Phonetics — the physical science of speech sounds

Intuition Beginner

Phonetics is the branch of linguistics that studies speech sounds as physical events: how the vocal tract makes them, how they travel through air, and how the ear decodes them. It asks not what sounds mean — that is phonology and semantics — but what they are as measurable signals.

Every spoken sound is the product of a moving airstream shaped by the tongue, lips, teeth, and jaw. The positions and motions of these articulators determine which frequencies the vocal tract amplifies, and those amplified frequencies are what a listener's ear detects.

Phonetics divides naturally into articulatory (production), acoustic (transmission), and auditory (perception) branches. This unit concerns the first two; auditory phonetics connects to psychology in 29.03.01.

Visual Beginner

A schematic of the human vocal tract showing the airstream path from lungs through the larynx and oral/nasal cavities.

The articulators — tongue body, tongue tip, lips, velum — move to shape the airstream into the resonant frequencies that distinguish one speech sound from another.

Worked example Beginner

Consider the vowel in "heed" versus "hod". The two vowels sound different because the tongue body sits high and forward for the first, low and back for the second.

Step 1. For "heed", raise the tongue body toward the hard palate; the oral cavity is small in front, large behind.

Step 2. For "hod", lower the tongue body and retract it; the oral cavity is large throughout.

Step 3. The different cavity shapes amplify different frequencies (formants): "heed" has a high first formant and a very high second formant; "hod" has a high first formant and a low second formant.

What this tells us: vowel quality is set by the geometry of the vocal tract, and that geometry is measurable.

Check your understanding Beginner

Formal definition Intermediate+

Phonetics is the empirical study of speech sounds as physical events ^{[Ladefoged & Johnson Ch. 1]}. Its three branches:

• Articulatory phonetics: the study of how the vocal tract produces sounds. The vocal tract is modelled as a tube whose shape is set by articulators (lips, tongue blade, tongue body, velum, jaw). Place of articulation (bilabial, alveolar, velar, etc.) names the constriction location; manner of articulation (stop, fricative, affricate, approximant, nasal) names the constriction type.
• Acoustic phonetics: the study of the sound wave produced. The vocal folds (when voiced) generate a periodic source spectrum; the vocal tract filters this spectrum, amplifying resonant frequencies called formants (F1, F2, F3).
• Auditory phonetics: the study of how the ear and brain decode the acoustic signal.

A phone is any speech sound, transcribed in IPA brackets [a]. A phoneme is a psychologically contrastive sound category, transcribed between slashes /a/; phonology (not phonetics) studies the phoneme systems of languages.

Counterexamples to common slips

• Phones are not phonemes. [tʰ] (aspirated) and [t] (unaspirated) are distinct phones; in English they are one phoneme /t/ (allophones), in Hindi distinct phonemes.
• Voicing is not the only laryngeal contrast. Some languages contrast breathy voice, creaky voice, and ejective phonation.
• Formants are not overtones of a fixed source. The formants are resonances of the vocal tract, which filters whatever the laryngeal source produces.

Key concepts Intermediate+

Source-filter theory (Fant 1960). The speech sound wave radiated from the lips is the product of (i) the laryngeal source spectrum and (ii) the transfer function of the vocal tract, treated as a resonant acoustic tube ^{[Stevens Ch. 1]}.

Explanation. The vocal folds (when vibrating) produce a source with a harmonic spectrum whose energy declines at roughly 12 dB per octave. The vocal tract, shaped by articulator positions, behaves as a tube with several resonant frequencies — the formants. Multiplication in the frequency domain means the radiated spectrum shows peaks at the formant frequencies, and the articulator positions determine those peaks. This separation of source and filter explains why the same vowel can be whispered (no laryngeal source) or sung at any pitch: the pitch tracks the source, the vowel quality tracks the filter.

Bridge. This result builds toward 51.02.01 (phonology), where the continuous acoustic signal is categorised into discrete phonemes, and appears again in 10.04.02 (electromagnetic waves and the wave equation), whose signal-analysis apparatus underlies the spectral decomposition of speech. The foundational reason source-filter theory works is that vocal-tract resonance is a linear wave phenomenon, and putting these together, this is exactly the bridge that unifies the biological description of speech with the physical description of sound, and the pattern generalises to the resonance-and-filter analyses used throughout acoustics and signal processing.

Exercises Intermediate+

Lean formalization Intermediate+

lean_status: none. Phonetics is an empirical science whose quantitative content draws on acoustic signal analysis; its correctness gate is articulatory and acoustic evidence, not formal proof.

