What a Screening Needs to Answer
A start-of-year articulation screening is a pass or refer decision, not a diagnosis. It should take three to five minutes per student and answer three questions: which sounds are in error and in which word positions, whether the student can produce something better with a model and a cue, and whether the errors make the student hard to understand. If your tool answers those three, it is doing its job.
Keep that boundary clean in your paperwork. A screening samples behavior; an evaluation gathers norm-referenced data, case history, observation, and parent and teacher input, and ends in a team eligibility decision. Never word a screening result so it implies a diagnosis, because only a full evaluation by a licensed SLP can determine whether a child has a speech sound disorder. Consent rules vary too: some districts treat screening as a general education activity, others require written consent first, so check your handbook.
Stimulability is the most useful column on the sheet. A student who can produce the target with a model and a cue has a different outlook than one who cannot, and that shapes both the referral decision and your first therapy target.
Who to Screen, and Who to Skip
Screening everyone collapses under caseload math, so prioritize. The highest-yield groups are kindergarten and any grade your district designates for universal screening, new enrollments with no records, teacher-flagged students, and students dismissed or placed on monitor within the past two years.
Two categories need care. Students already on your caseload need baselines, not screenings. And a student who speaks another language at home may produce English sounds in a way that reflects their first language rather than a disorder. Errors appearing only in English that match the home language's patterns generally point to difference rather than disorder, which is covered in English language learner or speech sound disorder.
Designing a Five-Minute Probe
Build one probe and use it identically all year. Consistency beats coverage, because a screening becomes far more useful when you can repeat it in eight weeks and compare directly.
Part one is twenty to twenty-five single words covering the sounds most likely to be in error at school age: s, z, sh, ch, j, th, f, v, l and r, each in initial, medial and final position where the sound allows, plus two clusters and the vocalic r variants (ar, or, er, air, ear, ire) that short screeners routinely miss. Part two is three to five sentences to repeat, loaded with the same targets, which shows whether accuracy survives a higher motor demand. Part three is thirty to sixty seconds of connected speech.
Use pictures for younger students and a printed list for readers, and note which, because a reading demand depresses performance in a student who struggles to decode. Assembling targets by sound and position is the slow part, so word lists by sound save real time. For the speech sample ask something answerable without effort, like how to play their favorite game, and skip summer vacation prompts, which produce three words. Listen while they talk and score right after.
Listening for Intelligibility, Not Just Errors
Two students with identical error inventories can be nowhere near the same clinical picture. A student with a lateral s distortion is usually completely intelligible; a student who drops final consonants and reduces clusters may produce fewer errors on your word list and still be hard to follow. Error counts alone mislead, which is why the speech sample is not optional.
Rate intelligibility on a scale you apply the same way every time, and record the listener condition. Four points works: understood easily, understood with occasional repetition, understood only with topic knowledge or repeated attempts, mostly unintelligible. Note whether you knew the topic in advance, because that quietly inflates the rating.
Look for pattern as well as inventory. When errors follow a rule across a class of sounds, such as back sounds moving forward or every cluster reducing to one, you are seeing a phonological pattern rather than isolated articulation errors. Phonological means the errors follow a rule; articulation means a specific sound is produced incorrectly. Patterns usually cost more intelligibility than the error count suggests.
Age Ranges, Not Cutoffs
The refer decision usually hangs on whether a sound is late enough to matter. Use ranges rather than one cutoff, because published norms differ by source and children vary widely within them. The broadly accepted picture: early sounds including p, b, m, n, h, w, d, t, k and g are typically in place by about three to four; later sounds including l, s, z, sh, ch, j and v tend to settle by about five to six; R and TH are usually last, often by six to seven, with TH sometimes running to seven or eight.
So a first grader still saying wabbit sits inside the range where many children are still developing R, and a kindergartner substituting f for TH is very common. Ranges vary by source and by child, and only an evaluation by a licensed SLP can determine whether a child has a disorder. The speech sound development chart is a useful handout when a teacher or parent wants them laid out.
Age is one input among several. Weigh it against stimulability, intelligibility, whether errors follow a rule, and whether speech is affecting classroom participation or early literacy. A student who is age-appropriate on paper but has stopped answering in class deserves a team conversation.
Never refer on age alone. Age, stimulability, intelligibility, error pattern and functional impact together make a defensible recommendation; any one of them by itself does not.
Documenting the Result
Write it up the same day, in a form short enough that you will actually do it: date, grade, age in years and months, which stimulus set and whether it was pictures or reading, the error inventory by sound and position, error type, stimulability per error, the intelligibility rating, and a disposition.
Keep the disposition to three options so it stays a decision. No concern at this time. Monitor, with a recheck date on your calendar. Refer to the team for consideration of an evaluation. A monitor disposition without a date is functionally a no concern disposition, because nobody comes back to it.
Word the write-up as observation and recommendation, never diagnosis. Errors observed on s and z in all positions, stimulable for s in isolation with a visual cue, intelligible to an unfamiliar listener with occasional repetition, referred to team for consideration of evaluation. That holds up in a meeting. A September screening is also the most valuable document you own when a concern surfaces in February, and for students who become yours it feeds the probes in baseline data in the first two weeks.
A Teacher Referral Form You Can Hand Out Monday
The quality of your referrals is set by the form you hand out. A form asking is there a speech concern gets a checkbox; a form asking for examples gets something you can act on. Keep it to one page, ask for specifics, and put a next-steps line at the top so teachers see the form as the start of a process rather than a request for services.
Most districts route classroom concerns through a general education problem-solving or multi-tiered support process before an evaluation referral, and parents may request an evaluation in writing at any time. Those procedures vary by state and district, so write that line from your own handbook. Teachers also need to know what is worth flagging, which is a training problem rather than a form problem: send a guide such as when to refer a student to the speech therapist with the form, then talk in the hallway two weeks later.
- Student name, grade, date of birth, teacher, today's date
- Three examples of words or sentences that were hard to understand
- How often you ask this student to repeat: rarely, sometimes, often, most of the time
- Whether peers understand this student in group work and at recess
- Whether the student avoids speaking or reading aloud
- Languages spoken at home, and who to contact about interpretation
- What you have already tried in the classroom, and what happened
- Hearing and vision screening status and date
Bringing It Together
A good screening season runs on repetition, not sophistication. One probe, used identically, five minutes, scored the same day, ending in one of three dispositions: no concern, monitor with a date, or refer to the team. Do that for two weeks in September and you decide from evidence, not memory, all year. The rest of the setup those weeks depend on is in the back-to-school checklist for school SLPs.
For students who move from screening onto your caseload, having practice pages and word lists for every sound in one place shortens the gap between the decision and the first productive session. The scored pronunciation games on SpeechTherapyMagic mark whether the target sound was correct on each attempt, an easy way to gather extra trials on a stimulability question. Creating an account is free, and it supports the protocol you already run.