Articulation

The Articulation Hierarchy: Steps From Isolation to Conversation (Free Chart)

What each level looks like, the criteria many SLPs use to move up, how to fade cues, and where home practice fits, with a one-page chart you can print.

September 25, 2026•16 min read

Most articulation therapy follows the same basic path. A student learns to make a sound on its own, then uses it in syllables, words, phrases, sentences, longer structured speech, and finally conversation. SLPs call this path the articulation hierarchy, and a great deal of school therapy runs on some version of it.

The idea is simple. The details are where students get stuck: when to move up, how much help to give at each step, and what practice at home should look like along the way. This guide covers all three, plus the research that has asked whether the ladder really has to be climbed one rung at a time. The free printable chart is just below.

The articulation hierarchy at a glance

ASHA’s practice portal on speech sound disorders describes treatment as a sequence of three phases. Establishment is eliciting the target sound and stabilizing it on a voluntary level. Generalization is carrying the sound over at increasingly challenging levels, such as syllables, words, phrases and sentences, and conversational speaking. Maintenance makes the sound more automatic and builds self-monitoring and self-correction. The articulation hierarchy is that middle phase broken into steps, with isolation as the place where the sound is first established.

Each step up asks more of the student. The utterance gets longer, the sound meets more neighbors, and attention moves away from the sound and toward the message. That last part is the real test: a sound is learned when it holds up while the student is thinking about something else.

Articulation Hierarchy Chart

Student Target sound Position

Start at step 1 at the bottom and climb. Examples use the S sound.

StepWhat it looks like (S)Common criterion to move upDate reached
7. ConversationThe goalTalk about the weekend, a game, a favorite showCorrect most of the time in conversation (often 80 to 90%), then with other people and places
6. Reading and structured talkRead a short paragraph aloud, describe a picture, retell a story80 to 90% across a few minutes of structured speech
5. SentencesSam sat on the bus. Then sentences the student makes up.80 to 90% in repeated sentences, then in self-made ones
4. PhrasesI see a sun; a big sun; my yellow bus80 to 90% in varied 2 to 4 word phrases, not only carrier phrases
3. Wordssun, soap; bus, house; pencil (one position at a time)80 to 90% in each target position, on words not practiced before
2. Syllablessa, see, soo; as, ees, oos; asa80 to 90% across several vowels, in each position you plan to teach
1. IsolationStart hereThe sound on its own: sssCorrect on most tries with little or no cueing (often 80 to 90%)

Cue levels, from most to least support

MaximalTouch or a tool to show placement, or saying it togetherModerateA model to copy, or spoken placement instructionsMinimalA gesture, a picture, or a one-word reminderIndependentNo cue. The student checks their own sound.

Criteria are common conventions, not rules. Many SLPs look for them across two or three sessions in a row, with little or no cueing, on words and sentences the student has not practiced.

The seven levels, step by step

The criteria below are the conventions many SLPs use, not research cutoffs. There is more on choosing a criterion in the next section.

1. Isolation

What it looks like. The sound by itself. For S, a long, steady sss. Sounds that cannot be held, like K and G, come out as a quick k or a k with a short vowel after it, which is one reason many SLPs start those sounds at the syllable level instead.

What is happening. This is establishment: teaching the movement. Most placement work lives here, such as showing where the tongue goes, shaping the new sound from one the student can already make, or watching in a mirror. Cueing is heaviest at this level, and that is expected.

Common criterion. The student produces the sound correctly on most tries with little or no cueing. Many SLPs look for about 80 to 90 percent.

Watch for. Staying too long. Isolation is a means to an end, and some students and sounds skip it. For R, many SLPs start from a syllable or word where the student already makes a good R, as our guide to R sound therapy explains.

2. Syllables

What it looks like. The sound paired with vowels in short nonsense syllables. For S: sa, see, soo with the sound first; as, ees, oos with it last; asa or eesee with it in the middle. Changing the vowel matters, because each vowel gives the tongue a different starting point.

