WEBVTT 00:00:00.000 --> 00:00:03.720 What if I told you that in some languages, the 00:00:03.720 --> 00:00:07.080 word for decentralization and the literal phrase 00:00:07.080 --> 00:00:09.720 for stabbing a spear into the dirt are actually 00:00:09.720 --> 00:00:12.619 mathematically identical at their core? Yeah, 00:00:12.740 --> 00:00:15.119 which is just, it sounds completely absurd when 00:00:15.119 --> 00:00:16.579 you first hear it. Right, because you usually 00:00:16.579 --> 00:00:18.140 don't think about the words coming out of your 00:00:18.140 --> 00:00:20.719 mouth as having some kind of mathematical formula. 00:00:20.859 --> 00:00:23.179 It's like, when we look at a massive skyscraper, 00:00:23.539 --> 00:00:26.519 we just see the shiny glass. Right. Right. and 00:00:26.519 --> 00:00:29.300 the solid steel exterior. Exactly, yeah. We don't 00:00:29.300 --> 00:00:32.840 see the rebar or the concrete foundation or, 00:00:32.840 --> 00:00:35.899 you know, the thousands of tiny metal brackets 00:00:35.899 --> 00:00:38.020 holding all those massive beams together. We 00:00:38.020 --> 00:00:40.479 just appreciate the finished product. We assume 00:00:40.479 --> 00:00:43.140 it's one solid, unchangeable structure. Right. 00:00:43.280 --> 00:00:45.399 But if you had x -ray vision, you'd suddenly 00:00:45.399 --> 00:00:48.299 see all that invisible architecture. You'd realize 00:00:48.299 --> 00:00:51.100 that this giant complex tower is actually built 00:00:51.100 --> 00:00:53.939 from a surprisingly small set of fundamental 00:00:53.939 --> 00:00:56.820 pieces. They're just repeating and combining 00:00:56.820 --> 00:00:59.880 in different ways. And that is exactly our mission 00:00:59.880 --> 00:01:02.840 for today's Deep Dive. We are taking X -ray vision 00:01:02.840 --> 00:01:06.560 to human language. We are going to drill down 00:01:06.560 --> 00:01:10.340 into the irreducible core of how we communicate, 00:01:11.180 --> 00:01:13.500 stripping away the surface to look at the hidden 00:01:13.500 --> 00:01:15.760 linguistic scaffolding underneath. I love that 00:01:15.760 --> 00:01:18.180 framing. Yeah, and by the time we're done here, 00:01:18.480 --> 00:01:20.939 you, the listener, you will never look at a simple, 00:01:21.120 --> 00:01:24.180 everyday word the same way again. Okay, let's 00:01:24.180 --> 00:01:26.739 unpack this. So before we can really understand 00:01:26.739 --> 00:01:29.159 the grand architecture of thousands of different 00:01:29.159 --> 00:01:31.159 human languages, we have to look at the individual 00:01:31.159 --> 00:01:33.739 bricks. We have to define what a root actually 00:01:33.739 --> 00:01:37.400 is. Right. And a root, which is sometimes called 00:01:37.400 --> 00:01:40.890 a root word or a radical, or an edamon, is the 00:01:40.890 --> 00:01:43.530 absolute core of a word. Like the baseline. Exactly. 00:01:43.629 --> 00:01:45.810 It's the primary lexical unit. Basically, it's 00:01:45.810 --> 00:01:47.849 a piece of meaning that cannot be reduced down 00:01:47.849 --> 00:01:50.409 any further. If you try to chop it up into smaller 00:01:50.409 --> 00:01:52.689 parts, it just completely loses its semantic 00:01:52.689 --> 00:01:54.769 meaning. It's the ultimate base layer. And the 00:01:54.769 --> 00:01:57.189 way linguists actually write this down is wild 00:01:57.189 --> 00:01:59.469 to me. Yeah, the notation. Yeah. They treat it 00:01:59.469 --> 00:02:02.469 almost like an equation. To avoid confusing a 00:02:02.469 --> 00:02:06.049 root with just a normal, fully formed word, linguists 00:02:06.049 --> 00:02:08.430 often use the mathematical radical symbol, like 00:02:08.430 --> 00:02:10.719 the the literal square root symbol. Which is 00:02:10.719 --> 00:02:13.099 so cool. It really is. So if someone