# Arabic's Closest Living Relative Is Hebrew

My Weird Prompts — an AI-generated podcast. Daniel Rosehill sends in a prompt; AI hosts Corn and Herman discuss it. Verify important claims independently.

- Episode: 6666
- Published: 2026-11-18
- Duration: 28:05
- Audio: https://episodes.myweirdprompts.com/audio/semitic-language-family-tree-28569.m4a
- Tags: ancient-world, language-learning, middle-east-history, writing-systems, translation
- Web page: https://myweirdprompts.com/episode/semitic-language-family-tree-28569/
- JSON: https://myweirdprompts.com/api/episodes/semitic-language-family-tree-28569.json

## Summary

Comparative linguistics upends geography in the Semitic language family. Using shared innovations rather than shared retentions, the Hetzron model dissolves the old South Semitic branch and places Arabic in Central Semitic alongside Hebrew, Aramaic, and Ancient South Arabian. Despite sharing three-consonant roots like K-T-B and the genitive chain construction, Hebrew and Arabic speakers still cannot understand each other. The discussion covers the four-thousand-year written record from Akkadian cuneiform to modern debates over North Arabian inscriptions.

## The prompt

> Zooming out from the geopolitics and looking purely as linguists: how is the Semitic language family put together? Hebrew, Arabic, Aramaic and their relatives. Which of them are most closely related to each other, how much core vocabulary and grammar do they share through common roots, and how mutually intelligible are they in practice?

## Show notes

Semitic is one branch of Afroasiatic, with over 460 million speakers—411 million of them native Arabic speakers, against roughly 5 million Hebrew speakers. The episode walks through how the family tree is actually built, and why the older three-way East/Northwest/South split collapsed. The diagnostic rule is the shared innovation: only features inherited from a recent common ancestor count, not retentions everyone kept. Under the Hetzron model, Arabic moves out of "South Semitic" into Central Semitic, making Hebrew its closest living relative and Aramaic and Phoenician its closest dead ones—the tree inverts the map. The episode also covers the shared machinery (triconsonantal roots, the genitive chain) and why a Hebrew speaker and an Arabic speaker share a branch but not conversations.

