top of page

Rapid Automatized Naming: What the Measures Actually Predict

Writer: Andrea Chernin
Andrea Chernin
Nov 2
28 min read

Updated: Nov 5

A rapid automatized naming test is the fastest thing in your kit and the hardest to act on. Five minutes, a card of letters, a stopwatch, and a number that predicts reading years out better than almost anything else you could administer in the same five minutes.


Then you have it, and the question arrives: what does a slow RAN license you to say, and what does it license you to do on Tuesday?


This article answers both from the publishers' own documentation. It compares what each instrument calls the construct, how each one scores it, where each one files it on the report, and what each publisher says you may conclude. All figures below were verified against primary sources on 2 November 2026.

The short version


  • Every major battery now carries a RAN measure, and no two of them build it the same way. The CTOPP-2 has four naming subtests and two composites. The RAN/RAS has six tests and no composite. Acadience RAN has three measures and one total. The KTEA-3 has two, filed in two different composites. The WJ V has five, in two clusters, one of which is for mathematics.

  • The scores are not interchangeable. One instrument gives you a scaled score with a mean of 10. One gives you elapsed seconds. One gives you how many items the student named in sixty seconds. There is no conversion between them and nobody publishes one.

  • There is no cut score. The one published set of cut points in this comparison is described by its own publisher as preliminary, covers kindergarten and first grade only, and is stated in seconds for one language version of one instrument.

  • Nobody who publishes a RAN measure claims you can teach it. Acadience says so in print, twice. Elizabeth Norton, who studies this for a living, says there is no known way to bolster it. That is not a gap in the evidence. It is the finding.

  • Which is why RAN appears on the screening charts and never on the progress monitoring charts. A measure that is not expected to change is not a measure you graph.

  • A slow RAN is not a diagnosis of anything. It is a risk indicator whose own best-known describer calls it a check engine light.


"Because RAN ability depends on a large number of perceptual and cognitive factors, one can think of RAN as the 'check engine light' that indicates a problem, but doesn't reveal the exact cause."

That sentence is from Elizabeth Norton's article for the Learning Difficulties Australia Bulletin, Volume 52, No 1, June 2020. It is the most useful sentence written about RAN for practitioners, and the rest of this article is an attempt to say what follows from it.

What makes a task a RAN task, and what disqualifies one


Norton's article defines the task by three features, and each one rules something out.


The items are in an array, not presented one at a time. Most tasks run 8 to 10 items per row across 4 to 5 rows, for somewhere around 40 to 50 items in total. Norton's reason for caring: the array "mirrors the process of reading connected text, as it requires similar sustained attention, eye movements, monitoring, and cognitive processing."


Tasks that show one item at a time, which the literature calls discrete trial naming, are "not nearly as strongly related to reading ability as RAN is," which Norton attributes to Logan and colleagues in 2011. So a flashcard drill is not a RAN task, however fast you run it.


The items are already familiar. The student has to know them cold. This is not a knowledge test wearing a stopwatch.


That constraint is what forces the stimulus choice by age. For a kindergartener who does not yet know letters automatically, objects or colours are the valid option. Once letters are secure, Norton notes that alphanumeric RAN is the stronger predictor, attributing the finding to Araújo and colleagues.


Time is the score. Not accuracy. Errors matter only because they cost time, and because the stimuli are supposed to be easy there are usually very few of them.


Here is the thing that catches people: the small set size, typically five or six distinct items, is an inheritance. Norton traces the design to tasks originally built for adults with aphasia who had lost some capacity to name familiar things, citing Cutting and Denckla in 1999. The grid you are holding was not designed for reading screening. It was adapted into it.

The one number the field agrees on, and its boundary


The RAN and reading relationship is one of the better-replicated findings in reading research, and it is worth stating with the numbers attached because the vagueness is where the overclaiming starts.


The largest meta-analysis is Araújo, Reis, Petersson and Faísca, published in the Journal of Educational Psychology in 2015, volume 107, issue 3, pages 868 to 883. Its published record reports 137 studies, 857 effect sizes and 28,826 participants, analysed under a random-effects model, with a pooled correlation of .43 and substantial heterogeneity.


Norton's article breaks the same literature down by outcome, and the breakdown is where the practical information lives:


  • Single word reading: .41 across 2,991 individuals, attributed to Swanson and colleagues in 2003, and .45 across 26,491, attributed to Araújo and colleagues in 2015.

  • Text reading, meaning sentence or paragraph: .45, across 2,798.

  • Reading comprehension: a range, from .45 across 1,550 in Swanson to .39 across 4,965 in Araújo.

  • Restricted to orthographically opaque languages like English: the association with reading accuracy was .44 across 8,913, and with reading fluency .55 across 6,565.


