Research interpretation
Reading SpeechLP’s articulation practice study: a clinician’s guide
An app can accept a practice card without establishing how a child’s articulation has changed. This distinction is central to reading SpeechLP’s new descriptive report.
SpeechTherapist.app · Released 2026-11-03 · Based on the SpeechLP manuscript dated November 3, 2026
The finding, with its limits
Among 101 returning recorded profiles and 660 matched profile-and-sound pairs, mean accepted-card accuracy was 69.4% in Month 1 and 74.9% in Month 2: +5.5 percentage points (95% CI 0.4 to 10.4). Recorded sessions and practice days declined.
These findings are preliminary and descriptive. Acceptance rates reflect the app’s automated speech recognition decisions, which have not been validated against SLP ratings, and may also reflect changes in app scoring over time. They should not be interpreted as clinical articulation accuracy, within-platform growth, or evidence of treatment efficacy.
What was actually measured?
The outcome was the proportion of eligible scored practice cards that the app accepted. A card could be accepted after repeated utterances and auditory modeling. The study did not count individual speech attempts or model replays, and the measure does not describe first-attempt accuracy, spontaneous naming or clinician-rated production.
Each recorded profile was compared on identical sound labels across two consecutive elapsed 30-day windows, anchored to its first observed eligible card. These are observed practice periods rather than a pre-practice baseline. Different words, sound positions, games and attempts could still occur within a matched sound. Profiles are records, not independently verified unique children.
Why scoring changes matter
Speech recognition and acceptance scoring changed during the January 22–October 7, 2026 observation window. The analysis pooled eligible historical app versions without an app-version adjustment. The contribution of those changes to the observed difference is unquantified.
A higher later acceptance rate can reflect a combination of production, task conditions, target mix within a sound and the scoring system. This report cannot separate those contributions. Its confidence interval describes uncertainty in the observed paired mean; it does not remove confounding or validate the outcome against SLP judgment.
More recorded practice was not the overall pattern
Recorded sessions and practice days were lower in Month 2. The cohort includes profiles with some later practice and matched sounds; it does not represent all families who started using the app. Frequency subgroups were exploratory, and their direct between-group gain intervals included zero. They do not establish a causal dose response.
Individual patterns also differed: 54 profiles had higher later means, 34 had lower means and 13 were unchanged. A positive cohort average is not a finding that all 101 children improved.
How to read practice reports in a clinical conversation
Start with the recorded task: target sound, word position, game, cueing and scoring context. Distinguish app acceptance from a clinician’s own observation of a production. Interpretation of generalization, maintenance and treatment outcomes requires evidence beyond a structured card score.
The report raises questions for further work: how app acceptance agrees with blinded SLP ratings, how performance varies by app version and task conditions, and whether independently measured speech outcomes change over time. This study does not answer those questions.
Read the complete source
Silver, S.; Fox, A.; Gottlieb, M.; Siu, M.; and Ndungu, J. Children’s Articulation Practice in SpeechLP Across Two Months. SpeechLP. Manuscript dated November 3, 2026. Author preprint; not peer reviewed.
Permanent record: doi:10.5281/zenodo.23932485.
The source provides the full manuscript and supplement, all 15 tables, figures, methods, uncertainty intervals, ethics statement, AI-assistance disclosure and aggregate downloads. Only the four aggregate CSV/JSON files are CC BY 4.0; manuscript and figure rights are reserved.
All study authors are affiliated with SpeechLP and receive compensation through equity, pay, or both. SpeechTherapist.app is operated by SpeechLP. This explanation concerns the separate SpeechLP app study; it is not a validation study of the SpeechTherapist.app clinical platform.
Research contact: alissa@speechlp.com.