wants to 00:02:13.099 --> 00:02:15.120 talk about the ancient Sanskrit root for the 00:02:15.120 --> 00:02:17.340 concept of being or becoming, they don't just 00:02:17.340 --> 00:02:19.819 write out BHU. They write the square root symbol 00:02:19.819 --> 00:02:23.020 and then boo. Right. It visually reminds you 00:02:23.020 --> 00:02:25.379 that you are looking at the foundational math 00:02:25.379 --> 00:02:28.020 of a language. It acts as a visual cue saying, 00:02:28.080 --> 00:02:31.599 hey, we have drilled down to the bedrock here. 00:02:31.680 --> 00:02:34.520 Yeah. And traditionally, these foundational bricks 00:02:34.520 --> 00:02:37.120 are divided into two categories. You have free 00:02:37.120 --> 00:02:40.860 morphemes and bound morphemes. And English, as 00:02:40.860 --> 00:02:43.199 a language, is heavily biased toward the free 00:02:43.199 --> 00:02:45.860 route. We really minimize what is called affixation. 00:02:46.430 --> 00:02:48.810 Affixation, meaning gluing prefixes or suffixes 00:02:48.810 --> 00:02:50.810 onto a word. We generally don't like doing that 00:02:50.810 --> 00:02:53.370 if we don't have to. Exactly. We strongly prefer 00:02:53.370 --> 00:02:56.030 our everyday words to be identical to their roots. 00:02:56.569 --> 00:02:59.849 Think of a free root like a fully formed independent 00:02:59.849 --> 00:03:02.110 Lego block. OK, I like that. So take the English 00:03:02.110 --> 00:03:05.509 word running. The root is run. You can just take 00:03:05.509 --> 00:03:07.750 run, set it down on the table all by itself, 00:03:07.750 --> 00:03:10.050 and it works perfectly in a sentence. It's free. 00:03:10.270 --> 00:03:12.909 Right. It doesn't need any help. But stepping 00:03:12.909 --> 00:03:15.419 outside of English, That Lego block approach 00:03:15.419 --> 00:03:17.680 is definitely not universal. Oh, not at all. 00:03:17.879 --> 00:03:20.300 Like, take polysynthetic languages. For example, 00:03:20.539 --> 00:03:22.900 Yupik, which is spoken by indigenous peoples 00:03:22.900 --> 00:03:26.599 in Alaska and Siberia. Yupik is an incredible 00:03:26.599 --> 00:03:29.520 contrast to English. In a polysynthetic language, 00:03:29.800 --> 00:03:32.099 words are formed through extreme inflection and 00:03:32.099 --> 00:03:34.240 compounding. Extreme compounding. Yeah, they 00:03:34.240 --> 00:03:36.639 basically take a single root and just endlessly 00:03:36.639 --> 00:03:39.229 stack bound elements onto it. They do this to 00:03:39.229 --> 00:03:41.930 convey time, subject, object, location, all of 00:03:41.930 --> 00:03:45.229 it. A single Yupik word can contain the exact 00:03:45.229 --> 00:03:47.430 same amount of grammatical information as an 00:03:47.430 --> 00:03:49.689 entire English sentence. Which is just mind -blowing. 00:03:49.889 --> 00:03:51.909 And because they rely so heavily on stacking 00:03:51.909 --> 00:03:54.110 all these bound elements onto a single base, 00:03:54.610 --> 00:03:56.689 their pool of free roots, meaning words that 00:03:56.689 --> 00:03:58.990 can just stand completely alone without any extra 00:03:58.990 --> 00:04:02.050 pieces attached, is remarkably small. Very small. 00:04:02.330 --> 00:04:04.969 Yupik restricts its free roots to no more than 00:04:04.969 --> 00:04:08.699 2 ,000 words. 2 ,000! They generate a massive, 00:04:09.139 --> 00:04:11.960 infinitely expressive language from just 2 ,000 00:04:11.960 --> 00:04:15.319 standalone pieces. That extreme efficiency really 00:04:15.319 --> 00:04:18.879 shows how much heavy lifting a root can do when 00:04:18.879 --> 00:04:20.860 a language is designed to maximize it. Totally. 00:04:21.120 --> 00:04:22.939 But let's talk about the other side of that coin, 00:04:23.120 --> 00:04:25.259 the bound roots. Because we actually have these 00:04:25.259 --> 00:04:27.319 in English, too, and they are kind of weird. 