WEBVTT Kind: captions Language: en 1 00:00:00.240 --> 00:00:03.360 This is one of the most famous  experiments in linguistics. 2 00:00:03.360 --> 00:00:05.280 One of these shapes is called “bouba”. 3 00:00:05.280 --> 00:00:08.320 The other is called  “kiki”. Which is which? 4 00:00:08.960 --> 00:00:11.680 This was invented in 1929,  using slightly different words, 5 00:00:11.680 --> 00:00:13.600 and it’s been refined over time. 6 00:00:13.600 --> 00:00:16.320 For most people, the  pointy shape is "kiki" 7 00:00:16.320 --> 00:00:18.480 and the rounded shape is "bouba." 8 00:00:18.480 --> 00:00:20.160 The majority of studies… 9 00:00:20.160 --> 00:00:23.040 Wow, that is a lot of citations. 10 00:00:23.040 --> 00:00:27.440 The majority of studies find that  pointy shapes are more associated with 11 00:00:27.440 --> 00:00:31.840 unvoiced plosives and front  vowels: so “ta, pi, ka”, 12 00:00:31.840 --> 00:00:34.000 and round shapes are associated with 13 00:00:34.000 --> 00:00:38.400 voiced plosives, nasals, and  back vowels, so, “bou, gǝ, no”. 14 00:00:38.400 --> 00:00:40.080 Okay, sure, but that’s  just English, right? 15 00:00:40.080 --> 00:00:42.080 We have words like  ‘point’ and ‘balloon’, 16 00:00:42.080 --> 00:00:44.400 maybe we’re just copying the  associations from the words   17 00:00:44.400 --> 00:00:45.360 we already know. 18 00:00:45.360 --> 00:00:47.120 One of the founders  of modern linguistics   19 00:00:47.120 --> 00:00:48.960 has an entire theory named after him, 20 00:00:48.960 --> 00:00:50.640 about there being no  relation between the   21 00:00:50.640 --> 00:00:52.400 form of a word and what it represents. 22 00:00:53.360 --> 00:00:55.600 But: there’s a study  where English speakers 23 00:00:55.600 --> 00:00:58.240 were given pairs of words  in a language from Peru. 24 00:00:58.240 --> 00:00:59.040 In each pair, 25 00:00:59.040 --> 00:01:02.080 one word was for a bird,  and one was for a fish. 26 00:01:02.080 --> 00:01:04.400 And the English speakers,  who didn’t know the language, 27 00:01:04.400 --> 00:01:06.880 who knew nothing even  close to the language… 28 00:01:07.600 --> 00:01:10.240 they could sort those  words into birds and fish   29 00:01:10.240 --> 00:01:12.160 a little bit better than chance. 30 00:01:12.160 --> 00:01:16.000 Not well, but out of hundreds of people  answering thousands of questions, 31 00:01:16.560 --> 00:01:20.560 they got 58% right. That is a  statistically significant result. 32 00:01:20.560 --> 00:01:23.760 And there’s another study where  Hebrew speakers were given pairs of   33 00:01:23.760 --> 00:01:26.480 Chinese characters with  opposite definitions, 34 00:01:26.480 --> 00:01:30.000 and asked to match up the  characters and definitions. 35 00:01:30.000 --> 00:01:33.440 Again, slightly better than  chance, about 55% right. 36 00:01:33.440 --> 00:01:36.400 And if people who speak  different languages have even 37 00:01:36.400 --> 00:01:39.840 a slight ability to figure out  completely unfamiliar words, 38 00:01:40.640 --> 00:01:42.080 well, that raises the question,   39 00:01:42.080 --> 00:01:45.040 do humans have some sort  of built-in associations 40 00:01:45.040 --> 00:01:48.160 between sounds and symbols  and things in the real world? 41 00:01:48.720 --> 00:01:52.800 “Bouba” and “kiki” have been tested  in a lot of languages, and, yeah, 42 00:01:52.800 --> 00:01:54.400 there seems to be something there, 43 00:01:54.400 --> 00:01:58.080 what researchers call a type of  "crossmodal correspondence" or   44 00:01:58.080 --> 00:01:59.360 "sound symbolism." 