Norton notes these are absolute values, and that in every case faster naming goes with stronger reading.


The prospective figures are lower and more honest, because they are the ones that answer the question a screener actually asks. Norton reports a meta-analysis by Hjetland and colleagues in 2017 in which RAN measured around age five correlated .34 with reading comprehension around age eight, across 3,746 individuals, and .37 with word identification across 3,285.


That is the boundary, and it has to be said in the same breath as the correlations. A correlation in the .3 to .4 range between a kindergarten measure and a third-grade outcome is a group-level fact about variance.


It is not a prediction about the child in your room. You cannot read a percentile off one of these tables and tell a family what reading will look like in three years, and neither can anyone else.


What the correlations do support is that RAN carries information your phonological measures do not. Norton makes the point directly: "RAN is not just a subcomponent of phonological awareness (PA), as RAN improves prediction of reading ability beyond PA measures alone," attributing it to Kirby and colleagues in 2010. If you already run a phonological battery, that incremental bit is the entire argument for adding naming to it.


The hypothesis that organised two decades of this work is the double-deficit hypothesis, introduced by Wolf and Bowers in 1999: that dyslexia could follow from a deficit in phonological awareness, in naming speed, or in both, with the double case the most severe.


Norton's summary of where the field landed is careful, and the caution is the useful part: the view has shifted toward a multi-componential account in which "a weakness in RAN can cause poor reading, but that RAN, phonological, and other deficits can co-occur and that profiles of children with dyslexia are highly heterogeneous."


Heterogeneous profiles is a researcher's way of saying the thing you most want is the thing you cannot have: a profile that tells you which child this is.

Five instruments, five different objects


Here is where the comparison earns its keep. These five are the instruments a practitioner is most likely to meet, and each of them has made different decisions about what rapid naming is.


The CTOPP-2 splits naming into symbolic and non-symbolic


The Comprehensive Test of Phonological Processing, Second Edition, by Wagner, Torgesen, Rashotte and Pearson, carries four rapid naming subtests. PRO-ED's own description of them is terse and worth having verbatim: Rapid Digit Naming "measures the ability to rapidly name numbers," Rapid Letter Naming letters, Rapid Color Naming colours, Rapid Object Naming objects.


Those four roll into two composites, and the split is the interesting decision:


  • The Rapid Symbolic Naming Composite Score combines Rapid Digit Naming and Rapid Letter Naming, and PRO-ED states it is available for all individuals.

  • The Rapid Non-Symbolic Naming Composite Score combines Rapid Color Naming and Rapid Object Naming, and the page describes it as an alternative "for young children, ages 4 through 6 year olds, not familiar with letters and numbers."


So the CTOPP-2 has encoded Norton's age rule into its composite structure. The alphanumeric composite is the real one; the colours and objects composite exists because four-year-olds cannot do the real one yet.


One detail on this page is worth more than its length, because it tells you something about the construct's history on this test. PRO-ED writes that in developing the second edition, "Confirmatory factor analyses were carried out and justified the presence of rapid naming subtests on a test of phonological processing."


Read that again. The publisher is telling you, in its own promotional copy, that whether naming speed belongs on a phonological processing test was a live enough question to need a factor analysis to settle. It is not obvious that these tasks measure phonology at all, and the test that most practitioners get their RAN scores from had to argue the point.


The CTOPP-2 is distributed through more than one storefront, which is worth knowing when you are checking a detail. PAR's own CTOPP-2 listing names the same four rapid naming subtests and the same two naming composites, and gives the age range as 4 years to 24 years.


The CTOPP-2's reported psychometrics, again from PRO-ED: average internal consistency above .80 for every subtest except Nonword Repetition at .77, composite internal consistency .85 or higher, subtest validity coefficients from .49 to .84 and composite coefficients from .65 to .76. The whole battery is listed at 40 minutes, Test Level B, with a 2013 copyright.


The RAN/RAS keeps six tests separate and adds one nothing else has


The Rapid Automatized Naming and Rapid Alternating Stimulus Tests, by Maryanne Wolf and Martha Bridge Denckla, is the instrument the construct is named after, and it makes the opposite structural choice to the CTOPP-2. Six tests, reported separately, no composite.


Four are RAN proper: Letters, Numbers, Colors, Objects. Each uses five high-frequency stimuli repeated ten times across five rows, for fifty items.


Two are RAS, and nothing else in this comparison has an equivalent. A 2-Set test alternates letters and numbers; a 3-Set test alternates letters, numbers and colours. The 2-Set uses ten distinct stimuli and the 3-Set fifteen, again repeated across five rows to fifty items.