00:04:27.500 --> 00:04:30.899 Yeah. Like in Spanish, you have the superlative 00:04:30.899 --> 00:04:33.779 adjective amplicimo, where the root amplo cannot 00:04:33.779 --> 00:04:36.500 stand on its own. It requires modification. But 00:04:36.500 --> 00:04:38.819 a great English example is the word interrupt. 00:04:38.959 --> 00:04:42.360 So the prefix is interrupt, meaning between, 00:04:42.439 --> 00:04:45.500 and the root is unrupt. But here's where I get 00:04:45.500 --> 00:04:49.379 confused. If unrupt is the root and it carries 00:04:49.379 --> 00:04:52.160 the core meaning of breaking or bursting, why 00:04:52.160 --> 00:04:54.120 can't I just go around rupting things? Right, 00:04:54.120 --> 00:04:56.519 you can't just say, I'm going to rupt that. Exactly. 00:04:56.579 --> 00:04:59.600 It always needs inter or core attached to it. 00:04:59.620 --> 00:05:01.759 Why is it trapped? Well, what's fascinating here 00:05:01.759 --> 00:05:03.639 is that this perfectly illustrates the hidden 00:05:03.639 --> 00:05:05.399 scaffolding we were talking about earlier. You 00:05:05.399 --> 00:05:08.279 cannot go around rupting things because English 00:05:08.279 --> 00:05:11.480 swallowed that specific Latin genetic code whole 00:05:11.480 --> 00:05:13.860 without ever granting it independent status. 00:05:14.199 --> 00:05:17.480 Ah. It is permanently bound. But its presence, 00:05:17.800 --> 00:05:21.160 across words like disrupt, corrupt, rupture, 00:05:21.420 --> 00:05:23.800 and interrupt, that proves that the root exists. 00:05:24.000 --> 00:05:25.959 It's an inherited piece of linguistic architecture. 00:05:25.980 --> 00:05:29.579 Right. Gives us a glimpse into how a root provides 00:05:29.579 --> 00:05:32.980 the genetic code. for a whole family of words, 00:05:33.360 --> 00:05:35.779 even if it never gets to step out into the daylight 00:05:35.779 --> 00:05:39.139 by itself. It's like linguistic DNA. But OK, 00:05:39.319 --> 00:05:42.600 English generally builds words by taking solid 00:05:42.600 --> 00:05:45.120 fixed Lego pieces and maybe snapping a prefix 00:05:45.120 --> 00:05:47.779 on it, which is a process called concatenative 00:05:47.779 --> 00:05:50.160 morphology, where you chain things end to end. 00:05:50.579 --> 00:05:52.740 What do you do if you don't have Legos? Right. 00:05:52.819 --> 00:05:54.759 Because some languages don't build with blocks 00:05:54.759 --> 00:05:57.019 at all. They build with skeletons. Yes. And now 00:05:57.019 --> 00:05:59.319 we are moving from concatenative languages to 00:05:59.319 --> 00:06:01.750 non -concatenative language. And the prime examples 00:06:01.750 --> 00:06:04.529 here are the Semitic languages like Arabic and 00:06:04.529 --> 00:06:06.990 Hebrew that use consonantal roofs. Instead of 00:06:06.990 --> 00:06:09.709 a fixed block like run or rupt, the foundational 00:06:09.709 --> 00:06:12.329 root of a word is typically just a skeleton of 00:06:12.329 --> 00:06:14.250 three consonants. Sometimes two, sometimes up 00:06:14.250 --> 00:06:17.110 to five, but typically three. So they build words 00:06:17.110 --> 00:06:19.629 by taking that three -consonant skeleton and 00:06:19.629 --> 00:06:21.990 injecting different vowels into the empty spaces. 