How do I track carryover from app practice to spontaneous speech?
With probe data you collect, not data the app supplies — Talk & Listen deliberately does not score, track or report on a child, so there is nothing to export. Baseline four contexts before home practice starts: structured clinic, app practice, parent-elicited at home, and spontaneous at home. Then read the gap between them rather than the totals. High accuracy in the app with low spontaneous production is stimulability without generalisation, and the response is to change the practice context rather than increase the dosage.
The short version of what this page is: we do not have your data, we are not going to pretend otherwise, and here is how the app fits into a carryover plan you run yourself.
Why There Is No Dashboard
Talk & Listen does not score, track or report on a child. There is no clinician portal, no exportable accuracy log, and no per-session summary — and this is a design decision rather than a roadmap gap.
Measuring speech progress properly means standardised administration and scoring by someone qualified to do it. An app inferring accuracy from taps would be producing a number that looks like a measure and is not one, and in a caseload context that is actively harmful: a reassuring figure delays a decision, and a discouraging one prompts a change of course that the data never supported.
It also means the app holds no data about any child, which is the same reason it needs no account and works entirely offline. The trade is deliberate: you lose an export button and gain a tool you can hand to a family without an information-governance conversation.
The Four Contexts
Carryover is a question about the distance between contexts, so the data has to be organised that way. Baseline all four before home practice starts — the spontaneous column with no pre-practice value is the single most common reason a home programme cannot be evaluated afterwards.
| Context | What it measures | Collected by | What it cannot tell you |
|---|---|---|---|
| Structured clinic | Accuracy under maximum support | You | Anything about use outside the room |
| App practice | Engagement and adherence. Whether the child will actually do it | Parent report | Accuracy — nobody is scoring, and the app is not either |
| Parent-elicited at home | Production with a familiar prompt, outside clinic | Parent, one target | Spontaneity — it is prompted by definition |
| Spontaneous at home | Carryover. The only column that answers the question | Parent tally, unprompted only | Little — but it needs a pre-practice baseline to mean anything |
Probe, Do Not Count Practice
The most available number in a home programme is dosage, and it is the least informative. "Twenty minutes most days" is adherence data. It belongs in the notes as context and nowhere near the outcome.
Keep the measurement on short probes you control: ten trials, same elicitation, same scoring rules, at a fixed interval. Everything else — app usage, parent impressions, the child's enthusiasm — is context for interpreting the probes rather than a substitute for them.
Reading the Shape
The pattern across columns is more informative than any single figure.
High in app, low spontaneous
Stimulability without generalisation, and the most common shape here. The app supplies a known target, a predictable prompt, a fixed pause and a visual cue — heavy scaffolding. Vary the context, not the dosage: same targets, less support, more partners.
Spontaneous rising, app unremarkable
Carryover happening. Easy to miss if you weight the practice context, and a good argument for not collecting app data at all — it would have pulled attention to the wrong column.
Flat everywhere at four weeks
The programme is not working as constructed. Check adherence first, since a plan nobody ran is not a plan that failed — then target selection, then whether the practice context suits this child at all.
What to Ask the Parent For
Less than feels sufficient. Parent-collected data degrades quickly with complexity, and the failure is not partial — sheets stop being filled in entirely rather than getting sparser.
- One target. Rotate weeks if you need more; do not add columns
- Tally marks only. Unprompted productions. No ratings, no notes, no times
- One week per sheet, on paper, somewhere already visible — a fridge beats an app for this, reliably
- Explicitly: prompted repetitions do not count. Say it twice; it is the instruction most often lost
The parent-facing version of this method — the same three-word tally, written for families rather than clinicians — is at how to tell if a speech app is actually working, if it is easier to send a link than explain it twice.
Where the App Genuinely Helps
Being precise about this, since the rest of the page has been about what it does not do.
- Adherence. The binding constraint on most home programmes is not the protocol, it is that practice stops happening. A practice context a two-year-old asks for is worth a great deal, and it is the one thing an app is genuinely good at.
- Repetitions. High-frequency production opportunities in a format that does not require a parent to run a session.
- No governance overhead. No account, no child data, no network. You can recommend it to a family without a data-protection conversation, which is not true of most of the category.
- Cost. $4.99 once per device, so recommending it does not put a family into a subscription. Relevant when you are advising across a caseload with mixed circumstances.
And what it is not: an assessment, a scoring instrument, a substitute for therapy, or a source of data about your client. The full boundary, including caseload licensing, is on the professionals page.
The Short Version
Baseline four contexts before you start. Keep the measurement on probes you run. Read the gap between the app column and the spontaneous column rather than either number alone — and when that gap is wide, change the practice context rather than asking the family for more minutes.
Ad-free & offline ready
