← all systems · see it in The System
Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies
Phonetics is a branch of linguistics that mainly concerns the articulation, sound wave properties, and perception of speech sounds.
Properties published scienceour proposal
Sourced fields, and our own readings.
| born at | field: 'as early as the 6th century BCE'; the sounds: contested |
| born precision | century (field); origin of the sounds contested by two orders of magnitude |
| boundary | Inside: speech sounds as phonetics describes them (phones, made by an airstream modified by the larynx and the vocal tract, carried as pressure waves, taken in by the ear and sorted into categories), the speech chain between talker and listener with the talker's own feedback, and the practice of phonetics (articulatory, acoustic and auditory phonetics, instruments, transcription with the IPA, phonetic description and typology, and their applied uses). Outside: phonology (how a language organises its sounds into phonemes; booked as a neighbour floor of language and grammar), words, grammar and meaning, the ear and brain as whole systems, the physics of sound in general, and sign languages' own walk (phonetics 'also addresses equivalent aspects of sign languages', booked as one channel here and walked in sys:sign-language). |
| nested level | A branch of linguistics (field side) and language's second floor (thing side): the vocal tract making the signal and the ear taking it in. It runs on people (talkers, listeners, phoneticians) and their instruments; speech sounds ride on acoustic waves in air. |
| scale | social |
Measures (32) published science
| information rate of speech | mean 39.15 bits/s, SD 5.10 (17 languages) |
| speech rate | mean 6.63 syllables/s, SD 1.15 |
| speech rate across speakers | about 4.3 to 9.1 syllables/s |
| information per syllable | 4.8 (Basque) to 8.0 (Vietnamese) bits/syllable |
| speech rate against information density | Pearson r = -0.71 correlation |
| speech envelope modulation peak | 4-7 Hz |
| cortical bands tied to speech | low gamma 25-35 Hz, theta 4-8 Hz, delta 1-3 Hz |
| delayed auditory feedback that disrupts speech | 100-200 ms |
| feedback response latency | 50-150 ms |
| auditory response after sound at the ear | 10-15 ms |
| transglottal pressure for voicing | 1-2 (98-196 Pa) cm H2O |
| subglottal pressure for phonation | 3-5 above supraglottal cm H2O |
| exhalation share of the breathing cycle in speech | about 60% at rest, about 90% in speech percent |
| oral air pressure peak on /p/ | 11.65 (controls) vs 8.10 (22q11.2DS, velopharyngeal dysfunction) cm H2O (median) |
| pressure wave duration on /p/ | 273.5 (controls) vs 237.0 (22q11.2DS) ms (median) |
| human hearing range | 20 to 20,000 Hz |
| consonants and vowels across the world's languages | about 600 consonants and 200 vowels; about 40 phonemes per language count |
| IPA segmental letters | 107 (since 2005) count |
| languages with ejectives / implosives | 16% / 13% percent of languages |
| languages with uvular consonants | an estimated 19% percent of languages |
| phonetic perceptual narrowing | between 6 and 12 months of age |
| macaque vocal tract configurations traced | 99 count |
| macaque phonetic potential against the 1969 estimate | eightfold ratio |
| the dawn of speech (contrasting vowel qualities) | beyond ~200 ka to over ~20 Ma; last common ancestor with Old World monkeys about 27 Ma years ago |
| first known study of phonetics | as early as the 6th century BCE (Sanskrit grammarians); Panini c. 350 BCE date |
| the term 'phonetics' in its present sense | 1841 year |
| International Phonetic Association founded / letter values made uniform | 1886 / 1888 year |
| Lombard effect discovered | 1909 year |
| McGurk effect first described | 1976 year |
| phonation onset pressure in the vocal-membrane model | 1238 Pa without membranes; down to 488 Pa with membrane length 0.1 cm |
| articulator tracking rates | EMA up to 400 Hz; EPG 100-200 Hz; X-ray 10-50 Hz; rtMRI temporal resolution 12 ms Hz / ms |
| whistled Turkish recognition | words about 70%; common sentences about 80-90% percent |
Open contentions (7) open contention
Where sources or readings disagree. Both stay on the map.