That alternating-stimulus design is a genuinely different task and it is worth knowing why it exists. Switching between categories adds a set-shifting demand that single-category naming does not have.


Norton's article notes the RAS variant and says plainly that "because traditional RAN is used more broadly, that is the focus here," which is an accurate description of the literature and also a warning: the 2-Set and 3-Set tests sit on thinner research than the four single-category ones.


PRO-ED's own claims for the instrument: it "identifies students who may be at risk for reading failure," converts raw scores to percentiles, standard scores and age and grade equivalents, and takes five to ten minutes for all six tests. The page also calls the RAN and RAS tests "the 'gold standard' of naming tests," which is a publisher's claim about its own product and should be read as one.


The norms are the number to hold onto. PRO-ED states the tests were normed on 1,461 individuals in 26 states, suitable for ages 5-0 through 18-11, with a 2005 copyright.


That date is independently corroborated. The Buros Center's catalogue entry for the RAN/RAS records the publication date as 2005 and the publisher as PRO-ED, and notes that the instrument has been reviewed in the Seventeenth Mental Measurements Yearbook. If a date on a publisher's own page matters to a report you are writing, Buros is the cheapest second opinion available.


Acadience RAN reports a time in seconds and tells you when to use it


Acadience RAN is the one instrument here designed from the outset as a screening add-on rather than as part of a diagnostic battery, and it shows in every decision.


Three measures: RAN Objects, RAN Letters, RAN Numbers. The administration sequence is fixed, and it is conditional. From Acadience's own Preliminary Cut Points for Risk document, carrying a February 2022 copyright:


"Students begin with RAN Objects and proceed to RAN Letters. RAN Numbers is only administered if a student discontinues on RAN Letters. The RAN Total score is the time to complete RAN Objects plus the time to complete RAN Letters. If the RAN Letters task is discontinued, then the RAN Total score is the time to complete RAN Objects plus the time to complete RAN Numbers."

So RAN Numbers is a substitute, not a third data point, and the RAN Total is always a sum of exactly two elapsed times.


The same document carries the only published cut points in this comparison. They are in seconds, they cover kindergarten and first grade, and they differ by time of year:


  • Kindergarten. Low-Risk is 195 seconds or less at the beginning of year, 139 or less at the middle, 125 or less at the end. At-Risk is 230 or greater, 170 or greater, and 154 or greater respectively. Some-Risk is the band between.

  • First grade. Low-Risk is 138 or less, 96 or less, 85 or less. At-Risk is 165 or greater, 124 or greater, 111 or greater.


Acadience states the cut points "are in seconds and applicable only to the English version of Acadience RAN," and calls them preliminary "because we will reexamine them once we obtain additional data." They were derived by logistic regression predicting the probability of being At or Above Benchmark on the Acadience Reading Composite Score at a later benchmark period.


Note what that means about the construct validity argument on offer. The cut points are not anchored to a reading outcome measured independently. They are anchored to a later score on the same publisher's own composite.


And the decision rule attached to the instrument is more restrictive than most practitioners realise. Acadience's Acadience Reading crosswalk document, published in January 2026, lists RAN among the Acadience Reading measures and labels it, in its own row, "Risk Indicator." The footnote to that row is the single most quotable sentence any publisher in this comparison has written about naming speed:


"RAN serves as an additional indicator of risk. RAN is not a measure that is expected to be repeated across each TOY but rather added to a single TOY testing window where the student's performance on other Acadience measures is Below or Well Below Benchmark. The addition of RAN provides more information about the risk profile of the student. RAN is not a skill which can be directly taught. As such, RAN is not expected to increase over time."

That document was assembled against one state education agency's guidance, so read the state-specific columns as applying to that state and nowhere else. The footnote itself is Acadience's statement about its own measure, and it is not state-specific at all.


Two things follow. First, the publisher's own recommended use is conditional: you add RAN when other measures are already below benchmark, not as a standing part of every screening window. Second, the publisher tells you in writing not to expect the score to improve.


The KTEA-3 files its two naming subtests in two different composites


The Kaufman Test of Educational Achievement, Third Edition carries two rapid naming subtests, Object Naming Facility and Letter Naming Facility, which Pearson's own product page describes as two measures of rapid automatized naming. Ages 4:0 to 25:11, qualification level B, 2014.


What the product page does not tell you is where those two subtests are reported, and that turns out to be the most practically important fact about RAN on this battery. Pearson's published KTEA-3 score report sample resolves it:


  • Object Naming Facility is reported inside the Oral Fluency Composite, paired with Associational Fluency.

  • Letter Naming Facility is reported inside the Orthographic Processing Composite, alongside Spelling and Word Recognition Fluency.