00:06:22.529 --> 00:06:24.910 Exactly. The consonants provide the broad abstract 00:06:24.910 --> 00:06:27.649 meaning, and the vowels are poured in to determine 00:06:27.649 --> 00:06:30.290 the specific grammatical function. Like, is it 00:06:30.290 --> 00:06:32.850 a noun, a verb, and an adjective? Right. It's 00:06:32.850 --> 00:06:34.529 a completely different way of engineering a word. 00:06:34.569 --> 00:06:37.050 Let's take Hebrew phonetics, for example. Look 00:06:37.050 --> 00:06:40.189 at the three -consonant root G -D -L, gimel, 00:06:40.329 --> 00:06:43.589 dalet, lamed. This abstract root carries the 00:06:43.589 --> 00:06:47.110 general sweeping idea of largeness. Yes. So if 00:06:47.110 --> 00:06:49.949 you inject an A and an O into the skeleton, you 00:06:49.949 --> 00:06:52.490 get gadal, which is the masculine adjective for 00:06:52.490 --> 00:06:54.930 big. And if you change the vowels, you change 00:06:54.930 --> 00:06:57.269 the grammatical function. But that core concept 00:06:57.269 --> 00:07:00.230 of largeness remains perfectly intact. Inject 00:07:00.230 --> 00:07:03.089 different vowels into GDL, and gadal becomes 00:07:03.089 --> 00:07:06.389 the verb he grew. Right. Higdal means he magnified. 00:07:07.029 --> 00:07:09.589 Maggillet is a magnifier. And gadal means size, 00:07:09.870 --> 00:07:13.060 and migdal means tower. All of those wildly different 00:07:13.060 --> 00:07:16.639 words, big, grew, magnified, size, tower, they 00:07:16.639 --> 00:07:19.779 all share the exact same GDL skeleton. But wait, 00:07:19.959 --> 00:07:21.759 you can't actually pronounce GDL without vowels. 00:07:21.779 --> 00:07:26.019 No, you can't. So is the root GDL a real word 00:07:26.019 --> 00:07:29.100 anyone would ever say out loud? Or is it just 00:07:29.100 --> 00:07:31.839 a raw abstract formula floating in the background 00:07:31.839 --> 00:07:33.699 of all these other words? Well, if we connect 00:07:33.699 --> 00:07:36.259 this to the bigger picture, you've just hit on... 00:07:36.009 --> 00:07:38.370 the most profound difference between a language 00:07:38.370 --> 00:07:41.129 like English and a Semitic language. Oh, really? 00:07:41.250 --> 00:07:44.149 Yeah. These continental roots are entirely abstract. 00:07:44.509 --> 00:07:46.329 You will never walk down the street in Tel Aviv 00:07:46.329 --> 00:07:49.730 and hear someone just say G -D -D -L. It doesn't 00:07:49.730 --> 00:07:52.829 exist as spoken speech. Wow. It acts as an invisible 00:07:52.829 --> 00:07:55.860 conceptual anchor. For speakers of these languages, 00:07:56.259 --> 00:07:58.100 the meaning of a word isn't just tied to the 00:07:58.100 --> 00:08:00.560 specific sounds they're making right now. It's 00:08:00.560 --> 00:08:03.139 explicitly tied to a shared underlying mathematical 00:08:03.139 --> 00:08:05.740 idea that connects a whole web of vocabulary. 00:08:06.120 --> 00:08:08.019 So a tower and a magnifier are fundamentally 00:08:08.019 --> 00:08:10.139 the same word built on different scaffolding. 00:08:10.379 --> 00:08:12.279 Exactly. But that brings up a really practical 00:08:12.279 --> 00:08:15.160 problem. If a society relies on these ancient 00:08:15.160 --> 00:08:17.939 abstract skeletons to build their words, how 00:08:17.939 --> 00:08:20.660 do they invent entirely new concepts. Do they 00:08:20.660 --> 00:08:23.279 have to invent entirely new three -letter skeletons 00:08:23.279 --> 00:08:25.319 from scratch every time a new piece of technology 00:08:25.319 --> 00:08:27.939 drops? No, not at all. They use what are called 00:08:27.939 --> 00:08:31.000 secondary roots. They take an existing root and 00:08:31.000 --> 00:08:34.059 they mutate it to create a new baseline for a 00:08:34.059 --> 00:08:36.500 slightly different meaning. This allows the vocabulary 00:08:36.500 --> 00:08:39.679 to expand infinitely without needing to invent 00:08:39.679 --> 00:08:42.940 entirely new architectural blueprints from scratch. 