45 00:01:59.360 --> 00:02:02.160 Correlations between  phonemes, the sounds we make, 46 00:02:02.160 --> 00:02:06.320 and traits like shape, texture,  brightness, size, or even taste. 47 00:02:07.200 --> 00:02:10.400 Maybe that’s down to cross-activation  between brain regions: 48 00:02:10.400 --> 00:02:14.240 researchers into synaesthesia  have spent a long time on that. 49 00:02:14.240 --> 00:02:16.400 Or it could be repeated association. 50 00:02:16.400 --> 00:02:18.480 An elephant makes a  deeper sound than a mouse. 51 00:02:18.480 --> 00:02:22.240 A large dog usually barks lower  and longer than a small dog. 52 00:02:22.240 --> 00:02:23.920 If you shout into a big, round cave, 53 00:02:23.920 --> 00:02:27.360 it’ll reflect back deep,  round, resonating tones. 54 00:02:27.360 --> 00:02:29.840 Shout into a tiny cave with  a lot of sharp angles in it 55 00:02:29.840 --> 00:02:31.600 and you’ll hear higher, sharper tones. 56 00:02:32.160 --> 00:02:36.480 Something hard and brittle is more  likely to make a sharp ‘kh’ sound, 57 00:02:36.480 --> 00:02:39.520 something like ‘kiki’, when you  hit it or break it or shatter it; 58 00:02:39.520 --> 00:02:43.760 something soft and round is more  likely to make a noise like “bouba”. 59 00:02:43.760 --> 00:02:46.880 I need to stress this is one  theory from a couple of papers, 60 00:02:46.880 --> 00:02:50.480 don’t take this as gospel, it’s  right to be skeptical about that. 61 00:02:50.480 --> 00:02:52.240 Plus, it doesn’t always work. 62 00:02:52.240 --> 00:02:54.400 A paper from 1975 shows the results of a   63 00:02:54.400 --> 00:02:57.360 bouba-kiki style test on Songe  speakers in Papua New Guinea, 64 00:02:57.360 --> 00:03:00.960 where the results were like they were  picking at random. No preference. 65 00:03:00.960 --> 00:03:06.000 And in 2017, another test on Syuba  speakers in Nepal, again: no preference. 66 00:03:06.000 --> 00:03:08.880 The likely reason is that  the nonsense words they chose   67 00:03:08.880 --> 00:03:10.720 could not exist in those languages. 68 00:03:10.720 --> 00:03:13.360 It’d be like giving English  speakers a test to choose between   69 00:03:13.360 --> 00:03:14.880 “ŋoba” and “tlet”. 70 00:03:14.880 --> 00:03:18.160 You can’t start a word  with ŋ or tl in English, 71 00:03:18.160 --> 00:03:19.680 so the choice doesn’t make sense. 72 00:03:20.400 --> 00:03:24.720 The frustrating thing is: there  isn’t much data on the failures. 73 00:03:24.720 --> 00:03:26.880 Researchers often don’t  publish their negative results, 74 00:03:26.880 --> 00:03:30.800 and besides it’s very difficult and  very expensive to give linguistic tests 75 00:03:30.800 --> 00:03:35.600 to people who’ve never been exposed  to any of the major world languages. 76 00:03:35.600 --> 00:03:38.240 But according to all those studies, 77 00:03:38.240 --> 00:03:41.760 that bouba/kiki distinction  is true for most people. 78 00:03:41.760 --> 00:03:45.120 There may be a link between some  sounds and some real-world properties. 79 00:03:45.120 --> 00:03:47.840 And that may be why English  speakers were able to distinguish 80 00:03:47.840 --> 00:03:50.320 birds and fish in that  Peruvian language: 81 00:03:50.320 --> 00:03:54.080 the birds’ names had more high,  front, non‐rounded vowels: 82 00:03:54.080 --> 00:03:56.720 /i/ and /e/, they had  more “sharp sounds”. 83 00:03:57.360 --> 00:03:59.840 And while there are  plenty of exceptions, 84 00:03:59.840 --> 00:04:04.000 birds have more sharp beaks and  claws than your average fish does.