- C-413 Where do thoughts come from?
booked: No theory is proven, so the map lists the main ones side by side and doesn't pick one. It shows where thinking is handled in the brain, not where thoughts start.
the other reading: see research/CONTENTIONS.md C-413 - C-2114 Does culture change how people hear the McGurk effect?
booked: The map books a large study finding no difference, with the earlier smaller studies beside it.
the other reading: similar McGurk frequencies across Chinese and American listeners (as booked in C-2114) - C-2459 Are the cardinal vowels real targets?
booked: The map keeps the competing readings; none is ruled.
the other reading: cardinal vowels as articulatory anchors by which phoneticians judge other articulations (the practical-phonetics tradition) - C-2460 Do speakers plan ahead for the next sound, or overlap them?
booked: The map keeps the competing readings; none is ruled.
the other reading: look-ahead / feature-spreading: variable gestural plans; coproduction: temporally invariant gestures overlapping in implementation - C-2500 Is speech produced step by step or in parallel?
booked: The map keeps the competing readings; none is ruled.
the other reading: serial stages; parallel, interactive processing (the encyclopedia: 'recent evidence shows this is inaccurate' for strictly independent stages) - C-2508 How do we perceive speech sounds?
booked: The map keeps the competing readings; none is ruled.
the other reading: motor theory; abstractionist theories; episodic (exemplar) theories - C-2509 When did the vocal tract become ready for speech?
booked: The map keeps the competing readings; none is ruled.
the other reading: laryngeal descent theory: the monkey tract cannot produce the quantal vowels [i], [u], [a]; the human tongue's 1:1 oral-to-pharyngeal proportions mattered (Lieberman 2017; Lieberman et al. 1969); Fitch et al. 2017 reply: they did not show the monkey tract cannot produce these vowels, only that they did not observe it; quantal vowels 'common but by no means universal'
Cross-scale patterns seen here (21) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies. Each links to every other system that shows it.
- same role, different builds (structural)
One job, putting a sound source into the vocal tract, done by different builds: the vibrating vocal folds (voicing), turbulence at a narrow constriction (frication), pressure released from a closure (stops), a closed glottis moved as a piston (ejectives, implosives), air rarefied between two tongue - same shape, opposite sign (structural)
One process, the closed glottis moved as a piston on the air above it, with its direction reversed: moved up it compresses the air (ejectives, 16% of languages), moved down it rarefies it (implosives, 13%). Booked also as a second form: the same loudness-control loop raises effort when feedback is m - same toolkit, independent builds (structural)
Vocal learning built independently in humans and song-learning birds from shared parts: convergent gene expression in a motor nucleus with a direct line to brainstem vocal neurons (human laryngeal motor cortex, songbird RA) and a striatal learning nucleus; also in cetaceans, pinnipeds, bats, elephan - symmetry breaking (name the cut) (structural)
A symmetric start (infants discriminate the phonetic units of all languages) broken at a named cut (the language heard from live people, between 6 and 12 months); the direction chosen sets what is built after: early native and non-native discrimination predict later language in opposite directions. - templating (structural)
One module reused unchanged: the IPA's letter values, made uniform across languages in 1888, so the same 107 letters transcribe any oral language; the same source-filter model describes any voice and drives synthesis and analysis. - the whole sets the local reference (structural)
Each vowel is read against the talker's whole range: the best normalisation accounts centre or standardise formants by talker, and listeners shift category boundaries to their partner's distinctions, so the same formant values mean different vowels from different talkers. - trade one property for another (structural)
Information per syllable is paid for in speed: across 17 languages the denser the syllable, the slower the speech (r = -0.71), so rates gather near 39 bits/s; whistled speech trades spectral detail for range; at soprano pitches vowel identity is traded for pitch. - Umwelt: each system carves its own slice of one shared medium (structural)
One acoustic space, many slices: each language carves its own categories from the same phonetic continuum (Japanese r/l as one category, English as two), naive listeners sort whistled vowels by their native language, human hearing covers about 20 Hz-20 kHz, chinchillas show speech categorical percep - who opens the door (structural)
The receiver installs the port ahead of time: native phonetic categories, committed by first-year exposure, wait for speech; non-native contrasts that meet no installed category pass unsorted (Japanese infants' r-l discrimination declines between 8 and 10 months while American infants' rises). - barrier-gates-complexity (partial)