00:08:43.120 --> 00:08:45.740 And this connects perfectly to that mind -bending 00:08:45.740 --> 00:08:48.679 hook from the start of the deep dive. Let's look 00:08:48.679 --> 00:08:52.419 at the Arabic root RKZ. Originally, this ancient 00:08:52.419 --> 00:08:55.179 root relates to the highly physical action of 00:08:55.179 --> 00:08:57.779 planting a lance or sticking a spear into the 00:08:57.779 --> 00:09:00.259 earth. A very specific ancient action. Yeah. 00:09:00.600 --> 00:09:02.580 And from that root action, they derive the word 00:09:02.580 --> 00:09:05.139 Markaz, which means center. And you can easily 00:09:05.139 --> 00:09:07.539 follow the cognitive leap there, right? Planting 00:09:07.539 --> 00:09:10.440 a flag or a lance establishes a central point. 00:09:10.639 --> 00:09:12.659 Right, but the evolution doesn't stop there. 00:09:12.889 --> 00:09:16.070 The language then takes the consonants from Markaz, 00:09:16.090 --> 00:09:20.070 M -R -K -Z and treats that as a brand new four 00:09:20.070 --> 00:09:22.830 -letter secondary root. And from that new mutated 00:09:22.830 --> 00:09:25.269 root they build entirely modern concepts. They 00:09:25.269 --> 00:09:28.049 get Markazi for central, Markazi for centralization, 00:09:28.450 --> 00:09:30.870 and even Lamarkazi, which means decentralization. 00:09:31.549 --> 00:09:33.830 They took a physical spear in the dirt, stretched 00:09:33.830 --> 00:09:36.610 the cognitive concept, mutated the consonants, 00:09:36.730 --> 00:09:39.210 and ended up with a modern geopolitical term 00:09:39.210 --> 00:09:42.309 for decentralization. And Arabic does this constantly. 00:09:42.250 --> 00:09:46.309 The root rj means way down. That mutated into 00:09:46.309 --> 00:09:48.870 the concept of a swing, which mutated into a 00:09:48.870 --> 00:09:52.639 secondary root for oscillated, or SR, which means 00:09:52.639 --> 00:09:55.460 mocked, mutated into the secondary root nekester, 00:09:55.600 --> 00:09:57.899 meaning made fun. And Hebrew does the exact same 00:09:57.899 --> 00:10:01.299 thing. The biblical Hebrew root for Stan is QNQM. 00:10:01.580 --> 00:10:04.159 From that skeleton, they built the word Mokum, 00:10:04.460 --> 00:10:07.320 meaning place. Right. Modern Israeli Hebrew then 00:10:07.320 --> 00:10:10.279 took the consonants from Mokum MQM and turned 00:10:10.279 --> 00:10:12.240 that into a brand new secondary root, meaning 00:10:12.240 --> 00:10:15.059 to locate. The language is endlessly recycling 00:10:15.059 --> 00:10:17.559 its own material to meet modern needs. And the 00:10:17.559 --> 00:10:20.070 linguist Gilad Zuckerman points out. This isn't 00:10:20.070 --> 00:10:22.629 exclusively a Semitic phenomenon. He compares 00:10:22.629 --> 00:10:24.789 it to the production of frequentative verbs in 00:10:24.789 --> 00:10:26.590 Latin. Oh, interesting. Yeah, the Latin word 00:10:26.590 --> 00:10:30.549 for to throw is iasio. Through grammatical stretching, 00:10:30.710 --> 00:10:33.590 that morphed into iactito, which means to toss 00:10:33.590 --> 00:10:36.740 about or keep bringing up. is taking a core root 00:10:36.740 --> 00:10:39.159 and pulling it like taffy to create a nuanced 00:10:39.159 --> 00:10:41.600 secondary meaning. Here's where it gets really 00:10:41.600 --> 00:10:43.940 interesting, because we do something conceptually 00:10:43.940 --> 00:10:46.539 similar in English, like taking the root verb 00:10:46.539 --> 00:10:49.980 to conduct, as in to conduct an orchestra, and 00:10:49.980 --> 00:10:52.919 mutating it into the noun conductor, which then 00:10:52.919 --> 00:10:55.759 functions as its own solid base for further word 00:10:55.759 --> 00:10:58.740 building. Exactly. But examining how these roots 00:10:58.740 --> 00:11:01.759 stretch and mutate leads us to what might be 00:11:01.759 --> 00:11:04.700 the most fiercely debated area in modern morphology. 