Write it at or slightly below the student's receptive language level, not their grade level and not their decoding level. Carol Gray methodology treats comprehension as the ceiling. In a 2024 survey of 16 school SLPs, pediatric OTs, and parents, 94% spent 30 or more minutes on a single story, and rewriting one because the language landed too high is a large share of that time.
Why receptive language and not grade level?
Because a student can read a sentence perfectly and still not picture what it means. Grade level describes where a student sits on a roster. Receptive language level describes what they actually understand when someone talks to them, and that is the channel a social story uses when an adult reads it aloud.
Carol Gray and The Gray Center are explicit that a story is written for the audience's comprehension, using vocabulary and structure the person already has. That rules out writing to the grade on the IEP cover page. It also rules out writing to a district reading benchmark, which measures decoding.
What is the difference between reading level and receptive level?
Reading level is decoding. Receptive level is understanding. These two numbers separate widely in the K-5 autistic population, and the gap runs in both directions.
| Student profile | Decoding | Receptive | Write the story at |
|---|---|---|---|
| Hyperlexic second grader | Grade 4 | Age 4 to 5 | Age 4 to 5 vocabulary, adult reads aloud |
| Non-reader kindergartner | Pre-primer | Age 5 | Age 5 vocabulary, picture-carried |
| Fourth grader with language disorder | Grade 2 | Grade 2 | Grade 2, matched |
| Fifth grader, strong language, poor decoding | Grade 1 | Grade 5 | Grade 5 content, read aloud or audio |
| AAC user, minimally speaking | Unknown | Estimate from directions followed | One idea per page, symbol-supported |
That last row is the one people avoid. When decoding is unknown, receptive is still estimable. Watch how many steps of a direction the student follows without a prompt, and write the story in units that size.
How do you find the receptive level without new testing?
Open the most recent evaluation and take the receptive subtest age equivalent, then round down one level. That number is already in the file. You do not need a new assessment to write a story, and waiting for one is how a story for next Tuesday's assembly gets written the night before.
If there is no usable evaluation, use a behavioral proxy. The longest direction the student follows independently, without gesture and without repetition, is a workable estimate of how much language they process at once. A student who follows "get your backpack" but not "get your backpack and line up by the door" is telling you to write one-clause sentences.
Do readability scores like Lexile or Flesch-Kincaid help?
Use them as a sanity check, never as the target. Readability formulas count syllables and sentence length. They cannot see abstract vocabulary, and abstract vocabulary is what breaks comprehension in a social story.
"My friends feel disappointed" scores as easy text. Every word is short. The concept of another person's disappointment is the hardest thing in the story. A formula will pass that sentence and a five-year-old will not.
The practical rule: run the check, and if the score is far above your target, the story definitely needs work. If the score is at target, that tells you almost nothing, and you still have to read it for abstraction.
From the same 2024 community survey: "Getting suitable pictures is 90% of the work." Reading level and images are the same decision. If the language sits at the student's level, the picture only has to confirm it. If the language sits above, the picture has to carry the whole meaning, which is why photo hunting expands to fill an evening.
What sentence length actually works in K-5?
Five to eight words for K-2, eight to twelve for grades 3 to 5, one idea per sentence at every level. These are starting points you adjust after the first reading, not rules.
Two specifics matter more than the word count. First, repeat the noun instead of using a pronoun. "Ms. Chen rings the bell. Ms. Chen waits" is clearer than the version with "she," and it costs one word. Second, keep the tense consistent. Sliding between "I go" and "I will go" inside one story adds a comprehension load that the student spends attention on instead of the content.
Can simplifying go too far?
Yes, and the failure mode is predictable. When a story gets cut down hard, the descriptive and perspective sentences go first, because they read as optional. What is left is a list of directives, which is a rule sheet with pictures.
AFIRM lists social narratives as an evidence-based practice for autistic learners aged 3 to 22, and the social narratives brief packet describes narratives that explain context and others' perspectives, not narratives that only issue instructions. Simplify the words. Keep the explanation.
The 2026 Frontiers in Psychology meta-analysis of 21 single-case studies reported a moderate overall effect, Tau-U of 0.743, strongest for children aged 7 to 12, with no significant effect on classroom adaptive skills. Effects like that come from stories that teach understanding. A stripped directive list may still be worth having, and it is not what the research studied.
How do you check the level was right?
Ask two questions after the second reading, one factual and one about someone else. "Where do we walk when the bell rings?" and "What does Ms. Chen want us to do?" take thirty seconds together.
Both answers correct means the level is fine. Factual correct and perspective blank means the vocabulary is fine and the abstract content needs a rewrite, usually by naming the feeling more concretely. Neither answer means the whole story is above the student, and the fix is shorter sentences and more pages, not a shorter story.
Do this on the second reading, not the first. First readings measure novelty.
Frequently Asked Questions
What reading level should a social story be written at?
At or slightly below the student's receptive language level, which is usually lower than their decoding level and lower than their grade placement. Carol Gray methodology treats comprehension as the ceiling, because a story the student can read aloud but cannot picture does not change behavior.
Is reading level the same as receptive language level?
No, and confusing them is the most common error. Reading level describes decoding, and receptive language level describes understanding. A hyperlexic student may decode fourth grade text and comprehend at a first grade level, so the story should be written for the lower number.
How do I find a student's receptive language level without new testing?
Pull the receptive subtest age equivalent from the most recent evaluation in the file, then round down one level. If no evaluation exists, use the longest direction the student follows independently in class as your working estimate.
Should I use a Lexile or Flesch-Kincaid score for a social story?
Use them as a rough check, not a target. Readability formulas count syllables and sentence length, and they cannot see the abstract vocabulary that actually breaks comprehension. A sentence like "my friends feel disappointed" scores as easy and is not.
What sentence length works for K-2 students?
Five to eight words per sentence, one idea per sentence, and one sentence per page for most kindergarten and first grade students. Second grade can usually handle two sentences per page if the second one repeats a noun rather than using a pronoun.
Can a social story be too simple?
Yes, in one specific way. Oversimplifying strips out the descriptive and perspective sentences that carry the reasoning, which leaves a rule list. Keep the sentence ratio and simplify the words instead of deleting the explanation.
How do I know the level was right?
Ask the student two comprehension questions after the second reading, one factual and one about what someone else in the story wants. If the factual answer comes back and the perspective answer does not, the vocabulary is fine and the abstract content needs rewriting.
One approach for school SLPs short on time is to keep a 5-tool stack: a methodology checklist for the descriptive-to-directive ratio, a slide template you reuse, a folder of stock photos sorted by scenario, an AI text drafter (ChatGPT, Claude, MagicSchool, or Emoquest for one-sentence-in story output), and a delivery format your district already uses (Google Slides or PDF). Add one more thing to the stack: a note field on each student in your caseload list with their receptive age equivalent. Whatever drafts the text, you are the one supplying that number, and having it at hand turns a rewrite into a first draft that lands.