The velum seals the nose so pressure can build behind oral closures (stops, fricatives) and oral and nasal sounds can contrast; the without-seal case is shown: with velopharyngeal dysfunction the oral pressure peak on /p/ falls (median 8.10 against 11.65 cm H2O), pressure waves are shorter and nasal - borrowed-then-held (partial)
Regulation of the infant's vocalising first comes from outside: mothers' contingent responses make babbling more mature (8 months), and live people, not television, teach phonetic categories; later the speaker holds the regulation inside, in auditory and somatosensory feedback control against learne - holding / entrainment (partial)
The speech stream sets the clock of the listener's cortical rhythms: syllable-rate edges reset auditory cortex theta, which entrains to the envelope (4-7 Hz) and nests gamma, packaging the input; a driver setting a follower's clock rather than a whole holding its parts. - lightning (partial)
A stored difference released at once: pressure builds behind a full closure and is released as a burst, the cue for stops; each burst is single and capped, so a firing form without a runaway. - namespace completeness (the map predicts its gaps) (partial)
The IPA chart is a map of places by manners: shaded cells are judged impossible (no voiced epiglottal stop: the cavity is too small), blank cells 'can be filled without much difficulty', and in 2005 a blank was filled with a letter for the labiodental flap; slots set by the chart's coordinates, not - outside-to-inside (partial)
The categories come from outside first: infants start with a universal discrimination and, from live speech around them between 6 and 12 months, take in their language's partition of the phonetic space; the DIVA model likewise learns auditory targets from heard speech before producing the sounds. - runaway from an unstable balance (partial)
Phonation starts at a threshold: below the onset pressure the folds stay at rest, above it self-sustained oscillation begins (onset pressure lower than offset in the experiments, regimes switching to irregular, possibly chaotic oscillation); the oscillation is capped as a limit cycle, so a capped fo - seed (partial)
A small notation that unpacks: 107 segmental letters plus diacritics and tone letters write the ~600 consonants and ~200 vowels of the world's languages, at any level of detail; three description parameters (place, manner, voicing) span the consonant chart. - The carry (partial)
The lookahead face: in anticipatory coarticulation a later sound's feature is produced during an earlier one, which look-ahead models place in the plan (contested against coproduction, C-2460); the wait face: speech breaks down when the own voice returns a syllable late (100-200 ms). - The Payment to Entropy (partial)
Every hand-off along the speech chain loses something: only part of the lungs' aerodynamic work becomes sound (vocal membranes raise vocal efficiency and cut the onset pressure in a model from 1238 to 488 Pa), the signal degrades with distance (whistles degrade less than spoken or shouted speech) an - causal ceiling (surface)
No unique way back from sound to movement: acoustic targets map onto many muscle configurations (the inverse problem), so the articulation cannot be read uniquely from the signal; the source adds the concern 'may be exaggerated', so booked as surface. - smooth to sharp in finite time (surface)
A smooth continuum heard as a jump: listeners hear only /ba/ or /pa/ along a voicing continuum; sharp in perception, not a singular value reached in finite time.
Valence: what it trades, and with whom (20) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Phonology: how languages organise the meaningless units their words are built from (phonemes in spoken languages; handshape, location and movement in signed ones), together with the field of linguistics that studies those systems via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Acoustic wave (sound: pressure waves in matter) via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Human (Homo sapiens) via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Language (human language) via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Grammar (the structure of a language, as linguistics describes it) via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Sign language (the natural signed languages) via sys:electromagnetic-field · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Alphabet (one letter per phoneme) via sys:written-record · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Human brain via sys:electrochemical-gradient · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Writing system (a script plus the rules binding it to a language) via sys:written-record · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Music via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies mixed
- Sound via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Respiratory system via sys:atmosphere-of-earth · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Ear via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Perception via sys:sensory-environment · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Fourier analysis (the practice of splitting signals into waves and rebuilding them, and the structure it works on) via sys:sensory-environment · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Naval sonar and ship noise via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies loss