So the KTEA-3's two RAN subtests never appear together. There is no naming composite on this test. One of them is averaged with a word-generation task and called oral fluency; the other is averaged with spelling and called orthographic processing.


If you are reading a KTEA-3 report looking for naming speed, this is the thing that will mislead you. An Orthographic Processing Composite in the below-average range tells you nothing on its own about whether the weakness is in spelling, in word recognition fluency, or in naming speed. You have to go to the subtest line.


The score report's own interpretive narrative concedes the structure in passing, and is worth quoting because it is Pearson telling you to do exactly this: "Consider performance on the KTEA-3 Oral Fluency composite (Associational Fluency and Object Naming Facility subtests) or the Letter Naming Facility subtest to cross-validate a possible weakness in rapid automatic naming (RAN)."


Cross-validate, note. Not establish.


The WJ V has five naming tests and one of them is for math


The newest entry in this comparison is also the most expansive. The Woodcock-Johnson V Technical Abstract, published by Riverside Assessments in 2025, describes five rapid naming tests in the WJ V Virtual Test Library. Four of them are new.


Riverside's own account of the decision is explicit about its warrant: "The author team developed four new rapid automatized naming (RAN) tests for the WJ V VTL: Rapid Letter Naming, Rapid Phoneme Naming, Rapid Number Naming, and Rapid Quantity Naming," with Rapid Picture Naming carried over from the WJ IV.


The abstract cites Araújo and colleagues, Kirby and colleagues, Koponen and colleagues, Norton and Wolf, Schneider and McGrew, and Swanson and colleagues in support.


They combine into two clusters, and the second one is the news:


  • The RAN–Reading cluster contains Rapid Phoneme Naming, Rapid Picture Naming and Rapid Letter Naming.

  • The RAN–Math cluster contains Rapid Number Naming and Rapid Quantity Naming.


A naming cluster built for mathematics is a first in this comparison. Riverside places the five tests in the processing speed and retrieval fluency broad abilities, with Rapid Phoneme Naming also loading on auditory processing and Rapid Quantity Naming on visual processing, and describes them as "sensitive to the early detection of reading and math disorders."


Rapid Phoneme Naming is the one to look at twice. Riverside's description: "The examinee has 1 minute to quickly pronounce phonemes (e.g., /m/) for single letters (i.e., graphemes) presented in successive rows." That is not naming a letter. It is retrieving a sound for a symbol under time pressure, which is a different operation from retrieving a name, and it is the only task of its kind in this comparison.


Two structural points about the WJ V matter for anyone reading its reports. The RAN tests sit in the Virtual Test Library rather than in the core cognitive battery, and Riverside is clear that the library "is not intended as a direct replacement for the WJ IV OL" but is a flexible set usable alone or alongside the other batteries.


And every one of these tests is a one-minute test. The score is how much the student got through in sixty seconds, not how long the array took.


WJ V norms were collected from 5,837 individuals aged 3 to 90 and over, during a nineteen-month window from February 2022 through August 2023.

Three scales, no conversion


Lay the five instruments side by side and the scoring metrics do not line up.


  • The CTOPP-2 gives you subtest scaled scores with a mean of 10 and a standard deviation of 3, and composite indexes with a mean of 100 and a standard deviation of 15.

  • The RAN/RAS is scored on "the amount of time required to name all of the stimulus items on each test," converted to percentiles, standard scores and age and grade equivalents.

  • Acadience RAN gives you elapsed seconds and a risk band. No standard score at all.

  • The KTEA-3 gives you subtest standard scores inside composites that are not about naming.

  • The WJ V gives you a count of items completed inside one minute.


Three of these are rate measures with the rate expressed three different ways: total seconds for a fixed array, standard score derived from total seconds, and items completed in fixed time. Those are not the same measurement operation, and a student's standing can move between them.


Nobody publishes a crosswalk between them, and you should not build one. If a prior evaluation reported a CTOPP-2 Rapid Symbolic Naming composite and you re-test with Acadience RAN, you have two numbers about the same construct with no defensible way to say whether the second is better or worse than the first. The correct move is to say so in the report.


Grade and age equivalents are offered by two of these instruments, which is a separate problem with its own history. If you are tempted to use one, Grade Equivalent Scores: What the Publishers Themselves Say collects what the publishers write about their own equivalents, and it is less flattering than you would expect.

Three norm dates, and one of them is over twenty years old


Set the copyright and norming dates next to each other, because this is where a cross-instrument comparison does something a single publisher's page cannot.


  • RAN/RAS: normed on 1,461 individuals in 26 states, 2005 copyright.

  • CTOPP-2: all new normative data collected in 2008 and 2009, 2013 copyright.

  • KTEA-3: 2014.