Why it helps. Syllables are the smallest piece of real speech. Many SLPs like nonsense syllables because they carry no old habit: a student may have said sun the old way many times, but has never said soo at all.

Common criterion. About 80 to 90 percent across several vowels and in each position you plan to teach next.

3. Words

What it looks like. Real words with the target in one position at a time: the start of the word (sun, soap), the end (bus, house), and the middle (pencil, missing). Many SLPs begin with whichever position is easiest for the student, often the start, and move from one-syllable words to longer ones. Clusters such as star, spoon, and snake often come once the single sound is solid, and they have their own S blend word list.

Common criterion. About 80 to 90 percent in each target position, checked on words the student has not practiced, with no cueing or teaching first. SLPs call this a probe. Drilled words get easier through repetition, so a list the student has practiced all month overstates the skill.

4. Phrases

What it looks like. The target word inside two to four words. Carrier phrases are the easiest start, because only one word changes each time: I see a sun, I see a bus. Then varied phrases: a big sun, my yellow bus, sip it slowly.

Why it matters. This is the first time the sound has to share breath and planning with other words, and many students who are accurate in single words slip here for a while.

Common criterion. About 80 to 90 percent across varied phrases, not only carrier phrases.

5. Sentences

What it looks like. Repeated sentences first (the SLP says it, the student says it back), such as Sam sat on the bus. Then sentences the student reads or makes up, for example in answer to tell me something about the sun. Making up a sentence is harder, because the student has to plan what to say and how to say it at the same time.

Common criterion. About 80 to 90 percent, including sentences the student creates.

6. Reading and structured talk

What it looks like. Longer speech with some structure around it: reading a paragraph aloud, describing a picture, retelling a short story, answering questions that pull target words (what would you pack for the beach?), or a game with set turns. For students who are not reading yet, picture description and retelling do the same job.

Why it matters. Attention shifts from the sound to the content, which is a close rehearsal for conversation. Reading also puts a lot of targets in front of the student quickly; our guide to reading practice in speech therapy covers how to choose passages.

Common criterion. About 80 to 90 percent across a few minutes of structured speech.

7. Conversation

What it looks like. Structured conversation first, in the therapy room, on a topic full of target words. Then free conversation. Then conversation with other people and in other places: the classroom, the lunch table, home. That last part is carryover, and some charts list it as its own eighth step.

Common criterion. Correct use in conversation most of the time (many SLPs look for about 80 to 90 percent in a conversation sample), and then the sound showing up outside the therapy room. Our carryover strategies for speech therapy cover this final stretch in detail.

When to move up a level

No study sets the right percentage for moving up. The criteria on the chart are a common convention, and a typical version has four parts:

  • Accuracy of about 80 to 90 percent at the current level.
  • Consistency across two or three sessions in a row, so one good day does not count as mastery.
  • Little or no cueing. Accuracy with a mirror and a model is a different skill from accuracy alone.
  • New items. Checked on a probe of words or sentences the student has not practiced.

If the student’s IEP goal names a level and a criterion, measure against that. Our guide to writing speech therapy IEP goals shows how to build the level into the goal itself, and speech therapy data collection covers quick ways to count trials in a busy session.

An illustration, not real student data

M. is working on S at the word level, sound at the start of the word.

  • Monday: 14 of 20 correct with a mirror.
  • Wednesday: 17 of 20 with a hand gesture as a reminder.
  • Friday probe, 10 untrained words, no cues: 9 of 10.

One more probe at 9 of 10 or better next week, and M. moves to phrases, where the SLP expects to bring a cue back for a while.

Moving back down is part of the process, not a failure. If accuracy drops well below the criterion after a step up, the usual fix is to add support at the new level or spend a few more sessions at the old one. The step up, the step back, and the probe results all belong in the progress report, because they show how the skill is actually building.

Cueing hierarchy: how to fade support

Every level of the ladder can be done with a lot of help or with none. So there are really two dials: how hard the task is (the level) and how much support the student gets (the cue). A cueing hierarchy is the order in which you turn the second dial down.