00:11:04.809 --> 00:11:07.990 Absolutely. We've talked about how roots provide 00:11:07.990 --> 00:11:10.370 core meaning, but when a root is just sitting 00:11:10.370 --> 00:11:13.009 there in the dictionary completely bare, does 00:11:13.009 --> 00:11:16.049 it actually have a grammatical identity? Is the 00:11:16.049 --> 00:11:19.730 concept of a chair fundamentally a noun or fundamentally 00:11:19.730 --> 00:11:22.490 a verb? Which brings us to the concept of category 00:11:22.490 --> 00:11:25.590 -neutral roots. This was born out of decompositional 00:11:25.590 --> 00:11:27.769 generative frameworks. Which is a heavy wall 00:11:27.769 --> 00:11:29.549 of jargon, let's be honest. Yeah, it's a mouthful. 00:11:29.590 --> 00:11:31.710 Let's break that down. Yeah. Decompositional 00:11:31.710 --> 00:11:33.789 simply means we are breaking language down into 00:11:33.789 --> 00:11:36.720 its absolute smallest parts to see how it works. 00:11:37.200 --> 00:11:39.120 Generative frameworks refers to the theories 00:11:39.120 --> 00:11:41.820 about how human beings generate or build sentences 00:11:41.820 --> 00:11:44.440 in their minds. Okay, so within that framework, 00:11:44.620 --> 00:11:47.000 the theory is that roots are totally blank slates. 00:11:47.279 --> 00:11:49.539 They hold almost zero grammatical information. 00:11:49.899 --> 00:11:52.379 They don't know if they are a noun or a verb 00:11:52.379 --> 00:11:54.919 until the surrounding syntax, like the actual 00:11:54.919 --> 00:11:57.220 sentence they are placed into, forces them to 00:11:57.220 --> 00:11:59.779 choose an identity. Right. Linguists generally 00:11:59.779 --> 00:12:02.379 agree that a root has some underlying conceptual 00:12:02.379 --> 00:12:04.960 meaning, but there is still heavy debate on whether 00:12:04.960 --> 00:12:07.639 these category neutral roots contain what is 00:12:07.639 --> 00:12:10.440 called a semantic type or an argument structure. 00:12:10.600 --> 00:12:12.899 Right, and to put that simply, an argument structure 00:12:12.899 --> 00:12:15.940 dictates who is doing what to whom. Like knowing 00:12:15.940 --> 00:12:18.840 a verb requires a subject and an object. Yes, 00:12:19.000 --> 00:12:21.600 and a semantic type is the inherent classification. 00:12:22.039 --> 00:12:24.700 Is it a physical thing, an action, a property? 00:12:25.100 --> 00:12:27.460 The blank slate theory argues the root contains 00:12:27.460 --> 00:12:29.419 none of that until the sentence builds a house 00:12:29.419 --> 00:12:31.820 around it. And the evidence supporting this blank 00:12:31.820 --> 00:12:34.000 slate idea is incredibly intuitive when you look 00:12:34.000 --> 00:12:36.659 at English. Look at words with covert morphology, 00:12:36.679 --> 00:12:38.379 meaning the word doesn't change its spelling 00:12:38.379 --> 00:12:42.590 or add prefixes. Words like dance, walk, Chair. 00:12:42.950 --> 00:12:46.169 Wardrobe. I want to dance. Verb. That was a great 00:12:46.169 --> 00:12:49.669 dance. Noun. Please chair this meeting. Verb. 00:12:49.909 --> 00:12:52.909 Sit in that chair. Noun. The exact same root 00:12:52.909 --> 00:12:54.730 is playing completely different roles without 00:12:54.730 --> 00:12:57.710 changing a single letter. It is category neutral. 00:12:58.149 --> 00:13:00.269 And even with overt morphology where we actually 00:13:00.269 --> 00:13:03.529 add prefixes or suffixes, the underlying root 00:13:03.529 --> 00:13:06.490 is still neutral. Yeah. The root employee becomes 00:13:06.490 --> 00:13:08.590 the noun employment or the verb to employ. The 00:13:08.590 --> 00:13:10.289 root character becomes the noun character or 00:13:10.289 --> 00:13:14.159 the The root itself is just raw conceptual potential 00:13:14.159 --> 00:13:16.500 waiting to be activated by its environment. It's 00:13:16.500 --> 00:13:18.419 wild, but let's look at Hebrew again to see how 00:13:18.419 --> 00:13:21.000 this blank slate idea functions inside a skeleton 00:13:21.000 --> 00:13:25.379 framework. Take the Hebrew root in, shin -mem. 00:13:25.659 --> 00:13:28.779 None. This root carries the general category 00:13:28.779 --> 00:13:31.120 neutral concept of a greasy, fatty material. 