## Sources

- [John Huehnergard, Arabic in Its Semitic Context, chapter 1 in *Arabic in Context* (Brill, 2017), doi 10.1163/9789004343047_002](https://almuslih.org/wp-content/uploads/2024/12/Huehnergard-J-Arabic.pdf) — retrieved 2026-11-18 (pp. 1, 3 read).
- [Wikipedia, Semitic languages](https://en.wikipedia.org/wiki/Semitic_languages) — retrieved 2026-11-18; last edited 2026-10-25.
- [Wikipedia, Afroasiatic languages](https://en.wikipedia.org/wiki/Afroasiatic_languages) — retrieved 2026-11-18; last edited 2026-11-16.
- [Wikipedia, Ethio-Semitic languages](https://en.wikipedia.org/wiki/Ethio-Semitic_languages) — retrieved 2026-11-18; last edited 2026-11-09.
- [Wikipedia, Aramaic](https://en.wikipedia.org/wiki/Aramaic) — retrieved 2026-11-18.
- [The Classification of the Languages of North Arabia: Remarks on the Semitic Language Family Tree of the 2nd edition of Routledge's *The Semitic Languages*, *International Journal of Arabic Linguistics*](https://revues.imist.ma/index.php/IJAL/en/article/download/18967/10334/48727) — retrieved 2026-11-18.
- [Amani Aloufi, A grammatical sketch of Soqotri, dissertation, 2016](https://openalex.org/W2513242099) — retrieved 2026-11-18 (abstract).
- [Holger Gzella, Aramaic, the English of the Levant in Antiquity, *The Bible and Interpretation*, January 2015](https://bibleinterp.arizona.edu/articles/2015/01/gze398010) — retrieved 2026-11-18; published 2015-01-02.
- [Holger Gzella, The Semitic Languages and Dialects III: North-West Semitic, University of Tübingen repository](https://tst-publikationen.uni-tuebingen.de/xmlui/bitstream/handle/10900/168832/Gzella_051.pdf) — retrieved 2026-11-18.
- [Edward Ullendorff, The Knowledge of Languages in the Old Testament, *Bulletin of the John Rylands Library* 44/2](https://www.manchesterhive.com/downloadpdf/view/journals/bjrl/44/2/article-p455.pdf) — retrieved 2026-11-18.
- [Reshef Shilon, Nizar Habash, Alon Lavie & Shuly Wintner, Machine Translation between Hebrew and Arabic: Needs, Challenges and Preliminary Solutions, AMTA 2010 student research workshop](https://aclanthology.org/2010.amta-srw.4.pdf) — retrieved 2026-11-18.
- [Shlomi Fish, Can speakers of Modern Hebrew understand Spoken Arabic and vice versa? FAQ](https://www.shlomifish.org/meta/FAQ/hebrew_and_arabic.xhtml) — retrieved 2026-11-18.
- [Rasha Kadry Abdelatti Mohamed Soliman, Arabic Cross-dialectal Conversations with Implications for the Teaching of Arabic as a Second Language, PhD thesis, University of Leeds, July 2014](https://etheses.whiterose.ac.uk/id/eprint/9119/1/PhD%20thesis%20-%20Rasha%20Soliman%20final%20version.pdf) — retrieved 2026-11-18.
- [Salim Abu-Rabia, The Role of Vowels in Reading Semitic Scripts: Data from Arabic and Hebrew, *Reading and Writing* 14:1-2, March 2001](https://eric.ed.gov/?id=EJ624562) — retrieved 2026-11-18 (abstract).
- [WordReference Forums thread, Aramaic vs Hebrew: mutual intelligibility](https://forum.wordreference.com/threads/aramaic-vs-hebrew-mutual-intelligibility.3230620/) — retrieved 2026-11-18 (anecdotal; used only for the ~70% roots estimate, flagged as unsourced).

## Transcript

Corn: Herman, before we get into Daniel's prompt, one ground rule. Nobody is allowed to say "Middle East conflict" for the next half hour.

Herman: That is a rule I can live with. Because the prompt itself is refreshingly pure on that front — Daniel wants us to zoom out from the geopolitics entirely and look at the Semitic language family the way a comparative linguist would. How is the family actually put together, who is closest to whom, how much core vocabulary and grammar do Hebrew, Arabic, Aramaic and their relatives share through the root system, and — this is the part I suspect he really wants — can a Hebrew speaker and an Arabic speaker actually understand each other in practice.

Corn: And the twist, which I'll spoil now because it's what makes the whole thing worth doing: the answer to "who is closest to whom" and the answer to "who can understand whom" are completely different answers. Arabic's nearest relatives turn out to be Hebrew and Aramaic, not the languages you'd guess from a map. And a Hebrew speaker and an Arabic speaker share a branch and approximately zero conversations.

Herman: Then let's build the tree properly, because the tree is the whole game.

Corn: Scale first. What are we holding?

Herman: Semitic is one branch of Afroasiatic, and Afroasiatic has six branches — Berber, Chadic, Cushitic, Egyptian, Omotic, and Semitic. Over five hundred million native speakers across the whole family, which makes it the fourth-largest language family on earth, behind Indo-European, Sino-Tibetan, and Niger-Congo.

Herman: Semitic alone is more than four hundred and sixty million speakers, and the shape of that number is lopsided to the point of comedy. Arabic is four hundred and eleven million native speakers across all its varieties. Then Amharic at thirty-five million, Tigrinya at just under ten, Hebrew at five, Tigre at a million, and Maltese at five hundred and seventy thousand. Arabic is the most spoken native language in both Africa and West Asia. Everything else in the family is a rounding error next to it.

Corn: Five million Hebrew speakers sharing a branch with four hundred and eleven million Arabic speakers. We're the boutique wing.