  • WJ V: norming data collected February 2022 through August 2023, 2025.


The instrument the construct is named after is working from a norm sample collected before any student currently in a K-12 classroom was born. The instrument most practitioners use for naming, the CTOPP-2, is working from data collected around 2008 and 2009.


Whether that matters is a real question rather than a rhetorical one, and the honest answer depends on whether naming speed in the population has drifted, which nobody in this comparison has published evidence about either way. The general problem is worked through in Outdated Test Norms: What Actually Makes a Score Too Old, including the specific case where an old norm sample is defensible and the case where it is not.


One more thing about the CTOPP-2 page, and it is a discrepancy rather than a criticism. PRO-ED's product page gives the CTOPP-2's age range as 4-0 through 24-11 and, two paragraphs later, describes the normative sample as "1,900 individuals ranging in age from 6 through 24 years."


The same page elsewhere says item additions made it "appropriate to administer the CTOPP-2 to 4-year-olds" and that the lower version "now covers the 4- to 6-year-old range."


Those statements do not reconcile on the lower bound as the page is written on 2 November 2026. The Examiner's Manual is where that gets settled, and if a four-, five- or six-year-old's naming composite is load-bearing in a report you are writing, that is the document to open rather than the product page.

Letter naming fluency is not rapid automatized naming


This is the single most common confusion in practice and it is worth settling with evidence rather than assertion, because the two tasks look almost identical on paper.


Both show letters. Both are timed. Both produce a rate. The difference is set size and repetition: a RAN array uses five or six distinct letters repeated many times, while a letter naming fluency probe uses a large, randomised set of different letters.


The instruments themselves treat them as two different measures. i-Ready Literacy Tasks, the Curriculum Associates K-6 set, lists Rapid Automatized Naming and Letter Naming Fluency as separate items on its own measure list.


NCII's Academic Screening Tools Chart reviews them as separate entries for the same tool, with "Rapid Automatized Naming – Letters" appearing for grades 1 through 3 and "Letter Naming Fluency" appearing separately. Acadience likewise lists Letter Naming Fluency and RAN as distinct rows on its crosswalk.


And the publisher makes the empirical claim directly. Acadience's cut points document states that "Initial research indicates that Acadience RAN provides additional information for educators above and beyond the information provided by the Acadience Reading Composite Score and also measures skills not entirely captured by Letter Naming Fluency," attributing it to Gray and colleagues in 2020.


So if a school tells you it already screens naming because it runs letter naming fluency, that is not the same measure, and the measure's own publisher says it is not.

There is no cut score, and the one published set is preliminary


This is the part that practitioners find hardest and that the research community is most unanimous about.


Norton states it without hedging: "Despite extensive research, there is no single test or single cutoff score that indicates that RAN is a problem."


Her reason is the useful part, because it transfers to every rate measure you own:


"RAN ability is a continuum, not a cliff; a child who scores in the 9th percentile is not all that different from a child who scores in the 10th percentile. However, if a cutoff of below the 10th percentile is chosen, one child would be identified while the other would not. Thus, RAN should be considered as a continuum and a piece of the puzzle with other assessments of language, reading, and cognition."

The only published cut points in this comparison are Acadience's, and they are limited in four ways at once by their own document: preliminary pending more data, in seconds rather than in a standard metric, kindergarten and first grade only, and valid for the English version only.


Which means that for every student above first grade, and for every instrument except one, there is no published threshold at which naming speed becomes a finding. You are left with a percentile, a confidence interval, and judgement.


If you want the part of this that generalises, the band around any score is the thing that decides most close calls, and Standard Error of Measurement: What That Band Means works through what a confidence interval does and does not entitle you to conclude. A naming score two points from a cut point is the exact case that section exists for.

Nobody who publishes a RAN measure claims you can teach it


Three independent statements, two of them from publishers with a commercial interest in saying otherwise.


Acadience, in its January 2026 crosswalk footnote: "RAN is not a skill which can be directly taught. As such, RAN is not expected to increase over time."


Acadience again, in its cut points document: "While RAN itself is not a teachable skill, the incremental validity of RAN means that it can be useful in identifying students who are at risk for reading difficulties."


Norton, on the question she says she is asked most often: "A question I am often asked is 'How can we improve that child's RAN score?' The answer is frustrating to hear, I'm afraid — we know of no way to simply bolster a child's RAN ability."


She notes that studies attempting to improve RAN through naming practice, reviewed by Kirby and colleagues in 2010, do not show "that training RAN leads to better outcomes than providing equal hours of reading intervention," and that naming ability appears to be "a relatively intrinsic or set characteristic of an individual."