00:13:31.379 --> 00:13:33.419 Yes, and according to the linguist Maya Arad, 00:13:33.740 --> 00:13:36.299 that consonantal root is only turned into a specific 00:13:36.299 --> 00:13:39.220 word due to pattern morphology. Pattern morphology. 00:13:39.360 --> 00:13:41.360 Right. Pattern morphology is the specific mold 00:13:41.360 --> 00:13:44.000 of vowels you pour the skeleton into. If you 00:13:44.000 --> 00:13:46.179 pour in into a verbal mold, it gets assigned 00:13:46.179 --> 00:13:48.210 verbal features. poured into a noun mold, it 00:13:48.210 --> 00:13:50.269 gets noun features. So depending on the vowel 00:13:50.269 --> 00:13:52.389 pattern you apply to shun, you get shunmen for 00:13:52.389 --> 00:13:55.350 oil, shunmen for cream, shunmen for fat, and 00:13:55.350 --> 00:13:58.289 shunmen for to grow fat. Exactly. They all vary 00:13:58.289 --> 00:14:01.450 semantically, but they all clearly stem from 00:14:01.450 --> 00:14:04.210 that exact same neutral concept of greasiness. 00:14:04.690 --> 00:14:07.590 So what does this all mean for how a human brain 00:14:07.590 --> 00:14:10.809 is actually wired? This raises an important question 00:14:10.809 --> 00:14:13.269 about language acquisition. How does a toddler 00:14:13.269 --> 00:14:15.620 learn to think? based on the root system they 00:14:15.620 --> 00:14:18.440 are born into. Yeah. Maya Arad argues that this 00:14:18.440 --> 00:14:21.360 creates two totally different types of cognitive 00:14:21.360 --> 00:14:24.279 workloads. Think about a toddler learning English. 00:14:24.919 --> 00:14:27.269 When they are categorizing the world, They're 00:14:27.269 --> 00:14:29.450 basically building totally disconnected boxes 00:14:29.450 --> 00:14:31.809 in their mind. Because the words look and sound 00:14:31.809 --> 00:14:33.929 completely different. Exactly. The word cream 00:14:33.929 --> 00:14:36.909 has absolutely zero phonetic or visual connection 00:14:36.909 --> 00:14:39.409 to the word fat. The child just has to memorize 00:14:39.409 --> 00:14:41.029 that these are two completely separate roots 00:14:41.029 --> 00:14:43.490 for two separate things. Their brain is categorized 00:14:43.490 --> 00:14:46.929 into highly specific isolated units. But a Hebrew 00:14:46.929 --> 00:14:49.450 speaking child is learning that the world is 00:14:49.450 --> 00:14:52.629 deeply interconnected. They don't memorize isolated 00:14:52.629 --> 00:14:56.870 boxes. They learn one single root. shum, and 00:14:56.870 --> 00:14:59.409 from that one conceptual thread they can unravel 00:14:59.409 --> 00:15:02.610 the concepts of cream, fat, and oil. It's amazing. 00:15:02.830 --> 00:15:05.029 It implies that the underlying mechanics of a 00:15:05.029 --> 00:15:08.230 language literally dictate the pattern recognition 00:15:08.230 --> 00:15:10.470 required of native speakers from early childhood. 00:15:10.669 --> 00:15:13.610 The English speaker's brain is heavily categorized. 00:15:14.029 --> 00:15:16.610 The Hebrew speaker's brain is deeply networked. 00:15:16.870 --> 00:15:19.269 And researchers Alexia Dew and Lon Dahl advance 00:15:19.269 --> 00:15:21.690 the claim that languages exist on a typological 00:15:21.690 --> 00:15:23.769 scale in this regard. It's not just categorized 00:15:23.769 --> 00:15:25.409 English on one side and networked Hebrew on the 00:15:25.409 --> 00:15:27.629 other. Greek, for example, sits right in the 00:15:27.629 --> 00:15:29.309 middle between the English method and the Hebrew 00:15:29.309 --> 00:15:31.470 method. Greek is a fascinating bridge between 00:15:31.470 --> 00:15:34.809 these two architectural styles. In Greek, roots 00:15:34.809 --> 00:15:37.149 generally cannot stand alone like the English 00:15:37.149 --> 00:15:40.460 word run. They're bound. They need affixes to 00:15:40.460 --> 00:15:43.360 become real words. But Greek doesn't use the 00:15:43.360 --> 00:15:45.659 vowel interlocking skeleton system of Hebrew 00:15:45.659 --> 00:15:49.460 either. A Greek root is a solid block, but it 00:15:49.460 --> 00:15:51.840 remains grammatically blank until an inflectional 00:15:51.840 --> 00:15:53.600 ending is snapped onto it. So it's a hybrid. 