Herman: You are, yes. And the family name itself is a biblical artefact, not a linguistic one. August Ludwig von Schlözer coined "Semitic" in 1781, from Shem, Noah's son, and Eichhorn popularised it in a 1795 article. But the comparative work predates the label by more than two centuries — Guillaume Postel published a Latin comparison of Hebrew, Arabic and Aramaic in 1538, and Hiob Ludolf later extended the comparison out to the Ethiopian languages.

Corn: So the noticing came first, the branding later.

Herman: And the earliest attested Semitic writing is Akkadian and Eblaite in cuneiform, around 2600 BCE — so this is a family with a written record stretching back four and a half thousand years. But here's the spine for everything that follows: the tree is a hypothesis. It's not a fact someone dug up. It's a reconstruction, and there is exactly one kind of evidence that counts for it, and that rule is going to tear down an older tree and put up a new one.

Corn: The rule.

Herman: John Huehnergard states it flat in his chapter on Arabic in its Semitic context: for establishing genetic relationships among members of a language family, only one type of feature is diagnostic, and that is the shared innovation.

Corn: Meaning a feature the languages share because they each got it from a recent common ancestor after the family had already split. Not something they both just... still have.

Herman: That's the distinction, and it's the whole ballgame. A shared innovation is new — it appeared in one ancestor language and was passed down to its daughters. If Hebrew and Aramaic both have it and Akkadian doesn't, that tells you Hebrew and Aramaic share a parent that Akkadian missed. A shared retention is the opposite: something from the proto-language that everybody simply kept. If all the daughters still have it, it tells you nothing about who is closest to whom, because nobody changed anything.

Corn: It's like two cousins both having their grandfather's nose. That proves they had the same grandfather. It doesn't prove they're closer to each other than to the cousin with the different nose.

Herman: That's a decent — no, actually that's exactly it. The retention is inherited face furniture. The innovation is the family scandal: the thing that happened after the inheritance, that only some of the children know about.

Corn: Our family has several.

Herman: Moving on. Here's why this rule mattered so much. The older model of the family, the one a lot of people still have in their heads, was a three-way split: East Semitic, Northwest Semitic, and South Semitic. East was Akkadian and Eblaite. Northwest was Hebrew, Aramaic, Ugaritic, Phoenician — the Levant. And South was everything south of that: Arabic, the South Arabian languages, the Ethiopian languages like Amharic and Ge'ez, all lumped together, mostly because they were in the south.

Corn: Grouped by where they sit on a map.

Herman: And by features that are shared retentions. Huehnergard goes through them specifically. The broken plural — Arabic has it, and the old argument said, look, it patterns with the other southern languages. But it's a retention, or possibly diffusion, not an innovation. The fāʕala derived verb form — same story. And the sound change of p to f — Arabic, Aramaic and the Ethiopian languages all did it, but that's a change so natural it happened independently all over the place; it doesn't diagnose a shared parent. If your evidence for a subgroup is things the proto-language already had, you've proven the proto-language existed, which nobody doubted.

Larry: Somewhere between the shared innovation and the shared retention, there is a moment. A moment that belongs only to you. Lamsa. The Semitic family candle. Whispered from a single wick, it holds four hundred and sixty million tiny lights, each one a native speaker of calm. Lamsa does not prove a subgroup. It proves you.

Larry: Breathe in the triliteral root of amber and fig. Breathe out everything that was merely retained. Lamsa. Close to Hebrew. Closer to Aramaic. Closest, somehow, to the quiet. Available wherever fine candles are kept, and also wherever they are not.

Herman: The candle again.

Corn: Third week running. The fourth-largest family deserves better copy, but here we are.

Herman: So the old South Semitic node collapses under the shared-innovation rule, and what replaces it is the Hetzron model — and Huehnergard reports this is the consensus among most Semitists today. Robert Hetzron's articles in the nineteen-seventies, Huehnergard's fifteen-feature paper in 2005, restated in the 2017 chapter. And the headline: Arabic moves. It sits not in the south but in a Central Semitic branch, defined by one specific innovation — the shape of the imperfective verb, the yafʕalu form. The prefixed present-tense conjugation. Central Semitic innovated that pattern, and its members are Arabic, the Northwest Semitic languages — Hebrew, Aramaic, Ugaritic, Phoenician and the other Canaanite languages — and the Ancient South Arabian languages.