Advanced results Master

Quantal theory (Stevens 1972). The relation between articulator position and acoustic output is nonlinear, with stable plateau regions: small articulatory changes within a plateau produce little acoustic change, but crossing a boundary produces a large jump. This explains why languages re-discover the same vowel and consonant inventories — the quantal regions are natural acoustic anchors ^{[Stevens Acoustic Phonetics]}.

Distinctive features (Jakobson, Fant, Halle 1952). Speech sounds are decomposed into binary features (+/- voice, +/- nasal, +/- continuant, etc.) grounded in acoustic and articulatory correlates. Features are the atoms of phonological theory and underlie the phonological rules developed in 51.02.01.

Coarticulation. Articulators move continuously and overlap, so each segment bears acoustic traces of its neighbours. Models of coarticulation (look-ahead, target-and-overshoot) are central to modern speech production theory and to speech-synthesis engineering.

Synthesis. Phonetics grounds the entire linguistic signal in physical measurement: the source-filter theory established here builds toward 51.02.01 where continuous acoustic variation is categorised into discrete phonemes, appears again in 10.04.02 whose wave-equation apparatus underlies speech-signal analysis, the foundational reason phonetics must precede phonology is that phoneme systems are abstractions over measurable acoustic events, the bridge is that the same resonance-and-filter pattern underlies both vowel production and the engineering of speech technology, and the pattern generalises to the feature-based descriptions used in every branch of structural linguistics; putting these together, phonetics is the empirical foundation of linguistics, the chapter that anchors language study in physical measurement.

Full proof set Master

Proposition (Formant scaling with vocal-tract length). For a vocal tract modelled as a uniform tube of length $L$ open at one end and closed at the other, the resonant frequencies are $F_n=(2n-1)c/(4L)$ for odd $n$, where $c\approx 343$ m/s is the speed of sound.

Derivation. A tube closed at one end and open at the other has standing-wave resonances with a pressure node at the open end and an antinode at the closed end. The longest such standing wave has quarter-wavelength equal to $L$, so ${\lambda }_1=4L$ and $F_1=c/(4L)$. Higher resonances occur at odd multiples of the quarter-wavelength, giving $F_n=(2n-1)c/(4L)$ for $n=1,2,3$. Doubling $L$ halves all formants; halving $L$ doubles them. This predicts an adult-male /ɑ/ vowel with $L\approx 17$ cm to have $F_1\approx 500$ Hz, $F_2\approx 1500$ Hz, $F_3\approx 2500$ Hz, in good first-order agreement with measurement.

Connections Master

• Phonology 51.02.01. The continuous phonetic signal is categorised by each language into a discrete inventory of phonemes; phonological rules refer to the features defined here.

• Signal analysis and waves 10.04.02. The source-filter theory of speech uses the same linear wave analysis that describes electromagnetic radiation, connecting phonetics to the broader physics of waves.

• Auditory and speech perception 29.03.01. How the ear and brain decode the speech signal is the subject of auditory phonetics, whose results constrain phonological theory.

• Speech technology. Speech synthesis, recognition, and speaker identification engineering all build on the source-filter-plus-features model developed here.

Historical & philosophical context Master

Modern phonetics was founded in the late nineteenth century by Henry Sweet, Eduard Sievers, and the IPA (established 1886), which standardised a one-symbol-per-sound transcription system. The International Phonetic Association's Principles (1888) fixed the doctrine that phonetics is the empirical foundation of linguistics ^{[IPA Handbook]}.

The source-filter theory was articulated and experimentally verified by Gunnar Fant in Acoustic Theory of Speech Production (1960), building on the earlier work of Hermann Helmholtz (On the Sensations of Tone, 1863) on vocal resonances. Kenneth Stevens's quantal theory (1972) explained the typological stability of speech-sound inventories, and the distinctive-feature framework of Jakobson, Fant, and Halle (Preliminaries to Speech Analysis, 1952) unified articulatory and acoustic description into a single feature set.

The discipline remains empirically rich: instrumental phonetics uses electromagnetic articulography, ultrasound, and MRI to measure vocal-tract shapes in real time, and computational models now synthesise and recognise speech at near-human accuracy.

Bibliography Master

@book{LadefogedJohnson2014,
  author = {Ladefoged, Peter and Johnson, Keith},
  title = {A Course in Phonetics},
  edition = {7},
  publisher = {Cengage},
  year = {2014},
}

@book{Stevens1998,
  author = {Stevens, Kenneth N.},
  title = {Acoustic Phonetics},
  publisher = {MIT Press},
  year = {1998},
}

@book{Fant1960,
  author = {Fant, Gunnar},
  title = {Acoustic Theory of Speech Production},
  publisher = {Mouton},
  year = {1960},
}