The Throat

Six storeys, deepest first.

Al-Farahidi did not begin his dictionary at alif. He began at ع, ʿayn, because he had worked out that it is produced deepest in the throat, and he ordered the whole language by how far down each sound is made. The book is named for that letter. Select a numbered place on the plate, or a storey of the tower, to see which structures meet there, what the air and the vocal folds are doing, how long the space in front of the constriction is — and why that is what makes the sound what it is.

The lips

The deepest. Al-Farahidi began one storey up, at ع ʿayn

The airway from the vocal folds to the lips, in six storeys, numbered deepest first — the order of al-Farahidi’s dictionary. The tower is the same airway, straightened.

I Storey I of VI · the deepest

Glottal ه hāʾ

The vocal folds themselves. The breath before any shape is put on it.

Where: The larynx, at the base of the throat: the vocal folds themselves, behind the Adam's apple.
What meets: The two folds narrow toward each other. For h they stay slightly apart and the air rushes through the gap; for the glottal stop ʔ they close completely and the air is held, then released.
The air: Comes straight up from the lungs and is shaped at the source. Nothing above the folds has moved yet.
The folds: Here the folds ARE the articulator. Held open and stiff they make the breath of h; snapped shut they make the catch in the middle of uh-oh.
The space in front: The whole tube — pharynx, mouth, lips — sits in front of this constriction, so the sound takes the colour of whatever vowel the mouth is already shaped for. h before ee and h before oo are two different hisses.
Why it sounds so: Low, breathy, unfocused: there is no narrow place above it to sharpen the noise. It is the sound of the airway with nothing added.

Al-Farahidi’s letters here

هhāʾh
ءhamzaʾ
اalifā

Where this follows al-Khalīl, and where it departs

Al-Khalīl had no glottal storey: hāʾ was his third throat letter, hamza and alif were airy letters. The floor is Sībawayh's aqṣā al-ḥalq (hamza, hāʾ, alif), and modern phonetics confirms it, so the base of the tower takes those three. ʔ is voiceless by definition — the folds are pressed shut, not vibrating; ɦ, listed for the Indic languages, is the only voiced sound here. Aspiration (the puff on Urdu kh th ph bh) and the ejective catch of Cusco Quechua (pʼ tʼ chʼ kʼ qʼ) are both glottal gestures, but each is listed, or folded, on the storey of its plain twin, not here. No vowels: alif appears only as a letter. Al-Farahidi did not begin here; Kitāb al-ʿAyn opens one storey up, at ʿayn.

The sounds

hh as in hawāʾ (air)ʔʾ, the catch in uh-ohɦthe murmured h of Urdu hāth voiced voiceless

hawāʾArabicair — the word al-Khalīl's airy letters, the hawāʾiyya, are named for
hālaUrduhalo, ring
hallowEnglishto make holy — Old English hālgian

One storey up, at the pharynx, the tongue root joins in and the breath becomes a squeezed, deeper-coloured sound — ʿayn and ḥāʾ.

II Storey II of VI

Pharyngeal ع ʿayn

The throat narrowed above the larynx. Arabic and Hebrew make consonants here; almost nothing in Europe does.

Where: The pharynx: the vertical passage behind the mouth, above the larynx and below the uvula.
What meets: The root of the tongue is pulled backward toward the back wall of the pharynx, narrowing the passage to a slot. The epiglottis folds into it.
The air: Forced through a tall narrow slot at the back of the throat, well below anything the lips or teeth can touch.
The folds: Vibrating for ʿ (ʿayn), open for ḥ (ḥāʾ). The squeeze sits just above them, so the voice comes out compressed.
The space in front: A long front space — the whole mouth — with the back space almost closed off. That drops the second resonance sharply, which is the 'dark' quality Arabic speakers hear as emphasis.
Why it sounds so: Compressed, dark, effortful. English has no sound here, which is why English speakers cannot hear ʿayn and why Iraq and Oman arrive without their first consonant.