Corn: So under the consensus, Arabic's closest living relative is Hebrew, and its closest dead ones are Aramaic and Phoenician.

Herman: And Ancient South Arabian, which is not the same thing as the modern South Arabian languages — that's a trap people fall into constantly. Sabaean, the language of Sheba, is Central Semitic's cousin. Soqotri, spoken today on Socotra, is something else entirely and much more distantly related.

Corn: This is the inversion I flagged at the top. Every intuition from geography says Arabic belongs with Amharic — Arabian peninsula, Horn of Africa, the hot countries, one cluster. And the method says no, Arabic's next of kin is the language of the Tanakh and the language of the Talmud.

Herman: The tree is not the map. And the deeper point is that the tree is not finished either. Two things worth being honest about. First, Amorite — a language known almost entirely from personal names and place names, attested from around 2100 BCE — Andrason and Vita ran the corpus against sixty proposed diagnostic features in 2017 and concluded the existing data fail to be conclusive. Sixty features and the answer is a shrug. Second, the general references haven't caught up. Wikipedia's Ethio-Semitic article still describes that group as the western branch of the South Semitic languages — which is precisely the node Huehnergard's argument dissolves. The specialist literature is the current position; the encyclopedia is a generation behind.

Corn: The reference work is a shared retention.

Herman: Ha — it is, actually. Nobody has innovated the Wikipedia page yet. And the specialist edge keeps moving: there's recent work in the International Journal of Arabic Linguistics arguing that the old North Arabian inscriptions, Safaitic and its relatives, aren't an independent branch at all but just another, perhaps archaic, variety of Arabic on a continuum with Hismaic and Nabataean Arabic. So even inside Arabic there's argument about what counts as Arabic.

Corn: Where does the modern Ethiopian branch sit, then, if South Semitic is dissolved?

Herman: Under the arrangement Huehnergard and Rubin set out, the modern Ethiopian Semitic languages, Central Semitic, and the Modern South Arabian languages — Soqotri and its relatives — sit together in a West Semitic sub-branch, with East Semitic, Akkadian and Eblaite, off on its own as the earliest split.

Corn: So that's the shape of the tree. The obvious next question is what all that relatedness actually buys you when somebody opens their mouth.

Herman: Start with what the family shares, because it's a lot. Huehnergard's checklist of what makes a language Semitic: three-consonant roots for most content words. A consonant inventory with pharyngeals, and triads of stops and fricatives — voiced, voiceless, and emphatic — the emphatic ones being the ones produced with a constriction in the throat, like the t and d and s of Arabic. Contrasting suffix-conjugation and prefix-conjugation of the verb, so a past built on suffixes and a present-future built on prefixes. Derived verb stems marked with n and t. Nouns with two genders, three numbers — singular, plural, and dual — and three cases. Default verb-first word order. And the genitive chain construction, the ʔiḍāfa, where you string nouns together without any word for "of."

Corn: I do enjoy that the family's signature move is that nobody says "of."

Herman: It's a elegant system. Arabic: bayt al-rajul, house the-man, "the man's house." Hebrew:beit ha-ish. Same construction, same order, no preposition, the first noun loses its definite article. That is a shared inheritance of grammar, not vocabulary — the machinery itself.

Corn: And the famous example of the machinery is the root.

Herman: The K-T-B root, and I want to do it properly because it's the clearest demonstration of what "shared vocabulary" actually means here. The root is a set of consonants, and in Arabic it's k-t-b, writing. You fill in vowels and affixes around it: kitāb, book. Kutub, books. Kātib, writer. Kuttāb, writers. Kataba, he wrote. Yaktubu, he writes.

