic justice competencies into assessment courses, requiring practice with dynamic assessment and family interviews. Employers and school districts need routine bias monitoring for any standardized or automated tool, with retraining or removal when error patterns emerge.
States should publish disaggregated eligibility data by race, language, and English learner status so overrepresentation and underrepresentation are visible and correctable. The goal is not to lower standards, but to make every standardized score answer a more accurate question: does this child have a disorder, or has the test misread their language?