Timothy Shanahan reached the same place in answer to a teacher asking exactly that question, in an April 2020 post on Reading Rockets: "that doesn't mean that you can teach RAN," noting that the one study that tried found no reliable impact on either RAN or reading, and that "no one who has studied it thinks that we can teach it successfully in any way likely to matter."


The clause to hold onto in all of this is Norton's comparative one: not that RAN training fails absolutely, but that it does not beat equal hours of reading intervention. That is the version of the finding that survives contact with a caseload. The relevant cost of an hour spent on naming drills is the hour of reading instruction you did not deliver.


Norton extends the caution to the adjacent temptation, noting that training processing speed and executive function skills "may lead to better performance on the training task, but rarely leads to meaningful, generalizable improvements," citing Melby-Lervåg and Hulme in 2013. That is the same finding this site has been through at length in Working Memory Training Programs: What the Evidence Says, and the two literatures say the same thing in the same shape.

Which is why RAN is on the screening charts and not on the progress monitoring charts


Here is a structural fact that follows directly from the one above, and that you can check for yourself in two minutes.


NCII's Academic Screening Tools Chart carries a reviewed rapid automatized naming entry. As the chart stands on 2 November 2026, i-Ready Literacy Tasks lists "Rapid Automatized Naming – Letters" for grades 1, 2 and 3, as a three-minute individually administered measure with manual scoring, reporting risk levels.


The chart's own columns for that entry are worth reading before you cite it. The grade 1 norm sample is described as regional without cross-validation and the grade 2 and 3 samples as national without cross-validation, and disaggregated reliability and validity data are recorded as not provided.


NCII's Academic Progress Monitoring Tools Chart contains no rapid automatized naming entry at all. Not one tool, at any grade, as of the same date.


That asymmetry is not an oversight. It is the construct doing what the publishers said it would do. A progress monitoring measure has to be sensitive to growth, because its whole function is to tell you whether what you are doing is working. A measure that its own publisher says is not expected to increase over time cannot do that job, and the review process reflects it.


So the operational rule falls out cleanly: naming speed belongs in your intake, and it does not belong on your progress graph. If you want the measures that do belong there, Progress Monitoring Tools Compared sets out four systems and their rules, and Progress Monitoring Decision Rules traces where the thresholds for acting on a graph actually came from.


One more thing worth noting: Acadience RAN itself does not appear anywhere on either NCII chart as of 2 November 2026, although Acadience Reading K-6, Acadience Reading 7-8, Acadience Math and Acadience Reading Pre-K: PELI all do.


The absence of a tool from the chart is not a judgement about the tool. It means it has not been through that review, and when a school asks you whether a measure is "on the NCII chart," that distinction is the whole answer.

A slow RAN is not a diagnosis


Flatly: no score on any instrument in this article identifies dyslexia, a learning disability, or any other condition. Not a slow RAN, not a slow RAN with a weak phonological composite, not a double deficit.


Identification is a determination made by a qualified professional from a full body of evidence, and the publishers of these instruments do not claim otherwise. PRO-ED's own framing of what the RAN/RAS does is that it "identifies students who may be at risk for reading failure."


May be at risk. Acadience calls its measure a Risk Indicator in the row heading. Pearson's interpretive text says to use naming subtests to cross-validate a possible weakness.


None of those is diagnostic language, and the gap between "at increased risk" and "has a condition" is the gap practitioners get asked to close in conversations with families. The honest answer is that a naming measure cannot close it.


What a screener can and cannot establish is worked through in detail in Dyslexia Screening Tools: What a Screener Can Tell You, and the related question of what you may do with somebody else's completed evaluation is in Psychoeducational Evaluation for Educational Therapists. The boundary between those two documents is the boundary this article sits on.

What a slow RAN does license you to do


Having stripped out what the measure cannot support, here is what it can, all of it drawn from the sources above rather than from inference.


It licenses closer monitoring, with a specific target. Norton's recommendation: "If a child is identified with low RAN at screening, they may be monitored more closely, especially for difficulties with reading fluency." That is a named outcome to watch, not a vague caution, and it follows the finding that in English the strongest RAN associations are with speeded and fluency measures rather than accuracy.


It licenses an expectation about what the fluency work will cost. A student with slow naming and intact phonology is more likely to be the one whose accuracy comes up and whose rate does not. Knowing that at intake changes how you set a goal and how you talk about the timeline, which is worth more than most intake data.


It licenses a reading of a plateau that would otherwise look like a failure of effort or instruction. This is the use I would defend most strongly. When decoding accuracy is secure and words-correct-per-minute has stalled, a naming score collected months earlier is the piece of evidence that makes the plateau legible rather than mysterious.