Three kinds of cues

Verbal

A model to copy, a placement instruction (keep your tongue tip behind your teeth), a one-word reminder, or a question that prompts a self-check.

Visual

A mirror, a mouth picture, a hand gesture, or a reminder card on the table. Watching the SLP’s mouth during a model is a visual cue too.

Tactile

Touch that shows where or how: a tool that marks where the tongue should go, or a hand on the throat to feel the voice switch on.

ASHA’s speech sound disorders page lists several of these among the techniques that raise awareness of the target sound or give feedback on placement: a mirror for visual feedback, gestural cueing for place or manner (for example, a long sweeping hand gesture for fricatives such as S and a short chopping one for stops such as T), and tools such as tongue depressors for tactile feedback on tongue placement. Not every cue has support, though. A tutorial on R therapy by Preston and colleagues (2020) lists tactile cues along the mylohyoid muscle, under the chin, among the elements of R therapy that currently lack scientific support.

From most to least support

Programs order the steps a little differently, but a typical articulation cueing hierarchy runs like this:

  1. Placement by touch or tool. The SLP shows the position with a tool or a touch.
  2. Say it together. The student says the target at the same time as the SLP, watching the SLP’s mouth.
  3. Model and copy. The SLP says it, the student says it right after.
  4. Placement instruction. A spoken reminder of how to make the sound, with no model.
  5. Visual cue. A mirror, a picture, or a gesture, with no words.
  6. Minimal reminder. One word (tongue), a tap on a reminder card, or a glance.
  7. Self-check prompt. How did that one sound? The student judges and fixes it.
  8. Independent. No cue. The student catches and corrects errors on their own.

The middle of the list is not fixed. Some students respond to a gesture better than to any spoken instruction, and the order should follow the student. For quick data, many SLPs group these into maximal, moderate, and minimal cueing and independent, as on the chart above.

How to fade cues without losing the sound

  • Keep most attempts correct. If the student is missing more than they hit, add a step of support. If they are hitting nearly every one, take a step away.
  • Fade one step at a time, and fade before moving up. A student who reaches the criterion only with a model is not ready for the next level yet.
  • Expect cues to come back at each new level. A student who was independent in words may need a model again in phrases. That is normal and usually short-lived.
  • Fade feedback, not only cues. Early on, feedback describes the movement (your tongue stayed behind your teeth that time). Later it only says whether the sound was right. Later still, the student judges first.
  • Write the cue next to the score. 17 of 20 with a mirror and 17 of 20 with no cues are different skills, and the difference is the progress you want to show.

The feedback point comes from motor learning research. In their R tutorial, Preston and colleagues (2020) describe blocked practice and frequent feedback about the movement as principles for the earliest stages of treatment, while the sound is being established. As Austermann Hula and colleagues (2008) summarize that research, immediate or frequent feedback boosts performance in the moment but can interfere with retention and transfer, while reduced or delayed feedback supports them. Their own study was in adults with acquired apraxia of speech and gave qualified support, so treat this as a guiding principle for children, not a recipe. The broader principles are reviewed in Maas and colleagues (2008).

How home practice fits each level

Therapy time is short, and much of the repetition a new sound needs has to happen between sessions. Research on home practice is thinner than you might expect, but it points the same way. In a study of 91 children aged 4 to 6 by Günther and Hautvast (2010), traditional articulation therapy was effective, and adding contingency management, a behavioral approach that rewards doing the homework, increased how often children practiced at home and made outcomes more consistent from child to child.

The rule of thumb many SLPs use is to send home the level the student already does well with little or no cueing, usually one step below the level being taught in the room. At home there is often no trained ear listening, so the practice should be at a level where the student produces the sound correctly on their own. Otherwise it can turn into practice of the error.