Corn: And in Hebrew the same root with the third consonant shifted to v, because Hebrew turned old b into v between vowels.

Herman: Between vowels and at the end of a word after a vowel, yes — a sound change every Hebrew speaker performs constantly without knowing it's a sound change. So k-t-v: katav, he wrote. Yichtov, he will write. Kotev, he writes or writer. Michtav, a letter. Hichtiv, he dictated.

Corn: The thing that lands for me there is that the shared item isn't the word. Nobody's word matches. It's the skeleton — three consonants that both languages know mean "writing," dressed differently.

Herman: That's why the shared vocabulary is better thought of as a list of consonant frames than a list of cognate words. And it's also why reading works differently than listening. Corn, you know this from living here — you can see a Semitic word you've never met and often guess the root, because the consonants carry the meaning and the script writes the consonants. Both Hebrew and Arabic script, being abjads, omit the vowels most of the time. Feasible precisely because the consonants are the primary carriers of meaning.

Corn: I'll say the one thing that needs saying about the old reflex, because you'll enjoy it: nineteenth-century Europeans treated Classical Arabic as basically Proto-Semitic come to life.

Herman: They nearly did. Huehnergard has a lovely passage on it — Arabic seemed to them the most archaic and conservative of all Semitic languages, and the earliest comparative studies almost treat Arabic as though it were in fact Proto-Semitic. Which is understandable — simple vowel system, this enormous consonant inventory, vast vocabulary, complex tense and mood. It looks like the showroom model. But of course Arabic is not Proto-Semitic. Its phonology is very conservative, yes, but other languages are more conservative in other respects — the consonant inventories of the Ancient South Arabian languages are closer to the reconstructed proto-system in some details — and Arabic has innovated heavily on its own account. The broken plural is an Arabic-side development, not a family heirloom. The n-prefixed passive stem is an Arabic speciality that Aramaic and Ethiopic never had.

Corn: A showroom model with aftermarket parts.

Herman: Though — and this is the part I love — Arabic preserved two fossils that only comparative Semitism explains, and they're beautiful. The first: the causative. The original Semitic causative marker was a prefix s. Very early, that s became h everywhere in most of the family — a sound change that only applied before a vowel. In Arabic's basic causative, Form IV, you see the h: ʔafʕala replaced the old s-form. But in Form X, istafʕala, the s survived — because in that shape the s was never pre-vocalic, so the sound change never touched it. The tenth form is a phonological bunker. The old causative marker sat inside it for four thousand years.

Corn: It survived by being in the wrong place at the right time.

Herman: The second one: the n of Form VII, yanfaʕil. That n goes back to Akkadian, where an n-prefix marked the ingressive — becoming, getting to be. Ya-n-paʕil: it came to be done. Arabic kept the n, lost track of what it meant, and built a whole verb form on a fossilized prefix. Herman from Connecticut, decades in Jerusalem, and I will tell you the Hebrew equivalent of that form barely exists — hitpa'el is a different animal — so Arabic is holding family heirlooms Hebrew threw out.

Corn: You've clearly been waiting all episode to say the bunker thing.

Herman: I've been waiting since Tuesday.

Corn: Now the intelligibility half. Because this is where Daniel's question gets its punchline.

Herman: Hebrew and Arabic share the Central Semitic node, share the root system, share the ʔiḍāfa, share the verb structure — and are mutually incomprehensible. The machine-translation literature just states it as a premise: Hebrew and Arabic are related but mutually incomprehensible languages with complex morphology and scarce parallel corpora. First-person testimony agrees. Shlomi Fish, a native Hebrew speaker who studied Literary Arabic for six years — six years — reports he still cannot understand the local spoken Arabic dialects except for a few words.

Corn: Six years of study buys you a few words.

Herman: Of the literary language, which then doesn't transfer to the street. And here's what makes it stranger: intelligibility isn't stable inside a single language, let alone between two. Modern Standard Arabic differs, in the literature's own words, in a non-trivial manner from the spoken regional dialects — which are described as the true native languages of Arabic speakers, and which differ quite a bit from each other. There's a whole research subfield on Arabic dialect identification for computational work, because a Moroccan and an Iraqi can be at odds in conversation.

