Our Practice Approach

Retrieval over Repetition

Retrieving a word from memory produces stronger neural change than repeating a word someone just said. The effort of retrieval is the mechanism. VoicePath prompts retrieval before providing the answer.

Spaced Practice

Items practiced across spread-out sessions are retained significantly better than massed repetition. Daily short sessions — rather than long weekly sessions — produce better long-term naming gains.

Multimodal Input

Visual, auditory, phonological, and motoric channels working simultaneously reinforce each other. VoicePath pairs images + audio + spelling + lip-synced avatar to maximize channel overlap.

Functional Communication

Transfer to real life requires practicing in realistic communicative contexts. Any mode of communication — speech, gesture, pointing — counts toward effectiveness. The goal is communication, not performance.

Picture Word Cards

Confrontation Naming

Confrontation naming — show a picture, patient attempts to retrieve the name — is the cornerstone of word-retrieval rehabilitation. Every card tap is confrontation naming: show image → tap → avatar models pronunciation.

Cueing Hierarchy

SLPs structure cues from least to most powerful: semantic cue → phonemic cue → whole word cue. The letter-by-letter spelling in Picture Word Cards delivers a visual-graphemic cue sequence — each letter is a cue building toward the whole word. Research shows phonemic cues are "extremely effective" for word retrieval.

Semantic Feature Analysis (SFA)

SFA asks patients to name features of a target word (category, use, location, action, association, appearance) to strengthen semantic network connections. Picture Word Cards provides the image and word; the visual representation activates semantic features.

Match the Label

Word-Picture Verification

Picture-word matching is a standard receptive language task sensitive to comprehension deficits. Match the Label directly implements this core assessment and treatment technique. Research shows progressive difficulty from single word → choose among 3 options → match within scenes is the evidence-based progression.

Retrieval from Distractor Sets

Presenting foils forces the brain to discriminate between semantically similar words (e.g., "apple" vs "orange"). This is more cognitively demanding and more therapeutically effective than binary yes/no tasks. The more semantically related the foils, the greater the treatment effect. Match the Label draws foils from within-category words — exactly the right distractor type.

Errorful Learning with Correction

Research shows that making and correcting errors drives deeper encoding than errorless protocols for many aphasia types. Match the Label provides immediate visual feedback on match attempts — implicit errorful learning with correction built in.

Build a Sentence

PACE Therapy

PACE (Promoting Aphasics' Communication Effectiveness) emphasizes that any mode of communication counts and that message effectiveness matters more than grammatical correctness. Build a Sentence's word-picking is a multimodal approach: visual word selection + avatar verbal output + reading along. The construction method removes pressure of spontaneous speech while still producing a communicative utterance.

Oral Reading for Language in Aphasia (ORLA)

ORLA involves repeated reading aloud with a clinician to reinforce phonological and semantic routes simultaneously. When the avatar speaks the built sentence and the user reads along or repeats it, this is the ORLA model: a multimodal simultaneous production task that reinforces both phonological decoding and semantic comprehension.

Script Training

Script training involves practicing a personally relevant set of sentences until they become automatic — greetings, requests, describing needs. Build a Sentence functions as script training when sentence categories map to functional communication needs: choosing what to eat, describing daily activities, making requests.

Rhythm & Words

Melodic Intonation Therapy (MIT)

VoicePath directly implements MIT: rhythmic external pacing, left-hand tapping (which engages right-hemisphere motor circuits to bypass left-hemisphere damage), unison production with a model, and progressive complexity from word to sentence. MIT is one of the most studied aphasia interventions for non-fluent aphasia. The left-hand tapping specifically activates the undamaged right motor cortex and right-hemisphere language areas, creating an alternative pathway for speech production. This is the strongest research-to-feature alignment in the entire app.

Speech Entrainment

Simultaneous speaking with an audiovisual model normalizes connectivity between Broca's area and auditory regions. The avatar provides both the acoustic model and visual lip-sync — precisely the multisensory combination research identifies as most effective. The "AT THE SAME TIME as your tutor" instruction correctly emphasizes synchrony over imitation.

Rhythmic Auditory Stimulation (RAS)

RAS uses an external rhythm as a temporal scaffold for motor output. Originally developed for Parkinson's gait rehabilitation, it applies directly to speech timing. The drum beat is this scaffold. Per-syllable IPA phoneme pronunciation via Azure Neural TTS ensures each syllable is acoustically correct on the beat — critical for motor pattern reinforcement.

Sing Along

Familiar Melody & Right-Hemisphere Activation

MIT was originally inspired by the observation that many patients with severe non-fluent aphasia who could not speak spontaneously could sing familiar song lyrics with correct articulation. The neurological reason: melody and rhythm processing is dominated by the right hemisphere, which retains language-capable regions even after left-hemisphere stroke. Familiar songs reduce cognitive load — the patient already knows the next word before it arrives, which relieves the motor planning burden that breaks down in Broca's aphasia. This makes singing easier than speech, and that easier pathway is what MIT exploits.

MIT Intoning Protocol — Unison Production

MIT does not require pitch-perfect singing. It uses intoning — exaggerated, slowed melodic prosody that approximates a sung quality. The clinical protocol: (1) therapist intones the phrase, (2) patient joins in unison, (3) therapist fades and patient continues alone. Sing Along implements steps 1 and 2: the avatar intones each line via SSML prosody control (raised pitch, reduced rate) and the patient sings along simultaneously — the same unison production that is the core of MIT. The left-hand tapping continues from Rhythm & Words, engaging the right-hemisphere sensorimotor network that controls both hand and mouth movements.

Nursery Rhymes & Song Structure in Speech Therapy

Nursery rhymes and simple folk songs are ideal MIT targets: short lines (4–7 syllables), strong predictable rhythm, highly repetitive structure, and cultural familiarity ingrained since childhood. The repetition within a single song is intentional — MIT protocols repeat each phrase multiple times to strengthen the motor engram. Songs also provide a phrase-level rather than word-level target, extending the MIT approach to connected multi-word utterances in a natural, low-anxiety context.

Public Domain Song Library

All 17 songs are confirmed public domain: 7 traditional nursery rhymes (pre-1900), 5 American folk songs (Oh Susanna 1848, Clementine 1884, Down in the Valley traditional, She'll Be Comin' Round the Mountain traditional, Skip to My Lou traditional), and 5 hymns and gospel songs (Rock of Ages 1763, What a Friend We Have in Jesus 1855, Blessed Assurance 1873, Jesus Loves Me 1860, The Old Rugged Cross 1913). Under U.S. copyright law, all works published before January 1, 1928 are in the public domain.