It licenses interpretive caution about the fluency norms themselves. If you are reading a student's oral reading fluency score against a norm table, a known naming weakness is part of why the score sits where it does. What those tables are and are not is in Oral Reading Fluency Norms: What the Table Does Not Say.


It does not license a naming intervention. Norton's practical advice when there is a weakness, given the absence of RAN-specific evidence-based practice, is "to work on building accuracy and then automaticity in all the other areas of language and literacy." Which is to say: the correct response to a slow RAN is good reading instruction, delivered at volume.


"A measure that nobody claims you can teach is not a target. It is a reason to expect the work to take longer, and to say so at the start."

Which of these you can actually buy


Worth a line, because it determines whether the question is live for you at all.


The CTOPP-2, the RAN/RAS and the KTEA-3 are all listed by their publishers at qualification level B. PAR's RAN/RAS listing gives B as well, and gives the age range as 5 years to 18 years, which agrees with PRO-ED.


The WJ V is distributed by Riverside Insights, which operates its own qualification scheme rather than a letter system. Acadience RAN and i-Ready Literacy Tasks are screening instruments with the access conditions typical of that category rather than publisher qualification letters.


What those letters mean, why they are not an industry scale, and which storefront's rules actually bind you is set out in Achievement Test Qualification Levels: What the Letters Mean. The short version: a letter is one publisher's purchasing rule, and two publishers' Bs are not the same B.

A decision sequence, not a decision


What this article will not do is tell you which naming measure to administer to a described student. That is a decision that depends on the referral question, the age, what has already been done, and what you are qualified to purchase. Here is the sequence instead.


  1. Ask what the naming score is for. If the answer is to monitor a fluency risk or to interpret a plateau, it earns its five minutes. If the answer is to establish a diagnosis or to pick an intervention, no instrument in this article will deliver it.

  2. Check what has already been administered. If a recent CTOPP-2 is in the file, the Rapid Symbolic Naming composite is already there and re-testing on a different instrument buys you a second, non-comparable number.

  3. Match the stimulus to the student, not to the form. If letters and digits are not yet automatic, the alphanumeric tasks measure knowledge rather than naming speed. Both the CTOPP-2 and Acadience RAN build that rule into their own structure; follow it.

  4. Decide in advance what a slow score will change. Write down, before you administer, what you will do differently if the score is at the tenth percentile. If the answer is nothing, do not administer it.

  5. Read the subtest line, not the composite. On the KTEA-3 especially, the composite containing a naming subtest is not about naming.

  6. Report the metric you actually used. Seconds, scaled score and items-per-minute are three different things. A report that says "slow rapid naming" without naming the instrument and the metric cannot be acted on by whoever reads it next.

  7. Never graph it. See the NCII charts above.


The nearest neighbour to this post is the one on what a pseudoword decoding test can and cannot show, which is the other construct that sits in nearly every dyslexia battery and gets read as more than it is. The parallel is close and the contrast is instructive: nonword decoding is a teachable skill with alternate forms, so it appears on both of the federal tools charts, while this one appears only on the screening chart.

Where this sits next to the other assessment posts


Naming speed is the construct that sits at the edge of several others, so it is worth saying what is covered where.


Phonological Awareness Assessment: Where the Construct Ends covers the three other CTOPP-2 composites and the question of how wide the phonological umbrella is on each instrument. The relationship to this article is structural: the CTOPP-2's phonological composites and its naming composites are on the same test and are measuring different things, which is exactly why the publisher ran a factor analysis.


Dyslexia Definition Compared: Five Bodies, Five Answers covers what the published definitions do and do not name as diagnostic features, which is the right place to check whether a given definition mentions naming speed at all.


Spelling Assessment Tools: What Each One Can Diagnose is the other article on this site that runs into the KTEA-3's Orthographic Processing Composite, from the spelling side rather than the naming side.


What "the Science of Reading" Actually Names is relevant for a narrower reason: naming speed is one of the better-evidenced findings in the reading literature and it appears in very few of the documents that claim the phrase.


And Educational Therapist vs. Learning Specialist vs. EF Coach is the role map underneath all of it, for the question of who administers and interprets which of these in which setting.

Where the published record runs out


This is the part of every article on this site where we say what we could not establish, because a confident sentence over a gap is worse than the gap.


Two of Acadience's own RAN technical documents are no longer retrievable. The Acadience RAN Assessment Manual and the Incremental Validity of Acadience RAN technical brief, both cited in and linked from Acadience's own cut points document, returned 404 pages on 2 November 2026, with a notice stating that the organisation recently launched a new website and that pages "may have moved or been retired in the process."


The Gray and colleagues 2020 incremental validity finding reported above is therefore stated here as the cut points document states it, and has not been read in its original. If that finding is load-bearing for a decision you are making, email Acadience and ask for Technical Report No. 28.