- Songbird (oscines, suborder Passeri) via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies even
- Macaque via sys:electromagnetic-field · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Linguistics (the field: the practice and the languages it studies) via sys:written-record · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
- Speech (human spoken language) via sys:acoustic-wave · Phonetics: the study of speech sounds (how they are made, carried through the air and heard, and how they are recorded and transcribed) together with the speech sounds it studies gain
Partners and parts published science
Built from / builds (10)
- Linguistics (the field: the practice and the languages it studies)
- Speech organs (lungs, larynx, upper vocal tract)
- Larynx
- Spectrogram
- Speech perception
- Phones: individual speech sounds
- Airstream mechanisms (pulmonic, glottalic, velaric)
- Formants: resonances of the vocal tract
- The International Phonetic Alphabet (transcription)
- Competition between the laryngeal descent theory and the speech-ready vocal tract
Sits inside / holds (3)
- Language (human language)
- Linguistics (the field: the practice and the languages it studies)
- Speech (human spoken language)
expressed by (15)
- Generation/Synthesis
- Scaffolding/Enablement (Meta-Tooling)
- Templating / Standardization
- Quality Improvement/Refinement
- Interoperability/Transduction (Bridging)
- Encapsulation & Modularity (Abstraction)
- Internal Coordination/Regulation
- Sensing/Scanning
- Mapping/Modeling
- Uncertainty Reduction/Learning
- Active Experimentation / Probing (Optimal Experimental Design)
- Multi-Scale Data Fusion / Triangulation
- Trail-Marking / Externalized Memory (Stigmergy)
- Social Learning / Public Information Use
- Quality Control & Error Correction
floor filled by (18)
- Phonology: how languages organise the meaningless units their words are built from (phonemes in spoken languages; handshape, location and movement in signed ones), together with the field of linguistics that studies those systems
- Linguistics (the field: the practice and the languages it studies)
- Electromagnetic field
- Atmosphere of Earth
- Acoustic wave (sound: pressure waves in matter)
- Human (Homo sapiens)
- Language (human language)
- Human brain
- Hair cell
- Ear
- Perception
- Spectrogram
- Formants: resonances of the vocal tract
- The International Phonetic Alphabet (transcription)
- Superior temporal gyrus
- Ventral sensorimotor cortex
- Auditory cortex
- Voice onset time
Kinds (6)
- Articulatory phonetics
- Acoustic phonetics
- Auditory phonetics
- Practical phonetics (ear and production training)
- Shiksha (Sanskrit phonetics)
- Phonetics of sign languages (manual-visual and tactile)
Notes published science
- Your skin listens too: an inaudible puff of air on the back of the hand or the neck, timed with a syllable, makes people hear 'ba' as 'pa', with no training at all (src:gick-derrick-2009).
- Fast languages and slow languages come out even: across 17 languages, the more information a language packs into each syllable, the slower its speakers talk, so they all carry about 39 bits per second (src:coupe-2019).
- For almost 50 years the textbooks said monkeys can't talk because of their throats, a claim based on a cast from one dead monkey; x-ray videos of living macaques showed their vocal tracts could make the sounds of speech, so the missing piece is the brain (src:fitch-2016), though Lieberman still argues they can't make 'ee', 'oo' and 'ah' (src:lieberman-2017).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
phonetics: speech sounds from the talker's breath through the air to the listener's categories, and as phoneticians record, describe and transcribe them (reader's view)
- L1: Speech in the air: pressure waves made by the articulators, carried to the listener (and the talking face in light, air puffs on skin)
- L2: Taking it in: eardrum, ossicles and cochlea sorting frequency by place; for the field, microphones and recorders
- L3: Acoustic cues: formants, fundamental frequency and harmonics, voice onset time, bursts, frication noise, durations; the cues listeners weight most
- L4: Phonetic categories: cues sorted into phones and features (manner first, then place, in the superior temporal gyrus), categorical perception, talker normalisation, and fusion with the face and skin
- L5: Speech sensorimotor control: a speech sound map driving feedforward commands, with auditory and somatosensory targets and error maps correcting them (the talker's own perception joined to production)
- L6: Phonetic description: places and manners, phonation types, vowel height, backness and rounding, prosody, airstream mechanisms; the world's ~600 consonants and ~200 vowels compared; the IPA; theories of production and perception
- L7: Use back on the world: speech between people at about 39 bits/s, and applied phonetics (language teaching, acting, speech pathology, speech recognition and synthesis)
the boardroom (shared floors)
- L3: Acoustic cues: formants, F0, VOT, bursts and noise; the cues listeners weight; cortical rhythms packaging syllables (theta) and phonemic detail (gamma)