Al-Farahidi’s letters here

عʿaynʿ
حḥāʾḥ

Where this follows al-Khalīl, and where it departs

This is where al-Khalīl began: ʿayn, then ḥāʾ. His throat group also held ه (filed on I, at the folds) and خ غ (filed on III, by the back of the tongue); those departures are stated on their storeys. Only two Arabic letters and two sounds are made here, and no everyday English, Urdu, Persian, Spanish or Japanese consonant: Urdu and Persian keep the letters ع and ح but say them as a vowel and as h; modern Israeli Hebrew has lifted ע and ח to ʔ and χ, while Mizrahi and Yemenite reading keep them here. The four emphatics of storey V (ṭ ḍ ṣ ẓ) borrow this squeeze as a second articulation, which is their 'darkness'.

The sounds

ʕʿ as in ʿaynħḥ as in Muḥammad voiced voiceless

عينArabicʿayn — eye, spring, and the letter itself
علمArabicʿilm — knowledge; the root under ʿālim, ʿulamāʾ, and Urdu ilm
ʿibrīHebrewHebrew — from ʿ-b-r, to cross over

Down a storey the tongue lets go and only the folds remain: h. Up a storey the back of the tongue lifts to the uvula and the sound turns gravelly: the French r, Arabic q.

III Storey III of VI

Uvular ق qāf

The back of the tongue against the uvula. The q in Qurʾān and Iraq, the kh in khān.

Where: The soft palate and the uvula, the fleshy tag hanging at the back of the roof of the mouth.
What meets: The back of the tongue rises to the uvula. A full stop for q; a rough, fluttering gap for χ and ʁ, where the uvula itself trembles in the airstream.
The air: Pushed through the gap between tongue-back and uvula, often setting the uvula vibrating — the growl of the French and German r.
The folds: Open for q and χ, vibrating for ʁ and ɢ. The Parisian r is voiced; the Scottish loch sound is not.
The space in front: The mouth in front is still long, so the noise is coloured low — but a little brighter than the pharynx, because the constriction has moved forward and the front space has shortened.
Why it sounds so: Gravelly, throaty, textured: a rough source deep in the mouth, heard through a long chamber.

Al-Farahidi’s letters here

قqāfq
كkāfk
خkhāʾkh
غghayngh

Where this follows al-Khalīl, and where it departs

Two of al-Khalīl's groups meet here: the lahawiyya ق ك and the last two throat letters خ غ, which he heard as throat sounds because their rasp is deep; Sībawayh moved them to the mouth end of the throat and phonetics puts them at the uvula, so the tower files them on III against al-Khalīl. The tower has no velar floor, so k, g, ŋ and the velar kh gh (x, ɣ) of Urdu, Persian, Spanish and Greek are filed here as 'the back of the tongue', a finger's width in front of q; Arabic speakers themselves vary between the velar and uvular value of خ and غ, so each fricative is listed at both depths, and Tehran Persian غ ق (ɢ at the start of a word, ʁ elsewhere) sits at the uvular one. The Urdu and Hindi aspirates kʰ gʱ are listed as sounds of their own, because a listener meets them in Latin spelling as kh and gh — the same digraphs as خ and غ — and must be able to tell khānā (kʰ) from khān (x), ghar (gʱ) from ghazal (ɣ). ɢ, which the storey prose already names, is listed for Persian. Two series fold under their plain twins rather than take chips of their own: the Cusco Quechua ejectives kʼ qʼ (with pʼ tʼ chʼ on their storeys) and the aspirate qh, which the show's Quechua words (Qusqu, qucha) do not carry; and the rounded velars of Akan (kw gw hw nw in kwan, hwɛ, nwoma), which are these sounds with the lips rounded, named in the hints. w also raises the tongue-back here but is filed at the lips, its visible half.