The RAN/RAS norm sample is reported only in total. PRO-ED gives 1,461 individuals in 26 states and no breakdown by age, race, ethnicity, parent education or region on its product page. For a 2005 instrument whose scores you may be comparing to a 2022 to 2023 norm sample on another test, that breakdown would matter, and it is in the Examiner's Manual rather than in anything public.


The CTOPP-2's lower age bound is stated two ways on the publisher's page, as described above. We have not resolved it and are not going to guess.


No instrument in this comparison publishes a cut score above first grade. That is not a gap in this article's research. It is a gap in the field, and the only honest thing to do with it is say so out loud rather than borrow a threshold from a different instrument.


The RAS tests sit on thinner evidence than the RAN tests, by Norton's own account of where the literature has concentrated. If you administer the 2-Set or 3-Set and report it, the interpretive base under that number is smaller than the one under RAN Letters.


And nobody has published evidence either way on whether population naming speed has drifted since 2005, 2009 or 2023. The norm-recency question in this specific construct is open.

Frequently asked questions


What is a rapid automatized naming test?


A rapid automatized naming test presents an array of highly familiar items, typically five or six distinct letters, digits, colours or objects repeated across four or five rows to about 40 to 50 items, and asks the student to name them all as quickly as possible without errors.


The score is the time taken, or on some instruments the number of items completed in a fixed minute. It measures how automatically a known name can be retrieved under continuous time pressure, not whether the items are known.


Does a slow RAN mean a student has dyslexia?


No. No score on any instrument discussed here identifies dyslexia or any other condition. Rapid naming is a risk indicator, and the publishers use risk language rather than diagnostic language for it: PRO-ED says its instrument identifies students who "may be at risk for reading failure," and Acadience labels its measure a Risk Indicator.


A slow naming score raises the probability of later reading difficulty at a group level and does not establish anything about an individual.


Can you improve a student's RAN score?


Every source consulted for this article says no, and two of them are the publishers themselves. Acadience states in print that "RAN is not a skill which can be directly taught" and that it is "not expected to increase over time."


Elizabeth Norton, writing for practitioners, says "we know of no way to simply bolster a child's RAN ability," and that attempts to train naming do not produce better outcomes than the same hours spent on reading intervention. The practical implication is that the correct response to a slow naming score is more and better reading instruction, not naming drills.


Is letter naming fluency the same thing as RAN?


No, and the instruments treat them as separate measures. A RAN array repeats a small closed set of five or six letters many times, while a letter naming fluency probe uses a large randomised set of different letters.


i-Ready Literacy Tasks lists Rapid Automatized Naming and Letter Naming Fluency as separate measures, NCII's screening chart reviews them separately, and Acadience's own documentation states that its RAN measure captures skills "not entirely captured by Letter Naming Fluency."


What RAN cut score indicates a problem?


There is no published cut score for rapid automatized naming above first grade on any instrument in this comparison. The only published thresholds are Acadience's preliminary cut points for kindergarten and first grade, stated in seconds and applicable only to the English version of their measure.


Norton states the research position directly: there is no single test or cutoff score that establishes that naming is a problem, because the ability is a continuum rather than a cliff.


Which instruments include a rapid naming measure?


Among those most likely to be encountered in practice: the CTOPP-2 has four naming subtests and two naming composites; the RAN/RAS has four RAN tests and two rapid alternating stimulus tests; and Acadience RAN has three measures reported as one total time.


On the broader batteries, the KTEA-3 has Object Naming Facility and Letter Naming Facility, reported in two different composites, and the WJ V has five naming tests split between a RAN–Reading cluster and a RAN–Math cluster. The scoring metrics are not interchangeable across any of these instruments.


Should RAN be used for progress monitoring?


No. As of 2 November 2026, no rapid automatized naming measure appears on NCII's Academic Progress Monitoring Tools Chart, while a reviewed RAN measure does appear on the Academic Screening Tools Chart. That asymmetry follows from the construct: a progress monitoring measure has to be sensitive to growth, and the publishers of naming measures state that growth is not expected.

Join the Illuminate educator network


Naming speed is a good illustration of why this work is harder than the materials make it look. The measure is quick, the research base is strong, the publishers are reasonably candid, and the thing you most want from it is the one thing it will not give you.


The practitioners who use it well are not the ones who found the lowest percentile. They are the ones who decided in advance what a slow score would change, wrote down which instrument and which metric produced it, and did not put it on a graph.


If this is the work you do, we write for practitioners: educational therapists, learning specialists and executive function coaches. Join the educator list and new practice-facing guides arrive as they publish.

Comments


bottom of page