Corn: Has anyone actually measured the cross-dialect comprehension?

Herman: Rasha Soliman's Leeds doctoral work did — she recorded eleven conversations among speakers of twelve Arabic dialects, natural conversations, and analysed how they managed. And the finding was that comprehension wasn't automatic. It was propped up by strategy: leaning on context, ignoring the non-content words, using root-and-pattern knowledge to reconstruct what a word must mean, and crucially, prior familiarity with the other dialect. People were working at understanding each other. The relatedness gives you raw material, not free comprehension.

Corn: So within one "language," speakers are doing active interpretive labour. Between two sibling languages, nothing happens at all.

Herman: And the biblical record has the cleanest intelligibility test I know, twenty-seven hundred years old. Second Kings, chapter eighteen. The Assyrian army is at the wall of Jerusalem, and the official — the Rabshakeh — calls out to the people in Hebrew, presumably to demoralise them, to get them to surrender. And Eliakim and Shebna and Joah, the Judaean officials on the wall, say to him: "Speak, I pray thee, unto thy servants in Aramaic, for we understand it" — speak Aramaic to us, because the people standing within earshot don't. The officials spoke Aramaic. The crowd spoke Hebrew. Two languages, one wall, and the negotiation turns on exactly that gap.

Corn: The first recorded instance of "let's take this offline."

Herman: Ullendorff made the observation I love most about that scene: the Assyrian officers' command of Hebrew must have been excellent if they were able to address the people assembled on the wall. Everyone in the scene is bilingual except the crowd. And Gzella's judgement on the underlying pair is that the Aramaic of Damascus and the Hebrew of Jerusalem were, in antiquity, fair to assume were not mutually intelligible — close cousins, same region, no shared conversation.

Corn: Which is roughly where Hebrew and Arabic are today. Same branch, different world.

Herman: Aramaic's own arc is the warning label on the whole topic. Between the mid-first millennium BCE and the spread of Islam, Aramaic was the international language of communication in the Ancient Near East — Gzella calls it by far the most important language in the region, running from Egypt to Afghanistan, Anatolia to Arabia. It was the Achaemenid chancellery language. It's the language of chunks of Ezra and Daniel — the Writing on the Wall is in Aramaic. And today it survives in scattered communities, several of them endangered.

Hilbert: The Writing on the Wall is Daniel chapter five. Not Second Kings.

Herman: It is, yes — chapter five of Daniel, I ran the two scenes together.

Hilbert: I've been keeping a list on the same subject, so it's on my mind. A notebook. Paper, indexed by consonant. Every occurrence of one root, logged in Hebrew, Aramaic and Arabic as I meet it — I mean met it in the wild, not in a dictionary. There's a rule: I don't look a root up until I've found it three times on my own. Broke the rule once, for the root with the ayin, and I still don't like talking about it.

Corn: Which root?

Hilbert: Doesn't matter. The point is the counts. The books promise you a root behaves — same skeleton, three languages, regular as bricks. Mine doesn't. The Aramaic side has a form that only shows up twice in the whole notebook, and one of those is a borrowing everyone agrees is a borrowing, so I don't see why it counts. Forty years in trades before this desk, mind — I indexed parts catalogues the same way, by part number, and a catalogue that lies to you about fitment is a worse catalogue, not a different philosophy.

Herman: Two attestations, one of them a loan — that's not a pattern, that's a coincidence with paperwork.

Hilbert: That's what I said to my cousin in Beersheba, who does this for a living, the actual linguistics, and he says the root is fine and the notebook is over-strict. He is not to be trusted on roots, or on anything involving his landlord. But he's probably right about the notebook. The index outgrew the binding in the spring. It sits next to the mixing desk now, under the meter bridge, and the pen is the good one, the one with the fine tip that fits between the columns — you try writing with a claw made for tearing open a termite mound and you'll take the fine tip where you can get it.