  • Isolation and syllables. Usually kept in the therapy room while the sound is new. If you do send something home, keep it very short and give the family a recording of the correct model to copy.
  • Words. A short list in one position, a few minutes most days, saying each word a few times. Mix positions once each one is solid. Our free articulation word list generator pulls words by sound and position, and the per-sound pages listed below sort words by position and number of syllables and print as one-page homework sheets.
  • Phrases and sentences. Each word list page also has phrases and sentences for that sound. Games with a repeated frame work well here: I spy something that starts with S, or In my suitcase I packed a sock.
  • Reading and structured talk. Each word list page has a short practice paragraph. Reading a page of a favorite book aloud, retelling a show, or describing photos from the weekend all count.
  • Conversation. Pick one short, predictable talking time a day, such as dinner or the drive to school, as target sound time. Agree on a quiet signal instead of correcting mid-sentence, and let the student check their own sound afterward. Our guide to speech therapy homework that students actually do has more ideas for keeping this going.

For the conversation level, there is direct research on what helps. In a study by Koegel and colleagues (1988), children who self-monitored their target sound only in the clinic did not use it correctly outside the clinic. Once they were also asked to self-monitor outside the clinic, generalization was rapid and widespread. An earlier study of children in public school speech therapy (Koegel, Koegel, and Ingham, 1986) found the same direction: once a self-monitoring activity was added, the children began using the correct sound in spontaneous speech outside the treatment setting. The top of the ladder is climbed where the student actually talks.

Motor chaining: the ladder in one sequence

On the traditional ladder, a student might spend several weeks at the word level before phrases are introduced. Speech Motor Chaining compresses the ladder. ASHA describes sessions in three phases, with a structured practice phase that runs in a set progression: syllables, one-syllable words, multisyllabic words, phrases, and self-generated sentences, followed by a randomized practice phase that varies the stimuli. The tutorial by Preston, Leece, and Storto (2019) describes it as building complex speech around core movements, with feedback type and frequency, practice variability, and stimulus complexity adjusted to the child’s performance.

Because each chain is built on one core movement, the sentence at the top contains the same syllable the student practiced at the bottom. For S, a chain might run: suh, sun, Sunday, see you Sunday, and then a sentence the student makes up with Sunday.

  • The whole ladder shows up early. A student can move from syllable to sentence within a single session, rather than meeting sentences weeks later.
  • Difficulty follows performance. The student moves up or back down the chain depending on how the attempts are going, which is the same logic as the traditional criteria, applied minute by minute instead of week by week.
  • Practice is dense. In the tutorial’s data, most children reached over 200 practice trials per session.
  • Isolation is not on the chain. Each session begins with a prepractice phase, which ASHA also calls elicitation, and the student moves on to the chain only after meeting a criterion there. Establishing a brand-new sound still comes first.

Our full guide to speech motor chaining covers the feedback schedule and how to choose chains.

Does the ladder have to be climbed in order?

ASHA describes three ways to organize targets. A vertical strategy gives intense practice on one or two targets until the child reaches a criterion level, usually conversation, before moving on. A horizontal strategy gives less intense practice on several targets in the same session. A cyclical strategy mixes the two, practicing each target for a set period. Taking one sound all the way up the ladder before starting the next is the vertical strategy, and it is not the only option.

Some research has tested mixing the levels instead of climbing them. Skelton (2004) studied Concurrent Treatment, which randomly intermixes easy and hard practice tasks. All four children, aged 4 to 7 with S errors, learned S quickly and generalized it to untaught items from syllables up to connected speech, and the author concluded that an easy-to-hard sequence may be unnecessary. In a later randomized study of 28 children aged 6 to 9 treated in small school groups, Skelton and Richard (2016) found that the treated children acquired their target sounds within 40 sessions of 30 minutes. These are small studies, but they suggest the ladder is a useful map, not a law.

In practice, many SLPs keep the ladder as the backbone and borrow from both ideas: most of a session at the student’s current level, a few attempts at the next level up, and self-monitoring tasks that go home.