The sounds

qq as in Qurʾānkk as in kitābgg as in goŋng as in singxkh as in khānχkh as in khubzɣgh as in Urdu ghazalʁgh as in Baghdādɢq as in Persian Qomkʰkh as in Urdu khānāgʱgh as in Urdu ghar voiced voiceless

قلمArabicqalam — pen; cognate with Greek kálamos, reed, and English calamus
قصرArabicqaṣr — castle; borrowed from Latin castrum, and returned to Spanish as alcázar
khānaUrduhouse, room — as in the -khana of dawakhana, the pharmacy

Up a storey the constriction jumps forward to the hard palate and the sound brightens dramatically: the hush of sh, the hiss of ch in German ich.

IV Storey IV of VI

Palatal ش shīn

The tongue raised to the hard palate. The sh that the Shaq episode is built on.

Where: The hard palate: the bony dome of the roof of the mouth, behind the ridge and in front of the soft palate.
What meets: The body of the tongue arches up toward the dome, leaving a wide flat channel. For ʃ the tongue tip also lifts and the lips round slightly, lengthening the front chamber.
The air: Spread thin over a broad arch and released as a hush rather than a whistle.
The folds: Open for ʃ and ç; vibrating for ʒ and j — the sound at the start of yes.
The space in front: A short front space now, so the energy moves high — but the wider channel and the rounded lips keep it from the sharp whistle of s. That is why sh is 'hush' and s is 'hiss'.
Why it sounds so: Bright, soft-edged, hushing. The place of Shaq, Shakira, shukr and śakti — three families, one sound.

Al-Farahidi’s letters here

شshīnsh
جjīmj
يyāʾy

Where this follows al-Khalīl, and where it departs

Al-Khalīl's shajriyya were ج ش ض: ج and ش stay; ي joins them from his airy letters, where Sībawayh already put it; ض leaves for storey V (reason given there). Placement decision: the postalveolars ʃ ʒ tʃ dʒ are filed here rather than at the ridge because the tongue body is domed to the palate — the tie-breaker the site has used since it built the Shaq episode on this storey — and the true palatals j ç ɲ ʎ are here on any reading. The alveolo-palatals of Japanese and Akan are listed as sounds of their own, ɕ with its affricates tɕ dʑ under one rule, since a listener meets them in two of the show's languages and the ear can tell them apart; Sanskrit ś is ɕ, while the Sanskrit and Hindi ca-varga (c ch j jh ñ) is written tʃ dʒ here by convention, though it lies close to tɕ dʑ. Spanish y (ʝ) folds under j. The Urdu and Hindi aspirates chh and jh are listed for the reason stated on storey III.

The sounds

ʃsh as in shukrʒzh, the s in measuretʃch as in churchdʒj as in jamīljy as in yesçch as in German ichɲñ as in niñoʎll as in Quechua llamaɕsh as in Japanese sushitɕch as in Japanese chidʑj as in Japanese Fujitʃʰchh as in Urdu chhoṭādʒʱjh as in Urdu jhīl voiced voiceless

شكرArabicshukr — thanks; Shakur, the grateful one
शक्तिSanskritśakti — power, capacity
shakeEnglishProto-Germanic *skakaną — a different root from either

Up a storey the tongue tip goes to the ridge behind the teeth, the channel narrows to a groove and the front space is shortest of all: s, t, d, n — the crispest sounds in any language.

V Storey V of VI

Alveolar ت tāʾ

The tongue tip at the ridge behind the teeth. The busiest storey in every language on earth.

Where: The alveolar ridge: the small shelf of gum just behind the upper front teeth. Touch it with your tongue tip.
What meets: The tongue tip or blade meets the ridge. A full seal for t, d and n; a narrow grooved gap for s and z; a light tap or trill for r in many languages.
The air: For s, a fine jet of air is aimed through the groove at the edge of the teeth: the teeth are what turn it into a whistle.
The folds: Open for t and s; vibrating for d, n, z, l, r. n also opens the nasal port behind the soft palate.
The space in front: The shortest front space on the tower — a few millimetres between ridge and lips — so the noise is the highest and sharpest the mouth makes.
Why it sounds so: Crisp, precise, forward. The most common place of articulation in the world's languages, and the home of the first consonant most infants say after the lips.

