For speech-language pathologists

Speech and language assessments at a glance: CELF-5, GFTA-3, PLS-5 and more

Published · Updated

The short answer

The most widely used children’s tests cover different ages and skills. According to their publishers, the CELF-5 assesses language from 5:0 to 21:11, the CELF Preschool-3 from 3:0 to 6:11 and the PLS-5 from birth to 7:11, and all three have separate screening tests. For speech sounds, the GFTA-3 covers 2:0 to 21:11 and the Arizona-4 18 months to 21:11. No single score diagnoses a disorder: ASHA calls for several measures, including speech and language samples, and US special education rules bar relying on any one measure.

The tests at a glance

Everything in this table comes from each publisher’s US product page (Pearson for the CELF, PLS, GFTA, KLPA and DEAP; WPS for the CASL-2 and Arizona-4), checked in October 2026. Editions and details change, so confirm against the manual for the edition you use.

Widely used children’s speech and language tests, from the publishers’ product pages
Test (publisher)AgesWhat it measuresScreener
CELF-5: Clinical Evaluation of Language Fundamentals, Fifth Edition (Pearson)5:0–21:11Semantics, morphology, syntax and pragmatics, in 16 standalone testsYes: CELF-5 Screening Test, 5:0–21:11, about 15 minutes
CELF Preschool-3: Clinical Evaluation of Language Fundamentals Preschool-3 (Pearson)3:0–6:11The language preschoolers need for classroom demands, including vocabulary, sentence structure, concepts, phonological awareness and pragmaticsYes: CELF Preschool-3 Screening Test, 3:0–6:11, 5–10 minutes
PLS-5: Preschool Language Scales, Fifth Edition (Pearson)Birth–7:11Auditory Comprehension and Expressive Communication, from pre-verbal, interaction-based skills to early literacyYes: PLS-5 Screening Test, birth–7:11, 5–10 minutes
CASL-2: Comprehensive Assessment of Spoken Language, Second Edition (WPS)3–21 yearsComprehension and expression across lexical/semantic, syntactic, supralinguistic and pragmatic language, in 14 standalone testsNone listed
GFTA-3: Goldman-Fristoe Test of Articulation 3 (Pearson)2:0–21:11Speech sound articulation in single words (Sounds-in-Words) and sentences (Sounds-in-Sentences), with intelligibility and stimulability sections and a criterion-referenced vowel checkNone listed
KLPA-3: Khan-Lewis Phonological Analysis, Third Edition (Pearson)2:0–21:11Use of 10 developmental phonological processes, analyzed from the child’s GFTA-3 responsesNone listed (it reuses GFTA-3 responses)
Arizona-4: Arizona Articulation and Phonology Scale, Fourth Revision (WPS)18 months–21:11Speech intelligibility, articulation and phonology, in three tests: Word Articulation, Sentence Articulation and PhonologyNone listed
DEAP: Diagnostic Evaluation of Articulation and Phonology (Pearson)3:0–8:11Articulation, phonology and oral motor skills, with each sound sampled at least twiceBuilt in: a 5-minute Diagnostic Screen

Test items, scoring rules and norms tables belong to the publishers and aren’t reproduced here: use the examiner’s manual for administration and scoring.

Subtests, scores and testing time

What standard scores and percentiles mean

A norm-referenced test compares a child with a large group of children of the same age or grade, and the result is usually reported as a percentile: a child at the 90th percentile scored as well as or better than 90% of that norming group (ASHA). Most of these tests also convert raw scores to standard scores. On the GFTA-3, for example, standard scores sit on a scale with a mean of 100 and a standard deviation of 15 (Pearson), so a score of 85 is one standard deviation below the mean, about the 16th percentile on a normal curve.

A low score isn’t the same as a disorder. In a 2006 review of the manuals of 43 child language tests, most manuals reported that children with language impairment scored above 1.5 standard deviations below the mean, and for more than a quarter of the tests within 1 standard deviation. Only 9 of the 43 reported sensitivity and specificity, and 5 of those reached 80% accuracy or better (Spaulding, Plante & Farinella, 2006). Many tests have been revised since, so check the current manual. In a survey of 364 SLPs, how often a test was used wasn’t related to its reliability, validity or accuracy; only its publication year was (Betz, Eickhoff & Sullivan, 2013).

Why one test doesn’t make a diagnosis

ASHA describes assessment as using a variety of measures and activities, formal and informal, which may or may not be standardized depending on the child’s background (ASHA). Screening shows who needs a full assessment; it doesn’t result in a diagnosis (ASHA). For speech sounds, ASHA describes using standardized tests alongside other sampling, because single-word testing may not reflect how a child sounds in connected speech (ASHA).

In US schools, IDEA’s evaluation rules say the same: an evaluation must use a variety of assessment tools and strategies, including information from parents, and must not use any single measure as the sole criterion for deciding whether a child has a disability or what their program should be. Assessments must also be chosen and given so they aren’t discriminatory on a racial or cultural basis (34 CFR 300.304).

See also what happens at a speech evaluation and ICD-10 codes for speech and language.

Dialect and bilingual speakers: where bias creeps in

Language samples: what tests miss

A language sample captures how a child talks in play, conversation, storytelling or explanation. ASHA lists measures such as mean length of utterance, type-token ratio, Developmental Sentence Scoring, clausal density and subordinate clauses, and describes them as complementing standardized tests (ASHA). For speech sounds, a connected speech sample from storytelling, picture description or conversation can be used to calculate percentage of consonants correct (ASHA).

Samples are recommended more often than they’re used. In a national survey of school-based SLPs, a third hadn’t used language sample analysis in the previous school year, most analyzed fewer than 10 samples, and the most common barrier was time (Pavelko et al., 2016). Even a short recorded sample, analyzed consistently, gives a baseline to measure progress against: see progress monitoring and developmental language disorder.

Questions SLPs ask

What is the age range for the CELF-5?

Pearson lists 5:0 to 21:11, with Reading Comprehension and Structured Writing from 9:0. For younger children, the CELF Preschool-3 covers 3:0 to 6:11.

What are the CELF-5 subtests?

Pearson lists 16 standalone tests, including Sentence Comprehension, Word Structure, Formulated Sentences, Recalling Sentences, Following Directions and the Pragmatics Profile. Clinicians give the tests the referral question needs.

Is there a CELF-5 screener?

Yes. The CELF-5 Screening Test covers 5:0 to 21:11, takes about 15 minutes and gives criterion scores with cut scores by age. A screen shows who needs a full assessment; it doesn’t diagnose.

What is the GFTA-3 age range?

Pearson lists 2:0 to 21:11, with separate norms for males and females. The Sounds-in-Words section takes about 5 to 15 minutes.

Is there a PLS-5 screener?

Yes. The PLS-5 Screening Test covers birth to 7:11, takes 5 to 10 minutes, and screens language, articulation, connected speech, social communication, stuttering and voice.

Sources

Join the Early Adopter Program