Place & Manner stands you in front of an anatomical model of a head, cut in half from nose to throat, and hands you three dials. Every consonant in English is an answer to three questions — where the airflow is squeezed (place), how the air gets past (manner), and whether the vocal folds are buzzing (voicing) — and the lab’s whole premise is that once you can set those three dials on purpose, the consonant chart stops being a table and starts being a map of your own mouth. It lives on the Speech Pathology library because that library describes sounds this way constantly; this is the room where the vocabulary becomes physical. One thing said plainly, because the game itself says it plainly: this is a phonetics teaching lab, not therapy, screening or assessment, and it never listens to your voice.
How to play
A prompt names a sound — the first sound in a word you know. You slide the place dial and watch the tongue travel: both lips for bilabial, lip against teeth for labiodental, tongue tip at the teeth for dental, the bumpy alveolar ridge where so much of English lives, then post-alveolar, palatal, velar, and finally glottal, where there is no tongue involved at all and the constriction is the vocal folds themselves. Manner picks how air gets past — a stop blocks it completely and releases with a burst, a fricative hisses through a narrow gap you can hold forever, a nasal drops the soft palate and sends air out the nose, a lateral lets it escape around the sides of the tongue, an approximant gets close but never close enough to make noise. Voicing is one switch: folds buzzing, or open and still.
Press SAY IT to hear the setting you built, then lock it in. Everything works by mouse or touch; on a keyboard, V toggles voicing, SPACE speaks, ENTER commits, R restarts, and you can drag to turn the model. The best moments are the minimal pairs: build s, then flip one switch to get z with a hand on your throat. One feature changes; everything else is identical.
What’s actually in the head
The model is a mid-sagittal slice — the same view a phonetics textbook draws — built as an upper boundary curve, a lower one, and a thickness, so it reads as a specimen mounted in cross-section. The tongue is not a set of poses; it bunches wherever the selected place is. The soft palate is a flap that genuinely drops out of the way for nasals, opening the nasal cavity drawn above the palate.
The sound is the part worth lingering on: nothing is recorded or sampled. Each consonant is synthesised — a glottal buzz shaped by three formant filters, a burst for stop releases, a voice bar during closure — and it is always spoken against the vowel /ɑ/ so the formant transition is audible. Each place carries its own second-formant locus, the actual acoustic cue listeners use to hear where in the mouth a consonant was made. When you slide the place dial and the sound bends, that bend is the phonetics.
How it was built
Place & Manner is hand-written WebGL and JavaScript with no engine, and the audio is a handful of Web Audio filters rather than a sample bank — the commit that landed it is titled, accurately, “three dials, one head, and the sound that falls out.” The head, tongue, palate and nasal cavity are drawn as extruded slabs behind the moving parts, driven directly by the three phonetic features rather than by canned animations, which is why an impossible setting looks impossible.
The boundaries were built in on purpose. The lab describes sounds; describing a sound is not the same as teaching someone to produce it. If you are worried about a child’s speech or your own, the game’s own closing screen says what we would say here: a licensed speech-language pathologist is the right next step — not a browser game.