- L4: Phonetic categories: features in the superior temporal gyrus, categorical perception, talker normalisation, audiovisual and aero-tactile fusion through causal inference
- L5: Speech sensorimotor control: speech sound map, feedforward articulator commands, auditory and somatosensory target and error maps, feedback corrections; ventral sensorimotor cortex organised by articulator
- L6: Phonetic description and theory: articulatory, acoustic and auditory descriptions, cross-language comparison, transcription with the IPA, models (source-filter, gestures, DIVA) and contested theories of perception
- L7: Use: talk between people at a shared information rate; applied phonetics in teaching, acting, speech pathology, speech recognition and synthesis
way in: Feeling one's own articulators (somatosensory feedback)
- L1: Contact and stretch as the articulators move; voicing can be felt at the larynx
- L2: Tactile and proprioceptive receptors; local stretch-reflex feedback to brainstem motor neurons
- L3: Afferents from face and vocal tract to ventral sensorimotor cortex
- L4: Somatosensory target and state maps in ventral supramarginal gyrus: a 22-dimensional expected feel for each sound
way in: Hearing one's own voice (auditory feedback)
- L1: The talker's own speech; delayed by about a syllable it disrupts speaking
- L2: The same ear; auditory responses within 10-15 ms of the sound
- L3: Auditory cortex: the response to one's own speech is suppressed against playback (SIS)
- L4: Auditory target and error maps in posterior superior temporal cortex: the sound compared with the formant bounds expected
way in: Hearing: speech sound in air
- L1: Pressure waves from a talker; the envelope rises and falls at the syllable rate (4-7 Hz)
- L2: Eardrum, ossicles, cochlea: the basilar membrane resonates by place; hair cells release transmitter onto the auditory nerve
way in: Imaging the articulators (x-ray, ultrasound, endoscopy, MRI, cortical electrodes)
- L1: The vocal tract moving during vocalisation, feeding and display
- L2: Instruments: x-ray video, ultrasound, endoscopy, electromagnetic articulography, real-time MRI
- L3: Tracings of observed configurations
- L4: Measures: the tract's phonetic space in F1 by F2
- L5: Models: tubes and resonance equations predict the acoustics of a posture
way in: Recording and acoustic analysis (the phonetician's instruments)
- L1: Pressure fluctuations at a microphone
- L2: Transduction and display: spectrogram by band-pass filter bank or Fourier transform
- L3: Analysis: voicing bar, formant tracks, inverse filtering to recover the glottal source
- L4: Measurement: VOT, formant frequencies, speech rate, across speakers
- L5: Corpora compared across languages: 17 languages' speech rates against information per syllable
way in: Sight (and touch): signs from hands, body and face
- L1: Hands, arms, upper body and head articulating signs; in tactile signing, felt by the hands
- L2: The eye: gaze on the face, the hands in peripheral vision
- L3: Sign parameters: location, movement, handshape, palm orientation, non-manual features
way in: Sight: the talking face (lips, jaw)
- L1: Light from the talker's mouth movements
- L2: The eye and visual system read the mouth; infants already follow adult faces
- L3: Visual cues to place: a seen 'ga' against a heard 'ba'
way in: The message to be spoken (inner channel)
- L1: A nonlinguistic message to be encoded (where it comes from is contested)
- L2: Lexical selection and phonological encoding: words assigned their phonemes (serial or parallel, debated)
- L3: Articulatory specification: phonemes specified for articulatory goals (closed lips, tongue position)
- L4: The speech sound map in left premotor and inferior frontal cortex: one cell per frequent speech sound
way in: Touch: air puffs from aspirated stops on the skin
- L1: A slight, inaudible puff of air, as from an aspirated 'p'
- L2: Cutaneous reception at hand or neck, without training
way out: Speaking: breath, voice and articulation into the air
- L1: Sound radiated into the air, heard by others and by the talker
- L2: The filter: the supraglottal tract's shape sets the resonances; closures, constrictions and the velum add bursts, noise and nasality
- L3: The source: lungs hold subglottal pressure; the vocal folds vibrate (or not); the glottis can act as a piston
- L4: Articulatory gestures: lips, jaw, tongue (a muscular hydrostat), velum, larynx coordinated
- L5: Motor commands from ventral sensorimotor cortex, timed as sequences
- L7: The intention to say a message
way out: Synthesising speech (from Pattern Playback to source-filter synthesis)
- L1: Sound made from pictures of spectrograms
- L6: A model of the signal: source and filter, or a painted acoustic pattern
way out: The visible articulation (the talking face)
- L1: Mouth movements seen by the listener
- L5: The same articulator commands
way out: Transcribing: writing speech in the IPA
- L1: Written transcription read by linguists, teachers, singers, actors and speech-language pathologists
- L3: One letter for each distinctive sound, with uniform values across languages
- L6: Choose the level of detail: one letter, or a letter with diacritics