Al-Farahidi’s letters here

تtāʾt
طṭāʾṭ
دdāld
ظẓāʾẓ
ذdhāldh
ثthāʾth
صṣādṣ
سsīns
زzāyz
ضḍādḍ
رrāʾr
لlāml
نnūnn

Where this follows al-Khalīl, and where it departs

Four of al-Khalīl's groups collapse into this storey because a six-floor tower cannot tell the ridge from the teeth: the niṭʿiyya ط د ت (the front of the palate at the roots of the teeth), the lithawiyya ظ ذ ث (the gums — interdental), the asaliyya ص س ز (the tongue tip) and the dhalqiyya ر ل ن (the tip's edge), plus ض moved down from his shajriyya because the ḍād heard today is dˤ, dāl's emphatic twin — thirteen letters, the headline tāʾ first and then al-Khalīl's groups in his order. Placement decision: dental, alveolar, interdental and retroflex all file here as the tongue tip within a centimetre — t d n are dental in Arabic, Spanish and Urdu and on the ridge in English; θ ð are the tip at the teeth; ʈ ɖ ɽ ɳ ʂ are the tip curled back behind the ridge (Urdu, Hindi and Sanskrit ṭ ḍ ṛ ṇ ṣ) and have no Arabic letter. The four emphatics tˤ dˤ sˤ ðˤ have their primary contact here and a secondary squeeze at storey II. Two ṭ's, and two ṣ's, must never be confused: Urdu ṭopī has retroflex ʈ, Arabic ṭālib has emphatic tˤ; Sanskrit Kṛṣṇa has retroflex ʂ, Arabic ṣabr has emphatic sˤ; likewise Urdu th dh (tʰ dʱ, breath) against Arabic th dh (θ ð, the teeth). Three r's: the trill r, the tap ɾ and the English ɹ, which touches nothing; dark ɫ is l with the tongue-back raised toward III. The aspirates tʰ dʱ ʈʰ ɖʱ ɽʱ are listed as sounds of their own (reason on III); the breathy sonorants nh mh lh of Hindi nanhā, tumhārā, kulhāṛī are treated as clusters with ɦ and not listed. ts is written without a tie bar, matching tʃ dʒ. The Nahuatl tl of tomatl and chocolatl is not listed: the words came into every language on the show, the sound did not. The latin label is capped at eight items, so the emphatics and retroflexes appear in the panel, grouped, not in the row.

The sounds

plaintt as in kitābdd as in dīnnn as in nūrss as in salāmzz as in zaytll as in laylrthe trilled r of perroɾthe tap of Spanish peroɹthe English r of redɫthe dark l of Allāhtsts as in tsunamiat the teethθth as in thinðdh as in dhikrcurled backʈṭ as in Urdu ṭopīɖḍ as in Urdu ḍabbāɽṛ as in Urdu baṛāɳṇ as in Sanskrit prāṇaʂṣ as in Sanskrit Kṛṣṇaemphatictˤṭ as in ṭālibdˤḍ as in Ramaḍānsˤṣ as in ṣabrðˤẓ as in ẓuhrwith breathtʰth as in Urdu thālīdʱdh as in Urdu dhobīʈʰṭh as in Urdu ṭhīkɖʱḍh as in Urdu ḍholɽʱṛh as in Urdu paṛhnā voiced voiceless

دنUrdudin — day
dinAkanname
diēsLatinday — the same root as the Urdu, and not the Akan

Up a storey and the tongue stops mattering: the lips take over, the front space disappears, and the sound goes from a whistle to a pop.

VI Storey VI of VI · the lips

Bilabial ب bāʾ

The two lips. The first sounds an infant makes, in every language, before any language.