Corn: The good pen has a location and the root has a criminal record.

Hilbert: The root does, yes. Fine tip pen, notebook, one rule broken in — a long while. My cousin says everybody breaks the rule. My cousin also says "transliteration is not a language," which for a man who owes his landlord money is oddly firm. I'll say this for the people who do the comparative work, though — the ones I've met, at conferences, at the desk in Beersheba — they argue about everything except whether the family is real. They know what they're doing. More than I could say for the parts catalogues.

Herman: The cousin who can't be trusted, vouching for the method.

Hilbert: Ineffectively, but he tries. The notebook's at forty-one entries now. Forty-two if you count the borrowing, which I don't.

Corn: You counted it anyway.

Hilbert: Of course I counted it. You don't leave a root uncounted just because you disapprove of it.

Herman: One root, forty-one entries, indexed by consonant, kept by hand. And the specialists argue about everything except whether the family is real — that's actually the honest state of the field, Corn. And that gets us to the part of Daniel's question people always reach for and nobody should.

Corn: The percentage.

Herman: "How much vocabulary do they share — is it sixty percent, is it eighty?" No published cognate-percentage figure exists for Hebrew versus Arabic, Hebrew versus Aramaic, or Arabic versus Aramaic. Comparative Semitists don't work that way — they work with shared innovations and isoglosses, individual features that cut the family at particular joints. The percentage framing is a popular-science import from lexicostatistics, the Swadesh-list school, which never seriously took root in Semitic studies.

Corn: There's a number floating around, though. I've seen seventy percent.

Herman: You have, because someone on a language forum once estimated that a reader of Biblical Hebrew would recognise about seventy percent of the roots in Biblical Aramaic. One person, on a forum, guessing. It might even be roughly right — but it's a guess wearing a lab coat, and it's not a measurement. What's actually true and checkable is better than the number: the shared material is a system, roots and patterns, not a word list. You don't inherit a dictionary. You inherit a factory.

Corn: A factory. And the factory's products don't resemble each other on the shelf.

Herman: Which is the clean answer to Daniel's last part. In practice — in the street, on the wall of Jerusalem, twenty-seven centuries ago or this afternoon — genealogical closeness buys you almost nothing at the level of conversation. It buys you something rarer: a Hebrew speaker and an Arabic speaker can both take an unfamiliar word apart, extract the consonants, and often land in the same semantic neighbourhood. Understanding the shape of a word is not understanding the word.

Corn: If you take one thing from this episode, take that split. The family tree tells you who descends from whom. It tells you nothing about who can talk to whom — Arabic and Hebrew share a branch and can't order coffee together, while Arabic's own dialects are so far apart that people study the gap for doctorates.

Herman: And the reason the split exists is worth holding onto: relatedness is carried by the grammar and the root system, which persist silently, while the actual surface — sounds, vocabulary, idiom — changes fast and constantly. The deep structure survives; the conversation doesn't. That's the Semitic family's whole trick.

Corn: The open question I'd leave with Daniel is the one the field is still working: the tree is a hypothesis under active revision, and some of the evidence is disappearing faster than it's being classified. Amorite is inconclusive on sixty features. The South Semitic node is still printed in general references. And Soqotri — a language sitting in that West Semitic branch we mapped out — is classified by UNESCO as severely endangered. Soqotri is the language people will be studying a century from now mostly from recordings, if the recordings get made.

Herman: A family that wrote in 2600 BCE, and the newest branch of it is the one running out of speakers. The tree's not finished being drawn, and part of the paper's being erased at the edges.

Corn: We should mention the producer credit — Hilbert Flumingtop produces this show, apparently while indexing roots by consonant at the same desk. Don't ask him which root. That's been My Weird Prompts. If today's tree got you curious, the whole archive is at my weird prompts dot com — and a review wherever you listen is worth more than you'd think.

Herman: We'll be back soon. Until then — shalom, salam, and shlama, three languages, one root.

Corn: One stubborn consonant frame. Goodnight.
