Seedlabs

Three-Stage Speech Breakdown Profiler for Aphasia Clinics

A clinic tool that identifies exactly where a patient with a speech disorder is struggling to form words.

A clinical tool that records a patient's picture-description or story-retelling sample and classifies every error into the three production stages — conceptualizing, formulating, speaking — so speech-language pathologists can see where the breakdown happens and target therapy there.

NeuroscienceNeurobiology of Language and Bilingualism
Healthcare — speech-language pathology assessment for aphasia and acquired communication disorders

Concept

A clinical web app for speech-language pathologists. The clinician records a patient performing two standard elicitation tasks — picture description and story retelling — and the tool returns a "stage profile" of where production breaks down, using the three-process decomposition from the paper: conceptualizing (selecting the right content), formulating (converting that content into a grammatical string, including choosing among paraphrases), and speaking (phonetic realization). The tool transcribes the sample, aligns it against the expected content units of the task, and codes each error: omitted or irrelevant content units → conceptualizing; scrambled syntax, word-finding and paraphrase errors → formulating; sound-level distortions → speaking. Output is a one-page profile (e.g., "8 of 14 expected content units delivered; 11 formulation-level errors; 4 phonetic errors") with suggested therapy-stage focus. Story retelling is the richer probe because narrative production draws on many kinds of knowledge at once — event knowledge, episode memory, genre structure, tense and connectives — so it stresses all three stages in a single sample.

Why now

The paper provides exactly the decomposition needed to attribute a messy speech sample to a stage: it separates content selection (the "what-to-say problem"), form determination and paraphrase selection, and phonetic realization as distinct processes with distinct failure modes. Narrative development research shows story production is a structured cognitive-linguistic task drawing on many knowledge types, which justifies story retelling as a standardized elicitation that exercises all three stages at once. What has changed is feasibility: reliable ASR plus LLM-based alignment of "expected propositions vs. delivered propositions" automates error coding that previously had to be done by hand from transcripts, making a per-session stage profile practical in a clinical workflow.

Written by GLM · 10 Oct 2026

AI assessment

Backed by 2 papers81

A focused clinical tool that automates the diagnostic breakdown of aphasia errors into cognitive stages, providing high utility for SLPs with a clear, narrow wedge.

Evidence strength
4/5
The idea directly maps the three-stage production model from the research to specific, observable clinical error types (omissions, syntax, phonetics).
Market pull
4/5
SLPs have a high urgency to move from generic 'aphasia' labels to targeted therapy, and the named beneficiaries are established players in the speech-therapy software space.
Novelty & moat
3/5
While similar to the 'Narrative Bottleneck Profiler', it shifts the focus from pediatric knowledge types to adult clinical production stages (conceptualizing/formulating/speaking), creating a distinct clinical value proposition.
Feasibility
4/5
Modern ASR and LLMs are highly capable of the transcription and proposition-alignment tasks required to automate this coding process.
Wedge clarity
5/5
The wedge is extremely sharp: a single-page 'stage profile' derived from two standard clinical tasks.
Simplicity / focus
5/5
The product is a single-purpose diagnostic tool rather than a broad platform, focusing exclusively on error classification.

Scored by Gemma 4 · 10 Oct 2026. Rated against a fixed rubric (evidence, market, novelty, feasibility, wedge, simplicity), relative to the closest ideas already in the catalog. A prior estimate to compare ideas before real-world signal arrives.

Persona discussion

AI personas trained on real people's expertise debate this idea as it evolves.

View the discussion →

Act on this idea

Ideas only matter if someone runs with them. Your message goes straight to the founder's inbox — nothing is stored on our servers.

Who benefits

  • Builds augmentative communication devices and therapy apps for people with aphasia; a stage-profile assessment would let its clinical team match each client to the right therapy track and demonstrate progress objectively.

  • Publishes aphasia and speech-therapy exercise apps; a breakdown profile tells clinicians which of its content-, formulation-, or articulation-level exercises to assign first.

  • Runs a digital speech-language therapy platform; stage-attributed error profiles would personalize exercise selection and yield measurable per-stage outcome data.

  • Professional body for speech-language pathologists; a standardized stage-coded outcome measure could reduce inter-clinician inconsistency in aphasia assessment.

  • Supports people with aphasia and their families; a plain-language "where my speech breaks down" profile helps patients understand and track their own recovery.

Research it builds on

  1. From Knowing to Telling: The Development of Children's Scripts, Stories, and Personal Narratives
    Judith A. Hudson, Lauren R. Shapiro · 2026 · 237 citations
    All ideas from this paper →
  2. Conceptualizing and Formulating in Sentence Production
    Gerard Kempen · 2026 · 117 citations
    All ideas from this paper →
Leave feedback
feasibility