00:15:54.000 --> 00:15:56.159 Yeah, it blends the solid concatenative block 00:15:56.159 --> 00:15:58.679 building of English with the strict category 00:15:58.679 --> 00:16:00.940 neutral dependency seen in Semitic languages. 00:16:01.100 --> 00:16:03.440 It just completely changes how you view the words 00:16:03.440 --> 00:16:05.960 coming out of your own mouth. We started this 00:16:05.960 --> 00:16:08.159 deep dive talking about invisible architecture. 00:16:08.720 --> 00:16:11.240 We've journeyed from the solid, standalone Lego 00:16:11.240 --> 00:16:14.159 blocks of English -free roots. We've uncovered 00:16:14.159 --> 00:16:17.559 the bound, trapped roots acting as genetic scaffolding, 00:16:17.860 --> 00:16:20.919 like erupt and interrupt. We examined the mathematical, 00:16:21.200 --> 00:16:23.679 invisible skeletons of Arabic and Hebrew, where 00:16:23.679 --> 00:16:26.620 abstract consonants act as the structural rebar 00:16:26.620 --> 00:16:30.080 for entire webs of vocabulary. And finally, we 00:16:30.080 --> 00:16:32.419 landed on the reality of category -neutral roots, 00:16:32.519 --> 00:16:35.279 these blank slates that have no idea as a noun 00:16:35.279 --> 00:16:37.799 or a verb until the sentence they are in commands 00:16:37.799 --> 00:16:40.259 them to take a shape, completely altering the 00:16:40.259 --> 00:16:42.340 cognitive workload of a child learning to speak. 00:16:42.779 --> 00:16:46.279 We take an impossibly complex world and reduce 00:16:46.279 --> 00:16:48.700 it down to a handful of irreducible concepts, 00:16:48.919 --> 00:16:51.179 and then we build infinite universes of meaning 00:16:51.179 --> 00:16:53.620 out of them. It's happening all the time, right 00:16:53.620 --> 00:16:55.940 beneath the surface. The next time you effortlessly 00:16:55.940 --> 00:16:58.960 use the word wardrobe to describe both the piece 00:16:58.960 --> 00:17:01.279 of furniture and the act of outfitting an actor, 00:17:01.659 --> 00:17:03.659 or the next time you hear a wildly different 00:17:03.659 --> 00:17:06.500 language being spoken, take a second. Appreciate 00:17:06.500 --> 00:17:08.740 the invisible linguistic engineering that your 00:17:08.740 --> 00:17:11.140 brain is executing in milliseconds. It really 00:17:11.140 --> 00:17:13.700 is an engineering marvel. It totally is. I want 00:17:13.700 --> 00:17:15.880 to leave you, the listener, with one final thought 00:17:15.880 --> 00:17:18.380 to mull over. We talked about how those ancient 00:17:18.380 --> 00:17:21.299 Semitic roots act as conceptual anchors. how 00:17:21.299 --> 00:17:25.380 the single Hebrew idea of largeness, GDL, spawned 00:17:25.380 --> 00:17:28.079 the modern words for tower, size, and magnifier. 00:17:28.180 --> 00:17:30.579 It makes you wonder, if you had x -ray vision 00:17:30.579 --> 00:17:32.539 for your own vocabulary, what would you find? 00:17:32.900 --> 00:17:35.720 How many of our modern, highly complex technological 00:17:35.720 --> 00:17:39.480 ideas are secretly just ancient, abstract concepts 00:17:39.480 --> 00:17:42.599 wearing a new grammatical structure? What invisible 00:17:42.599 --> 00:17:45.000 linguistic architecture is shaping your reality 00:17:45.000 --> 00:17:45.480 right now?