When a different map fits better

  • The sound comes easier in one context. ASHA notes that some phonetic contexts can make a sound easier to produce correctly. An SLP may start from the syllable where the sound already comes out right and use it as the building block for longer words and phrases.
  • The errors follow a pattern. When many sounds are affected by the same pattern, such as leaving off final consonants, SLPs often use phonological approaches built around sound contrasts or cycles, which ASHA describes separately.
  • Motor planning is the problem. Childhood apraxia of speech calls for its own treatment approaches; see our guide to childhood apraxia of speech.
  • The sound is not expected yet. Before starting any ladder, check the speech sound development chart by age.

Where LumaSpeech fits

The hardest part of any hierarchy is the time between sessions, when the student is supposed to keep practicing at the right level without an SLP in the room. LumaSpeech is built to keep students on track in that gap, at the levels where home practice does most of its work.

  • Words, sentences, and motor chaining. Students practice at the word, sentence, and motor chaining levels, for any sound in any position, with feedback on the target sound as they go.
  • Conversation. In articulation carryover conversation, the student has a real conversation with Luma and their target sounds are counted. Misses are re-modeled naturally, the way a good conversation partner would say the word back, and never corrected mid-conversation. Our guide to AI conversation practice in speech therapy explains how this works.
  • For the SLP, less time on paperwork. Recordings of each attempt to re-listen to, per-sound tallies and the transcript from conversations, and a session log where you record trial counts with the cueing level and cue type, and whether each set was a probe or treatment trials. That data appears on each goal and feeds the progress report.

Isolation and early syllable work stay with the SLP, where the teaching happens.

Frequently asked questions

What is the articulation hierarchy in speech therapy?

The articulation hierarchy is the sequence of levels a student moves through when learning a speech sound: the sound in isolation, then syllables, words, phrases, sentences, reading or structured speaking tasks, and conversation. Each level adds length and pulls more of the student's attention toward the message, so the new sound has to hold up under more load. ASHA describes the same idea as generalization at increasingly challenging levels, from syllables through conversational speech.

What accuracy should a student reach before moving up a level?

There is no single research-based cutoff. A common convention is about 80 to 90 percent correct across two or three sessions in a row, with little or no cueing, measured on words or sentences the student has not practiced. Districts, programs, and IEP goals may set their own criteria, and the right number depends on the student and the sound.

Do you always have to start with the sound in isolation?

No. Isolation is useful for establishing a brand-new movement, but many SLPs start at the syllable or word level when the student can already produce the sound there, or start from a context that makes the sound easier. Stops such as K and G are hard to say alone without adding a vowel, so they often start in syllables. Some treatment research also mixes easy and hard levels from the start instead of climbing one at a time.

What is a cueing hierarchy in articulation therapy?

A cueing hierarchy is the order in which an SLP reduces support. It usually starts with the most help, such as a touch or tool to show placement or saying the sound together, then moves to a model to copy, placement instructions, a visual cue or gesture, a one-word reminder, a prompt to self-check, and finally no cue at all. Cues can be verbal, visual, or tactile, and the order of the middle steps depends on the student.

At what level should articulation homework start?

Many SLPs set home practice at a level the student already does well with little or no cueing, usually one step below the level being taught in therapy. For most students that means word lists first, then phrases, sentences, reading, and short conversations as each level becomes solid. Isolation and early syllable work usually stay in the therapy room while the sound is new, so the student is not practicing the error at home.

How is motor chaining different from the traditional hierarchy?

The traditional hierarchy moves a student up one level at a time, often over weeks. Speech Motor Chaining practices one core movement from a syllable up to a self-generated sentence in a single sequence, raises or lowers difficulty based on how the student is doing, and follows it with randomized practice. The levels are the same idea; the pace and structure are different.

Sources

Keep every level moving between sessions

LumaSpeech keeps students practicing their target sounds at home, from words and sentences up to real conversation, at the level their SLP assigned, and sends recordings and progress back to the therapist.

See how LumaSpeech supports articulation practice →