Where: The lips. The outermost storey; the only one you can see.
What meets: Upper lip against lower lip. A complete seal for p, b and m; a narrow gap between the lips for the ɸ and β of Spanish and Japanese.
The air: Stopped behind the closed lips and released in a small burst (p, b), or sent through the nose while the lips stay shut (m).
The folds: Open for p, vibrating for b and m. Put a hand on your throat and say pa, then ba: the difference is entirely here, not at the lips.
The space in front: There is no front space at all — the constriction is at the exit. The burst is broad and soft, coloured by the whole tube behind it.
Why it sounds so: Soft, blunt, plosive: a release rather than a hiss. Babies find it first, which is why so many words for mother begin with m.

Al-Farahidi’s letters here

بbāʾb
مmīmm
فfāʾf
وwāww

Where this follows al-Khalīl, and where it departs

Al-Khalīl's shafawiyya ف ب م, with و brought in from his airy letters where Sībawayh placed it, between the lips. Placement decision: labiodental f v ʋ are the lower lip at the upper teeth and file with the lips as their visible half, not at V; w, whose tongue-back rises at storey III, and the Yoruba and Igbo kp gb, which shut the lips and the tongue-back at once, file here because the lips are what is seen. p and v do not exist in Arabic, which borrows them as b and f, but a listener meets them in Urdu, Persian, Hebrew and every European language on the show; Persian and Urdu added پ for p. Spanish and Japanese narrow the lips without sealing them (β, ɸ); Hindi and Urdu hold و between v and w (ʋ), which is why wālā and vālā are one word and one chip. The lip-rounding of Akan kw gw hw nw tw dw (kwan, hwɛ, nwoma, twa, dwene) lives here, but each of those is a storey I, III or IV sound with the lips rounded, and is named in that sound's hint. pʰ and bʱ are listed as sounds of their own (reason on III); the Cusco ejective pʼ folds under p. The Sanskrit and Hindi pa-varga sits here.

The sounds

pp as in papabb as in baytmm as in māmāff as in fikrvv as in veryww as in waladβthe soft b of Spanish haberʋthe v/w of Hindi Vedaɸthe f of Japanese Fujipʰph as in Urdu phalbʱbh as in Urdu bhāīk͡pp as in Yoruba pupag͡bgb as in Yoruba gbogbo voiced voiceless

papanurseryfather — everywhere, independently
ماماArabicmāmā — mother
*ph₂tḗrPIEfather — the reconstructed form, one asterisk

This is the top of the tower. Every place below it moves the constriction back into the mouth and throat, lengthening the front space and lowering the sound — until, at the folds, there is only breath.

Why this is here

Phonetics is anatomy.

Speech is breath in motion. Every word anyone has ever spoken was air, pushed from a hollow interior and shaped into sound by a column of soft tissue that a physician spends years learning by name. The places where that column narrows — the folds, the pharynx, the uvula, the palate, the ridge behind the teeth, the lips — are the places where languages differ, and the places where they change.

Al-Farahidi understood this in the eighth century well enough to build the first dictionary of Arabic on it. He did not begin at alif. He began at ع, ʿayn, because he had worked out that it is produced deepest in the throat, and he ordered the entire language by how far down each sound is made. The book is named after that letter: Kitāb al-ʿAyn.

The tower is the same six places, stacked. It is the show’s mark for a reason. A tower of sounds, deepest at the base, is what the story of Babel is about — and it lets the site carry the thesis of one origin without drawing a family tree, which is a diagram the show has decided to be careful with.

Two of the storeys matter most to the argument. The palatal storey is where Shaq, Shakira, shukr and śakti all live: three families, one sound, one root — the breath meeting the wall, which is the third episode. The bilabial storey is where every infant on earth begins, which is why papa and mama recur in every language there is, and why the oldest word on The Descent is the one a child says first. The mouth is the same everywhere. That is where the case for one tongue begins.

Six storeys, deepest first.

Phonetics is anatomy.

Every root, in your inbox.