What Counts as a Usable Articulation Baseline
A usable articulation baseline is a percent correct score tied to three things: a specific sound, a specific word position, and a specific level of the hierarchy. For most students that means one ten-word probe per sound per position, scored in about ninety seconds, plus one conversational sample of two to three minutes. Two short probes per student across the first two weeks is enough to write a present levels statement, choose a starting level, and defend both at the annual review. You do not need to repeat a standardized test, and you do not need to spend September assessing instead of treating.
Baseline data eats sessions when we treat it as an event, a testing week that must finish before therapy starts. Run it as a layer instead. Your first week sessions already have warm up minutes and a get to know you activity in them, and a ten-word probe fits inside that structure without announcing itself.
What makes the number useful in March is consistency, not length. Same words, same elicitation, no model, same scoring rule. Write those conditions on the sheet the first time so future you does not have to reconstruct them.
The Ten-Word Probe: One Sound, One Position, Ninety Seconds
Ten items is the right size for one unglamorous reason: the math is free. Each word is worth ten percent, so you read percent correct straight off the tally, and ten words fit inside a single game turn.
Pull the ten from a fixed list and then freeze them. The word lists by sound are organized by position, so you can lift ten initial /s/ words in a minute. Keep them to one or two syllables, and avoid words carrying a second target sound, since an /r/ word with an /s/ in it muddies the score.
Present the probe with no model: picture naming for pre readers, a printed list for second grade and up, delayed imitation only as a last resort. Imitation inflates accuracy, so if you modeled, mark it and treat the number as a ceiling.
- Score plus or minus only. Approximation ratings are not reproducible six months later.
- Note the substitution for the first three errors: w for r, th for s, t for k.
- Do not cue or repeat. Circle the box if a student self corrects unprompted.
- Stop at ten, even when the number looks worse than you expected.
Freeze your probe words on day one and reuse the identical ten at every progress check. A baseline you cannot reproduce is a number, not a measure.
Probing Up and Down the Hierarchy to Find the Working Level
A word level score tells you where the student is, not where to start teaching. For that you need one probe above or below it. The hierarchy runs isolation, syllable, word, phrase, sentence, and conversation, and week two is for finding the highest rung a student can stand on with effort.
Start at word level with almost everyone, then move by rule of thumb. At or above roughly eighty percent with no model, probe phrase and sentence next session. Under roughly thirty percent, drop to syllable and check stimulability at isolation. Those cut points are clinical convention rather than a research finding, so adjust for the student.
Stimulability at isolation is the most valuable extra data point you can collect this month. A student who produces a clear /r/ in isolation with a visual or tactile cue needs a generalization goal. One who cannot produce it with maximum cueing needs an elicitation goal, and the two look nothing alike on paper.
- Isolation: five trials with your best cue, and note which cue worked.
- Syllable: ten consonant vowel combinations in one vowel context.
- Word: the frozen ten, one position at a time.
- Phrase: the same ten words in a carrier phrase.
- Sentence: ten sentences, one per probe word.
- Conversation: a short sample, scored later from a recording.
Scoring Four Students at Once Without Stopping the Group
Mixed groups are where baseline plans die, and the fix is structural. Give every student a turn that produces scoreable speech, give the waiting students something silent to do, and use a tally fast enough to run one handed.
Group by sound for two weeks wherever your schedule allows. Even a partial regrouping, three /s/ students in one slot and the /r/ students in another, means one probe list on your clipboard instead of four. Trade slots with yourself, then put the schedule back.
Then make the scoring invisible. Turn the probe into the game turn: a card flip, a spin, a move on a board. Students say their ten words to earn the turn, you slash the misses, and nobody experiences it as testing. See our post on first weeks games for mixed groups.
There is also a data source you may already have. Practice tools that score each attempt as it happens, like the pronunciation games on SpeechTherapyMagic, log every production a student makes while your attention is on the child beside them. They do not replace a probe you administered, but they catch practice trials between probes that otherwise vanish.
- Mark errors only. Assume correct and slash the misses.
- One row per student, ten boxes per row, so the group fits on one page.
- Record the students you are least sure about. A phone catches what you miss.
- Never score conversation live. Collect the sample and score it after school.
A Data Sheet That Fits on One Page
The sheet you will actually use is one page per group per week, not one page per student per session. Print a stack in August and clip one inside each group folder.
Across the top put the week, the group, and the activity. Down the left, one row per student with the target sound and position, the level probed, ten numbered boxes, and a notes column. Keep percent correct in a narrow column at the far right so the number always lands in the same place. The notes column earns its space: the substitution you heard, the cue that worked, and anything that changed the conditions, such as illness or a probe you had to model.
- Header: week, group, activity, your initials.
- Row label: student, sound, position, hierarchy level.
- Ten boxes: blank for correct, slash for error, circle for self correction.
- Percent column: count slashes, subtract from ten, add a zero.
- Bottom strip: date of the speech sample and your intelligibility note.
A Two-Week Baseline Schedule You Can Keep
Spread the work so no single session becomes a testing session. Two weeks covers a full caseload if you hold yourself to two probes per student.
If you also inherited students who have never been screened, run your start of year screenings before this window rather than inside it. Screening answers who needs a closer look; baseline answers where to start with the students you already serve.
- Days 1 and 2: routines, rapport, and group rules only. No data.
- Day 3: word level probe for the first half of the caseload, inside a game.
- Days 4 and 5: word level probe for the second half. Score recordings Friday afternoon.
- Days 6 and 7: the follow up probe, one rung up or down.
- Day 8: speech samples for the students whose intelligibility is the real concern.
- Days 9 and 10: catch up absences, transfer the numbers, start treating.
Writing the Baseline Into Present Levels
A baseline is only worth collecting if it lands in the paperwork in a form you can compare against next year. Write it as one sentence with all four variables in it.
Here is a template to copy: On September 18, given ten pictured words with no model or cue, Student produced initial /r/ correctly in 3 of 10 opportunities (30 percent). At the syllable level with a visual cue, accuracy was 7 of 10 (70 percent). In a three minute speech sample, Student was intelligible to a familiar listener with occasional repetition.
That sentence hands you the level to start at, the cue to fade, and the comparison point for the next probe. Turning it into a measurable annual goal is the next step, and our articulation goal bank has the target language. IEP format requirements vary by state and district, so match your local template.
Write the baseline sentence the same day you collect the data. A percentage with no conditions attached becomes uninterpretable within about a week.
Bringing It Together
Baseline data does not need its own phase of the year. Ten frozen words, one page per group, two probes per student, and a short speech sample where intelligibility is the concern will give you a defensible present levels statement and a starting level for every articulation goal on your caseload. If you are still assembling your September plan, the back to school checklist for school SLPs covers the scheduling and paperwork around it.
For the drill between probes, printable articulation drill worksheets and the scored practice games on SpeechTherapyMagic can carry the repetition, so your session minutes go to teaching the sound while data collection stays a ninety second layer underneath. The word lists there cover every English speech sound including blends and vowels, so your probe list and your practice list come from the same place.