How to measure progress in speech therapy: from isolated sound to spontaneous speech
In many therapy centres, progress is recorded using a system designed for ABA: correct trials, prompted trials, incorrect trials. For speech therapy, this system says too little. A child might pronounce the isolated sound "r" correctly and still not use it at all when speaking. Both situations would show up, in an ABA table, as a "correct trial".
In speech therapy, the question isn't just "did they succeed?", but at what level did they succeed, and how close were they to the target. Here is a simple way to record exactly that, session by session.
Why correct and incorrect trials aren't enough
A speech therapy target passes through several stages before it reaches everyday speech. A sound practised well in the therapy room, in front of the mirror, is not yet a sound used spontaneously at the dinner table or at nursery.
What's more, there are steps between "incorrect" and "correct" that matter. A child who produces an approximation of the sound is making progress, even if they're not yet producing it correctly. A child who only succeeds right after your model is at a different point than one who succeeds on their own.
The five levels
Let's take the sound "r" as an example:
- Isolated sound: the child produces "r" on its own, separate from other sounds.
- Syllable: "ra", "re", "ri", then syllables where the sound comes at the end, "ar", "or".
- Word: "rabbit", "carrot", "car", with the sound at the start, in the middle and at the end of the word.
- Sentence: "The rabbit runs to the river." The sound now appears in connected speech, but still rehearsed.
- Spontaneous speech: the child uses the sound correctly while telling a story, playing or asking for something, without paying attention to it.
By recording the level you're working at in each session, you can see straight away where the child stands on this path, not just how many trials they got right.
Four responses, not two
For each trial, record one of four responses:
- Incorrect: the target is missing or replaced with another sound.
- Approximate: the child gets close to the target, but the production isn't yet correct.
- After a model: they succeed immediately after you've shown them how.
- Correct: they succeed on their own, without a model.
The difference between "after a model" and "correct" shows how independent the child is. The difference between "incorrect" and "approximate" shows progress that would otherwise stay invisible, especially in the first few weeks.
What progress means
Progress is read at the level you're currently working at: the proportion of correct responses rises from one session to the next.
There is an important detail to watch for. When a child moves up to a new level, say from syllables to words, the proportion of correct responses often drops at first. That isn't a regression, it's a harder task. For that reason, compare sessions only within the same level. A graph that mixes levels together may show a drop at exactly the moment the child has taken a step forward.
When to move on to the next level remains a decision for the speech therapist. Many teams set a criterion, for example a certain number of consecutive sessions with consistently correct responses. What matters is that the criterion is the same for every therapist on the team.
What this looks like in a session
- Choose the target and the level you're working at today.
- For each trial, record one of the four responses.
- At the end, add a short note: what helped, what tired the child, what you'll try next time.
It takes a few seconds per trial, and by the end of the month you have a clear picture, not just an impression.
What you show parents
Parents understand levels easily. "We're working on words now, and the child succeeds on their own in most trials" says more than a percentage without context. You can also give them homework suited to the level: if the child is working on words in the therapy room, it makes no sense to ask for whole sentences at home.
In TempoApp
In TempoApp, speech therapy programmes are recorded exactly this way: you choose the level, from Isolated sound to Spontaneous speech, then each trial as Incorrect, Approximate, After a model or Correct. Progress is calculated only from sessions at the current level, so moving up to a new level doesn't show up as a regression, and parents see it in their own portal.