READY: 36 SYLLABLES
Within-Word Mean TP
1.00
Deterministic syllable pairs
Between-Word Boundary TP
0.33
Statistical dip at boundary
Consonant Plosive Contrast
+3.8 dB
Acoustic onset salience
Canine Boundary Discrim.
88.4%
Simulated d' = 2.18

Continuous Syllable Ribbon & Boundary Detection

Click any syllable to trigger isolated formant synthesis and inspect transitional probability
Within Word Word Boundary (TP Drop) Current Auditory Playback
Statistical Transitional Probability Profile P(Si+1 | Si)
Simulated Canine Auditory Cortex ERP (Cz/Fz) N1-P2 & Boundary MMN (μV)

Canine Lexical Habituation & Probe Test

Test whether the canine auditory system treats learned words differently from part-words (spanning a boundary) and non-words.

3 Validated Probes Ready
Stream ready for auditory statistical analysis.

How Canine Statistical Learning Works

Just like 8-month-old human infants in classical developmental linguistics (Saffran, Aslin, & Newport), domestic dogs (Canis familiaris) are exposed to fluid, unbroken streams of human speech without pauses between words.

Recent canine electrophysiology and fMRI studies demonstrate that dogs track conditional probability: given syllable A, how frequently does syllable B follow? Syllable pairs within words have high probability (~1.0), whereas boundary transitions drop to 0.33 or lower.

The Role of Consonants & Acoustic Boundaries

Human speech research has shown that vowels carry prosody, but consonants provide the temporal scaffolding for lexical boundaries. Recent canine brain studies show dogs specifically recruit auditory cortical mechanisms to parse consonant onset bursts.

When consonant onsets coincide with transitional probability troughs, dogs register pronounced mismatch negativity (MMN) and enhanced P2 components, demonstrating subconscious word segmentation before semantic comprehension begins.

Auditory Formant Synthesis Architecture in this Lab

This simulator uses the Web Audio API to construct synthetic vocal-tract resonances. Each consonant (e.g. /b/, /d/, /p/, /t/, /g/, /k/) is modeled via short bandpass-filtered noise bursts and rapid frequency modulations of Formant 1 (F1) and Formant 2 (F2), followed by sustained harmonic vowel formants (/a/ = 730/1090 Hz, /i/ = 270/2290 Hz, /u/ = 300/870 Hz, /o/ = 500/1000 Hz, /e/ = 530/1840 Hz).

Event-Related Potential (ERP) Electrophysiology Model

The simulated neural trace reproduces canonical auditory vertex potentials measured in canine studies using non-invasive scalp electrodes: the sensory N1 (~100 ms) and P2 (~200 ms) obligatory waves, superimposed with a distinct Mismatch Negativity (MMN) wave peaking around 280–350 ms specifically when transitional probability dips, indexing boundary detection.

Enjoy this tool? Build your own with Super